Energy Conservation in Industries. Applications and Techniques.

Energy R&D Programme
of the
Commission of the European Communities
Energy Conservation in Industry -

Applications and Techniques
Proceedings of the Contractors' Meetings held in Brussels o
10 May, 17 June, 1, 7 and 22 October 1982
D. Reidel Publishing Company

Dordrecht / Boston / Lancaster
Energy Conservation in Industry
Energy R&D Programme of the
Energy Conservation in Industry

Proceedings of the Contractors' Meetings held in
Brussels on 10 May, 17 June, 1, 7 and 22 October 1982
Edited by
H. Ehringer, G. Hoyaux and P. A. Pilavachi

Directorate-General for Science, Research and Development,
PARL EUROP. Biblioth.
N. C.
Com. 3 9. Ito
k 4
A MEMBER OF THE KLUWER ACADEMIC PUBLISHERS GROUP
EUA Ski f
Library of Congress Cataloging in Publication Data
Main entry under title:
Energy conservation in industry.
At head of title: Energy R&D programme of the Commission of the
European Communities.
"Organization of the Contractors meeting by Commission of the
European Communities, DirectorateGeneral Science, Research and
Development, Brussels"T.p. verso.
1. IndustryEnergy conservationCongresses. I. Ehringer, H.
II. Hoyaux.G. III. Pilavachi, P. (Petros), 1944 . IV. Com
mission of the European Communities. V. Commission of the
European Communities. DirectorateGeneral for Science, Research and
Development.
TJ163.27.E5 1983 621.042 833149
ISBN 9027715807
Organization of the Contractors' meetings by

DirectorateGeneral Science, Research and Development, Brussels
Publication arrangements by
DirectorateGeneral Information Market and Innovation, Luxembourg
EUR 8436
Copyright 1983, ECSC, EEC, EAEC, Brussels and Luxembourg
LEGAL NOTICE
Neither the Commission of the European Communities nor any person acting on behalf of the
Commission is responsible for the use which might be made of the following information.
Published by D. Reidel Publishing Company

P.O. Box 17, 3300 A A Dordrecht, Holland
Sold and distributed in the U.S.A . and Canada

by Kluwer Boston Inc.,
190 Old Derby Street, Hingham, MA 02043, U.S.A .
In all other countries, sold and distributed

by Kluwer A cademic Publishers Group,
P.O. Box 327, 3300 A H Dordrecht, Holland
All Rights Reserved

No part of the material protected by this copyright notice may be reproduced or utilized
in any form or by any means, electronic or mechanical, including photocopying,
recording or by any informational storage and retrieval system,
without written permission from the copyright owner.
Printed in The Netherlands
PREFAC E
The main objectives of the Community's Energy policy consist

Cf securing a sufficient energy supply for the present and
future demand of its Member States and in reducing the Com-
munity's dependence on imported energy through the more ra-
tional use of energy and a broader diversification of supply.
This requires a full set of common efforts at all levels,
including energy research and development.
In the framework of its strategy for scientific and technolo-

gical research the Commission of the European Communities has
launched several R + D programmes in the field of energy.
These programmes are implemented either directly in its
Joint Research Centre or by concluding contracts with re-
search institutions in the EC Member countries.
One of the most important short and medium term objectives of

the present four-year Energy R + D Programme (1979-1983)
approved by the Council of Ministers on 13.IX.1979 aims at
fostering energy conservation technologies in the three main
energy consuming sectors : domestic, industry and transport.
In the European Community more than 35 % of the energy is

consumed in Industry. About 23 % of the imported oil is used
in this sector. Two thirds of this energy are used for hea-
ging purposes. Consequently, there is a high potential for
residuary heat recovery.
In order to enhance rational use of energy and energy conser-

vation in the Member States of the European Community, it is
necessary, among other things, to promote the development of
applications and techniques in industry, in particular as
regards energy intensive branches.
The Commission has therefore concluded with industries, public

organisations and universities of the EC Member States a
number of R + D contracts dealing with the application of
energy saving techniques in industry. The progress of work
achieved in these contracts was reported in contractors
meetings, which took place respectively on 10 May, 17 June,
1, 7 and 22 October 1982 in Brussels.
The papers presented at these meetings and the conclusion of
the discussions are collected in the present book, which is
intended to inform the Member States, research bodies and
other interested parties about existing R + D work in this
area.
I should like to address my sincere thanks to all the authors

of papers and speakers of these meetings. I should also thank
the chairmen of these meetings, Dr. J. CARRASSE, Dr. G. LIMIDO,
Mr. J. A. KNOBBOUT and Mr. D.A. REAY, v:ho animated the dis-
cussion with comments and questions. Last but not least, I
extend my thanks to the editors of the proceedings, who sum-
marized the content of the meetings and the discussions.
May I express the hope that the work reported will contribute
to rapid progress in the development and practical application
of energy saving techniques in industry, thereby contributing
to rational use of energy throughout the European Community.
Dr. A. STRUB
Director
Head of the Energy R + D Programme
ATTENDANCE LIST
1. CONTRACTORS
SESSION I : Metallurgy
M. BAFFIGI FIAT
M. VOLMI R.N.U. RENAULT
M. MICHEL R.N.U. RENAULT
E. FELDER ARMINES
J.H. SCHISSLER INSTITUT NATIONAL POLY-
TECHNIQUE DE LORRAINE
C. BAK CENTRE DE RECHERCHES
PONT A MOUSSON
T.J. DENNIS BRITISH ALUMINIUM CO. LTD.
D . P. MCGEEHIN AERE HARWELL
SESSION II : Cement and glass-ceramic industry
P. COROMPT CEA. GRENOBLE

B. GAILLARD CEA. GRENOBLE
J. GROSJEAN IRSID
E.P. SORRENTINO IRSID
C. PAILHES LAFARGE COPPEE RECHERCHE
R. MARGRITA LAFARGE COPPEE
D . P.S. ROGERS IMPERIAL COLLEGE OF SCIENCE
AND TECHNOLOGY
SESSION III : Texti Le Industry
R.B.M. HOLWEG T.N.O. FIBRE RESEARCH INST.

M. G LASTRA T.N.O. FIBRE RESEARCH INST.
J.G. ROBERTS SHIRLEY INSTITUTE
SESSION IV : Food Industry
L.W..H. DE VRIES TNO

D . A.H. BLOKSMA TNO
T.L. ONG TNO Div. for Nutrition
and Food Research
SESSION V : Chemical Industry
M. SPOTO MONTEDIPE SPA.

L. G AZZI SNAH PROG ETTI
C. RESCALLI SNAM PROG ETTI
Prof. A.C.C. TSEUNG THE CITY UNIVERSITY
Dr. A. ANTONIAN THE CITY UNIVERSITY
F. AUDIBERT I.F.P.
D.F. G IBBS I.R.D.
V.J. PATER THE BRITISH CARBONIZATION
RESEARCH ASSOCIATION
SESSION VI : Catalysis
J. DRAPIER UNIVERSITE DE LIEGE

Prof. A.J. HUBERT UNIVERSITE DE LIEGE
P. JACOBS KATHOLIEKE UNIVERSITEIT LEUVEN
P. STRUIFF KATHOLIEKE UNIVERSITEIT LEUVEN
Prof. J. UYTTERHOEVEN KATHOLIEKE UNIVERSITEIT LEUVEN
J. GROSMANGIN CIE FRANAISE DE RAFFINAGE
C. MARTY CIE FRANAISE DE RAFFINAGE
P. CHICHE CERCHAR
J. GAGET CERCHAR
Y. JACQUIN IFP
SESSION VII : Microwaves
J.F. COURTOY EUROFLOOR

L. HAAG EUROFLOOR
W. VAN LOOCK UNIVERSITEIT GHENT
M. DE POURCQ UNIVERSITEIT GHENT
Prof. P.J. LUYPAERT KATHOLIEKE UNIVERSITEIT LEUVEN
SESSION VIII : Energy Management
P. BERGER ACEC
Dr. P. KIRSCH ACEC
J.F. DE GREEF SCK/CEN
2. Commission
Dr. H. EHRINGER DG XII

G. HOYAUX DG XII
Dr. P.A. PILAVACHI DG XII
P. ZEGERS DG XII
3. Chai rmen
Dr. J. CARRASSE expert SYNTEC-INGENIERIE FRANCE

Prof. Dr. J.A.A. KETELAAR expert UNIVERSITY OF AMSTERDAM
J.A. KNOBBOUT expert
expert TNO - The Netherlands
Dr. J. LIMIDO expert I.F.P. - ENSPM FRANCE
D. REAY I.R.D. - UK
C O N T E N T S
Preface
Attendance List
Introduction
H. EHRINGER, Commission of the European Communities
SESSION I : METALLURGY
Contractors meeting held on 22 October 1982

Chairman: J. LIMIDO
Summary of Session I 4
Economy storage in forged parts

Rgie Nationale des Usines Renault (R.N.U.R.)
Socit des Aciers Fins de l'Est (S.A.F.E.)
M. VOLMI, R. EL HAIK, L. BACKER, F. LARDET 6
Development of bainitic nodular iron for the construction of

speed gears for the car industry
Fiat Auto S.p.A.
M. BAFFIGI, S. CORSO, F. RABEZZANA 32
Lubrication by slags of the continuous casting of steels

ARMINES
E. FELDER, Ch. NIGGEL 51
Economie d'nergie par mise en oeuvre de nouveaux cycles

thermiques appliqus des pices de fonderie dmoules
chaud juste aprs la fin de solidification
Centre de Recherches de Pont--Mousson
C. BAK, R. BELLOCCI
Institut National Polytechnique de Lorraine
J.M. SCHISSLER 62
Reduction in energy requirements for converting liquid aluminium

to semi-fabricated rolled products
The British Aluminium Company pic
T.J. DENNIS 79
Composite metal jointing

AERE Harwell
P. McGEEHIN 92
SESSION II - CEMENT AND GLASS-CERAMIC INDUSTRY

Chairman: J. CARRASSE
Summary of Session II 118
Etude par traceurs des paramtres de transfert du systme

rgulant la composition chimique de l'alimentation des fours
ciments afin d'optimiser leur rendement
Commissariat l'Energie Atomique
P. COROMPT, B. GAILLARD, J. GUIZERIX, M. LAUNAY 120
Echangeur thermique cyclones, mise au point d'une mthode

pour dfinir et amliorer ses performances
Lafarge Coppe Recherche (L.C.R.)
C. PAILHES
Centre d'Etudes Nuclaires de Grenoble (C.E.N.G.)
R. MARGRITA, SANTOS GTTIN 133
Study of a treatment of oxygen steelmaking slag in view of its

use by the cement industry
Institut de Recherches de la Sidrurgie Franaise (IRSID)
R. PAZDEJ, J.C. GROSJEAN, M. HAMMAN
Lafarge Fondu International (LFI)
C M . GEORGE
E.P. SORRENTINO (Lafarge Coppe) 147
Ceramic materials from molten blast-furnace slags by direct

controlled cooling
Imperial College of Science and Technology
P.S. ROGERS, J. WILLIAMSON, J.F. BELL, M. CAMPBELL 156
SESSION III - TEXTILE INDUSTRY

Summary of Session III 168
Industrial application of continuous wet-on-wet treatment of

fabrics webs
TNO - Fibre Research Institute
R.B.M. HOLWEG 170
Low energy preparation processes for textiles
Shirley Institute
J.G. ROBERTS 184
SESSION IV - FOOD INDUSTRY
Contractors meeting held on 17 June 1982

Chairman: D.A. REAY
Summary of Session IV 198
Energy saving in the bakery by improvement of energy efficiency

and recovery of waste heat
TNO - Institute for Cereals, Flour and Bread
L. DE VRIES 200
Energy saving in the soybean extraction industry by reducing

the steam consumption for desolventizing - Toasting and
drying extracted beans
TNO - Institute CIVO - Technology
T.L. ONG 216
Energy saving in edible oil processing plants by application of

a total-energy (TE) system - Part I: Soybean oil extraction plant
TNO - Institute CIVO - Technology
T.L. ONG 230
SESSION V - CHEMICAL INDUSTRY

Chairmen: J. LIMIDO and J.A. KNOBBOUT
Summary of Session V 238
Modification of cracking furnaces of existing plants to increase

yields of valuable products and to reduce fuel consumption
Montedipe S.p.A.
M. SPOTO 240
Selective removal of HJ3 from gaseous mixtures containing C0 p

SNAMPROGETTI S.p.A.
C. RESCALLI 253
Energy conservation in the chlor-alkali industry

The City University
A.C.C. TSEUNG, J.A. ANTONIAN 263
Thermal depolymerisation of waste tires by heavy oils

- Conversion into fuels
Institut Franais du Ptrole
F. AUDIBERT, J.P. BEAUFILS 275
A critical evaluation of anaerobic fermentation of waste products
(A management study)
International Research & Development Co. Ltd.
M.F. ANDERTON, D.F. GIBBS 286
The recovery, treatment and utilization of solid bearing

effluents emanating from coke ovens
The British Carbonization Research Association
V.J. PATER, G.A. WADE 299
SESSION VI - CATALYSIS
Contractors meeting held on 10 May 1982

Chairman: J. LIMIDO
Summary of Session VI 316
Catalytic fuel production using water and coal

Universit de Lige
J. DRAPIER, A.J. HUBERT, Ph. TEYSSIE 318
Fischer-Tropsch synthesis of hydrocarbons in the gasoil range

Katholieke Universiteit Leuven
P. STRUYF, P.A. JACOBS, J.B. UYTTERHOEVEN 331
Valorisation des asphaltes par hydrognation trs haute

svrit sur catalyseur entran
Centre d'Etudes et Recherches des Charbonnages de France
(CERCHAR) et Institut Franais du Ptrole (IFP)
P. CHICHE, J. GAGET, P. GIULIANI, Y. JACQUIN 3M9
SESSION VII - MICROWAVES
Contractors meeting held on 17 June 1982

Chairman: D.A. REAY
Summary of Session VII 372
Application of microwave heating and hot air for the continuous

production of plasticized PVC sheets
EUROFLOOR S.A.
A. ROUSSEL, L. HAAG 375
Microwave properties of some building materials in view of an

oven design
State University of Ghent
M. DE POURCQ, C. DE WAGTER, W. VAN LOOCK 387
Development of a microwave applicator to dry thin structures -
First semi-annual report
Katholieke Universiteit Leuven
J.P. LUYPAERT, P. REUSENS, H. LEMMENS 3%
SESSION VIII - ENERGY MANAGEMENT

Summary of Session VIII 406
An assessment of the energy needs of a heavy electrical

products factory with a view to their reduction
A.C.E.C. (Ateliers de Constructions Electriques
de Charleroi)
P. BERGER 108
Economic study of the energy exchange between factories in

an industrial site
C.E.N. - S.C.K.
J.F. DE GREEF, A. WOUTERS 423
I N T R O D U C T I O N
by
Hermann EHRINGER
Industry consumes 35? of the final energy in the European Community.

About 23? of the imported oil are used in this area. Two thirds of this
energy are intended for heating purposes. In spite of intensive rational-
ization efforts undertaken by industry in the last decade, there is
still an important energy saving potential in all industry branches,in
particular in the energy intensive ones. Therefore, the European Community
supports R & D work on energy saving techniques for application in
metallurgy, chemical, cement, food and textile industry.
The purpose of the contractors meetings is to give the project leaders the
opportunity to report on the progress of their work and on the results
achieved. Moreover, these meetings serve to establish a link between
contractors working in the same area and to strengthen cooperation on a
European level.
The papers presented at these meetings and the discussion will be recorded
in proceedings which will be published in order to meet the provisions
laid down in the Council regulation EEC/2380/74 of 17 September 1974.
SESSION I : METALLURGY
Chairman: J. LIMIDO
Summary of Session I
Economy storage in forged parts
Development of bainitic nodular iron for the construction

of speed gears for the car industry
Economie d'nergie par mise en oeuvre de nouveaux cycles

thermiques appliqus des pices de fonderie dmoules
chaud juste aprs la fin de solidification
Reduction in energy requirements for converting liquid

aluminium to semi-fabricated rolled products
Composite metal jointing
-3-
SUMMARY OF SESSION I
METALLURGY
*) . . , **) ., , .***)
J.A. Knobbout , J. Lirmdo , P.A. Pilavacrn
Broadly speaking, Metallurgy is an energy-intensive sector and has a great

potential where energy conservation is concerned.
Substantial savings in energy or raw materials can be made by improving the

process of moulding or forging steel or pig-iron products. This is
achieved by using new, simplified quenching and annealing processes for
hardening the steel.
The following three projects have this aim.
Hot strip_p_ed_caiQ2 The research work carried out by the CENTRE DE

RECHERCHES D E P N T ~ A ~ M 0 U S S 0 N with the participation of the ECOLE D ES MINES
DE L'INSTITUT POLYTECHNIQUE D E LORRAINE, is aimed at developing new thermal
treatment techniques for treating cast nodular pigiron which has been
removed from the moulds while hot. Specific quenching techniques may lead
to the same high mechanical properties, while the usual intermediate
cooling and heating stages are avoided. The first results have shown that
the procedure is feasible. Nevertheless the new thermal cycles which
yield remarkable energy savings (these have still to be assessed) neces
sitate optimization of the conditions to ensure that the structures
required are in fact obtained.
Forged_p_arts The second project, undertaken by the REGIE NATIONALE D ES

USINES RENAULT (R.N.U.R.) aims at achieving savings of energy and raw
materials by forging steel parts for motorcars. The following two routes
are used :
D irect forging of bars obtained by continuous casting, thus avoiding
the rolling phase, and
D irect treatment after forging without the reheating phase.
The practical application of these new cycles requires close cooperation

with the forging subcontractors of R.N.U.R. The first contacts have
already been made and the technology of the new equipment developed will
be transferred to and controlled by the subcontractors.
*) TNO, The Netherlands

**) IFPENSPM, France
***) Commission of the European Communities
aD_D.2dy.kar._.20 ~ The a ^ m ^ *^ third project given to FIAT Lies in
developing bainitic nodular iron, as an alternative to hardened, tempered
and casehardened 18 Cr Mo 4 steel. This material would meet the specifi
cations for speed gear manufacture but would require less energy and raw
materials. Tests regarding mechanical properties and machinability have
yielded positive results; however, deformation problems which appear
during the thermal treatment have still to be solved.
Ly2Il2D_2_tt!_2DDy2y2_itD2 ~ The fourth project given to

ARMINES, aims at improving the application of continuous casting of steels
which, as compared with the classical route, i.e. ingot casting and
rolling, yields large energy savings and gives better surface quality.
An improvement is expected through a better formulation of the mould powder
(slag) of which the lubrication action is important. The contribution of
the ARMINES research is to build a reliable model for a rational formula
tion of casting powders.
2DyEID2_i:9y2'_iy!!!Dy!D_2_2E!I22S2'_!2S2'_E22'y2I ~ ^n t'le
aluminium industry, BRITISH ALUMINIUM Co. Ltd. has undertaken research
work the scope of which is to develop methods of producing a coiled strip
product which are alternatives to the conventional method of direct chill
(DC) casting hot rolling. Continuous thinstrip casting methods appear
to be particularly worthy of investigation, since they eliminate the
scalping, preheating and hot rolling processes required with the conven
tional method and hence offer a potential for substantial energy savings.
Comgosie_meal_jo2ni.n2 ~ In the transport sector, one way to save energy

is to reduce the weight of motoicars and trucks. A clear way to reach
this objective is to make more extensive use of composite materials and of
adhesive bonding techniques. Besides, this would provide an improved
design. The aim of the project entrusted to AEREHARWELL is to develop
proven design procedures for structural joints between composites and
metals. The project is supported by 15 firms including manufacturers of
motorcars.and trucks, adhesives, and composite materials. Three typical
potential applications have been chosen : a torsional joint, a chassis
bracket and the hinge attachment of a hatch backdoor.
First results indicate the value of the project and the importance of
industrial support for an extensive implementation of the results.
ECONOMY STORAGE IN FORGED PARTS
Authors M. VOLMI, R. EL HAIK, L. BACKER, F. LARDET
Contract number EEB-I-115-F (S)
Duratxon 30 months 1-3-1981 - 1-9-1983
Head of project : R. EL HAIK - Direction des Laboratoires R.N.U.R.

assisted by
L. BACKER - Direction des Dpartements Mtallurgi-
ques - S.A.F.E
F. LARDET - Direction des Mthodes Centrales Forges/
Traitements R.N.U.R
Contractor : REGIE NATIONALE DES USINES RENAULT (R.N.U.R)
SOCIETE DES ACIERS FINS DE L'EST (S.A.F.E)
Address : 8-10 avenue E. Zola - 92109 BOULOGNE BILLANCOURT
40, rue de Paris 92100 BOULOGNE BILLANCOURT
(Usines HAGONDANGE - 57301
Summary
The present overall economic situation led use to reanalyse the manufac-
ture of forged parts, in order to find solutions enabling to reduce
their energy content while maintaining their utilization specifications.
Two main topics were studied :
1.- Suppression of rolling process with the parts being directly forged
from cast bars having correct surface finish. The lesser centre compact-
ness connected to this type of product is reduced through hot working
For a hot working ratio of about 4, the mechanical properties obtained
are sufficient. Two axisymetric parts (wheel hub, differential crown
wheel) have given satisfying results in endurance tests.
2.- Reduction or suppression of heat treatment after forging
Three types of direct treatment after forging are applied in our produc-
tion processes :
- Isothermal annealing (gears in 30 CD 4)
- Classical annealing (ex : connecting rods in 45 M 5)
- Water quenching (ex : steering knuckle in 20 MB 5)
Research work enabled to define for each case, the cooling laws ensuring
the best compromise between part quality (hardness, structure, endurance
impact strength) and machinability, as well as to plan the industrial
equipments necessary to carry out these treatments.
Regarding direct isothermal annealing, the relatively coarser grain
obtained ensures a substantial gain during machining. The final quality
treatment (carbonitriding quenching), enables to obtain the structure
of the present mass-produced parts.
-6-
1.- Introduction
Les circuits de fabrication industriels pour pices de forge ont t
tablis pour tenir un niveau lev de performance et garantir une grande
fiabilit du produit. Les cycles de transformation et de traitement concer-
ns entranent un engagement nergtique important.
La conjoncture actuelle nous a amen rexaminer les possibilits de
bilans nergtiques plus judicieux.
Le but vis dans ce travail de recherche a donc consist, tout en
maintenant les caractristiques d'emploi des pices, rduire pour chaque
opration de fabrication les engagements matire et la consommation nerg-
tique. La recherche s'est donc oriente vers la suppression ou la rduc-
tion de certaines phases du cycle de fabrication de la pice sans affecter
les caractristiques de celle-ci.
Les deux voies principales suivantes ont t tudies :
a) La suppression du laminage avec un forgeage des pices partir de
barres coules
b) La rduction ou la suppression du traitement thermique aprs forge
trois types de ces nouveaux traitements couvrent l'ensemble des cas
des pices forges pour nos fabrications
- le recuit isotherme direct
- le recuit classique direct
- la trempe directe l'eau ou l'huile
2.- Suppression du laminage et forgeage partir de barres coules

Une barre brute de coule prsente des caractristiques mcaniques
(rsistance, tnacit, fatigue) faibles et htrognes, un tat struc-
tural donn, du fait de la grosseur de grain, de l'absence de compacit
du mtal, de la sgrgation, e t c . .
Le corroyage apport par le laminage, puis le forgeage, permet de
supprimer ces inconvnients et relve les proprits mcaniques. Le for-
geage correct apporte en plus un fibrage accentuant l'amlioration pour
les contraintes appliques dans le sens des fibres.
La connaissance de l'volution des proprits en fonction du taux
de corroyage montre que cette volution est asymptotique et que, au-del
d'un certain taux de corroyage critique, l'nergie dpense est pratique-
ment perdue.
Il existe donc des familles de pices forges o le corroyage dans
la zone sollicite est suffisant pour atteindre cet optimum par forgeage
partir d'bauche non lamine.
C'est donc pour vrifier cette ide que des essais ont t entrepris
en utilisant des fournitures aux dimensions adquates.
2.1.- Mode d'obtention des barres coules

Deux technologies de ralisation de ces produits sont en concurrence
- La coule continue verticale centrifuge (C.C.C.) qui permet de
raliser des barres de diamtre mini de l'ordre de 90 mm. Ce produit
a tat de surface correct est apte au forgeage. La figure 1 prsente
une installation de coule CCC et des macrographies sur coupe d'un
acier 16CD4
- La coule continue horizontale (CCH) qui permet de raliser des
fournitures des diamtres plus compatibles avec les besoins des
forgerons de l'automobile (diamtresde 30 90 m m ) . Cependant, les
conditions d'extraction de la barre coule induisent des replis
superficiels circulaires suivant un "pas d'extraction", dfavorables
pour le forgeage direct (bien que certaines geometries pourraient
s'en accommoder si le "dfaut rsiduel" aprs forge disparat par
usinage). Les aciristes CCH font de gros efforts actuellement pour
attnuer cette "singularit".
La Figure 2 prsente une installation de coule CCH et l'aspect
de la surface de barre coule en 32 C 4 0 46.
2.2. Remarques propos de la gamme de forgeage de ce produit

De nombreuses machines forger sont encore alimentes avec des
lopins dcoups froid et chauffs au four gaz ou par induction.
Les diverses fournitures de l'tude (32 C 4 0 46, 30 CD 4 0 59,
40 CD 4 0 75, 16 CD 4 0 90) n'ont pas permis de raliser des lopins
correctement cisaills froid mme aprs adoucissement par traite
ment thermique des barres. Dans ce cas,seul le sciage garanti une
qualit de coupe convenable n'entranant pas de dfaut sur pice
forge. Cependant l'alimentation de machine forger horizontale
multiposte, en barre chauffe par induction permet un cisaillage
ais du lopin haute temprature (i\>1200C).
2.3. Etude des caractristiques d'emploi de ces fournitures

Les deux aspects suivants ont t examins :
2.3.1. Caractristiques mcaniques sur prouvettes
Ces essais sont raliss pour divers taux de corroyage sur
lopins contreforges par crasage du lopin en galet ou par rduction
de section du lopin en tirage la forge pour des taux de corroyage
de 0 10.
L'tude qui a port plus particulirement sur la resilience
en long et en travers, sur les caractristiques de traction R, E,
A, l'tat tremp et revenu pour 1000 MPa a permis de vrifier
sur les aciers 16 40 CD 4et sur 32 C 4 que ces caractristiques
croissent assez rapidement entre l'tat coule (corroyage 0) et
des taux de corroyage de l'ordre de 3 4 o elles se stabilisent
et deviennent analogues aux valeurs obtenues sur des produits lami
ns classiques.
La Figure 3 prsente les graphiques d'volution de ces ca
ractristiques
2.3.2. Essais sur pices
Pour l'tude de la tenue en fatigue, les essais sur pices
nous ont paru plus judicieux. Par ailleurs ces essais ont permis
galement de vrifier le comportement du matriau aux diffrents
stades de la fabrication, (mise en forme, usinage, qualit des struc
tures aprs traitement thermique ou thermochimique) et s'assurer
qu'aucun dfaut ou dformation anarchique n'affectaient la qualit
de la pice avant sa mise en essai sur un banc d'endurance.
Les deux premires pices choisies, moyeu de roue en 37 C 4
tremp et revenu pour 1000 MPa et la couronne de diffrentiel en
16 CD 4 cment ont t ralises partir de mtal CCC en diamtre
90.
Ces pices relativement massives gomtrie de rvolution
s'accommode assez bien d'un forgeage partir de barres coules 0 90.
Comme le montre les photos de la Figure 4, les zones les plus sol
licites (zone de la denture pour la couronne de diffrentiel
photo 9242, et plateau et raccordement ft/plateau pour le moyeu
photo 9241) sont des zones fort corroyage lors de la mise en forme
et concernent par ailleurs la partie la plus compacte du lopin ini
tial. Sur la zone centrale de la pice, moins corroye, on retrouve
les vestiges de la partie centrale du lopin. Toutefois, cette zone
moindre compacit disparat lors de l'usinage de l'alsage.
Les essais d'endurance ont t raliss au banc d'organe
dans des conditions qui refltent les sollicitations les plus sv-
res rencontres sur vhicule. Les rsultats ont t satisfaisants
pour ces deux pices dont le comportement a t analogue celui
des pices de la srie.
L'introduction de cette fourniture dans nos fabrications
va se poursuivre par le suivi d'essais industriels sur plusieurs
milliers de pices.
D'autres pices sont en cours d'tude et plus particulire-
ment partir de fournitures obtenues en coule continue horizontale.
3.- Rduction ou suppression du traitement thermique aprs forge

Les pices brutes fabriques en forge chaud subissent aprs forgea-
ge un refroidissement jusqu' l'ambiante avant d'tre reprises pour un
traitement thermique ultrieur.
Jusqu' prsent, ce refroidissement tait impos soit par des con-
traintes pratiques lies 1'loignement des installations de traitement
des machines de forgeage, soit et surtout pour assurer une rgnration
de l'acier par affinage du grain mtallique par rchauffage ralis
une temprature suprieure au point de transformation AC 3. Le souhait
du mtallurgiste, de limiter le grossissement du grain de l'acier lors
du forgeage (vitant ainsi toute rgnration ultrieure), s'est vu exauc
en partie par l'apparition de nouvelles nuances d'aciers grce la pr-
sence d'lments tels que le niobium, le vanadium et le titane par exemple
faible teneur (0,10% environ).
Des travaux rcents nous ayant montr qu'un grain relativement gros-
sier n'est pas obligatoirement nfaste la tenue mcanique d'une pice
(endurance, rsistance au choc, etc..) pourvu que la sous-structure reste
fine ; il nous a paru intressant de nous engager dans cette voie qui
permet d'utiliser des aciers classiques vitant le surcot des nuances
additions particulires.
Il apparat donc possible de traiter thermiquement une pice au sor-
tir de la machine de forge en utilisant la chaleur disponible dans la
pice.
3.1.- Remarques propos du choix du cycle de refroidissement

Ces remarques concernent plus particulirement les deux types de
recuits cits.
Pour chaque nuance d'acier , il existe des abaques temprature/temps
qui permettent de prvoir ce que sera la constitution micrographique
d'un acier l'ambiante pour un cycle donn de refroidissement,
que ce soit par transformation de l'austnite dans des conditions
isothermes (abaque T.T.T., transformation, temprature, temps) ou
dans des conditions de transformation par refroidissement continu
(abaque T.R.C).
La connaissance de ces abaques a dj permis, dans le pass
de dterminer les cycles de refroidissement sur des matriaux aust-
niss par rchauffage au-dessus de AC 3 (avec une surchauffe de
l'ordre de 50C conduisant un grain austnitique normal).
Les pices prises en charge pour refroidissement partir
de la "chaude de forge" auront obligatoirement un grain austnitique
de taille diffrente (voqu ci-aprs).
Ces mmes abaques seront utiliss pour le choix de la loi
de refroidissement approprie, sachant que la dformation chaud
de la pice entrane un certain affinage de la structure.
3.2.- Remarques propos de l'affinage du grain lors de la dformation
chaud
Le chauffage du lopin de mtal avant forgeage se fait temprature
leve (u 1200C), bien que le sjour cette temprature soit
court, le grain d'austnite obtenu est grossier.
Pendant le formage de la pice, le grain d'austnite qui se forme
est uniquement le rsultat du mcanisme de recristallisation pendant
et tout de suite aprs la dformation.
Le rsultat de la recristallisation est influenc par toute une
srie de grandeurs, savoir :
- la temprature de prchauffage du lopin ,
- la temprature de dformation,
- le mode et le taux de dformation,
- la vitesse de dformation,
- le temps de maintien la temprature de dformation aprs dfor-
mation,
- la forme de la pice (zones de massivets diffrentes et taux
de dformation diffrents).
L'obtention d'une grosseur de grain d'austnite bien dfinie dans
une pice dans ces conditions prsente donc quelques difficults.
Pour illustrer ces deux remarques, nous avons choisi un cas simple
de forgeage de lopin en acier XC 42 dont les rsultats sont prsen-
ts sur les Figures 5 et 6.
3.3.- Remarques propos de 1'usinabilit du produit

La gamme de fabrication d'une pice forge chaud comporte nces-
sairement un usinage d'importance variable avec la pice concerne.
Cet usinage peut tre l'opration ultime avant montage. Dans ce
cas, les traitements thermiques (recuit, trempe et revenu) confrant
les caractristiques d'emploi sont raliss sur pices brutes de
forge (cas des bielles, porte-fuse, e t c . . )
Dans d'autre cas, le traitement thermique ralis sur pice forge
est appliqu pour obtenir une structure apte l'usinage (recuit
isotherme par exemple), c'est le cas de la pignonnerie o le traite-
ment de qualit (cmentation ou carbonitruration trempe) est ralis
en final sur pice usine.
Dans le dveloppement des traitements directs aprs forge, il est
donc important que, paralllement l'tude des proprits mcani-
ques du produit, on vrifie que cette nouvelle technologie de trai-
tement conduise des produits bonne usinabilit.
Le critre d'acceptation retenu par l'atelier usineur reste videm-
ment le bon comportement en atelier de lots d'essai significatifs.
Cependant, la mise au point du cycle de traitement direct se fait
d'abord sur chantillon prouvette dans des conditions de laboratoi-
re avant de passer aux essais sur pices en atelier.
L'apprciation de 1'usinabilit sur petits chantillons fait appel
l'essai suivant mis au point au Laboratoire de la RNUR il y a
une dizaine d'annes.
-10-
Critre d'usinabilit D 0,20 (MATHON - RENAULT)
1
mal al usine
amer 100C6.60HRC
L'essai est effectu comme l'indique le schma ci-dessus. On utilise

un outil de capacit de coupe rduite en acier 100 C 6 ; ceci permet
d'excuter des essais de dressage avec usure complte de l'outil
des diamtres compris entre 6 et 40mm. Le relev du profil usin
qui devrait tre un plan si l'outil ne s'usait pas donne le recul
du bec d'outil par usure en fonction du diamtre usin. Le critre
d'usinabilit D 0,20 retenu est donc le diamtre usin correspondant
un recul de 0,20 mm dans des conditions de coupe dtermines.
Le critre d'usinabilit D 0,20 est en relation avec la vitesse
de coupe employe en pratique. Si 1'usinabilit D 0,20 crot, la
vitesse de coupe crotra approximativement comme celle-ci, mais
elle crotra dans des proportions diffrentes pour chaque type d'u-
sinage.
.- Recuit isotherme direct (recuit pour usinage)

3.4.1.- Recuit isotherme classique d'aciers 30 - 35 CD 4
Les recuits isothermes industriels raliss dans les meil-
leures conditions ne sont que "pseudo isothermes" (semi-continus
et semi isothermes). En effet, il n'est pas toujours facile de ma-
triser de faon relativement homogne l'abaissement rapide de temp-
rature d'un plateau de pices de 50 100 kg entre 900C et 850C
pour satisfaire aux lois de refroidissement imposes par les abaques
TTT ou TRC (Figure 7 ) .
Malgr cette singularit, ces traitements permettent depuis
des annes de fournir aux ateliers d'usinage des produits struc-
ture ferrite-perlite de morphologie correcte avec des dispersions
de durets compatibles avec les conditions d'usinage des ateliers
de fabrication.
3.4.2.- Recuit isotherme direct

Au cours de travaux prliminaires importants, nous avons
tents de reproduire des structures analogues la srie en abais-
sant rapidement la temprature des pices de 1100C 650C en les
plongeant dans un four lit fluidis, mais ce cycle dlicat ma-
triser nous a conduit opter pour un mode opratoire capable d'as-
surer une meilleure reproductibilit de traitement. Nous avons
donc dfini un cycle de refroidissement deux paliers isothermes
825C et 650C permettant sur un pignon en 30 CD 4 d'obtenir d'une
manire parfaitement reproductible sur toutes les pices d'une fa-
brication une structure ferrite-perlite gros grain parfaitement
bnfique en usinage et restaurable lors du traitement final de
carbonitruration trempe de la pice.
La figure 8 prsente le schma du cycle et les structures
ralises. Des essais significatifs une chelle industrielle
(3000 pices ont t ralises suivant cette technologie. Deux fours
lits fluidiss sont accols une machine forger horizontale
(6x4) de la SAFE.
La figure 9 prsente le schma d'un four lit fluidis
et une photo de l'quipement d'essai.
Compte tenu de la capacit des lits fluidiss 825C et
650C (200 pices environ), deux campagnes de 15 lots de pignons
en 30 CD 4 et 30 CD 4 R (nuance hautement resulfure) ont t rali-
ses. Les divers contrles de duret et structure montrent une bonne
homognit de ces caractristiques.
Les mesures d'usinabilit D 0,20 MATHON-RENAULT vrifient
sur de nombreux chantillons le gain d'usinabilit apport par le
gros grain, gain qui vient se cumuler avec les amliorations appor-
tes par la teneur en soufre plus leve dans le cas du 30 CD 4 R
la figure 10 prsente les rsultats de duret et d'usinabilit.
Les deux lots de 3000 pices suivis en atelier d'usinage
et de traitement ont conduit des rsultats trs satisfaisants,
un gain notable en usinage a t constat ; cependant l'atelier
devra s'accommoder pour l'vacuation de certains copeaux morpho-
logie particulire.
Par ailleurs, les contrles dimensionnels aprs traitements
thermochimiques ont permis de s'assurer qu'aucune dformation anar-
chique n'tait craindre avec cette technologie.
L'tude se poursuit par la mise au point d'un quipement
industriel capable d'absorber en continu les cadences de la machine
de forge.
Recuit classique direct aprs forge

Pour des pices de faibles rsistances 600 800 MPa, un recuit
de normalisation est le traitement de qualit retenu pour assurer
ce niveau de caractristiques sur des nuances telles que l'A 60
ou le 4 5 M 5. Nos bielles sont actuellement en 45 M 5 et c'est donc
sur cette nuance que nous avons dcid de commencer l'tude du re-
cuit direct aprs forge.
Le recuit de normalisation actuel consiste raustniser jusqu'
850C pendant 30 minutes, les pices en vrac dans des paniers et
les laisser refroidir l'air calme (R.A.C). Le refroidissement
en lot assure pour l'ensemble des pices une certaine loi de refroi-
dissement permettant d'avoir une structure ferrite-perlite exempte
de constituants intermdiaires de trempe du type bainite.
Dans le cas du recuit direct, le traitement consiste prendre en
charge chaque pice au sortir de la machine de forge une tempra-
ture de 1000 1100C , d'abaisser sa temprature assez rapidement
jusqu' 800-850C pour viter que le grain affin par forge risque
de regrossir par sjour trop haute temprature. Ensuite il faut
assurer sa transformation structurale ferrite-perlite entre 800
et 550C dans le dlai impos par l'abaque T.R.C, du 45 M 5, qui
est de l'ordre de 15 mn, ce qui parat difficile tenir par simple
refroidissement air calme des pices l'unit. Il faut donc freiner
ce refroidissement.
La figure 11 prsente l'abaque TRC du 45 M 5 ainsi que les cycles
de traitement "recuit classique" et "recuit direct aprs forge".
La loi de refroidissement "direct" ainsi ralise a permis d'obtenir
des pices de duret correcte et homogne, une structure morpholo-
gie particulire mais acceptable pour l'usinage (essai d'usinabilit
D 0,20). Par ailleurs, ces pices essayes en fatigue en traction
compression ont eu un comportement satisfaisant et analogue celui
des pices de srie.
La figure 12 prsente les rsultats de structure et d ' usinabilit
3.5.1.- Mise en oeuvre d'quipements d'essai pilote chez les for-

_ grons
L'une des difficults importantes^, surmonte pour les essais
pilotes a consist mettre en place chez les divers forgerons sous-
traitants un quipement d'essai (tunnel de refroidissement calorifu-
ge) permettant sur plusieurs milliers de pices (quantit ncessaire
pour un suivi d'usinage industriel) de raliser le cycle de refroi-
dissement tel qu'il a t dfini et cela de la premire la derni-
re pice.
En marche industrielle cela signifie que le dmarrage d'une
cadence de bielle avec refroidissement direct implique une mise
en rgime de temprature du tunnel avant que la premire pice y
soit introduite pour qu'elle puisse accomplir son refroidissement
suivant le cycle cit. Des couples thermolectriques placs en dbut
milieu et fin de tunnel ont permis de saisir un "profil de tempra-
ture optimis" de l'enceinte pour raliser notre cycle de refroidis-
sement.
Pour nos essais, le "profil de temprature" a t obtenu
par apport pralable de pices chaudes sur le tapis convoyeur avan-
ant dans le tunnel. En marche industrielle, un matriel plus so-
phistiqu serait ncessaire (prchauffage lectrique avec asservis-
sement de ce chauffage et de l'avance du tapis par les couples
thermolectriques) pour rpondre avec souplesse aux alas de la
fabrication (incidents sur machine de forge, e t c . . )
La figure 13 prsente l'un des tunnels de refroidissement
ralis pour ce traitement ainsi qu'un relev des profils de temp-
rature.
3.5.2.- Ralisation des lots de pices pour suivi d'usinage
Plusieurs lots de pices ont t raliss sur diverses ins-
tallation. Ceux pour lesquels le cycle de refroidissement a pu tre
bien matris (durets homognes pour l'ensemble des pices du lot
structures correctes sans constituant bainitique) ont conduit
des rsultats satisfaisants en usinage et analogues ceux obtenus
sur pice de srie.
De nouveaux essais sont en cours avec un matriel de refroi-
dissement mieux adapt aux conditions industrielles et sur des biel-
les en acier 45 M 5 S (nouvelle nuance resulfure adopte pour sa
meilleure usinabilit).
Trempe directe aprs forge

L'ide de tremper directement une pice aprs forge afin d'conomi-
ser le rchauffage d'austnisation avant trempe n'est pas rcente,
mais des difficults d'ordre technologique avaient frein jusqu'
prsent son dveloppement. Les perfectionnements de la mcanisation
du matriel de forge, la mise en place d'quipement de chauffage
mieux adapt associ une pyromtrie plus fiable (prchauffe par
induction des lopins avec rgulation des tempratures) ont permis
une meilleure matrise et donc une meilleure reproductibilit des
tempratures tous les niveaux de la gamme de forgeage de la pice
assurant ainsi une plus grande rgularit dans les dbits et les
tempratures des pices forges
Des travaux antrieurs nous ayant montr que la trempe directe aprs
forge pouvaient conduire des produits ayant un niveau de caract-
ristiques quivalent celui obtenu par le traitement classique,
nous avons donc dcid d'tudier ce traitement sur une pice repr-
sentant des tonnages trs importants dans nos fabrications : le
porte-fuse.
Simultanment l'tude a port sur pices et sur prouvettes pour
examiner le comportement de l'acier 20 MB 5 utilis pour cette fa-
brication.
Le porte-fuse est une pice "de scurit" qui doit rpondre des
exigences de qualit o les trois aspects suivants sont concerns
a) la duret (1000 MPa) obtenue par trempe et revenu
b) la tenue au choc (resilience)
c) la tenue en endurance
Les essais entrepris ont permis de vrifier que le grain plus gros-
sier inhrent la trempe directe n'est pas nfaste la qualit
du produit, la sous structure restant relativement fine. Pour ce
matriau tremp et revenu, la resilience est dj naturellement
surabondante et la faible baisse de resilience enregistre en trempe
directe permet de garantir aux pices une tenue au choc suffisante
mme aux basses tempratures, d'aprs l'tude de la temprature
de transition de la resilience.
La figure 14 prsente les structures obtenues, ainsi que les courbes
sur la temprature de transition de la resilience
La tenue en endurance vrifie sur porte-fuse s'est rvle trs
satisfaisante et l'homognit des rsultats peut s'expliquer comme
suit :
- la trempe l'eau directe de pice l'unit dans des conditions
parfaitement reproduites
- la meilleure trempabilit de ce matriau en "trempe directe" con-
duisant des durets moins disperses
- une dcarburation moindre et un tat de surface moins perturb
du fait de l'absence du second rchauffage pour austnisation
avant trempe
Un essai industriel sur 5000 pices a permis de s'assurer de la
bonne matrise de la "chane thermique" de cette technologie. Le
suivi en usinage de ces pices a donn des rsultats satisfaisants,
1'usinabilit du produit se situe au mme niveau que celui des four-
nitures de srie.
Par ailleurs, il s'est vrifi qu'un "usinage panache" de pices
"tremp direct" et de pices "tremp classique" tait parfaitement
compatible avec les rglages d'usinage de la chane actuelle.
Compte tenu de la meilleure trempabilit d'un matriau en "trempe
directe", l'tude s'oriente vers l'emploi de nuances plus conomi-
ques (moins charges en lments d'alliage).
Par ailleurs, d'autres pices sont en cours d'tude en trempe
l'huile directe avec des essais de substitution de matires moins
onreuses pour l'application de la trempe l'eau directe.
-14-
16 CD4 0 90 et rond forg (U = 8)
9151
Coule continue centrifuge ( C.C.C.)

[ Sy^rmc sabor)
WAi
I ^-- ToW,
! ~1-^

I
\ / / / / \ I
__ FIGURE 1
-15-
Coule continue hor7onl"ale (C .C .H.)
boîn -"
mitre rod*"
EM (brasiate elcfc-.-mo9r.que)
cr
A bora lingon*.- - extraction par oscillaTion
j J a S.l-c.u^ (rfre.tli**or par ca") enTrainanT sJnei circulaire*
av un pov Je 20 50
b a r r e s en 32 C4 0 46
.C' ' 'A t" '
FIGURE 2
-16-
Evolution de resilience en long Evolution de R e t E avec
avec le taux de c o r r o y a g e le taux de c o r r o y a g e
Etat T r e m p e Revenu p o u r 1000MPa MPat ( 850 R v 600)
500
Evolution de ./.
27. i
70
60
50
A7.' Evolution de A /.
oorre coule
16CD*t 90 m
30CDA 5 9 ma,
40CD^ . 7 5 mm
3 2 C 4 A 6 mo.
10 Os.
s
Figure 3
V^/
couronne
moyeu
Figure 4
18-
Schma de a r<zcrisl"alisal"on aprs Formage cKaud
I emps .
x
Ho 100 = 0% , 57% , 71% , 85%
Vfr-
Figure 5
19-
Acier XC 42 prchauffas 30mn 1150C
'2&^&{ l'i Vir"
fe-i "1
V
-tj&mj*
&*%:* f Ha
Drormation 0% Dformation 5 7 ^ 1000 "C Dformoton05% 100C TC

RAC RAC RAC
Grain . 0 . 2 Grain . 3 . 5 Gram 7 . 8 cjcjs 5_6
Figure 6
35 C Dr Abaque TRC
35 CDA- Abaque T T
A :& F : ferrite C : carbures M : merl'ensil
A:ausfnire F" : f e r r i t e C : carbures M . martens i fe.

e'c
\
\
\
Acj
600
Aci
100
KO
Ms
300
A *M
tW
0 1 10 50 0 2C 3 50 '0' IO'
0 1 10 50 IC O TC O 500 10' 10"
M I I I I I Il l l l l l l
1 2 "- 1h 2h 4h 8l
Figure 7
1050M100"
30 mio 900'
1 650'
RAC
Racuf ISO ocl-ucl
1050M100*
30mi 8 2 5
W 650'
Recuit" 2 paliers ISO direct aprs forg
H
%- 110/00 110/70/50/20
7.- 27 62
duref HB_ 17 7_187 192.202
r o < ( 5 0 % j 6r 5.7 r ex (35/) Gr 7.
structure. ( p L G r 4 6
L PL. Gr 6.
Do 2
' ^lurel-HB_ 18,0/177 18,1/192
Fiqure 8
-22-
Schma d'un four li Fluidisi
Couple
DISPERSION DU BILLAGE SUR L'ENSEMBLE DES LOTS
RECUIT ISO DIREC T SUR PIGNON DE 3 e m t
b.!! 41 41 42 42 43 43 44 44 45
L-t- 50CD4 Norr b r e 13 34 62 50 63 34 13 1
BV. 354 en 30C D4 ef 30 CD4 R
Lr 30CD4R NOT tr, 3 5 4 2 12 14 45 33 22 Deux Essais Industriels de 3000 pices (14 lofs)
? 1-', k FoureheHe. <J c. bill USINABILITE D.0,20
.5
s u r pignons couvra.nl la dispersion de duref
60
\\l\ 30 CD4:S= 0,030

Duref HB D.0,20fmmJ
5 0 - 179 19,03
187 18,81
192 19,15
200 18,78
0 .

1
I
I
207 16,51
', 223 17,08
34 -
32
J t
30CD4R.-S=0,073
2B
24
l\
I
I
/
\ A
I

*
i
'
Duref HB
187
D.0,20 (mm)
21,18
30CD4 / '\ * 202 19,11
20
30CD4R . 1 217 19,13

IG ' 1
/ ' \ S f r u c f u r e : Grain f e r r i f e (<*) 7.9
13- perlife (P.L) 3.7
-12
<' \
e
3
i '' \ RECUIT ISO DE SERIE SUR 30 CD4
4
D.0,20 :17,03 m m (183 HBJ S=0033
3
a ,|/ \ UI.
S f r u c f u r e : Grain Ferrife (j 8.9

11 4 * 43 43 s 44 U 5 45 ii' 4.' perlife (P.L) 7.8
B u aii K B ?17 110
Figure 10
4515 Abqquz TRC
A:ciusfcnitc F: f c r r i f e Cicarbures M :
O l 10 50 100 200 500 10 10' 10'

Imn 2 1m 11, 2h ^h 8 ^rh
Cycles du iTaifemenr a V 4 5 M 5 rzcuir direct et- normalis
1150'
LWW\~1050
30 min 8 5 0 '
1150"
-AM^1050"
Figure 11
25-
Structure du 45M5 recui" direcl" apres forge compare au 45 5 normalis
V6
/Recuit direct" aprs forge ( V o i r graphique j Normalisation 30min 850 C RA C

Ho-H,100,e,Z, 35 A 07. 35 A-O t
Mo
i lopins simulant' la deformahon en pied de bielle)
s t r u c t u r e ^ Perrite cellulaire et aciculaire ( l 5 2 0 ^ ) / ferrite grains trs fins et si
: siDiemenr equiaxes o r . i u .
xdlimitant des amas de perlite lamellaire ^ siblement" quiaxes Gr. 10.11
[ (~35%)perlite lamellaire fine
IGr 5.7 *t 3 . 5 (D
DO 2
' %uretH&- 10,96/262 11,23/235
Figure 12
TOPIS TRANSPORTEUR MMS TUNNFI CLQPIFU&E
REFROIDISSEMENT DIREC T 5 FORKE DE BIELLES EN 4 5 5
Cou piti
. TUNNEL C ALORIFUGE SAFE BIELLES 45 M
Figure 13
-27-
2 0 M B 5 POUR PORTE FUSEE R-. 1000 M
1
Temprature de Transition de la Resilience
en do J / c m
20*
15
10 l. + Hvanu Seria
T. H. Dirachz -* R e v e n u
T . E . DirecTc + R e v a n u
1 1 \ 1 1 1 1 1
90 7 60 50 A0 30 20 0 > iQ 20
Temperatur O esi en * C
- r- f - - ^'raclure aprs _^ - __ /. .
7- E c/ss/que ~ / t- c*'rect
D,<io/ ipres rei/env pour f- JoppM/ja
'HB "sUes ">>6s/3oZ
Figure 14
-28-
DISCUSSION
Question - Mr SchissLer, INPL Nancy : 1. Quelle est l'utilisation du

brassage lectromagntique ?
2. Quels sont les problmes rencontrs par l'utilisation du
bore sur la modulation de la vitesse de trempe la plus faible
et l'amlioration de la trempabilit.
Rponse 1. La plupart des aciries coule continue "classique"

(verticale ou courbe) laborant des aciers fins et spciaux
sont amenes utiliser le brassage lectromagntique (B.E.M.)
pour matriser au mieux la qualit de leurs fournitures.
Ce brassage prend en charge le lingot en cours de solidifica-
tion quand la partie centrale encore liquide reprsente 1/4
1/3 de la section totale. Il permet d'obtenir une cristal-
lisation plus fine, une meilleure homognit chimique et
attnue notablement les vestiges d'une "fine retassure
centrale". L'amlioration apporte peut tre importante pour
certaines nuances d'acier.
La technologie propre la coule continue centrifuge permet
dj une bonne matrise de la qualit du produit solidifi.
Il est probable cependant que des quipements de B.E.M. adapts
ce type de fournitures pourraient avoir des effets favorables.
En coule continue horizontale certains aciristes ont signal
L'emploi du B.E.M., toutefois la position horizortale de la
barre coule implique des conditions particulires de brassage.
2. L'emploi du "bore trempabi li t" s'est dvelopp par des

raisons d'conomie. Il a remplac le chrome dans certaines
ruances de nos fabrications de pices l'tat tremp et
revenu pour R = 800 1000 M P a ) .
Dans ces conditions des essais de trempe dans des milieux
faible drasticit (conduisant des structures htrognes
avec prsence de ferrite) n'ont pas entran de baisse sur
des caractristiques telles que La resilience ou La tenue
en fatigue.
Question - Mr Baffigi, Fiat : 1. La coule continue centrifuge et la

coule horizontale entranent des dfauts internes qui en
limitent l'utilisation. Etes-vous d'accord ?
2. Avez-vous eu des problmes dans L'emploi de L'acier au bore ?
Rponse 1. Dans les conditions d'utilisation de ces fournitures teLles

que celles dcrites dans notre expos, les pices choisies
permettent de garantir un taux de corroyage suffisant dans les
zones haute sollicitation. Dans ce cas la partie centrale
de la barre coule moins compacte, se retrouve moins compacte
sur la pice parce que situe dans une zone moins corroye
(elle est d'ailleurs limine par usinage).
Toutefois, si ces barres coules sont transformes en barres
lamines et pour des taux de corroyage analogues ceux
appliqus sur produits issus de coule continue classique,
ces fournitures ainsi lamines permettent tous les forgeages
habituels.
-29-
2. Aprs des essais sur prouvettes, dfavorables, L'emploi de
ce type d'acier en cmentation et en carbonitruration a t
exclu (grosseur de grain anarchique entranant des dformations
aprs traitement, existance d'une "anomalie de couche" par
formation de carbonitrure de bore dans le cas de la carboni
truration).
Cependant l'tat tremp et revenu,avec ou sans trempe aprs
chauffage par induction, l'emploi des aciers au bore s'est dve
lopp avec des nuances type 37 3, 20 HB 5, 38 MB 5.
Seules quelques mises au point en usinage ont t ncessaires.
Question Mr Felder, Armines : 1. Yatil un avantage considrer la

coule continue horizontale ? L'tat de surface semble tel
qu'il implique un scalexage du produit, ce qui pnalise forte
ment d'un point de vue nergtique ce procd.
2. Les structures bainitiques ontelles une meilleure usinabi
lit que la structure ferriteperlite ? Une tude mene avec
CreusotLoire sur du 20 NCD 2 nous a montr que les efforts
d'usinage de la bainite sont infrieurs ceux de la ferrite
perlite.
Rponse 1. La coule continue horizontale (C.C.H) permet l'obtention

de barres coules des diamtres jusqu' 30 mm, dimensions
parfaitement compatibles avec la fabrication par forgeage de
pices automobiles habituellement ralises en barres lamines
classiques. Comme nous l'avons signal, les aciristes C.C.H.
ont dj ralis de gros progrs en ce qui concerne l'tat de
surface de la barre. D 'autres amliorations en cours devraient
terme conduire un produit acceptable.
Il est certain qu'actuellement toute "rparation" de la surface
des barres par "scalexage" (meulage) ou crotage est pnalisante
2. D ans notre expos nous avons prcis que nos bielles en acier
45 M5 taient utilises aprs usinage sur une structure de
recuit de normalisation, c'estdire une structure ferrite
perlite. D e ce fait, la loi de refroidissement de ce traitement
devait viter l'apparition de constituant bainitique; leur
prsence accidentelle entranant une dgradation importante de
l'usinabilit du produit.
De nombreux travaux RNUR SAFE mens depuis plus de 10 ans
permettent de proposer globalement un classement d'usinabili t
pour certaines fournitures. Cette usinabilit dcrot dans
l'ordre des structures suivantes :
Ferrite + perlite >Bainite ^Ferrite + bainite ^
Ferrite + perlite + bainite.
Un mmoire C.N.A.M. de Mr Luciani prsent le 3 juillet 1975
MaizireslesMetz, intitul : "Etude de l'influence de la
structure accessible en microscopie optique sur l'usinabilit
des aciers de construction mcanique" apporte un clairage
intressant sur ce sujet.
Comment - Mr Dennis, British Aluminium : Electromagnetic casting is well
established in the aluminium industry. The use of this
technique has eliminated the necessity for scalping aluminium
alloys of a high copper content which previously required
12.5 mm depth of scalping on the major faces when the direct
chill semi continuous casting process is used.
This has given considerable energy gains in eliminating
necessity of remelting light scalping scrap and associated
melt losses due to oxidation.
Comment - Dr Ehringer, CEC : Inductive heeting of metals depends, among

other things, on its electrical and thermal conductivity. These
characteristics are quite different as regards iron and
aluminium.
Question- Mr Knobbout, Expert : Quel est le taux de l'conomie de

l'nergie ?
Rponse Le prsentation qui vient d'tre faite pour cette tude ne peut
donner, pour le moment, qu'un tat d'avancement des mises au
point techniques et technologiques pour assurer la qualit
mtallurgique et mcanique des pices. Il est prmatur de
prciser un chiffrage de l'conomie d'nergie pour ces fabri-
cations.
Je rappelle cependant que globalement, pour l'ensemble des
pices de forge concernes dans la production de la CEE, soit
2.400.000 T/an, une estimation d'conomie d'environ 20 % de
la consommation nergtique avait t avance, soit 230.000 TEP/an.
-31 -
EVELOPMENT OF BAINITIC NODULAR IRON FOR THE CONSTRUCTION OF SPEED GEARS
FOR THE CAR INDUSTRY
Authors M. BAFFIGI, S.CORSO, F.RABEZZANA
Contract number EE-B-1/149/I
Duration 1/9/1980 - 31/9/82
Head of project S.CORSO, Fiat Auto S.p.A.
Contractor Fiat Auto S.p.A.
Address C.so G. Agnelli, 200 - Torino - Italy
Summary
Considerations of economic (to reduce the cost of individual components)

and energetic nature (use of materials which require less energy both to
produce and to machine them) have led the automotive industry to search
for alternative materials to replace hardening and tempering and caseha-
rdening steels.
Spheroidal cast iron is capable of meeting these requirements and this
paper is intended to analyse the industrial feasibility and possible pro
blems connected with the use of this material for gear applications (be-
vel gears for a rear drive differential) in replacement for carbonitri-
ded 18 CrMo 4 steel.
In order to confer to spheroidal cast iron bevel gears mechanical and
dynamic characteristics similar to steel, it was necessary to treat the
machined parts with the ferritization, austempering and shot-peening pro
cesses.
The use of spheroidal cast iron resulted in a 75 % saving on tool cost
in the machining stage. However, considerable problems emerged concern_
ing deformations of the finished parts, with scraps amounting to about
50 % of totale production.
The survey has made it clear that if cast iron is to be used for gear ap
plications, further efforts are needed to align both the heat treatment
and machining cycles with the characteristics of the new materials
-32-
1. Introduction
The development of alternative materials for the manufacturing of automoti
ve components has followed two main trends in the past few years : reducing
the alloying elements in hardening and tempering and casehardening steels,
and replacing hardening and tempering steels with spheroidal cast iron.
One potential application, which seemed impossible up to a few years ago,is
the utilization of spheroidal cast iron for vehicle power line components
where high specific pressures and flexural load stresses are involved.
The development of improved manufacturing techniques, heat treatment methods
and casehardening processes has permitted to impart to spheroidal cast iron
such mechanical and wear resistance characteristics very similar to those
found in casehardening steels used for 'gear manufacturing.
The major advantage obtained by using spheroidal cast iron in place of case
hardening steels are: improved machinability, energy saving in the manufactu
ring of blanks and reduced part weight.
In view of these potential benefits, a development and testing program was
undertaken concentrating on crown wheel and pinion gears of a rear drive ve_
hiele differential, currently made from carbonitrided 18 CrMo 4 steel.
This program was designed to define and compare both cast iron heat treat
ment cycles and the types of casehardening cycles apt to confer to cast i
ron functional properties very similar to casehardening steel, as well as
the machinability of cast iron compared to gear steel.
2. Heat Treatment Cycles and Mechanical Characteristics

In order to confer to spheroidal cast iron the functional characteristics
suitable for the specific application of gear manufacturing it is necessary
to treat it with a ferritization process designed to reduce the pearlitic
content to < 5 % and to guarantee constant deformation after the bainitic au
Stempering treatment.
The cast iron used was of the Gh 654805 type with the following percent
chemical composition: C = 3.87 %; Si = 2.65 %; Mn = 0.4 %; Ni = 4 1.5 %; Mg=
0.3 %; S = 0.012 %; P= 0.02 %.
The mechanical characteristics of the cast blank meet the UNI 4544 standard:
surface hardness HB = 240280 kg/mm2; spheroid density = 390 nVmrn 2 ; sphe
roid diameter = 1530 jjm.
Two separate ferritization cycles were tested: a high temperature cycle A
(Fig.1) and a low temperature cycle (Fig.2). The following parameters we
re applied :
CICLE A CYCLE
Heating to 900C for 5 hrs Heating to 740C for 7 hrs
Cooling at 780C for 1 hr Cooling in still air
Cooling at 630C for 1 hr
Cooling at 427C for 1 hr
Cooling in still air
Expansions after the ferritization treatment, as measured by dilatometrie

analysis, were as follows:
0.54 % after the high temperature cycle (A )
0.35 % after the low temperature cycle (B)
33
The CCT curves (Fig. 3) were determined for spheroidal cast iron in order
to define the cooling rate and austenite/bainite transformation temperatu
re to be applied in the austempering process.
The following parameters were used in the austempering cycle (Fig. 4 ) :
Heating to 910C for 2 hrs

Quenching in oil at 235C
Maintaining at 235C for 21/2 hrs
Cooling in air
Expansion measurements taken after heat treatment revealed that the sum of
final expansions was practically the same whether the high temperature or
low temperature ferritization cycle was applied (0.78 % and 0.79 % respec
tively) .
For the ferritization treatment it was decided to adopt the high temperatu
re cycle, as the larger deformations which take place during the ferritiza
tion stage can be eliminated during the machining of parts, and the subse
quent hardening stage causes only deformations due to the austenite/bainite
structural transformation.
The steel used for the tests was the 18 CrMo 4 type and was carbonitrided
by applying a cycle similar to that used for standard production gears
(Fig. 5 ) :
Heating to 880C for 2 hrs

Diffusion at 880C for 1 hr 20 min.
Stabilization at 830C for 1 hr
Oil quenching at 120C
Tempering at 180C for 2 hrs
The carbonitriding depth was 0.5 to 0.7 mm. The mechanical properties of
spheroidal cast iron after the high temperature ferritization and austemp
ring treatment are compared with those of carbonitrided steel in Table I.
The comparison between the two materials shows that the core strength and
surface hardness of cast iron are similar to those of steel and are within
the limits (even though near minimum values) set by manufacturing specifi
cations.
A more detailed characterization of the materials involved was then made by
means of toughness and fatigue tests.
3. Toughness and Fatigue Limits

The dynamic properties of spheroidal cast iron were compared with those of
carbonitrided 18 CrMo 4 steel in the following tests :
. 3point bending fatigue test : The specimen tested (Fig. 6) simulated a

gear tooth. The data were processed using the Stair Case statistical method.
For this test, a 10 ton MTS electrohydraulic machine was used. The values
recorded at the completion of the tests were 299+244.5 N/mm2 for carbonitri
ded 18 CrMo 4 steel, and 155+127.4 N/mm2 for bainitic spheroid cast iron.
The comparison of the 3point bending fatigue limits for the two materials
clearly indicates that bainitic spheroidal cast iron cannot replace steel
as a gear material without undergoing subsequent hardening treatment, as
34
its low fatigue limit value would cause an early failure of the gear tooth.
. Measured resilience: This test was made using an AMSLER pendulum with
maximum available energy of 235 J; the recording of data monitored by
strain gauges mounted on the pendulum IZOD bob was made using an oscillo
scope equiped with a memory for a photographic recording of the time/load
diagram. The specimen used is shown in Fig. 7; the specimen shape can be
considered as a large module gear tooth as its dimensions (thickness and
tab to specimen body junction radius) are comparable with those found in
gears.
The test results, summarized in Table II, show the low toughness of cast'
iron in comparison with steel.
The maximum oimpact stress values for cast iron are 130% lower than
steel values. This means that between two gears of the same dimensions ma
de from the two materials tested, the one made from bainitic spheroidal
cast iron will fail under much lower stresses than those normally withs
tood by carbonitrided steel.
Moreover, a 18.7 joule energy is sufficient to break a cast iron specimen,
against the 86.5 joules required to break a steel specimen (a 360% increa
se).
. Rate of growth of a fatigue crack: To complete the dynamic characteriza

tion of bainitic cast iron, the rate of growth of a fatigue crack was tes
ted according to ASTME399 standard (specimen as shown in Fig. 8) using an
electrohydraulic MTS machine.
The law of growth of a fatigue crack was formulated according to the Paris
expression da /dN = Co , and by plotting the da/dN and values for
steel and cast iron on a double logarithmic diagram the straight lines
shown in Fig. 9 were obtained.
It is apparent that the propagation rate of a crack is greater in cast
iron than is steel, the coefficient = 4.93 value for cast iron being
about double the value for steel, which is = 2.33.
4. Surface Hardening of Bainitic Spheroidal Cast Iron

The low toughness and fatigue strength values obtained from bainitic sphe
roidal cast iron test pieces in comparison with carbonitrided 18 CrMo 4
steel raise critical problems concerning the use of cast iron for automo
tive gear applications.
The strength properties of this material must be improved by means of sur_
face treatments such as shotpeening and rolling to obtain the desired
strength. These treatments produce structural changes in the surface la
yers and permit to achieve dynamic characteristics similar to those found
in carbonitrided steel. The testing was performed using specimens of 18Cr
Mo4 carbonitrided steel and bainitic spheroidal cast iron; the latter ma
terial was surface treated with the shotpeening and rolling methods.
The operatinf conditions of the two surface treatments were as follows:
SHOTPEENING : pellet 0 110 shot, 170 shot

pellet material casehardened steel
intensity 6 Almen A; 16 Almen A
35
2 2
air pressure 55 N/mm ; 60 N/mm
ROLLING pressure 5000 N, 9000
No. of turns 3
The steel and cast iron specimens were tested for 3point bending fatigue
strength (to determine the D value), while the quantification of the
amount od compressive stresses due to the surface hardening treatments
was made by measuring the residual stress using a SIEMENS type SMD 2000
machine.
The results obtained are summarized in Table III.
By relating the dynamic component da of the fatigue limit as a function
of shotpeening intensity and rolling pressure, the diagram shown in Fig.
10 can be obtained. This indicates that :
shotpeening has a greater effect on improving the fatigue strength of
cast iron than rolling does;
the a values for the 16 Almen A shotpeened specimens are or 6 % lower
than those for carbonitrided steel;
the 6a values for the 9000 pressure rolled specimens are 2r30 % lo
wer than those for carbonitrided steel
The different fatigue performance of shotpeened and rolled specimens is

confirmed by the fractographic study of the broken pieces. Fig. 11 shows
the results of a scanning microscope analysis which indicate that in the
shotpeened specimen the crack originates at a deeper level than in the
rolled specimen. Moreover, if the measured residual stress values are pio
tted against the <5 a values (Fig. 12), the importance of the compression
treatment becomes even more apparent; in particular, the following points
can be noted :
cast iron which is only austempered (tensile stress) yields the minimum
d a value;
the câ values increase with the increasing of compression stress values;
the maximumiSa value for cast iron specimens ( ~ 6% lower than the value
for carbonitrided steel) is obtained with a maximum compression stress va
lue of 248 N/mm2. This value approaches that of carbonitrided steel(284
N/mm 2 ) ;
practically equal a values (cast iron rolled at F = 9000 and cast

iron shotpeened at 6 Almen A) correspond to residual stress values very
similar to each other if the measuring scatter is taken into account.
Having established the superior performance of 16 Almen A shotpeening

over other surface treatment methods, a nulber of specimens thus treated
were selected to determine the toughness parameters typical of instrumen
ted resilience (Table IV).
The results obtained indicate that treated cast iron yields higher values
than nontreated cast iron ( ^. 65%) but in any case lower ( 4 22 %) than
the values obtained from carbonitrided steel.
36
5. Machinability Tests
The last phase of the reseacrh program consisted in a set of comparative
machinability tests both at the laboratory and production level.
The following materials were used for these tests :
ferritic spheroidal cast iron (perlite < 5%) . average HB=145 (143148)
ferritic/pearlitic spheroidal cast iron . average HB=201 (178255)
austempered annealed 18 CrMo 4 steel . average HB=152 (149156)
A metallurgical characterization was made which indicated that spheroidal

cast iron components showed a good distribution of graphite spheroids both
in the ferritized and pearlitic material, and the structures met the stan
dards.
5.1. Comparison of laboratory machinability tests

A set of laboratory tests was designed to simulate different types of ma
chining operations (with their related problems) which would permit to
establish a comparison among the various test materials in a short time
and at low costs, and would supply preliminary indications on the correct
use of the new material in production.
In particular, in the test case of the bevel gears for a rear drive vehi
cle differential, the machining cycle was analyzed to identify the two ma
chining operations most suitable for a machinability evaluation, that is:
bevel cutting of crown wheel and pinion gears using a GLEASON machine
(milling operation)
drilling od screw holes.
Milling Operation
The evaluation of cast iron machinability was made by means of milling te
sts using modular cutters made from 5 % Co high speed steel (diameter 0 =
70 mm, No. of teeth Z=12).
The test procedure consisted in cutting 3 mm deep and 250 mm long grooves
with one return stroke with the cutter sliding on the machined surface.
This sequence was repeated a number of time corresponding to 100 minutes
machining. The wear of each cutter tooth was measured at regular intervals.
Fig. 13 shows a schematic representation of the machining cycle and the
geometry of the test cutters.
Table V lists the cutting conditions selected for the test. The analysis
of the machinability test results indicates that in the milling operation
ferritic cast iron is 40 % better than pearlitic cast iron in terms of
tool wear (Fig. 14). The tools used to machine pearlitic cast iron show a
greater scatter of the wear values measured; by reference to the percenta
ge ratio between deviation and average, the value is 12 % for ferritic
cast iron and 35 % for perlitic cast iron. When the comparison is made be
tween the two cast iron varieties and steel (at 35 m/min speed), it can
be seen that pearlitic cast iron has an advantage over steel of about 100%
while ferritic cast iron reaches the 200 % level (Fig. 15).
37
Drilling operation
To evaluate the performance in drilling operations of cast iron materials
in comparison with 18 CrMo 4 steel, tests were performed using 8.5 mm dia.
twist bits with cutting speeds varying between 30 and 50 m/min and with
0.34 to 0.15 mm/turn feed, using a 5 % emulsion coolant. The comparison
was made couting the number of holes drilled until the avarage wear of
cutters was V = 0.3 mm.
The drilling test showed] the advantages of ferritic cast iron over both
pearlitic cast iron and steel. ..
The advantage of ferritic cast iron over the pearlitic variety was in the
order of 400 % for the same number of holes up to a tool wear level of
V B = 0 . 3 mm (Fig. 16). Under the same machining conditions, the bits used
to machine the steel material chipped after 40 holes; thus ferritic cast
iron would have a 1500 % advantage over steel. However, these cutting cojn
ditions are not acceptable for steel material. Other tests were performed
with reduced feed rates, and under these new conditions the difference in
productivity of perlitic cast iron over steel was 130 % (Fig. 17).
For drilling steel material the reduced feed rate of s = 0.15 mm/turn was
considered satisfactory (Fig. 17).
5.2. Comparison of machinability at the industrial level

The comparison of production machinability between ferritic spheroidal
cast iron (greater machinability advantages) and austempered annealed 18
CrMo 4 steel (the currently used material) was made by machining 1300 be-
vel gear pairs made from spheroidal cast iron.
An analysis of the machining cycle for the crown wheel and rear axle redjj
ction pinion led to the identification of the two most important opera-
tions in terms of tool costs and machining time:
- crown wheel : . roughing and semifinishing boring

facing of housing support planes
drilling, boring and tapping of screw holes
rough gear cutting
finish gear cutting
- bevel pinion facing of the two ends

turning of stem and conical head
rough gear cutting
finish gear cutting
The results obtained from machining two homogeneous lots of the two test
materials confirmed the laboratory test results (see Table VI); in parti-
cular:
Turning, boring and facing operations:

. In these machining operations performed with carbide insert tools, gene-
rally coated, and exasperated cutting conditions (Vt=100-350 m/min, a =
0.2-0.6 mm/turn) tool productivity improvements of 100-200 % were obtained
with tool wear reduced by 50-70 %;
. morever, when machining ferritic cast iron nocritical problems were ex-
perienced with chip formation and clogging, especially in the turning op
ration using casehardening steel.
Drilling and tapping operationd:

. In the drilling operation, performed with twist bits made from high
speed steel (0 8.3) and cutting conditions not optimized for cast iron ma
chining (Vt=15 m/min, a =0.13 mm/turn), all the available cast iron sam-
ples (1300) were machined with a negligible amount of wear in comparison
with the limit productivity of 500 steel pieces.
Thus a minimum 300 % improvement was obtained if the insufficient lot si-
ze and not-optimized cutting conditions for cast iron are taken into con-
sideration;
. In the tapping operation, performed with high speed taps (0 8.8) with a
geometry and cutting conditions suitable for steel machining (Vt=17 m/min,
a=0.14 mm/turn), the productivity in cast iron machining went from 500
pieces to 300 pieces, with a 50 % to 70 % less wear;
. In this operation, too, chip formation was better when machining cast
iron; however, a greater contamination of coolant with materiale partic-
les was experienced.
Gear cutting operation:

. The gear cutting of crown wheel and pinion was performed using Gleason
0 6" cutters. Both these parts require a roughing and a finishing opera-
tion with widely different cutting conditions (pinion roughing: Vt= 40 m/
min; crown wheel roughing: V-(-=32 m/min; pinion finishing: V-(-=46 m/min;
crown wheel finishing: V t =6.6 m/min. Also in this case the results obtai-
ned indicate the greatly improved performance of cast iron, with producti
vity increases up to 7 times higher and wear reduced by one half.
These results were quantified in terms of savings on tool and sharpening

costs amounting to a 75 % reduction or Lit. 270 per gear pair.
6. Checking Deformations
One lot of 200 bevel gear pairs (Fig. 18) was treated with the ferritiza-
tion process described above and then machined with the same parameters
used for machining steel material.
Dimensional measurements were taken on this lot to check deformations af-
ter hardening.
The gear pairs were then austempered and measured again for a further di-
mensional checking.
Crown wheels had varying deformations up to 0.5 mm the housinf seat dia-
meter. However, even though these deformation are considerably large,
they are sufficiently constant and it is felt that the desired design di-
mensions can be achieved by calibrating the blank machining cycle.
The pinions had the same dimensional anomalies noted for the crown wheels
with the same chance of correction in the machining stage, but they are
more critical with regard to the straightening of the stem. Fifty percent
of the parts produced cannot be recovered through the straightening ope-
ration because of elastic resilience; the ramaining 50 % require a time
which is double the time currently needed, and even then the required to_
lerances cannot be achieved.
To overcome this problem it appears that austempering with a locking fi-
xture is necessary to prevent the occurrence of large deformations and
to reduce to a minimum the straightening operation, which proved to be ye
ry critical in the industrial process.
7. Conclusions
The industrial feasibility study concerning cast iron bevel gear pairs
for a rear drive vehicle differential (currently made from 18 CrMo 4 4
steel) led to the following conclusions:
- Spheroidal cast iron must undergo a ferritization treatment to 9COC

for 5 hours followed by austempering in order to confer to this material
metallurgical properties, such as surface hardness and core strength, com
parable to those obtained from 18 CrMo 4 steel a,fter a carbonitriding
treatment;
- Compared to steel, the dynamic characteristics of bainitic spheroidal

cast iron have values are lower with regard to both fatigue strength (48%
less) and toughness ( 6 impact lower by 130 %) ;
- In order to upgrade the dynamic characteristics of spheroidal cast iron

the material must undergo surface hardening treatments. The shot-peening
treatment applied with 16 Almen A intensity is capable of improving the
dynamic characteristics of cast iron to values approaching those of steel
(22 % less toughness and 6 % less fatigue strength);
- In the laboratory machinability tests, ferritic spheroidal cast iron

yielded tool wear rates 100 % (or more) lower than ferritic-pearlitic
cast iron;
- The comparison with 18 CrMo 4 casehardening steel at the annealed state

showed both at the laboratory and industrial level that ferritic cast
iron permits to increase cutting conditions and productivity up to a 75%
saving in tool costs;
- Dimensional measurements taken on a 200-piece lot of bevel gear pairs,

machined with the parameters used for machining steel parts, showed the o
currence of deformations, which in the case of the pinion unit remained at
critical levels even after the straightening operation (50 % of production
rejected because off tolerance).
Considering these results, it can be stated that the characteristics of

spheroidal cast iron, austempered and shot-peened, are inferior to those
of casehardening steel of the type currently used for gear manufacturing.
The machinability tests performed on a large lot of bevel gear pairs made
from spheroidal cast iron have permitted to quantify the savings obtaina-
ble in terms of tool costs., but at the same time they indicated the occu
rence of critical deformations on the finished parts, which can be only
-40-
partially eliminated with the straightening operation.
For these reason, the use of spheroidal cast iron in the field of gear
manufacturing is possible provided that the necessary changes are made
in the heat treatment and machining cycles.
The work done, however, served to highlighta wide range of possible uses
open to spheroidal cast iron which, with adequate heat treatment and sur-
face hardening cycles, can be successfully employed for components requi
ring a high fatigue strength. These possibilities must be examined at the
design engineering level so that the design a new product can be made ta-
king into account the dynamic-mechanical properties of spheroidal cast
iron.
-41 -
Temperature (C) Temperature (C)
Fig. 1 - Ferritisation cycle A Fig. 2 - Ferritisation cycle
Temperature (C)
1000-
800
600
400
200
Time (h)
Fig. 3 - CCT diagram Fig. 4 - Austempering cycle
2 li"
0
(
20
Temperature (C)
R r 25
6 9
Time (h)
Fig.6 3point bending fatigues
Fig.5 Carbonit red cycle testspecimen
5"
I 1 1
7 1 I 1
" 1 I
Lo 1
1 1
1
I
l 1 CTO
75
(
1
_
J
^ r '. f -
Ci
H
In
O
0
o <0

1 R M 25
in 6C
3
<
MS
Fig.7 Measured resilience Fig.8 Rate of growth of a fatigue

specimen crackspecimen
shotpeened
rolled
18CrMo4 steel
Dinamic component
Cast'iron ta (N/mm2)
da/dN (mm/cycle)
cast iron
2 ^ s I.02IO Ak '
12 16
Shotpeening intensity(A lmen)
I I " 1 1 1
1er 4 8 12 16
4 ( mm'3,
Rolling pressure ()
Fig.9 Crack propagation in rela Fig.10 Dynamic component versus

tion toload intensity surface hardening
Surface
Fig. 11 Scanning microscope analysis
44
Dinamic component
CD a (N/nnm2) 18 O M o 4 steel
* Cast iron shotpeening 16 Almen A
260.'
Cast iron shotpeening 6 Almen A
D Cast iron rolling F=9.000
Cast iron rolling F=5.000

220
O Cast iron
180
D
140
100
320 240 160 80 +80 +160
Residual stress RfN/irn^)
Fig. 12 Dinamic component in relation to residual stress
chips section
o. 25"
Fig. 13 Schematic representation of the cutters test (mm)
45
ferritic matrix
ferritic/perlitic matrix
Cut time (min)

Wear (mm) 150
i .
0.5-
0.4- . 10O
0.3-
^
0.2. 50-
0.1. . -~~ ,
/ .
(
O r-fc CU
18CrMo4 ferritic ferritic
40 3 120
steel perlitic matrix
Time (min)
matrix
Fig.14 - Milling operation Fig.15 - Milling operation
18CrMo 4 steel
* ferritic matrix
ferritic/perlitic matrix
Wear (nm) Feed rate (mm/giro)

k
0.4.
0.30
0.3
0.2.
0.15
0.1.

0 400 800 1200 18CrMo4 ferritic
n holes steel perlitic
matrix
Fig.16 - Drilling operation Fig. 17- Drilling operation
-46-
..liga -w
iipijliii|iiii
Fig. 18 - Bevel gears

Tab.I Mechanical properties of steel and cast iron
Bainitic 18 CrMo 4 Manufacturing

Materials pheroidal carbonitred
specifications
cast iron steel
Core strenght
1365 1400 1100 + 1500
(N/mm2)
Surface hardness
57 60 58
(HRc)
Bainite.Martensite, Sup.=Martensite
residual austenite and residual auste
Structure nite
and graphite
nodules Heart=Bainite and
martensife
Tab. II Toughness properties of steel and cast iron
Tensil Striking Maximum Breack energy

Materials stress time impact stress
(N) 2 (J)
(m sec.) (N/mm )
18CrMo4 carboni
42.000 0.65 2325 86.5
tred steel
Bainitic spheroi 18.500 0.32 1000 18.7
dal cast iron
Tab. Ill Dinamical properties of steel and cast iron
Surface Fatigue Residual

Materials Treatment limitCO * stress
Hardening
(N/mm2) (5? (N/mm2)
155.8+127.4 + 143+10
shotpeening
213 + 173 241+49
Spheroidal 6 A lmen A
Austempering shotpeening
cast iron 282.7+231.3 248+10
16 Almen A
Rolling 183 +150.2 78.4+20
F=5.000
Rolling 211 +172.4 215.6+29
F= 9.000
18CrMo4 Steel Carbonitred 298 + 244.5 = 240 + 20

0 = (medium stress) + C>a (dinamic component)

D m '
48
Tab. IV- Toughness properties of steel and cast iron
Tensil Striking Max imum Breack energy

Materials time impact stress
stress 2 (J)
(N) (m/sec) (N/mm )
Cast iron 18.500 0.32 1.000 18.7

Shot-peened 30.000 0.32 1 .800 24.7
cast i ron
18CrMo4 Steel 42.000 0.65 2.325 86.5
Tab.V - Cutting condition at the laboratory level
Ferritic- Cast iron

Operative conditions Ferritic-Per
lite steel
Cutting speed (m/min) 55 35

RPM (rev./min) 250 159
Feed (rrm/min) 630 305

Feed/tooth (mm/tooth) 0.21 -
Depth of cut (mm) 3 3
Cutting fluid oil oil
Depth of chip equivalent (mm) 0.019 0.019
Tab.VI - Comparison of machinability at the industrial level
Operation Cast iron 18CrMo4 steel

Piece
D Wear
roductivity (mm)
3
roductvity Wear
firm)
Roughing boring 250+300 3.5+0.7 80+100 1 .5
Facing 250 0.2 100 0.4
Crown Dr illing,boring-tapping 1300 - 500 -

wheel
Rough gear cutting 1300 D.4+0.5 180 0.9+1
Finish gear cutting 1300 0.2 200 0.5
Facing 3200 - 1600 -

Bevel Turning 400 0.4 200 1
pinion
Rough gear cutting 1300 3.4+0.5 200 0,9+1
Finish gear cutting 1300 0.2 300 0.4
-49-
DISCUSSION
Question Dr Ehringer, CEC : Estce que vous avez pu varier La vitesse

d'usinage et tudier Les conomies d'nergie ou de temps qui
peuvent en rsuLter ?
Rponse Le tabLeau V donne Les conditions des tests effectus et mentionne

notamment Les variations de vitesse de coupe obtenues.
L'appareiLLage utiLis pour Les tests tant prvu L'origine pour
L'acier n'a pas permis d'utiLiser des vitesses de coupe suprieures
Question Mr Bak, C R . PontA Mousson : QueLs sont :

1. Le cycLe du traitement thermique raLis pour faire de La
bainite ?
2. La composition chimique de La fonte mise en oeuvre pour
obtenir La bainite ?
3. La dimension des pices traites ?
Rponse 1. Chauffage 910 C pendant 2 heures.

Trempe L'huiLe 235 C et maintien 235 C pendant 2 h 1/2,
suivi d'un refroidissement dans L'eau caLme.
2. Gh 654805
C = 3.87 % Si = 2.65 % Mn = 0.4 % Ni ^ 1.5 %
Mg = 0.3 %; S = 0.012 %; = 0.02 %
3. Les pices traites ont approximativement Les dimensions

suivantes :
Couronne 0 internes 100 mm 0 externe 120 mm
Pignon Longueur ci 300 mm
Nombre de dents 11/41 Pignon/couronne
Question Mr FeLder, Armines : Le fait que La pice se casse Lors de son

dressage finaL n'estiL pas de mauvais augure pour sa tenue en
service ? Dans L'hypothse o Le traitement finaL rendrait
inuti Le Le dressage, estce que La pice en acier subit Le mme
dressage ?
Rponse Le dressage finaL est indispensabLe pour assurer une mise La

cote de L'engrenage rpondant aux tolrances finales imposes
par les conditions de fonctionnement.
La resilience du matriau, quoique de 22 % infrieure l'acier,
peut tre admise car l'engrenage est protg par un botier des
chocs ventuels et par un renfort de La pice.
50
LUBRICATION BY SLAGS OF THE CONTINUOUS
CASTING OF STEELS.
Authors E. FELDER, Ch. NIGGEL
Contrat bumber EE-B-117 F
Duration 24 months : 1 October 1980 - 30 September 1983
Head of project E. FELDER, CEMEF (Ecole des Mines de Paris)
Contractor ARMINES
Address 60, Boulevard Saint-Michel

F 75272 PARIS CEDEX 06
Summary
We have developped a theoretical analysis of the lubrication by slags

of the continuous casting of slabs by coupling a thermal analysis of the
system with a model of the viscous flow of the slag between shell and
mould. This model provides a good estimation of the thermal flux
extracted by the mould for slags of low viscosity. For slags of higher
viscosity, comparison of the analysis with experiments suggests that
the fusion process of the slag above the meniscus controls its flow
rate. Therefore we analyse the fusion experiments performed in
laboratory and build with these results a model of the fusion process
of slag in continuous casting which we shall couple with the first
model.
1. INTRODUCTION
This work is devoted to the study of the phenomena induced by the slag
lubrication in the continuous casting of steels and aims to predict the
influence of the slag properties on :
the heat flux extracted by the mould
the friction between the mould and the steel shell
the defects of the casted product
This paper describes the work performed at the present day and is
divided into two parts :
A theoretical analysis of the viscous flow of the slag between
mould and shell and the comparison of the results obtained under the
assumption of maximal flow rate with the measurements of thermalflux
performed by IRSID /l/.
A theoretical interpretation of the fusion experiments made by
IRSID /3/ in order to predict the flow rate of high viscosity slags in
continuous castings.
2. THERMOMECANICAL ANALYSIS OF THE SLAG FLOW BETWEEN SHELL AND MOULD

2
.1 Theoretical model / 2 /
The nomenclature is given in figures 1 2 and tables I, II.
The figure 1 describes the system. We analyse the heat flow and the
slag flow under the meniscus, near the middle of the broad face.
. Therefore we suppose that the shell is applied against the slag
with the hydrostatic pressure :
= P A g X (1)
where g is the gravity acceleration
. The temperature gradient across the slag thickness is assumed
constant, of value (figure 2 ) .
= K(T 3 T A)/ h (2)
the viscosity of the liquid slag is taken as :
( = n0/g(T/Tj,) (3)
I g(s)= SY (si)
So the flow rate (figure 2) is :
h
h_ rdj>
q = U dy
Idx + ] (s 3 ) + h V G 2 (S 3 ) (4)
with s3 = T 3 / T L
;.(s)
1
^-\\

(s-i)J
(vi)2g(v)dv
[/ g ( v ) ( v l )
J' g(v) dv
dv

^ ( v l ) g ( v) d v
G 2 (s) = 1
(s1) f S g ( v ) dv
J
l
. The solid slag is supposed rigid. Its thickness is dedu ced from the
thermal gradient and the water temperature :
h = K(T )/ . / (5)
s e c
52
. We analyse the steel solidification with the equation of the global
conservation of the energy for a shell's slab :
h +e
P dx H (6)
A[V C A (VT) *y] ; c
ht
is the convective heat transfer and the overheating ;
. In a first model (figure 2 ) , we suppose that the thermal gradient is

constant across the shell. There fore
yeh
= + / (1 <_ Y =
0 ) (7.1)
e
. A second model takes into account for the curvature of the temperature
distribution under the form :
A A
The system of equations (1) (7) is resolved with a computer program

/2;A/.
2.2 Theoretical results and comparison with experiments
A theoretical analysis of the equations Tl) and TJ provides that the
slag flow rate is limited to a finite value q which corresponds to the
initial value of film thickness
h = h(x=0) which verifies the equation
|| ( x _0) ^ ^ (8)
We have calculated the distribution of h, h , and e under the

s
assumption (8). The values of the various physical parameters are gathered
in the tables I and II.
For a low viscosity slag (A ), we observe that the second thermal
model gives better results than the first (figure 3 ) . The other theoretical
results are therefore deduced with the 2" model.
The theoretical influence of the convected heat in steel liquid and
the overheating is small, in agreement with the experimental observations/2/.
For the same1slag, the theoretical influence of the casting velocity V
on the profiles of the thermal flux and shell thickness e is in good
agreement with the experimental ones (figure 4 ) .
We see on figures 5 that the agreement is poorer as the slag viscosity
increases ; the theory under estimates greatly the thermal flux for slags of
high viscosity (E,D).
As the theory predicts a very little influence of casting velocity on ,
in contradiction with experimental results, the discrepancy increases with
speed.
In the table III, we compare the theoretical slag consumption (slag
mass (Kg) used per ton of casted steel) under maximal flow rate condition
and the experimental one : the latter is not very well known ; moreover a
significant part of the powder disappears as gaz during the fusion and is
trapped in the oscillation marks of the casted slag ; so we suppose
that the consumption due to hydrodynamic flow is about half the total
consumption. The table III suggests that the slag of low viscosity flows at
the maximate rate, but it is not the case for the high viscosity slag D ;
there fore h is overestimated and underestimated.
3. ANALYSIS OF THE FUSION PROCESS OF THE SLAG

3.1 The fusion experiments
. In order to explain the experimental observations (table III) , we
must assume that /4/
- The no-heated wall has a thermal conductance (figure 6 ) .
- For high viscosity slags as D, the contact between slag and the
heated wall A is not perfect and can be described with a thermal
conductance
. A simplified analysis of the thermal flow provides the liquid slag
thickness cT +L
1
h- i-i- 2 -! (9)
= ( - L/C)/( + ) = maximal slag temperature
where\ r / 8 a_ cT -
W +
, /1, + ^ ^ (,cT o .+TL ) MI do)
2a W R pc o
_ 2
for W = Q/(TT R p) and a = K/(pc)
. A mechanical analysis of the system provides that the flowrate is

controlled by the deformation under the gravity force of the solid slag
which we assume pseudoplastic, with a pseudoplasticity index m and a
consistency G(T). The mean slag velocity W verifies :
T
L
G_ = J G(v) dv = | pg H(3 R, 1 + m [L W (3+ )] (11)
b -L ~^- i m
S
_ The values of G deduced from the experimental observations,
W and e, with these formula are intable IV.
3.2 Comparison_with_the continuous casting conditions.
We can deduce from the effective consumption C t a mean value of the slag
velocity above the meniscus (table IV) : it is much greater than in the
fusion experiments ; due to the high value of slab's width B, we can
neglect the thermal losses in the mould (a 0) and estimate the liquid
slag thickness h above the meniscus with the formula (9), (10) : it is of
the same order of magnitude than in fusion experiments if we suppose that
the contact between the slag and the liquid steel is thermally equivalent
to the contact slag graphite in the experiments.
At present day, we try to analyse the mechanical behaviour of the slag
above the meniscus in order to estimate the fusion flow rate.
4. CONCLUSIONS AND FUTURE WORK

The results of the hydrodynamic model is in good agreement with
experiments for low viscosity slags. For high viscosity slags, we shall
54
develop the fusion model and analyse the deformation of the solid part of
the slag in the lower part of the mould (figure 2 - 3 d model) wich induces
high friction. Beside this, we try to analyse the phenomena induced by the
mould oscillations.
In parallel, we shall participate in experimental investigations
performed by IRSID in the SOLMER plant and which could provide further
measurements of slags performances (heat flux, slag consumption, friction,
product defects).
REFERENCES
IM H. LARREQ, M. WANIN, M. MANGIN, M. NICOLAS

Rsultats d'exploitation de la lingotire du coule continue de
la SOLLAC - Journe Sidrurgiques ATS 1981
IH Ch. NIGGEL, P. GILORMINI

Progress Report nl (Juin 1982)
/3/ P. RIBOUD, H. LARRECQ, ROUX

Private communication (September 1982)
Ih I E.' FELDER, Ch. NIGGEL

Progress Report n2 (in preparation)
SLAG /I/ A D E
: 1300C 0.1 0.14 0.401 0.532

(P1)1500C 0.036 0.05 0.128 0.158
T L () 1180 1210 1350 1390
0 (PD 0.1222 0.1177 0.102 .1018
4.66 4.16 3.32 3.42
Table I: Viscosity characteristics of the studied slags
Material density heat Fusion

Thermal
Kg/m3 capacity conductivity
Point {K) Enthalpy
J /Kg W m-1 K-
MJ/Kg
Steel pA=7000 C=670 K. L =0.247

A A = 32 Tp=1809
A
Slag =2700 C =1100 T L =0.7

= 1.5 L (I)
Copper - - c = 390 1 -
I -
Table II: Thermomechanical characteristics used in calculations

Slag consumption C (Kg/t) in Fusion
SLAG continuous casting experiments
hydrodynamic theory experiments Q h

q q (g/mn) (mm)
max
V == 1 1.5 m/mn total effective
(50%)
regular
A 0.22 0.27 0.92 10
0.5 0.25
irregular
D 0.38 0.52 0.61 14
Table III: Influence of the nature of the slag on the slag

consumption C in continuous casting and the fusion
experiments (figure 6)
FUSION EXPERIMENTS CONTINUOUS CASTING FOR

SLAG
c e = 0 .25 Kg/1
experimental thermal conductances c o n s i s t e n c y

values: W m" 2 C~' (m = 0 . 2 )
WQln/s) h(mm)
(B=0.2 m)
W h a
A B
Gs(Pa
0 2
SU,)
V o)
(ym/s) (mm)
A 2.9 10 0 28 1,850 13 ,2.7(V)
D 1.9 14 240 28 2,050 13 .4( >

|.3.1(a A 240)
Table IV: Interpretation of the fusion experiments and the

consumption in continuous casting
57
lag
^ Solid licjuid
-hN
Fusion.
*"i*fiJtrahen
x^^N IA
Figure 1 : Geometry of the system
^
Temperature
1
Te V. Kt y
0 cv
velocita
J
V
njould otJ liquici shell liquid
waltr cobber stee.
1sthicltl 2dnodel 3d*odtl
Figure 2: Schematization of a t r a n s v e r s a l s e c t i o n (X > 0)
58

2.
\
/
Hvft
2 \ \ 2 d mod J
1 0.3 0.5 m. o.? 0 . 3
. 6
s l model ^ V / " " ^ 2
\
*y*y'
. 2 ~ _ ^ \ A.S ^yy
y y ^
V = i
cm - ^
0i 0.3 5 m. 7 X 0.9 yy
i . /y y / Yj.5-
yy yy
SU3 ^ ~^= 5 m/ma / vin />>-
o.S // < , - -~"

///y
Figure 3: Improvement of the thermal model of the
steel solidifications
r
O.J 0.3 0.5 m 0.T X 0-3
VVveorv experience / /
Figure A: Theoretical and experimental profiles

of the heat flux (j> and the shell tickness
e (slag A)
0 1 Figure 6: The fusion experiments 12/
0.3 0-5" m .- 0.3
Figure 5: Theoretical and experimental influence

of the nature of the slag on the heat
flux profile and the mean heat flux
extracted by the mould
0.6 .2 mnn J. 4 -
-rheor>
e*frer-i tuce 11 I
DISCUSSION
Question Prsident : Y atiL une corrlation simple pour un laitier

entre la consistance l'tat solide et sa viscosit l'tat
liquide ?
Rponse La consistance G d'un corps solide caractrise la loi de

comportement du corps :
1
2 2 p 2 2 ' *
S G
^ =3 IIZ-.JI
L

i,j *
o j S
S
t jj sont li
j les composantes du dviateur de contrainte
l j j sont les
i composantes des vitesses de dformation
3 j = 13.
m est l'indice de pseudoplasticit

m = o corps plastique (de Miss) G = limite d'lasticit
m = 1 corps newtonien G/3 = viscosit du corps.
. fn
Par exemple, en cisaillement simple : *Z = G y .
G et m dpendent de la temprature du corps. On pourrait

imaginer que, lors de la fusion d'un corps, qui s'effectue dans
un certain intervalle de temprature, G et m varient continment
et donc qu'il existe une relation entre viscosit l'tat
liquide et consistance l'tat solide. Mais ce problme ne
semble pas avoir t tudi.
En toute rigueur, cette comparaison entre diffrents corps
devrait faire intervenir la temprature homologue T/Tp o est
la temprature absolue de sollicitation et Tp la temprature
absolue de fusion. Pour les alliages mtalliques, la contrainte
d'coulement dpend avant tout de T/Tp et d'une nergie
d'activation thermique Q s caractristique de la structure de
l'alliage. Ceci suggre qu'une corrlation^ 4T/Tp doit
exister pour les phases liquides et doit faire intervenir une
chaleur d'activation thermique caractristique Q(. de liaison
l'tat liquide du corps considr. Mais la Littrature ne
semble pas fournir d'informations sur les relations entre Q s
et Q L .
ECONOMIE D'ENERGIE FAR MISE EH OEUVRE DE NOUVEAUX CYCLES THERMIQUES
APPLIQUES A DES PIECES DE FONDERIE DEMOULEES A CHAUD JUSTE APRES LA FIN DE
SOLIDIFICATION
Auteurs : *C. BAK, *R. BELLOCCI, **J.M. SCHISSLER
Contrat numro : EEB-1-120 F
Dure : 24 mois 15 septembre 1981 - 15 septembre 1983
Chef de projet : R. BELLOCCI, Centre de Recherches de PONT-A-MOUSSON
Contractants : * Centre de Recherches de PONT-A-MOUSSON

et adresses B.P. 28 - 54700 PONT A MOUSSON
** Institut National Polytechnique de Lorraine

Laboratoire de Mtallurgie associ au CNRS N 159
Parc de Saurupt - 54042 NANCY CEDEX
Resum
Le but de cette tude, mene sur une fonte graphite spheroidal, est
de supprimer toute phase de rchauffage des pices moules destines
tre traites thermiquement, par une opration de dmoulage
chaud, suivie par un cycle thermique contrl. Cette nouvelle mise en
oeuvre est appele "cycle thermique conomie d'nergie" par
opposition aux cycles thermiques conventionnels.
Pour aboutir l'tude comparative des tats initiaux de haute temp-
rature, obtenus par trempe l'eau depuie la temprature de dcochage
chaud ou depuis la temprature de rchauffage ; nous avons entre-
pris la dtermination de tous les paramtres exprimentaux de cette
tude :
- choix et dimensionnement des prouvettes
- conditions de faisabilit de l'opration de dcochage chaud
- slection des diffrents cycles thermiques
- ralisation exprimentale de ces cycles thermiques
- examens structuraux et analyse des premiers rsultats comparatifs.
L'ensemble de ce travail reprsente une tape prliminaire indispen-
sable l'tude des structures ferritique, perlitique et bainitique.
Les examens montrent qu'un dcochage chaud suivi d'un maintien mme
court haute temprature augmente nettement la trempabilit des
prouvettes massives par rapport un traitement conventionnel.
Nous avons galement vrifi la possibilit de mise en oeuvre indus-
trielle d'un cycle conomie d'nergie, sur un tube de diamtre 370
mm et de 1 m de long, labor par coule de la fonte centrifuge sur
une coquille mtallique.
-62-
Summary
The goal of this study on a spheroidal graphite oast iron was to

avoid reheating castings, prior to heat treatment, by using hot
ummolding followed by a controlled thermal cycle. This new process is
called "energy saving thermal cycle" as opposed to conventional
thermal treatments.
In order to compare the different initial high temperature
structures, observed after water quenching either from the unmolding
temperature or from the reheating temperature, we first determined
the following experimental parameters :
- choice of test samples and of their dimensions
- conditions of hot unmolding
- selection of thermal cycles
- carrying out of experimental thermal cycles
- structural examinations and analysis of first results.
This work was a necessary step for the study of the ferritic,
pearlitic and bainitic structures. It showed that even a short high
temperature holding after hot unmolding is enough to increase
hardenability of massive samples markedly as compared to conventional
treatments. We also tested the industrial feasibility of such an
energy saving treatment on a one meter long 370 mm diameter cast iron
pipe centrifugaly cast on metalic mould.
1. Introduction
Dans 1 industrie de la fonderie, une tendance actuelle qui se dve-
loppe est l'obtention de pices en fonte graphite spheroidal rpondant
des exigences mcaniques prcises, en relation avec une structure contr-
le de la matrice (structure bainitique par exemple).
Dans le cas gnral d'une fahrication classique, l'issue du refroi-
dissement des pices coules en sable, les constituants en prsence dans
la matrice sont essentiellement la ferrite, la perlite et le graphite. La
proportion de ferrite et de perlite dpend la fois de la massivet et de
la composition chimique de la fonte. A partir de cet tat brut de coule,
l'obtention d'une structure dfinie ncessite dans la majorit des cas un
rchauffage pour porter toute la masse de la pice en phase austnitique.
Cette tape, suivie par la mise en oeuvre d'un refroidissement contrl
aboutissant la transformation structurale dsire, reprsente le traite-
ment thermique conventionnel de la fonte.
Notons cependant que le traitement thermique des pices moules dont
le but est d'amliorer les caractristiques mcaniques de 1 alliage, ne
peut pas tre considr comme une simple adaptation des traitements ther-
miques classiques appliqus aux alliages corroys. Les facteurs qui s'op-
posent cette similitude sont d'ordre structuraux et technologiques. Un
alliage moul possde une structure chimiquement htrogne qui reflte
l'historique de sa solidification. De plus il necessite une operation de
dmoulage ralise habituellement une temprature voisine de la tempra-
ture ambiante, avant tout rchauffage.
Compte tenu de la dpense nergtique importante que reprsente ce
rchauffage, il semble peu judicieux de laisser refroidir une pice de
fonte qui vient d'tre mise en forme haute temprature, pour la rchauf-
fer ensuite. Le but de cette tude est de supprimer toute phase de r-
chauffage des pices moules destines tre traites thermiquement, par
une opration de dmoulage chaud, immdiatement suivie par un cycle
thermique contrle. Nous appelerons cette nouvelle mise en oeuvre "cycle
thermique conomie d'nergie ou cycle PAC" par opposition aux cycles
thermiques conventionnels.
Les caractristiques mcaniques obtenues aprs mise en oeuvre d'un
cycle conomie d'nergie doivent tre d'un niveau au moins gal celui
rsultant de la mise en oeuvre d'un cycle thermique conventionnel.
Les 3 axes de recherche qui ont t dvelopps sont les suivants :
- Etude de la faisabilit de l'opration de dcochage chaud
- Dfinition des traitements thermiques appliqus partir d'une matrice
de haute temprature
- Caractrisation, aprs trempe l'eau, de 1'ex-structure austnitique
vierge telle qu'elle peut se prsenter au moment du dcochage vers
1000 C ou aprs un maintien 1000 C (en comparaison une structure
austnitique rgnre au cours d'un cycle thermique conventionnel.
2. Description des installations et techniques exprimentales

L ensemble des installations necessaires pour realiser cette tude se
compose de la partie laboration et traitement de la fonte, de la partie
moulage, d'une aire de coule et de dcochage des pices, des fours de
traitement thermique et des machines centrifuger la fonte.
2.1. Elaboration et mise en forme de la fonte

La fonte servant a cette etude est labore par fusion dans un four
induction de 500 kg moyenne frquence, partir de fonte neuve de
haut-fourneau. A l'issue de l'laboration et des divers traitements, cette
-64-
fonte peut tre mise en forme par coule et solidification dans un moule
de sable ou dans un moule permanent (coquille mtallique par exemple). La
technique d'obtention de pices par coule en moule sable est de loin la
plus usite. Elle permet d'laborer des pices de forme simple ou comple-
xe, de quelques centaines de grammes plusieurs centaines de kg. Si nous
prenons le cas de la solidification de pices en coquille mtallique, elle
peut se faire en coquille mtallique soit statique, soit dynamique. C'est
cette deuxime technique qui est essentiellement mise en oeuvre dans notre
Socit pour la ralisation de pices en fonte graphite spheroidal. En
plus des elaborations en moule sable, nos essais porteront sur l'obtention
de structures dfinies dans le cadre de la coule en coquille mtallique
dynamique. Dans l'optique de l'obtention d'une matrice bainitique, nous
avons t amens quiper la machine centrifuger d'un systme de
refroidissement par aspersion d'eau l'intrieur du tube lorsque ce
dernier atteint une temprature voisine de 1000C.
2.2. Installation de traitement thermique

La realisation des cycles de traitement thermique conduisant l'ob-
tention des structures dfinies a ncessit l'implantation de 2 fours de
traitement thermique lit fluidis dont le chauffage est assur par des
rsistances lectriques. Ces fours, d'une puissance de 45 kW possdent un
mouffle de diamtre utile gal 600 mm pour 800 mm de hauteur utile. Leur
domaine d'utilisation se situe entre la temprature ambiante et 1100C.
Dans le prolongement de ces fours qui ont t enterrs, nous avons plac
des chassis nous permettant de positionner les moules de sable destins
contenir la fonte (figure 1 ) . Cette aire de remplissage constitue gale-
ment l'aire de dcochage chaud, puisque sur cette installation expri-
mentale, les deux oprations de coule et de dcochage sont ralises sans
mouvement du moule. Lorsque le cycle thermique l'impose, nous installons
proximit des fours un bac, contenant 200 1 d'eau, destin raliser une
trempe sur les pices dcoches ou traites thermiquement.
Les manipulations et les transferts de pices temprature leve
sont assurs par un palan fix sur un monorail situ dans l'axe des fours
et de l'aire de coule.
2.3 Equipement de mesure

Les cycles thermiques rels sur pices sont suivis par un enregistre-
ment de la temperature l'aide de thermocouples inserrs au coeur des
pices et 2 mm de la surface et relis un enregistreur de type x-t.
Cette technique nous a permis galement d'tablir un talonnage des fours
lit fluidis de manire raliser, aux diffrentes tempratures d'uti-
lisation, des cycles thermiques reproductibles qui permettent une tude
comparative prcise entre cycles conventionnels et cycles conomie
d'nergie.
Dans le cas de la coule de la fonte sur coquille mtallique, durant
le cycle thermique, la temprature est releve de manire soit continue
par un pyromtre optique, soit ponctuelle au cours du temps par l'interm-
diaire d'une canne de mesure de temprature par contact.
2.4 Examens mtallurgiques

Apres traitements thermiques, des chantillons sont prlevs dans des
zones particulires des pices puis examins au Laboratoire de Mtallurgie
LA 159 de l'Ecole des Mines de NANCY. Toutes les techniques classiques de
preparation d'chantillon, d'observation et d'analyse de la fonte gra-
phite spheroidal sont mises en oeuvre.
-65-
3 Avancement des travaux
La premiere phase des travaux a ncessit une tude prliminaire
importante pour fixer les paramtres indispensables une tude comparati-
ve rigoureuse entre les diffrentes caractristiques obtenues l'issue de
cycles thermiques conventionnels et de cycles thermiques conomie
d'nergie. Nous allons examiner quelques uns de ces paramtres.
31 Composition des fontes

La fonte industrielle qui a t choisie a la composition suivante :
3,6 % C ; 2,2 % Si ; 0,16 % Mn et 0,05 % P. La quantit trop faible d'l-
ments d'alliages trempants ne nous permet pas dans nos conditions opra-
toires d'atteindre le constituant bainitique de la matrice des fontes.
Pour ce cas, nous nous orienterons vers une composition de fonte faible-
ment allie par addition la fonte de base d'lments comme le manganse,
le nickel ou le molybdne.
3.2. Type d'prouvettes et dcochage chaud de pices coules en moule de

sable
De maniere simuler le comportement de pices industrielles au cours du
traitement thermique, les essais devaient porter sur des prouvettes pr-
sentant par leur gomtrie des vitesses de solidification et de refroidis-
sement diffrentes. Afin d'isoler les principaux paramtres qui dpendent
de la gomtrie des prouvettes, nous avons adopt 2 prouvettes de massi-
vit trs diffrentes.
- Pour contrler les gradients thermiques l'intrieur des pices, nous
avons choisi des prouvettes de forme cylindrique de diamtre gal 18
mm et 80 mm.
- De manire ne tenir compte que des changes thermiques radiaux, l'tu-
de est mene sur un secteur D gal au diamtre d'prouvettes de hauteur
minimale gale 3D.
Dans le but de disposer d'un certain nombre d'prouvettes d'essais
mcaniques de manire tablir une comparaison entre les diffrents
cycles thermiques, nous avons labor les jets de diamtre 18 mm sous la
forme de grappes regroupant 16 jets (figure 2 ) . Pour obtenir un mtal sain
dans la zone de prlvement des chantillons, diverses coules exprimen-
tales nous ont permis de fixer la hauteur des jets 350 mm pour le diam-
tre 80 mm et 65 mm pour le diamtre 18 mm.
Dans le cas de la coule en moule sable, la faisabilit de l'opra-
tion de dcochage chaud est lie plusieurs facteurs dont la nature du
sable de moulage utilis. Deux catgories de sables ont t testes : le
sable vert et les sables lis l'aide d'agglomrants chimiques. Dans
cette dernire catgorie, nous avons choisi un sable obtenu par le procd
Promex (le pep-set).
A l'issue du dmoulage chaud de la grappe de jets de diamtre
18 mm, nous avons pu constater que l'adhrence trs importante du sable
vert ne permet pas d'obtenir une pice suffisamment propre pour tre imm-
diatement traite thermiquement. Par contre, durant la solidification et
le refroidissement d'une fonte moule dans une empreinte de sable ralise
l'aide du procd Promex, la rsine qui lie le sable est brle au
contact du mtal. Aprs dcochage chaud, comme froid d'ailleurs, il ne
subsiste sur les pices qu'une mince pellicule peu adhrente de rsidu
calcin. Le comportement favorable de ce sable peut tre tendu tous les
autres procds utilisant des agglomrants chimiques tels que les sables
lis l'aide des rsines furanniques, les sables issus des procds
Ahsland ou Croning.
33 Type d'prouvettes dans le cas de la solidification en coquille
mtallique
L elaboration de pices en fonte coules en coquille mtallique
semble particulirement bien adapte la mise en oeuvre de cycles thermi
ques conomie d'nergie aprs dcochage chaud car elle permet d'ex
traire facilement et rapidement des pices haute temprature dont les
surfaces sont propres. Dans notre cas, la technique employe qui est la
centrifugation de la fonte liquide coule sur une coquille mtallique en
rotation, permet d'obtenir sur nos installations soit des tubes de 370 mm
de diamtre, 12 mm d'paisseur et 1 m de longueur soit des tubes de 110 mm
de diamtre, 15 mm d'paisseur et de longueur gale 220 mm. Dans un
premier temps, pour dmontrer la faisabilit du processus d'obtention de
structures contrles par mise en oeuvre de cycles conomie d'nergie,
nous avons entrepris le traitement thermique de tubes de grand diamtre.
34 Droulement des cycles thermiques conduisant aux diverses structures

341 Obtention des diverses matrices
On peut prvoir le domaine d'existence des diffrentes phases et leur
obtention soit partir des courbes TTT (I), soit partir des courbes
TRC. Rappelons que les courbes TTT(l) reprsentent les diffrents domaines
de la transformation de l'austnite ralise par maintien isotherme la
suite d'une trempe tage depuis une temprature suprieure 850C, jus
qu' la temprature du maintien isotherme. Les courbes TRC reprsentent
les domaines de la transformation de l'austnite par refroidissement
continu ; l'tat de dpart tant toujours une phase austnitique une
temprature suprieure 850 C. Nous avons plac sur les figures 3 la
reprsentation schmatique des courbes TTT(I) et TRC d'une fonte, ainsi
que le profil des diffrents cycles thermiques capables de promouvoir des
structures entirement ferritiques, perlitiques, bainitiques ou martensi
tiques. Ces diffrents cycles ont t numrots et regroups sur le
tableau I. Nous pouvons remarquer que la phase ferritique obtenue
l'issue du cycle thermique 7 est ralise par une ferritisation indirecte
constituant cycles thermiques type de transformation

cycle thermique 1 transformation en refroidissement
Ferrite continu
cycle thermique 6 graphitisation en maintien iso
cycle thermique 7 therme
Perlite cycle thermique 2 transformation en refroidissement
continu
cycle thermique 3 transformation en refroidissement
Bainite (S) continu
cycle thermique 5 trempe tage et transformation en
maintien isotherme
Martensite cycle thermique 4 transformation en refroidissement
continu
Tableau I : Rcapitulatif des diffrents cycles thermiques conduisant

l'obtention des diverses structures.
34.2. Slection des cycles thermiques

. Cycles thermiques conventionnels
Tous les cycles thermiques que nous avons reprsents, hormis le
cycle 7, ont pour point de dpart la phase austnitique. En consquence,
un cycle thermique classique ncessite dans tous les cas sauf un, un
rchauffage pralable des pices, dont le but est d'atteindre, dans la
67
masse de ces dernires, l'tat austnitique. Ce rchauffage est ralis
la temprature de 1000C.
La matrice des fontes prsente, comme tous les alliages mouls, des
htrognits chimiques importantes induites au cours de la solidifica-
tion. Ces htrognits, qui affectent largement les transformations
structurales et les proprits mcaniques de la fonte peuvent tre att-
nues par un maintien prolong en phase austnitique. L'influence du degr
d'homognit de la phase austnitique est un paramtre que nous exami-
nons. Dans un premier cas, les cycles thermiques comportent une austniti-
sation 1000 C durant le temps ncessaire pour atteindre l'quilibre en
temprature dans l'ensemble de la masse des pices. Dans le second cas,
nous prolongeons d'une dure de 20 mn 1 heure le maintien prcdent
1000 C de manire modifier de faon sensible la rpartition des lments
d'alliage. C'est dans ce but que la temprature de 1000C, qui peut para-
tre leve, a t choisie.
. Cycles thermiques conomie d'nergie
La mise en oeuvre des cycles thermiques classiques s'applique des
pices qui sont au dpart du cycle la temprature ambiante. Nous appe-
lons cycles thermiques conomie d'nergie, des cycles qui s'appliquent
des pices dont la temprature de dpart est suprieure 800C de manire
ce que les constituants en prsence soient essentiellement l'austenite
et le graphite. Cet tat de dpart austnitique, qui permet de supprimer
les phases de rchauffage qui existent dans un cycle thermique classique,
est intercept au cours du refroidissement des pices, juste aprs solidi-
fication, par un dcochage la temprature de 1020 C. A partir de cet
tat de dpart austnitique, 3 possibilits s'offrent nous pour raliser
des cycles thermiques identiques aux cycles thermiques classiques.
- enclencher directement les cycles thermiques aprs dcochage 1020C
- raliser un quilibre thermique 1000 C et enclencher les cycles ther-
miques
- aprs avoir obtenu un quilibre thermique 1000 C, maintenir les pices
durant 20 mn 1 heure cette temprature, puis enclencher les cycles
thermiques.
Notons que la premire possibilit reprsente le cas extrme le plus
favorable sur le plan nergtique.
Nous avons plac sur la figure 4 l'organigramme des diffrentes
combinaisons ralises pour l'obtention des structures par mise en oeuvre
des cyclee thermiques conventionnels et des cycles thermiques conomie
d'nergie. Nous avons slectionn les cycles 6, 2, 5 et 4 pour obtenir
respectivement les phases ferritique, perlitique, bainitique et martensi-
tique. Des essais prliminaires de traitement thermique par mise en oeuvre
de cycles thermiques classiques nous ont permis de prciser les conditions
d'application de ces diffrents cycles :
- cycle 6 (ferrite) : trempe tage et maintien isotherme de 1 heure
720C
- cycle 2 (perlite) : trempe en lit fluidis temprature ambiante
- cycle 4 (martensite) : trempe l'eau
- cycle 5 (bainite) : trempe tage suivie d'un maintien isotherme entre
300 et 400 C sur une fonte de composition adapte.
35 Exploitation de cycles thermiques exprimentaux

351 Essais raliss sur un tube de 0 370 mm
Dans un premier temps, nous nous sommes attachs vrifier la possi-
bilit de mise en oeuvre industrielle d'une fonte structure bainitique
applique un tube de 0 370 mm et de 1 m de longueur ralis par coule
-68-
de la fonte centrifuge sur une coquille mtallique. Le cycle 5 que nous
avons mis en oeuvre au cours du refroidissement du tube aprs sa solidifi-
cation ncessite un refroidissement intense de la phase austnitique,
depuis 1000 C jusqu' une temprature voisine de 450 C. Aprs de nombreux
essais, nous avons pu raliser cet-te cintique de refroidissement d'une
manire reproductible. Nous avons ensuite recherch une fonte dont la com-
position chimique est compatible avec la cintique de refroidissement pr-
cdemment dfinie. La composition qui a t retenue est la suivante : 3,5
% C ; 2,7 % Si ; 0,4 % Mn ; 0,75 % Cu ; 0,1 % Mo. Au cours du traitement
thermique, aprs l'arrt de la phase de refroidissement vers 450C, on
procde aux oprations de dcochage et d'enfournement rapide du tube,
suivies d'un maintien de 1 heure 400 C. Ceci nous a permis d'obtenir une
structure typiquement bainitique comme en tmoigne la micrographie de la
figure 5
Nous avons ensuite procd une tude comparative de l'aspect struc-
tural et des caractristiques mcaniques observes d'une part sur cette
structure obtenue par mise en oeuvre d'un cycle thermique conomie
d'nergie, et d'autre part sur la structure obtenue par mise en oeuvre
d'un cycle thermique conventionnel. Le cycle thermique conventionnel a t
ralis par raustnitisation la temprature de 1000 C des chantillons
prlevs sur un tube. Ces chantillons ont subi une trempe tage en bain
de sel la temprature de 400 C, conformment au cycle 5 La structure
issue de ce cycle thermique conventionnel est reprsente sur la figure 6.
Les caractristiques mcaniques associes ces structures ont t
prsentes sur le tableau II.
Cycle conventionnel Cycle conomie d'nergie

. dpart : temprature . dpart : tat austnitique
Diffrentes ambiante . dbut de refroidissement :
phases des austnitisation : 1000C
cycles 1000C/30 mn . fin de refroidissement :
thermiques . trempe tage en bain de 450C
sel 400C . dcochage et enfournement
400C
. maintien : 1 heure . maintien : 1 heure
E 0,2 55 daN/mm 68 daN/mm
2
Rt 88 daN/mm 85 daN/mm2
A % 8 % 6 %
Tableau II : Cycles thermiques et caractristiques mcaniques compares des
traitements thermiques conventionnels et des cycles thermi-
ques conomie d'nergie.
Ce premier stade a eu un rsultat positif puisque nous avons pu
mettre au point, aprs dcochage chaud, un traitement thermique permet-
tant l'obtention de caractristiques mcaniques similaires celles obte-
nues par un traitement thermique conventionnel.
Etant assur de l'application possible de cette nouvelle technique,
nous avons entrepris l'tude structurale de l'tat initial obtenu par
trempe martensitique suivant le cycle 4 sur les chantillons labors dans
les moules sable.
352. Cycles thermiques raliss sur prouvettes coules en moule sable

Les diffrents cycles thermiques qui ont t raliss sur les prou-
vettes coules en sable ont pu tre suivis grce l'enregistrement des
courbes de type 0 = f(t). Nous possdons les deux valeurs extrmes, en
fonction du temps, du gradient thermique qui existe le long d'un diamtre
des cylindres.
-69-
L'enregistrement, depuis la temprature de coule, des courbes de
refroidissement des prouvettes refroidies en moule sable jusqu' 20C et
des prouvettes dcoches 1020C et refroidies l'air, nous a permis de
connatre les conditions thermiques dans lesquelles est effectue l'opra
tion de dcochage qui a lieu 1020C, et de confirmer la faisabilit de
cette opration qui peut tre ralise totalement en phase austnitique.
En vue des bilans nergtiques futurs, ce sont les prouvettes massi
ves qui prsentent le plus d'intrt. Dans ce but, nous avons plac sur
le3 figures 7a, b et c les cycles thermiques l'issue desquels nous obte
nons :
la phase austnitique brute de dcochage (figure 7a)
la phase austnitique dcoche 1020*C et rquilibre thermiquement
1000C (figure 7b)
la phase austnitique ralise de manire conventionnelle par rchauffa
ge de la structure depuis 20C, jusqu' la temprature de 1000C (figure
7c).
Ces diffrents tats reprsentent les points de dpart des cycles
thermiques repre A, B et B' sur la figure 4 Nous avons port sur le
tableau III les lments essentiels une comparaison nergtique pour
atteindre ces points de dpart.
Cycles thermi ques Oprations Temps de sjour dans

le four 1000C
Cycle D.A C A / 0
Cycle D.A C B Equilibrage en temp rature 18 mm
Cycle D.A C C Maintien d'homog 20 mn 38 mn
nisation 1 h 78 mn
Cycle conven rchauffage 45 mn
tionnel B'
Cycle conven rchauffage + 20 mn 65 mn
tionnel C' maintien d'homog 1 h 105 m
nisation
Tableau III : Aspect compar des divers cycles thermiques.
La poursuite de ces cycles thermiques est en cours, mais il nous a
sembl ncessaire de raliser auparavant une tude comparative de ces
tats de haute temprature qui donneront naissance aux diverses structures
l'issue des cycles thermiques complets.
4. Rsultats et analyse
Les examens micrographiques qui ont t raliss sur les prouvettes
trempes a 1 eau a partir de 1000 C a l isue des diffrents cycles ther
miques montrent une evolution importante de la trempabilite. Nous avons
porte sur la figure 8 la proportion de martensite value le long d'un
rayon, en fonction du cycle thermique. On peut constater d'une part que
dans le cas des prouvettes dcoches chaud, la trempabilite augmente
avec le temps de sjour a 1000 C, ce qui est pas le cas pour les prou
vettes qui ont subi un traitement conventionnel, et d'autre part que la
trempabilite des prouvettes dcoches chaud et maintenues 1000*C est
nettement superieure a la trempabilite des prouvettes rchauffes a
1000 C. Cette dernire constatation est confirme par l'examen des courbes
de duret que nous avons traces sur la figure 9
Dans la mesure o toutes les prouvettes ont t labores partir
d'une mme coule et que nous avons vrifi qu'un sjour de 20 mn 1000 C
ne modifie pas d'une manire sensible les sgrgations qui existent,
70
figures 10 et 11, nous pouvons carter l'ventualit d'une variation de
trempabilit en relation avec la composition chimique. Actuellement, pour
expliquer ce phnomne, la seule hypothse que nous pouvons mettre est
relative l'tat des contraintes internes qui existent dans les pices,
en fonction du cycle thermique. D'ailleurs, les prouvettes dcoches
chaud ne tapent pas aprs trempe l'eau alors que les prouvettes
rchauffes 1000 C se sont casses en deux aprs la trempe.
5 Conclusions
Si nous reprenons les 3 groupes de cycles thermiques que noue avons
examins :
- Groupe 1 : dcochage chaud suivi d'un refroidissement contrl (A)
- Groupe 2 : dcochage chaud suivi d'un maintien haute temprature,
puis mise en oeuvre d'un refroidissement contrl (B et C)
- Groupe 3 : cycle thermique conventionnel (B' et C ) .
On peut noter que sur le plan conomie d'nergie, l'intrt crot du
groupe 3 vers le groupe 1. Sur le plan mtallurgique, dans le cadre de
l'obtention de structures comme la perlite, la bainite ou la martensite,
structures qui ncessitent une trempabilit plus ou moins importante, la
tendance qui se dgage semble montrer qu'un dcochage chaud suivi d'un
maintien mme court haute temprature est particulirement intressant.
Un tel cycle thermique correspond au groupe 2.
D'ores et dj, on peut affirmer que sur les plans nergtiques et
mtallurgiques, la mise en oeuvre d'un cycle conomie d'nergie semble
particulirement bnfique. Les consquences de ce premier rsultat vont
tre suivies attentivement sur les diffrents cycles thermiques aboutis-
sant aux structures ferritiques, perlitiques et bainitiques qui sont
actuellement en cours de ralisation.
Ces premiers rsultats ont pu tre obtenus aprs une tude prlimi-
naire importante. Cette dernire nous a permis de fixer soigneusement tous
les paramtres exprimentaux de cette recherche. Nous avons en particulier
dtermin exprimentalement les cycles thermiques qui aboutissent chacu-
ne des structures tudies. Cette mise en oeuvre exprimentale a pu tre
mene bien aprs l'analyse des diffrentes possibilits et la rsolution
des problmes concernant :
- le choix et le dimensionnement des prouvettes
- les conditions de faisabilit de l'opration de dcochage chaud
- la recherche des paramtres de fluidisation conduisant une bonne
reproductibilit des cycles thermiques
- la slection des diffrents cycles thermiques.
Enfin, nous nous sommes attachs vrifier la possibilit de mise en
oeuvre industrielle d'un cycle conomie d'nergie adapt l'obtention
d'une matrice bainitique sur un tube de diamtre 370 mm et de longueur
gale 1 m, labor par coule de la fonte centrifuge sur une coquille
mtallique. Les examens micrographiques raliss sur diffrents prlve-
ments nous ont confirm la faisabilit de cette opration.
Figure 1 : Vue de l'installation de traitement thermique
0 80
rti
11 i i ' i ' i i rU
H L 018
Zones de prlvement
des chantillons
Figure 2 : Type d'prouvettes utilises au cours de l'tude

Jet 0 80 mm et grappe de 16 jets 0 18 mm.
&S{**: Ferrite'
Cycle 7
COURBE TRC (OIRht TTT(I)
Figure 3 : Representation schmatique des courbes TRC et TTT(I) des fontes
Cycles thermiques Cycles thermiques

conomie d'nergie conventionnels
Maintien d'homognisation
a 1000 C
Dparti
temprature
voiiine de1000C
Figure 4 : Aspect comparatif des cycles thermiques-conventionnel s et

des cycles thermiques conomie d'nergie
-73-
^ ^ ' ' ' >" -',--' ^'i t ^
- - *
$W?-*/ ".!'- ';'..f - ^ - * >:^V:v-g'
>:#&#".-j!
Figure 5 : Structure bainitiq ue obtenue sur un tube centrifug de

0 370 mm, l'issue d'un cycle thermiq ue conomie
d'nergie
10
*: tftfSfcu.
Figure 6 : Structure bainitiq ue obtenue sur un tube centrifug de

0 370 mm l'issue d'un cycle thermiq ue classiq ue
c
: - , 1
fi
E
fi 1000
SS
//
800-
//
600
. 1
,7
il
1
400- 1 ' il
il
200 ' il

ti 1 >
- 1 1

K40mn
I0l8mn
Figure 7 : Diverses possibilits d'obtention de la matrice austnitique de haute temperature

sur un jet de 0 80 mm
par dcochage 1020C (figure 7a)
par rquilibrage thermique 1000C (figure 7b)
par rchauffage depuis la temprature ambiante jusqu' 1000C (figure 7c)
Maintien 1000C avant trempe
iioo
C>cle DAC Cycle classique
* 3 8 mn
rechauffe en
18 mn * 40mn
G 0 mn
Cote(fnm)
Figure 8 : Pourcentage de martensite relev sur un rayon des jets de

80 mm en fonction du cycle thermique avant trempe
o Cycle classique
Cycle DAC (38mnlOOOC)
Figure 9 : Influence du cycle thermique sur la duret mesure

sur des jets de 0 80 mm
^
v.:

Si
Cellule
Figure 10 :
Rpartition du Si et du Mn
entre une cellule et un
eoo joint de cellule de 1'ex
structure de solidification.
Analyses ralises sur un
jet de 0 80 mm, dcoch
1020C et tremp eau
Nodule 10
d graphite
Distance en
J
M M^N. <* '
(Mn)
y
* y 0,6
\" y
0.5
/
2 0,4
10
Figure 11 : Variation du rapport de concentration des lments Si et Mn
enregistre entre le joint de l'excellule (J) et la matrice
(M) en fonction du temps de maintien en phase austnitique
(par rchauffage 1000C)
DISCUSSION
Question - Dr. Ehringer, CEC : Est-ce que vous avez L'intention de faire
des tests d'usinage ?
Rponse Nous raliserons des tests d'usinage sur Les structures

ferritique, bainitique et perlitique qui sont en cours de
ralisation.
Question - Mr Zegers, CEC : Is the technique of heat treatment with a

fluid bed already applicable on an industrial scale or are
there still technical and economical problems ?
Rponse Si l'utilisation du lit fluidis basse temprature semble

maintenant au point, la technique de chauffage en lit fluidis
haute temprature est plus rcente.
Il est ncessaire d'accumuler une bonne exprience du lit
fluidis utilis haute temprature pour juger des possibili-
ts d'application industrielle d'une telle technique.
Dans notre cas, l'application rcente des fours ne nous a pas
encore permis de connatre parfaitement les performances ou
les points faibles de ce matriel.
Question - Mr Felder, Armines : Serait-ce les rsidus du moule qui changent

la courbe de trempabilit du produit ?
Rponse Les rsidus du moule sur les pices (quand ils existent) sont
trs faibles et trs peu adhrents.
1. Le milieu en mouvement et lgrement abrasif que reprsente
le lit fluidis permet d'liminer trs tt ce rsidu au cours
du maintien haute temprature.
2. Les pices qui ont subi un cycle thermique conventionnel
sont coules dans des moules identiques ceux des pices
dcoches chaud. Le mme rsidu existe sur ces pices.
On ne peut donc pas considrer que le rsidu du moule sur les
pices, quand ce rsidu existe, influence la courbe de
trempabilit du produit.
REDUCTION IH ENERGY REQUIREMENTS FOR CONVERTING LIQUID
ALUMINIUM TO SEMIFA BRICA TED ROLLED PRODUCTS
Author J Dennis
Contract number EEB1102GB
Duration 24 months 1 October 1980 30 September 1982
Head of project J Dennis
Contractor The British Aluminium Company pio
Address Chalfont Park

Gerrards Cross
Bucks SL9 OQB
Summary
Examination of the conventional DC castinghot rolling route

for the production of semifabricated rolled products from
liquid aluminium has confirmed that it is an energy intensive
process. The alternative processing routes of thin strip
casting should consume less energy because they dispense with
the scalping, preheating and hot rolling operations of the
conventional route.
A total energy audit of six typical rolled products produced

by both routes was undertaken to determine the potential
energy savings. These audits included the effect on energy of
scrap generated, irrecoverable metal losses and yields for
each product in addition to the process energy per tonne at
each stage of manufacture.
It was confirmed that product energy savings of up to 37$

were possible using thin strip casting methods. These savings
were less than expected because of increased cold mill passes
and long high temperature homogenising anneals which partially
offset the energy gains of eliminating the scalping, pre
heating and hot rolling operations of the conventional route.
The energy requirements of both processing routes for all six

products selected were related to a typical product mix of
40,000 tonnes per annum. It was established that the
potential energy saving using the thin strip casting route to
achieve this output would be approximately 25$ or 3 million
therms equivalent to 315 517 G J per annum.
1 . Introduction
1.1 The bulk of the semi-fabricated rolled products manufactured by the
aluminium industry in the past has been produced by the conventional DC
casting-hot rolling process. The apparently high energy requirement
associated with this route compared with the other alternative processes
for the conversion of liquid aluminium to rolled products initiated this
detailed assessment of the potential energy savings.
Five alternative processes were considered worthwhile investigating.
These were continuous thin strip casting using (l) roll caster (2) block
mould caster (3) fixed mould caster (4) spray casting-rolling and (5)
pellet compacting by rolling. The first of these processes is well
established on a production basis with machines manufactured by Seal
Engineering, Hunter Engineering and Alusuisse installed throughout the
world. The four other processes, however, are not used extensively and
the processes concerned must be fully investigated before their considera-
tion for bulk production.
From the point of view of reduced energy consumption these alternative
manufacturing routes are particularly attractive because they eliminate the
energy intensive scalping, preheating and hot rolling processes of the con-
ventional route. Figure 1. illustrates the possible reduction in the
number of processes.
In addition to reducing the energy input because of a reduction in
manufacturing processes, however, the alternative methods should give rise
to less scrap and therefore less scrap remelting should be required per
tonne of product produced. Scrap remelting also gives rise to increased
melt losses which can be considerable for certain types of scrap such as
sawing swarf and scalpingsparticularly if associated with the higher
magnesium aluminium alloys.
To establish the amount of energy required to produce a tonne of a
given product, therefore, consideration must be given not only to the
energy input at each stage of manufacture in the process used but also the
energy requirement for remelting scrap and replacing irrecoverable metal.
In view of the fact that primary metal is reputed to require 250 GJ of
energy per tonne to produce (to be verified in the contract work programme)
the importance of reducing scrap arisings and subsequent melt losses can
be appreciated.
The amount of scrap generated and melt losses incurred are directly
related to the number of processing stages, the type of scrap produced
and the aluminium alloys and tempers involved.
For the purpose of this investigation, therefore, six rolled products
were considered covering a typical range of alloys tempers and forms.
These are 1200 soft temper and 3003 H24 temper foilstock coil used for
thin gauge foil and container foil respectively, 1200 H H and 5251 H34
temper sheet, 3103 H18 corrugated S12 profiled sheet and 2024 T3 clad
sheet. The first five products can be produced by the roll caster
technique but it is envisaged that only spray casting could be used to
produce clad 2024 alloy.
1.2 The following work programme, therefore, was undertaken for this
investigation into the possible reduction in energy requirements for
converting liquid aluminium to rolled products.
(a) Examination of the conventional DC casting-hot rolling method of
production and the alternative methods of continuous thin strip casting
using moving moulds (roll and block) fixed horizontal mould, spray
casting-rolling and pellet compacting techniques.
(b) Establish equipment used and processing details of both the DC casting-
hot rolling and continuous thin strip casting methods of production for
-80-
the six products chosen for investigation.
(c) Determine yields, scrap generated and melt losses incurred in both the
DC casting-hot rolling and thin strip casting methods of manufacture for
the six products specified for comparison purposes.
(d) Determine the total energy input of the DC casting-hot rolling and
appropriate thin strip casting methods of production for each stage of
processing of the six products concerned for comparison purposes.
(e) Obtain samples of products produced by alternative methods of manufac-
ture for simulated cold rolling and property determination. The necessity
to use additional heat treatment or other operations when using the alter-
native production methods of thin strip casting should be established.
(f) Apply the findings to a typical rolling mill product mix of 40,000
tonnes per annum to establish the total potential annual energy savings.
(g) Establish the further reduction in energy requirements that could be
obtained by casting thinner and faster thus avoiding additional cold
rolling passes entailed by present strip casting thicknesses.
2. Examination of the DC casting-hot rolling and thin strip casting methods

of manufacture
The considerable differences in plant and energy requirements of the
conventional and non conventional routes for producing semi fabricated
products from liquid aluminium are best seen by comparing the DC casting-
hot rolling process with the established roll caster method of thin strip
casting.
The roll caster method was chosen in preference to the other
alternative methods because it was established on a production basis and
capable of producing 6 mm thick strip in the first five alloys of the
typical product mix detailed in 1.1. The remaining product 2024 clad T3
sheet cannot to date be produced on a production basis by any process
other than the conventional DC casting-hot rolling method. It is thought
likely that the spray casting-rolling technique could probably be developed
for this product and, therefore, a detailed examination of the energy
requirements of this process was also undertaken.
2.1 Description of DC casting-hot rolling process

(a) Melting and casting - Rolling blocks for this process are either cast
at a smelter or at the plant producing the semi fabricated rolled product.
In the case of the six products chosen for this investigation, the
1200 and 3103 alloys are cast at a smelter and the 3003, 5251 and 2024
alloys at the casting shop of the rolling mill plant concerned.
The energy benefit of casting at the smelter is obvious because liquid
metal from the reduction cells can be used for part of the charge compared
with 100$ solid ingot and scrap which has to be remelted at the rolling
mill. The resultant energy saving, therefore, is considerable using liquid
smelter metal but will apply to both conventional and non conventional
methods of processing and in both cases only a holding furnace is required.
Solid scrap ingot and hardners, however, constitute the charge at the
rolling mill casting shops and it is customary to use an oil fired melting
furnace for converting the solids to liquid metal and alloying to the
specification required. Degassing and cleaning of the metal is achieved by
using a BACO furneless in line degassing unit between the holder and DC
casting machine. This unit reduces the hydrogen content of the metal to an
acceptable level and also removes inclusions. To maintain the correct
casting temperature of the metal, however, it is also necessary to heat
this unit using gas burners.
Metal is then fed via launders into the vertical moulds of the semi-
-81 -
continuous casting machines and casting carried out at the rate of two to
five blocks at a time depending upon the capacity of the filtering equip-
ment and the size of the blocks being cast. Blocks of 1200, 3003 and 3103
alloys are cast 510 na thick but 5251 and 2024 alloys are cast 410 mm
thick. Blocks cast at the smelter are cast two in the length and those at
the rolling mill casting shop in single lengths.
(b) Scalping - If required by the end product concerned cast blocks from
both smelter and rolling mill sources are machined on their two major
surfaces to remove casting imperfections and undesirable structure.
(c) Preheating - The next process in the DC casting-hot rolling route is
preheating prior to hot rolling. Rolling blocks are charged into a batch
pusher type gas fired preheating furnace. These furnaces are capable of
heating 25 blocks each of 6 tonne weight from room temperature to 610 C and
also to maintain that temperature. The temperature used and the "soak"
tines depend upon the products concerned. Deep drawing 3003 material
requiring fine grain size high elongation values and good temper letting
down properties in the final foil, have to be homogenised for long periods
at a high temperature. The similar European alloy 3103 used for profiled
cladding sheet, however, only has to be heated to the hot rolling tempera-
ture to give the properties required for building products.
(d) Hot breakdown mill rolling - After preheating the rolling blocks are
discharged at the required rolling temperature on to the hot line tables
prior to breakdown rolling in a two high single stand 3,000 H.P. reversing
mill. This mill reduces the block thickness to 25 mm or thinner plate in
9 to 20 passes, depending upon the alloy and the entry and exit thickness
required at the hot strip mill.
The outgoing plate temperature from this breakdown mill can vary from
450 to 320 C depending upon the product concerned.
The next operation in the hot line is to shear the nose and tail end
scrap of the hot rolled blank. This squares the end of the blank for
subsequent rolling at the hot strip mill but also removes laminated metal.
The hot rolled and sheared blank can be edge trimmed if necessary at the
entry trimmer of the hot strip mill but this was only found necessary for
5251 and 2024 alloys.
(e) Hot strip mill rolling - The final hot rolling process is carried out
in a three stand, four high, tandem hot strip mill powered by 4,000 ]..P.
motors on each stand. This mill reduces the blank to thicknesses of from
4.0 to 2.5 Dim depending upon the alloy and thickness cf the final coil and
sheet product. If necessary the hot rolled coil can be further edge
trimmed before coiling on a drum mandrel using a steel belt wrapper.
The hot rolled coil product for the purpose of this investigation can
be compared directly with the products of alternative continuous strip
casting processes. Subsequent processing to achieve the final product form
temper and dimensions, therefore, will be similar for materials from both
processes except for additional cold rolling passes and heat treatment
operations required for the thicker strip cast product.
2.2 The roll caster thin strip casting process

The initial processes for melting and holding for each of the products
concerned are similar to those described for the conventional route. This
process also requires the use of a system for cleaning and degassing the
metal and this can be of the BACO furneless in line degasser type.
The casting process is entirely different, however, and requires
strict control of temperature to - 5 C of the required casting temperature.
Details of this method of production were obtained by witnessing the
operation of a Seal 3C Jumbo caster (see Figure 2) . In this process liquid
-82-
metal of the alloy concerned is transferred via a launder to a refractory
nozzle extending the full width of the coil to be cast and placed immediate
ly in front and between two internally water cooled rolls positioned one
above the other (see Figure 3) Demineralised water was used for this
purpose and the energy content of obtaining this was included in the audit.
The water cooled rolls were 960 mm in diameter and over 2,000 mm long
to permit the casting of this width strip down to 6.0 mm thickness although
production had been standardised at 8.0 mm thickness for the casting
witnessed. The large diameter rolls and efficient internal water cooling
permitted casting speeds of 1.4m per minute or a weight of 2.7 tonnes per
hour to be achieved when casting 1200 alloy 1,500 mm wide by 8.0 mm thick.
Metal solidification in the roll gap was extremely rapid (10 mm per second)
giving rise to a super saturated solid solution of the alloy concerned and
different property characteristics from those of the same alloy produced
by the DC casting.
The metal solidified prior to the minimum roll gap and a hot reduction
of up to 40$ was possible before the strip left the caster. The rolls and
caster, therefore, were designed to permit this deformation over a wide
range of alloys.
Lubrication of the rolls during casting w a s brought about by two
traversing lubricators which automatically sprayed graphite suspended in
water along the roll length.
The issuing cast strip controlled to close thickness tolerances was at
a temperature of approximately 380 C and passed through an upcut shear (for
shearing the required coil weight) to a tight drum upcoiler. Coils are
removed before the leading edge of the next coil reaches the coiler.
All of the five products selected are being regularly produced using
this method, but as the alloy content of the alloys increases, the casting
speeds have to be reduced. The wider solidification range of 5251 alloy
containing 2.25$ magnesium necessitates using 50$ of the casting speed of
1200 alloy of low alloy content.
2.3 Spray castingrolling process

The wide solidification range of 2024 alloy would reduce the casting
speed of the roll caster process to an unacceptable level. In view of this
the spray casting process, although not used for producing 2024 clad
material on a production basis, shows considerable potential.
A processing method using the spray casting technique was, therefore,
formulated to give an end product of 2024 alloy clad on each side with
99.5$ purity alloy, 1.25 metres wide by 2.0 mm thick. . This product could
then be used as reroll for producing 1.0 mm thick T3 sheet.
A notional diagram of the plant layout envisaged is given in Figure 4.
The intention would be to first spray cladding on the belt to give 5$ of
the total thickness and then to spray core 2024 alloy followed by a further
cladding layer of 5$ to give a total thickness of 2 mm. Such an arrange
ment could be fed by two 7.5 tonne electric channel furnaces working
alternatively and capable of tilting to deliver a constant rate of metal.
These furnaces would be fed by an oil fired 20 tonne reverberatory furnace.
A one tonne coreless induction furnace could provide sufficient material
for the 5$ cladding required and this furnace would feed two hold and four
electric resistance furnaces of 0.5 tonne capacity.
For the width of material 1.25 metres and deposition rate of 5 tonnes
per hour a set of three sprays for each layer will be required each spray
covering one third of the width of the coating. The coating would be
deposited on a belt passing over and driven by water cooled rollers.
Thickness will be set by the belt speed which will be 0.145 metres per
83
second for a 3 mm deposit. One major drawback with this method, is the
necessity to overspray at least 10^ of the total width to ensure uniformity
of thickness in the resulting reroll after hot rolling, giving rise to
large quantities of scrap. The strip would be over 90$ dense at this stage
prior to feeding into a hot rolling mill for consolidation and reduction to
2 mm coils.
3. Determination of yields, scrap generated and melt losses incurred

It will be seen from the description of the conventional and non
conventional routes in 2.0 for six selected products that the DC casting-
hot rolling process produces a thinner reroll product than the roll caster
method for the five alloys produced by this route. The thicker strip cast
material necessitates at least one and up to three additional cold rolling
passes with increased energy consumption, scrap arisings and melt-losses
thus partially offsetting the energy savings of eliminating scalping, pre-
heating and hot rolling of the conventional process.
After cold rolling the roll caster product to the same gauge as the
hot rolled product, subsequent processing is mainly the same for the
products selected. However, in the case of alloys that have to undergo
critical forming operations such as 3003 container foil, the metallurgical
investigation associated with this contract confirmed that it is necessary
to introduce an homogenising anneal to produce the required properties in
the roll cast product.
In the case of spray casting, the reroll 2024 clad product is 2.0 mm
in thickness compared with the 5.0 mm thickness of the hot rolled product
and therefore requires less cold mill passes to achieve the final sheet
thickness.
Complete processing schedules were compiled, therefore, for the six
products selected for both the DC casting-hot rolling and thin strip
casting methods of production. Determination of the scrap generated at
each manufacturing stage and melt losses incurred on remelting, enabled the
yields for each product by the different processing routes to be
calculated.
The yields for each of the six products by conventional and non
conventional routes for comparison purposes are given in Table I together
with the scrap generated and melt losses incurred per tonne of product
produced.
These findings applied to a typical product mix of 40,000 tonnes of
the six products selected, confirmed an improvement in yield of 75^ using
the thin strip casting routes giving a reduction in scrap and melt losses
of 7230 tonnes per annum.
4. Determination of the energy input at each stage of manufacture

4.1 DC casting-hot rolling route
Using improved metering equipment for oil, gas and electricity
consumption at the smelter and rolling mill plant a comprehensive energy
audit of all the processes used in the production of the six selected
products was carried out.
The work involved recording the actual energy consumption as the
products concerned were being processed through the individual items of
plant in order to obtain the energy expended per tonne of product produced.
It was then possible to determine from these figures, using the
previously calculated yields for each of the products concerned, the total
process energy per tonne of end product produced.
In addition, of course, the energy associated with irrecoverable metal
losses and the induction melting of swarf and scalpings as a separate
operation also had to be included in the total energy audit.
The energy required to replace metal irretrievably lost was reassessed
with the experience of modern smelter technology.
The energy requirement to produce one tonne of liquid aluminium from
bauxite ore was assessed to be 243 GJ per tonne instead of the figure of
250 GJ per tonne previously quoted.
From the above information, the total energy requirement in GJ per
tonne of each product produced was calculated and included in Table II for
comparison with the thin strip casting process. The process energy in this
table includes the increased energy requirement when scalpings and swarf
are double melted.
4.2 Thin strip casting routes

(a) Boll caster route - This route was used for the following products:-
1200 soft temper foilstock, 3003 H24 temper foilstock, 1200 H H sheet,
3IO3 H18 profiled sheet and 5251 H34 sheet. Discussions with the equipment
manufacturers which included both Hunter and Seal Engineering together with
the purchase and processing of three continuous strip cast coils to foil-
stock, permitted the determination of the energy requirements of this method
of processing for each of the products concerned.
The energy of the additional cold rolling passes necessitated by the
thicker gauge of the rolled cast products compared with the hot rolled
thickness of the conventional process was also calculated. In the case of
3OO3 alloy it was also necessary to include the energy of an homogenising
anneal of the coil after roll casting to prevent coarse grain developing
on subsequent annealing. For the same reason it was not possible to anneal
1200 material until it had received at least 30$ cold work after casting.
The process energy and metal loss replacement energy associated with
the production of the five products selected, were calculated and included
in Table I for comparison with the DC casting-hot rolling route.
It will be seen that the reduction in energy compared with the conven-
tional route extended from 27 to over 37$ for all the products except 3003
which only gave an 8.12$ improvement. The reason for this was the
inclusion of an homogenising anneal to enable the. metallurgical require-
ments to be met which offset the gain obtained by eliminating the block
homogenising of the conventional process.
(b) Spray casting-rolling route - This method of processing has not been
used on a production basis for producing 2024 clad sheet. It was felt,
however, that it's ability to thin strip cast alloys having a wide
solidification range and the possibility of spray casting the cladding,
should be investigated in view of the considerable energy consumption of
the conventional method of producing this product.
In view of the considerable experience of Swansea University college
staff in the use of the spray cast process, particularly Professor A R E
Singer, who developed the method, it was agreed that they should carry out
an energy audit of the hypothetical process advanced by the writer. This
was satisfactorily completed and enabled the energy consumption of using
this method for the production of 2024 clad T3 sheet to be calculated for
comparison with that of the DC casting-hot rolling method.
The large reduction in process energy of approximately 30$, however,
was considerably offset and reduced to 16.1$ by the large metal loss
included in the energy audit. This was due primarily tc the metal loss
associated with 10$ overspray which also increased the scrap generated. In
addition, the melt loss of the melting and holding operation also seemed
excessive at 5$ and 2$ respectively for 2024 and 99.5$ purity alloys,
particularly in view of the fact that induction furnaces were used for
-85-
holding and melting of cladding material. Development of the process
should enable a considerable reduction in these losses to be achieved with
further energy savings.
4.3 Energy savings applied to a typical product mix

The energy savings per tonne of product produced were applied to a
typical rolling mill product mix of 40,000 tonnes per annum. It was
established that the use of the thin strip casting routes could give rise
to a total potential energy saving of approximately 23$ or 3 million therms
per annum compared with the DC casting-hot rolling route, (see Table III)
5. Conclusions
The main objective of this work was to establish a less energy
intensive process than that of DC casting-hot rolling for the conversion of
liquid aluminium to semi-fabricated rolled products. The thin strip cast-
ing technique achieves this aim and the potential energy savings together
with other significant findings are given below.
5.1 The roll caster thin strip casting process is established for the
production of 1200 scft temper and 3003 H24 temper foilstock coil, 1200
H14 temper sheet, 3103 H18 profiled sheet and 5251 H34 sheet.
5.2 The spray casting-rolling process is proposed for the processing of
2024 T3 clad material but has not yet been developed to the production
stage for this product. The process has considerable potential for energy
saving, however, and is well worthwhile further development.
5.3 Thin strip casting and subsequent conventional processing can give
rise to substantial energy savings of up to 37$ compared with the conven-
tional DC casting-hot rolling route for the products examined.
5.4 The main factors contributing to this reduction in energy are (i) the
elimination of the scalping, preheating and hot rolling processes and (ii)
the improved yields and subsequent reduced scrap remelting, giving rise to
lower metal losses with corresponding metal replacement energy savings.
5.5 The extent of the energy savings using the thin strip casting process
is reduced by the additional cold rolling passes required by some of the
products to achieve the hot rolled thickness of the conventional process.
In addition, the introduction of continuous strip cast coil homogenising
to overcome the metallurgical effects of rapid solidification and achieve
the required properties in the final 3003 H24 foilstock, considerably re-
duced the potential energy savings of this product.
5.6 Future work should be aimed at developing the thin strip casting
processes to produce thinner, say 2.5 mm cast coil at a faster rate for
the products concerned. In addition, alloys should be developed to take
advantage of rapid solidification, giving a continuous strip cast
structure suitable for subsequent processing without the necessity for
high temperature homogenising anneals.
5.7 Metal losses, although considerably reduced by the roll caster route,
are still high due to oxidation on remelting scrap from subsequent
processes. An investigation into means of reducing these losses in view
of the metal replacement energy requirement of 243 GJ per tonne (revised
figure) from the bauxite ore, would appear to be well worthwhile.
5.8 The energy savings of the thin strip casting route when applied to a
typical rolled products mix of 40,000 tonnes per annum resu]ted in a
reduction in energy of approximately 25$ or 3 million therms per annum
equivalent to 315 517 GJ per annum.
-86-
TABLE I COMPARISON OF SCRAP GENERATED. METAL LOSS INCURRED AND YIELDS USING THE CONTINUOUS STRIP CAST AND
CONVENTIONAL DC CASTING/HOT ROLLING ROUTES
DC/Hot Mill Route Thin Strip Cast Route

Product Scrap generated Melt loss Yield Scrap generated Melt loss Yield
Tonnes Tonnes % Tonnes Tonnes %
1200 Soft Temper Foilstock O.3O5 0=014 75.81 0.2578 0.0129 78.70
3OO3 H24 Temper Foilstock 0.406 0.017 70.35 O.3429 0.0143 3.69
1200 H H Temper Sheet 0.472 0.016 67.21 0.3568 O.OI42 72.94
3103 H18 Profiled Sheet 0.600 0.018 61.80 O.5O45 0.0167 65.74
5251 H34 Temper Sheet O.72I 0.035 56.94 0.4590 O.O25O 67.37
2024 T3 Clad Sheet 2.207 0.089 30.67 1.26 0.128 44.25
N.B. The scrap generated and metal losses are given in tonnes per tonne of the product produced.
TABLE II COMPARISON OF THE ENERGY REQUIREMENTS OF DC CASTING/HOT ROLLING AND THIN STRIP CASTING ROUTES
DC/Hot Rolling Process Thin Strip Casting Reduction % Reduction

Product Gj/Tonne Product Gj/Tonn in Energy
Gj/Tonne Product
I
Melt Melt
Process Loss Total Process Loss Total
1200 Soft Temper Foilstock 15.048 3.402 18.450 8.367 3.135 11.502 6.948 37.66
3003 H34 Temper Foilstock 18.239 4.131 22.370 17.079 3.475 20.554 1.816 8.12
1200 H H Temper Sheet 16.768 3.888 20.656 10.178 3.451 13.629 7.027 34.02
3103 H18 Profiled Sheet 17.469 4.374 21.843 1.694 4.058 15.752 6.091 27.89
5251 H34 Temper Sheet 28.337 8.505 36.842 18.839 6.075 24.914 11.928 32.38
2024 Cl- d T3 Sheet 86.790 21.627 108.417 59.927 31.000 90.927 17.49 16.13
N.B. 1 KWH = 0.012 GJ (assumes generating efficiency of 3fi)

TABLE III COMPA RISON OF TOTAL ENERGY REQUIREMENTS OF DC CASTING/HOT ROLLING AND THIN STRIP CASTING
ROUTES FOR AN ANNUAL OUTPUT OF 40.000 TONNES OF SEMIFABRICATED ROLLED PRODUCTS
Tonnes Gj/Tonne Gj/Annum Saving

per DC/ Thin DC/ Thin GJ
Product Hot Strip Stri) per
Hot
Annum Roll Cast Roll Cast Annum
1200 Soft Temper Foilstock 15 000 18450 11.502 276 750 172 530 104 220 37.66
3003 H24 Temper Foilstock 5 000 22.370 20.554 111 850 102 770 9 080 8.12
1200 H U Temper Sheet 6 000 20.656 13=629 123 936 81 774 42 162 34.02
3103 H18 Temper Profiled Sheet 5 000 21.843 15.752 109 215 78 760 30 455 27.89
5251 H34 Temper Sheet 5 000 36.842 24.914 184 210 124 570 59 640 32.38
2024 Clad T3 Sheet 4 000 108.417 90.927 433 668 363 708 69 960 16.13
Grand Total 40 000 30.99 23.10 1 239 629 924 112 315 517 2545
N.B. 1 KWH = 0.012 GJ (assumes generating efficiency of 30$)

DC.Casting/ Hot Rolling process
DC PRE- BD END EDGE TANDEM
SCALP HOMOGENISE HOT
CAST HEAT ROLL | SHEAR TRIM HOT ROLL
500 L ?2_ 11
COIL
AS
EROLL
8
THIN
STRIP SHEAR
CAST
Thin Strip Casting process
Flow diagram comparison of the conventional D.C. casting/hot rolling and

thin strip casting routes
The numbers indicate typical metal thickness in mm
Figure 1
Figure 2
-90-
Continuous Strip Casting
10-12 Rolls
14 Feed Nozzle
20 Liquid Metal
24 Cast Strip
Li Distance between Nozzle and Roll Gap
T1 Roll Gap
Figure 3
COATING TUNDISH
CORE TUNDISH
COATING TUNDISH
TRPLE ATOMISERS
SEALED CHAMBER
PROTECTIVE
ATMOSPHERE SNOJT
CONTINUOUS BAND
TEMPORARY SUBSTRATE
STRIP PASSUNE
Notional layout of a 3 layer Spray Rolling installation.
Figure 4
-91 -
COMPOSITE METAL JOINTING
Author : P. McGEEHIN
Contract number EE-C-4-261-UK
Duration : 24 months 1 July 1981 - 30 June 1983
Head of project P. IfcGEEHIN
Contractor : AERE Harwell
Address : Materials Engineering Centre

AERE Harwell
Didcot
OxDn
UK
SUMMARY
The Composite-Metal Jointing Working Party is jointly funded by

the Commission of the European Communities and a group of 15
companies. The aim of the work is to develop proven design
procedures for structural joints between composite-materials
and metals, with emphasis on adhesive bonding techniques.
The joint geometries and materials being investigated have
been chosen because of their applications potential in the
automotive and aerospace industries. To date, design computer
programmes have been written and a large programme of materials
testing commenced. Test data are currently being used to design
the specified joints, and a programme of manufacturing is under
way which will allow the performance of component joints to be
subsequently established, and the design procedures validated.
-92-
1. INTRODUCTION
The Composite/Metal Jointing Working Party has been formed under the
aegis of the European Commission's Energy Conservation R & D Subprogramme.
The overall objective of the work is to give vehicle and aircraft designers
greater confidence to use light .and structurally efficient composite
materials; this will lead to a reduction in weight and a consequent decrease
in transport fuel consumption.
The Commission is providing 50% funding, and the balance is obtained

from the members of the Working Party, who each pay the same membership fee.
There are presently 15 members of the Working Party, summarised in Table 1.
As can be seen, major European companies in .the aerospace, automotive and
raw material supply sectors are represented, together with two UK Government
Departments (Industry and Defence).
Technically, the aim of the work can be expressed quite simply. It is

to develop proven design procedures for structural joints between composites
and metals, giving emphasis to the use of structural adhesives. As well be
described in Section 2, joint geometries and all the materials have been
chosen to reflect specific potential applications for composite/metal joints
in vehicles. There are three major facets of the technical work, which can
be summarised as follows:
(i) Materials testing, and in particular testing the adhesives,

statically, In fatigue and as a function of environmental ageing
(time, temperature and humidity). This will provide data for:
(li) Engineering design studies on the chosen joints after continuum

mechanics computer programmes have been written and validated
against the literature and using finite element methods.
(iii) Testing of designed joints after manufacture, in order to

establish the general principles which underlie the whole
programme, and indicate appropriate failure criteria which might
be used in practice.
In addition, a programme of non-destructive inspection of joints is

planned.
The project work is undertaken by the Materials Engineering Centre at

Harwell on behalf of the members, and it is managed by four-monthly meetings
at which progress (summarised in a previously circulated report) and
technical difficulties are discussed, and future actions agreed.
2. COMPONENTS AND THEIR SPECIFICATIONS
The work is split into two, roughly equal halves. In the first part,
torsional joints are being considered, and in the second part two different
types of bracket joint are being investigated.
2.1 Torsional Joints
2.1.1 Introduction
The spectrum of composite-metal torsional joints which may be needed

ranges from small driveshafts or torsion bars of ~25 mm diameter and
-93-
~ 300 mm length to truck propshafts of ~ 100 mm diameter and several metres
in length, as well as some applications in the aerospace field, e.g. for
helicopter driveshafts. In order to cover the spectrum of sizes and
materials likely to be used in these applications two model components have
been selected. The first is a small car front wheel driveshaft where the
emphasis on design is torque transfer over a short distance within space
constraints and at fairly elevated temperatures. The second is a truck
propshaft where the emphasis is on vibrational modes and critical length
problems where composites have weight advantages and in some cases can
eliminate heavy and expensive bearings.
Specifications have been provided by Working Party members for a wide

range of driveshafts, particularly for front-wheel-drive car applications.
A study of these revealed that the maximum load conditions placed fairly
high stresses on the existing metal components although these load
conditions were infrequent. The dynamic nature of the loading and the
freedom of the shaft to accelerate, however, would be likely to cause a
reduction in these load conditions. In addition the steel materials often
have a toughened outer layer or prestressed compression on the outer surface
which would tend to inhibit the propagation of fatigue cracks. The
application of composite materials to these components needs to be similarly
optimised and take account of the true loading spectrum including thermal
and minor secondary stresses.
The specification for the components cover the following parameters:
Loading
Fatigue spectrum
Minimum whirling frequency
Length
Constraint on diameter
Service temperatures.
2.1.2 Front Wheel Driveshaft
A typical set of design requirements for a front wheel driveshaft of a

small saloon car has been taken as:
(i) Maximum torque 100 Nm

(ii) Cycled torque (10 6 Hz) 400 Nm
(iii) Whirling frequency Above 150 Hz
(iv) Length 620 mm
(v) Constraint on diameter 40 mm
(vi) Temperature range -40 to 160 Deg C.
The design values for a typical steel driveshaft to meet this specification
are given below:
Diameter = 28 mm
Max. shear stress = 232 MPa
Cycled shear stress (1 MHz) = 93 MPa
Whirling frequency = 148 Hz
Torsional stiffness = 138 Nm/Deg
Weight = 3.0 kg
The metal end fitting for the driveshaft component is conceived as a

constant velocity knuckle couple as shown in Fig. 1 made from a steel which
is used for forging of shafts and is widely available in many sizes. A
design for this component in composite materials would use a laminate with
fibres oriented at +45 to maximise torsional strength and stiffness. Since
the component would be fabricated most easily in tubular form the two
dimensional variables of thickness and outside diameter must be selected.
These two variables are limited by (i) the constraint on the maximum
diameter of torsional buckling, which will occur if the tube is too thin,
and (ii) the strength requirement. A design in continuous glass fibre
exceeds the limitation on diameter, requiring an outside diameter of 45 mm.
A design in high strength carbon fibre however can meet this limitation.
The optimisation of design allows failure due to shear stress to occur just
before failure due to buckling. The critical whirling speed should also be
kept above 150 Hz. Such a design optimised for minimum weight to meet the
above has been obtained.
2.1.3 Truck Propshaft
In order to consider a wide range of torsional joints a truck propshaft

was considered as a potential application for composites which provides a
different emphasis on material requirements and scale. The design
requirements are typically:
(i) Maximum torque 4932 Nm

(ii) Cycled torque (10 6 Hz) 1973 Nm
(iii) Critical whirling frequency Above 50 Hz
(iv) Length 1.36 M
() Constraint on diameter 85 mm
(vi) Temperature range -40 to 50 Deg C
The specification of a steel tubular propshaft which meets this requirement

is:
Diameter = 76.2 mm
Thickness = 3.05 mm
Max. shear stress = 200 MPa
Cycled shear stress (1 MHz) = 80 MPa
Buckling stress = 974 MPa
Critical whirling frequency = 114 Hz
Torsional Stiffness = 976 Nm/Deg
Weight = 7.48 kg
The metal end fitting for the propshaft component is conceived as a

coupling as shown in Fig. 2 made from an aluminium alloy which is used
widely in the automotive industry for high strength casting and which has
good corrosion resistance.
The design of a composite shaft to meet these requirements at the

lowest material cost whilst still giving substantial weight savings uses a
glass fibre composite wound in a tube at laminate angles of + 45. This
takes the torque loading adequately but cannot achieve the required critical
whirling frequency. However the addition of some axial carbon fibre
enhances the axial modulus sufficiently to allow the specification for
critical speed to be met. In this instance a volume ratio of glass to
carbon of 7:1 gives a suitable performance, but for longer shafts more
carbon would be needed. The chosen hybrid composite design gives a weight
saving of about 30% whilst meeting the required specification. This weight
saving could be further improved if a larger diameter had been allowed in
-95-
the specification.
2.2 Bracket Components
2.2.1 Introduction
The general type of composite-metal joint under consideration in this

part of the project is the application of composite materials in a bracket
type of component where a joint can take bending and shear loading. The
various joining options cover both mechanical fixing and adhesive bonding
including combinations of these two methods. For joints subjected to
relatively low stresses the per unit cost of the joining method is a
dominant design consideration but for higher stresses the structural
integrity is also of importance.
The cost of moulding inserts and the poor fatigue performance of

mechanical fixings makes this method less attractive for composites than for
metal-metal joints. On the other hand the extreme modulus mismatch between,
for example, a glass composite and steel makes the load distribution through
an adhesive bond very uneven unless due consideration is given to designing
the joint carefully. In addition the poor performance of adhesives against
peeling stresses makes the use of combined mechanical and adhesive bonding
desirable in some cases. For these reasons the bracket/top hat type of
joint forms a challenging exercise in design and a useful background for the
comparison of joining methods.
In the first type of joint being considered, a glass fibre composite

bracket fixed to a metal structural section of the type used for a truck
frame is considered. The type of loading to which this joint is subjected
is considered to the the loads expected from an eccentrically fitted mass
subjected to bump and corning accelerations. The second type of joint is
concerned with the joint between a metal hinge and a glass fibre composite
rear hatchback door. The loadings arise from distortion of the body during
severe bump loads and in addition the service loading of repeated opening
and closing of the door.
The specifications for these components cover the following

parameters:
Loading magnitudes and directions

Loading fatigue spectrum
Size constraints
Mechanical fixing locations
Form limitations from manufacturing routes
Temperature range of operation.
2.2.2 Chassis Bracket
A typical specification for a bracket found on the longitudinal

structural members of a typical truck has been selected. Such brackets are
used to mount most of the fixtures on the vehicle including the cab, engine,
suspension, fuel tank and battery.
This type of bracket is subjected to considerable peel loading due to

acceleration forces in cornering and it is therefore likely that mechanical
fixing in the form of bolting or riveting will be an additional requirement
to adhesive bonding as the latter is particularly weak for this type of
loading. Pure mechanical fastening on thesother hand has proven to be often
-96-
inadequate due to its poor fatigue performance and tendency to provide
stress concentrations. The combination of the two joining modes has proved
in the aerospace industry to perform very much better than either of the
modes used separately.
Typical loading magnitudes and directions are given in Fig. 3. These

loads are the maximum values encountered as a result of a suspended mass at
the illustrated loading point being accelerated by bump loading for the
vertical load case and cornering loading for the horizontal load case.
The frequency with which loadings of various magnitudes are applied is

illustrated by Fig. 4. This graph is a nominal representation of the
loading frequency spectrum and serves to indicate that the most severe loads
for these bracket components are many times in excess of the normal service
loads represented by the decrease in magnitude at the 1,000 cycle point.
Typical spatial limitations require the width of the joint to be 120

mm. The maximum depth of the section is also taken to be 120 mm to prevent
excessive protrusion from the side of the vehicle. The nature of the metal
structural component is such that no use can be made of the top and bottom
flanges for joining as these must remain clear. Any mechanical fastening
should be made in such a way that holes made in the metal component are near
the centre line or neutral axis to minimise any strength reduction.
A geometry as shown in Fig. 5 has been selected as a model system.

This is a bracket of top hat section made from either a woven glass laminate
or a 40% glass weighting SMC composite. The thickness of the glass fibre
composite is taken to be a maximum of 6 mm for ease of moulding and reliable
structural properties. The moulding could be profiled in depth over the
area of the joint. Full use of the available space has been made because of
the high bending loads on the bracket which made the design as large as
possible to avoid failure of the composite. The temperature range of the
extremes of the local environment in the region of the bracket is taken to
be from -40 to +120C.
2.2.3 Hatch-back Door Attachment Bracket
A weight saving application for composites is in door frames where use

use of SMC in particular is being considered. An added incentive for weight
saving is the reduction in lifting force required to open a hatchback door.
The hinge attachments for such a door can be difficult to mould and the SMC
would not perform well as a bearing surface. One option for avoiding this
problem is to use a metal hinge attached to the composite by some means.
Mechanical fastening would usually require moulded inserts in the SMC to
avoid any unsightly feature on the outside of the door. However, moulded
inserts would lead to additional cost which might be avoided by the use of
adhesive bonding.
The hinge is subjected to forces during assembly, opening and closing

operations and road usage. The latter forces arise from bump conditions
where the body shell distorts and the force is transmitted through the
hinges and the rubber seal to the body shell as the door resists deformation
in its own plane. A specification showing forces considered to be suitable
for test purposes is given in Fig. 6. The maximum loadings given occur
infrequently. However the door is obviously subjected to a high frequency
of opening and closing operations. The test specification will subject the
door to 20,000 cycles at a closing speed of 1.6m/s at the level of the lock.
The width of the hinge in the lateral direction is taken to be a
maximum of 50 mm. No protrusion to the outside of the door of any part of
the hinge assembly is permitted. The part of the door in the region of the
hinge is limited to a width of 25 mm as indicated in Fig. 7 where nominal
overall dimensions are also given. The thickness of the glass fibre
composite could be profiled slightly in the moulding process. The ribbed
sections can also be placed to suit the needs of the joint. The metal
component would ideally be made in a form compatible with a stamping and
bending method of manufacture such that the thickness of the metal hinge
attachment would be constant.
A geometry as shown in Fig. 7 has been selected as the model system.

This is part of the top section of an SMC door. A piece 100 mm long is to
be used as a sample of the structure in order to avoid making large test
pieces. This part of the door is taken to be a ribbed box section as shown
in Fig. 7. Use can be made of all of the inside of the box section and the
face toward the body shell for making the joint. The door material is SMC
with a 25% glass fibre weighting in a polyester resin with filler. The cure
temperature will be 140C. The temperature range of the extremes of the
local environment in the region of the bracket is taken to be from -30 to
+85C.
3. MATERIALS TESTING
3.1 Introduction
Major emphasis in the materials testing is centred on the adhesives.

They are being considered, from the point of view of the initial design
studies, as 'just another material' between the two adherends; their
stress-strain characteristics must be obtained for use in the engineering
design studies. In further tests, their adhesion to adherends is checked
and both the adhesive and its bonding qualities are determined as a function
of various ageing treatments.
Adhesives have been selected on an exemplary basis, with a choice being

made from the various commercially available products in each particular
class. Two different sets of adhesives were chosen to reflect the different
loading situations encountered in the two different types of joint. The
following types of adhesive are being Investigated:
one and two part epoxide

toughened acrylic
urethane
n i t r i l e phenolic film
epoxy film.
In all, ten different adhesives are being studied.
In the following sections the various tests being employed are

described in more detail. In brief they are:
thick adherend shear test to characterise the mechanical properties

of the adhesives.
lap shear tests to investigate the ageing behaviour of adhesives and

interfaces.
Boeing wedge test for accelerated environmental durability studies.
The interpetation of test results, and in particular their validity, is an

important part of the work. Concern about spurious results has led to
detailed investigations of trace contaminants left on the surface of
composites after the removal of protective peel ply materials, and on the
moisture content of SMC. The latter can migrate into curing adheslves and
give poor bond strengths.
3.2. Thick Adherend Shear Test
The thick adherend shear test is recommended for producing the shear
stress/shear strain elasto-plastic data on the adhesives which are needed in
non-linear stress analyses. A diagram of the test specimen is shown in
Figure 8. The adherend material is of secondary importance as the test
measures the shear properties of the adhesive. Thick steel adherends are
being employed and data are being obtained for each of the ten different
adhesives at -20C, room temperature, 50C, 100C and 150C. The effects of
environment are being investigated by comparing results for wet and dry
conditions at room temperature and one elevated temperature. Each test is
undertaken in triplicate.
In this test, very small displacements are measured, and the uniformity
of the dimensions of the bond line are critical to quantitative use of the
data in design. Considerable practical difficulties have been overcome in
setting up this test for the large number of samples being studied, and the
test is now yielding reliable design data, like that illustrated in Figure
9.
As can be seen from Figure 9, the regions of elastic and plastic

deformation are identified, and importantly for design, the elastic limit.
Results are regarded as completely valid only when purely adhesive failure
is observed when the specimen fractures.
3.3 Lap Shear Tests
The lap shear test is not recommended for producing quantitative design
data but is a useful test for comparing the durability of joints. The
double lap shear test has been selected as more relevant to the torsion
joints, and the single lap test (with the element of peel) for the bracket
joints (see Figure 10 for the specimen geometry). In the two parts of the
work, the metals are being tested against the CFRP/GRP hybrid (in the case
of the torsion joints) and the 40% E/glass polyester SMC for the brackets.
The various adhesives are being tested after manufacture of the

specimens, and then after ageing for periods of three and twelve months in
humid conditions (95% RH) at room temperature and one elevated temperature.
Three different surface treatments are being considered:
(i) An aerospace quality finish. For aluminium this follows the

phosphoric acid anodising treatment descibred in various Boeing
specifications; for steel the H3PO4 etch described in BS 5350:
1976 is being used;
(ii) An abraded finish obtained by vapour blasting aluminium and sand

blasting steel;
-99-
(iii) A contaminated surface, obtained by smearing an abraded eurface
with general purpose light machine oil.
In Figure 11 a typical set of data for various adhesives is given.
3.4 Boeing Wedge Test
The Boeing wedge test is being investigated because it offers the

possibility of an accelerated environmental durability test. Figure 12
shows the geometry: a wedge is driven between the adherends to produce a
crack. The stressed specimens will be exposed to an environment of 55C,
95% RH and the increase in the length of the crack measured after various
periods of time. It is reported that the test gives a good estimate of the
relative stability of joints to moisture, and it may be possible to use this
test to accelerate the bond durability of these parts of the testing
programme. Emphasis is being given to the metals as environmental
degradation is expected to be more severe with them than with the
composites. A luminium and steels will be surface treated in the same way as
for the lap shear tests, except in the case of steel where the effect of
silane coupling agents to enhance interface durability is being considered
empirically.
Typical data are shown in Figure 13. Crack length is converted to

fracture energy using conventional analysis and this is then plotted against
time. Clearly, the higher the fracture energy, the greater the loadbearing
capability of the joint.
3.5 Fatigue and Impact Peel Tests
In an extension to the programme, it is proposed to investigate

adhesive fatigue characteristics using the thick adherend test geometry. It
is hoped this will lead to a situation where the fatigue life of a given
joint can be predicted, bearing in mind the stresses and environmental
conditions. The principle is that just as the static strength of the joint
can be predicted using the continuum mechanics approach coupled with
suitable measured static properties, then both the fatigue strength and
lifetime can also be predicted using pure shear fatigue properties as input
data in design computer programs provided that no other mode of failure
occurs (e.g. substrate failure). This behaviour may not be applicable when
the joint is fatigued in the presence of moisture where interfacial
degradation between adhesive and metallic adherend is likely to occur.
It is proposed that room temperature, dry fatigue testing of one

adhesive be used initially although in principle this may be extended to
include different temperatures and environments. Results from similar and
dissimilar adherends will be compared. The procedure will be to obtain the
quasistatic shear stressstrain (f, 0 curve and to select three shear
stress levels X\, "X a n d 3 ^1 is in the elastic region, 3 in
the plastic region and X- is intermediate between the two extremes (see
Fig. 14). Positive/negative shear stress cycling (under load control) at
these stress levels will be used in order to reduce creep in the adhesive
caused by positive cycling only. The results would then be plotted as in
Fig. 15. In addition to obtaining the fN curve it is also necessary to
measure X t curves as a function of the number of cycles (N). This would be
done by selecting 2 values of at each stress level and after fatiguing
loading the material to failure. These tests will be carried out up to
10 cycles.
The impact peel test is useful in helping to anticipate how the
adhesives, and therefore joints, will behave under shock loading; the chosen
adhesives will be investigated as a function of adherend and temperature.
4. ENGINEERING DESIGN
4.1 Introduction
Finite Element Analysis is well established as a technique for

modelling stresses in elastic continua. It is also becoming a part of the
design process for highly stressed parts in the automotive field. However
the complexity of the modelling has often limited its use to only those
areas of the most critical importance. The application of Finite Element
Analysis to adhesive bonds has been accomplished successfully in certain
applications although the technique does have disadvantages for this area.
Firstly the thin bond lines need to be modelled by a large number of
elements in order to ensure accuracy which makes the analysis expensive.
Secondly for more compliant adhesives the limitations on strain of standard
FE codes are surpassed and a more complex non-linear code may need to be
used. However, the method can take 3-D effects fully into account, and
geometry is also freely variable.
In contrast, the continuum mechanics approach, which has been adopted

for the main parts of this study, is cheap and simple to use. Furthermore
programmes giving quite accurate output can be run on microcomputers. The
method has been developed in the USA, especially by L.J. Hart-Smith of the
Douglas Aircraft Corporation, and codes are avialable which form the basis
of the design codes for the two types of joint being considered in the
project.
Adhesive bonding has the advantage of not impairing the strength of the
substrate in addition to being efficient in terms of low weight and
increased stiffness. However the distribution of stresses within an
adhesively bonded joint subjected to tensile load have been shown to be
uneven along the length of the bond and there is a marked stress
concentration. This stress concentration is still more pronounced for
joints between dissimilar adherends. Therefore the calculation of the
stresses within the joint is essential in the use of design procedures to
ensure that the appropriate factor of safety may be applied.
As indicated above, the two main techniques which can be used to

predict stresses within adhesive bonds are finite element analysis and
classical analysis by continuum mechanics. For the general case of a
composite-metal joint analysis by either of these techniques should be
capable of taking into account non-linearity of the adhesive, orthotropic
adherends, shear deflection in the composite and thermal stresses. The use
of such techniques as part of the design procedure should also allow for
changing geometrical parameters. It is in this last respect that the two
techniques differ in their capabilities. Continuum mechanics codes based on
classical analysis can be set up for a particular range of problems for
rapid solution with varying dimensions within that range. Finite element
analysis however whilst being almost infinitely variable in the range of
geometries which it can tackle is time consuming with respect to even fairly
minor changes in geometry. The detailed results which it provides are used
mainly for analysis of bondline stresses on a research basis rather than as
a design method.
101
4.2 Continuum Mechanics Method
This type of analysis involves setting up a series of equations

describing an element of the bond line, leading to a differential equation
which can be solved either analytically or, for more complex problems, by
computer methods. Transfer of load across the element (tensile, bending and
shear loads) is considered, with the load lost at one boundary being
equalised by gains elsewhere. The method, in its most recent and
sophisticated form, can account for shear and peel stresses, orthotropic
adherends, some non-linear effects and thermal stresses. However not all of
these factors have been applied to all of the geometry types required for
design and there are many gaps in the solutions available, particularly for
the non-linear regime. When the Working Party was formed, solutions for the
joint geometries chosen were not available.
4.3 Typical Results
The development version of the code for non-linear analysis of profiled

torsional joints has been demonstrated on an Apple microcomputer. The
results were compared with a classical solution of the joint in Fig. 16 for
elastic analysis. The programme gave good agreement as illustrated in Fig.
17.
The effect of non-linearity of adhesive for a shear stress-strain

characteristic shown in Fig. 18 is given in Fig. 19 for increasing load. It
illustrates the need for non-linear analysis in order to predict ultimate
failure load. The effect of bondline thickness (Fig. 20) and adhesive
stiffness (Fig. 21) on shear stress distribution has been shown for the
reference joint (Fig. 16) in order to demonstrate the potential capability
for a rapid evaluation of joint parameters. It should be emphasised that
the stresses would be compared against different allowables (as obtained in
the thick adherend tests) for different bondline thicknesses and adhesive
stiffnesses during a formal design exercise.
The programme has been used to examine the stress pattern for simple
lap joints for the specifications of driveshaft and propshaft. Different
types of adhesives or adhesives at different temperatures give a very wide
range of stress-strain characteristics, as measured by the thick adherend
shear test. These differing characteristics provide a guide to the probable
performance of the adhesives in fatigue since extensive non-linearity is
known to result from mechanisms which would lead to failure in fatigue.
By performing stress analysis on the simple lap joints with elastic

adhesive properties and looking at the highest stress in the joint a curve
of points can be plotted which defines the region below which an adhesive
should be linear. This curve is illustrated in Fig. 22 for the driveshaft
and Fig. 23 for the propshaft as curve "A". The loads used for this
analysis correspond to those levels which are seen for a high number of
fatigue cycles. The same approach for non-linear analysis using the maximum
conceivable loads gives curve "B" in Figs. 22 and 23. The region between
"A" and "B" is one which the adhesive characteristics must cross if the
adhesive is to survive the application of the maximum load.
By examining these curves it is possible to consider both the

performance of various adhesives and also adhesives at different
temperatures. It should be noted that curve "B" is strictly a fuction of
the shape of the adhesive stress-strain curve but this only makes a very
marginal difference to the values and is fortunately not too significant for
design purposes. The curves show, in this illustrative example, that for
the simple geometries considered all the adhesives plotted would meet the
design criteria at room temperature. However at higher temperatures this is
no longer the case. An improved design which removes the stress
concentrations would be needed for the higher temperatures, particularly so
that elasticity is maintained beyond curve A. Also factors of safety would
have to be applied to the design.
An interesting point is that although the more compliant adhesive is of

lower strength, it gives rise to lower stress concentrations which more than
compensate in these particular joints. An equivalent effect to increased
adhesive compliance is gained from increased bond thickness and this effect
can be predicted by altering the adhesive characteristics by a scale factor
on the horizontal axis. However the maximum strain properties of the
adhesive may change with increase in bond thickness and this must be taken
into account.
Development of the second computer code which analyses bracket joints

under bending, tensile or shear loads has progressed from the theoretical
formulation of the equations to a working computer code. The present form
of the program can perform linear analysis of bracket joints under all three
types of loading. Further developments to include non-linear effects and
also stepped joints will commence after the validation of the code has been
completed.
5. QUALITY ASSURANCE AND NON-DESTRUCTIVE INSPECTION
Adhesives are sometimes difficult materials to work with, and regular

quality assurance tests are performed in order to be certain that they
remain within the manufacturer's specification. Perhaps more importantly,
inspection of joints when made, but prior to testing is crucial to the
correlation of design work with test results. Unfortunately, in the case of
composite/metal bonded joints there are no easily applied techniques.
In the case of many metal structures, and even for some non-metallic
materials, it is possible by purely non-destructive means to assess the
strength with a very high degree of confidence. There is however no such
test for the strength of adhesively bonded joints. Nevertheless since 1950
many successful aircraft and aerospace components have been produced by
adhesive bonding, and if more comprehensive tests could be developed this
would be the generally preferred technique. Honeycomb-cored materials have
been particularly successful, with both metal and composite skins. The
procedure followed in these applications Involves these main points:
(i) Quality control of adhesive resins.
(ii) Inspection of surfaces to be bonded by e.g. the Fokker

Contamination Tester. This is essentially a surface potential
probe.
(iii) A programme of destructive tests on the design adopted.
(iv) Inclusion of test coupons for peel, shear or tensile tests as

appropriate in each production batch.
(v) Design of components to optimise inspectability.
- 103-
(vi) Use of such NDT techniques as are available.
(vii) Repeated inspection after use involving fatigue and environmental

degradation by e.g. thermal and humidity cycling.
In the case of composite/metal joints problems arise because of

component dimensions and because the composite is physically similar to the
adhesive. Amongst the techniques being considered are coin tapping,
radiography (X-rays and neutrons), ultrasonics and thermography. None
appear immediately applicable, and further development will be required.
6. FUTURE WORK
Future work will be concerned with (i) execution of the materials

testing programme and interpretation of the test results; (ii) conclusion of
the joint design exercises after further development of the computer codes;
(iii) manufacture of component joints for testing, and (iv) interpretation
of the component test results and refinement of the design procedure.
7. CONCLUSIONS
The work programme and early results of the Composite/Metal Jointing

Working Party have been described in outline. It is hoped that the work
will achieve its objective, in increasing designer confidence in the use of
both structural composites and structual adhesives, leading to a reduction
in the weight of vehicles and aircraft, and consequently a reduction in
energy consumption in transport.
On a less lofty plane, it is also hoped that the work will lead to
manufacturers and users of adhesives having more quantitative understanding
of the design problems and their methods of solution, allowing them to
specify either their needs or products with greater precision.
ACKNOWLEDGEMENTS
The progress to date on the Working Party programme owes much to the
enthusiasm and ability of Richard Lee (Materials Development Division) and
Andy Patterson (Engineering Projects Division). In the former Division,
Dennis Bowen and Clive Phillips provide encouraging and expert supervision,
and Roger Davidson gives much assistance with testing methods. Bob Forgan
(EPD) contributed much to the early thinking, and Bob Rudham supervises the
EPD contribution. Bill Reynolds (NDT Centre) provides the expertise in
non-destructive inspection methods.
yoke fitting
CV. Joint
FIG. 1 COMPOS ITE DRIVES HAFT COMPONENT FIG. 2 COMPO

S ITE PROPS HAFT COMPONENT
Truck channel section
20 KN
.Loading position
S KN - ^ ' O 6 KN
-100%
I
100 1,000 10,000 100,000
FIG. 3 LOAD S ON BRACKET COMPONENT FIG. 4 FREQUENCY SPECTRUM OF LOADING ON THE BRACKET
: 2000
ZEE
250 out
Truck channel section
76mm
_Q 500 out
2000
6mm 6 mm 6mm
SECTIONAL VIEW ON A-A

Composite bracket Force applied
to door in open
or closed
position
FIG 5 - GEOMETRY OF MODEL BRACKET
500
FIG. 6 - LOADS ON HINGE ATTACHMENT

550
800
1350
<3&r
3 M 25 60
20
Moulded 25
stiffening rib
Rear windowM
I SECTION B -B
(enlarged)
All dimensions in mm.
FIG 7 - GEOMETRY OF SMC HATCHB ACK DOOR MODEL SYSTEM
0-31 dia.
O
05>
J I00625*
->- 25
' :'S:25"
10' 075
3-125*
70"
FIG 8 - MODEL HICK-ADHEREND SINGLE LAP SHEAR SPECIMEN
107-
70 r -10
- 4 0 21 50 100 150
-9
(C)
- 4 0 C Cohesive +E
21 17-7 140 10-5 14
IMftl -8
TSE
3-78 3-85 3-1 6 9 20
-7
22-3 3 1
IMPoi 60 475 3-95
TF 30 150 20-2 60 50 -6
G
0 4 5 0-15 0-07
[GRj 0-55 0-46
-4
-3
40 50 60 100
TT (7.)
FIG.9 - THICK ADHEREND SHEAR TEST-TYPICAL RESULTS
Double overlap joint
Single overlap joint
5-99
1 hole 600 dia
II O
^ i
II o.
02-53
8256
Dimensions as
shown in table 1
T~
Note: Remove sharp corners
Rates to be flat t o a edges, a n d b u r r s
tolerance of C 0 5
All dimensions are m millimetres.
FIG. 10. SPECIMENS FOR SINGLE AND DOUBLE OVERLAP JOINTS

(FROM BS 5 3 5 0 : PART C5 : 1976 ) .
108-
800
60
700
E - Etched -50
600 A - Abraded
C - Contaminated
500 - 4 0 :
S
- 400
-30
300
-t-20
200
-10
100
E A C E AC
Adhesive
FIG.11. AGED SINGLE LAP SH E A R RESULTS (3MONTH S 2 1 C / 1 0 0 / . R H ) .
Aluminium or stainless steel wedge

i-OiO-03
0-125 NOMINAL
I
Lo-25'iO
Wedged crack extension s p e c i m e n
length = crack growth after exposure
FIG.12. GEOMETRY OF TH E BOEING WEDGE TEST.

Key
Etched -
Abraded - o
Contam -
Silane - D 11
600- 18
f " 500 L
E
o 400 t
300-
o 200 0.
3
O
100-
5 10 l50 I 00 I I 150 I 200 I 250

1Day 10 Days
2 Days 3Days A Days 5 Days 6Days 1Week
Time ( h o u r s / d a y s )
FIG. 13 TYPICAL RESULTS DERIVED F ROM THE WEDGE TEST.
FIG. H . THREE SHEAR STRESS LEVELS F OR FATIQUE

EVALUATION
2 Log
FIG.15. ADHESIVE SHEAR F ATIQUE DATA.
110-
t =01 4 (101-6mm) t =0-1"
(254mm) 1 (2-54mm)
L =A (101-6mm)
t a = 0-508mm
,/
T
G = 79-3GPa
7
G=22-2GPa
(Aluminium ) (Composite)
FIG . 1 6 .
5
1
Distance along bond (mm) 101-6
FIG.17. ELASTIC ANALYSIS COMPARISON OF CONTINUUM MECH ANICS TORSION MODEL WITH
ANALYTICAL RESULTS ( A D A M S PEPPIAT).
111 -
Shear
stress (MPa)
30
I
20 30
Strain V
FIG 18 CHARACTERISTIC OF A MODEL ADHESIVE
101-6
Distance along bond ( m m . )
FIS. 19 - NON - LINEAR ANALYSIS V A R I A T I O N WITH INCREASING TORQUE LOAD
-112-
30r
t a = 0-2 mm.
t a = 0-508 mm.
20- ta = 0 75 mm.
0.
U) 10
101 6
Distance along bond (mm.)
FIG 20 - NON LINEAR ANALYSIS VARIATION OF BOND LINE TH ICKNESS
FIG 21-NON-LINEAR ANALYSIS VARIATION OF ADH ESIVE STIFFNESS
-113-
L e n g t h of b o n d = 5 0 m m
Bondline thickness = 0-3mm
Solid s t e e l inner a d h e r e n d
Adhesive c u r v e s at room temp. (21C)
X ( 'I. )
F I G . 2 2 . DESIGN CURVE -DRIVESHAFT
L e n g t h of b o n d = 7 5 m m
Bondline thickness =0-3mm
T u b u l a r a l u m i n i u m adherend 10mm t h i c k
A d h e s i v e curves at room t e m p . ( 2 1 C )
100
F I G . 2 3 . DESIGN CURVE - PROPSHAFT

Table 1
Industrial Members of the Composite/Metal Jointing Working Party
Aerospace Industry Automotive Industry Materials Suppliers
Rolls Royce British Leyland BP

Westland Helicopters Ford Shell
FIAT Ciba-Geigy
Peugeot Permabond
Renault
Volvo
Volkswagen
115-
DISCUSSION
Question - Dr Ehringer, CEC : Did you intend to do also some thermal

cycling tests at low temperatures ?
Answer Yes, but modest because of expense. Shock loading is most

important and this will be studied. It is hoped that the more
comprehensive materials test will allow a prediction of low
temperature performance.
Question - Mr Dennis, British Aluminium : In view of the fact that the

thinnest possible adhesives are desirable is it possible that
mechanical damage of adhesive before joining could give
galvanic corrosion between dissimilar metals ?
Answer The interplay of adhesive mechanical properties and glue line

thickness allows thicker bonds to be achieved which will help
in ensuring separation of metallic components which may
corrode if brought into contact.
Question - Mr Baffigi, Fiat : Which are the energy savings which you
foresee with composite materials if you consider the fact
that the cost of components made of composites is very high
and really competitive ?
Answer The energy saving from using composites is well known. The
economic advantages are less clear. The situation is different
for each component :
1. Propeller shafts - removing the centre bearing - will soon
be economically viable
2. Drive shafts are further away
3. Some hatch-back doors are commercial
4. The chassis bracket is attractive because the manufacturing
volume is intermediate (from the point of view of the car
manufacturer).
-116-
SESSION II - CEMENT AND GLASS-CERAMIC IND USTRY
Summary of Session II
Etude par traceurs des paramtres de transfert du systme

rgulant la composition chimique de l'alimentation des fours
ciments afin d'optimiser leur rendement
Echangeur thermique cyclones, mise au point d'une mthode

pour dfinir et amliorer ses performances
Study of a treatment of oxygen steelmaking slag in view of its

use by the cement industry
Ceramic materials from molten blast-furnace slags by direct

controlled cooling
117-
SUMMARY OF SESSION II
CEMENT AND GLASS-CERAMIC INDUSTRY

*)
J. Carrasse , P.A. Pilavachi
The cement, brick and ceramics industries in aLL Community countries

consume Large amounts of energy. Consequently, any energy saving in these
industries is also a significant cost saving.
The major element in production costs is the energy being consumed in a

high-temperature process. In the manufacture of cement, e.g. almost 90 %
of the energy input of the final product is consumed in the high-temperatu-
re kiln.
In this context, C E . A . (COMMISSARIAT A L'ENERGIE ATOMIQUE) is investi-

gating methods of optimizing energy consumption in a cement kiln, by
studying the mass transfer of the raw materials during preparation and
baking, together with an analysis of the kiln operating parameters.
LAFARGE COPPEE RECHERCHE and C.E.N.G. are studying the mass transfer of
gases and raw materials during their passage through the cyclone heat
exchangers.
IRSID and LAFARGE FONDU INTERNATIONAL are developing a treatment of slags

from the oxygen steel-making process with a view to using them in the
cement industry.
Finally, IMPERIAL COLLEGE proposes the use of molten blast-furnace slags

for the production of a glass-ceramic material.
Feed regulation of a cement kiln - The fluctuations encountered in the

composition of the raw material fed to a cement kiln must generally be
compensated by an over-baking which leads to energy losses. By developing
a new method of marking the raw material, C.E.A. was able to study the
efficiency of the homogenising silos and the passage of the material
through the kiln as well as to obtain a better understanding of the
operation of the crusher. In addition the mixing efficiency of the
pre-homegenising stockpiles was evaluated.
In a second step, C.E.A. will investigate the correlations between the

composition of the raw materials : the energy consumed per ton of clinker
produced, the quality of the clinker and the hazards of kiln operation.
An accurate knowledge of the mass transfer mechanism in the kiln would
also be obtained. The results of this study would be used to achieve
stabilization of the composition of the raw material; this would yield
substantial energy savings.
*) SYNTEC-Ingnirie, France
-118-
Cyclone heat exchangers ~ T ê cyclone heat exchangers are used to preheat
the feed to a rotating kiln by counter-current heat exchange between the
coLd materials before they enter the kiln and the hot exhaust gases before
they are discharged to the stack. LAFARG E COPPEE RECHERCHE and C.E.N.G .
are trying to improve the efficiency of this process. They hope to
achieve this by studying the mechanical performances of the cyclones, i.e.
the residence time of the materials, the efficiency of separation, the
pressure gradient, etc.
Steel-making slag in the cement industry ~ Slags from oxygen steel-making

have not yet found any large application; they generally contain between
1 and 2 % alumina. If the alumina content could be raised to about 10 %,
this product would have sufficient hydraulic properties to be used in the
cement industry as an additive to the Portland cement.
IRSID and LAFARG E test the methods to increase the alumina content of the
initial slags at the Dunkerque steelworks of Usinor. Alumina is added to
the slag in the form of an alumina-containing flux (camflux) in the
convector and by addition of bauxite and aluminium waste in the slag pot,
their melting being ensured by an oxygen injection. Alumina concentrations
greater than 10 % may be obtained by this technique.
The hydraulic properties of the product are however quite poor and further
studies should be carried out to obtain a better understanding of the
mechanisms involved.
Ceramic material from molten blast-furnace slags. - IMPERIAL COLLEG E in

London is developing a novel process for the production of a glass-ceramic
material "Silceram" from molten blast-furnace slags. The process involves
a direct controlled cooling from the melt instead of the two-stage cooling
and reheating treatment used for conventional glass-ceramics. A pilot
plant capable of melting a 100-kg batch has been designed and constructed,
together with casting and moulding equipment. A first series of trials
has enabLed the best conditions for casting to be established; there is
now nothing to hinder starting continuous production at the pilot-plant.
Tests to determine the properties of the material have been carried out
in the laboratory. The results obtained, particularly the information on
the rates of erosion, confirm that the new material is suitable for
industrial application. It would be used as a lining for conveyors or
pipes which carry erosive gas-entrained solids.
119-
ETUDE PAR TRACEURS DES PARAMETRES DE TRANSFERT DU SYSTEME REGULANT LA
COMPOSITION CHIMIQUE DE L'ALIMENTATION DES FOURS A CIMENTS AFIN
D'OPTIMISER LEUR RENDEMENT
Authors P.COROMPT, B.GAILLARD, J.GUIZERIX, M.LAUNAY
Contract number EE-B-I-I4I-F
Duration 30 months 1 September 198O - 30 April I983
Head of project J.GUIZERIX - SAR-LAT-Centre d'Etudes Nuclaires de

Grenoble
Contractor Commissariat l'Energie Atomique
Address CEA-Centre d'Etudes Nuclaires de Grenoble

SAR-LAT - 85 X
F - 38041 GRENOBLE CEDEX
Summary
The fluctuations encountered in the composition of the raw mate-

rials in some cement factories must generally be compensated by
an overtaking which leads to energy losses.
An optimization method of the energy consumption is investigated
by a survey of the transfers of the raw materials during their
preparation and baking, together with an analysis of the kiln
operation parameters.
The research is carried out in the cement factory at LE BOUCAU

near BAYONNE.
The development of a marking method of the raw materials enabled:
- the study of the crusher operation, the efficiency of the homo-

genising silos and the progression of the load in the kiln
- the beginning of an evaluation of the mixing capacity of the

pre-homogenising stokpiles.
The correlations existing between the energy consumption per ton

of cement clinker produced, the parameter showing the cement clin-
ker quality, the composition of the raw material and the hazards
of the kiln operation are investigated in parallel using the pro-
cess tables of the cement factory. An accurate knowledge of the
mechanisms involved in the kiln operation shoud optimize the energy
saving related to a stabilization of the composition of the raw
material.
1.1. Introduction
Dans certaines cimenteries se posent des problmes de stabilit
des compositions minralogiques et chimiques des matriaux entrant dans
la fabrication du cru. Une partie des variations de composition rencon-
tres doit tre compense par une surcuisson, source d'une consommation
supplmentaire d'nergie.(1)
Les mthodes de marquage de matriaux utilises en gnie chimique
pour dterminer les valeurs des paramtres caractristiques des trans-
ferts de matire ont t appliques l'tude des circuits de prpara-
tion du cru. Une modlisation de ces circuits, accompagne d'une iden-
tification des paramtres servant au pilotage de la cuisson du clinker
permettra de diminuer la consommation d'nergie partir la fois de
l'obtention d'une plus grande stabilit de la composition du cru et
d'une meilleure prvision des consquences d'un ala de production en
carrire au niveau de l'entre du four.
Ralise sur un cas particulier, l'tude servira la mise au point
d'une mthode gnrale d'analyse et d'amlioration des conditions de
fonctionnement des cimenteries.
1.2. Description du site d'essais

Il est apparu que la cimenterie du BOUCAU prs de BAYONNE prsen-
tait des conditions d'approvisionnement ncessitant un ajustement per-
manent de la composition du cru. Avec l'accord de la Socit des Ciments
d'ORIGNY dont dpend cette cimenterie, une tude des conditions de
transfert et de prparation des matriaux crus ainsi qu'une recherche
de l'influence de leur composition sur la conduite du four ont pu tre
entreprises.
Cette cimenterie est alimente par deux carrires, l'une de cal-
caire presque pur, l'autre de calcaire siliceux trs htrogne. Les
deux carrires sont implantes 40 kms en amont de l'usine prs d'un
affluent de l'ADOUR.
L'essentiel des matriaux est ainsi amen par barges jusqu' la
cimenterie. Un chantillonnage continu des matriaux pendant le dchar-
gement des barges permet, aprs analyse, d'ajuster en carrire la compo-
sition des barges suivantes aux besoins du tas de prhomognisation.
Mais en raison de la dure du trajet, l'ajustement dcid partir du
dversement de la Nime barge d'un tas et des prcdentes, ne peut tre
ralis que sur la barge portant le numro N+2.
Seuls quelques ajouts correctifs : pyrites de fer, alumine, etc..
sont introduits en usines sur le tas de prhomognisation. La reprise
d'un tas de prhomognisation d'environ 12.000 tonnes, son broyage
et l'homognisation de la farine obtenue par silos de 800 tonnes dans
lesquels peuvent tre introduits d'ultimes ajouts correctifs aprs ana-
lyse, permet d'alimenter le four durant 7 jours.
La production quotidienne de l'unit quipe d'un four rotatif
de 60 mtres fonctionnant en voie sche avoisine 900 tonnes de clinker.
La consommation spcifique d'nergie volue gnralement entre 85O et
1000 thermies par tonne de clinker.
1.3 Essais raliss et rsultats

Le programme de recherche comporte les 7 tapes suivantes :
- Mise au point du marquage des matriaux
- Slection des lments de circuits de prparation de cru tudier
- Dtermination des Rponses Impulsionnelles de ces lments de cir-
cuits
121-
Modlisation des transferts de matriaux
A nalyse du pilotage du four en vue d'identifier les schmas de con
duite qui amnent les surconsommations d'nergie et de les corriger
Gnralisation de la mthode en vue de son application d'autres
cimenteries
Etude conomique.
La ralisation de la premire tape a essentiellement comport:
une tude de la rsistance mcanique des roches de carrire et la re
cherche des conditions statistiques de traage :
Pour marquer les matriaux de carrire, il fallait raliser des
blocs de mortier de rsistance convenable pour simuler les roches et
tracs dans leurs masses l'aide de substances adaptes. La mise au
point de l'opration a ncessit une dtermination pralable de la r
sistance mcanique et des rductions granulomtriques successives de
chaque roche travers les concasseurs et broyeurs traverss par ces
matriaux. Voir Figure 1.
Un traage doit permettre de marquer par exemple les 12.000 tonnes
d'un tas de prhomognisation. Mais les mesures de concentration en
traceur sont gnralement effectues en aval sur des chantillons attei
gnant au plus quelques kilogrammes. Il est donc indispensable que les
matriaux tracs injects l'entre du circuit de transfert tudi
soient, aprs broyage par exemple, diviss en une quantit suffisante
de particules marques pour que chaque chantillon en contiennent un
nombre statistiquement reprsentatif. La rpartition des particules
marques dans la masse de matriaux est en effet soumise des fluctua
tions alatoires de distribution. Si le nombre moyen de particules mar
ques renfermes dans un chantillon est N, les nombres de particules
marques effectivement contenues dans une srie d'chantillons prsen
teront une distribution alatoire autour de ^ cart type y/i. Il faut
donc que soit assez grand pour que les fluctuations alatoires de
distribution ne masquent pas les dfauts de mlange des matriaux qui
doivent tre mis en vidence.
Des essais prliminaires ont permis de mettre au point :
Un traage des blocs de mortier l'aide d'un lment fluorescent,
la Rhodamine Vit.
Un traage des matriaux pulvrulents des ajouts au moyen d'un lment
chimique, le Bichromate de Sodium.
Un traage de la farine de cru au moyen d'une poudre de verre de mme
granulometrie marque l'Or et activable en racteur nuclaire.
Un traage des conglomrats issus du tas de prhomognisation par
tir de blocs de mortier marqus l'Or,convenablement concasss pour
avoir sensiblement la mme granulometrie et activs en racteur nu
claire .
La seconde tape de la recherche a permis de slectionner parmi
les circuits de transfert servant la prparation du cru :
Le circuit de broyage
Les silos d'homognisation
Le tas de prhomognisation et ses circuits d'alimentation depuis
le concasseur de carrire et les apports d'ajouts.
Au cours de la troisime tape, il a pu tre procd :
a) Une tude de la Distribution des Temps de Sjour des matriaux dans
les diffrents circuits du systme de broyage. Voir Figure 2.
b) Une tude de l'efficacit d'un silo d'homognisation en procdant
une injection instantane de traceur aprs son remplissage et au
suivi de la rpartition de ce traceur, aprs homognisation, durant
sa vidange et celle du silo de stockage correspondant. D'importantes
variations de concentration de traceur dans la farine issue de ce
silo appellent quelques remarques sur son efficacit. Voir figures
3 et 4.
c) Une tude complmentaire sur la progression de la charge en cours
de cuisson dans le four rotatif. Voir figures 5,6 et 7
d) Une injection de blocs de mortier marqus en carrire et la mesure
de la distribution du traceur au moment du dchargement des barges
en usine et lors de la reprise du tas de prhomognisation. Bien
que les mesures de concentration en traceur aient prsent quel-
ques problmes, des questions doivent tre poses quant au degr
d'homognit introduit par ce systme. Voir Figures 8,9 et 10.
1.4 Analyse et discussion des rsultats
A/ Rsultats de l'tude du Broyeur : Les abscisses des centres

de gravit des courbes traduisant l'volution au cours du temps des
taux de comptage donns par les 8 dtecteurs rpartis sur le circuit
de broyage aprs l'injection l'entre du systme des conglomrats
de ciment marqus l'Or 198 ont permis de dterminer que :
- l'essentiel de la sparation entre particules assez fines pour cons-
tituer la farine de cru et les lments devant tre renvoys vers
les chambres de broyage est effectu par le turbosparateur n 2 et
le cyclone n 2.
- 40 % du dbit de matriaux sortant des chambres de broyage y sont
renvoys.
- le temps de sjour moyen des matriaux dans le broyeur demeure voisin
de 20 minutes.
- le systme de broyage constitue galement un bon mlangeur l'chelle
de la vingtaine de tonnes de matriaux qu'il contient.
B/ Rsultats de l'tude d'un silo d'homognisation : Deux campa-
gnes d'essais ont t ralises sur le mme silo. Lors de chaque cam-
pagne, le traceur, des spherules de verre marques l'Or 198 prsentant
une courbe granulomtrique convenable, tait incorpor un demi seau
de farine de cru dvers ensuite sur le contenu du silo. Ce contenu,
85O tonnes de farine de cru, tait alors homognis par insufflation
d'air la base du silo. Puis le silo d'homognisation tait vid dans
le silo de stockage implant en dessous.
Les deux campagnes de mesures ont donn des rsultats sensiblement
voisins, bien que lors de la premire campagne l'homognisation se
soit tendue sur lh45 et la vidange sur 150 minutes, tandis que pendant
la seconde campagne ces dures se sont respectivement abaisses 1
heure et 100 minutes. Dans les deux cas c'est vers le milieu de la vi-
dange du silo qu'apparaissent des surconcentrations en traceur. Ainsi
que le montre la figure 3 prsentant les rsultats de la seconde campa-
gne, les teneurs en traceur de la farine de cru peuvent varier du simple
au double lors de la vidange.
Les rsultats des mesures effectues la sortie du silo de stoc-
kage en vidange permanente en direction du four, sont prsents figure
4. Ils mettent en vidence un important court-circuit, la farine pn-
trant dans ce silo en ressort immdiatement. D'autre part, les teneurs
123-
en traceur sont plus leves au dbut de la vidange. Ce phnomne semble
une consquence du court circuit.
Le court circuit provient de ce que la farine restant des vidanges
prcdentes demeure colle en biseau contre les parois du silo de stoc-
kage laissant apparatre un cone vide en entonnoir qui s'enfonce jus-
qu' la goulotte de sortie. La nouvelle cargaison de farine dverse
au sommet du silo ressortira immdiatement sa base. Ce n'est qu'une
fois le silo de stockage partiellement rempli qu'un certain mlange
s'effectue. La farine sortant du silo de stockage au dbut de la vidange
du silo d'homognisation provient donc quasi exclusivement du silo
d'homognisation, ce qui explique sa plus forte concentration en tra-
ceur.
Bien que la masse de farine marque dverse sur le contenu du
silo d'homognisation puisse paratre faible, un ajout atteindra en
effet toujours au moins une tonne, l'tude montre que le systme d'homo-
gnisation a quelques difficults assurer un mlange parfait. Ces
difficults semblent indpendantes de la dure de l'homognisation.
Pour mieux juger de l'efficacit de ce silo il reste dterminer num-
riquement, partir de la distribution de la figure 3, la rpartition
de concentration de l'un de ses constituants que prsenterait un lot
de farine de cru aprs incorporation d'un ajout correctif.
C/ Rsultats de l'tude du fonctionnement du four : La figure 5

prsente le schma de principe du four et l'implantation des dtec-
teurs de traceur. Deux essais successifs ont t raliss par injection
d'oxyde de Lanthane 140 pulvrulent l'entre du four. La vitesse de
rotation du four atteignait 1,7 tour par minute lors du premier essai
et 1,1 tour par minute pendant le second essai. Cette dernire vitesse
constitue pratiquement la vitesse normale d'exploitation du four.
La figure 6 prsente la distribution des dates de passages des
points remarquables de la vague de traceur devant les dtecteurs. La
courbe relative au dplacement du centre de gravit de cette vague per-
met de dterminer la rpartition dans le four des matriaux en cours
de cuisson. Cette rpartition est donne sur la figure 7 Entre l'entre
du four et l'abscisse 36 mtres, la masse de matriaux par mtre de
longueur dcroit de 500 kg 300 kg environ probablement en raison de
la dcarbonatation. Puis entre 36 et 40 mtres cette masse s'lve
brusquement prs de 20 tonnes par mtre. Il semble alors qu'en cer-
tains points 80 % de la section du four puisse tre occups par du clin-
ker en fusion.
Mais il convient surtout de remarquer que le temps moyen de sjour
des matriaux dans le four atteint 1 heure et demie. En fait les 10
% des matriaux qui transitent le plus rapidement n'y demeurent pas
plus de 45 minutes. Les 10 % qui y sjournent le plus longtemps y pas-
sent plus de 4 heures. Il semble regrettable, bien qu'il soit impossible
de prconiser un palliatif, que les temps de sjour des matriaux dans
le four soient si diffrents. Une partie des matriaux, trop rapide,
risque d'etre mal cuite, une autre partie trop lente est vraisemblable-
ment cause d'une surconsommation d'nergie sans profit.
D/ Rsultats de l'tude du pouvoir mlangeur du tas de prhomo-

gnisation : L'tude a comporte 2 phases : l'injection d'un lot de
blocs de mortier trac l'aide d'un lment fluorescent (Rhodamine
Wt) dans le circuit d'alimentation partant de la carrire pendant la
constitution d'un tas de prhomognisation et l'talement, sur ce tas
aprs son achvement, de bauxite marque au bichromate de sodium.
Les blocs de mortier marqus ont t introduits dans le concasseur
de carrire durant le remplissage d'une barge. Les prlvements effec-
tus la sortie de la station d'chantillonnage ont permis d'valuer
la distribution des matriaux provenant de ces blocs dans la cargaison
de la barge. Comme le montre la figure 8 l'essentiel de la masse de
traceur est apparu pendant le dchargement de la 350me la OOme
tonne de matriaux. Les 250 tonnes les plus fortement marques ont t
rparties en au moins 15 couches sur le tas de prhomognisation, ce
qui a d faciliter leur mlange avec l'ensemble des matriaux.
Des chantillons ont galement t prlevs la sortie du broyeur
durant toute la reprise du tas de prhomognisation. La figure 9 pr-
sente l'volution des concentrations en Rhodamine Vit observes. La me-
sure de teneur en traceur de chaque chantillon a t ralise par fluo-
romtrie de sa solution de lixiviation. L'efficacit de ces lixiviations
posant quelques problmes, les fluctuations de concentration en traceur
observes sur la figure 9 ne peuvent tre attribues avec certitude
un dfaut de pouvoir mlangeur du tas de prhomognisation.
Il en est de mme de l'volution des teneurs en chrome hexavalent
manifeste par ces mmes chantillons et prsente sur la figure 10.
Deux procds de dosage ont t utiliss : colorimetrie des solutions
de lixiviation et analyse par spectromtrie de flamme. D'importantes
distorsions apparaissent dans les rsultats.
L'tude du pouvoir mlangeur du tas de prhomognisation devra
donc tre perfectionne, en utilisant un traceur radioactif dont les
concentrations dans les chantillons peuvent tre mesures directement.
2.CONCLUSIONS
La recherche ralise a pour but la mise au point d'un procd
d'tude par traceur des transferts de matriaux de la carrire au four
en vue d'amliorer la rgularit de la composition du cru de faon
viter les surcuissons et faciliter la conduite du four pour conomiser
du combustible.
Il a t possible de raliser :
- la simulation des roches de carrire par des blocs de mortier marqus
et de la farine de cru par des billes de verre traces l'Or.
- la slection des circuits de fabrication de cru tudier.
- l'tude d'une partie des transferts de matriaux, de leur prparation
et de leur mlange.
L'tude a fait ressortir quelques problmes au niveau de l'effica-
cit des processus de mlange. Ces points sont en cours de confirma-
tion.
Pratiquement la plus grande partie des quatre premires des sept
tapes de la recherche est ralise.
La ralisation des trois dernires tapes va porter la fois sur
la recherche, partir des tableaux de marche du four, des corrlations
existant entre la consommation d'nergie par tonne de clinker produite,
les paramtres traduisant la qualit de ce clinker, la composition
chimique et minralogique du cru et les alas de conduite du four et
sur l'valuation de l'interfrence des problmes de mlange soulevs,
s'ils sont confirms, avec le pilotage du four.
BIBLIOGRAPHIE
1 - Influence des matires premires de cimenterie sur la consommation

de combustibles - Dr. GOUDA Ingnieur en Chef, FULLER Co. USA
CIMENTS, BETONS, PLATRES, CHAUX, N 3/78,712
Carrire d'ARANC OUl Cernere de UJICHF
Danien Yp r s len
Ca)cat re . Culcaire
feuillet
(3e* n i v e i '
carrire)
FIG. 1 D urets compares des chantillons de roche prlevs en carrire
FIG. 2 ETUD E BROYEUR: Schma gnral du systme

de Broyage Principe de fonctionnement et im
plantation des dtecteurs
126
impultJoM par McoAt ( P r c i s i o n d e s Mesures > O.S 2 c t r t s typt p r s )
1000
FIG. 3 Etude de l'efficacit d'un Silo d'Homognisation 2me campagne

d'Essais Evolution des taux de comptage donns dans une mme gomtrie
de mesure par des chantillons de mme poids recueillis la sortie du
Silo d'Homognisation
( P r c i s i o n des O. 2 c a r t t y p e prs)
1000-
cm d'
Si IO
200 400
FIG. 4 Etude de l'efficacit d'un Sila d'Homognisation 2me campagne

d'Essais Evolution des taux de comptage donns dans une mme gomtrie
de mesure par des chantillons de mme poids recueillis la sortie du
Silo de Stockage
FIG. 5 Etude Four rotatif Schma de principe montrant les emplacements
des dtecteurs de rayonnements gamma utiliss: A, B, ..., C Caractristiques
du Four: Longueur 60 m, Diamtre extrieur 3,85 m, Epaisseur d'acier de
la virole 25 mm, Epaisseur garnissage => 20 cm (neuf)/ 10 cm (us),
Epaisseur croutage (valeur moyenne environ 10cm)
Centro de (rvitp
de la va(ue
FIG. 6 Etude Four rotatif Essai 2 Vitesse de rotation du Four 1,1 tour
par minute. Distribution des dates de passage des points remarquables de la
vague de traceur devant les dtecteurs
128
( M ItMH )
il
funJ Lr
i " TW ' r^-
FIG. 8 Etude du pouvoir mlangeur d'un tas de prhomognisation.

Distribution au dchargement d'une barge d'une injection instantane
de matriaux marqus faite l'amont du concasseur de carrire.
0*
0 o
I fOMT ( M
FIG. 7 Etude Four rotatif Essai 2 Vitesse de

rotation du Four 1,1 tour par minute. Rpartition
des matriaux dans le Four calcule par segments de
2 mtres de longueur
Concentrations relatives
en traceur
9000 1 0 0 0 0 tonnes de Matriaux

broys
FIG. 9 Etude du pouvoir mlangeur d'un tas de prhomognisation.

Distribution la sortie du broyeur d'une injection instantane de
matriaux marqus faite l'amont du concasseur de carrire
Concentration en Cr
(en ppm)
S
idi "TT Hos.iE.- p.ir
t \ h
.los.iis p,ir spoi'troet r i e
''[JM
Un

8000 1 0 0 0 0 tonnes de at
hroys
FIG. 10 Etude du pouvoir mlangeur d'un tas de prhomognisation,

ribution la sortie du broyeur de matriaux d'ajout marqus et aies
Distri
sur le tas de prhomognisation aprs son achvement
130
DISCUSSION
Question D r. Pilavachi, CCE : Pouvezvous donner un exemple de La faon

dont vous pourriez rectifier un mauvais mlange dans Le si Lo
d'homognisation ?
Rponse En fait Le problme devrait tre tudi avec les cimentiers,

voire avec Les concepteurs des si Los. Cependant, titre
d'exemple on pourrait envisager :
a) d'insuffler de l'air plus longtemps encore que dans le
cas presentacela ne semble pas tre La solution;
b) d'insuffler un plus fort dbit d'air pour mieux brasser
la poudre de cru. Mais La consommation d'nergie lectrique
ncessaire aux compresseurs devrait tre examine;
c) de modifier la squence d'insufflation d'air.
I l y a 2 x 4 = 8 points d'arrive d'air La base du siLo et
ils sont utiliss squentiellement selon un mode de permutation
bien dfini;
d) peuttre d'effectuer une combinaison des trois propositions
cidessus;
e) enfin peuttre de redfinir les nombres, les positions et
Les formes des entres d'air. Mais il s'agirait alors d'une
tude complte.
Question Prsident : 1. Ecoulement dans Le four : peuton reprer Les

paramtres d'exploitation au moment de La mesure qui
doivent tre pris en compte pour expliquer les irrgularits
d'coulement et La formation d'anneaux ou de boulets par
fusion : granulometrie du cru et composition; configuration
du four; conduite du brleur, forme de flamme et profil des
tempratures de flamme; nature du combustible (gaz ou
charbon) ?
2. Peuton avancer une valuation de l'influence
des diverses irrgularits mises en vidence par les
mesures sur La consommation d'nergie ?
Rponse Les deux questions devraient tre tudies avec des

cimentiers. Nous nous efforcerons de Le faire au moins en
partie dans Le reste de La prsente tude. Cependant,
titre indicatif, iL est possible de penser :
Pour la question (1) que la formation d'anneaux et de
boulets a t tudie depuis fort Longtemps par divers
cimentiers et que Le problme ne parat pas encore avoir
t rsolu.
131
Pour Le reste de La question (1), nous aLLons justement
nous efforcer d'anaLyser L'infLuence d'une partie au moins
de ces paramtres sur La consommation d'nergie dans La
suite du contrat et ne pouvons donc L'heure actueLLe
rpondre pLus prcisment.
- Pour La question (2) que La consommation d'nergie peut
tre accrue de queLque pour cent par Les irrguLarits
signaLes.
132-
ECHANGEUR THERMIQUE A CYCLONES, MISE AU POINT D'UNE METHODE
POUR DEFINIR ET AMELIORER SES PERFORMANCES
Auteurs R. MARGRITA (C.E.N.G.), SANTOS COTTIN (C...G.)

C. PAILHES (L.C.R.)
Numro du contrat EEB1106 F
Dure du contrat 18 mois : 1er Juin 1981 30 Novembre 1982
Chef de projet C. DOUVRE, LAFARGE COPPEE RECHERCHE
Contractants LAFARGE COPPEE RECHERCHE (L.C.R.) et CENTRE D'ETUDES
NUCLEAIRES de GRENOBLE (C.E.N.G.)
Adresse LAFARGE COPPEE RECHERCHE G.I.E.
B.P. 8
F. 07220 VIVIERS sur RHONE
Abstract
The study discusses cyclones situated in a preheater tower above a

cement kiln.
This preheater insures the thermal preparation of the material by a
contraflow heat exchange between the stream of cold material
descending towards the kiln and the hot gas drawn up towards the top
of the tower by fans.
The purpose is to reduce the energy consumption in the preheater
tower ; Energy can be saved by diminishing the power consumed by fans
and by a better use of the heat brought by the gas. The efficiency of
the thermal exchange is affected by the separation efficiency of the
cyclones.
However there is a lack of information about internal working para
meters of these separators (separation efficiency, pressure drop,
retention time of the material in the cyclones, etc ...) since a
suitable method of investigation was not available.
An experimental method has been perfected, using radioactive tracers
in order to investigate these parameters.
A computer programme has been designed and evaluated by the experi
mental data.
This programme can now be used to optimise the cyclone geometry for
a better separation efficiency and a lower pressure drop.
133
1.1 - Introduction
L'industrie du ciment est une grosse consommatrice d'nergie.
A titre indicatif :
- un four en voie humide consomme, pour produire 1 tonne de clinker, en-
viron 1400 thermies ( 140 kg de fuel ou 1625 kWh)
- un four en voie sche consomme, toujours pour 1 tonne de clinker, en-
viron 800 thermies ( 80 kg de fuel ou 930 kWh).
Ces dernires annes un effort important a t accompli pour passer du
procd en voie humide celui en voie sche. Cependant des amliorations
restent faire quant aux performances de ce dernier.
Une unit de fabrication en voie sche se compose de 3 parties
(cf figure 1)
- un changeur thermique cyclones, ou prchauffeur (cf figure 2)
- un four rotatif o se produit la clinkrisation
- un refroidisseur permettant de rcuprer une grande partie des calo-
ries du clinker
La recherche porte sur l'changeur thermique cyclones.
Les connaissances sont relativement limites en ce qui concerne les
paramtres de fonctionnement internes de 1'changeur (rendement de spara-
tion des cyclones, taux de recirculation et temps de sjour de la matire
aux diffrents niveaux du prchauffeur) car on ne dispose pas de mthode
d'investigation approprie.
L'accs ces paramtres s'est fait par des essais exprimentaux in
situ raliss en commun par le CENTRE D'ETUDES NUCLEAIRES de GRENOBLE
(C.E.N.G.) et le laboratoire LAFARGE COPPEE RECHERCHE.
L'unit sur laquelle les essais sont mens appartient au groupe
LAFARGE COPEE et est situe l'usine du TEIL (ARDECHE - FRANCE).
1.2 - Description des essais exprimentaux et des installations

Le prchauffeur cyclones de l'usine du TEIL comporte deux tours en
parallle composes chacune de quatre tages de cyclones (cf photo 1 ) .
Chaque tage possde un appareil sparateur sauf l'tage suprieur qui
en comporte deux identiques et en parallle, ces derniers devant assurer
un dpoussirage correct des gaz sortant de la tour.
Nos mesures n'ont port que sur une seule de ces deux tours. Elle
tait donc compose conformment la figure 2 de :
- deux cyclones suprieurs en parallle appels CIN et CIS l'tage 1
- un cyclone C2 l'tage 2
- "une boite fumes" entre la sortie du four et le carneau conduisant
les gaz l'entre du cyclone C4.
Une tour de prchauffage ou changeur cyclones permet l'change
thermique contre courant entre les gaz chauds remontant du four vers le
haut de la tour et la matire "froide", ou cru, injecte son sommet et
descendant vers le four. On appelle cru le produit de base du ciment c'est
-dire un mlange argilo-calcaire qui se transforme en clinker la suite
de son passage dans la tour d'changeurs cyclones puis dans le four rota-
tif.
Les cyclones sont l pour raliser la sparation gaz-matire chaque
tage du prchauffeur. Dans ces conditions l'idal serait que chaque cyclo-
ne effectue une sparation parfaite gaz-matire, (rendement de sparation
= 1) afin que chaque particule du cru traverse directement la tour de haut
en bas sans "revenir sur ses pas".
134-
En effet une particule, qui, un tage quelconque de la tour n'est
pas capte par le cyclone correspondant cet tage, est emporte par les
gaz ascendants. Ce grain de matire va donc repasser dans le cyclone sup-
rieur, qui est un niveau thermique infrieur, et constitue ainsi une
masse recirculante qui fait baisser le rendement de l'change thermique
dans l'changeur (cf figure 2 ) .
Dterminer le rendement de sparation d'un cyclone revient mesurer :
- le dbit de matire entrant
- le dbit de produit spar et recueilli sa base
- le dbit de produit non capt et emport par les gaz (cf figure 3 ) .
Ces mesures ont t ralises par le C.E.N.G. L'opration consiste
injecter en haut de la tour, l'endroit mme o le cru naturel est introduit
quelques grammes d'un traceur radioactif. On suit son volution travers
la tour au moyen de sondes places la base de chaque cyclone (cf figure
2 ) . Le traceur retenu pour les derniers essais d'Avril 1982 fut le cru
mme, irradi en racteur nuclaire, l'lment radioactif form dans ces
conditions tant le sodium 24 de priode 15 heures.
Les quantits irradies varient de 1 g 10 g suivant les chantillons.
Un banc d'talonnage des activits a t install sur le site exprimental
afin de pouvoir comparer entre elles les activits au moment de l'injection
et effectuer les corrections ncessaires pour que les courbes enregistres
(cf figure 4) soient comparables entre elles d'un essai l'autre.
Plusieurs chantillons de cru brut activ, c'est--dire de mme granu-
lometrie que la matire de base, ont t injects.
A chaque passage de matire radioactive devant les sondes, ces derni-
res fournissent des impulsions qui, aprs enregistrement, permettent de
tracer les courbes reprsentatives de l'activit radioactive en fonction du
temps (cf figure 4 ) .
On dduit de ces courbes les temps de passage et les dbits de matire
dans chaque cyclone.
Le spectre granulomtrique du cru normalement inject dans la tour
s'tale entre 1 \im et 300 pm. Etant donn sa diversit, on a effectu aussi
des injections d'chantillons par classes granulomtriques afin d'tudier
le comportement spcifique des particules suivant leur grosseur.
Ces classes granulomtriques sont les suivantes :
- particules infrieures 5 pm
- particules entre 40 et 50 pm
- particules entre 160 et 200 pm.
Paralllement ces essais avec traceur radioactif, le laboratoire
L.C. Rech a effectu des mesures concernant les caractristiques des gaz et
de la matire chaque tage de cyclones constituant la tour de prchauffa-
ge. Ces mesures sont destines fournir des informations complmentaires
et comparatives celles obtenues avec traceur radioactif. En effet un
modle informatique a t labor par L.C. Rech permettant de calculer le
rendement de sparation et la perte de charge d'un cyclone partir :
- des caractristiques gomtriques du cyclone
- des donnes concernant les gaz (dbit, temprature, masse volumique,
viscosit)
- des donnes sur la matire (dbit, masse volumique, granulometrie).
Les informations recueillies ont donc servi de donnes ce programme
dont les rsultats ont pu tre compars aux rsultats exprimentaux.
Mesures concernant les gaz
Les mesures de dbit gazeux ont t faites l'aval des cyclones sup-
rieurs, dans le sens d'coulement des gaz, dans une canalisation rectiligne
o le fluide est le moins charg en matire de la tour (cf figure 2 ) .
Les relevs de pressions dynamiques l'aide d'un pitot ont permis de
dterminer le dbit en Nm3/s traversant la tour.
L'analyse des gaz le long de la tour a permis de quantifier les en-
tres d'air parasite et les relevs de tempratures chaque tage de
cyclones ont servi au calcul du dbit rel en m3/s traversant chaque ap-
pareil ainsi qu' la dtermination de la masse volumique et de la visco-
sit.
Mesures concernant la matire

Le but est d'obtenir, par prlvements d'chantillons de matire en
isocintisme, le dbit et les caractristiques physiques (granulometrie,
masse volumique, composition chimique) des particules entrant et sortant
de chaque cyclone.
L'isocintisme n'a gure t possible qu' l'aval des cyclones sup-
rieurs o l'coulement est suffisamment rtabli (non tourbillonnant) pour
que les mesures soient reprsentatives. La concentration en poussires a
pu ainsi tre value la sortie du prchauffeur. Dans le reste de la
tour les prlvements ont t faits par pompage l'aide d'une canne sans
pouvoir assurer 1'isocintisme des prlvements. Nous avons tent partir
de ces chantillons et de leur analyse chimique d'tablir le bilan par
cyclone d'un traceur chimique, le K2O, contenu naturellement dans le cru.
Le bilan massique de cet lment au niveau de chaque sparateur permettait
thoriquement de retrouver le dbit de matire entrant et sortant de
l'appareil.
Le dpouillement de ces essais n'a pas donn les rsultats escompts,
tant donn la trs faible concentration en cet lment dans la matire et
sa fluctuation suivant les chantillons.
1.3 - Rsultats obtenus, exploitation et interprtation

Rsultats des mesures avec traceurs radioactifs
Les paramtres dcrivant le passage direct de la matire dans les
cyclones sont :
- t a : le temps d'arrive du traceur
t a correspond au temps mis par le traceur pour atteindre chaque pied
de cyclone depuis le top d'injection.
- t m : reprsentant le temps pour lequel les enregistrements prsentent
leur maximum (cf figure 4)
- l'amplitude du sommet des courbes exprime en choc/sec (c/s).
Le tableau I donne les valeurs du paramtre t a pour chaque cyclone de
la tour et pour les diffrentes injections ralises.
-136-
temps d'arrive t a en secondes
cyclones
classe granuloratrique de
l'chantillon inject
CIN
en \im C2 C3 C4
CIS
160 200 4,8
11,6 20,0 27,0
4,8
A0 60 4,8
11,7 18,0 26,0
4,7
infrieur 5 Pm 4,7
12,0 18,5 26,5
4,7
4,0
cru (injection 1) (1300) Pm 11,6 19,0 26,0
4,1
4,6
cru (injection 2) (1300) 11,0 19,7 26,0
4,6
TABLEAU I
Ce tableau montre qu'il n'existe pas de diffrence significative dans

les temps d'arrive, t a , du traceur au pied des cyclones pour les diff
rentes granulomtries tudies.
Les deux autres paramtres conduisent la mme constatation.
Le passage direct de la matire dans les cyclones se prsente ainsi
comme un mouvement global sans sgrgation granulomtrique significative.
Ce rsultat va dans le sens de la modlisation des phnomnes de spa
ration adopte par L.C. Rech.
En effet les concentrations de matire dans les gaz tant trs leves
(de l'ordre de 1 kg de matire par kg de gaz) des tranes importantes de
poussires se crent sur les parois des carneaux. Cela provient du fait
qu'un courant gazeux ne possde qu'un pouvoir portant limit vis vis de
la poussire. Au del d'une concentration limite une partie des particules
de matire se dpose sur les parois des conduites et est dplace sous
forme de tranes. De ce fait une quantit importantes de particules,
grosses et fines confondues est plaque sur les parois du cyclone ds
l'entre de l'appareil. Cet effet bnfique la sparation gazmatire
s'ajoute la centrifugation dans des proportions non ngligeables (on
value par le calcul 80 % la part de ce phnomne la sparation globale
dans les cyclones du prchauffeur du Teil). Cette forme de sparation doit
agir globalement sur la matire sans distinction granulomtrique contraire
ment la centrifugation, c'est ainsi que l'on explique les temps de pas
sage identiques observs pour toutes classes de particules.
Energtiquement ce phnomne a son importance car il ncessite de
faibles vitesses d'entre des gaz dans les cyclones, donc de moindres
pertes de charge. Plus la vitesse des gaz l'entre du sparateur sera
faible, plus les dp8ts de matire seront importants.
La centrifugation au contraire ncessite des vitesses importantes,
gnratrices de pertes de charge, de faon obtenir des acclrations
centrifuges suffisamment grandes pour plaquer les particules les plus
fines possible sur les parois du cyclone.
137
Nous tudierons dans ce sens, l'aide du modle, les paramtres go
mtriques susceptibles d'augmenter les performances d'un cyclone en tant
que sparateur tout en diminuant sa perte de charge.
Dtermination des dbits de matire et rendements des cyclones

Soient respectivement Qfr et Qs les dbits de matire entrant et sor
tant au pied d'un cyclone. Le rendement de l'appareil est par dfinition :
Qs 04 x< 1
A la suite d'une injection Ao, instantane, de traceur radioactif au
sommet de la tour de prchauffage, les dbits d'activit radioactive Si et
Ei mesurables au pied et l'entre de chaque cyclone sont proportionnels
aux dbits de matire et on a pour les quatre appareils les relations :
Ci g i = l , 2, 3, 4
Les entres "Ei" peuvent tre dfinies partir des relations suivan
tes :
El = g]A0 + E 2 (1 P 2 )
E2 = SX + E3 (1 3
E3 = S2 + E4 (1 - 4 >
C
E4 = S3 + X
X reprsente le dbit de matire ou d'activit remontant avec les gaz

du four et pntrant dans le cyclone infrieur C 4 (cf figure 5 ) .
A partir de ces relations, on a ainsi le rendement des cyclones :
S2
o.
r sx + s 2 + s 4
S3
o,
VJ
S2 + S3 + X - S4
S4
0
" S3 + X
S^, S2, S3, S4 et X sont fournies par les surfaces des
Les valeurs S^
courbes enregistres l'aide des sondes aau pied de chaque cyclone.
Pour les cyclones suprieurs, Cl, la dtermination du rendement aurait
ncessit l'talonnage de la sonde situe dans le circuit sortie tour, ce
qui n'a pas t ralisable (cf figure 2 ) .
C'est donc la valeur estime par les mesures du laboratoire LC Rech
qui sera donne.
Le tableau suivant donne les valeurs des rendements des cyclones d
termins suivant la mthode cidessus pour les diffrents chantillons
injects.
Classe granulomtrique Rendement des cyclones
des chantillons injects
en (>2 P3 P4
Cru 1 (1 300) 0,89 0,90 0,80
Cru 2 (1 300) 0,90 0,84
40 60 0,86 0,82
infrieur 5 ym 0,62 0,89 0,68
TABLEAU II
Les lignes "cru 1" et "cru 2" donnent les rendements des cyclones C2,
C3, C4 pour le cru comprenant son spectre granulomtrique complet.
La figure 5 donne la rpartition des masses de matire travers la
tour, dduite des essais avec traceur radioactif.
Synthse des rsultats acquis par injection de traceurs radioactifs
a) Description du passage de la matire dans les diffrents cyclones
(valeurs temporelles)
On montre en particulier que les passages directs de la matire dans
les cyclones sont indpendants de la granulometrie.
b) Quantification d'un dbit de matire en provenance du four (matire
entrane par les gaz) (cf figure 5 ) ,
c) Dtermination des dbits de matire dans les diffrentes parties
de la tour (cf figure 5)
Calcul du rendement des cyclones pour diffrentes classes granulome
triques .
Rsultats des mesures du laboratoire L.C. Recherche

Dbits gazeux et caractristiques physiques des gaz le long de la tour
Cyclones
Caractristiques
des gaz
C 1 C 2 C 3 C 4
Dbit moyen l'en
tre du cyclone en 21,37x2* 52,01 59,61 65,16
m3/s
Temprature moyenne
dans le cyclone en 410 580 720 830
C
Viscosit dynamique
3,04 105 3,56 105 3,92 105 4,20 105
en kg/m s
Masse volumique en 0,57 0,46 0,40 0,36

kg/m3
TABLEAU III
(*) Il y a deux cyclones Cl en parallle.
139
Les mesures concernant la matire
Granulometrie du cru
dimensions des
2 4 8 16 32 64 100 160 200 315
grains en
% en masse en pas
4 13 29 47 65 76 82 91 95 99
sants cumuls
TABLEAU IV
La mesure du dbit de poussires sortie tour par prlvements d'chan
tillons de matire en isocintisme dans les gaz a donn les rsultats
suivants :
concentration : 0,06 kg/Nm3
dbit gaz : 17,08 Nm3/s
dbit de poussire sortie tour : 1,02 kg/s soit (0,1 kg/kg de clinker)
A l'aide de cette valeur et du dbit de matire entrant dans les cy
clones Cl, dtermin par radioactivit, on a estim le rendement de spa
ration des cyclones suprieurs : pi = 0,95.
Rendements thoriques des cyclones dtermins partir du modle infor

matique
Deux mcanismes sont modliss pour rendre compte de la fonction
sparation du cyclone :
la centrifugation, qui met en jeu l'acclration centrifuge due la
mise en mouvement tournant des gaz dans le sparateur
les tranes de matire sur les parois du sparateur qui sont disso
cies du mouvement gazeux ds leur entre dans l'appareil du fait des con
centrations leves de particules dans les gaz.
La perte de charge due une singularit quelconque dans une conduite
est proportionnelle la pression dynamique des gaz dans une section de
rfrence donne. Le coefficient de proportionnalit caractrise la gom
trie de l'obstacle et ne varie pas avec le dbit gazeux. C'est ce coeffi
cient que nous avons modlis.
On montre que plus la gomtrie du cyclone se prte crer de grandes
vitesses tourbillonnaires dans le vortex central du cyclone, plus la perte
de charge sera importante. Or les grandes vitesses sont recherches pour
augmenter la centrifugation dans les cyclones dpoussireurs.
On retrouve ces rsultats dans le tableau suivant qui donne la perte
de charge et le rendement de sparation (thorique et exprimental) de
chaque cyclone du prchauffeur tudi.
140
rendemerit (%) perte de charge(mmCE)
Cyclone
thorique exprimental thorique exprimental
C 1 96 95 160 183
C 2 89,6 89 126 90
C 3 87,6 90 107 100
C 4 86 8085 110 95
TABLEAU V
On remarque que les cyclones Cl qui sont conus pour dpoussirer le
mieux possible les gaz ont effectivement le meilleur rendement de spara
tion mais aussi la plus grande perte de charge.
Le programme montre que leur efficacit de sparation par centrifuga
tion est une fois et demie suprieure celles des cyclones infrieurs C2,
C3, C4, mais leur perte de charge aussi.
Cependant, le rendement de sparation global des 3 cyclones infrieurs
n'est pas une fois et demie plus faible que celui des cyclones Cl. C'est le
mcanisme des dpts de matire l'entre des cyclones C2, C3, C4 qui aug
mente bnfiquement leur rendement de captation. Le programme montre que
les vitesses au niveau du vortex sont plus faibles que dans le cas des cy
clones Cl ce qui explique les pertes de charge plus faibles des 3 cyclones
infrieurs.
Conclusions
Cette tude a pour premier objectif la mise au point d'une mthode
exprimentale permettant de chiffrer le rendement d'un cyclone et ses para
mtres de fonctionnement.
Des injections d'un traceur radioactif ont permis de quantifier les
dbits de matire aux diffrents niveaux de la tour de prchauffage tandis
que des mesures plus traditionnelles donnaient les informations sur les
conditions de fonctionnement de chaque cyclone.
Un programme informatique modlisant le rendement et la perte de char
ge d'un cyclone a t crit. Les rsultats de ce modle ont t confronts
aux mesures exprimentales. L'analogie des rsultats constitue un test de
validit satisfaisant pour le programme.
La deuxime partie de l'tude consistera utiliser le modle pour
optimiser la gomtrie des cyclones de faon obtenir de meilleurs rende
ments avec de moindres pertes de charge. Il serait ainsi possible de rdui
re la consommation nergtique au niveau des ventilateurs de tirage et du
combustible brl par une meilleure prparation de la matire dans la tour
de prchauffage.
141
-
VHITC PC T>ROPUCTIOM D e CLINKER
3>'N FOUR JE C I M E N T E R I E EN VOIE SECHE
TREtOBUrFEOR A
G A Z CNRUS CYCLONES
HATIERC
SO*n.
FOU ROTATIF
TU/fRE _
RrrMwssrv*
PlGrU"R
1 IIUE
ECOULEMENT D E S S A Z HANS ON CVCLON V E R T I C A L - A
ENTREE TANGENTIELLE
EVACUATION P E S 5 * 7 ET D E S
e-HTRET TANGENTIELLE
POUSSIRES FINES HOU
(GAZ C T M A T I E R E )
COLLECTEES
SUR-VE-RSE
MOUVEMENTS
S S CONDA 1 R E S
TRATET SPIRAL E S
S VERS LE B AS
J O p e OUTUQE
VERS LE CENTRE
PLONGEUR
TRTET CH SPIRALE
ASCENDANTE SANS
LE TUBE CENTRAL
CLONE
ur
VENTILATEUR M
MOUVEMENTS
TIU*
SECONDAIRES.
m7Er
7 - m s ,
HEIUCUUTIK.
- TMIET 4 A Z .
POSITIONS JES
SCTECTEURS: T R E M I E J>e RECEPTION
4 N0IUJJ1SUDJJ3JI s D E LA I-IUTIERE
SORTIE T O U R SEPAREE
COU. M t>o*DS fto M " -O'" .o.nMTV.lo.1 O tuo mo'ioecTiF
COURBES NORMEES
Cyclone 1 Nord
Cyclone 1 Sud
Cyclone 2
Cyclone 3
Cyclone 4
10 20 3B 10 50 60 70 0 30 100 110 120 130 110 150 160 170 160 130 200
LHFHRGE Le Tell 31 " a r s I9B 2 Fl&uRE If
J>ETTS D E nnTv*e EM \**& pcfJT* Dfcv.pi"nsu*i
ESSAI N' 1 : "CRU 1"
0,85 kg/s
Pour les 2 cyclones C. (0,086 kg/kg de KK)
S. = 18,41 kg/ '
(1,86 kg/kg .KX)
A = 17,03 kg/s
(1,72 kg/kg de KK)
E, = 20,S6 kg/s
' (2,05 kg/kg d CK)
S' * 2,23 kg/s
(0,23 kg/kg de KK!
S?,= 2,2< kg/s

(0,23 kg/kg l e KK)
, = 23,54 kg/s
(2,38 kg/ko d KK) S = 18,44 kg/
' (1,87 kg/kg de KK)
S, = 21,27 kg/s
(2,15 ko/kg de 100 S ' , = 5,10 kg/s
* (0,S2 kg/kg de KK)
E = 25,14 kg/s
* (2,55 kg/kg de KK)
X = 3,87 kg/s S =20,05 k g / .

(0,39 kg/kg da KK) (2,03 kg /kg de KK)
Fi&UTiE 5
-144-
V """V
Vue gnrale du prchauffeur

cyclones et du four rotatif
l'usine du TEIL
Vue d'un cyclone dans la tour

de prchauffage
cyclone vertical entre
tangentielle spirale
mitralem "êntis.
145-
DISCUSSION
Question Mr. Bell, Imperial College : What percentage of < 5 \

tracer particles were carried out by the exhaust gases,
and what percentage passed through the tower into the kiln ?
Answer We have not quantified the percentage of parti c les<5

that were carried out by the exhaust gases but we have
detected only for these kind of particles a signal at the
exit of the tower.
Question Prsident :
1. Quelle est la relation entre le rendement thermique et le
rendement mcanique de sparation des cyclones ?
2. L'efficacit de sparation des cyclones est due 80 aux

phnomnes de dcantation, ceci ne peutil conduire chercher
optimiser sparment les changes thermiques et la sparation
gazsolides ?
Rponse 1. Globalement l'change thermique dans la tour de prchauffage

est un change contre courant entre les gaz chauds ascendant
et la matire froide descendant vers le four. La temprature
de la matire s'lve au fur et mesure qu'elle descend la
tour.
A chaque tage du prchauffeur les cyclones assurent la

sparation gazmatire afin que les gaz continuent leur
ascendance et que la matire progresse vers les tages inf
rieurs.
Cependant la sparation n'est pas parfaite et pour chaque

cyclone il existe une masse de particules qui n'est pas capte
et que les gaz emportent dans leur mouvement ascendant. Ces
particules donnent naissance un rtromlange ou recirculation
qui dgrade l'efficacit de l'change thermique du prchauffeur
compte tenu que l'on considre l'change thermique quasi
parfait entre gaz et particules chaque tage du prchauffeur.
Une tude antrieure a tabli la correspondance suivante :

Chaque point de rendement de sparation gagn par cyclone
reprsente un gain de 0,5 thermies par tonne de clinker
produite.
2. Effectivement la sparation par dpt de matire de l'entre

du cyclone est un phnomne qui ne favorise pas l'change
thermique dans l'appareil du fait que la matire chute en
vrac ce qui restreint sa surface d'change.
D'ailleurs, avant que ce mcanisme ne soit mis jour en ces

termes on pensait dj que l'change thermique se faisait en
grande partie dans les conduites menant les gaz et la matire
d'un cyclone l'autre plutt que dans les cyclones mmes.
C'est ainsi que certaines usines ont essay d'augmenter au

maximum le temps de sjour gazmatire dans Les carneaux
reliant deux cyclones entre eux.
146
STUDY OF A TREATMENT OF OXYGEN STEELMAKING SLAG
IN VIEW OF ITS USE BY THE CEMENT INDUSTRY
Authors R. PAZDEJ, 3.C. GROSJEAN, M. HAMMAN (IRSID)

C M . GEORGE (Lafarge Fondu International)
E.P. SORRENTINO (Lafarge Coppe)
Contract number EE-B-4-183-F
Duration 2<f months - 1 June 1981 - 31 May 1983
Head of project R. PAZDEJ (IRSID)
Contractors Institut de Recherches de la Siderurgie Franaise (IRSID)

Lafarge Fondu International (LFI)
Addresses IRSID
185, rue Prsident Roosevelt
78105 Saint-Germain-en-Laye Cedex (France)
LFI
157, avenue Charles de Gaulle
92521 Neuilly-sur-Seine (France)
Summary
This paper presents the results obtained up to now of an IRSID - LAFARGE

study on a treatment of oxygen steelmaking slags in view of their use by the
cement industry.
Alumina is added to the slag by Camflux additions in the converter
and by bauxite and aluminum waste in the slag pot, their melting being ensured
by an oxygen injection. Alumina contents greater than 10 % may be obtained
by this technique.
The hydraulic properties of the product are however quite poor, and
further study should be made in order to better understand the mechanisms
involved. The role of Camflux on the metallurgy and refractory wear in the
converter should also be studied before a pilot plant in an industrial setting
could be planned.
1. INTRODUC TION
Oxygen steelmaking slags from the refining of low phosphorus hot metal
have not yet found any large scale use except for the recycling in the blast
furnace. These slags contain generally between 1 and 2 % alumina. If this con
tent could be brought to about 10 %, this product should have sufficient hydraulic
properties for a use in the cement industry as an addition to the Portland cement.
The use of such a product would bring an energy gain to the cement industry
as it should decrease the fuel consumption during the klinker firing operation.
The addition of an alumina bearing product, such as some bauxite,
to the liquid slag would be a way to obtain such a product. The heat necessary
for the melting and dissolution of the alumina bearing product could be generated
by the oxidation of the divalent iron in the slag through an injection of gaseous
oxygen. On the other hand, if the temperature of the slag at the end of the
treatment is lower than the initial temperature, part of the sensible heat of
the slag would have been recovered.
However the bauxite additions are limited by the available heat. In
case the final alumina content is too low, two techniques can be tested :
to increase the available heat by adding an exothermic agent in the
slag pot as aluminum scraps for example ;
to increase the A l,. content of the initial slag by adding an alumina
containing flux (Camffux) in the converter.
A thermal balance shows that an alumina content of about 10 % can
be hoped for by using a combination of these various techniques.
2. DESC RIPTION OF THE EXPERIMENTAL SETTING

The various techniques described above have been tried in the steel
shop no.2 of USINORDunkirk.
In order to have a setting as simple as possible and, as a consequence,
to keep the research costs level as low as possible, an existing steel ladle treat
ment stand has been used for the trials. However, due to this, the possibilities
to make trials were reduced as the slag treatment could be done only when
the converter associated to the steel ladle treatment stand was not in use.
The experimental setting is shown on Fig. 1. It is mainly composed
of :
a 20 t slag pot ;
an oxygen lance. The outer diameter of the steel tube is 20 mm and
its length was 4.9 m for the first campaign and 5.4 m for the next
two campaigns because of the very rapid wear observed. It is covered
with pure alumina (Purotab, from KERFRA NCE) and the outer dia
meter of the lance is 180 mm on a length of 4 or 4.5 m ;
an oxygen line (diameter 20 27 mm) with various regulation and
control devices : 2 manual regulating valves, 1 safety electromagnetic
valve, 2 pressure gauges. The flowrate was measured with a rotameter
during the first campaign and with a sonic orifice during the next
two campaigns ;
a temperature sensor Positherm from ELECTRONITE. During the
third campaign the temperature was measured with an optical pyro
meter ;
a steel bar for slag sampling.
3. C HRONOLOGY OF THE RESEARC H

The study was planned in three main phases :
1/ Preliminary trials on a small number of slag pots in order to control
the feasibility of the process and to get an estimation of the prime
cost of the product ;
2/ Testing of the addition of Camflux in the converter. Its effect on
148-
the metallurgy and the r e f r a c t o r y wear ;
3/ Building of a pilot slag t r e a t m e n t stand in order to optimize the
process.
This paper presents only the results obtained during the f i r s t phase
of the study. During this phase, eight slag pots have been treated in three t r i a l
campaigns.
3.1. Oxygen injection (first campaign)

During the f i r s t campaign, 3 slag pots have been t r e a t e d . The t r e a t m e n t
itself consisted only of an oxygen injection in the liquid slag and the trials showed
there were no slag ejections and a temperature increase of about 100C was
observed after a 20 min. injection.
A summary of the results is given in Table I.
Table I - Oxygen injection results
Pt $ % Fe oxides % Fe3+ Temperature (eC)

% Fe oxides
initial final initial final initial final
1 17.5 25.1 29 40 1615 1740
2 20.0 20.0 25 70 1600 1680
3 15.0 17.4 37 45 1595 1630
For the pot no.2, heat was generated only by oxidation of the divalent
iron. On the other hand, for the pots no.l and 3, where the lance was immerged
deeper in the slag, the t o t a l iron oxides content increased during the t r e a t m e n t .
This may be due to an oxidation of a steel skull in the bottom of the pot.
3.2. Oxygen injection and bauxite addition (second campaign)

A bauxite (45 to 50 % A l - O j addition was made during the oxygen
injection in both pots treated during the campaign :
- pot addition of 600 kg bauxite in 12 t slag - 50 k g / t slag
- pot no.5 addition of 1400 kg bauxite in 20 t slag - 70 k g / t slag.
The bauxite was dried at IRSID before addition.
It has not been possible during this campaign to get all the data : a
sample of the i n i t i a l slag in the pot no.4 could not be taken as the slag was
solid on its surface and the temperature could not be recorded during the t r e a t -
ment of the pot no.5 as the recorder was not working properly.
The evolution of the composition and temperature of the slag in the
pot no.k is shown in Fig. 2.
A temperature drop of about 100C is observed when the bauxite is
added to the slag. The temperature at the end of the treatment (duration 30
min.) is slightly higher than the i n i t i a l one. The alumina content, very high just
after the bauxite addition (about 10 % ) , decreases and is stable by the end of
the treatment (about 4 %).
The main results for this campaign are summarized in Table I I .
149-
Table II - Results of the oxygen injection and bauxite addition trials
Bauxite % Fe ox i des %Fe3+/Fe oxides Temperature(C) % AU

Pot % added
3-
(kg / t ) i n i t i a l final initial final initial final initial final
4 50 21.1 22.6 48 46 1600 1625 3.8
5 70 23.2 24.6 33 67 1550 3.7 5.9
The heat is generated during the treatment by the oxidation of both

metallic and divalent iron. It should also be noted that the oxidation level of
the slag (% Fe 3 )Fe oxides) is very much higher when it is cooled than at the
end of the treatment (76 for both pots).
For pot no.5 where the initial A UO, content of the slag is known,
the final A I..O, content is the same as given By mass balances.
3.3. C amflux addition in the converter, oxygen injection and bauxite

addition (third campaign)
The goal of this third campaign, during which three pots were treated
(pots no.6, 7 and 8), was to obtain high final A l,, contents of the slag. So,
Camflux additions were made in the converter (Camflux is an alumina bearing
flux, whose composition is given in A nnex I) and aluminum scraps were added
in the slag pot.
The main data for the trials during the third campaign are given in
Table III.
Table III - Results of the t h i r d campaign
Pot CAMFLUX Bauxite Al scraps XFe ox ides F e 3 + / F e oxides Temperat ure (C ) Ml2 0
added 3
(kg/t slag) (kg/t slag) initial Final initial final initial final initial final
(kg/t steel)
6 0 ISO 20 16.4 19.4 32 52 1605 1590 2.8 6.0
7 S 155 15 18.B 26.8 31 57 1595 3.8 10.5
8 10 190 7.5 26.2 39.0 37 67 1615 1520 3.3 13.7
The addition of Camflux in the converter allowed an increase of the

initial A 1 ? 0_ content of the slag. For pot no.8, this increase is lower than calcu
lated because of the large stoppings observed during the oxygen blow in the
converter. The two heats made are evidently not enough to conclude on the
effects of Camflux additions on the metallurgy and refractory wear in the conver
ter. The stoppings observed show however that a harder blow is necessary in
order to keep the slag in the converter. With this harder blow, low iron oxides
contents of the slag would also be obtained with all the benefits on the iron,
manganese and aluminum yields.
During the treatment in the pot, the same observation concerning a
large increase of the iron oxides contents is made. High A 1.0, contents of
the slag ( > 10 %) have been obtained by this technique, showing its interest
for a use of steelmaking slags by the cement industry.
150-
3.4. Hydraulic properties of the alumina enriched slags
A study of the hydraulic properties of the slags was made on pot no.5
treated during the second campaign (% Al-O, = 6.85) and on three pots of the
third campaign : a non-treated pot (as a reference) % A 1 ? 0 , = 2.4 and the pot
no.6 (without Camflux addition in the converter) % AlJO., = 6 and pot no.7
(with an addition of 5 kg/t steel Camflux in the converter) % A L O , = 10.
3.4.1. Mineralogical study

The X-ray diffraction and electronic micro-analyzer analysis show that
alumina was dissolved and that it enriched first the calcium ferrite phase, with
a formation of calcium aluminoferrites (C,A. _F. ). Then, when the AlÔ,/
FeÔ, ratio gets higher than 1.2, calcium aTumiriates (C.~A 7 ) favourable to
the hydraulic properties of the product are formed. On tne other hand, two
phases containing no alumina, dicalcium silicate and magnesio-wustite, are present.
3.4.2. Ability of the slag to be crushed

The products obtained after treatment are very hard and their crushing
requires a large energy spending. The fineness had to be limited to 3500 cm 2 /g
for economic reasons, although wished finenesses were 3500, 4500 and 6000 cm 2 /g
A less fine granulometry lowers however the reactivity of the product.
3.4.3. Strength properties of the slag

As the slag contains hydraulic phases C-,5 , CÂF, C | 7 A 7 , the binding
properties of the slags only were studied with various gypsum ratios : 0, 2,
3 and 4 %. The short-term strengths of the reference samples and of no.6 samples
are practically zero and, on the long term, they are destroyed totally by inflation.
On the contrary, slag no.7 shows good strengths after 2 days (Table IV) ;
gypsum addition improves the short-term strength but this increase subsists
only slightly on the long term.
Table IV - Strength of the samples (M Pa)
REFERENCE *6 a 7
STRENGTH flexion compression flexion compression flexion compression
Xgypsum Q 0 0.2 0.8 1.6
2 0.2 0.3 0 0.4 1.5 4.0
3 0.2 0.4 0 0.4 1.8 5.8
4 0.2 0.4 0.2 0.4 2.0 5.8
Penetration strength of mixtures containing 30 % slag and 70 % Portland

cement are given in Table V.
The results in this table show that the slags are harmful to the setting.
The alumina addition improves greatly the strength properties of the mixture,
but does not compensate this harmfulness. This phenomenon was not observed
nowever for slag no.5 (% AlÔ, = 6.8), which has the same properties as a blast
furnace slag when mixed with Portland cement as shown on Fig. 3.
- 151
Table V - Penetration strengths (kg/cm 2 )
Water/Cement : 0.35 2 days 7 days 28 days

Portland cement only 82 240 370
PC + 30% inert 57 170 260
PC + 30% reference 6.4 17.1 34
PC + 30%. # 6 4.0 11.8 27.6
PC + 30% # 7 55 144 181
It seems therefore that all slags do not have the same properties and
that some of them contain harmful elements. The further studies should therefore
be aimed at finding an explanation to these phenomena and a remedy.
t. ECONOMIC CALCULATIONS
The prime cost of the products may be estimated from the data collec-
ted during the trials.
This evaluation was made for a 300.000 t/year production and it is
given for 1 t of the final product, knowing that 1 t of primary slag gives 1.2 t
of product with a 10 % Al O content.
Table VI gives an account of the hypothesis and the calculations of
the prime cost.
Table VI - Economic calculations
CONSUMPTION UNITS COST PER TON

PER TON OF SLAG OF PRODUCT
Camflux 50 kg/t 800 FF/t 33 FF/t

Bauxite 150 kg/t 150 FF/t 15 FF/t
Al scraps 0 or 10 kg/t 2000 FF/t 0 or 17 FF/t

Oxygen 20 m 3 /t . 20 FF/m3 3 FF/t
Personnel .058 h/t 50 FF/h 3 FF/t
Total Cost 58 or 75 FF/t
This prime cost may only be approximated as other costs are difficult
to know right now, for example maintenance and lance refractory costs. On
the other hand, on the gain side, lime gains and an increase of the iron, manga-
nese and steel deoxidizing agent yields is expected from the use of Camflux
in the converter.
The selling price of the product has not yet been determined as the
cement industry must get large quantities of the product in order to determine
exactly its hydraulic properties. A prime cost of 60 to 75 FF/t product looks
however reasonable and justify a further study. The trials during this study should
be made on a pilot t r e a t m e n t stand and the study should be aimed at lowering
or suppressing the Al scraps additions and the lance r e f r a c t o r y wear.
5. CONCLUS IONS
The study on the t r e a t m e n t o oxygen steelmaking slag has shown up
to now that it was technically possible to obtain A l - , , contents of the slag
in the 10 % range by combining an addition of Camflux in the converter (5 k g / t
steel) and an oxygen injection and bauxite addition in the slag pot.
The prime cost of the product obtained has been estimated to 60-75 F F / t
and justify a further study.
The product obtained after t r e a t m e n t is very hard and d i f f i c u l t to
crush. Its hydraulic properties and the strength of a mixture w i t h Portland
cement are quite poor except for slag pot no.5. Further study must be made
to understand better this phenomenon.
Chemical composition of the C A M F L U X a n d the bauxite

(%)
Prodin Fe S1O2 CaO 23 Ti02 MgO Cr203 H20

tot
CAMFLUX 11.3 1.5 37.3 36.5 1.6 7.8 0.08 0.05 0.024 -
BAUXITE 16.2 3.5 4.8 51.9 2.5 I.I 0.08 0.10 0.15 11.0
To the temperature
recorder To the oxygen main
Steel shop working (loor
Sampling floor
Oxygen lance
Steel bar
Temperature sensor
slag pot
Liquid slag
FIGURE 1 : Diagram of the oxygen injection experimental setting
-153-
Start Oxygen End
Temprature injection
CC) bauxite addition
% Fe
1
1640 -
20
Time (min)
FIGURE 2 : Pot // 4 Evolution during the treatment of the temperature,

the alumina content and the oxidation of the slag
(Oxygen flowrate : .15 m'/s)
S t r e n g t h (MPa)
100
50
2 4 7 28
T i m e (days)
F I G U R E 3 - E v o l u t i o n of the compression strength of samples made w i t h blast

f u r n a c e slag, steelmaking slag and Portland cement during t i m e
DISCUSSION
Question - Dr. Rogers, Imperial College : Calcium aluminate in Portland

cement produces a "flash set" if gypsum is absent. Does
gypsum react in the same way with the calcium aluminates and
alumino-ferites in your material ?
Answer In Portland cements gypsum reacts with C3A to give ettringite

(hydrated tricalcium aluminate calcium sulfate,
C,A - 3CaSC>4 - 32H 2 0). In the case of the steel slag, it
gives a sort of Iron containing ettringites : we know
little concerning the formation of this compound but we
expect it behaves in the same way.
Question - Mr. Giraud, Lafarge-Coppe : Sur le plan de l'conomie du

systme pour le sidrurgiste, quelle peut tre la rpercussion
sur la tenue des rfractai res, sur la consommation de
rfractaires et la priodicit des arrts pour rfection du
garnissage ?
Rponse L'utilisation de Camflux au convertisseur devrait permettre une

meilleure formation du laitier et d'obtenir les mmes rsultats
mtallurgiques avec une teneur en fer du laitier plus faible.
Ceci pourrait avoir un effet favorable sur la tenue des
rfractaires, bien qu'il soit en partie compens par la
prsence de larges quantits d'alumine dans le laitier qui
ont l'effet inverse. Le gain restera de toute faon trs
limit en terme de francs/t. En ce qui concerne la priodi-
cit des arrts, leur importance est surtout sensible en
priode de haute conjoncture.
Par contre, les gains les plus apprciables devraient se situer
au niveau des rendements fer et manganse et du niveau
d'oxydation de l'acier, conduisant une diminution de la
consommation d'agents dsoxydants.
Ces points restent dmontrer exprimentalement (suite du
programme).
155-
CERAMIC MATERIALS FROM MOLTEN BLAST-FURNACE SLAGS
BY DIRECT CONTROLLED COOLING
Authors P. S. ROGERS, J. WILLIAMSON, J. F. BELL,

M. CAMPBELL.
Contract number EE-B-4-184-GB
Duration 1st May 1981 - 31st December 1983
Head of Project Dr. P. S. Rogers
Contractor Imperial College of Science and Technology
Address Department of Metallurgy and Materials Science

Royal School of Mines
London, S.W.7 2BP.
Summary
The paper describes progress made in the development of a novel

process for the production of a glass-ceramic material
('Silceram') from molten blast-furnace slag by a heat treatment
which involves direct cooling from the melt instead of the
two-stage reheating treatment for conventional glass-ceramics.
A pilot-plant for the production of glass-ceramic castings has
been designed, constructed and commissioned. This consists
of an oil-fired tilting tank furnace capable of melting 100 kg
batches, a heat treatment kiln, casting and moulding equipment
and facilities for storing raw materials and finished product.
The viscosity of the molten mixture used for 'Silceram'
production has been determined as a function of temperature.
The data show that the fluidity under the melting conditions is
satisfactory and that the shape retention after casting is good.
Physical and mechanical property determinations have been made
in the laboratory with the new material. The resistance to
erosion by gas-borne particulate solids has also been studied
and compared with data for other materials. These investigations
have shown the new material to be attractive as an erosion-
resistant ceramic for industrial applications.
156-
1. Introduction
Glassceramics are materials which are shaped by glassforming
techniques but, by suitable heat treatment, are transformed to fine
grained ceramics of low porosity. The overall energy required for the
manufacture of glassceramics is less than that needed for the equivalent
sintered ceramics. In the conventional glassceramic process, the
components are melted together and cooled to a glass which is then
subjected to a reheating treatment; the material is held at two
successively increased temperatures for nucleation of crystals and the
development of finegrained ceramic microstructures.
At Imperial College, research has previously been conducted into the
role of transition metal oxides as nucleating agents in glassceramics.
This research has provided an understanding of the chemistry of the
nucleation and growth processes involved in controlled crystallisation
and has enabled predictions to be made for the selection of suitable
nucleating agents. A major advance has been made in the development of
a particular combination of nucleating agents which allow the conversion
of a silicate melt to glassceramic during a controlledcooling heat
treatment, removing the necessity for first quenching to a glass and then
reheating to complete the crystallisation. This novel heat treatment
programme thus provides a further saving in the energy requirement.
The silicate melt employed is a calcium oxidemagnesiaalutninasilica
mixture of suitable composition with a combination of iron oxide and
chromium oxide as nucleating agents. The mixture can be made up from
pure oxides, but is more conveniently manufactured from iron blastfurnace
slag with additions of sand, magnesite and nucleating agents. The use of
slag accelerates homogenisation during melting, and is additionally
attractive because it provides a high value outlet for a low cost
industrial byproduct. If molten slag freshly produced by the blast
furnace can be used as a constituent raw material, a substantial saving
in thermal energy would be added to the process advantages already
mentioned.
The microstructure of glassceramic materials is associated with a
number of desirable properties which are otherwise difficult to achieve.
Some SEM micrographs of polished and etched surfaces of 'Silceram'
specimens are shown in Figure 1, which are based on pureoxide slag and
slag/ash components respectively. In the case of 'Silceram', the
property particularly associated with the microstructure is resistance to
erosion by fluidborne particulate solids. This property, coupled with
the favourable tensile strength and hardness, makes the material
particularly useful under industrial conditions which lead to wear due
to erosion; examples of such conditions occur at electricity generating
stations in the transportation of coal and ash, at mines in the
transportation of coal, spoil and ores, and at steelworks in the
transportation of furnace burden.
The objective of the project is to develop a pilotplant melting and
heattreatment facility which is capable of producing glassceramic
('Silceram') by the novel process. The pilotplant is to be used to
establish the optimum melting and casting conditions for smallscale
production, and to provide samples for insitu trials at industrial sites.
Laboratory testing is also to be done to establish physical, chemical and
mechanical properties.
2.1. Description of PilotPlant

The construction of the pilotplant has been completed, and the plant
157
is now being commissioned. The site occupies an area of 1,000 square
feet, and in addition to accommodating the equipment, provides space for
the storage of raw materials and the finished glass-ceramic product
('Silceram'). Figure 2 shows the general layout of the pilot-plant, and
includes the oil fired melting furnace and the electric heat-treatment
kiln. These are shown individually in Figure 3.
The melting furnace, constructed on site, is an oil-fired tilting
tank, capable of melting 100 kilogramme batches of the raw materials with
a bath depth of 5 to 10 cm. This is to be operated continuously with
occasional pouring of the melt for specimen production. The furnace is
fired with kerosene using a single automatic burner, manufactured by
Nu-Way Eclipse Ltd. The melt temperature of 1450C is maintained by
operation with an oil/air mixture, but boosting of the air with oxygen can
be provided to assist during melt-down and to control the oxygen potential
of the melt. The design of the furnace and the installation of refractory
bricking were carried out after consultation with the refractory supplier
(Steetley Refractories Ltd.) and after compatability tests had been
performed in the laboratory. Figure 4 shows the lining construction,
which consists of a magnesia-chrome hearth and side-walls and a high-
alumina arched roof, backed with firebrick. Thermocouples have been
placed in the side-walls, roof and stack and burner quarl to monitor the
temperature during operation.
The heat-treatment kiln is a commercial model (Catterson-Smith)
obtained second-hand, repaired and reconstructed in the production area.
This has a truck hearth and an internal volume of 0.6m3 which is
sufficient for the heat-treatment of castings from a full batch of melt.
It has a maximum operating temperature of 1100C, so that the heat-treatment
temperature required by the process, 900 - 950C, is well within the
capabilities of the kiln.
The forming methods for 'Silceram' prior to heat-treatment include
casting into metal moulds and sand casting. Sand moulds have been
prepared for the casting of tiles both horizontally and vertically; the
binder used in the preparation of these moulds was sodium silicate. A
quantity of shell moulds has also been obtained. The viability of each
of these casting techniques has been confirmed using a smaller 30 kilogramme
melting furnace made available for a short period at British Steel
Corporation, Teesside.
2.2. Compositions and Raw Materials

'Silceram' can be made from melts with quite a wide range of com-
positions and can incorporate a number of different raw materials. To
date, waste materials including blast-furnace slag, minestone, coal slag
and pulverised-fuel ash have been used as major components of 'Silceram'
batches whilst still maintaining good mechanical and physical properties.
One composition, chosen as a reference due to its uniformity of structure
and superior mechanical properties and made from pure oxides, is given
below :
Component Wt %
Si02 56
CaO 20
MgO 14
A1 2 0 3 10
With 3 parts Fe 2 0 3 and 1.7 parts Na2 CrO.,
added per 100 parts as nucleating agents.
When using blast-furnace slag as a major component of the batch, a
number of different compositions have been successfully produced, one
of which is given below.
Component Wt
Blast-Furnace Slag 70
Sand 20
Magnesite 10
With 2 parts Fe 2 0 3 and 1.7 parts

Na2 CrOi, added per 100 parts as
nucleating agents.
The choice of raw materials for the pilot-plant study was made with
a number of considerations in mind, including suitability of chemical
composition (checked by test melts), ease of homogenisation and cost.
Materials eventually chosen and obtained include components for both
the slag-based batches and pure oxide test batches; these are listed
in Table I. All are uniform powders except the blast-furnace slag.
TABLE I. Raw Materials
Component Source
Blast-Furnace
<3mm (Llanwern Slag Co.)
Slag
Si02 Redhill 65 Sand (British Industrial Sand)
CaO Calcium Carbonate (Blue Circle Ltd.)
MgO Magnesium Carbonate (Steetley Minerals Ltd}
A1 2 0 3 Alumina Trihydrate (BACO)
Fe,0,
"2 " 3
Red Oxide (J. Winter & Co.)
Cr 2 0 3 Sodium Dichromate (Brit. Chrome
Chemicals Ltd.)
2.3. Physical Properties Measurement

Compressive strength has been measured on an 'Instron' testing
machine at a crosshead speed of 0.2mm" . Specimens used were 6mm diameter
cylinders cut by a diamond borer from a tile of 'Silceram'. The mean
breaking load of the eleven specimens tested was 853 MNm"2 217 MNm" 2 .
Modulus of rupture has been obtained from the breaking load in three
point bending tests with eight bar specimens of dimensions 2 4 25mm.
These were tested at room temperature on a Lloyd's J.J. tester at a
crosshead speed of 2.5mm min .
159-
Fracture toughness was determined by standard single-edge notch bend
tests with specimens 2 4 25mm and notch depth 2mm. Eight specimens
were tested, again using a crosshead speed of 2.5mm min" 1 . A value 10%
lower was obtained by the hardness indentation method.
Young's modulus was derived from measurements obtained using strain
gauges fixed horizontally and vertically on two faces of a 10 10 40mm
specimen loaded in compression on an 'Instron' testing machine at a cross-
head speed of 2.5mm min 1 .
The coefficient of linear thermal expansion was determined using a
vitreous silica dilatometer after calibration with pure silver. Specimens
were 5 5 15mm with one hemispherically shaped end. Heating and
cooling rates were approximately 100C per hour.
Thermal conductivity measurements were carried out by the Lee's disc
method from room temperature to 130C, with two specimens each
approximately 50mm diameter and 8.7mm thick.
TABLE II. Physical Properties of 'Silceram'
Density 2.9 gm cm
Porosity zero
Compressive Strength 850 M N m - 2
Modulus of Rupture 180 MNm" 2
2
Fracture Toughness (K ) 2.1 MNm
1C
Young's Modulus 1.3 10 5
Hardness (Vickers) !m"
Coefficient of linear thermal 7.5 l O ^ K - 1
expansion
(room temp, to 500C)
Thermal Conductivity
1. 76 Wm"1 -1
Coefficient of Restitution
0.910
*Ball Compression
> 40 kN
*Deep Abrasion Resistance (PREN 102)
62 mm 3
Coefficient of Friction dry 0.39
(against coal) wet 0.30
Activation Energy for Viscous Flow 194 KJ m o l - 1
Viscosity at 1350C 5.95 Nsm -2
All values determined at Imperial

College except (*) at B.C.R.A.
Viscosity () - temperature (T) characteristics of the melt were

obtained by a modified form of the falling sphere method. A molybdenum
sphere is suspended by a thin molybdenum wire from an electronic balance,
and is immersed in the melt contained in a molybdenum crucible. The bob
is moved by a motor at a fixed rate, and the drag is recorded by the
balance. The viscosity is calculated by application of Stokes' Law.
The activation energy over the temperature range 1370 to 1600C has been
calculated for the 'Silceram' melt.
The physical property data are collected in Table II and the viscosity
and thermal expansion data are shown graphically in Figure 5.
-160-
2.4. Erosion Resistance Data
One property of 'Silceram' which is of particular interest is the
resistance to the impact of small solid particles borne in a gas stream,
more commonly referred to as erosion-resistance. Samples have been
supplied to a number of organisations for testing and some of the results
obtained are shown in Table III, together with comparative results for
other materials.
TABLE III. Comparative Erosion-Resistance of Materials

Oata from E. Raask, C.E.R.L., Leatherhead)
Material Erosion Wear (mg k g - 1 )
Mild Steel 15.5

Cast Basalt 27.0
'Ni-Hard' 4.2
'Silceram' (Pure Oxide) 4.0
Alumina (75% Al 2 0 3 ) 5.3
Alumina (97.5% A1 2 0 3 ) 0.6
Test Conditions
Impacting Material : 125 to 150 quartz grains

Velocity : 27.5 m s _ 1
Angle of impaction : 45
Due to the invariance in the testing conditions of existing erosion-

test rigs, a continuous sand-blasting rig which was built at Imperial
College was used to determine the effects of different variables on the
erosion rate of 'Silceram' and other materials. One variable of
considerable importance is impact angle, and Figure 6 shows the behaviour
of typically ductile and brittle materials as well as the curve obtained
for 'Silceram'. This type of behaviour is most advantageous for
materials to be used in certain erosive conditions. A number of other
variables are also under examination, including impact velocity and
erodant flow rate. Apart from the investigation of erosion processes,
the test rig has been used to give comparative data from other materials
and will be used as a quality control test. Some data obtained from
the rig are shown in Table IV.
-161 -
TABLE IV. Comparative Erosion-Resistance
(Data from Imperial College)
Material Mass Eroded (mg) Test Conditions
'Silceram'
5.6 Impact Angle : 45
(Pure Oxide)
" Velocity : 14.5m/sec
'Silceram'
6.1
(Minestone based) Sand Flow Rate : 140g/min
'Silceram' Total Load : 820g
6.5
(Slag based)
Erodant : 100-150um Sand
Cast Basalt 17.0
'Slagsitall'
(USSR) 8.2
Recrystallised Alumina 2.3
3. Discussion and Conclusions

The construction of the pilot-plant has now been completed and the
raw materials have been acquired. The suitability of these constituents
as components of the batch feed to the melting furnace has been confirmed
in the laboratory and in brief melting and casting trials on the
20 to 30 kilogramme scale in a furnace made available by B.S.C., Teesside.
These trials also helped to establish the best melt compositions and the
best conditions for casting from a tilting tank furnace. There appears
to be no obstacle now to the commencement of a continuous production
programme at the pilot-plant.
The viscosity data obtained over a range of temperatures between
1250 and 1600C confirm that the melting conditions and the casting
procedures chosen are the most suitable. The melt has adequate fluidity
at 1450C, but has rapidly increasing viscosity as the temperature falls
to that of the heat treatment kiln (900-950C) which is necessary if the
shape is to be retained during crystallisation.
The work on physical properties has been on samples of 'Silceram'
made on the laboratory scale; these values will be compared with data
from pilot-plant samples. The information on the physical properties, and
particularly on the comparative rates of erosion, confirms the suitability
of the new material for industrial application as a lining for conveyors
and pipes carrying erosive gas-borne solids.
) SCA
b)5CPF2
c)5CF5
Figure 1: Microstructures of three "Silceram" Compositions Figure 2: General View of Pilot Plant Furnaces
) Melting Furnace and Sand Moulds.
0 10 2 0 30
F = Firebrick I 1 1
MC = Magnesia Chrome R e f r a c t o r y [ ( R ) = R a m m a b l e ]
HA = High A l u m i n a Refractory [ (C)= Castable ]
c ) Melting and H eat Treatment Furnaces
Figure A: Diagram of Melting Furnace Construction

30 60
IMPACT ANGLE
1300 1400 1500

MELT TEMPERATURE ( C )
Viscosity - Temperature Relationship for Silceram IMPACT ANGLE
Figure 5
1 1 1
-Â- Silceram
o- --O- Alumina recryslallised )
- -
IMPACT ANGLE
- -
Figure 6: Erosion against Impact Angle Plots for:
- (a) Typically Ductile Material
(b) Typically Brittle Material
(c) "Silceram" (SCF 5)
I 1 1
0 200 400 600 00 1000
TEMPERATURE ( C )
Linear Thermal Expansion of S i l c e r a m
(compared with A l u m i n a )
DISCUSSION
Question - Dr. Pilavachi, CEC : What were the main difficulties you had ?
Answer It is very difficult to develop a new idea beyond laboratory

research because an industrial concern wishes to see the
results of full scale trials before investing money, but
industrial money is normally needed for these trials.
The financial support of the Commission has enabled us to
break this loop an take the idea to the next stage towards
commercial exploitation.
SESSION III - TEXTILE INDUSTRY
Summary of Session III
Industrial application of continuous wet-on-wet treatment of

fabrics webs
Low energy preparation processes for textiles
167-
SUMMARY OF SESSION III
TEXTILE INDUSTRY
J. Carrasse, P.A. Pilavachi
A Large proportion of the energy requirements of the textiLe industry are

concentrated in the finishing sector. About 50 % of the energy consumed
per ton of fabric by this sector is used in the wet processing stages of
desizing, scouring, bleaching, dyeing, etc.
Considerable savings in energy and hence in processing costs can be
achieved. Opportunities for energy savings can be seized in different
ways, e.g. by shorter processing times, lower processing temperatures,
using less water, combined processing stages, elimination of intermediate
drying between successive wet processes and others.
The two projects on the textile industry under the Commission's R 8 D

Programme investigate such means of energy saving. The aim of the
research proposed by the NETHERLANDS ORGANISATION FOR APPLIED SCIENTIFIC
RESEARCH (T.N.O.) is to develop an industrial application of continuous
wet-on-wet treatment of fabrics, avoiding indermediate drying.
The objective of the SHIRLEY INSTITUTE project is to develop a sequence of
low-energy pathways for the preparation of cotton and polyester/cotton
fabrics.
Industrial application of continuous wet-on-wet treatment of fabric webs -

The impregnation of a fabric is often carried out on a pad. For this
process, use is made of dry fabrics. The possibility of impregnating wet
tissues directly, thus avoiding the "energy-expensive" intermediate
drying stage is being investigated by T.N.O. Three methods are being
tested in the laboratory :
Experiments on a pad : wet cloth can be impregnated on a pad, the

concentration of the soLution being adjusted. T.N.O. determined the
optimal values of this concentration as a function of impregnating
conditions : type of cloth, speed, initial water content of the cloth,
type of substance.
Experiments on engraved rollers : Satisfactory results were obtained on a

semi-technical printing machine. Initial results indicate that : the
effect of cloth speed is small; the amount of solution applied depends
on the roughness of the cloth; the pressure between the printing roller
and the cloth is an important parameter.
Impregnation by means of foam : In some textile mills, chemicals are

applied to a dry cloth by means of foam. This method could be rendered
suitable for the impregnation of wet cloth. Experiments have been
carried out by T.N.O. on a semi-technical installation constructed by
the Fibre Research Institute. Applying the foam by means of the vertical
slit method, the chemicals were applied fairly evenly across the width
of the cloth. Penetration of the chemicals into the interior of the cloth
was fairly good. The process conditions should, however, be selected
carefully.
Experience gained in the Laboratory will enable experiments and applica-
tions to be carried out in industrial plants.
Low-energy preparation processes for textiles ~ Surveys of typical

processes encountered in the industry show that the energy requirement
for desizing, scouring and bleaching is commonly between 10 ans 30 GJ
per ton of product.
It is envisaged by the SHIRLEY INSTITUTE that processing can be achieved
with a very low-energy consumption - 5 GJ per ton - by eliminating all
processes involving heating without imposing longer processing times.
Three lines of experimentation are being followed to reduce energy
requirements and still maintain short reaction times : steam purging;
combined desizing, scouring and bleaching; catalysed bleaching.
Experimental work has demonstrated the value of the steam purging technique
as a means of obtaining rapid full impregnation of loomstate fabric. This
is an essential feature of rapid processing at this stage.
Using steam purging, combined desizing, scouring and bleaching has been
achieved with a 24-hour reaction period at ambient temperature. The
extent of reaction after 5 hours suggests that acceleration of the three
reactions can be obtained. Two separate approaches were tried : a
two-stage application of a foamed hypochlorite solution followed by the
peroxide bleaching step; a single-stage application of peroxy salts
incorporated with the hydrogen peroxide bleach; further work will be
required to obtain the optimum minimum times.
A literature survey has been made on the use of activators or catalysts

for bleaching textiles. The more important substances identified were
urea und tungstates. Experiments have started to study urea. In pad-
steam experiments, initial results suggest that reaction times can be
halved, but considerably more work is required.
INDUSTRIAL APPLICATION OF CONTINUOUS
WET-ON-WET TREATMENT OF
FABRICS WEBS
Author Ir R.B.M. Holweg
Contract number EEB-1-139-N
Duration 27 months, July 1, 1981 - October 1, 1983
Head of project Ir M. Glastra
Contractor Netherlands Central Organisation

for Applied Scientific Research
Address Fibre Research Institute TNO

97 Schoemakerstraat
2628 VK Delft, the Netherlands
Summary
In this paper the results are described of an investigation carried out

with the aim to find a method enabling the reliable and continuous im-
pregnation of a fully wet fabric in practice.
The impregnation of a fabric is often done on a pad. For this process
use is made of dry fabrics. If use could be made of a wet fabric, energy
savings could be obtained as a result of the ommission of a drying stage.
In this investigation an impregnation process, frequently carried out in
industry, is considered. This includes the impregnation of dry fabrics
with a resin to make the fabric crease resistant.
It will be investigated in which way this impregnation can be carried
out in practice without applying a preliminary drying stage.
In the laboratory pilot and full scale experiments are being performed.
Several methods by means of which wet cloth can be impregnated are
tested. In this way experience will be gained to enable the performance
of the selected impregnation process in practice.
The results of these experiments in which three different types of
fabrics are involved will be described.
At present preparations are also made to be able to perform experiments
in Dutch mills.
-170-
1. INTRODUCTION
In manufacturing a dyed or a printed cloth from grey material the
cloth is subjected to a multiplicity of treatments.
There are four characteristic treatments which are performed in the
sequence shown in Fig. 1. The sequence is repeated several times dur-
ing the manufacturing of a dyed, or a printed cloth. In each sequence
the cloth is impregnated with other types of chemicals. The reacting
and washing conditions are different as well. In most cases the cloth
is dried after each sequence.
Imp re g-
Cloth Reacting Washing Drying
nating
Fig. 1 Sequence of operations carried out several

times in a textile finishing mill.
When a large batch is impregnated with a resin or a dyestuff, the im-

pregnation is carried out continuously on a pad. To avoid inferior re-
sults it is necessary to dry the cloth first.
The pad (cf. Fig. 2) consists of a trough and a squeezing device. The
solution of water and the chemicals to be applied to the cloth are put
into the trough. The cloth is guided through the trough over rollers,
after which the excess of solution is squeezed off.
In this way some thousands of metres of cloth are dyed continuously.
If it were possible to impregnate continuously wet cloth on a pad energy
savings could be obtained by omitting the drying stage.
LIQUOR SUPPLY
SQEEZ1NG DEVICE
CLOTH
Fig. 2 PRINCIPLE OF A PAD
In some cases wet-on-wet impregnation, i.e. impregnation of wet cloth

with a solution, is already applied. This refers firstly to treatments
which are carried out in a discontinuous manner, secondly to continuous
treatments where low-cost substances are applied and where the accuracy
of the application is not very important.
There are several methods by means of which wet cloth can be impreg
nated continuously:
1. Methods where the cloth is immersed in a bath.
1.1 Impregnating the wet cloth on a pad in a modified way.
1.2 Impregnating the wet cloth on a suction drum machine by means
of which intensive contact between cloth and bath is obtained.
2. Methods where relatively small quantities of liquor are applied to
the cloth without submerging the cloth in a bath. These methods are
generally known as MinimumApplication techniques or MAtechniques.
Small quantities can be applied e.g. by means of the following
methods:
2.1 Engraved rollers.
2.2 Foam. Foaming agents are added to the solution to be applied
and the solution is thoroughly mixed with air. Different tech
niques can be used to apply the foam.
In advance of this investigation a research project, supported by the

Commission of the European Communities, concerning wetonwet impreg
nation of fabrics, was already performed (contract no. 39278EENL).
In this project three of the above mentioned methods, i.e. the methods
1.1, 1.2 and 2.1, were tested and compared. It was concluded that the
most promising and simple way to impregnate wet cloth in practice is
the method where use is made of engraved rollers. But it is also pos
sible to impregnate wet cloth on a pad.
A suctiondrum machine constructed to intensify the contact between
cloth and impregnating solution, which machine should make the impreg
nation of wet cloth easier, should have to be improved.
In continuation of the above mentioned project an impregnation process,
which is carried out in industry, starting with dried cloth, will be
considered. It will be investigated in which way this impregnation can
be carried out in practice without a preliminary drying stage.
2. DESCRIPTION OF THE INVESTIGA TION

The present investigation can be divided into three stages.
In the first stage an enquiry will be made of the industrial impregna
tion processes which could be performed with wet cloth without inter
mediate drying. One of these processes will be selected and studied in
detail.
In the second stage pilot or full scale experiments will be performed
in the laboratory. Three different methods to impregnate wet cloth will
be tested. In this way experience will be gathered to enable the perfor
mance of the selected impregnation process in practice, without applica
tion of a preliminary drying process.
In the third stage trials will be carried out in industry. One or two
methods will be tested in industry. The results of the usual procedure
will be compared with the results obtained by impregnating wet cloth.
At present the first stage is completed and the second is nearly com
pleted. For the third stage preparations are being made.
172
3. RESULTS OBTAINED SO FAR
3.1. Enquiry in the textile industry and selected process
3.1.1. Enquiry
An enquiry has been made in the Dutch textile industry which
showed great interest in wet-on-wet processes to save energy. It was
concluded that mills are interested in applying the following substances
to wet cloth without previously drying the cloth: resins, optical bright-
eners, softeners, water repellent compounds and dyes.
It was pointed out during the survey that the above mentioned substances
could be applied to wet cloth if a reliable and accurate method would be
available. In particular for dyeing a high accuracy is needed. So it is
of great importance to pay attention to the accuracy with which the sub-
stances are applied in the tests.
3.1.2. elected_p_rocess
The process selected for the present investigation is the impreg
nation of fabrics with a resin. This process is of general interest and
is carried out frequently.
The aim of the resin impregnation is to make the cloth crease resistant.
After the (dry) cloth has been impregnated with the resin the following
treatments are carried out in practice: firstly, the cloth is dried and
secondly the cloth is treated at a high temperature for a short time to
fix the resin to the fibres of the cloth.
A lot of resins are available for the crease resistant finish of a fabric.
An often used resin was selected for the investigation.
In impregnating fabrics with a resin a number of other chemicals are ap
plied at the same time, from which the catalyst is a very important one.
The catalyst is needed for the fixation of the resin to the fibres of the
cloth. For this purpose magnesium chloride is often applied. In the labo
ratory tests the selected resin and magnesium chloride are applied in
stead of a complete set of chemicals. In practice experience will be
gathered in applying all chemicals at the same time.
The chemical solution applied to fabrics for obtaining a crease resistant
finish is almost colourless. A green dyestuff is added to the impregnat
ing solution in the experiments in order to enable a visual inspection of
the process during the experiment.
So the following substances are applied in an laboratory experiment:
a resin
magnesium chloride
a green dyestuff.
The physical properties of these substances are quite different. So a
deeper understanding of the effect of the type of substance in the wet
onwet process is obtained. In this way the knowledge obtained during
this investigation will be useful for a wider field of application.
In general each mill manufactures a large number of specialized products.
Therefore, it is of interest to study the effect of cloth type in the
wetonwet impregnation. In the investigation three different cloths are
involved: a light print cloth, a poplin and a work clothing. Composition
and weight per square metre of these fabrics are given in Table I.
173
Table I Composition and weight per square metre of the fabrics
Weight per
Fabric Composition square metre
(bone dry)
Light 100 % cotton about 0.10 kg

print cloth
Poplin 100 % cotton 0.12 kg
Work clothing 35 % cotton 0.25 kg
65 % polyester
3.2. Pilot and full scale experiments in the laboratory
3.2.1. Introduction
Of the several methods by means of which wet cloth can be impreg-
nated the following three are being tested in the laboratory:
impregnating the wet cloth on a pad in a modified way;
impregnating the wet cloth by means of engraved rollers;
impregnating the wet cloth by means of foam.
The results obtained so far will be described further.
3.2.2. Experiments on the pad

The pad is usually used to impregnate dry cloth. However, wet
cloth can be impregnated on a pad as well, if some precautions are taken.
Both the starting concentration of the solution in the trough of the pad
and the concentration of the solution supplied to the trough during the
impregnation have to be adjusted. To be able to adjust these concentra-
tions a parameter has to be determined. A method to determine this para-
meter was developed in the preceding investigation. In this method a
(relatively) small length of cloth is impregnated without supply of
solution to the trough of the pad and samples of the bath are analyzed.
The parameter depends upon the type of substance, the type of cloth and
the impregnating conditions.
In the present investigation a large number of parameter determination
were carried out, using the three types of cloths and impregnating these
cloths with a solution containing the resin, the dye and magnesium chlo-
ride. The effect of cloth speed on the value of the parameters was also
examined.
These determinations were carried out on a pad of technical scale (cf.
Fig. 3 and Fig. 4 ) .
In each determination one of the three types of cloth was wetted and then
guided through the trough of the pad. During the experiment bath samples
were taken and analyzed.
Some hundreds of metres of cloth were used in these experiments.
The resin content of the bath samples taken was determined by means of
the analysis of the nitrogen content using the Kjeldahl method.
The magnesium chloride content was determined by analyzing the chloride
using a Potentiometrie method.
The dyestuff content was determined spectrophotometrically.
From these contents the parameter is calculated for each substance applied.
Fig. 3
Photograph of the pad
-
front view.
Fig. 4
Photograph of the pad
- back view.
To be able to calculate these parameters the initial water content of

the cloth as well as the content after the impregnation should be known
also.
Some of the data obtained will be discussed below:
- effect of initial water content of the cloth;
- effect of cloth speed;
- effect of type of substance.
Effect of initial w ater content of the cloth

In Fig. 5 some of the results of the impregnation of the poplin cloth are
presented. This figure shows the results of two experiments in which only
the initial moisture content of the cloth differed. All other conditions
remained unchanged.
The values found for the parameter (PAR) of the dyestuff and the magne-
sium chloride are plotted against the initial water content of the wet
cloth (Rj.). According to theoretical considerations the values of PAR of
each substance should be on a straight line where the tangent of incline
is 1. So there appears to be a fairly good agreement between theory and
0.4
0.5
0.3
0.4
0.2
0.3
0.1
0.2
30 40
Cloth speed (m/min)
0.1
Fig. 6 I M P R E G N A T I O N OF WORK CLOTHING
ON T H E P A D .
0.2 0.4 0.6 0.8 1.0

E F F E C T OF C L O T H S P E E D ON T H E
RI X 103(m3/kg)
PARAMETER PAR
Fig. S I M P R E G N A T I O N OF POPLIN ON T H E P A D . NOTE

VALUES FOUND FOR THE PARAMETER PAR VERSUS BECAUSE THE INITIAL MOISTURE CONTENT ( R I )
OF THE CLOTH DIFFERED SOMEWHAT IN EACH
THE WATER CONTENT IN THE CLOTH BEFORE
EXPERIMENT THE VALUES OF PAR ARE CORRECTED
IMPREGNATING ( R I ) TO VALUES WHICH WOULD HAVE BEEN FOUND WHEN
R I EQUALS 0 . 4 5 0 X 1 0 " 3 M 3 /KG
practice. This was also found in the investigation carried out earlier.
The correlation between PAR and the initial moisture content is of great
interest for practical application of wet-on-wet impregnation.
The initial moisture content of the wet fabric to be impregnated may vary
somewhat. The above mentioned correlation means that no extra determina-
tions of the parameter need be made. Considering the mathematical decryp-
tion of the process it even means that the starting concentration need
not be changed in practice when the initial moisture content of the cloth
varies.
Effect_of_cloth_sgeed
In the case of work clothing the effect of cloth speed is shown in Fig. 6.
In this figure the results of the parameter determinations using work
clothing at different cloth speeds are plotted. The effect of cloth speed
is small. So, in practice, changes in cloth speed have no great effect on
the process.
Effect of type of substance

The effect of the type of substance can be seen in Fig. 5 and Fig. 6. A
correlation between the data obtained appears to exist. For each cloth,
independent of cloth speed, the value of PAR for the dyestuff, divided
by the value of PAR for the chloride is fairly constant.
The correlation found can be used in future experiments to correct data
when the analysis is not very accurate. Moreover, the number of determi-
nations can be reduced, because the value which has to be known can be
calculated.
3.2.3. Impregnation of wet cloth by means of engraved rollers

In the preceding investigation experiments with engraved rollers
were performed on a semi-technical printing machine (cf. Fig. 7 ) . The ex-
periments on this machine were continued to examine the effects of cloth
speed and the type of cloth.
Fig. 7
Photograph of the
printing machine.
Before performing the experiments the printing machine was modified in

order to make it more suitable for the impregnation of wet cloth.
TO DRYER
RUBBER
BAND
BACK
(, ) GRAY
CLOTH TO BE
PRINTED
PRINTING
ROLLERS
Fig. 8 OUTLINE OF THE FOUR-COLOR

PRINTING MACHINE
Rubber WET CLOTH TO BE

IMPREGNATED
CLOTH
Fig. 9 DIAGRAM OF THE PRINTING MACHINE

FOR WET-ON-WET IMPREGNATION
The modifications can be derived from comparing Fig. 8 with Fig. 9.
Fig. 8 shows a diagram of the arrangement when cloth is printed by means
of four engraved rollers. Fig. 9 gives a diagram of the arrangement made
for the application of a solution to a wet cloth by means of one engraved
roller. For this arrangement no use was made of back gray or rubber band.
Furthermore the layers of fabric wrapped around the pressure bowl were
replaced by a layer of rubber material.
Several experiments were carried out. As in the experiments on the pad
three types of cloth were used. For each experiment the cloth was wetted
first with water on a laboratory pad.
The solution applied contained the same substances as in the experiments
using the pad. The solution was applied to the wet cloth by means of a
roller which was engraved by milling. Fig. 10 shows part of a roller en
graved in this way. The roller used had the following engraving charac
teristics:
40 holes/cm
depth of holes 100 .
Fig. 10
Photograph of the
surface of a roller
engraved by milling.
The penetration of the solution into the cloth was observed. This was sat
isfactory in the case of print cloth and poplin. For work clothing, which
is much heavier than the other materials, a roller is required which can
supply a larger amount of solution.
After analysis of the samples of the impregnated cloths the following con
clusions could be drawn.
The effect of cloth speed is small, which is shown in Fig. 11.
Comparing the results obtained with different cloth types it seems that
the amount of solution applied depends on the roughness of the cloth.
In addition it can be concluded that the pressure between the printing
roller and the cloth is of interest. This pressure is measured by means
of a pressure gauge. A more reliable electronic measuring system would
be recommended.
179
80 100
Cloth length (m)
Fig. 11 A M O U N T OF DYESTUFF APPLIED ON THE
POPLIN CLOTH BY AN ENGRAVED ROLLER
3.2.A. Impregnation of wet cloth by means of foam

In some textile mills chemicals are applied to a dry cloth by
means of foam. In this way the amount of water applied to the cloth is
half as much as applied on the pad. So energy can be saved in the drying
process.
Using foam is a rather new technique for applying chemicals to a cloth.
This method could be made suitable for the impregnation of wet cloth as
well.
In applying foam to a dry cloth the nature of the cloth appears to have
effect. The same type of cloth may show a difference in the tendency to
absorb water. This may be attributed to a difference in conditions dur-
ing a preceding drying stage and gives variations in results in foam ap-
plication. In the case of wet cloth such difference in the nature of the
cloth will not affect the result. This is an advantage in the application
of foam to wet cloth.
A foam is obtained by adding a foaming agent to the solution containing
the chemicals to be applied. In order to get satisfactory results the
type of foaming agent has to be selected carefully. The solution is foam-
ed up by blowing air in the solution and by thoroughly mixing the air and
the solution. This is performed on a special machine. Next the foam is
applied to the cloth. The amount of foam applied to the cloth should be
homogenous across the width of the cloth. For the application of the foam
to the cloth several techniques have been developed.
A few experiments have been carried out on a semi-technical installation.
This installation, constructed by the Fibre Research Institute TNO, has
been designed with the aim to compare several foam application methods.
The installation is mainly used for experiments in which dry cloth is im-
pregnated. Compared with industrial installations this machine mainly
differs in width. The width of the cloth may not exceed 0.35 m on the
semi-technical installation.
Two different methods were used for the impregnation of wet cloth. These
methods are schematically shown in Fig. 12 and Fig. 13. Fig. 12 shows
where the foam is applied to the cloth by means of a roller. Foam is
applied to a rubber roller and then to the cloth.
Fig. 13 shows a diagram of the application by means of the vertical slit
method. The foam in the slit has a subatmospheric pressure, which results
in penetration of the foam into the cloth passing the slit.
The foam applied in the experiments was composed of the same substances
as the solution used for the experiments with the pad and the printing
machine (a green dyestuff, a resin and magnesium chloride). In addition
the foam included two foaming agents.
Experiments were carried out using a poplin cloth saturated with water
on a pad.
During the experiments, using the method shown in Fig. 12 the roller,
which was not driven, slipped sometimes. The friction between the roller
and the cloth is insufficient when wet cloth is used.
However, it can be stated that in the case of a driven roller the impreg-
nation would be rather homogeneous across the width. Moreover, an ade-
quate penetration of the chemicals into the cloth would be obtained.
In applying the foam to the wet cloth by means of the vertical slit meth-
od (cf. Fig. 13) a fairly good evenness across the width of the cloth was
obtained. The penetration of the chemicals into the interior of the cloth
was fairly good as well. Whether or not sufficient results are obtained
largely depends on the process conditions, which have to be selected
carefully.
It may be concluded that wet cloth can be impregnated fairly well by
means of foam.
3.3. Preparations made for experiments in industry
Experiments in Dutch textile mills will be carried out with a pad

and with engraved rollers. Installations on which the experiments can be
carried out are available.
For the impregnation of wet cloth on a pad an installation is available
consisting of a heavy duty squeezing device and a pad. First the wet
cloth can be squeezed off as well as possible on this installation. Then
the wet cloth can be impregnated. The squeezing off and the impregnation
can be carried out in one continuous operation.
Wet cloth can be impregnated with engraved rollers in industry on an in-
stallation which is in use for applying small quantities of solution to
dry cloth.
Preparation are being made in order to be able to perform experiments in
industry. So it is necessary to install equipment on the pad enabling the
continuous measurement of the moisture content of the cloth before and
after the impregnation.
4. CONCLUSIONS
Experience is gained in the laboratory to impregnate wet cloth with a
resin. The experience will enable the performance of the resin impregna-
tion of wet cloth in practice.
Cloth
Rubber
rotler
Flg. 12 FOAM APPLICATION

WITH A ROLLER
g Onl Rollers
Fig. 13 FOAM A P P L I C A T I O N WITH

A VERTICAL SLIT
-182-
DISCUSSION
Question - Chairman : When studying, the effect of cloth speed with the
three different cloth types, it might be of interest to plot
the results versus weight cloth speed (kg/m.min = linear
cloth speed (m/min) specific weight of cloth (kg/m2)) instead
of versus linear cloth speed (m/min; cf. Fig. 6 ) .
Answer In this study with three different types of cloth weight, cloth
speed might be a very useful parameter. The results obtained
will also be plotted versus weight cloth speed.
Question - Dr. Pilavachi, CEC : In case of the engraved rollers would the
thickness of the material have an effect on the amount of
dyestuff applied ? If so, how can you overcome this problem ?
Answer When the thickness of the cloth is not uniform the amount
applied will differ over the surface. This could be overcome
by applying the solution to a rubber roller, which then
applies the solution to the cloth.
Question - Dr. Ehringer, CEC : Would it be possible by means of deeper

engravings of the roller to adapt the quantity required to
the specific weight of the cloth ?
Answer Yes, rollers can be engraved with different depth by using the
engraving technique (i.e. milling) by means of which the roller
used in the experiments has been engraved.
Furthermore, a machine with two or three successive rollers
could be made. One roller is used for a light cloth, all of
them are used when a heavy cloth has to be impregnated.
Question - Dr. Roberts, Shirley Institute : How important is the squeeze

pressure of the last treatment befowe wet-on-wet padding ?
How should the pressure relate to the pressure of the squeeze
in the wet-on-wet padding ?
Answer Before wet-on-wet padding the cloth should be squeezed to a

water content as low as possible. For this a heavy duty
squeezing device is needed, which preferably is placed in front
of the pad. The water content before padding should be such
that this content can be increased with 20 to 30 percent by
weight on the cloth during padding without a cloth being
obtained from which the solution drains off. After padding
the cloth should be sligtly squeezed.
-183-
LOW ENERGY PREPARATION PROCESES FOR TEXTILES
Author : JOHN G. ROBERTS
Contract number : EEB-I-HO-UK
Duration : 30 months 1 January I98I - 30 June I983
Head of Project : Dr JOHN G. ROBERTS
Contractor : Shirley Institute
Address : Shirley Institute

Didsbury
GB - MANCHESTER M20 8RX
Summary
The objective of this project is to develop a sequence

of low-energy pathways for the preparation of cotton and
polyester/cotton fabrics. In this progress report work
is described which demonstrates the value of steam purging
of fabric prior to achieving full impregnation of loomstate
fabric in the first stage of processing. The possibility of
combining desizing, scouring and bleaching in a single
processing step has been demonstrated with a difficult fabric.
The single stage process still requires acceleration to
bring overall reaction times down from 2k hours to a more
convenient level. The objective is less than 1 hour.
Initial experiments with modified processing and use of
accelerators suggest that this will be a reasonable objective.
Surveys of existing processes show that current industrial
energy requirements for the preparation processes are between
10 and 30 GJ/tonne. Combined single-stage processing with
final washing and drying as the only heating stages would
only require 5 GJ/tonne.
-184-
1 Project Objective
The objective of this project is to develop a sequence of
optimum low-energy pathways for the preparation of cotton and
polyester/cotton fabrics.
2. Introduction
The final stages of production of textile materials are
application of colour by dyeing or printing and the application
of special finishes to impart properties required in the textile
to suit it for its end use. Before any of these processes can
be applied, textiles as produced by weaving, knitting or other
techniques require preparation. Typically, preparation processes
include:
1) sin'ein;j to remove surface fibres,

2) desizing to remove polymeric materials applied
to strengthen and lubricate fibres during fabric
forming processes,
3) scouring to remove impurities including oils,
fats and waxes,
and k) bleaching to improve the ground colour of the
fibre and to provide a uniform base.
These processes convert an inherently-coloured, sometimes

dirty, unwettable material into an absorbent, clean and white
fabric, ready to receive colour and chemical finishes.
2.1 Preparation Processes

The singeing process is a relatively simple one, surface
fibre is removed by a burning process which is achieved by
parsing the fabric through a gas flame. The energy requirement
IT consequently low. As fire is a risk in this process it is
normal for the fabric to be passed directly into a quench bath
which is frequently used as the impregnating step for the next
process, namely, desizing.
To remove size it must either be swelled in water or broken
lown by acid or an enzyme. Sizes are polymeric materials,
commonly, starch based; but synthetic materials such as
polyvinyl alcohol or Polyacrylamide are being used. Synthetic
sizes are generally soluble or are highly swollen in water and
thus do not require degradation. It is the starch or starch
derivatives which require breaking down to render them soluble.
Whether the fabric is impregnated with water to swell and
solublize, or with some agent to break down the size (acid,
an enzyme solution or an oxidizing agent), this reaction is
time and temperature dependent so that, in practice, a period
of heating or a longer dwell in the cold preceed a washing-off
process.
The next stage is scouring. This is an alkaline treatment
which saponifies and removes fats and waxes and solubilizes
pectic cell-wall material in the cotton fibre. It is this stage
which drastically affects the wettability of the fabric. In
the case of polyester/cotton fabrics the level of alkalinity
must be controlled so that the polyester component is not
significantly hydrolysed. Again this process is time/temperature
dependent and the way in which it is carried out varies from
long time (y*2k hrs) in the cold to one or two minutes at
110C under pressure. This stage is usually followed by a
washing off process in which hot water is used.
The fabric is now cleaner and more wettable but is quite
highly coloured and this colour is removed oxidatively in the
bleaching process. The chemical most commonly used is hydrogen
peroxide though sodium hypochlorite and sodium chlorite also
find use.
As with scouring the bleaching process is performed
on a wide variety of types of machine with equally widely
differing time/temperature characteristics.
A final washing off in hot water is followed by drying.
In some processing sequences intermediate drying can occur
but this is now usually avoided.
Other processes may be encountered in this sequence, in
particular mercerization for cotton fabrics and heat setting
for polyester/cotton but these are not included in this project.
3 Survey of energy use in preparation processes

In order to establish a base line of energy use in these
preparation processes five mills have been surveyed (Table 1) and
calculations have been made to determine the energy content
of individual steps in the preparation sequence. In addition,
data on estimates made in the early 1970's has been re-calculated
to enable comparisons to be made (Table 2 ) .
3.1 Singeing Information from two mills shows that the singeing
process using gas-burning singers requires 0.3 to O.'t GJ/tonne
of fabric processed. The main variable which will effect this
value is the weight of fabric which is processed.
3.2 Desizing normally follows singeing in that fabrics are

quenched in a solution containing an enzyme which is used to
break down the size. After a suitable delay the fabric may be
washed off (a process with an energy use of about 5 GJ/tonne)
or may pass directly to the next stage.
33 Scouring Scouring processes vary considerably, in the

scale of operation, in the conditions used, and the type of
machinery employed. Batches of fabric, in some cases up to
't tonnes, are processed in kiers where caustic liquors are
circulated for reaction timee of several hours. In continuous
processing a single web of fabric may be processed on a roller-
bed steamer where the reaction time is measured in minutes.
In most cases scouring is followed by washing-off to remove the
impurities which have responded to the treatments applied.
186-
Table 1
Energy use on preparation processes determined by survey
Processes Source Energy use

(GJ/tonne)
Scour - pad steam 6.2

Wash off 5.5
Bleach - pad/steam Mill 6.2 25.3
Wash off 5-5
Dry 1.9
Scour - Markal and Wash off 13.2

Bleach - roller bed steamer Mill 11.8 26.6
Dry 1.6
Rope - Kier scour and bleach

and wash off 7.0
Hill
10.7
Dry 3.7
J-Box scour and bleach

and wash off 8.1
10.1
Dry 2.0
Vaporloc pressure bleach

and wash off Mill 8.2
Table 2
Estimates of energy use in typical preparation processes
Processes Source Energy use

(GJ/tonne)
Two stage pad-roll (peroxide bleach) Estimate 1^.5

Two stage pad-roll (cold chlorite) Estimate 13.4
Pressure bleach Full White Estimate 21.6
Half White Estimate 20.7
Half White Estimate 14.5
polyester/cotton
Continuous J-Box two-stage Estimate 16.1
Batch on beam scour and bleach Estimate 11.2
Kier boil Estimate 12.3
Note: All estimated values include scour-bleach and drying steps
- 187-
As far as can be judged at present, the scouring and washing-
off stages have similar levels of energy use. Typical results
suggest that the energy requirement for scouring cotton fabric
is about 6 GJ/tonne. Solvent scouring for polyester/cotton by
the MARKAL process is similar, but kier and J-box processing
are considerably lower. Washing-off requires an additional
5 to 6 GJ/tonne.
3.'t Bleaching The type of machinery and the way in which fabric
is handled in bleaching is generally similar to that in scouring,
and hence the energy requirements are also similar. Washing-off
is an important part of the processing and is again found to
require about 5 GJ/tonne. This energy requirement is highly
dependent on the water usage in the process and can also be
significantly influenced by the use of a heat exchanger to
recover heat from the outgoing effluent.
35 Drying, which is usually on cylinders, varies in the energy

requirement between 1.9 and 37 GJ/tonne. This variation is, in
part, a reflection of variation in fabric width and weight
and the moisture content.
3.6 Overall, the preparation stages made a significant

contribution (10 to 30 GJ/tonne) to the energy use in a dyeing
and finishing works where polyester/cotton or cotton fabrics are
processed. Hence, any modifications in processes which result
in reduced energy use will be highly desirable provided that
production rates, quality and other costs do not suffer.
^. Reduction in Energy Used in Preparation Processes

It is envisaged that processing can be achieved with a very
low energy use. A reasonable target is for elimination of all
heated processing with the exception of the singeing process,
final washing off and drying. An overall energy use of some
k to 5 GJ/tonne would then be possible.
This would be achieved by cold reactions in desizing,
scouring and bleaching both of which are technically feasible
but suffer from the requirement of lengthy processing times.
Combined processes have been developed but for reasonable
reaction times require elevated temperatures. Three lines of
experimentation are being followed to reduce energy requirements
and still maintain short reaction times. These are :-
1) steam purging
2) combined desizing, scouring and bleaching
3) catalysed bleaching
5 Steam Purging
A significant factor in all preparation reactions is the
poor wettability of fabric in the early stages of processing.
It has been shown that steaming fabric displaces air from
interstices within the fabric. If immersion occurs without
further contact with air then a fabric can be fully wetted out.
The principle of steam purging1 is described in British Patent
1,^68,028. Although very successfully used in a limited number
of applications, it has not been exploited in the early stages
of preparation processes.
Steam purging is used primarily to enable complete wetting-out
to occur when impregnating textile materials. In previous work
it has been proved effective with a wide range of materials which
could not be completely wetted out using conventional techniques.
The principle is a simple one. Fox, Marshall and Stewart (J. Soc.
Dyers & Colourists, I967, 83, *+93-51) showed that if air was
removed from a fabric by evacuation, then even so-called
unwettable fabric, such as loomstate cotton, could be thoroughly
"water-logged" so that impregnation processes then become viable.
In the case of steam purging, fabric passes through a chamber
with a countercurrent flow of steam, which displaces air in the
fabric. The fabric outlet of the unit dips into the impregnating
trough where the steam in the fabric condenses with complete
impregnation of the fabric. The effect is analogous to the
vacuum impregnation technique.
Experimental equipment has been made and is being evaluated.
(Fig 1.). Typical improvement in pick-up of impregnating liquors
by loomstate cotton fabric is from kh% by conventional padding
to 9-0% when steam purging precedes impregnation.
Thus,thorough wetting out of loomstate fabrics is possible
and this factor is seen as important in determining the time of
reaction of any applied solutions and the concentration of
applied solutions.
6. Combined Desizing, Scouring and Bleaching

In the preparation sequence, major energy savings can be
achieved if the three stages of processing, i.e. desizing,
scouring and bleaching, can be combined with a single washing-off
step at the end of the sequence. If, in addition, the preparation
stages can be achieved at ambient temperature, then the energy
requirement would be at an absolute minimum. These possibilities
are currently being examined using a seedy cotton 3/1 twill
(80 ^6, 12/10's) containing sago/tallow size. E xperiments have
been made in which fabric was first steam purged and then passed
directly into bleaching baths. In the first series of experiments
the concentration of hydrogen peroxide, concentration of stabilizer
(sodium silicate), and time of reaction were varied. Typical
results are outlined in Table 3 It can be seen, that with a
peroxide concentration of 50 ml (100 vol) per litre and a reaction
time of 2k hours, that promising levels of bleaching can be
obtained. When stabilizer was present the level of bleaching was
improved and was accompanied by a fairly good level of seed
removal. Levels of absorbency of residual size (measured as
starch equivalent) and residual fat and wax content were
acceptable. Thus with a standard bleaching bath, with impregnation
aided by steam purging, it has been possible to obtain combined
189-
Table 3
Combined desizing, scouring and bleaching with impregnation aided

by steam purging
NaOH (g/1) 25 25 5 25 25 25
H 2 0 2 ml of 100 vol/1 25 25 50 50 50 50
Stabilizer (ml/l)
Sodium silicate
lone None None None 12 12
.Reaction time (hr) 5 24 5 24 5 2k
Whiteness
TAUBE 42.0 49.2 55.8 57.9 64.4 64.2 72.3
Absorbency (sec) 00 4.6 6.7 5.'t 4.3 4.3 5.5
Starch equivalent (5'J) 79 3.5 2.8
Fat and wax (%) 0.75 0.45 0.41 0.43 0.42 0.44 0.41
Seed removal poor poor poor mod mod fairly

erate erate good
190
desizing, scouring and bleaching in the cold in 2k hours.
The results after 5 hour treatments are also encouraging and
further work is in progress to investigate the possibilities
to both improve the levels of seed and size removal and to
further shorten the reaction time.
In a parallel series of experiments a two stage application
of bleaching agents is being examined. In a first treatment
fabric is steam purged then directly passed to a bath of foamed
sodium hypochlorite solution. A fter a suitable reaction time
the fabric is fully impregnated with a hydrogen peroxide solution.
Sodium hypochlorite solutions can be readily foamed in a
COWIE and RIDING laboratory foam generator (Fig 2.) which
operates on the high shear stirring principle. A typical foam
of high stability can be made from a solution containing:
300 g/l sodium hypochlorite

5 /1 ELA a xanthan gum foam stabilizer
and 5 g/l LANK0F0L a surface active foaming agent
A typical blow ratio for the foam is 4.8:1

After steam purging and passage through the foam the
fabric picks up kJ by weight of foam based on the dry fabric
weight. Thus approximately 3' hypochlorite is added to the
fabric. The fabric was batched cold for 1 hour during which
time considerable bleaching occurs but with incomplete
removal of both seed and size. The fabric is then padded through
a stabilized hydrogen peroxide solution 25 ml (100 vol) per litre
picking up an additional 23'^ by weight of this solution. A
further 5 hour reaction time produced levels of size and fat
and wax removal equivalent to conventional processing (Starch
Equivalent 1.6% 3endered Fat and Wax 0.38/J). Good absorbency
1.5 second wetting time and reasonable levels of whiteness
(Y value 88, Taube 70.7) Seed removal was only moderate
and improvements in this and in whiteness are required.
Condensate
shield
Nip seal
Condensate
Steam ports
Condensate
Figure 1 Steam purging unit Figure 2 Cowie and Riding foam generator
7. Catalysis or activation of bleaching
A literature survey has been made on the use of activators or
catalysts for bleaching textiles and related cellulosic products
with hydrogen peroxide. Materials identified fall broadly into
two classes, organic nitrogen-containing compounds and heavy
metal compounds. Materials of both classes have been examined
in the context of textiles and found to enhance bleaching actions.
The more important materials in each class were respectively urea
and tungstates. These materials have been taken as the starting
point in an examination of the potential for accelerating
bleaching reactions.
Experiments have commenced with the examination of urea
under two conditions, (a) long liquor ratio, and (b) pad-steam.
In the long-liquor ratio experiments, desized and scoured
cotton fabric was heated to 62C for 5 hours at a pH of 10.7
in a liquor containing hydrogen peroxide (10 ml/1 100 vol.),
caustic soda (0.12 g/l), and sodium silicate (.O.kk g/l). Parallel
experiments were made with, and without, urea (20 g/l). The
extent of bleaching was measured by the development of whiteness
and is expressed in terms of Y-values. Reasonable levels of
bleaching were found both with and without urea (Y = 87) after
1 hour reaction and this increased (Y = 91) after 5 hours in
the presence of urea. Without urea this increase after 5 hours
was small (Y = 88). The slight improvement caused by urea in
these long-liquor experiments was probably caused by urea exerting
a buffering effect with a consequent better holding of the
initial pH value. Several experiments showed broadly similar
results. It is concluded that under these conditions urea is un-
likely to produce any significant shortening of reaction times.
However, in the case of pad/steam experiments, initial
results suggest that urea has a greater effect, with the
possibility that reaction times can be halved. Duplicate sets
of cotton fabric samples were impregnated to a 90% pick-up with
a solution containing hydrogen peroxide (30 ml/l, 100 vol.),
sodium hydroxide (0.3 g/l) and sodium silicate (1.1 g/l) with
and without addition of urea (20 g/l). These fabrics were then
steamed in a Werner-Mathis steamer for either 5 or 10 min and
the extent of bleaching measured by development of whiteness.
Equivalent levels of whiteness were obtained in 5 min with
urea and 10 min without urea.
8. Conclusions
There are a variety of ways in which cotton and polyester/
cotton fabrics are processed to achieve desizing, scouring and
bleaching. The variation can be in the process itself, in the
type of machinery used as well as in the time/temperature profile
used. There is considerable scope for energy saving by
modification of processing routes.
Surveys of typical processing encountered industrially
shows that the energy requirement of desizing, scouring and
bleaching is commonly between 10 and 30 GJ per tonne of
production.
The objective of this project is to find ways in which the
processing can be optimised in low-energy pathways. An overall
energy requirement of 5 GJ per tonne of production forms, it is
believed, an achievable target.
Experimental work has demonstrated the value of the steam
purging technique as a means of getting rapid full impregnation
of loomstate fabric. This is an essential feature of rapid
processing at thisstage. Using this technique combined desizing,
scouring and bleaching has been demonstrated using conventional
bleaching solutions. This has been achieved in a 2k hour
reaction in the cold. The extent of reaction - 5 hours suggests
that acceleration of the three reactions (desizing, scouring
and bleaching) can be obtained. Experimental evidence supports
this view in two separate approaches. In the first a two stage
application of oxidizing solutions; a foamed hypochlorite
solution is followed by the peroxide bleaching step. In the
second approach other peroxy-salts are incorporated with the
hydrogen peroxide bleach. Further work is required to obtain
optimum reaction in minimum times.
A further aspect which requires considerably more work is
the possibility of catalysing reactions to obtain significantly
shorter overall processing times.
DISCUSSION
Question - Dr. Ehringer, CEC : Do you see any problems or difficulties

in combining several processes ?
Answer No problem is seen in combining the processes of desizing,

scouring and bleaching, in fact there are advantages to
be gained in saving energy and time of production.
Question - Chairman : 1. What is the steam pressure applied in the

steam purging section ?
2. Can you foresee mechanical recompression
of vapour ?
Answer 1. The pressure of steam in the steam purging apparatuses

is atmospheric.
2. Mechanical recompression of vapour has not been considered
but its value is doubtful as the out-flow from the unit is
small in volume and consists of a mixture of air and steam.
194-
Question - Mr. Holweg, TNO : What wiLL be the minimum amount of steam
for the purging operation ?
Answer In the pi Lot-scale equipment the amount of steam required

for purging fabric has not been measured. It is however
very low. The equipment has small volume and the pressure
is atmospheric, the volume flow has only to be sufficient
to displace air entrapped in the fabric. Measurements will
be made on the full scale equipment which is being built.
Question - Mr. Glastra, TNO : Do you carry out dyeing or printing

experiments with the fabrics pretreated according to your
single stage preparation method ?
Answer At this stage of the project we have not dyed or printed

experimental fabrics. Fabric absorbency is measured and
this parameter is related to dyeing or printing performance.
No special problems relating to dyeing and printing are
expected.
195-
SESSION IV - FOOD INDUSTRY
Chairman: D.A. REAY
Summary of Session IV
Energy saving in the bakery by improvement of energy efficiency

and recovery of waste heat
Energy saving in the soybean extraction industry by reducing

the steam consumption for desolventizing - Toasting and
drying extracted beans
Energy saving in edible oil processing plants by application of

a total-energy (TE) system - Part I: Soybean oil extraction plant
197-
SUMMARY OF SESSION IV
FOOD INDUSTRY
D.A. Reay , P.A. Pilavachi
The food industry is a significant user of energy in aLL Community

countries, although the types of foodstuff being processed varies from one
region to another. This industrial sector also differs from most of the
other major energy users in that the size of processing unit can be very
large (e.g. a brewery or canning plant) or very small (a bakery behind a
high street shop). Many and varied approaches to energy conservation may
therefore be needed.
This is reflected in the projects being funded within the European Commis-
sion Energy R 8 D Programme; (note that other projects of interest to the
food industry can be found in categories such as "Combustion"). All
three projects being supported in the "Food Industry" category are being
undertaken by the NETHERLANDS ORGANISATION FOR APPLIED SCIENTIFIC RESEARCH
(T.N.O.). The first project, concerned with energy savings in bakeries,
is managed by the INSTITUTE FOR CEREALS, FLOUR AND BREAD, and reveals much
data on the several uses for energy in bakeries, both large and small, and
the conservation measures which can readily be undertaken.
The DIVISION FOR NUTRITION AND FOOD RESEARCH within T.N.O. has responsibi-
lity for the two other projects. Aspects of the soyabean processing
industry, which also came under scrutiny within the 1st Energy R 8 D
Programme, are being studied, with particular attention being paid to
reducing steam consumption in desolventizing-toasting and drying of
extracted beans.
The use of total energy plant is topical in several Community countries,

and T.N.O. is studying such systems as applied to edible oil processing
plants, where in some processes the balance of heat and electricity demand
appear to match that available from total energy installations.
Ener2y_Savings_i.n_BakeMes - Approximatively 7000 TJ of energy are used

annually by bakeries in the Netherlands. Half of the bread production is
carried out in small bakeries, of which there are 3600 in the Country, and
an idea of their size can be gained from the fact that only 7 per cent
employ more than 10 staff. It is therefore important that these bakeries,
as well as the larger centralized units, are studied.
The main source of energy is natural gas, the vast majority of which is
used in the baking ovens. These ovens vary considerably in efficiency,
and also each oven has its "good" and "bad" days. An efficient day of
operation can result in 25-30 % less energy usage than days when production
planning was Less than ideal. In addition to improved scheduling, better
ignition and the use of flue dampers could give worthwhile savings.
*) International Research and Development, UK
-198-
Electricity usage was comparatively small, but the cost could be high.
Savings could be made in the important area of refrigeration, and in
lighting. The study will now concentrate on industrial bakeries.
Ener9y_Sav2n2s_2n_S0y.abean_Extracti.0n - The soyabean extraction process

involves a number of stages, many of which are energy-intensive. Five
Community countries each processed in excess of 1 million tonnes of
soyabeans in 1981, and the total amount processed in the Community was
over 10 million tonnes. Steam usage is particularly high, and it is here
that T.N.O. is investigating energy saving techniques.
Two processes are being studied, toasting and drying. In toasting,

combining this with heat-conditioning and using combustion gases is a
possibility, although direct use of some combustion gases may not be
possible because of contamination. In drying, superheated hexane could
be used for desolventizing, giving an inherently dryer product, hence
minimising further energy inputs.
Savings of the order of 25 %, representing in excess of 100 kg steam/tonne

beans, could be realized if this development is successful.
Total Energy Plant in Edible Oil Processing - Provided that the electricity
supply authorities are co-operative, the installation of total energy
packages at factory sites, giving both electricity and heat, can be cost-
effective. It is necessary to identify the demand for electricity and
heat, and to quantify the grade of heat needed and the fuel available on
site before deciding on the type of system to be adopted, if any.
T.N.O. is studying three principal production processes, those for soya-

bean oil extraction, refining, and margarine production. Of the plants
studied to date, margarine production is not attractive, but the high
utilization (6000-7200 hour per annum) and attractive heat/power ratio of
6.9 suggests that the use of total energy in soyabean extraction plants
could be feasible.
The three projects forming this section of the Energy R & D Programme show
how a detailed analysis of specific processes can lead to identification of
cost-effective energy saving measures. Such studies represent a logical
progression from the more general audits carried out in the past in these
sectors.
ENERGY SAVING IN THE BAKERY BY IMPROVEMENT OF ENERGY EFFICIENCY
AND RECOVERY OF WASTE HEAT
Author ING. L. DE VRIES
Contract number EEB-1-148-N
Duration 36 months 1 January 1981 - 31 December 1983
Project leader Ing. L. de Vries
Contractor Institute for Cereals, Flour and Bread TNO
Address Institute for Cereals, Flour and Bread TNO

P.O. Box 15
6700 AA WAGENINGEN, The Netherlands
Summary
An investigation to determine possible ways of energy saving in

bakeries and of recover and re-use of waste heat, was started in
October 1981. The investigation is supported by the ECC, the Dutch
Bakery Trade Association and the Dutch Ministry for Economic Affairs
It is executed by the Institute for Cereals, Flour and Bread TNO,
in cooperation with the VEG Gasinstituut NV. For the investigation,
measurements will be made in 7 bakeries with batch-ovens, 5 bakeries
with continuous ovens and 3 industrial-confectioner's. The results
of 5 bakeries belonging to the first group bakeries, representing
50 % of the Dutch bread production are published in this paper.
By analysis the energy used in various stages of the bread making
and confectionery making processes showed that 75 % of the total
energy consumption was used in the baking process. A saving of
25 - 30 % of the baking energy could be archieved by a better pro-
duction planning, 10 % by substitution of pilot flames by spark
ignition and 10 % by closing of the gas exhaust duct in periods
in which the burners were switched off. Of the total consumption
of electricity 30 - 55 X is spent on freezing and refrigeration.
A saving of 10 - 20 % in this latter energy is possible.
1. Introduction
Each year a total of appr. 7000 TJ of energy is used in the produc
tion of bread and confectionery in the Netherlands. Partly because pro
duction is so very decentralised not enough has been known up to now about
the use of energy in this particular sector.
The object of this investigation is to determine and evaluate pos
sible ways of energy saving and of recovering and reusing heat in the
bakery. Prime importance has been attached to those processes or places
where a large amount of energy is used or lost. A second important aspect
is to outline situations where energy can be saved by the application of
very simple means.
The investigation is subsidized by the Bakery Trade Association, the
Dutch Ministry for Economic Affairs and the EEC. Supervision is in the
hands of a committee represented by the Netherlands Bakery Foundation, the
Netherlands Foundation for Information on Energy, and the Ministries of
Agriculture and Economic Affairs. Execution is in the hands of the Insti
tute for Cereals, Flour and Bread TNO, in conjunction with the VEGGas
instituut NV.
Procedure and Testing

Allowing for the share of production of the different types of baker
ies, it has been decided, after consultation with the supervisory commit
tee, to investigate in 7 small bakeries (with batch production of bread
and confectionery), in 5 industrial bread bakeries and 3 industrial con
fectioner's. The 12 bread bakeries in which the investigations were to
take place were selected on the basis of statistical data contained in
"The Structure of the Bakery Industry in the Netherlands" compiled and
published in 1981 by the Netherlands Bakery Foundation.
Table 1 indicates the data used. First of all the bakeries were di
vided into two main groups, viz. those with batch process ovens and those
with continuous process ovens.
In 1980 59 % of the total bread production took place in bakeries
with batch process ovens and 41 % in bakeries with continuous process
ovens. Within each of these main groups the bakeries were divided into
5 groups according to the size of their bread production. The confection
ery production of these same bakeries has not been included in Table 1.
It is 4 10 X of the bread throughput.
2. Method of Operation in the Bakery Sector and Method of Determination of

Energy Consumption
2.1. Type of Business

The number of businesses in the group of bakeries with batch process
ovens is appr. 3,800, representing 59 % of the Dutch bread production, and
as such the most important subsector. Only 7 % of these businesses employ
more than 10 persons.
In these bakeries not only the baking itself is noncontinuous, but
the preparatory stages like mixing, dividing, rounding, fermentation, etc.
are also noncontinuous. In the majority of these bakeries these opera
tions are mechanized. All the bakeries have refrigerator and/or freezer
facilities that are used to store 80 % confectionery and 20 % bread pro
ducts. A large percentage of confectionery is made at the beginning of the
week, stored in the freezer and sold at the weekend. This means that there
is more time available for bread production during the weekend. The bread
production on Friday and Saturday is 40 50 % of the total weekly pro
201
auction. In most of the bakeries bread is produced from Monday to Friday
from appr. A - 10 a.m., followed by confectionery goods up to 3 p.m.
Between Friday and Saturday, production begins at appr. midnight and ends
on Saturday around noon. This means that in the course of a production
week the ovens are "idle"/stand-by at baking temperature for 50-60 hours.
When production has finished the ovens are switched off and allowed to
cool. They have to be heated up anew to baking temperature each day.
2.2. Preparation of Bread

The basic principle of the preparation of bread is given in Table 2,
plus the energy input. As regards the preparation a distinction may be
made between conventional dough preparation and processing and (chemical-
ly) activated dough preparation. (Table 2)
Activated dough preparation with the addition of cysteine results in
a shorter processing time than is normal with the conventional method.
Table 3 shows the processing times per stage of preparation; varia-
tions are possible.
2.3. Selecting of the Bakeries for the Investigation

A telephone survey was conducted among 24 batch production bakeries
and their suitability assessed on the basis of:
- their willingness to cooperate in the investigation;
- size of production;
- location of the bakery and its machinery/equipment.
Subsequently 14 bakeries were visited and their production apparatus
and quality of production evaluated. Finally 7 bakeries were selected for
the investigations.
2.4. Measuring Apparatus Used

one Philips 12-point recorder type PM 8236
two Honeywell 12-point recorders type Versapunit
one thermo-anemometer Wallac GGA 23S
one thermo-anemometer Lambrecht
waterflow meters type Topas PMW
Binos infra-red gas analyzer for CO-CO amke Leybold-Hereaus
electronic hand thermometers
electric industrial timers Landis and Gyr
kWh meters
The gas consumption was continuously measured using a 12-point re-
corder. To this end the 220 V gas valves of the burner installations were
fitted with relays, over which per burner or burner setting a pre-selected
voltage was passed to the recorder. This signal yielded a record of the
times the gas valves were open. The gas consumption per minute with each
burner setting (gas valve) was also determined. The gas consumption per
unit of time is determined by multiplying the time the gas valves were
open, by the gas consumption per minute.
3. Progress of Work and Results

The investigations in the 7 bakeries belonging to the first group
"bakeries with batch process ovens" have been completed, and the results
from 5 of the businesses have been processed. These bakeries are repre-
sentative of a group of 3,650 firms, which together make up 50 % of the
bread production in the Netherlands.
Table 4 shows the nergy consumption figures for these 5 bakeries,
sub-divided per production groups bread and confectionery. These figures
-202-
relate solely to the production apparatus and production areas and the
office/staff facilities connected with the same. The amount of energy
used per bakery was measured over a period of 7 consecutive days.
The figures relating to the production apparatus and ovens were meas-
ured over 6 days, with the exception of the refrigerators and freezers.
The energy consumption of the shop that often adjoins the bakery and
of the equipment in the same has not been included in the table.
Table 5 shows the amount of natural gas and electricity per bakery as
a percentage of the total amount of energy used. The energy used in the
baking process is also expressed as a percentage of the total energy con-
sumption. It is interesting that in all the businesses investigated appr.
75 % of the total energy consumption is used in the actual baking process.
In bakery no. 18 it should be noted that the consumption of natural
gas was lower and that of electricity higher than in the other bakeries.
This is because in this particular bakery 50 % of the bread and confec-
tionery is baked in an electrically heated oven and 50 % in a gas-heated
oven. In bakery no. 1 there was also a small-capacity electric oven for
confectionery. The other ovens were gas heated, with in some cases the
circulation ventilators being driven electrically. The electricity re-
quired to do this was measured separately using KWh meters and then in-
corporated into the total energy consumption.
As already mentioned, the consumption of electricity for apparatus,
lighting, etc. was determined per apparatus or connected groups of appa-
ratus using separate KWh meters. It was impossible to determine the
energy consumption for different dough types, such as white flour and
whole meal, because the conditions in the various bakeries were not con-
stant. It was also impossible to determine reliable differences in energy
with respect to the baking of different bread types. It was, however,
possible to determine the respective energy consumption for bread and
confectionery products.
3.1. Electricity
Table 6 gives an overview of the consumption of electricity by the
major items of machinery and the lighting. The percentages have been cal-
culated from the total energy consumption measured per week and per proc-
ess stage; so as to be able to compare the various bakeries, the energy
consumption of the ovens was not included; this is recorded separately
in KWh.
In bakeries 5 and 9 the electricity consumption for the proofing
cabinets was low or nil, because they were assembled adjacent to the oven.
In bakery 5 the fermentation cabinet was situated very close to the ovens,
so that the amount of electricity required for heating purposes was very
slight. Hence the low percentage of electricity consumption for dough
processing.
In all the bakeries of this group the energy consumption by refriger-
ators and freezers is high, in most cases amounting to 50 % or more of the
total energy consumption.
Refrigerators and Freezers

The high energy consumption of refrigerators and freezers is largely
caused by:
a. opening doors unnecessarily frequently to put away/bring out small
quantities of products;
b. putting in products that are too warm, which causes an unnecessary load
on the cooling unit. In addition, evaporation of water from warm pro-
ducts results in an extra deposit of ice on the evaporators. Apart
from the decreased efficiency of the cooling system the evaporator has
to be defrosted several times in 24 hours;
c. the insulation is not thick enough;
d. doors that don't shut properly and isolation leakages, shoddy mainte
nance of the refrigeration system, such as insufficient refrigerant,
and fouled condensor, with the result that heat loss is inefficient;
e. unnecessary continuous door heating.
In one bakery there was such a high deficiency of refrigerant that
the refrigerator worked continuously (24 hours). If a normal operating
time of 16 hours a day were to be achieved, this would represent a pos
sible saving of 33 % for this particular installation.
Freezer units are usually supplied with builtin door heating as a
standard feature. There is no switch mechanism whereby the door heating
can be switched on/off as required, so that in practice the door heating
works continuously. The investigation shows that if the door heating is
disconnected, a 19 % saving in energy consumption can be achieved.
With a good sized evaporator (T < 7 C) and proper use of the in
stallation, the evaporator only needs to be defrosted 1 24 hours and
this can be done using an automatic time clock set to work on the cheap
night rate.
Using simple measuring equipment and gauges (theromocouple or pres
sure controller) it is possible to determine the point in the defrosting
cycle at which the evaporator is icefree, and the electric or hotgas
defrosting process can be automatically switched off. This avoids un
necessary use of energy and the temperature in the freezer will not rise
more than is necessary.
3.2. Use of Natural Gas

The percentage shown in Table 4 indicate that the amount of natural
gas used in the production of bakery products is high, varying between
70 90 % of the total energy consumption. Natural gas is used for cooking,
general heating and water heating and for baking. In Table 7 these dif
ferent uses are listed separately and given as percentages of the total
gas consumption
In bakeries with batch process ovens the work is not continuous;
after production the energy supply to the ovens is shut off and the ovens
allowed to cool down.
At the start of each working day the ovens are more or less cool and
energy has to be supplied to heat them up to the required baking temper
ature. The investigation showed that all the ovens reached the required
baking temperature before the products were ready to be baked. The energy
consumption during these periods of standing idle must be regarded as
waste, varying between 5 and 14 % of the total amount of energy used for
baking per week. The energy used to heat up the oven is accumulated in
some way, to a greater or lesser extent depending on the oven type, to be
transmitted to the product at a later point in time.
The amount of energy required to bring the oven up to baking temper
ature varies between 15 and 50 % of the total energy used per day by the
oven, depending on the accumulative capacity and the oven temperature.
In most cases the oven temperature decreases during baking and a part
of the accumulated heat in the oven is used for baking. The amount of
energy that has to be supplied to bake a batch of products consists of
the amount of energy consumption by the oven during baking plus the cal
culated amount of energy that would be needed after baking to bring the
204
empty oven back up to its original temperature. The amount of energy that
is needed to compensate the measured energy can be calculated using the
heat-up curve for the baking oven. This method is depicted in Table 8.
Idle periods also occurred during the production because a following
batch of products was not ready to be baked.
Following the bread production, confectionery products are usually
baked at a lower temperature. Here too it was found that the oven was
often kept up to temperature for long periods with confectionery only
being baked incidentally.
Due to the length of the dough preparation stage in bread production
it is impossible to plan batch production without any loss of time between
batches, but considerable improvements can be made amounting to a 20 %
energy saving in the baking stage.
4. Analysis of Results and Comments

With the help of continuous energy consumption measurements related
to the fluctuations in temperature in the baking ovens and the quantities
and kinds of products baked, the daily energy consumption can be divided
up as follows:
1. Energy loss before baking; energy loss during the period between the
required baking temperature being reached and the baking of the first
batch.
2. Energy consumption bread baking; energy consumption during bread pro-
duction, including energy loss during idling between batches and com-
pensation for the difference in temperature in the baking oven.
3. Energy consumption confectionery; energy consumption during confection-
ery production including energy losses during idling between batches
and compensation for the difference in temperature in the baking oven.
4. Cooling losses; energy loss due to cooling down of the oven between
two consecutive production days during the period when no energy is
being supplied. The energy loss due to cooling is equal to the energy
that must be supplied to the oven to bring it back up to the same baking
temperature as the temperature it was at the time baking stopped on
the previous day. The energy loss during cooling can be calculated from
the heat-up curve.
5. Energy difference; the total of the energy amounts 1 - 4 will show a
slight positive or negative difference compared with the amount of
energy supplied to the oven per day. This is due to the differences
in the degree of cooling that can fluctuate from day to day. The sum
of these differences per day taken over 6 days is more or less equal
to zero.
Table 9 shows the energy consumption per bakery and per oven per
production day.
Energy measurements of this kind have shown that with an average
assortment of bread and confectionery the energy consumption required
to bake 165 kg of bread dough (= 100 kg flour) and 100 kg of confectionery
dough/batter is virtually the same.
In this table the bread production is expressed in kg flour and con-
fectionery production in kg dough/batter. Also indicated is the average
daily energy consumption for baking 100 kg flour for bread production
and 100 kg dough/batter for confectionery production. It can be seen that
there are great differences in the amount of energy used per day to bake
an equal amount of product. Analysis of the results has shown that these
differences can be largely attributed to loss of energy during longer or
205-
shorter periods of idling between 2 successive batches and only very
slightly to changes in the composition of the product assortment. Idling
times between batches occur because the dough production is not suffi-
ciently adapted to the oven capacity.
Allowance should be made for this factor when planning the layout
of a bakery. Hence e.g. the capacity of the proofing cabinet must be at
least twice that of the oven. Given a baking time of 30 mins. and a
proofing time of 60 mins (see Table 3 ) , a proofing cabinet capacity that
is less than twice the oven capacity will always result in the ovens
having to stand idle.
The point at which it is decided to start heating up the oven can
be determined in relation to the processing time, so that the loss of
energy before baking begins, can be kept to a minimum or even eliminated
altogether. Here too considerable differences are measured from day to
day (see Table 9 ) .
If the daily production in these bakeries were to be planned like
that found on the most efficient day in the investigation period, an
energy saving of 25 - 31 % of the baking energy could be achieved (for
calculations see Annex 1).
In view of the wide bread and confectionery assortment in these
bakeries and the often long processing times, it will not always be pos-
sible to achive the calculated saving. A saving of minimum 20 % is cer-
tainly possible with better production planning.
4.1. Energy Saving in the Ovens due to Technical Measures

Many gas-fired batch process ovens are fitted with pilot flames that
burn continuously. This was also the case in bakery 5, which had 2 iden-
tical ovens.
The energy consumption of the pilot flames at times when the ovens
were not being used for baking is indicated in Table 9 under oven 1 + 2 =
1030 MJ per week. The total gas consumption of the ovens was 10,294 MJ.
Substitution of continuously operating pilot flames by e.g. spark igni-
tion would result in a saving of 10 % of the total energy consumption
of both ovens.
It was noted that the energy consumption per 100 kg flour was con-
siderably higher in oven 1 than in oven 2. The reason for this was the
heat exchanger in oven 1, where a baffle plate for the flue gases had
burnt through. Inspection of the burner installations by experts of the
VEG-Gasinstituut revealed that insufficient attention is paid to the
maintenance of the burner installations of most ovens. Regular mainte-
nance and control of the ovens will certainly result in energy saving.
Another source of energy loss in baking ovens was ventilation
through the oven burner section at times when no gas was being combusted.
It was found that 8 - 10 % of the energy supplied to the oven is lost
again due to ventilation via the chimney. This loss can be eliminated
by fitting a valve or sliding partition into the flue gas exhaust duct,
which would automatically close off the exhaust duct when no energy is
being supplied to the oven.
4.2. Recovery of Heat
Cooling Gas Boilers

The waste heat on the condensor side of a frezer installation can
be used to heat up e.g. tap water using a cooling gas boiler. The cost
price of a cooling gas boiler, the capacity of which is adapted to the
-206-
freezer installations in the bakeries, is appr. f 6,000, including
assembly into the cooling system. Of the bakeries investigated, the
greatest amount of energy was used to heat up the water in bakery 9, viz.
1169 MJ per week. The amount of natural gas that is supplied to this
, ., . 52 1 169 ,,,, 3
boiler per year is = = 1731 m .
. 3
At the present cost price of f 0,50 per m of natural gas the pay
600
u J f . ;
back period for a cooling gas boiler is 7 = = 7 years.
This is not including the loss of interest over the invested capital.
Economizers
It is obvious that the ovens are large users of energy. The investi
gation revealed that the burner efficiencies vary between 65 80 % with
flue gas temperatures of 275 C 350 C. Use of economisers in the
ovens' flue gas exhaust duct is a way of recovering energy, provided
this energy can be used elsewhere in the bakery.
The amount of energy used to heat up the bakery was very little,
because of the relatively large amount of heat given off by the ovens
by radiation and convection. In this type of bakery a baker's shop and
living quarters are usually attached to the production area, so that
during the time that the rooms needed to be heated (appr. 35 weeks/year)
the energy that is recovered can be used for ambient heating and heating
of tap water. During the rest of the year only the tap water will need
to be heated. With the economisers that are currently available appr.
55 % of the heat contained in flue gases can be recovered as nett heat.
The cost price of an conomiser for batch process ovens, incl. 300
liter indirect boiler, pumps, safety and control devices + assembly, is
appr. f 12,000. f 15,000. .
Calculation Pay back Time

Example: bakery no. 9
Oven I:
energy input 15,446 MJ/week
energy loss as flue gas 4,264 MJ/week
nett possible saving 0.55 4263 = 2,344 MJ/week
energy consumption tap water boiler 1,169 MJ/week
with 50 % boiler yield nett 0.5 1169 = 585 MJ/week
nett left for heating 2344 585 = 1,759 MJ/week
1759
gross with 70 % central heating boiler yield _ = 2,512 MJ/week
gross per year
52 1169 3
tap water 1,731 m
35.1
35 2512 3
heating 2,505 m
35.1
3
total 4,236 m
Pay back period excl. loss of interest
12000 c ,
4236 0.5 = 5 7 y e a r s
207
5. Conclusions
Each year a total of appr. 7000 TJ of energy is used in the produc-
tion of bread and confectionery in the Netherlands. Partly because pro-
duction is so decentralised not enough has been known up to now about the
use of energy in this particular sector.
The object of this investigation is to determine and evaluate the
possible ways of energy saving and of recovering and re-using heat in
the bakery. The investigations in the 7 small bakeries have been com-
pleted and the results from 5 of the businesses have been processed.
These bakeries are representative of a group of 3,600 firms, which to-
gether make up 50 % of the bread production in the Netherlands. In addi-
tion they also make some confectionery, varying between A and 10 % of the
bread sales.
5.1. Natural Gas

Of the total amount of energy used for the production of bread and
confectionery, the amount of natural gas is very high, varying between
70 and 90 %. The amount of electricity varies between 30 % and 10 %.
With the present cost price of Hfl. 0.5/m for natural gas and
Hfl. 0.20/KWh for electricity, the percentage of natural gas in the total
energy costs varies between 37 % and 63 % and that of electricity between
63 % and 37 %.
About 75 - 90 % of the natural gas is used for the actual baking
process. It has been found that a great deal of energy is lost during
periods when the ovens have to be kept at the right temperature but no
products are actually being baked.
In the bakeries that were investigated the ovens were heated up too
early each morning so that the right baking temperature was reached at
a time when the products weren't yet ready to be baked. Keeping the ovens
up to temperature until they were needed resulted in an energy loss of
5 - 14 % of the total energy consumption used for baking.
Energy was also lost between batches, because the ovens were left
idle. A better planning of the dough production in relation to the oven
capacity would result in a considerable saving of energy. If daily pro-
duction were to be planned like that found on the most efficient day in
the investigation week, an energy saving of 25 - 30 % could be achieved
for the baking process alone.
Substitution of continuously operating pilot flames by e.g. spark
ignition would result in a saving of 10 %,, and if the flue gas exhaust
duct were closed off when the burner was switen off, a further 10 % could
be saved.
Recovery of waste heat from flue gases from ovens using a flue gas
conomiser is another technical possibility. The cost price of acqui-
sition and installation of a flue gas conomiser and ancillary equipment
for the ovens in the various bakeries is f 12,000, - f 15,000,. The
use of an conomiser can result in a nett recovery of appr. 55 % of the
energy contained in the flue gases. The use of an conomiser will largely
depend on the extent to which the recovered energy can be re-used and
the degree of financial advantage that will result.
5.2. Electricity
In the bakeries that were investigated 30 - 55 7. of the total elec-
tricity consumption went on freezing and refrigeration. In many bakeries
a saving of appr. 19 % could be obtained by disconnecting the unnecessary
-208-
door heating. Better and regular maintenance of the refrigeration equip
ment would also result in most cases in a saving of 10 15 %.
ENERGY BA LA NCE FOR BA KING OVENS
energy input baking area :

nominal 14 m*
actual used 80 /.
product: 200kg white panbread
energy consumption in V
due exhaust 32,2 /.
water evaporation 21,5/

//////>
*/'///'. Into product 12,0/.
f'///; Into pans 4,6 /.
{////////
{"'/'/'//' transmission loss 29,7/
' ' / / , / .'.'II
wy/u'A
energy used for
baking 38,1 /139MJ
energy input baking area :

nominal 7,3 m'
actual used 100 /.
product: 103kg white panbread
energy consumption in /
flue exhaust 25 /.
energy loss
through chimney
4 8 / . 124MJ oven exhaust 7 /.
water evaporation 16 /.
into product 6,3/.
into pans 4,3/.
energy loss
in bakery transmission loss 39,4/.
5 2 / . 134 M J
energy used for

baking 2 8,6/.. 74 M J
209-
ANNEX I.
Examples of calculations of energy saving in the baking process.

Bakery no. 1 : bread production in flour 1461 kg/week
67 kg/week
confectionery production in dough
1528 kg/week
total
total energy consumption per week 8158 MJ

lowest measured energy consumption 281 MJ/100 kg
, ,t , . 1528100
281 4294 MJ
calculated energy consumption
cooling losses + 1478 MJ
+ 168 MJ
total 5940 MJ
8158 - 5940
Possible energy saving 100 = 27.2
8158
Bakery no. 18:
Oven no. 1: bread production in kg flour 1215 kg/week
(electrically heated) confectionery production in kg dough 162 kg/week
total 1377 kg/week

total energy consumption 3206 MJ/week
, , _ , ... 1377 1 17 MJ
calculated energy consumption 1611 MJ
100
12 MJ
total 2227 MJ
3206 2227
Possible energy saving 100 % = 30.5 %
3206
0ven_no^_2: bread production in kg flour 1468 kg/week
(gas-fired) confectionery production in kg dough 179 kg/week
total 1647 kg/week

total energy consumption 9176 MJ
- A '_ '647 299
4925 MJ
calculated energy consumption 100
8 MJ
total 6315 MJ
9176 1 5
Possible energy saving " 100 % = 31.2 %
-210-
Table 1 . DUTC H BREADBAKERIES DIVIDED INTO GROUPS WITH BATC H OVENS AND
CONTINUOUS OVENS.
Bakeries with batch ovens
Number of
Number of Bread production National bakeries for
bakeries tons/year bread production energy
X investigation
900 0 60 5
1600 60 120 17 1
700 120 200 12 2
450 200 400 14 2
150 400 1200 11 2
3800 59 7
Bakeries with continuous ovens
40 400 1200 3i
19 1200 2000 3!
19 2000 2800 5
16 2800 4000 5}
27 > 4000 23J
121 41 5
Table 2. FLOW DIAGRAM BREAD AND ROLLS PROC ESS
electric power Is used in oil stages
bulk handling
ol Hour
- mixing

I dough transpori

dividing
rolls and buns
rounding
fermentation
moulding
steam | proofing
J baking pans
not. gas "
baking
steam T_
depanning
slicing
wrapping
freezing fresh bread
-211 -
Table 3.
Process time in minutes

Processphase Conventional Chemical dough
process development
mixing 10 - 25 5 - 10
bulk fermentation 20 - 45
dividing-rounding + +
fermentation 45 - 30 15 - 30
moulding + +
proofing 60 - 80 60 - 80
baking 30 - 50 30 - 50
total process time 3 - 4 2 - 3

in hours
Table 4. PRODUCTION AND ENERGY CONSUMPTION PER WEEK.
Bakery code 1 5 8 9 IB
Production in kg
Bread 2170 2520 4375 7310 3985
Confectionary 100 420 320 410 340
Energy consumption in MJ
Bread i nel. baking 7418 9414 11912 20757 11838
Confectionary i n d . baking 1599 1591 1444 2095 1523
Hot water 439 394 355 1169 877
Heating system 807 1144 2253 1271 1335
Refrigeration and freezing 617 1287 727 1153 655
Miscellaneous 281 165 140 351 358
Total 11163 13946 16830 26791 16563
Electricity 1478 2227 1328 4209 5015
Natural gas 9685 11719 15502 22582 11548
-212-
Table 5. ENERGY CONSUMPTION (IN HJ ANO IN PERCENTAGES).
Bakery code 1 5 8 9 18
Baking energy bread 7039 8842 11439 18927 10993

Baking energy confect. 1495 1452 1315 1693 1389
Non specified energy 2629 3652 4076 6171 4181
Total used energy 11163 13946 16830 26791 16563

Electricity 1478 2227 1328 4209 5015
Natural gas 9685 11719 15502 225B2 11548
Baking energy 76.5 73.8 75.8 77.2 74.8

Electricity % 13.2 16-0 8.0 15.7 30.0
Natural gas % 86.8 84.0 92.0 84.3 70.0
Table 6. ELECTRICAL ENERGY CONSUMPTION.
Mixing 8.9 6.5 14.6 10.9 6.8

Dough processing 10.0 1.5 4.7 19.0 5.0
Proofing % 8.4 - 13.0 1.0 17.6
Depanning + pan greasing I - - - 1.2 -
Slicing/wrapping X - - 0.3 - -
Confectionary machines t 1.9 2.9 7.0 4.4 2.9
Refrigeration/freezing 1 56.0 64.4 54.8 30.8 40.1
Lighting X 14.8 21.9 5.6 15.7 27.6
Miscellaneous X - 2.8 - 17.0 -
Sub total * 100 X kWh 306.1 555.0 368.7 1041.1 453.6
Electricity baking
ovens kWh 104.6 63.7 - 274.8 939.0
Total kWh 410.7 618.7 368.7 1315.9 1392.6
Pan washer
Table 7. NATURAL GAS CONSUMPTION.
Natural gas in MJ 96B5 11719 15502 22582 11548
Cooking X 2.9 1.4 0.9 1.6 3
Heating of water X 4.5 3.0 2.3 5.2 7.6
Heating system X 8.3 9.8 14.5 5.6 11.6
Baking ovens X 84.3 85.8 82.3 87.6 77.8
-213-
Table Relation between oven temperature and energy consumption
3 0 0 *C
oven temperature
T^Ni-
y\
200 /
/
/
/
100 y
1--- I I" !-'|".;.:.:.;.|>enerav consumption
h e a t i n g up
A . .a baking
| baking e n e r g y - D A
e n e r g y loss b e f o r e baking = C
Table 9. ENERGY CONSUMPTION OF B AKING OVENS AND B AKED PRODUCTS PER WEEK.
BAKERY CODE 1 5 a 18
1 2
OVEN NO. (3 days) (3 days)
1 + 2 1 2
BREAD in kg flour 1461 491 372 1696 2945 1215 1468
CONFECTIONARY in kg dough/batter 100 48 422 318 162 179
ENERGY CONSUMPTION OVEN IN MJ 8158 3177 1050 10294 12754 3206 9176
ENERGY CONSUMPTION divided into:

1. Losses before baking 745 347 192 639 352 1282
2. Baking bread 4648 2162 913 8344 1903 5450
3. Baking confectionary 1119 63 608 1315 335 1054
4. Cooling losses after baking 1646 397 212 2456 616 1390
5. Pilotflame in nonbaking period 208 173 1030
Energy consumption per 100 kg
flour for breadbaking:
Monday MJ/100 kg 300 289 133 453
Tuesday " " 319 300 134 431
Wednesday " " 281 496 228 257 144 447
Thursday " " 279 468 284 365 117 352
Friday " " 306 382 231 224 212 299
Saturday " " 369 275 188 356
-214-
DISCUSSION
Question - Chairman : Concerning the interest of bakeries in energy

conservation, was this their first involvement with energy
conservation scheme ?
Answer Yes - in general the bakeries had not been aware of their
energy losses prior to our visits.
Question - Chairman : As an alternative to the economizer, did you

consider use of a spray recuperator, which could be cheaper
Answer No, although this would be a possibility.
Question : The amount of fluorescent light appears to be high in some

of the plants.
Answer Yes, this is overdone in some cases.
Question : The energy use for refrigeration appears to be high in the

sectors.
Answer Yes, a good technique for detecting when the frosting level
on the coil is high would be useful in contributing to
savings in this area.
Question : Do you see any applications for microwaves in this area ?
Answer It would be cheaper.

ENERGY SAVING IN THE SOYBEAN EXTRACTION INDUSTRY BY REDUCING THE STEAM
CONSUMPTION FOR DESOLVENTIZING - TOASTING AND DRYING EXTRACTED BEANS
Author : Ir T.L. Ong
Contract number: EEB-I-I38N (B)
Duration : 36 months 1 January I98I - 31 December 1983
Head of project: Ir T.L. Ong
Contractor : The Netherlands Central Organization for Applied

Scientific Research TNO
Address : Institute CIVO-Technology TNO

Utrechtseweg k
370U HE ZEIST
Summary
In the soybean extraction process, steam is required for heat-con-

ditioning, extraction, desolventizing-toasting, drying and for distil-
ling hexane out of the oil-in-hexane solution.
The aim of the research is to reduce the steam consumption in the
following operations:
a. in the toasting operation by about 20 kg/ton of beans, by combining
this operation with the heat-conditioning step and utilizing com-
bustion gases instead of steam;
b. in the drying operation by about 30 kg/ton of beans, by desolven-
tizing the defatted flakes by means of superheated hexane vapour,
resulting in soybean meal of low water content which hardly needs
any drying.
In this context the optimal specific energy consumption of the pro-
cesses concerned has been assessed. A considerable amount of steam
can be saved if toasting and desolventizing could be accomplished in a
relatively dry atmosphere. In this context a test circuit for toasting
soybeans at low moisture contents has been designed and constructed.
A survey has been made of the average available quantity and also of
the important minor components of the combustion gases from steam-
boilers fired with'natural gas, fuel oil and coal respectively. The
combustion gases of fuel oil and coal are not suitable for direct
heating soybeans, because they contain considerable amounts of hydro-
carbons and fly ash.
-216-
1 .1 Introduction
In the EEC soybeans rank at the top of the list of oilseeds being ex-
tracted. The amount of soybeans and other oilseeds processed in the EEC is
presented in table 1.
For the production of soybean oil and meal, soybeans are consecutively:
cracked, dehulled - in making soybean meal 1+9 % protein, heat-conditioned,
flaked and extracted by means of hexane; the defatted flakes, containing
30-Uo % hexane, is subsequently desolventized and toasted - to improve the
nutritional value - by means of steam; finally the wet extracted meal is
dried by means of hot air (see figure 1 ) .
In this process steam is required for heat-conditioning, extraction, desol-
ventizing-toasting, drying and for distilling hexane out of the oil-in-
hexane solution (miscella).
In the manufacture of low-phosphorus degummed soybean oil, the flakes
- prior to extraction - are cooked at a moisture content of about 16 % and
a temperature of about 110 C - to partially deactivate the enzymes - and
subsequently dried to about 11 % moisture and cooled to about 60 C. These
flakes are extracted in the usual way and - after desolventizing - still
have to be toasted to improve its nutritional value (see figure 1 ) .
The aim of the research is to reduce the steam consumption in the fol-
lowing process steps:
a. the toasting process by about 20 kg/ton of beans, by combining this oper-
ation with the heat-conditioning step, and utilizing combustion gases
instead of steam;
b. the drying process by about 30 kg/ton of beans, by desolventizing the
defatted flakes by means of superheated hexane vapour, resulting in soy-
bean meal of low water content which hardly needs any drying.
In this context it is essential to assess the optimal specific energy
consumption in the processes concerned. Consequently, in the period
1 January 1981 - 31 March 1982 the optimal specific energy consumption of
the following processes has been assessed:
a. heat-conditioning soybeans;
b. desolventizing defatted flakes by means of steam;
c. desolventizing defatted flakes by means of superheated hexane vapour;
d. toasting soybeans;
e. drying desolventized-toasted soybean meal.
At the same time, in accordance with the pertinent research programme,
the following have also been accomplished:
a. design and construction of a test circuit for toasting soybeans by means
of hot vapours;
b. design of a test circuit for desolventizing defatted soybeans by means
of superheated hexane vapour ;
c. survey of the available quantity and the properties of the combustion
gases from steam-boilers fired with natural gas, fuel oil and coal res-
pectively.
1 .2 Description
1.2.1 Optimal energy consumption for heat-conditioning soybeans
Heat-conditioning of cracked soybeans at 60-80 C - prior to flaking -
is necessary to plasticize the kernel particles, so good flakes can be
formed. The kernel particles are flaked by passing through a set of smooth
rolls travelling at different speeds. The assessment of the energy consump-
tion for heat-conditioning soybeans was made for the standard vertical
stacked kettle type and was based on the following process conditions:
plant capacity of 1000 tons per day,
1000 kg soybeans (12 % water) yields: 180 kg oil + 820 kg meal (15 %
water),
cracked "beans entering heat-conditioner at 10 C and 12 % moisture,
cracked beans leaving heat-conditioner at 12 % moisture and at tempera-
tures of 60 C, TO C and 80 C respectively.
The results are summarized in table 2.
1.2.2 Optimal energy consumption for desolventizing defatted flakes by means
of steam
Soybean flakes are normally extracted with hexane at about 50 C.
The assessment of the energy consumption for desolventizing defatted flakes
by means of steam was based on the following process conditions:
plant capacity: 1000 tons per day;
temperature of defatted flakes leaving extractor: 50 C;
hexane content of defatted flakes: 30 %, 35 % and 1+0 % respectively;
specific heat (Cp) of hexane: 2,5 kJ/kg.K;
specific heat (Cp) of soybean meal containing 12 % water: 2.0 kJ/kg.K;
steam condensing on the flakes: 101.3 kPa and 100 C;
temperature of vapours leaving desolventizer: TO C, T5 C, 80 C and
85 C respectively.
It is normal practice to blow open steam onto the hexane-wet defatted
flakes, so as to desolventize the material. The steam condenses on the
flakes and the condensation heat evaporates the hexane.
The energy/steam consumption for desolventizing defatted flakes, and the
resulting water content of the desolventized material are shown in table 3.
1.2.3 Optimal energy consumption for desolventizing defatted flakes by
means of superheated hexane va-pours
A flowsheet of the proposed desolventizing system is shown in figure 2.
The main elements of the system are the desolventizer no. 1, the flake
separator no. 2, the circulating blower no. 3 and the vapour superheater
no. h. It is intended to utilize the excess hexane vapours for the heating
coils of the 1st evaporator.
It can be assumed that the circulating vapours principally consist of
hexane, with traces of water and air; they are heated in the superheater to
150 C. The amount of heat and superheated hexane of 150 C for desolven-
tizing defatted flakes are shown in table h. Operating parameters are:
- hexane content of extracted flakes: 30 %, 35 % and Uo % respectively;
- temperature of vapours leaving desolventizer: TO C, T5 C, 80 C and
85 C respectively.
1.2.1* Optimal energy consumption for toasting soybeans
Soybeans are toasted (cooked) to improve their nutritional value; the
process is normally carried out in a steam atmosphere at about 105 C.
The temperature of the flakes leaving the desolventizing section of the
desolventizer-toaster (DT) is raised to about 105 C. The heat for denatu-
ration the enzymes and proteins in the beans has been estimated at Q0 kJ/kg.
The amount of heat for toasting flakes, depending upon the desolventizing
temperature, is shown in table 5
1.2.5 Optimal energy consumption for drying desolventized-toasted soybean
meal
Depending on the hexane content of the defatted flakes and the desol-
ventizing temperature, the water content of the desolventized flakes
amounts to 21-26 % (cf. table 3 ) .
The assessment of the steam and air consumption for drying the desolventized-
toasted soybean meal was based on the following:
- meal dryer in which the drying air is reheated inside the apparatus;
- meal entering dryer at 100 C and 21-26 % moisture;
- meal leaving dryer at TO C and 13 % moisture;
- air entering dryer at 10 C, 80 % RH, O.OO62 kg H20/kg dry air;
-218-
air leaving dryer at TO oc, 70 % RH, 0.2093 kg H20/kg dry air.
The results of the assessments are summarized in table 6.
1.2.6 Design and construction of a test circuit for toasting soybeans fry
means of hot vapours
The aim of the research is to reduce the steam consumption in the
toasting process by combining this operation with the heatconditioning
step. The process should preferably be carried out at the lowest possible
temperature to avoid damage to the essential amino acids (lysine and
methionine) and at the lowest possible moisture content so as to avoid
the necessity of drying. It has been reported that optimum conditions for
flaking cracked beans are: 1011 % moisture content, and about 70 C tem
perature. Besides, the temperature of the flakes entering the extractor
should not exceed the boiling point of the solvent hexane i.e. about 69 C
to avoid excessive pressure in the apparatus. Based on these considerations
a proposed block diagram for the preparation of beans prior to extraction is
shown in figure 3. In this context it is crucial to know the optimum temper
ature and relative humidity of the vapours for the preheating, toasting and
holding operations respectively, in relation with the contact time. In this
connection a test circuit for toasting as shown in figure h has been
designed and constructed.
1.2. Design of a test circuit for desolventizing defatted soybeans by
means of superheated hexane vapour
The objectives of the experiments with this test circuit are to find
out :
a. the optimum temperature of the hexane vapour;
b. how far toastedextracted flakes can be desolventized by means of hexane
vapour ;
c. residual hexane and moisture content of the desolventized flakes;
Based on these considerations a test circuit as shown in figure 5 has been
designed.
1.2.8 Survey of the average available quantity and of the important minor
components of the combustion gases from steamboilers, fired with
natural gas, fuel oil and coal respectively
The amount and properties of the exhaust gases of steamboilers, fired
with natural gas, fuel oil and coal respectively have been measured by the
Heat and Refrigeration Engineering Division of TNO Apeldoorn. This div
ision has a special coach equipped with instruments for continuous analyzing
stack gases for: nitrogen oxides, carbon monoxide, soot, dust, sulfur oxides,
temperature etc.
1.3 Results
The results of the assessments of the optimal energy consumption for:
heatconditioning soybeans, desolventizing defatted flakes by means of steam,
desolventizing defatted flakes by means of superheated hexane vapours,
toasting soybeans and drying desolventizedtoasted soybean meal are summar
ized in the tables 2,3,^,5 and 6 respectively.
The results of the survey of the average available quantity and of the
important minor components of the combustion gases from steamboilers, fired
with natural gas, fuel oil and coal respectively are summarized in table 7
1.h A nalysis of results and comments

The steam consumption based on 1 ton of beans in the most important
operations of today's extraction process are:
1*5 kg for heatconditioning at 70 C,
96 kg for desolventizing defatted flakes (35 % hexane) at 80 C,
219
59 kg for toasting defatted flakes at 105 C,
12 kg for drying the meal from 2l* % to 13 % moisture.
The total steam consumption based on 1 ton of "beans is 32 kg.
The main features of the new proposed process are:
a. combining heatconditioning and toasting;
b. desolventizing defatted flakes with superheated hexane; probably an ad
ditional desolventizing (stripping) with superheated steam or some inert
gas (nitrogen or carbon dioxide) will be necessary.
The steam consumption based on 1 ton of beans in the most important oper
ations of the new proposed process will be:
79 kg for heatconditioning at 110 C,
39 kg (90 MJ) for denaturation/toasting the enzymes and proteins,
97 kg (22J MJ) for desolventizing defatted flakes (35 % hexane) at 80 C,
6 kg (estimated) for stripping residual hexane in the meal.
The total steam consumption based on 1 ton of beans is 221 kg, which is
106 kg less than that of today's process.
However, several questions remain to be answered; among them, the most im
portant are:
a. is it possible to toast the beans at 12 % moisture;
b. what are the optimal conditions for toasting with regard to holding time,
temperature and moisture;
c. how is the extractability of the toasted materai.
The results of the survey of steamboilers fired with several kinds of
fuels indicate that the combustion gases of heavy oil and coal contain con
siderable amounts of hydrocarbons and fly ash.
2. Conclusions
The steam consumption for drying the desolventizedtoasted meal is
considerable. In the example given in chapter 1.1 it was assumed that the
defatted flakes contain 35 % hexane; desolventizing at 8o C with open steam
resulted in flakes with 2k % water, in which case 127 kg of steam/ton of
beans was needed to dry the material to 13 moisture content. However, it
also happens in some extraction plants that the defatted flakes contain 1*0 %
hexane, and that desolventizing takes place at 85 C. In this case the de
solventized flakes will contain 26 % water and 15** kg of steam/ton of beans
will be needed to dry the flakes to 13 % moisture content. Therefore a con
siderable amount of steam can be saved if desolventizing could be ac
complished in a relatively dry atmosphere.
Besides, it is necessary to toast the beans at the lowest water content

feasible, so as to avoid the necessity of drying. Experiments with the newly
constructed toaster should disclose the optimum conditions for toasting the
beans.
The stack gases of steamboilers fired with heavy oil and coal contain
considerable amounts of hydrocarbons and flyash. This will prohibit the
use of these gases for direct heating soybeans, since the meal and oil are
normally intended for edible purposes.
Table 1 Soybeans and other oilseeds processed in the EEC in the years
1980 and 1981
1980 IQ81
soybeans other total soybeans other total
country oilseeds oilseeds
(io3t) (io3t) (I0 3 t) (I0 3 t) (io3t) (io3t)
Belgium/Luxembourg 812 100 912 1220 200 11+20

Denmark 32 78 1*05 180 93 273
France 856 977 1833 5H6 IO36 1582
West Germany 3791* 1929 5723 3068 2030 5098
Ireland 2 6 8 3 5 8
Italy 1393 In 8 1811 1293 78 1371
Netherlands 3002 201 3203 2686 326 3012
United Kingdom 1138 61*2 1780 1059 1*71 1530
Total II32I 1*351 15675 10055 1*239 11*291+
Table 2 Optimal energy consumption for heatconditioning soybeans of 10

moisture and 10 C, at temperatures of 60 C , 70 C and 80 C
respectively
Heat conditioning Energy consumption Steam consumption

temperature (C) (MJ/ton beans) (kg/ton beans)
60 83 36
70 103 1*5
80 123 53
221
Tatole 3 Optimal energy and steam consumption for desolventizing extracted
flakes, as well as the resulting water content of the desolven
tized flakes; operating parameters:
hexane content of extracted flakes: 30 %; 35 %; ko %
- temperature of vapors leaving desolventizer: 0 C, T5 C,
80 C and 85 C
1
hexane content energy consumption steam consumption water content

of extracted to evaporate (101.3 kPa, 100 C) of desolven
flakes {%) hexane (MJ/ton toeans) to evaporate hexane tized flakes
(kg/ton heans) (%)
0
temperatu re of vapors leaving des olventizer: 70 C
30 167 70 21
35 202 85 23
ItO 21+2 102 21*
temperature of vapors leaving desolventizer: 75 C
30 178 76 22
35 21k 90 23
l+o 255 108 25
30 188 81 22
35 22U 96 21
1+0 266 i iU 25
30 200 86 23
35 237 102 2U
Uo 28O 121 26
222
Table h - Amount of heat and superheated hexane of 150 C for desolventizing extracted flakes;
operation parameters are - hexane content of extracted flakes: 30 %, 35 % and Uo
temperature of vapors leaving desolventizer: 70 C, 75 C, 80 C and 85 C
temperature of vapors leaving desolventizer ( C):
70 75 80 85
hexane
content of hexane hexane hexane hexane
extracted heat vapors heat vapors heat vapors heat vapors
flakes {%) MJ/t. beans kg/t. beans MJ/t. beans kg/t. beans MJ/t. beans kg/t. beans MJ/t. beans kg/t .beans
30 167 837 ' 178 952 188 1076 200 1233

35 202 1010 211* 1139 22U 1280 237 1U59
Uo 2U2 1211 255 1359 266 1518 280 1722
Table 5 - Optimal energy consumption for toasting flakes - containing
about 22 % water - of TO C, 75 C , 80 C and 85 C respectively
at 105 C
desolventizing amount of heat amount of steam

temperature for toasting flakes (101.3 kPa, 100 C)
(C) (MJ/t. beans) for toasting flakes
(kg/t. beans)
TO 153 6T
T5 1M 63
80 135 59
85 126 55
Table 6 - Optimal steam and air consumption for drying desolventized-

toasted soybean meal of 21-26 % moisture down to 13 % moisture
moisture content steam consumption air consumption

of meal {%) kg/t. beans kg/t. beans
21 8T 105
22 100 160
23 113 520
21 127 580
25 110 6io
26 15U T05
Table 7 - Average available quantity as well as important minor components

of combustion gases from steam boilers, fired with natural g a s ,
fuel oil and coal respectively
emission (mg/fuel unit)

stack
gases
sulfur nitrogen (nn/
fuel carbon oxides oxides hydro- fly- fuel
fuel unit monoxide as S0 as N0 carbons ash unit)
3
natural gas 30-300 15 1000-1700 10-60 10 12-13
nm
heavy oil grade 6 kg 300 50.000 5000 300 2000-3000 15
coal kg 500 12.000 10.000 150 I5.OOO 16
-224-
Figure 1 Flow diagram of a soybean extraction plant producing soybean
meal and crude degummed oil of standard (200 ppm ) and low
(20 ppm ) phosphorus content
soybeans
V7
cleaning
cracking
(dehulling)
cracked beans
tempering/conditioning at 6 80 C
J'
flaking

full fat flakes
for producing oil for producing oil of

of low phosphorus standard phosphorus
content content

:
wetting and tempering
at 110 C
i extraction
drying . I
defatted flakes full miscella
desolventizingtoasting filtration
i .
drying distillation
grinding
I 1.
degummmg
cooling
soybean
* meal 1
crude degummed
1*1* % or 1*9 % protein drying
soybean oil
-i> phosphorus
standard or low
hexane vanors
to heating
coils of
1st evaporator
=* en)
a s scrubbing
liquid
hexane HDXIHD=
vapors
fi
X
to scrubber
4 discharge lock
flakes
from
extractor
Hl S
Y superheated
hexane
jfe.
K3
desolventized
flakes to
flake stripper
Figure 2 - Desolventizing system by means of superheated hexane
1 desolventizer h vapor superheater

2 flake separator 5 system pressure control valve
3 circulating blower f< vapor scrubber
226-
Figure 3 - Proposed block diagram for the Dreparation of
soybeans prior to extraction
flow of soybeans
flow of heating gases
-.-.-.- cooling air
I cleaned beans
[conditioned to 12 moisture
cracking
preheating
toasting
heater
holding
relative
humidity
adjustment
z
precooling
I
cooling
1
toasted cracked
beans 60-70 C
11-12 % moisture
to flaking and extraction
-227-
flow Tiixing relative ie iter beans
meter heater chamber humidity toaster
adjustment
O O
air
C7\ p , out
ill ss V

steam
Figure U - Test circuit for toasting soybeans
flow mix]ng heater extracted condensor

meter chamber flakes
desolventizer
^ exhaust
hexane
Figure 5 Test circuit for desolventizing extracted soybeans

DISCUSSION
Question - Chairman : Is the problem with hydrocarbons similar to that

experienced in malting plant in the UK.
Answer Yes.
Question - Chairman : Soyabean manufacturing is decreasing. Is this

because soyabeans are being replaced by other products ?
Answer No. The decrease is due partly to the recession and partly
due to the rise in value of the US dollar. A soyabean plant
in the UK has recently been closed.
Question - Dr. Pilavachi, CEC : What is the hexane content ?
Answer In soyabean meal it is 0.05% or less. In soyabean oil it

should be lower than 0.1%.
-229-
ENERGY SAVING IN EDIBLE OIL PROCESSING PLANTS
BY APPLICATION OF A TOTAL-ENERGY (TE) SYSTEM
PART I: SOYBEAN OIL EXTRACTION PLANT
Author : Ir T.L. Ong
Contract number : EE-Dl-303-N(B)
Duration : 18 months 1 October 1981 - 31 March 1983
Head of project : Ir T.L. Ong
Contractor : The Netherlands Central Organization for Applied

Scientific Research TNO
Address : Institute CIVO-Technology

Utrechtseweg 48
3704 HE ZEIST
Summary
In public electricity generation stations only 34-40 % of the total
primary energy input is converted Into electricity; the major part
of the energy input is lost in the transport system and particularly
with the low pressure exhaust steam from the turbines.
The aim of this research is to investigate whether savings in pri-
mary energy carriers by application of a total-energy system can be
implemented in the following types of edible oil processing plants:
a. soybean oil extraction; b. refinery; c. margarine manufacture.
This preliminary report is primarily concerned with the soybean oil
extraction plant. In this context the actual ratio of the optimal
steam/power consumption and also the operating times of the steam
and electricity consuming units have been determined in a soybean
oil extraction plant of 1000 tons per day capacity. It was found
that In this plant the heat/power consumption ratio (6.9) and the
number of operating hours per year (6000-7200) are high; these
findings are favourable for the installation of a TE-system.
-230-
1.1 Introduction
It is well known that in public electricity generation stations only
34-40 % of the total primary energy input is converted into electricity;
60-66 % of this energy input is lost. A small part of the losses is due
to the transport system, such as resistance in the cables and trans-
formers; however, the major part is in fact waste heat discharged with
the low pressure exhaust steam from the turbines.
Efforts to utilize this waste heat for town heating are not always suc-
cessful, because of the high investments for transporting the hot water
over long distances.
On the other hand edible oil processing plants - like extraction
plants, refineries etc. - utilize huge amounts of low pressure steam
(3-10 bar), beside electricity.
In the past the application of total-energy (TE) systems was feas-
ible only in big plants, having a capacity of at least 10 MW. However
recently reliable small TE-units, driven by gas or diesel engines become
available. Although the mechanical efficiency of these small units
(0,3-1,0 MW) is lower (< 30 %) than that of a big power station
(1000 MW), the application of a small unit In edible oil processing
plants could be advantageous because of the possibility to utilize the
waste heat, raising the total efficiency - based on the primary energy
input - to about 70 %, which is nearby two times that of a big power
station. Therefore substantial savings on primary energy can be achieved
when an edible oil processing plant generates its own electricity and at
the same time can utilize the waste heat on the spot.
However several questions have to be answered first, before the instal-
lation of a TE-system can be considered; the most important are:
a. choice of the proper type and capacity
b. the necessary investments
c. achievable savings
d. payback period
The aim of this research is to investigate whether savings in pri-
mary energy carriers by application of a total-energy system can be im-
plemented in the following types of edible oil processing plants:
a. soybean oil extraction
b. refinery
c. margarine manufacture
To achieve these goals the following data should be determined and
measured in each type of plant mentioned above:
a. the actual ratio of the optimal steam/power consumption
b. a breakdown of the operating times of the steam and electricity con-
suming units
c. based on the determinations and measurements of a. and b., the proper
type and capacity of TE-system (steam turbines, gas or diesel engines)
can be chosen
d. the necessary investments for the TE-unit, building, safety and noise
reducing measures
e. the energy and cost savings, including the payback period of the in-
vestments
f. safety and environmental, technical and non-technical barriers
In order to obtain the necessary actual data a close co-operation
with a soybean extraction plant, a refinery and a margarine plant re-
spectively is crucial.
In approaching several companies to cooperate in this project, a soybean
extraction plant (capacity about 1000 tons per day) and only recently
also a margarine plant (capacity about 200 tons per day) were found pre
pared to make this investigation possible. Consequently, in the period
1 October 1981 31 March 1982 the investigation could be carried out
only in the soybean extraction plant mentioned above.
1.2 Determination of the actual optimal heat/power consumption ratio in a

soybean extraction plant of 1000 tons per day capacity
1.2.1 Description of the soybean extraction process
Soybeans are normally processed to produce oil and meal by the pro
cesses shown in figure 1.
After cleaning the beans, they are adjusted to about 10 % moisture and
cracked into 48 pieces. For the preparation of high protein meal, the
hulls are removed by air classification. The cracked beans are softened
by tempering/conditioning with steam and subsequently rolled into flakes
having a thickness of 0.20.4 mm. The flakes are extracted in an extrac
tor in countercurrent with miscella of diminishing strengths and finally
with a wash of clear hexane to reduce the residual oil content to < 1.0 %.
The defatted flakes, which contain 3040 % hexane are desolventized and
subsequently toasted in a socalled desolventizertoaster (DT). The meal
discharging out of the DT at over 100 C and at a moisture of 1822 %
is usually dried to 1214 % moisture. The dried meal is consecutively
ground, cooled and sent to storage.
The socalled "full miscella" from the extractor, containing 2035 % oil
in hexane is filtered; the solvent is removed in consecutively the first
and the second stage evaporators.
1.2.2 Determination of the actual optimal heat/power consumption ratio

The extraction of soybeans is normally a continuous process; this
means that in normal operation (except during breakdowns, shutdowns and
startups) the steam and power demands are constant.
The plant under consideration has just finished a steam conservation pro
gramme, consisting of improving insulation, installing economizers, re
covery of the heat of the air leaving the meal dryer etc. These resulted
into a reduction of the natural gas consumption per ton of beans from
38 m 3 to 22 m 3 .
The plant normally operates 250300 days 24 hours = 60007200 hours a
year.
As described in chapter 1.2.1 the most important energy consuming oper
ations are: cracking, tempering/conditioning, flaking, extraction, desol
ventizing/toasting, drying, grinding, distillation/evaporation and obvi
ously pumps and conveyors for transporting beans, flakes and meal.
Besides, ventilators to keep the dust and hexane concentration in the
buildings below the maximum allowable concentration (MA C value) consume a
lot of electricity.
The energy consumption of the apparatuses concerned have been determined
by means of standard steamflow and kWhmeters during 2 consecutive weeks
at a production of 40 tons of beans per hour.
1.3 Results
The results of the determinations of the energy consumption of the
apparatuses concerned, expressed in MJ/ton of beans, are summarized in
table 1.
232
1.4 Analysis of results and comments
The heat/power consumption ratio in the extraction plant under con
sideration is 693/100.8 = 6.9 (cf. table 1 ) . Since the average hourly
production is 40 tons, the generator should have a capacity of 100.8
40/3.6 = 1120 kW.
There are two possible TEsystems viz.:
a. a gasturbine of for example KHDBRD type MIA 0 1 , capacity 1100 kW
b. three gaspiston motors of for example KHDBRD type B/GAM 816, ca
pacity 400, 400 and 300 kW respectively.
The gasturbine consumes 629 m^ natural gas per hour, yielding 1100 kWh
electricity and 28000 kg exhaust gases of 500 C. When these gases are
cooled in a special steamboiler to 180 C, the available heat for
steam production will be about 9000 MJ.
The overall yield of the system is (1100 3.6 + 9000)/(629 31.5) =
0.65. The three piston motors consume 378 m^ natural gas per hour,
yielding 1100 kWh electricity and 8000 kg exhaust gases of 600 C. When
these gases are cooled in a special steamboiler to 180 C, the avail
able heat for steam production will be 3400 MJ. The overall yield of the
system becomes (1100 3.6 + 3400)/(378 31.5) = 0.62.
Contrary to the gasturbines, the piston engines need a cooling water sys
tem; a considerable part of the fuel energy is lost with the cooling
water. This makes the overall yield of the piston motors somewhat lower
than that of the gasturbine. On the other hand the piston motors produce
less exhaust gases (8000 kg) than the turbine (28000 k g ) .
Conclusion
The high heat/power consumption ratio (6.9) and the high number of
operating hours (60007200 hours/year) in the extraction plant under con
sideration are favourable for the installation of a TEsystem. The inves
tigation will be continued to determine which type of TEsystem (turbine
or piston motor) is the most suitable. Besides, the necessary investments
and the payback period still have to be calculated.
Table 1 Breakdown of the energy consumption in a 1000 tons per day

soybean extraction plant MJ/ton of beans
heat (MJ) electricity (MJ)*)

cracking 7.2
tempering/conditioning 119 3.6
flaking 36.0
extraction 40 3.6
desolventizing/toasting 297 7.2
drying 158 7.2
grinding 7.2
distillation (including usage of
the heat of the vapours from the DT) 79
pumps, conveyors etc. 14.4
ventilators 14.4
693 100.8
*) 1 kWh = 3.6 MJ
233
Figure 1 - Flow diagram of a soybean extraction plant
soybeans
"T"
cleaning
cracking
1
(dehulling)
I
cracked beans
I
tempering/conditioning
I
flaking
extraction - full
miscella
defatted flakes filtration
deeolventizing-toasting distillation
drying |crude soybean oll|
grinding
cooling
soybean meal 44
or
soybean meal 49
-234-
DISCUSSION
Question : What are the relative capital costs of the plant, and have
you considered producing steam at higher pressure and
passing it through a turbine to produce electricity ?
Answer Prices of turbines are Lower than piston engines. The high
capacity of the exhaust gases puts the price up again,
however. Also, for a high pressure boiler, we would require
a water treatment system, which would be very expensive.
Question - Dr. Pilavachi, CEC : Where have total energy systems been used?
Answer A brewery has installed a total energy system, and glucose

and starch plants and sugar refineries have such instal-
lations. We have found that total energy plants would not
be of interest in margarine plants.
-235-
SESSION V - CHEMICAL INDUSTRY
Chairmen: J. LIMIDO and J.A. KNOBBOUT
Summary of Session V
Modification of cracking furnaces of existing plants to increase

yields of valuable products and to reduce fuel consumption
Selective removal of HJ3 from gaseous mixtures containing C 0 ?
Energy conservation in the chlor-alkali industry
Thermal depolymerisation of waste tires by heavy oils

- Conversion into fuels
A critical evaluation of anaerobic fermentation of waste products

(A management study)
The recovery, treatment and utilization of solid bearing

effluents emanating from coke ovens
-237-
SUMMARY OF SESSION V
THE CHEMICAL INDUSTRY
J.A.A. Ketelaar , J.A. Knobbout, J. Limido, P.A. Pilavachi
A certain number of contracts were grouped together under the heading

"CHEMICAL INDUSTRY"; the common feature here being their relevance to
chemistry or chemical engineering. This was a logical step to take even
though they had different objectives : the reduction in energy consumption
or the access to new sources of energy, fossil or otherwise. This link
between the different problems to be solved lead to lively discussions
among the participating contractors.
Energy Saving
Two projects stand out :
Cracking furnaces - The first concerns research work undertaken by

MONTEPIDE to improve the performance of steam cracking installations which
form the basis for petrochemicals production. This research project aims
at increasing the flexibility of the feedstock and the selectivity of the
olefin by using an improved coil design. At present, the new experimental
coil is under construction and testing will follow. It is expected that
the quantity of feedstock required and the energy consumption can be
reduced. In addition, substantial cost savings can be achieved, especially
in view of the rising cost of both raw materials and energy.
Chlor-alkali industry - The CITY UNIVERSITY has undertaken the second

project, which aims at reducing the energy consumed in the chlor-alkali
industry. The very large chloialkali electrolysis industry is by its
nature highly energy-intensive. A small reduction in cell voltage of
e.g. 150 mV on 3 V at a given current density represents a saving in
electric energy of 5 %.
The overvoltage of the cell is determined by the overvoltages at both
electrodes, especially the cathodic overvoltage required for the
hydrogen evolution, and by the Ohmic resistance of the electrolyte (and
diaphragm or membrane). Applying a catalyst to the surface of the cathode
reduces the overvoltage. However, it is necessary that the catalytic
activity remains unaffected for a long period under the prevailing cell
conditions.
Alternative Energy Resources
Four projects closely examine the possible access to new energy resources
*) University of Amsterdam
Selective removal of H2S from gaseous mixtures containing CO2 - T h e o b j e c -
tive of the first research project, entrusted to SNAMPROGETTI, was to
valorise sour gas resources. Hitherto this was difficult because of the
acid compounds - CO2 and H2S - contained in these gases. SNAMPROGETTI
investigated the set-up of a process for the selective removal of H2S.
The solution of the problem lies in the choice of solvents and in the
optimization of the parameters involved. A certain number of pairs of
compounds were selected on the basis of laboratory work and will be
evaluated on a pilot scale.
The following two work programmes concentrate mainly on the recuperation

of combustible products.
Scrap tyres processing Th e p roject entrusted to the INSTITUT FRANCAIS DU

PETROLE covers the transformation of waste tyres into a liquid fuel which
will correspond to the specification of heavy fuel oil. Without any
pretreatment, the tyres are completely depolymerised in oil - preferably
an aromatic oiI - at high temperature. The products obtained, gas,
gasoline and fuel are refined and can be used directly as liquid fuels.
The cost/benefit ratio achieved by an industrial unit is attractive such
as to encourage the development of this process.
Solid bearing effluents - T h e project carried out by the BRITISH CARBONIZA-

TION RESEARCH ASSOCIATION is aimed at the recovery, treatment and utili-
zation of solids - containing effluents emanating from coke ovens.
Two types of effluent are considered : the preheat slurries and the char-
ging main slurries.
The charging main slurries sometimes contain a high concentration of tar.
Handling of these slurries is now possible up to a tar content of 30 X.
Dewatering of the slurries requires two steps : the first is a flocculation
stage and additives are required to speed up dewatering; the second is a
mechanical dewatering stage, which brings the moisture content down to 30 '/..
Mixtures of charging main slurries and preheat slurries are more difficult
to dewater and this results in an increase in the moisture content. As
dewatering of these fines was found to be practicable, the investigation is
now extended to the integrated system.
Anaerobic fermentation - Finally, the critical evaluation of anaerobic

fermentation of waste products, carried out by INTERNATIONAL RESEARCH AND
DEVELOPMENT Co, is an important project. It is hoped that concrete infor-
mation will give a clear picture on the real potential of such a resource
of combustible gas.
It follows from this work that :

the upper limit of the possible contribution of waste products to energy
demand is significantly below 3 % of the Community's energy requirements.
Almost all European countries allocate significant funds to work on
anaerobic fermentation. Of recent years, Little progress has been made
in basic microbiology and biochemistry. The economics of the system are
influenced by the size of the installation, the volume and nature of the
wastes, the transport costs, and especially by the regularity of the flow
of wastes. The main incentive to install such a system is obtaining of
solutions to the environmental problems; energy saving is a secondary
matter.
MODIFICATION OF CRACKING FURNACES OF EXISTING PLANTS TO
INCREASE YIELD S OF VALUABLE PROD UCTS AND TO RED UCE
FUEL CONSUMPTION
Author M. Spoto
Contract number EEB-1-109-I
Duration 35 months 1 September 1981 to
31 July 1984
Head of project Ing. M. Spoto, Research and Technology
Contractor Montedipe S.p.A.
Address Montedipe S.p.A.
Via Rosellini 1 7 - 1 20124 Milano
Summary
For a long period of time, and up to 1973, Virgin
Naphtha availability in Europe was higher than demand
and its price lower than other petroleum products.
Therefore almost all steam cracking plants were based
on Virgin Naphtha as primary feedstock.
Many of the existing cracking monotubular furnaces,
which are the heart of an Ethylene plant, designed
within the above framework, do no longer comply with
the new market constraints : reduced V. Naphtha
availability, higher feedstocks and energy costs.
The aim of this research is to design a new radiant
coil which could improve olefins yields and reduce
feedstock and energy per unit of product and give a
maximum flexibility on feedstocks.
At the present stage of our research, a new radiant
coil design has been carried out to replace the
monotubular existing one.
Coil construction is in course.
After mechanical completion (February 1983)> experi
mental work will start.
240-
1 . Introduction
Rising costs of both raw material and energy require a

continuous technological effort to reduce feedstock and
energy consumption per unit of products. All chemical
processes must be continuously improved and the higher
the material and energy flow are, the more important
will be the results.
It is easy to realize the attractiveness of a steam
cracking plant : a unit of nominal capacity of 500,000
t/year consumes about 450,000 toe/year and 1,500,000
t/year of feedstock.
Since 1973 Montedipe has undertaken an energy saving
programme achieving a 35% cut in energy consumption at
its P. Marghera steam cracking plant (I).
In addition, in most cracking plants in the world, the
technological development of cracking coils design and
the use of new alloys allowed a 40% cut in the feedstock
required per unit of ethylene produced (II).
Nevertheless, many of the existing cracking furnaces,
equipped with the obsolete monotubular vertical coil,
do no longer comply with new production requirements :
high conversion and selectivity as well as feedstock
flexibility.
In the past, looking forward to obtaining the maximum
ethylene production, plants operators increased
cracking temperature as far as the upper limit value.
A further increase in cracking severity causes a selec
tivity loss and a rise in the coking rate because the
monotubular coil is characterized by a long residence
time and involves a high pressure drop. At a higher
conversion degree, valuable products such as propylene and
butadiene continue to decrease, ethylene degradation
starts and aromatics continue to rise (III).
Deposits inside the tube of heavy aromatics and further
dehydrogenation cause coking and involve frequent shut-
downs for cleaning.
Substantial improvement can be achieved only with a
reduction of contact time and pressure drop which both
depend on the coil geometry. The alternative technology
to the coil cracking, which has not been entirely success
fully developed and has not yet found industrial use,
does not fall within the aim of our research work.
Finally, it is worth noting the role of feedstock flexi
bility of a cracking furnace. Feedstock availability and
price become more and more critical, expecially
-241
for Virgin Naphtha which accounts only for 20% per crude
barrel and its use as motor fuel is about 16% (IV).
For a long time our Company, has been engaged in the
cracking area : the existing monotubular experimental furna
ce has allowed to make the best use of the monotubular coil
technology; now, starting from the beginning of the coil de
sign, we want to do more than in the past. The pyrolisis
section is the heart of an ethylene plant (V) and thus the
retrofitting of existing monotubular coil furnaces is
mandatory.
2. Description
To understand how we are going to retrofit the existing
furnace let's describe its present configuration.
2.1. Existing furnace
The existing cracking header is a typical Lummus
SRT-1 single coil design, with the following charac
teristics (VI ) :
- convection zone
14 tubes, finned or bares type, internal diameter :
67 or 91 mm; length : 338O mm
radiant zone
10 tubes, internal diameter : IO8 mm; length : 70OO
8040 and 8710 mm.
The furnace is equipped with 24 wall burners and
has a nominal feedstock capacity of 2,500 kg/h.
Feed enters into the top of the convection zone
where it is mixed with dilution steam at a fixed
ratio. The mix, fully vaporized and heated at
about 600C by combustion gases, enters the radia
ting zone where cracking reactions occur. The
cracking gases are finally quenched with oil in a
mixer before going to the primary fractionator.
At present, it is possible to perform effluent
analysis only once or twice a day because of the
time required by the manual sampling and analysis.
2.2. Furnace modifications

- General considerations
As it is well known, the optimum of olefins yields
is achieved at a fixed conversion degree with the
lowest residence time and the lowest hydrocarbon
-242-
partial pressure (HCPP).
Under these conditions, selectivity of cracking
reactions to the desired products is at the maxi^
mum value and condensation reactions are discou
raged. At a fixed feed rate and severity degree,
both residence time and HCPP depend on the absolute
pressure level at radiant coil outlet, steam dilu-
tion rate, coil geometry and lay-out.
Coil design
According to the above mentioned concepts, a study
has been carried out to replace the monotubular coil
of the existing semiscale furnace with another one
of improved performance.
By using a split coil geometry and a new alloy it is
possible to reduce both the contact time and the
hydrocarbons partial pressure.
New alloys, operating at a higher tubewall tempera
ture, make it possible to improve heat fluxes and
to reduce the coil length (VII-VIII).
A team of Montedipe ' s experts has worked out Sj3
veral coil arrangements (IX) looking for the opti
mum S/V (surface/volume) ratios along the reactor.
At the inlet, where the coking rate and gases
specific volume are still low, splitted coils of
small diameters can be used : the pressure drop and
residence time are low and heat fluxes are high.
At the outlet, where gas specific volume is higher
and the coking rate increases, small tubes must be
collected in a single coil of larger sectional area:
the pressure drop and tubewall temperature are still
in the desired range.
The next step was to simulate the performance of
the proposed coils with SPYRO program by Pyrotec
Company. SPYRO is a reliable theoretical model to
predict furnace performances. Actually it predicts
the profiles of the gas and tubewall temperature
and pressure along the coil. Furthermore, it also
predicts the effluent composition, the coking rate
inside the coil as well as the heat flux and the
residence time.
This model simulates the entire cracking reaction
involving more than 2,000 reactions taking into
account 110 radical and molecular species.
Analysis of simulation has shown that the reactor
with the best performance is a multiple diameter
-243-
coil () with the following tubes layout :
4 small bore tubes in the first and second pass (ID
64 mm)
1 larger diameter tube in the third and fourth pass
(ID 130 mm)
The total length is 31 meters, the radiant volume O.4
cubic meters and total surface is 24 sq. meters.
The selected coil, with respect to the existing mono
tubular coil of the experimental furnace, is characte
rized by lower contact time, lower pressure drop, hi
gher selectivity and better feedstock flexibility.
Feedstocks such as Virgin Naphtha (boiling range
38157 c C, specific gravity O.709), paraffinic Gas Oil
(boiling range 224365C, specific gravity 0.809) and
LPG (normal butane, isobutane mix) have been tested
and the product yields (wt %) are shown in Table XI.
The construction material of the tubes will be the
new alloy Manaurite XA produced by "Fonderies et A ci
res du Manoir". Manaurite XA is an iron, nickel
(33/40$) chromium (20/27$) alloy with special additio
nal elements (A l, Ti, Nb) and is characterized by im
proved resistance to the carburation and creep. In
addition, the new alloy, because of the passivation
of the alluminium, can extend the time between two dj;
coking operations.
Instrumentation
The temperature profile inside the coil will be conti
nuously monitored by six new recorders located through
the radiant coil. A new differential pressure recorder
will make it possible to follow coking rate (coke
deposits reduce the inside diameter increasing pressu
re drop).
Sampling of gases
The furnace will be equipped with an online process
Chromatograph designed by Montedipe. A complete analy
sis cycle is carried out in one hour and gives quanti
tative values, as wt % on total gas flowing out of the
furnace, of the following products : methane, ethylene,
ethane, propylene, propane, C cut, benzene, toluene,
xylols, styrene, C aromatics, residue,> 200C cut.
3. Experimental work
After mechanical completion of the project, experimeri
tal work is planned. The experiments will be designed
to elaborate results by a regression computer model.
244
The aim of the experimental research is to correlate
yields of olefins products and coking rate with cracking
temperature, feedstock rate aid quality, steam dilution ra-
tio .
The expected reduction in dilution steam ratio, at con
stant olefins yield and coking rate, will be evaluated.
Conclusions
The main actions of our research programme dealing
with the "modification of cracking furnaces of existing
plants to increase yields of valuable products ant to
reduce fuel consumption" are : the new coil design, the
coil construction, the experimental work.
At the present stage of our research, the basic enginee
ring of the modified furnace has been carried out. The
existing radiant monotubular coil (VI) will be replaced
by a new variable profile coil.
By using a simulation program (SPYRO, kinetic model), se
veral coils, designed on the basis of Montedipe's expe
rience, have been tested. The selected coil (X) consist-
ing of a multiple diameter tube arrangement, is more
selective in olefinic products than the existing one (XI).
The coils geometry and the new alloy employed will reduce
fouling rate. The new coil will be equipped with new
instruments which make it possible to follow the tempera
ture profile and coking rate inside tubes. A special
process Chromatograph, developed by Montedipe, continuou
sly analyses the gases flowing from the furnace. The
expected performances of the new coil should give remar
kable reduction in the dilution steam ratio at constant
olefins yield and coking rate.
-245-
PORTO MARGHERA CRACKING PLANT
ENERGY CONSUMPTION
1973 1975 1980 1983 1986
Figure I
?
-
T-760-780C T-810-830C T-830-840=C
Mat: ace. inox Mat: HK40 Mat: Manaunte
(18 Cr 8 Ni) (25 Cr 20 Ni) (25 Cr 23 Ni)
5.0-
3.0-
m 2.0-
1975-H980
Figure II
-246-
CHEMICAL USES OF CRUDE OIL
HAPHTHA C BAC KIH6
TYPICAL VARIATIMI OF MtOMKT HSTBInmOH WnU X V E g f f y
CRUDE OIL
(FAM . souttce") 100
GAS OIL NAPHTHA FUEL OIL

30 20 50
\ STMYV.IME
MOTOR FUEL PETROCHEMICAL

14 USE
6
/K^^"^^
Figure IV
/ 1 ^^-~-^PROt>yUENE ETHYLENE PLANT: ENERGY CONSUMPTION B ALANCE
3200 3200
t
CRACKING STEAM
REACTIONS FURNACES SUPERHEATER BOILERS
1 CfeUT& ritMfc +
LOSSES
900 1700
0 1 3 4 5 6 7
StVERtTy
6400 260C
Figure I I I 9.000
Basis: total energy consumpion 9.000 Kcal

Kg. Ethylene
Figure V
EXPERIMENTAL FURNACE EXPERIMENTAL FURNACE
OLD MONOTUBULAR C OIL GAS SAMPLING SIMPLIFIED DRAWING AN PROCESS ARRANGEMENT
f
Lu -STACK
DTW DILUTION STEAM
FEED
H -. OLD
MONOTUBULAR
COIL
- CONVECTION ZONE
T\ T\ T\ T\
QUENCH
MIXER
TO PRIMARY FRAC TIONATOR
OF INDUSTRIAL PLANT
BURNERS
1- -FIRE BOX
-QUENCH OIL
FUEL GAS |
FURNACE WALL
STACK
DILUTION STEAM
TO PRIMARY FRAC TIONATOR

OF INDUSTRIAL PLANT
QUENCH OIL
- FURNAC E WALL
TW TW TW UTW
Figure VII
Figure VI
COILS ARRANGEMENTS
T(C) ETILENE
850- OLD MONOTUB ULAR COIL
Inside surlace - 25.23 m3 PROPILENE
Inside Volume - 0.68 m3 V(m S/V(1/m) wallr gas("C) # ( M C ) Pfkg/cinl
Total length 74 m
Pressure drop 1.15 kg/cm'
800- 1
15.3 0.367 41.69 998 810 0.367 0.37 41.19
750- Cracking gas temperature protile Kcal
luti
22.9 0.586 39.1 925 816 0.586 0.41 40.13

700- 100.000-
80.000-
/ """ "" 1 60 000-
650- 1 22.7 0.468 48.5 957 823 0.468 0.45 40.8
40.000-
1
-" - H
-T
20000-
21 0.434 48.4 943 829 0.464 1.1 40.68

l i l i 1 1 1
C 01 02 03 04 05 06 08 isec.)
07
T(C) NEW SPUT COIL
Inside surface 19.34 m'
Inside volume 0.41 m3
Total length - 32 m
EHD raSLI 20.8 0.388 53.66 952 842 0.434 0.47 40.81
Pressure drop - 0.3 kg/cm3
cSLI 17.6 0.34 51.76 969 826 0.38 0.59 41.56
16.8 0.32 52.6 984 837 0.356 0.67 41.52
nSU 19 34 0.41 47.2 956 827 0.41 0.45 41.26
Figure IX
Figure VIII
EXPERIMENTAL FURNACE
NEW RADIANT SPLIT COIL GAS SAMPLING
-i r
/ vv
B
DDD
TWUTW UTW
Figure X
NEW SPLIT COIL VERSUS OLD MONOTUB ULAR

COIL PREDICTED PERFORMANCES. (FEEDSTOCK FLEXIB ILITY)
NEW SPLIT OLD MONOTUBULAR
COIL COIL
V.N. LPG GAS OIL V.N. LPG GAS OIL
METHANE 16.30 20.94 10.32 17.94 22.36 11.20
ETHYLENE 26.40 28.38 29.53 25.27 27.50 28.16
PROPYLENE 14.70 18.02 16.42 14.03 17.46 15.93
BUTADIENE 4.34 2.94 5.66 3.98 285 5.23
BENZENE 8.00 3.17 5.03 8.66 3.76 6.12
Figure XI
250
DISCUSSION
Question - Dr. Pilavachi, CEC : Do you have any pilot plant or laboratory
experimental results which confirm the order of magnitude of
the theoretical findings for the modified coil ?
Answer We have no experimental results which can confirm the theoretical

predictions of the SPYRO model for the proposed coil. However,
it is not important to prove absolute product yield values
because we are interested to the difference in performances
between the old and the new coil.
Question - Dr. Gazzi, Snamprogetti : How is the flow rate to parallel

lenghts subdivided and controlled ? If no control is utilized,
are there temperature differences at parallel lenght outlets ?
Answer The flowrate is not controlled in the parallel coil, but we

think that there are no problems in flow rate distribution or
in the temperature values. At present, we have 14 furnaces
operating with split coil geometry without any problem.
Question - Dr. Ehringer, CEC : Did you also take into account the
economical question, i.e. Did you make a cost calculation ?
Answer Yes. We think that the payback period per furnace modification
is 1/2 year. The cost of modification is about 400 million
Lires.
Question - Mr. Knobbout, Expert :

1. The cracking furnace layout is based on heat flows and as
the heat is radiant - thus a nearly constant heat flow - what
is the result in diminishing the throughput under the design
conditions ?
2. What is the price of the new material in comparison with
the standard material and for instance Stainless steel ?
3. Is there a relation between the composition of the steel
and the rate of coking ?
Answer 1. With split coil geometry we have a reduction in the exchange

radiating surface f r o m ~ 2 5 m2 (old monotubular coil) to 19,3 m 2
(new split coi I ) .
Heat fluxes and wall temperature values are higher for the split
coil geometry.
The cracking gas temperature is also higher, especially in the
first pass (of the inlet) where the residence time is low.
2. The price is 15 % higher than standard material.
251 -
Question - Prof. Ketelaar, Expert : To what extent do the calculated and
experimental values for the different products correspond for
the old coil arrangement ?
Answer Thechoiceof SPYRO model in order to simulate our alternative

coils designs has been supported by its reliability resulting
from several literature references (see, for instance
A.G. Goossens et al. Oil Gas J. Sept. 1978, page 89, M. Dente
et al. Comp, and Chem. ENG., 1979, (3) page 61).
In addition, with respect to our research Program, a sample of
comparison between experimental and predicted results for the
old coil arrangement is reported below :
Yields Gaso Fuel

C2H2
(wt X)
H
2 CH
4 C
2H6 C
3H4 C
3H6 C
3H8 C
4H6 C
4H8 C
4 H 10 line oil
Coil outlet
temp.(835C) 0,74 17,91 0,39 26,11 4,31 0,30 13,95 0,63 4,19 3,89 0,28 20,31 6,99
Experimental
Coil outlet
temp.(841C) 0,81 17,39 0,35 25,47 4,78 0,63 13,74 0,54 3,94 3,64 0,29 21,37 7,05
Prediction
Severity index (C^/C^) : EXPERIMENTAL = 0,534

PREDICTED = 0,539
RESEARCH PROGRAM WORK
STAGES * H962 ISS 198
^ ^
s
W/y,y,y//m
STAGE -MEW COIL bEMGN
iTACiEl-CO\L CONSTftUCrtON
&WGE -EXPEtUMENTM. PttOCQAW
-252-
SELECTIVE REMOVA L OF H S FROM GA SEOUS MIXTURES CONTAINING CO
2 2
Author : C. Rescalli
Contract number : EEB11301
Duration : 30 months; 1 September 1980 28 February 1983.
Head of project : L. Gazzi; SNA MPROGETTI S.p.A. GAS Department
Contractor : SNAMPROGETTI S.p.A.
Address : SNAMPROGETTI S.p.A.

20097 SAN DONATO MILANESE, Via Alcide De Gasperi, 16
Summary
The selective removal of HpS from gaseous mixtures containing CO2 is

a problem which has been examined for a long time and which is not yet ef
ficiently solved.
The research proposal concerns a highly selective process based on the
difference between the reactivity of 2 with respect to CO2 towards ter
tiary anies in suitable anhydrous solvents, with energetic advantages sin
ce CO2 is left in the main gas stream. Possible applications are very nu
merous in the field of natural gas and in the preparation of synthesis gas.
An experimental laboratory work has been carried out as first point
of the research program about said selective separation of H2S.
Various aminesolvent combinations have been tested and, from vapor
liquid equilibria determinations at various pressures and temperatures, a
first selection among different absorbing media has been made.
1.1 Introduction
The selective removal of the H2S from gaseous mixtures containing
COo is currently carried out by means of:
- oxidative processes (Gianmarco, Streetford, and so on)
- liquid gas washing processes based on the most favourable kinetics of
absorption of the H 2 S with respect to C 0 2 in water-Methyldiethanolamine
(MDEA) solutions.
The industrial application of these processes has met with considera
ble difficulties for both ecological problems and for the considerable
costs increase due to selectivity losses when a complete removal of the
H2S is required.
Our purpose is to develop a simplified process with a more complete
H 2 S removal.
It is well known that primary and secondary amines react with C 0 2
according to different mechanisms:
- by hydration followed by bicarbonate formation,
- by carbamate formation.
Tertiary amines react by bicarbonate formation only. On the contrary
there is no difference between primary, secondary and tertiary amines in
their behaviour with respect to H 2 S which reacts by salt formation only.
Our R. and D. program is based on two hypothesis:
- a solution of a tertiary amine in a reasonably anhydrous organic solvent
cannot react with C0 2 , because of the low H 2 0 concentration,
- the same solution will allow the salt formation of H 2 S with amine since
ionization of H 2 S is possible.
In practice a part of C 0 2 will always be absorbed with the H 2 S becau-
se both of the physical absorption phenomena and the reaction with the
small quantities of water: the water is present in the solvent, can enter
with the gas phase, and therefore its concentration can increase in the
amina-solvent solution according to the operating conditions.
The main steps of our job have been:
- to set up a laboratory equipment for evaluating said hypotesis,
- to screen the most selective solutions among tertiary amines and organic
solvents for a further test on bench-scale,
- to study the influence of small concentrations of water on the solution
selectivity.
1.2 Description of apparatus and measuring equipment

The quantities of H 2 S and C 0 2 present at equilibrium, in the various
systems reported in Table I, have been determined by means of the experi-
mental appartus described in Fig. 1:
- a cell for the gas-solution contact (10)
- two autoclaves for feeding the gas - H 2 S or C 0 2 - (5) (5a).
This experimental apparatus allows the determination of amount of
gas present in solution without chemical analysis. Each gas is measured
separately.
-254-
The equilibrium cell consists of a stainless steel autoclave, of the
capacity of 337,7 cm 3 in a isothermal water bath at 25 C + 0,1. Stirring
is assured by a high couple magnet.
For the exact measure of the quantity of gas absorbed by the solvent,
the manometer scale of the feed autoclave has been calibrated at 0,5
kg/cm intervals, by weighing the effluent gas.
Referring to Fig. 1, the experimental procedure can be summarized as
follows: 150 cm 3 of solvent are charged in the absorption cell ( 10 ); the
cell is immersed in the thermostatic bath ( 9 ) and connected through val
ves (7) and (7a) with the autoclaves containing the gas to be adsorbed.
After connection vacuum is made,through valve (8) in the cell and in
the connecting line in order to vent inert gases (time 2' at 1 Tor ca.).
Heating of bath (9) at the operating temperature is then started and
once this temperature is reached, the cell is ready for absorbing the gas.
Through valves (7) and (7a) gas to be absorbed is contacted with the
solution of cell (10) where is kept stirred and, through the manometer (3)
of the gas feed, the flow of the gas from the autoclave to the solution is
monitored.
Equilibrium is attained when the system pressure is constant for 10' .
When measurements at different temperatures are required, valves (7) and
(7a) are closed, the temperature of bath (9) is set at the desidered value,
and when this value is reached, the same procedure is repeated.
The system pressures are recorded at the beginneg and at the equili
brium points. From these values through the calibration curve, the gas ab
sorbed in the contact cell is determined (in order to know exactly this
quantity, the gas phase in the ceiling of the cell must be substracted).
Following the above procedure, the absorbed gas at the equilibrium
conditions is calculated by utilizing the formula:
P.V
gr gas in solution A ( .Pm)
KT
100
100 gr solution W
Where:
A = weight of the gas lost by the autoclave (5) or (5a) (gr)
= cell pressure (ata)
V = ceiling volume (It)
ata It
= universal gas constant (0,082 )
, s K m
= temperature (K)
Pm= gas molecular weight (gr)
W = weight of the solution charged in the measuring cell (10) (gr).
Absorbing solutions have been tested obtained by mixing the following
compounds:
Amines Solvents
Methyldiethanolamine (MDEA ) Sulfolane (SOL)
Dimethylethanolamine (DMEA ) NMethylimidazole (NMI)
Diathylethanolamine (DEEA ) NMethylpyrrolidone (NMP)
NMethylmorpholine (NMM) Tetramethylurea (TMU)
2Ethoxyethanolamine (ETEA ) NMethylformammide (NMF)
1Dimethylamine propan2ole (DMPA ) Dimethylsulfone (DMS)
l,l,lNitrileTripropan2ole (TPA )
Dimethylaminomethylacetate (DMA M)
Dimethylamino potassium acetate (DMA K)
The values of HS and CO2 at equilibrium conditions have been determi

ned for each solution tested at several temperatures (40, 60 and 80 C)
and pressures (115 ata). In order to compare among them the various sy
stems we calculated a selectivity value in this defined as:
gr H S in 100 gr of solution
S =
gr CO in 100 gr of solution
Where the weight of H 2 S and CO, are derived from the experimental
data at the equilibrium conditions of temperature and pressure.
For every solution the selectivity S can be related to the tempera
ture and the pressure by the relation:
S = a + bT + cP (a)
where = C, = ata and a, b, c are empirical parameters determined by
regressing the experimental data; the average value of the error is about
5%.
The absolute quantity of absorbed H2S also can be rapresented by a
relation analogous to (a) :
Q = a' + b'T + c'P (b)
where T = "C, = ata, and Q = gr H2S/100 gr solution.
The average value of the error of the regression for a', b' and c'
is about 12%.
In Table I for each tested system the following data are reported:
the composition of amine, organic solvent and water,
the values of the regression coefficients a, b, c of the selectivity
formula (a),
the quantity of H 2 S absorbed by 100 gr of solution, at 40 C and 4 ata
(calculated value from relation b ) .
In Figures 2+6 are reported the curves representing the quantity of
H 2 S and C 0 2 absorbed at the equilibrium conditions at various temperatu
res and pressures respectively in:
NMP/DMEA/AQ in ratio 64/34/2 weight
NMI/DMEA/AQ " " 64/34/2
SOL/DMEA/AQ " " 64/34/2
SOL/DMPA/AQ " " 64/32/2 "
SOL/DMEA/AQ " " 60,5/35/4,5 "
256
1.4 Analysis of results and comments.
From the results reported in Table I, we observe :
- the most selective solutions are the NMI-DMEA-AQ and NMP-DMEA-AQ; however
when low concentrations of residual H2S are required, as it often happens
in industrial applications, the SOL-DMEA-AQ system should be the most se-
lective.
Among the amines beside DMEA, the DMPA is the most promising.
- Water has a marked effect in lowering the selectivity to unacceptable va-
lues.
- By considering the solvent power, the DMEA and the DEEA are the most inte
resting amine compounds.
2. Conclusion.
The laboratory experiments carried out up to now showed the possibili-
ty of selectively removing H2S from a gaseous stream containing also CO2,
by using solutions of tertiary amines in nearly anhydrous organic solvents.
Interesting results have been obtained by the use of mixtures of Dimethyl-
ethanolamine in Sulfolane, N-Methylpyrrolidone and N-Methylimidazole con-
taining ca. 2% w of water. These mixtures should now be tested in a pilot
plant in order to collect further informations on their behaviour and on
the real possibility of using them in an industrial plant, informations
from which ultimately depends the final decision about industrial exploita-
tion.
Particularly in piloL plant fuLure tests will be carried out in ordei
to determine and compare, at standard conditions, and for the different ab-
sorbing media proposed :
- the minimum ratio solvent stream/acid gas feed (kg/Nm^) to recover from
the top of the absorber a gaseous stream combining 1 ppm of HpS;
- C 0 2 absorbed with H 2 S and discharged from the botton of the absorber due
to both physical absorption phenomena and reaction with the water present.
Once the preferred absorbing medium will be defined, further tests will
be carried out in order to optimize the cycle.
-257-
Tab. 1: TESTED SOLUTION CHARACTERISTICS
COMPOSITION H S SELECTIVITY (S=a+bT fcP)
MO SOLVENT POWER
liil
Organic solventi A mine Water a 1 b 1 (**)
Al SOL 4,9002 0,02152 0,0868 4,9
A2 NMI 10,5117 0,05008 0,0789 10,5
A3 NMP 8,5315 0,03772 0,0007 13,0
Bl MDEA 4,3492 0,02495 0,1362 19,1
Cl SOL DMEA(34%w.) 2%w. 7,7112 0,02085 0,1951 14,8
C2 NMI DMEA " Il
11,4435 0,05104 0,1601 15,4
C3 NMP DMEA " M
12,2681 0,08991 0,0258 16,6
C4 TMU DMEA " II
9,6989 0,06535 0,0232 13,9
C5 NMF DMEA " 4,1776 0,00985 0,2595 13,7
C6 SOLDMS() DME
A " 2%w. 6,9185 0,01019 0,0906 15,0
Dl SOL MDEA(34%w.) 4,6802 0,00073 0,0268 9,7
D2 SOL DMEA " 7,7112 0,02085 0,1951 14,8
D3 SOL DEEA " 7,3671 0,03485 0,0368 13,3
D4 SOL NMM " 5,4978 0,01557 0,0592 7,7
D5 SOL EtEA " 6,9737 0,03778 0,0206 9,4
D6 SOL DMPA " 8,6341 0,05442 0,0160 11,1
D7 SOL DMAM " 4,7971 0,01478 0,0687 6,8
D8 SOL TPA " 4,6946 0,01298 0,0330 5,0
Eli SOL MDEA(35%w.) 5,4724 0,00464 0,0640 8,5
E12 SOL MDEA(34%w.) 2%w. 4,6802 0,00073 0,0268 9,7
E13 SOL MDEA(33%w.) 5%w. 2,8064 0,02363 0,0551 9,7
E21 SOL DMEA(34%w.) 2%w. 7,7112 0,02085 0,1951 14,8
E22 SOL DMEA(35%w.) 4,5%w. 4,8260 0,01848 0,2668 15,4
E23 SOL DMEA " 18%w. 0,1622 0,05303 0,1865 15,7
FI DMEA " 65%w. 0,7556 0,00267 0,0012 14,6
F2 DMAK(30%w.) 70%w. 0,6666 0,00361 0,0983 6,3
(*) 50% SOL. 14% DMS; (**) gr H S/100 g sol; 40 C, 4 ata

Apparatus for vapor liquid equilibria determinations
-Kxl

1) Inlet valve
2) Outlet valve
3) (sA) (6) manometers
4) (4A) (9) thermostatic baths
5) (5A) Feed autoclaves
7) (7A) valves for collecting feed autoclaves (5) (5A) to absorption cell
8) Valve to vacuum, for feeding solutio and outlet gas from absorption cell
10) Absorption cell with magnetic stirrer
11) Magnetic driver apparatus
FIGURE 1
SELECTIVE ABSORPTION H2S - C02
-259-
SELECTIVE ABSORPTION H2S-C02
sao'c
FIGURE 3
SELECTIVE ABSORPTION H2S-C02 80C
~C02 SOLVENT I DMEA
FIGURE 4
-260-
IS-
SELECTIVE ABSORPTION H 2S-C02 êoc
SOL -64 weight'/.
-H 2,S
SOLVENT DMPA = 34
* -C> H\ 20 ,2
eoc
-40C
gr. abs.gas/lOOgnsol.

20
FIGURE 5
SELECTIVE ABSORPTION H 2S-C02

H }S (SOL -60S weighty.
o -CO, SOLVENT IDMEA -35
,80C [H2 " S ,80C
10
FIGURE 6
-261
DISCUSSION
Question - Prof. Ketelaar, Expert : Have you also made measurements

(with your present equipment) on mixtures of COj and HjS.
As the solutions of the gases are not ideal, the absorption
from a mixture will not follow exactly from the results for
the separate gases.
Answer Up to now our measurements were only made with pure HjS and CO2.
Our future program will include :
- tests in a pilot plant for the most promising amine-solvent
combinations
- tests in an equilibrium cell for the multicomponent system
which will be chosen from the previous step.
Question - Mr. Gibbs, IRD : 1. Can the amine solution be regenerated ?

2. Can you indicate the costs of regeneration ?
Answer 1. Yes, the amine solution can be easily regenerated by a

usual stripping operation.
2. At present it is not possible to indicate the cost of
regeneration because by now we have only carried out experi-
mental laboratory work for equilibrium data determination.
Question - Dr. Pilavachi, CEC : Since you recycle the amine the economics
of implementing this process should be good. Is this correct ?
Answer Yes, in this respect the process we are testing now is much
the same as other amine processes. When a selective HjS
removal is made, it should be more economic than conventional
processes, because investment and energy consumptions are
related to K2S only, and not to both HjS and COj.
Question - Chairman : From the practical point of view, is there in your

process a problem of water content adjustment ?
Answer Yes, as in other washing processes we have to introduce at the

top of the absorber column the water carried away from the
system by the H2S rich stream.
Comment - Mr. Knobbout, Expert : Under the heading of "Advanced heat

pumps" page 312 of the Energy Conservation booklet "Heat pumps"
you will find a project by IFP which also uses CO2 and amines.
They had some difficulties in measuring P-T-X in an absorption
cell. But by now they have solved these problems and have
increased the speed of their measurements.
It is suggest to contact IFP to discuss their experience and
to exchange information.
262-
ENERGY CONSERVATION IN THE CHLOR-ALKALI INDUSTRY
Authors : A.C.C.Tseung, J.A. Antonian
Contract number : EEB/1/111/80 UK (H)
Duration : 24 months 1 Dec 1980-30 Nov 1982
Head of Project : Prof. A.C.C.Tseung
Contractor ; The City University
Address : Chemical Energy Research Unit

Dept. of Chemistry,
The City University,
Northampton Sq.,
London, E.C.1V OHB
United Kingdom
Summary
One of the most promising methods for reducing the electricity

consumption of diaphragm chlor-alkali cells is to replace the inefficient
hydrogen evolving mild steel cathodes with efficient ones. Two types
of low over-voltage, H2 cathode coatings based on Ni-Co-S have been
developed and tested in chlor-alkali environment. The electrodeposited
and the Teflon bonded coatings both gave similiar results: -99mV and
-96mV vs the Dynamic Hydrogen Electrode (D.H.E.) at 300 mA/cm2 , 80C,
10% NaOH, compared to over -350 mV for standard steel cathodes. Long
term tests up to 1300 hrs have shown that such coatings gave stable
performance but corrode at open-circuit and appropriate solutions are
discussed.
1. Introduction
The chloralkali industry is a large energy consumer, accounting for
about V/% of the total industrial electricity consumption of developed
countries.
Traditionally, two types of electrolysis cells are used: The
mercury cell which gives a high quality concentrated caustic and the diaphragm
cell which produces a more dilute caustic which needs concentrating.Concern
over mercury pollution has caused a swing away from mercury cells towards
diaphragm and membrane cells. Membrane cells use a cationic membrane as
the separator, e.g. Dupont's Nafion or Asahi's membrane which produces
high purity, high concentration caustic directly from the cell. However,
the worldwide overcapacity of chloralkali plants has meant that the
number of new plants coming on stream will be low in the next few years.
Therefore, our primary aim has been to develop low overvoltage cathodes
capable of retrofitting existing diaphragm cells.
2. Background Information
The theoretical cell voltage for the electrolysis of brine is 2.1V.
However, ohmic losses and polarization of the electrodes result in an oper
ating voltage of 3.2V for a modern diaphragm cell using a dimensionally
stable anode (DSA : Ru02Ti02on Ti), asbestos diaphragm and mild steel cathode.
Since the anode polarization is already minimal, ca. 30mV, it was decided
to concentrate our work on low overvoltage H2 cathodes. At present, the
mild steel cathodes are relatively inefficient, ca. 350 mV.Attempts by
previous workers to reduce cathode overvoltage by using Raney nickel or
nickel blacks have not been successful since these catalysts invariably
suffered oxidation during shutdown with the formation of non conducting
oxide films which cannot be reduced when the plant is put on stream again.
Tseung et al (14) have shown that Teflon bonded NiCoS,4 electrodes
gave /cm2 ,85C, 45% KOH at 100 mV vs D.H.E.Such electrodes have been
tested for 5800 hrs in another EEC supported programme (26)on water electro
lysis. In the present study, we have investigated two approaches for
coating chloralkali cathodes: A) An in situ method for coating existing
mild steel cathodes for retrofitting existing diaphragm cells by electro
deposition. B) The use of Teflon bonded coatings on nickel screens. This
development is primarily for membrane cells.
3. Experimental
3.1 Electroplated Electrodes
A Wattstype bath similar to that described by Hine et al (7)was/sed.
The bath composition is: 263 g/1 NiS0 4 .6H 2 0; 47.5g/l NiCl2 .6H20;30 g/1 H3 B03
and 30 g/1 KCNS for NiS platingmodified NiCoS plating solution: 263 g/1
NiS0a.6H20;526 g/1 CoS0 4 . 6H2 0;47. 5 g/1 NiCl2 .6H20;95 g/1 CoCl2 .6H20;30 g/1
H3 B03 and 30 g/1 KCNS;. modified CoS plating solution: similar to NiS solut
ion, except that cobalt salts are used instead of nickel salts Before plating,
the mild steel screens were first pickled in a 1:1 HC1/H20 solution and then
they were degreased in concentrated NaOH and washed thoroughly. They were
then plated at current densities of 50150 mA/cm2 for one hour , 4050 C.
After plating the electrodes were washed thoroughly, dried and weighed.
Ni mesh substrates were only degreased but not pickled.
3.2 Teflon bonded electrodes

A) Coprecipated NiCoS 4
The original NiCoS 4 catalyst was prepared by sulphiding freeze
dried NiCo2 0/4 at 350C. This method is costly and inconvenient. Furthermore,
work has shown that high activity NiCo2 S,4 catalyst can be prepared by the
reduction of nickel cobalt nitrate solution with potassium polysulphide(56);
264
this was later replaced by sodium sulphide which yielded a more stable cat-
alyst. 50 g of sodium sulphide was dissolved in 150 cc of distilled water.
The solution was then filtered. 15 g of Ni(N03 ) 2 6H2 0 and 30 g of Co (N03 )
6 H 2 0 was dissolved in 50 cc of distilled water. The sodium sulphide solut-
ionwas added to the nitrate solution with continuous stirring. The catalyst
was dried under nitrogen at 125C overnight.
B) Electrode fabrication
The catalyst was mixed with ICI Fluon dispersion and pasted onto
100 mesh nickel screens. The catalyst/Teflon ratio was 10/3 in all cases.
The catalyst loading was between 40-60 mg/cm2 . The electrode was dried
at 100 C for 1 hr, followed by final curing at 300C for 1 hr in air.
The electrodes were 1 cm2 in all cases. A 1 mm thick nickel wire was spot
welded onto the electrode to facilitate current collection.
3.3 Characterisation of coatings

A) Sulphur analysis
The sulphur content of the electrodeposits was determined by a
Carlo-Erba Elemental Analyser Model 1006, using a sulphur channel with
a Poropak QS column.
B) EDAX
This was found to be the most suitable method for examining the
elect .odeposit and for the determination of the Ni-Co-S ratio on the elect-
rode.
C) X ray diffraction
A Phillips powder diffraction camera and Mo eradiation were used.
In the case of electroplated electrodes, the coatings were, scraped off
and ground with a morar and pestle.
D) SEM
The coatings were examined by scanning electron microscopy to study
the morphology of the electrodes.
3.4 Electrochemical Testing

A) Test Cell
Half cell measurements were carried out in a U shaped Pyrex glass
cell. For tests above 70 C, a Teflon cell was used. The reference elect-
rode was housed in a Teflon Luggin capillary assembly.A 20 cm2 nickel sheet
was used as the counter electrode. Tests were carried out initially in
5N NaOH, 80 C. Subsequent tests were done at 70" C, 10% NaOH and 80C .
17% NaCl, 15% NaOH to duplicate more realistically the operating conditions
in a membrane and diaphragm cell respectively.
B) The reference electrode
The major part of our work was carried out using a platinized
platinum Dynamic Hydrogen Reference Electrode (D.H.E.). The likely presence
of impurities in the electrolyte meant that a few precautions had to be
taken,i.e. the DHE was only introduced in the cell during polarization
measurements;the polarities of the platinized platinum black electrodes in
the DHE assembly were frequently reversed to oxidize the impurilies
DHE was checked against a RHE (-58 mV at room temp) before and after each
measurement. Any significant shift was taken as a sign of poisoning and the
DHE was either recycled (cleaned and replated) or replaced.
C) Measurement of ohmic drop
The ohmic drop was measured by the interruptor method, using a Hg
wetted relay and all the results were corrected for iR.
265-
3.5 Life-tests
Two test solutions were used for life-tests. In the case of
15wt% NaOH, 17wt % NaCl solution, it was necessary to minimize the variation
in electrolyte composition over prolonged periods, since chlorine is also
generated at the counter electrode (graphite anode). Therefore, a 10
litre solution was stored in a plastic reservoir and pumped through the cell
via a peristaltic pump. The composition of solution was periodically
analysed and adjusted regularly.
Since the next generation of chlor-alkali cells are likely to be
membrane cells where the concentration of CI in the catholyte is expected
to be minimal, we carried out further long terms tests mainly in 10% NaOH.
A closed one compartment cell fitted with a condenser with a H 2 /0 2
catalytic recombinationdevice was used for long term testing in NaOH.
This ensured that the rate of water lost was minimal and frequent 'top-ups'
were unnecessary.
4 Results and Discussion
4.1 Electroplated Ni-Co-S
Electroplating existing cathodes with a catalytically active layer
offers great advantages since it uses techniques already in widespread
use in industry. Once the process is developed, it requires minimal care;
it allows for more accurate control of the process variables and can be
coated on 'the spot', thus minimising handling and transportation charges.
In addition, the ever-present risk of damaging the catalyst layer during
transportation will be reduced. Initially, electrodes were made out of
1 cm2 , 100 mesh Ni screens plated with Ni-S, Co-S, and Ni-Co-S respectively.
The performance of such electrodes showed a great improvement over the
results obtained on uncoated 100 mesh Ni screens, with Ni-Co-S coating
giving the best performance. (Fig. 1, Table 1 ) . Similar results were
obtained when mild steel mesh was used as the substrate.
Table I Performance of electrodeposited coatings on Ni and steel

(5N NaOH, 80 C, iR corree ted)
Material Loading Performance S %
mg/cm2 lA/cm2 0 .25A/cm2
mV vs DHE
Ni mesh N/A -254 -
Ni-S on Ni mesh 55.1 -143 -80 4.31
Co-S on Ni mesh 52.3 -143 -80 4.51
Ni-Co-S on Ni mesh 112.3 -126 -69 4.53
ditto 199.4 -126 -62 4.57
ditto 49.0 -124 -69 4.47
ditto 54.8 -126 -69 4.30
ditto 96.0 -126 -69 4.53
Mild steel Grid N/A -375 -247

Ni-Co-S on steel 117.0 -127 -70
ditto 148.5 -126 -63
ditto 143.9 -121 -60
Effect of catalyst loading

Table 2 shows the performance of Ni-Co-S to be reproducible
and that there was very little difference in performance over a range
of loadings from 50-200 mg/cm2 . Loadings in the region of 20 mg/cm2 gave
poorer results possibly due to incomplete coverage. (Fig. 2) Examination
by SEM revealed dendritic growth at higher plating currents and that
coatings plated at 50 mA/cm2 appeared to be smoother.
-266-
Table II Effect of catalyst loading on performance
Material Plating condition Loading Performance

mA/cm2 mg/cm2 /cm 2 A /cm 2
0.25
(mV v s DHE, 80C,5N NaOH)
NiS on Ni 100 55.1 143 80
ditto 50 15.4 225 155
NiCoS on Ni 100 49.0 124 69

ditto 50 19.7 191 90
ditto 50 54.8 126 69
ditto 199.4 126 62
Composition of NiCoS
The composition of the electrodeposit as determined by EDAX is given
in Table III and indicates that the Co to Ni ratio is not constant. A poss
ible explanation could be the presence of cracks in the surface coating.
This would give a higher Ni signal since the nickel substrate would also
be accessible to the electron beam. Since the measured sulphur content
is only about 34%, far lower than the 40% in the thiospinel NiCoS/4,
it is evident that the coating is composed of a mixture of NiCoS.The
evidence available to us at present is not sufficient to allow us to
postulate a formula for the coating.
Stability of NiCoS coating

i)Co dissolution
Electrodes left on load for a number of days did not exhibit a loss
in performance. However, when left at open circuit, there was a bluish
coloration of the electrolyte. After five hours at open circuit,the
electrolyte was found to contain 22 ppm of Co, i.e. a substantial
proportion of the cobalt content of the electrode. No trace of nickel
was found. A s soon as current was passed through the cell, the
cobalt redeposited on the cathode. Table IV gives the surface composition
of electrodes which have been left on open circuit, followed by cathod
izing for yz and 1 hr respectively. The results showed that Co redeposits
on the surface of the cathode, thus resulting in a Co enriched surface.The
electrode that was cathodized for the longest time has the lowest surface
composition of Ni. Scanning electron microscopicinvestigation shows the
existence of cracks, most probably due to mechanical attrition.lt is
important to note that the performance of the electrodes remained unchanged
(Table V ) .
Table III Surface compositionof NiCoS electrodeposit as determined

by EDAX
Sample Atomic Ratio as determined by EDAX

Co Ni S
A 58.34 41.66 3.33
61.1.6 35.36 3.33
C 67.73 34.39 2.88
D 56.62 40.16 3.22
Table IV EDA X examination of electrodeposit after Co dissolution

and redeposition
267
Test conditions Co Ni Fe S
Control 61.16 35.36 0.15 3.33
Cathodized for 5hr
after Co dissolution
at OCV 77.32 21.10 0.67 0.91
Cathodized for 1 hr
after Co dissolution
at OCV 88.50 9.86 1.67 0.98
Table V Performance of NiCoS after Co dissolution at OCV 5NNaOH, 80C)
Electrode Loading Performance

iR corrected
2
mg/cm /cm2 A /cm2
0.25
mV vs DHE
I.a. Original NiCoS 112.3 126 69
b. After lossof Co
and new electrolyte 123 68
c.Using old electrolyte 125 60
II a. Original NiCoS 199.4 126 69

b. After loss of Co 120 60
III a. Original NiCoS 49.0 126 69

b. Bubble air at 80
for 5 hrs, OCV, 125 69
Table VI Stability and performance of electrodeposits
(5N NaOH, 80 C, iR corrected)
Electrodeposit loading Performance Sulphur content

mg/cm2 Initial Final Initial Final
mV vs DHE
NiS (5 days) 54 143 127 4.63% 3.04/3.71%
NiCo S(2 days) 115.6 126 114 4.47% 2.86%
ii) Sulphur dissolution

Table IV indicates considerable loss of sulphur. To find
out whether this occurs at OCV or during electrolysis , a NiS and a
NiCoS electrode were left on load at /cm2 for up to 5 days. Microelem
ental analysis showed that there was significant sulphur loss in both
cases.
iii) Presence of iron
The source of iron detected by EDAX (Table IV) was thought to be from the
NaOH electrolyte. However, as no reduction in performance was observed,
iron may not be a serious poison in the present case,
iv) Effect of heat treatment on stability of the catalysed coating
An attempt was made to stabilize the NiCoS coatings by
heat treating the electrodeposits at 300400 C under N2 .Two electrodes
were plated simultaneously; one was then heated and the other one used as
the control. The results in Table VII and Fig.3 showed that heat
treated coatings gave worse performance and they still corroded on
open circuit,
v) Performance of NiCoS electrode in the presence of Cl~.
268
Two electrodes were plated simultaneously; one was tested in 10% NaOH and
the other one in 15% NaCl, 17% NaOH at 75 C. The results were similiar
and they both gave similiar performance. (Table VIII and Fig. 4 ) .
Table VII. Effect of heat treatment under nitrogen on the performance

of NiCoS electrodeposits (5N NaOH,80 C)
Heat treatment Loading mg/cm2 Performance

/cm2 A 0.25 /cm2
mV vs DHE
96 126 69
300C 89.1 222 140
400C 75.1 267 200
Table VIII Performance of NiCoS in the presence and absence of NaCl
Testing conditions Loading Performance

,80 C mV vs DHE
mg/cm2 A0.3 /cm2 0.2 /cm2
A 0.1 /cm2
A
10% NaOH 51 99 87 68
15% NaCl, 17% NaOH 51 96 78 54
4.2. Tefon bonded NiCoS,4

The use of Teflon bonded electrodes for gas evolution reaction
has been studied in great detail by Tseung and Vassie(8) and also by Tseung
and his colleagues in the EEC hydrogen programme (No. 1617611 EH U and
EH/ 109). It was found that only the top surface of a porous electrode
could function during gas evolution reactions, since the gas evolved
inside the pores would push the electrolyte from the pores, rendering the
interior of the porous electrode ineffective. On the other hand, Teflon
bonded electrodes have dry, porous Teflon channels for gas to escape and
therefore more of the interior of the porous Teflon bonded electrode can
function electrochemically. Apart from providing higher performance,
Teflon bonded electrodes should be less susceptible to poisoning by imp
urities, since a significant portion of the electrode can still function if
the surface of the electrode is poisoned. Teflon bonded electrodes developed
at The City University use 100 mesh nickel screen as the current collector.
The closely woven nickel screens enable the Teflon/catalyst coating to bind on
to nickel wires, thus ensuring good adherence of the coating to the
substrate. On the other hand, if Teflon bonded electrodes are supported
on coarse steel mesh, the performance is likely to be worse, since the
adhesion to the substrate may not be as good and the current collection
efficiency will not be as high as fine nickel mesh substrates.Figs. 4 and
5 show that the performance of Teflon bonded NiCoS 4 electrodes compare
very favourably with electroplated NiCoS electrodes.
4.3 Scaling up of the electrodes

The next step was to attempt the preparation of larger electrodes.
3cm 3cm electrodes on a range of supports suitable for use in the chlor
alkali industry were prepared both by the electroplated and Teflon bonded
coating methods. The supports used were: Ni mesh; standard woven mild
steel; expanded mild steel (SUS 304L); Ni plated SUS 304L and expanded Ni
mesh. The results shown in Table IX indicate that satisfactory electrodes
could be prepared by electroplating on Ni mesh and mild steel substrates.
However on shiny surfaces such as Ni plated SuS 304 L, the electroplated
coatings dropped off regardless of whatever precautions were taken. In the
caseof Teflon bonded electrodes, satisfactory adhesion could be achieved
269
on nickel mesh and expanded nickel, substrate On the other hand, it
was necessary to roughen the surfaces of the other substrates before
adequate keying of the coating to the substrate could beachieved. One
satisfactory method was to dip the substrates in nickel cobalt nitrate solution
and decomposing the nitrates at 400" C in air to form N1C0G4, semiconduct-
ing oxidesln all cases, it was observed that the electrodes had to be
first cathodized at low current densities of 10-20 mA/cm2, for a few
hrs before they were used at higher current densities. This is most
probably due to the need of reducing the sulphide to activate the electrodes
If this precaution was not taken, the coatings peeled off from the substrate.
4.4 Durability
The results for electroplated and Teflon bonded coatings deposited
on various substrates are shown in Table IX.
Table IX Performance of electroplated and Teflon bonded coatings on

various metal substrates used in industry
75 C, 10% NaOH, iR corrected

Electroplated Ni-Co-S Teflon bonded NiCoS.4
mV vs D. H. E.
Support Loading 0.3A, loading 0.3A/cm2

mg/cm2 mg/cm2
Woven mild
steel 57.6 -140
57.9 -108
Ni fine mesh 109.5** -129 22.96 -126
57.5 -99 23.09 -96
Ni plated 57.5* -117 18.44 -132
59.7* -118
Expanded nickel
mesh 53.7 -120 21.03 -164
Expanded SUS
304L 59.1* -185 20.18 -168
56.2* -145
Flattened SUS
304 L - - 22.53 -140
* Coatings dropped off.

** Electrode polarized rapidly; N.B. At 75C, the value of the D.H.E. is -33
mV vs R.H.E. (reversible H 2 electrode).
Fig. 6 shows the life test results obtained for two Teflon
bonded NiCoS.4 cathodes (supported on 100 mesh nickel screens. A graphite
rod was used as the counter electrode. The cathodes were tested in
15% NaOH, 17% NaCl, 0.25A/cm2 , 95 C. The results have not been corrected
for IR. There was a significant improvement over the first two days
of operation and thereafter the performance was stable over the remain ing
test period of 420 hrs.
-270-
Since the next generation of chlor-alkali cells will be membrane cells
in which the chloride ion concentration in the cathode compartment is
likely to be minimal, we have also tested an electroplated Ni-Co-S
coating on 100 mesh nickel screen and tested it at 70 C, 10% NaOH. To
date, there was only 10 mV change in over-voltage over a test period
of 1300 hrs.(300 mA/cm2 )
4.5 PoJBnning of-.Hz evolution cathodes in chlor-alkali cells.

One of the most signicant poison in chlor-alkali cells is Fe ions
coming from the cell hardware. Since Pt is known to be poisoned readily
by Fe, we have studied the stability of electroplated and Teflon bonded
Pt black electrodes tested in {JaOH solution saturated with iron sulphate
The Pt loading was 18-22 mg/cm2 . The results shown in Fig. 7 indicated
that both the plain nickel mesh and the electroplated Pt on nickel
mesh electrodes were significantly poisoned by Fe. However, there was
almost no effect on the Teflon bonded Pt black electrode and the Teflon
bonded NiCoS.4 electrode. Surprisingly, the electroplated Ni-Co-S
electrode was also not affected by Fe. Whilst it is easy to explain the
immunity to poisoning of the Teflon bonded electrodes on
account of the fact that any surface coverage of the top surface by
impurities would still leave a significant part of the available catalyst
in the interior of the electrode functioning, it is difficult to see
why the electroplated Ni-Co-S porous coating could function effectively
once the top surface is covered.One possibility is the formation of
FeSy on the surface . FeSy has been shown to be active for Revolution (9).
It is possible that the Fe ions react with the S ions leached out from the
Ni-Co-S coating. However, in a real operating environment, there will
not be S ions available in the continuous flowing brine solution.
5. Conclusions and suggestions for further work

The results obtained in this study have shown that considerable
reduction in cathode over-voltage can be achieved by the use of Ni-Co-S
coatings. However, Co is likely to be corroded on open circuit, since it
will become anodic relative to nickel or iron. One possibility is to ensure
ensure that the cathodes are always cathodically polarized by a small
cathodic current on open circuit. However, this may not be practical in
an industrial environement. An alternative approach is to develop
coatings which are corrosion resistant under open circuit conditions.
A proprietary cathode catalyst h been developed in the EEC H2programme
conducted at The City University which is immune tocorrosion at open
circuit in 85 C, 45% K0H and this catalyst will be tested under
chlor-alkali conditons in the near future. In addition, we will be
working closely with various European chlor-alkali manufacturers to
evaluate such coatings in terms of performance, life, susceptibility to
poisoning and energy saving.
6. References
1. A.C.C.Tseung, S. Jasem Electrochimica Acta, 22,31, 1977.
2. A.C.C.Tseung, M.C.M. Man, S. Jasem and K.Yeung 'Seminar on hydrogen as
an energy vector' EEC Report EUR 6085, Brussels, Oct., 1978, p.255.
3. A.C.C.Tseung, M.C.M. Man, British Patent, 1174765,1979.
4. A.C.C.Tseung, M.C.M.Man, Electrochemical Society, Extended Abstracts,
Vol.79-1, 687, 1979.
5. A . C . C . T s e u n g , P.Rasiyah, M . C . M . M a n , K . L . K . Y e u n g , 'Optimization o f
g a s evolving T e f l o n bonded e l e c t r o d e s ' Final R e p o r t (1st N o v . , 1 9 7 7 - 3 1 s t D e c .
1 9 7 9 ) E E C C o n t r a c t N o . 311-77-11 E H U k .
-271 -
6. A.C.C.Tseung, P. Rasiyah, J.. A ntonian, Progress Report to EEC, Contract
EH/B 109, July 1980.
7. F. Hine, M. Yasuda and Watanabe, Denki Kagaku, 41(7), 401408, 1979.
8. A.C.C.Tseung, P.R.Vassie, Electrochimica Acta, 21, 315318, 1976.
9. A.I. Onuchukwu, Electrochimica A cta, 27, 529533, 1982.
7. A cknowledgement
The authors would like to acknowledge the help given by Drs. D.B.
Hibbert and N.Fouzder as well as colleagues in the chloralkali industry.
mV '
vs
DHE 5N NaOH, 80 C
300
5N NaOH, 80C IA/cm2

NiCoS
NiS
200' 0.25A/cm2
O NiCoS
NiS NiS
NiCoS
Ni mesh
100
Mild steel
iR corrected
iR corrected
"
1 1 1 * r
50 100 150 200 mg/cm
0.1 0.2 0.3 0.4 V vsDHE
Fig.l Performance of NiCoS and NiS Fig. 2 Effect of catalyst loading
5N NaOH, 80C, iR corrected
N2 atmosphere
10% NaOH, 80C
15% NaCl, 17%NaOH,

80C
T
50 100 mV vs DHE
100 200 300 400 C
Fig.4 Performance of NiCoS in NaCl
Fig. 3 Effect of heat treatment
NaOH solution
272
1.0
70C
A/cm' 1 7 % NaCl, 1 5 % NaOH ^nV vs DHE
160
55mg/cm
1) 55 rag/cm (2) 48.9mg/cm
J.75 iR corrected
Electrode 1 7 % NaCl, 1 5 % NaOH, 95C
left over
night in air 50 C 120

1.50. f(2)
90 (1)
No iR correction
0.25 25C Graphite counter
40
electrode
tfZa-* r^cS^ 1 >

0 100 200 mV vsDHE 0 100 200 300 400 hr
rig. 5 Performance if Teflon bonded "ig. 6 Long term tests of Teflon
NiCo2S4 bonded NiCo^S electrodes
V vs S.C.E.
1.6

A Ni screen
1.5 .
1.4 Electroplated Pt
Teflon bonded NiCos
Electroplated NiCoS 4
1.3
Teflon bonded Pt
1.2 i ^
10 15 20 hrs
Fig. 7 Electrode performance as ,. function of time in FeSO

saturated, 1 0 % NaOH, room temp., 0.3A /cm* ,
273
DISCUSSION
Question - Mr. Limido, Expert : What are the common poisons encountered
in the chLor-aLkaLi industry ?
Answer The usual poisons are Fe, Mg and Ca.
Comment - Chairman : It is already usual practice in the chlor-alkali

electrolysis to keep on the voltage (at very small current
density) in order to prevent corrosion.
Answer I agree with your comment and this would ensure that our
Ni-Co-S could be used without fear of corrosion.
Question - Dr. Pilavachi, CEC : Using these coatings could you reduce the
extent of required brine purification ?
Answer No, since it is important to get rid of the impurities as much

as possible for the sake of getting high quality caustic.
Question - Mr. Limido; Expert : Are there similarities between the electrodes
used in your application and the ones appropriate for fuel-cells,
at least at the structure and construction level ?
Answer Yes. The concepts and technology of fuel cells are used
extensively in this study.
THERMAL DEPOLYMERISATION OF WASTE TIRES
BY HEAVY OILS. CONVERSION INTO FUELS
Authors : F. AUDIBERT, J.P. BEA UFILS
Contract number : EEB4182F
Duration : 22 months 1 Oct. 1980 31 July 1932
Head of project : F. AUDIBERT
Contractor : Institut Franais du Ptrole
Address : .P. n 3 69390 VERNAISON France
Summary
The French Institute of Petroleum has developed a process of valorisa

tion of waste tires into fuels.
The process consists in treating the whole tires with heavy hydrocar
bons which transfer the heat for reaching the required temperature
and dissolve the oligomers resulting from devulcanisation and depoly
merisation reactions.
The tire is converted into moderate amount of gas and gasoline and
mainly into heavy oil loaded with finely dispersed carbon black.
Results have been obtained on a large size pilot plant making the
process directly appliable to the industrial scale.
275
1. Introduction
For a long time researchers have been involved in trying to valorize
waste rubber. Their attention has particularly focused on reclaimed mate-
rial mixed with fresh rubber. For ten years efforts have been made in this
field for both environmental an energy saving purposes.
Thus, since 1974 a chemical engineering team at the University of
Compigne (Professor M. Gelus and his assistant J.M. Bouvier) has been
investigating the thermal degradation of rubber on a laboratory scale.
Because of the promising results obtained, the Institut Franais du Ptrole
was assigned the development stage.
2. Development phase
The process aimed at consists in treating whole tires of all types
with heavy hydrocarbons which transfer the heat required to reach around
380 C and dissolve the oligomers resulting from devulcanization and depo-
lymerization reactions.
To gain better understanding of the field of rubber and tires, I.F.P.
became associated with the Michelin Manufactures. This company has provi-
ded financial aid, and support has also been obtained from european and
french public institutions (European Economic Community and Agence
Nationale pour la Rcupration des Dchets).
The development work was done during two distinct periods. The first
one was completed in 1979, using a medium-size pilot plant representative
of the contact phenomenon. The second one started early in 1982 in a large
pilot plant suitable for problems of scaling up.
Both plants were operated in batch vessels designed for 8 hours of
complete operations.
The medium size pilot plant included a batch vessel with a capacity
of a few cars tires and has demonstrated the feasibility of treating whole
tires.
2.1 Large pilot plant description -- Process description -

The large pilot plant built at the end of 1981 is described in the
inclosed figure and comprises the following main parts :
- the whole tires (100 to 300 kg) are put in a basket (R.) ;
- the contacting oil representing a total volume of around 600 1 per batch
is recirculated in the main loop with the pump Pj at a flow rate of 30 to
60 m /h depending on the viscosity of the bulk liquid ;
- the circulating liquid is heated in the electric heater E. up to 380 C
within 3 h ;
- the liquid is sprinkled in the vessel making contact with tires by tri-
ckling and involving much less liquid than with complete immersion ;
- as the different reactions proceed involving thermal cracking of rubber
and of the oil the light compounds produced condense in the air exchanger
E while the gas goes through the condensing part and is volume metered
before exiting ;
- when the depolymerisation reactions are over, the bulk liquid phase inclu-
ding the compounds resulting from the tire degradation is cooled in the
air exchanger E_ down to around 100 C ;
- at this stage of advancement of the operation the liquid in the main
loop, which is quite comparable to viscous fuel oils is diluted with the
light hydrocarbons coming from the bottom of E which favorably decrease
the viscosity and the pour point of the fuel. The role of E is to remove
light gasoline in order to set the flash point of the fuel oil at the
specification ( / 70 C) and to keep its viscosity in the heavy fuel oil
range (110 cSt-450 cSt at 50 C). If necessary E can supply an additional
-276-
amount of gasoline to fit the energy requirements of the plant which could
be normally provided by a mixed-burner (gas an gasoline).
2.2 Material balance

For each operation the rubber stock is placed in contact with an
amount of fresh oil representing about three times the quantity of rubber
by weight. When the temperature increases and reaches the range required,
the amount of gas and light compounds formed is also dependent on both tem-
perature and residence time. This is the reason why heating is stopped as
soon as the rubber has depolymerized when the temperature reaches 370-
380 C. Obviously any fresh oil distilling before the required reactor tem-
perature will supply additional condensates.
We describe hereunder a material balance with somme data included
within a given range because of the differences existing in the type of
tires and oils than cans be treated. In our experiments we used the oils
described later on in this paper. For instance, when the heavy fuel oil of
the market was heated to 380 C, no more than 10 % by weight was distilled.
-? gas 4 to 6
tires 100 -> gazoline 8 to 10
materiel balance "* fuel oil (including
oil 300 by weight carbon black)
* waste 15 to 25
As a first approach the oil can be considered to act as a solvent and

heat transferring agent.
In an industrial plant the effluents from the reactor would >e separa-
ted in two flash drums in serie. The amount of gasoline removed from the
first one will be ajusted so as to supply (with the gas) the energy requi-
rements of the plant. The second separator acts as a debutanizor.
gas 4
I
to 6
^
3S
tire J
* middle
I
.lates
gasoline
*"8 to 10
recycleddistil
back to
the reactor
50 to 150*
oil
fuel oil
waste 15-25
X amount sensitive to temperature level and residence time and of course to

the distillation curve of the contacting oil.
-277-
2.3 Contacting oil analyses
As said above in the text, various oils can be used as long as they
are available at low price and have a moderate amount of volatile compounds
below 380 C.
Beside waste oils which also give good results, two kinds of oil have
been systematically explored and are described below :
Aromatic extracts Heavy fuel

from lube manufacturing oil of the
150 ssu 400 ssu market
Specific gravity at 20 C 0.997 0.998 0.984

Viscosity at 50 C (cSt) 104 300 362
Viscosity at 100 C (cSt) 8.47 22.85 33
Pour point C + 12 + 27 0
Flash point C - - > 70
Conradson carbon wt % 0.22 2.2 13.7
Asphaltenes (wt %) - - 5.3
Sulfur (wt %) 4.7 4.9 3.7
5 % distill. C O 402 477 240
Table I
-278-
2.4 Products analyses
Gas
Originated from
150 ssu extract 400 ssu extract heavy fuel oil

+ + +
earth work tire car tires earth work and
truck tires
CO 1.2 3 1.4
CH 8.7 19.1 6.9 !
4
C 1.6 3.1 2.8
2
C 1.3 1.6 1.1
2H4
C 6.2 8.5 4.1
2H6
H20 1.3 2.8 1.2
H2S 8 8.6 4.5
C 4.3 4.2 3.2

3H6
C 8.8 8.5 6.4
3H8
C 27.2 19.5 22.5
4
nC 7.6 6.9 5.0
4
C 7.5 3.7 11.8
5
nC 5 11.5 5.8 24.0
C
2.5 2 2.7
C 1.7 1.3 1.8

6
undetermined 0.5 2.4 0.6
remark straight run straight run test following
a blank one
same conditions
Table II
Compared to blank tests operated in the same conditions we have twice the amount
of H S due to sulfur from tires, less propylene and propane and a double
amount of butnes.
Gasoline :
Typical analyses are given for gasoline having a 200 C and point.
Data are included within a range because of some composition variation due
to altogether rubber depolymerisation and cracking of the oil.
ASTM
Initial point 73 C
5 % 105
10 % 111
30 % 125
50 % 137
70 % 149
90 % 165
95 % 173
E.P. 194
Specific gravity at 20 C 0.770 to 0.800

Nitrogen ppm 500 to 1 000
Sulfur wt % 0.6 to 0,.8
Bromine number gBr/100 g 50 to 80
Maleic anhydride value mg/g 10 to 20
The bromine number and maleic anhydride value provide information on

the olefin and diolefin contents which are about :
olefins 33 to 53 %
diolefins 2,5 to 5,5 % (include also some aroma
tic rings with olefinic
branched chain)
Fuel oil :
The socalled product corresponds to when the contacting oil has dis
solved the rubber and contains no more gasoline and gas.
The thickening effect of thedepolymerised gum dissolved is largely com
pensated for by the diluent effect of the middle distillate recycled back
to the fuel oil after each operation.
It should be borne in mind that these middle distillates come essen
tially from rubber depolymerisation into oligomers and also from some
cracking of the oil.
It is very interesting to compare the analyses of the fuel oil obtai
ned to those of the corresponding contacting oil. (Table I)
280
150 ssu 400 ssu heavy fuel oil
+ + from the market
rubber rubber +
rubber
Specific gravity at 20 C 1.025 1.019 1.007

Vise, at 50 C (cSt) 90 266 223
Vise, at 100 C (cSt) - - -
Pour point C - 9 + 3 - 3
Flash point C 152 114 146
Conradson carbon % 7.14 9.3 18
Asphaltenes wt % - - -
Sulfur wt % 3.95 3.88 3.14
Zinc wt % 0.38 0.32 0.48
Table III
3. Problems solved in development

Reactor :
The reactor is made of carbon steel and operated at atmospheric pres-
sure. Because flammable and foul-smelling products are raised to high
temperature, it must be tight.
The door of the pilot reactor is a reinforced thick steel plate, and
some elasticity of the seal was required. For this reason several kinds of
seals were tested, and viton was selected. Some cooling was required to
prevent destruction of the seal. On an industrial unit the tightness would
be ensured by bright-parts making possible more efficient cooling and the
use of a wider choice of seals.
Mixing of the reactants :
In the pilot reactor the liquid was sprinkled on the tires placed in
a fixed position in the basket. Another way of contacting could consist
either in sprinkling the liquid through a rotating sprinkler or making the
basket rotate slowly in order to improve the mixing of the reactants.
Main pump characteristics :
When depolymerisation occurs gas evolves in the bulk liquid circulated
with the main pump. The presence of the gas might drop or stop the sucking
action of the pump. This problem has been fully solved by focusing atten-
tion on the pump characteristics and its proper location in the loop.
Heater :
No special attention except the observation of a sufficient liquid
velocity. After several months of batch operations we found clean pipes.
-281
4. Advantages of the process
- The whole tire treatment is shown to be feasible so that shredding
is no longer necessary (cars, trucks, earthwork vehicles) ;
- The trickling contact between oil and tires involves moderate hold-
up of the liquid and shortens the time required for heating and cooling
the bulk liquid ;
- Practically any kind of hydrocarbon can be used as long as the oil
doesn't distill too much before 360 C or so ;
- The rubber is converted into fuel oil and forms a homogeneous liquid
with the contacting oil and then is processed in conventional equipment.
No dcantation of the carbon black was ever observed in all the liquid fuel
oils obtained ;
- The resultant fuel was burned with success in a 6 300 MJ/h boiler.
The carbon black did not affect the quality of the smoke ;
- On the other hand various rubber wastes can be treated such as com-
posite metallic and rubber parts from the transport vehicle industry. This
latter case is of great interest for recovering the metallic portion ;
- The operating conditions enable viscous fuel oils to be valorized
by decreasing their viscosity and pour point as the result of oligomer
formation and the additional cracking of the hydrocarbons.
5. Main features of an industrial plant

The large size pilot plant drecribed above gives a most representative
view of what an industrial unit will be.
For economic reasons a batchwise process requires paying special at-
tention to the stream time with regard to the equipment cost. For this
reason it may be more advantageous to associate a couple of reactors timed
so that only one heater could be used twice a shift. Only two flash drums
make the light effluents separation.
On the other hand the gas and gasoline produced can be used on the
site to feed the heater.
Concerning storage it would be most advantageous to have several tanks

for the various possible contacting oil. The economic evaluation hereunder
assumes existing storage facilities for the feed assuming that in general
the site should be chosen either for liquid fuel oil facilities or scrap
tire disposal at low or zero price.
6. Economic evaluation
The investment cost of an industrial plant for 4 000 t/y waste-tire
capacity scaled up from up-to-date results approximates FF 3 400 000 B.L.
(start 1982). To this cost we must add the off-sites cost which is largely
dependent on the location.
The proximity of the raw material supply or the contacting oil pro-
duction site or eventual existing storage facilities decrease this cost.
Assuming such a situation and evaluating the storage and handling
of tires at FF 700 000 the total cost including engineering, process book
and royalties would reach FF 5 140 000. The data for the annual operating
statement shown in the figure can be used to calculate a payout time of
3.8. years and a rate of return on investment capital after taxes of
16,5 %. We assumed an 8 years life for the equipment with a linear depre-
ciation cost.
-282-
VALORISATION DES PNEUMATIQUES USES
CONVERSION EN COMB USTIB LES
-t*i mun if III

1/
"--
UNITE DE DEMONSTRATION IFP

Tires Total investment Fuel oil produced
4 000 t at FF 100/t FF 5 140 000 15 200 t at FF 1 100/t
FF 400 000 FF 16 720 000
Fixed costs
Oil Metallic waste
FF 620 000
12 000 t at FF 1 100/t 600 t at FF 200/t
FF 13 200 000 Fuel oil FF 120 000
FF 550 000
Textile waste
Total.. FF 13 600 000 Working capital 200 t at FF O/t
FF 370 000
Total FF 16 840 000
Return on investment after taxes : 16.5

Payout time : 3.8 years (2.5 years before taxes)
* including 3 operators for direct labor.
DISCUSSION
Question - Mr. Gibbs, IRD : Can you use waste engine oil as the heat
transfer medium and up grade it to fuel oiL in the process ?
Answer Yes, such a use is economically very attractive. Furthermore

the waste oil is separated from its light and undesirable
compounds.
Question Mr. Limido, Expert : Are there some difficulties with the fuel
oil specifications due to the ash or contaminants content ?
Answer
Comment - Dr. Ehringer, CEC : According to your information about

100 - 200 000 t of waste tyres are available per year in
France. If 50 % of these tyres are retreated in economically
viable installations of 4000 to/year about 20 such installations
would be required. This supposes also that the transport
problem of waste tyres is solved ?
Answer For problems of transportation costs it would be recommended

to built such plants in industrial areas. Unless financial
aid would be given for collection.
-284-
Question - Dr. Gazzi, Snamprogetti :
1. Is it possible to use the produced oiL as a tyre solvent ?
2. Has the produced oil some health objections, for instance
is it cancerigenic ?
Answer 1. Normally not, because the operation is conducted so that

the produced oil is saturated with depolymerised rubber.
2. The produced oil has the same health objections as the
contacting oil. The process does not add undesirable effects
on health.
Question - Prof. Ketelaar, Expert :

1. What happens with the zinc in the oil when it is burned.
Part will be deposited on walls but part will also escape
2. Are there still (traces) of substances e.g. accelerators
etc. present in the oil from tyres ?
Answer 1. Host of the zinc escapes in the smoke but can be separated
with cyclones. If the produced fuel oil is mixed with
comparable amount of commercial fuel oil, zinc pollution is
normally avoided (due to zinc dilution).
2. We have not made any determination of such possible traces.

We focused our attention to the fuel oil specifications for
burning.
Question - Mr. Limido, Expert : What is the potential production of

fuel from used tyres (e.g. for France or EEC) ?
Answer The potential production of fuel oil from waste tyres is in

the range 100 000 to 200 000 t/year for France.
285-
A CRITICAL EVALUATION OF ANAEROBIC FERMENTATION OF WASTE PRODUCTS
(A MANAGEMENT STUDY)
Authors : M.F. ANDERTON, D.F. GIBBS
Contract No. : EE-B-4-180-UK(H)
Duration : 2D Months 1 Jan 1961 - 31 Aug 1982
Head of
Project : Mr. D.F. Gibbs
Contractor : International Research S Development Co.Ltd.
Address : Fossway,
Newcastle upon Tyne
Tyne and Wear
NE6 2YD
Summary
The main aim of this study has been to evaluate whether the technology
of anaerobic fermentation when applied to waste products can produce a
significant energy saving within the Community in the context of current
primary energy needs. In carrying out the study the authors have reviewed
the current state-of-the-art of the technology and through visits and other
contact with experts in the field have gained an impression of the specific
interests and. current research being pursued in individual Community
countries. Throughout the study, particular interest has been taken in
those factors which may affect the implementation of anaerobic fermentation
techniques in the Community, especially in connection with energy
conservation or produccin.
It can be concluded from the technology review that anaerobic
fermentation has reached a level of development which, in terms of the
efficiency of biogas production, has not significantly advanced over the
past three to five years. Most current research is aimed at improving this
performance either from a better understanding of the microbiological/
biochemical processes involved or by improvements in plant design. From
estimates of available wastes, it is concluded that the maximum possible
energy contribution that anaerobic digestion could make is no more than 3%
of the Community's primary energy demand. However, in practical, terms, when
aLL of the energy requirements of transport, preprocessing storage etc.
are taken into account, the net benefits will be marginal. Overall the
real value of anaerobic digestion lies with its environmental applications.
-286-
1 Introduction
The fermentation of waste products to produce methane has been known
for many years; historically in a non-scientific fashion when 'marsh gas'
was a mysterious and frightening natural phenomenon, and in recent years
more rationally as the potential benefits of 'biogas' have come to be
realised, anaerobic fermentation is' particularly applicable to agro-
industrial wastes because their highly biodegradable character makes
biological treatment an attractive process in terms of efficiency and
economy. The worldwide energy crisis made it fashionable to undertake wide
ranging research and development and applications studies on anaerobic
digestion, not only in connection with waste treatment projects linking
energy conservation with pollution control, but also more generally on
energy production from biomass in the broadest sense, including energy
crops, forestry and algal systems.
The ever-increasing volume of literature on bioenergy is evidence of
the high level of current interest in this area in general and in anaerobic
digestion in particular.
However, in submitting their proposal for a management appraisal of
anaerobic digestion to the Community, the authors felt that a great deal of
time, money and effort was being directed towards too many projects, some
of which appeared to have no positive aim and many of which either over-
lapped in content or were duplications of one another - a view which was
strengthened during the course of the project.
The primary intention of the study was to place the anaerobic
fermentation of waste products for energy conservation (and to a lesser
extent other biomass products] into perspective with current European energy
and agricultural policies. In addition, it was felt that such an appraisal
would show whether the fashionable enthusiasm for anaerobic digestion had
been misguided in some areas.
In planning the various stages in the study, it was evident that the
first major task was to carry out a comprehensive state-of-the art review
to provide a foundation for the project. This was backed up by parallel
discussions with experts in many of the main centres of activity in several
of the individual member states as this was felt to be the most effective
way of collecting the most relevant information on anaerobic digestion
research and development, commercial practices and other factors affecting
its exploitation in the Community countries.
In considering subsequent stages of the project it was necessary to
bear in mind that, in scientific terms, anaerobic digestion originated as a
pollution control method which provided a balanced stabilised effluent for
disposal. Detailed consideration of the conservation or production of
energy by anaerobic digestion techniques is comparatively recent in terms
of research and development activities and application studies; and whilst
it is felt that there may be significant benefits to be obtained by using
anaerobic digestion as a pollution treatment system, there are many barriers
which exist to implementing it in terms of both energy saving and energy
production, particularly on a Community scale, where substantial national
and multi-national finance would be involved.
Consequently, one of the main objectives of the project has been to
identify these barriers and assess their importance in economic and
political terms particularly in the light of the Community's current and
expected future energy demand and waste arisings. The benefits to the
Community in exploiting anaerobic digestion are clearly dependent both on
its significance as a potential contributor to energy demand and on the
limitations imposed by the availability of agricultural wastes and crops;
and relevant information on both these vital topics is presented in the
final project report. Dther factors affecting exploitation such as
environmental considerations, agricultural policy, climatic uncertainties,
etc. are also dealt with briefly.
The conclusions of the study summarise the salient results of the
authors' investigations in the following areas: the present status and
future development of anaerobic fermentation as a technology; its role in
the context of national policies and the Community; and the barriers
preventing its full exploitation as a source of energy. Arising from these
conclusions the authors submit a number of recommendations for consideration
by the Commission.
The project, as outlined above, has now been completed and the results
are shortly to be published as a final report* ' from which the present
severely abridged version has been condensed. Most of the detailed
findings have thus had to be omitted from this summary document although
the conclusions and recommendations appear in a largely unchanged form.
2 State of the art review

2.1 Sources of information
The state of the art study accounted for a considerable proportion of
the total effort committed to the project and involved an in-depth review
(covering over one thousand references of which some one hundred and fifty
were subjected to detailed study), attendance at relevant conferences,
considerable correspondence and other contacts with individuals and
organisations and, most important of all, direct personal contact through
visits with a large number of experts in the field.
This latter activity was important not only in providing an input to
the review but also in obtaining a view of the current situation on
anaerobic fermentation in the member countries and their opinions on future
prospects and barriers to exploitation.
In the final report on the project (and in an existing interim report' '
published in 191) the information obtained by the above means has been
summarised in a state-of-the-art review under a number of well defined
subject headings: namely, biomass and biomass potential (of which waste is
one particular aspect); basic microbiology/biochemistry; digester types;
sewage/sewage sludges; agricultural wastes; food residues; municipal
refuse/landfill; and other sources of methane production. Limitation of
space in this present summary report allows no more than a brief reference
to one or two salient points in each of the above areas of interest.
2.2 Biomass and biomass potential
World net photosynthesis produces annually approximately ten times the
world's annual energy needs but extraction of this energy in a convenient
form is a major hurdle. However, setting aside this difficulty, for
temperate industrial regions like Europe having a high energy and food
demand, biomass could never be expected even in the most favourable of
circumstances to produce any more than a small fraction of the Community
energy needs. On a more restricted front however, the conversion of
biomass waste by anaerobic fermentation processes is relevant to European
conditions although its potential contribution to total energy needs is
still small.
2.3 Basic science and technology
Despite extensive work carried out on the basic microbiology and
biochemistry of anaerobic digestion there are still many uncertainties
regarding the precise behaviour of the bacteria involved in the process and
of the biochemical pathways concerned. Achieving a complete understanding
of their behaviour is difficult owing to their inherent nature; as strict
anaerobes they are difficult to isolate and study, their metabolism is not
fully understood and they have a restricted acceptability of substrates.
Significant improvements in digester technology are perhaps unlikely to be
achieved until a more detailed understanding of their behaviour
characteristics is Known.
2.4 Digester types
The potential applications of anaerobic digesters are wide ranging: to
reduce pollution loading; to conserve energy and/or produce energy; to
reduce objectionable odours; or to stabilise sludges for disposal. Each
of these applications (or combinations) has an impact on digester design.
Current digester designs range from simple, historically ancient digesters
which are still used in great numbers in China (with over seven million in
use), India and developing countries, through the classic digester of
sewage treatment to modern digesters based on improvements derived from
R S D into more efficient waste treatment/energy conservation systems.
Generally speaking the design of pollution treatment digesters is simpler
than that for energy generation systems but all depend on adequate
information regarding the substrate, dilution rate, thermal performance
(i.e. heat loss rate), retention time and temperature.
Some of the newer systems which are briefly reviewed in the final
report include anaerobic filters, anaerobic contact digesters and the upflow
sludge blanket (UASB) process. Hore recent developments on fluidised bed
and similar techniques are also covered.
2.5 Sewage and sewage treatment systems
The importance of sewage treatment and disposal in a European context
is self evident and it is therefore hardly surprising that much of the early
work on anaerobic digestion was undertaken with this application in mind,
particularly in the UK and Germany. Because of the nature of raw sewage
(mainly its low solids content) anaerobic processes are largely used for
treating sewage sludges rather than for primary treatment. Despite this
restriction they are an important feature of many modern treatment plants
and provide substantial quantities of biogas. Typically, it is estimated
that London's ten largest sewage works (serving seven and a half million
people) could produce 2SD,Q00 Pl-Vday of gas from an input of about
6B0 tonnes/day of sludge to offset a substantial proportion of the power
requirements for primary treatment. Current uncertainties regarding such
applications for anaerobic digestion concern not so much technical
capability/performance as economics and, in this respect there appear to be
divergent views on the minimum plant size required to justify investment in
digestion equipment.
More speculatively, as far as technical developments are concerned,
some workers have suggested that plant based on the UASB digester may
eliminate the need for the aerobic stage of sewage treatment. If so, this
would be of considerable economic importance.
2.6 Agricultural wastes
Animal wastes and crop residues are a highly important potential
source of energy at local level with one estimate indicating that some
95D million tonnes of Community agricultural waste was generated in 1979.
It is significant that most development work in the field of anaerobic
digestion has been concerned with agricultural wastes, the major part of
which has been carried out in the US. Much of the information in the report,
particularly with regard to animal wastes, is based on US experience and
this suggests that the required scale of plant to deal with animal wastes
on an economic basis is considerably larger than any in Europe. From a
European point of view, pig and poultry wastes are a more important aspect
of animal waste treatment and most of the larger pilot/demonstration plants
in Europe have been designed to handle pig wastes; very little work on
poultry waste has been identified.
-289-
Until recently little information was available on the anaerobic digestion
of crop wastes but, with increasing awareness of the need for energy
conservation and better utilisation of available materials, this situation
is changing. Agricultural crop wastes in Europe are dominated by straw
(5 tonnes produced in 1976) but the 'surplus' availability is difficult to
estimate. Unfortunately straw is unsuited to anaerobic digestion techniques
in an untreated state and much of the current work in this area (including
a number of CEC contracts) is concerned with improving methods of pre-
treatment.
2.7 Food and food processing residues
Food processing wastes tend to be characterised by large water flows
with high biological oxygen demand [BOD ) containing dilute sugars and
starches. They are therefore particularly suited to anaerobic digestion
techniques primarily as a means of reducing pollution problems although
recovery of valuable materials can be a bonus in certain cases.
Commercial plants exist and are in use in many industries; notably in
the sugar industry whilst others include the fermentation industry itself
(breweries, distilleries, cheesemaking), fruit and vegetable processing and
slaughterhouses.
2.8 Municipal refuse and landfill
Again much of the work in this area has been carried out in the US and
is thus, strictly speaking, outside the scope of the study but the main
features of US developments have been covered in the final report for the
sake of completeness. Plants using anaerobic digesters have been developed
or studied by the Institute of Gas Technology and Dynatech but the most
advanced is the Pompano Beach, Florida plant based on the Refcom process
which has two 31,3GG M 3 digesters.
In Europe, domestic and urban refuse is mostly incinerated as landfill
sites become scarce, with some being pyrolysed or used to make refuse
derived fuel (RD F). Landfill sites can be regarded as 'natural' anaerobic
digesters as the natural breakdown of organic wastes (oncethe small amount
of oxygen is used up) produces methane. Uncontrolled, this can lead to
dangerous situations through the gas seepage and this has led to development
of methods for the collection of gas and to the development of sites based
on the profitable collection of biogas.
2.9 Other sources of methane
Although outside the scope of the study, as they are energy crops
rather than wastes, the possibilities of methane production from algae
(Chlorella, kelp, seaweeds, etc.),from other plants such as water hyacinths,
and from peat have been briefly reviewed.
3 Anaerobic fermentation in Community countries

3.1 Sources of information
As mentioned in Section 1 above, the most valuable source of
information on activities in the Community countries has been through direct
personal contact with active investigators either by visits to recognised
centres of excellence or at conferences and backed by correspondence with
other centres where visits were not possible.
One major difficulty experienced in this part of the investigation was
the recurrent problem of confidentiality when discussing the performance/
economics of commercial equipment or processes and in most cases this has
prevented a detailed economic assessment being made of many of the processes.
3.2 Belgium
Three major institutions were visited in Belgium: namely the
University of Ghent; The University of Louvain-la-Neuve; and the
University of IMamur. In each establishment, discussions were held with
the senior researchers in the appropriate fields.
-290-
In summary, because of forward looking Government policies, Belgium is
in the position of having an advanced research programme in biomass and
anaerobic digestion. They also have significant internal and industrial/
commercial funding for much of the development and demonstration work. It
is felt that the results of many of the research programmes, if verified by
additional financial and technical data, will not provide encouragement for
long-term investment in anaerobic digestion within the Community. However,
the individual research programmes currently in hand should provide good
hard data which will be particularly useful locally and which could be
exploited successfully in the developing and poorer countries.
3.3 Denmark
Work in Demark on three demonstration projects for large farm units has
shown both the potential and problems of using anaerobic fermentation
technology for animal waste treatment. Their economic evaluations show that
many assumptions have to be made, the validity of which cannot be easily
determined but it is apparent that construction costs and operating costs
are generally too high and gas utilisation is underestimated whilst
environmental advantages are beneficial but cannot be easily calculated.
One means of enhancing benefits from anaerobic fermentation that has
always been assumed and is being studied by many groups is that the digested
slurry as compared with raw slurry will improve crop yields when used as a
manure. A two-year study in Denmark on this subject has concluded that there
is no signficant increase in yields over conventional non-digested slurry.
3.4 Eire
A fairly intensive programme in research and development and applied
technologies is being carried out in Ireland at two main centres - University
College, Galway and at several of the Agricultural Research Institute
stations such as that at Johnstown Castle. Funding for most of this work
comes from the Irish National Board for Science and Technology and from the
CEC's Solar Energy Programme.
Much of the work undertaken in Ireland concerns the development of a
two-stage digestion system which can be easily managed in situ on any farm
and can, in theory, be used to treat either animal or crop residues. The
basic unit is an efficient hydrolysis reactor followed by an anaerobic
filter for short retention time methanation. This is similar to the
research being carried out at Wageningen in Holland and in Denmark. The
projected benefits from the operation of a two-stage system for high solids
substrate digestion include reduced capital costs and greater methane yields.
3.5 France
Two institutions were visited in France, the Institut National de
Recherche Agonomique (INRA) and the Institut National Polytechnique de
Lorraine,whilst detailed information was received from the Commisariate a
l'Energie Solaire (COMES), the authority responsible for the national
programme on renewable energy sources.
Although it has been felt in France that it is behind its neighbouring
countries in methanisation techniques, it is catching up rapidly and has a
thriving national programme of research and development and application,
largely funded by three ministries - Research and Technology, Agriculture
and Energy. Results to-date indicate that agro-industrial wastes are
suitable for anaerobic fermentation waste treatment and also wastes from
the larger farms, ideally with more than five hundred pigs or one hundred
cows.
Generally, it is believed that present technologies are too expensive
and that heat production is the best use of biogas, but this in itself
requires a minimum number of animals (e.g. thirty cows). However, before
installing an anaerobic treatment system account must be taken of equipment
-291 -
already existing (such as that for manure handling), its operation must not
take up too much time, the end use of the biogas must be understood and
costed and the agronomic use of the residue must be considered: points
which are not particularly well understood by many of the plant manufacturers.
The French are also aware of the major problem which exists throughout
the Community, that there is, to-date, no flexible, commercially viable
continuous fermentation system for what the French call 'fumier' - that is
cattle waste, inclusive of its straw, bedding, etc.
3.6 Germany
Three organisations were visited in Germany: Agrarund Hydroteknik(AHT)
in Essen; FAL in Braunschweig; and KTBL in Darmstadt. In addition, the
2nd International Conference on Anaerobic Digestion in Travemunde was
attended. The German programme on anaerobic digestion and biogas appears
to be very progressive. Although some university projects are financed
under the terms of CEC contracts (e.g. Wagener's work on Mariculture at
Jlich) most research and development appears to be as a result of intensive
Government backing. The economic assessments to-date indicate a trend that
in terms of energy production and saving, biogas systems will not be
economic for a number of years yet, depending upon basic fuel prices. The
degree of Government involvement (despite certain cutbacks) in the German
programme, suggests that additional funding for research and development
may not be required, particularly as it is known that a number of industrial
organisations are also active in this and related topics.
3.7 Italy
Despite a number of requests for information on the status of anaerobic
fermentation in Italy, no response was obtained and the information in the
final report is based solely on published data and a few personal
communications.
Italy produces quite a high proportion of agricultural residue within
the Community but the average farm unit size (and therefore animals/farm)
is very small. This militates against any serious investment in anaerobic
fermentation technology from an economic point of view and is no doubt the
reason why Pirelli and the CRM (the Italian National Research Council] are
funding research on low cost systems for small dairy farms.
It is believed that there are a number of digesters working on large
pig farms in Northern Italy with a common biogas collection line, but
detailed performance data are not available.
Published information also indicates that several groups in Italy are
working on two-stage digestion systems and on the basic problem of breakdown
of ligno-cellulosic wastes.
3. Netherlands
Visits were made to the Agricultural University in Wageningen and to
TNO, Zeist from which the general impression was gained that the trend in
the Netherlands seems to be away from agricultural wastes towards more
unusual substrates for anaerobic digestion treatment, particularly those
encountered in dealing with a wide range of waste waters.
Views were also expressed that the rapid commercialisation of UASB
type digester systems to large plant scale is possibly premature in the
light of the limited amount of intermediate testing completed. This opinion
was backed by various comments that were made at Travemunde at the 2nd
International Conference on Anaerobic Digestion.
Considerable emphasis was placed on commercial opportunities in third
world countries for the export from Europe of simple, low cost hardware and
expertise, but industries in Holland and elsewhere in the Community appear
reluctant to do so because the returns on investment are too long term.
-292-
3.9 United Kingdom
Visits in the United Kingdom included five in England to Leicester
Polytechnic, the Agricultural D evelopment Advisory Service, Water Research
Centre and Warren Springs both in Stevenage and ETSU (The Energy Technology
Support Unit); three visits were made to Wales to the Polytechnic of Wales,
University College, Cardiff and the Cardiff University Industrial Centre;
and one visit to Scotland to the Rowett Research Institute in Aberdeen.
In addition, the Brighton Conference on Energy from Biomass was attended in
November 1980.
In summary, it was clear from the many discussions that there are
active research programmes on anaerobic fermentation being undertaken
throughout the United Kingdom. D espite this activity, there is no
comprehensive national policy on anaerobic digestion and there is unlikely
to be any strong Government commitment to anaerobic digestion as a means of
energy conservation or as an alternative source of energy. There was
however some indication tnat the need for demonstration systems is being
recognised, but this may be affected by a recent Government decision to
reduce funding for most alternative energy research programmes.
4 Waste arising in the Community and their relevance to anaerobic

fermentation
Clearly, in evaluating anaerobic fermentation of wastes in a Community
context, one important factor to be examined is the potential contribution
that anaerobic digestion of suitable wastes might make to Community energy
requirements and this requires a comparison between energy requirements on
the one hand and the availability of suitable wastes on the other. Both
subjects were considered in the study but space considerations limit
comment in this report to a summary of the sources of suitable waste. As
far as energy requirements are concerned, suffice to say that a figure of
930 million tons of oil equivalent [0) was taken as a measure of the
total Community primary energy requirements in 1980.
With regard to the availability of suitable waste within the Community
(and its potential as a source of energy) the authors, in reviewing the
relevant literature, were unable to reconcile the wide variability in
published estimates. These variations are the inevitable outcome of the
necessary extrapolations and assumptions that must be used in converting
incompletely detailed raw agricultural statistics into estimates of usable
energy. The problems in forecasting arise because of uncertainties in such
factors as the mean daily output from different animals, the usable/
fermentable fraction of total waste produced, its dry content and
proportion of organics, the average efficiency of conversion and the net/
gross energy ratio. In view of these uncertainties the authors used the
CEC's 1981 Agricultural Report, together with some recent French and UK
data to make a simple, approximate estimate of the maximum possible energy
potential of fermentable wastes so that it might be put in perspective with
Community energy requirements.
Three categories of waste were considered; animal wastes, crop wastes
and agro-industrial wastes. The main exclusions in the first category were
sheep and goats as their wastes are essentially non-collectable whilst,
in the second category, forestry wastes and straw were excluded as suitable
fermentation techniques are not available. In examining the third category
of waste, recent French data was taken as a typical basis for extrapolation
to the Community as a whole.
With these underlying ground rules the 1981 CEC Agricultural Report
statistics and the French data referred to above, were used in conjunction
with UK figures for methane yields to give an upper gross limit for
potential energy from waste in the Community of approximately 50 MTOE
-293-
corresponding to about 5.25% of the primary energy requirements.
However, taking into account the many factors that make this latter
figure an overestimate (e.g. only a proportion of the total waste can be
collected, seasonality of crop wastes, inefficiencies in the production and
use of biogas, the energy input required by collection systems, the existing
contribution of gas from sewage and urban waste treatment, etc.), the
authors consider that a more realistic upper limit for the contribution
anaerobic digestion of waste to energy needs is significantly below 3%.
5 Conclusions
For clarity the conclusions from the study have been grouped in a
number of categories and no special significance attaches to the order in
which they appear.
5.1 The basic technology of, and equipment for anaerobic fermentation
5.1.1 There has been little progress over the past three-five years in
the basic microbiology and biochemistry of anaerobic fermentation in terms
of achieving greater yields of biogas by means of increased understanding
of the processes involved. Little improvement is evident in the generally
quoted average yield of 0.33 of gas produced for every kilogramme of
volatile solids digested.
5.1.2 Nor is there any significant evidence of improvements in gas
yield by the alternative approach of advancements in the technology of
equipment, although pilot scale studies in Demark and Eire on two-stage
digester systems for the treatment of straw show some promise.
5.1.3 There is a lack of convincing evidence on the economic
justification for specific anaerobic digestion installations and the authors
doubt whether it is possible to generalise about the economics as each
application must be judged according to its merits.
5.1.4 With the exception of major sewage treatment digesters which,
although well established, are considered to be a special case, there are
few full scale digesters, other than agro-industrial types, operating in
the Community to allow any significant technical or economic assessments
to be made in the context of this or similar studies.
5.1.5 To-date, little attention has been paid to waste handling
equipment required to feed a digester or to the downstream requirements
after digestion - in particular to the use of discharged sludges/slurries
and the storage and use of the biogas.
5.1.B Cattle wastes form the largest usable proportion of current waste
arisings but the inevitable presence of straw in the waste is a major problem
as there is at present no low cost, flexible, off-the-shelf systems available
to deal with straw-containing waste.
5.2 Anaerobic fermentation in the context of national policies
5.2.1 With the apparent exception of the United Kingdom, Eire and
Italy, significant funds are allocated to anaerobic fermentation studies in
the other member states as part of their substantial national energy
programmes.
5.2.2 As part of their investigations the majority of the member
states have conducted their own surveys and forecasts of wastes arising
within their own borders. When estimated in terms of their individual
energy consumptions, the figures are remarkably variable and at variance
with those calculated at a Community level. This is perhaps a reflection
of the dangers associated with the interpretation of the large volume of
statistics available within the Community and individual countries and the
assumptions that have to be made in carrying out what can become complex
calculations.
5.2.3 The scientific/technological status of the individual national
-294-
programmes varies considerably with Belgium being one of the most advanced
in terms of research and development. Comercially, the situation is
difficult to assess: whilst many organisations have at some time been
involved with anaerobic fermentation plant, many have either subsequently
withdrawn or are concerned with a single experimental/trial plant.
Consequently, only a few companies in the Community can be regarded as
being actively engaged in the anaerobic fermentation plant business on a
significant commercial scale.
5.3 Anaerobic fermentation within the Community
5.3.1 Work carried out during the study has indicated that the maximum
possible contribution that anaerobic fermentation of waste could make to
the Community's energy requirements is no more than 3%*of the total primary
energy demand. It is unlikely therefore that anaerobic fermentation of
waste will be a significant factor in reducing the Community's dependence
on overseas sources of energy supply.
5.3.2 However, because of the geographical distribution/concentration
of wastes, it is quite reasonable to expect that useful savings can be made
at a local level within agricultural communities, provided the most suitable
applications are selected by a realistic evaluation procedure.
5.3.3 Similarly there may well be worthwhile applications for anaerobic
fermentation in the treatment of certain agro-industrial waste waters which
would be of financial benefit to the company concerned and of environmental
benefit to the Community as a whole.
5.3.4 The financial structure of the European farming community is so
dependent on the Common Agricultural Policy that the availability of, or
return on capital may be insufficient to justify the installation of
anaerobic digesters and related equipment and if the Community wishes to
promote the use of such equipment then the possibility of offering
financial incentives will have to be examined.
5.3.5 The research and development activities associated with anaerobic
fermentation are spread widely through the Community and the sources of
funding are also spread widely both on a national and Community level.
This is a situation that inevitably Leads to duplication in research
- as evidenced by much of the information currently presented at conferences,
meetings and in the relevant literature published throughout the Community.
5.3.6 Although not strictly within the scope of the project, the
authors consider it appropriate to sound a cautionary note on the enthusiasm
in some quarters for the prospects for energy production on a substantial
scale through the anaerobic fermentation of specially grown crops. For
a variety of reasons but especially in connection with the central issue
of land availability and competition with food production and forest
products, the authors firmly believe that energy production by this means
is not a viable proposition on any significant scale in a European context.
5-4 Factors affecting the exploitation of anaerobic fermentation
5.4.1 The effective implementation of anaerobic fermentation for
agricultural waste'ideally requires the installation of relatively
expensive equipment, good waste handling systems, etc. on large farm units.
In reality a large proportion of the Community farm holdings are small
labour intensive units essentially incapable of justifying or providing
the necessary capital investment. Even for the larger farm units, the
* see comment added to summary
-295-
present economic trends in interest rates, energy costs, feed costs, etc.
are such as to inhibit capital investment in equipment of a non-essential
nature; and applications which would otherwise be ideal for anaerobic
digesters may therefore remain unexploited until the present financial
restraints are lightened.
5.4.2 As yet, there are few specific legal restraints on the design
and operation of digesters but most of the Community countries are active
to a varying extent in the preparation or implementation of codes of
practice or other safeguards. It seems quite possible that when these
codes are finally issued the designers, manufacturers, installers and
operators of anaerobic fermentation systems will have to comply with safe-
guards which will add considerably to the capital and running costs of such
units. This could further inhibit their exploitation except in the most
favourable economic circumstances.
5.4.3 Any decision to install anaerobic fermentation equipment to deal
with agricultural waste will be (or should be) based on an economic
evaluation taking into account the volume and nature of the waste, the
nominal value of its disposal and the value and regularity of the energy
generated set against the investment required. In these circumstances
clearly the effect of climatic variations is important and in the extreme,
should the main crop fail or be substantially reduced, the installation
will fail to pay for itself.
6 Recommendations
B.l It is recommended that CEC funding for anaerobic fermentation
technology be reviewed by the Commission as the authors believe that
significant reductions are possible in the light of the level of funding
already available in the member countries from national and other sources
and in view of the relatively small contribution that the process can make
to energy savings in the Community.
6.2 Such funding as remains should be directed towards research in specific
areas that will advance the technology of anaerobic fermentation
particularly in those areas which may be beneficial to other technologies
in the longer term. Should this lead to significant improvements than there
is no reason why funding levels could not be increased again to assist in
exploiting the improvements.
6.3 It is imperative, if the research is to be cost effective, that it be
closely co-ordinated within a planned programme of objectives and projects
so that unnecessary duplication is avoided. It is suggested that work in
each of the three main areas of interest, namely; basic microbiology and
biochemistry; design and ergonomics; and economic evaluation be
centralised in a limited number of centres of excellence.
6.4 Whatever decisions may be taken regarding the scope of the chosen
programmes of work or their co-ordination, it seems likely from the nature
of the work that it will continue to be carried out in institutions mainly
of an academic character. It is therefore recommended that appropriate
industrial advisors and possibly potential users be given the opportunity
of commenting on the objectives of the proposed programmes and on the
progress of the work to ensure that the ultimate aim of commercial viability
is kept in mind.
6.5 It is recommended that the CEC should examine the commercial prospects
-296-
for supplying the technology to developing nations - the so-called third
world countries as there seems to be a large market in these countries for
simple, low-cost anaerobic fermentation systems backed up by effective
technical assistance. Few commercial concerns are pursuing these
opportunities either because of lack of knowledge of the demand or doubts
about the commercial returns. If the Commission can provide encouragement,
perhaps through market intelligence to quantify the demand and/or by
financial incentives to counter commercial difficulties then it is believed
the long term prospects for business are considerable. In this respect it
is of interest to note that Eastern bloc countries are actively marketing
low cost systems in these countries.
References
(1) Anderton, M.F. and Gibbs D.F., IRD 82/20 Critical evaluation of
anaerobic fermentation of waste products (a management study)
- Final report. Contract No. EE-B-4-18D-UKH - in preparation.
(2) Anderton, M.F. (191) IRD /TM/81-426. Interim Report - Critical
evaluation of anaerobic fermentation of waste products (a
management study) Contract No. EE-B-4-18Q-UKH.
DISCUSSION
Question - Chairman : Is there sufficient information and basic knowledge

available to direct the research and fix the priorities in this
field ?
Answer Yes. Studies which need to be executed before further technology

development takes place are :
1) Fundamental
a) Elucidation of the biochemical pahtways involved in the
production of volatile fatty acids.
b) Establishment of the mechanisms of methane production from
volatile fatty acids.
c) Identification of the most effective organisms concerned
with methanogenesis.
d) Studies on their physiology with particular reference to
operating in mixed cultures.
2) Applied studies
a) Utilisation engineering - A programme to ascertain the
most efficient and cost effective options for biogas usage
is vital.
Until such time as these studies show real promise, resources
should not be deployed in any significant way on empirical
development work.
Experts and institutions exist in the Community who could
make the necessary contribution.
-297-
Question - Prof. Ketelaar, expert : The "waste" from sugarbeets is used
completely,at Least in the Netherlands, as a quite valuable
cattle feed. In the production of patato starch the proteins
in the effluent are already partially extracted and used.
With pig manure the negative "value" is quite high
thus the biogas value may be secondary to the disposal of this
waste.
Answer Yes. This is so in many cases. Not many "residues" are total
waste. What we have shown is the total if all residues were
available for fermentation.
Question - Chairman : You have made an extensive overview and have many
contacts with research people involved in this field. I should
like your personal view on the future development in this field ?
Answer If the problems of removal of H ? S can be solved and effective

systems use of biogas devised, then there is a real future
for use in the process industries, i.e. brewing, food
processing, etc.
Comment - Dr. Ehringer, CEC: In the conclusion of your report you are
stating that the maximum possible energy contribution that
anaerobic digestion could make is no more than 3 % of the
Community's primary energy demand. This conclusion seems to
me much too optimistic. As a matter of fact practically no
energy is saved. Activities in this area are done mainly due
to environmental considerations. Transport and economical
considerations are other essential items to be taken into
consideration.
Answer This figure of 3 % relates to the gross amount if every

conceivable waste arising were available for fermentation.
Further, it does not take into account the energy needed to
fulfill the logistic requirements. Taking all these factors
into consideration, a figure of 0.5 % is nearer a functional
reality. It is agreed that the environmental aspect is the most
important and overall very little energy is saved.
298-
THE RECOVERY. TREATMENT AND UTILIZATION OF
SOLID BEARING EFFLUENTS EMANATING FORM COKE OVENS
Authors : V.J. PATER, G.A. WADE
Contract number EE-B-4-181-UK
Duration : 24 months 1 July 1981 - 30 June 1983
Head of project V.J. PATER
Contractor : THE BRITISH CARBONIZATION RESEARCH

ASSOCIATION
Address : Wingerworth
Chesterfield
Derbyshire S 42 6 JS
UNITED KINGDOM
SUMMARY :
A system for the recovery of the significant losses of coal fines (from
both the preheater and the oven during charging) associated with the use
of preheated coal in coke ovens is highly desirable because of the
considerable loss of energy and the possible pollution problem.
Utilizing the comprehensive facilities at BCRA, initial trials were

undertaken to ascertain the level and nature of these losses. Further
investigations were carried out on the treatment of this waste using
commercial pilot-scale dewatering equipment. Problems of charging-main
slurries containing high ( 25 per cent) tar concentrations were overcome
by the simple expedient of adding crushed coal to the effluent thereby
permitting the commercial equipment to treat a wide range of slurries.
Flocculation of effluents was found helpful in preparing the thickened
slurry prior to final dewatering by filtration. Thickened charging-main
slurries containing up to 30 per cent tar were successfully treated in
the commercial pilot-scale equipment to yield products containing 30
per cent moisture. Mixtures of charging-main and preheater slurries
were similarly treated but the product moisture increased as the pro-
portion of preheater fines increased.
As primary and secondary dewatering of these fines has been found to

be practicable, the investigation is being extended into handling, pre-
paration and utilization of recovered fines.
-299-
1. INTRODUCTION
The preparation and charging of preheated coal to the coke oven is
associated with significant losses of coal fines from the preheater (in
the waste gas), from the oven during charging (in the oven gas) and,
where a pipeline system with a bleedoff is used as the hot coal con
veyance system, in the steam bleedoff.
The total quantity of carryover can exceed 2.0 per cent of the coal
input and thus constitutes a significant proportion of the scarce and
expensive material (coking coal) used. Hitherto, plants in the UK and
abroad where the preheating process is operated have not succeeded in
recovering this material in usable condition and have adopted the simple
expedient of dumping. A ccordingly, there is a considerable loss of
energy in a form of the .valuable raw materials, handling charges are in
creased, and there is a potentially serious pollution problem.
Preliminary research at BCRA^' (involving the use of filtration
and sedimentation centrifuges and vacuum drum filtration) has provided
encouraging results in the recovery of coal fines but a further and
greater effort is required to enable a final and viable solution to be
achieved.
2. DESCRIPTION OF TESTPLANT FACILITIES

The testplant facilities provide for research on the carbonization
of coal on a scale directly comparable with commercial cokeoven practice.
They consist of two large test ovens of different chamber height, com
prehensive equipment for blending and pretreatment of coal charged to the
ovens (including two types of preheater and equipment for charging by
gravity and pipeline) and a complete installation for the recovery of
primary byproducts. Various smaller test ovens provide facilities for
special studies on preheating and extrahightemperature carhonization
and for the assessment of coking pressures. Comprehensive laboratory
support is provided for all testoven operations. A flow diagram of the
largescale testplant facilities is given in Fig. 1.
3. A MOUNTS AND CHARACTERISTICS OF SOLIDS CARRYOVER

3.1. Cerchar preheater
In the preheater, over 8o% of the gas leaving the hotcoal cyclones
is recirculated to the combustion chambers and the remaining 20% is dis
charged to atmosphere via a baffled waterspray washer. The liquid ef
fluent from the scrubber is a copious but weak slurry of finely divided
coal solids. The latter was found to amount to 0.5% (d.b.) or more of
the coal throughput and have the typical characteristics shown in Table
1. The total amount of solids in the liquid effluent from the scrubber
varied from 0.2 to 0 g/100 ml. The fineness of the material (55%
^0.008 mm) limited the range of choice from the available recovery tech
niques. Because of the weakness of the effluent slurry it was evident
that prethlckening would be a prerequisite of any recovery apparatus or
process.
32. Chargingmain of the oven

During oven charging, to avoid contamination of the primary by
product and associated plant, a separate sprayed chargingmain, served
by the second (cokeside) ascension pipe, was used solely for the oven
charging operation. Solids and some tar were precipitated by the sprays
during charging and were run off as a slurry to a recovery system.
Unprecipitated gases were exhausted from the main and discharged to at
mosphere via a venturi scrubber. Liquid effluents containing solids and
300
tar were collected, and quantities and characteristics were determined.
The amounts and properties of tar-free solids lost during charging under
the best operating conditions are given in Table 2. Effluents from the
charging-main contained tar in varying proportions up to ^ 0 % by weight.
k. METHOD AND EQUIPMENT USED

*t.1. Collection and measurement of solid carryover
The system for collection and treatment of charging-main carryover
is as shown in Fig. 2. The slurry passes via a seal box, which prevents
air ingress when the plant is not in use, to a primary settling tank
which serves as a decanter-cum-liquor clarifier; there the solid floats
and sinks are trapped, and the intervening clarified liquor can overflow
to recirculation or to waste. After the oven charge is completed, and a
short settling time, the sinks can be run off, the intervening liquor
discarded and then the floats can be run off - usually joining the sinks
for recovery. By an arrangement not shown, the same manifold can be fed
with thickened carryover slurry from the preheater - for which the bag
filters are equally effective. This method constituted the very first
attempts to collect, measure and analyse the total solid carryover from
each Test Plant preheat or oven charge made use of filters improvised
from ordinary close-weave hessian bags. Up to the present, no better,
cheaper or more flexible method has been found for making individual
measurements of quantities and qualities of carryover on the Test Plant.
Each bag rapidly acquired an internal skin of solids, forming a very
efficient, if slow, filter which operated with negligible loss of solids.
After draining for some hours the contents of the bags were weighed and
sampled, the samples being analysed for moisture and size distribution
and, in the case of charging-main solids, tar concentration.
k.2. Testing procedure

Past investigations had shown that although a commercial continuous
(pusher) filtration centrifuge could successfully dewater thickened
charging-main slurries, its performance was severely limited by the
presence of high concentrations of tar in the slurry (circa 20 to 259a
w/w on dry solids basis). In searching for factors influencing the con-
centrations of tar in carryover effluent from the charging-main, many
individual determinations were made of the quantity and quality of the
dry oven carryover, in which the oven was retained on the charging-main
for varying periods of time i.e. to vary the contact time of carryover
with rich crude gas. From the results of these trials the graph shown in
Fig. 3 was produced in which the concentration of tar in dry carryover is
plotted against time of "oven-on-the-main" period. Fig. 3 indicates a
roughly linear relationship.
This information was then utilised by retaining the oven on the main
for known periods of time, so as to produce carryover effluents with
roughly predictable tar concentrations.
Carryover slurries of high tar concentration (up to 29% w/w dry
solids basis) were produced to which controlled additions were made of
varying weights of crushed coal (80# - 1 mm) in order to reduce the tar
concentration by dilution. The apparatus used is shown diagrammatically
in Fig. 2, coal being added at one of the three points shown. Experiment
showed that tar concentration in the original carryover slurries could
best be controlled by the addition of pro rata weights of crushed coal at
the seal box during the process of oven charging. During this period,
turbulence at this point was at its maximum thus ensuring more homogen-
eous dispersion of the additive coal in the slurry.
-301 -
After dcantation in the separation tank, solid floats and sinks
were discharged to the agitated process tank (the intervening clarified
liquor being discarded) and fed to the filtration (pusher) centrifuge.
Earlier experiments using thickened preheater carryover in ad-
mixture with charging-main carryover as feed to the same centrifuge had
shown only a limited success, the ultra-fine grading of the preheater
carryover having resulted in relatively high loss through the centrifuge
screen and the production of an unacceptable centrate.
Laboratory-scale trials were carried out at BCEA in collaboration
with manufacturers in order to assess the effectiveness of (a) a filter
press (b) a belt press and (c) a novel approach to double belt press
apparatus known as the Squeezor, in dewatering thickened carryover from
the charging-main and from the preheater washer. Both of these mater-
ials were tested separately in bench-scale apparatus after pre-treatment
with flocculating agents. During intensive trials, encouraging results
were obtained and it was concluded that, although the sludges were con-
sidered difficult to dewater and filter presses would be unlikely to be
mechanically or economically viable, there was considerable evidence to
suggest that success would be achieved by use of the Squeezor.
As a result of these findings, arrangements were made for further,
more-extensive, trials using an industrial pilot-scale Squeezor instal-
led at the BCRA test plant. Fig. k shows a cross-sectional diagram of
the machine, indicating the working principles. It consists of three
sections, the first being a flocculator, the second is the drainage
section (incorporating a belt with coarse openings) and the third is
the pressing section, which uses a pair of belts of finer weave.
The materials from the charging-main and the Cerchar washer were
fed to separate decanter tanks, from where the solids were directed to
the main sludge-mixing tank. Steam was used in the tank to regulate
the temperature of the feed. The feed from the sludge tank was trans-
ferred through Mono-type sludge pumps to the Squeezor. A recycle
facility on the discharge side of the pump controlled the flow to the
Squeezor. A recycle facility on the discharge side of the pump con-
trolled the flow to the Squeezor. Both anionic and cationic polymers
were used; anionic polymer was introduced at the inlet to the floc-
culator, whereas cationic polymer was introduced inside the flocculator
chamber to effect proper flocculation. Slurry from the outlet of the
flocculator was then treated in the Squeezor.
Another pressure-assisted filtration machine tested was the
Ecobelt, which is also a development of the Double Belt press; in this
case however, pressure is applied continuously and mechanically rather
than intermittently and pneumatically (Fig. 5)
In the operation of a commercial Ecobelt, thickened and pre-
flocculated slurry is fed on to a primary belt which moves at an appro-
priate speed to permit maximum drainage before the floes are trapped by
the secondary or overbelt. The sandwich then passes over rollers of
decreasing radius, becoming subject to increasing pressure from the
tension and flexing of the belts. Final and greatest pressure is
applied as the sandwich passes over a larger drum and is covered by a
series of side-by-side V belts which are'stressed by hydraulic movement
of the outboard V belt roller. On the commercial-scale machines,
pressure approaching 9 bar can be achieved, but on the pilot-scale
machine pressure was applied by means of calibrated adjustable springs
capable of a maximum of 5-75 bar. After final pressing, the belts
separate and the product is discharged as a cake, the two belts being
spray-washed on the return strand.
-302-
5. CONSIDERA TION OF RESULTS
The addition of crushed coal was successful in reducing the con
centration of tar in hightar chargingmain slurries. From original
slurries containing up to 30$ tar (dry solids basis), the reduction in
tar concentration was pro rata to the weight of coal added provided
that adequate dispersion was ensured.
By feeding chargingmain carryover slurries containing different
tar concentrations to the pusher centrifuge it was established that at
tar concentrations of 15$ or less (dry solids basis), a dry product
containing about 15$ moisture could be achieved with a high rate of
throughput. A s tar concentration increased above 15$ (dry solids
basis) throughput became progressively restricted until at or about
25$ tar concentration, centrifuge cakes failed to slide over the
rotating screens and the machine became blocked.
Results of the Squeezor trials are given in Table k and the _
throughput figures given refer to the total filtration area of O.36 m .
In all cases two polyelectrolytes, AS 37 (anionic) and P W f K
(cationic), were used in order to flocculate the sludge fed to the
Squeezor. Because of pumping problems, some difficulty was experienced
in maintaining a steady flow and accurate dosing; nevertheless, suf
ficient information was collected to assess the performance of the
machine on all the sludges tested. On a commercialscale plant, with
large pipelines, flows and pump speeds would be easier to control, and
flow and dosing problems should be minimized.
Trial A indicated that the moisture content of the product could
be reduced to the order of 32 per cent at a throughput of I38 kg/mTi
under the best operating conditions. Further tests on a sludge
(prepared as for trial A) were undertaken with adjustments to the cake
thickness and cycle time. This trial (B) indicated that thicker cakes
produced greatly improved throughputs at over 200 kg/m with only
marginal changes in the moisture content.
The next series of trials (C, D and E) concentrated on varying the
tar concentration of the slurry. Even at high tar content (29 per cent)
the product moisture was as before at almost 3I per cent and the
throughput increased to 28o kg/m 2 .
Trials concerning mixtures of chargingmain and Cerchar slurries
(F and G) indicated that both the product moisture and throughput were
deleteriously affected as the proportion of Cerchar fines increased.
Trial confirmed that the Squeezor was not capable of treating Cerchar
fines alone.
Slurry brought from an industrial plant consisted of roughly equal
proportions of float and sink materials from settling ponds. The exact
proportions of chargingmain and Cerchar fines were not known, but
estimated to be about 50 to 60 per cent of the latter. Trial R shows
that, with good flocculation, a product of 355 per cent moisture
content could be produced at a throughput of 27*t kg/m 2 h.
Table 5 gives a summary of the results obtained with Ecobelt. The
pattern is similar to that of earlier work, product moisture varying
from just under 30$ on chargingmain slurry and increasing with the
addition of greater proportions of preheater carryover. On commercial
machines working at higher squeeze pressures, it may be expected that
product dryness would be improved.
6. CONCLUSIONS
By the use of additives, solidsbearing effluents from the charging
of ovens with preheated coal can be thickened in commercial decant
303
ation units yielding a liquor clarified sufficiently for recircula-
tion, and wet solids of controlled tar content for further treatment
2. In effluents produced during the charging of ovens with preheated
coal, tar concentrations are related to the contact time between
rich crude gas and the effluent slurry.
Tar concentrations in such effluents may be reduced by
a) reducing the contact time between effluent and rich crude
gas
b) the addition and mixing of appropriate weights of coal
crushed to - 1 mm with the original high-tar effluent.
3 Thickened charging-main slurries containing up to 30 per cent tar
(dry solids basis) can be successfully flocculated and dried to
about 30 per cent moisture content in pressure-assisted double
belt filters. The product is a breakable cake up to 10 mm thick
which is easily handled.
k. Mixtures of charging-main and preheater slurries can be similarly
treated but the product dryness increases as the proportion of pre-
heater fines in the mixture decreases.
5- Flocculated slurries from the preheater were more difficult to
treat and could be dried only to about k2 to 50 per cent moisture
content.
6. Flocculation was successful by using what is considered to be
fairly high rates of dosage of proprietary flocculating agents.
This can add appreciably to the expense of the process. As a
result of research by the manufacturers, new agents appear reg-
ularly on UK and European markets, and it is evident that a
diligent search for the most suitable agents is the best way of
controlling the costs of additives.
7. The two machines tested gave broadly similar performance. The
Hoesch Ecobelt had the advantage of simpler construction and fully
continuous operation. The Squeezor had the advantage of complete
sealing of the edges of the pressed area, thus ensuring that even
with the most mobile of floes there was no loss at the edges when
pressure was applied.
The results of the Phase 1 trials have shown that primary

(thickening) and secondary dewatering of the fines is practicable.
Accordingly, the investigation is being extended into handling, prep-
aration and utilization of recovered fines. (Phase 2 ) .
7. ACKNOWLEDGEMENTS
The British Carbonization Research Association wish to acknowledge
the excellent co-operation and collaboration given by the following
companies:
Degremont Laing Ltd - Borehamwood and Paris
The Alfa-Laval Company Ltd - Brentford
Eberhard Hoesch 8c Sohne - Duren - West Germany
Allied Colloids Ltd - Bradford
Reference
1) Carbonization Research Report 62 - BCRA. December 1978
304-
Table I * pical characteristic* of Cerchar-preface ter carryover
solids
Ash Vii.h. 25(6)t

Volatile mutter '..d.h. 23(31-33)
BS swelling number '/i<8>
l*iinmiltHi \izt v. Cumulativi'. %

>2S4 2.4
>20.l 8.2
>I6.0 11.8
>I2.7 19.4
>I0.1 28.7
>8.0 45.1
>6.4 66.9
>S.O 83.6
>4.0 95.6
t Figures in parentheses are values for parent charges.
Table 2 Typical amounts and characteristics of (tar-free) solids lost during charging (best operating conditions!
Cerchar-preheated. Rosin-preheated, Pipeline charges:

Solids (dry. lar-free) pipeline- and gravity-charged steam bleed-off
gravity-charged
Amount. % of charge weight =0.41.0 <0.4 =.3
Size, mm
>3.35 % 0.4
>2.0 % 0.7 Nil 2.5
>1.0 % 2.0 1.0 7.5
>0.5 % 7.0 4.0 15.0
>0.25 % 17.0 7.0 21.0
>0.125 % 30.0 21.0 28.0
>0.063 % 55.0 40.0 35.0
Ash.d.b. % 7.0(6.0)* 8.0(6.0) 8.0(5.0)
Volatile mailer, d b . % 25.0 22.0(33.0) 27.0(31.0)
(31 0-33.0)
BS swelling number 1 to 3 (8) Vi (8) 8(8)
Tar content % variabl : 10to40 Nil
Figures in parentheses are values for parent charges
-305-
Tibie 3 Test data: fines carryover from the charging main
Duration on Wel Water in Added Dry Tar in

( hurgi' mi. charging solids, wel solids, coal, solids dry Remarks
main,t kg It kg (including solids,
min-sec tar), kg It
17 AV/74/80P 13-0 197 41.5 Nil 115.2 8.1

17 AV,79 80P 10-0 * Nil 4.2
1309 21-0 * - Nil 25.0
1310 14-0 # Nil 10.9 Good run
1311 28-0 # - Nil 22.3
17 AF /80 7-0 Nil 4.2
17/AF 3 80 6-0 400 34.7 Nil 261.2 1.2
17/AF.4/80P 17-30 231 38.3 Nil 142.6 14.6
I7/AF/5/80P 25-0 255 39.0 Nil 155.6 24.9
17 AF 6 80 25-0 80kg <3mm 25.4 Poor mixing,
at mixer bad run
lank
17, AF/7/80P 25-0 - # 80kg <3mm 18.4 Good mixing,
at seal b o x * * improved run
I7/AF 9/80 25-0 * 53kg <3mm 11.5 Hard cake,
AF 27kg loss in
Cerchar** centrate
I7/AF/1I/80P 25-0 * 80kg <3mm
at seal b o x * *
15.4 Good run
I7/AF/I2/80P 25-0 # 80kg < l m m

at seal b o x * *
8.4 Very good
run
I7/AF/I3/80P 11-05 346 48.8 Nil 177.2 6.1
17/AF/I4/80P 25-00 * 53kg AF 5.2 Good run
O r a m 27kg
Cerchar** "
17/AF/I5/80P 11-10 318 51.5 Nil 154.2 6.5
17/AF/23/80P 8-0 282 30.3 Nil 196.5 5.3
I7/AF/24/80P 8-0 312 35.2 Nil 202.2 3.3
I7/AF/25/80P 8-0 304 37.4 Nil 190.3 12.9
I7/AF/26/80P 8-0 295 41.2 Nil 173.5 21.7
no reliable carryover data available, solids dewatered by centrifuge.

##coal added al the seal box.
tcharging times of 4.3 to 8.9 minutes, depending on charge-bin pressure and charging-steam flow rate.
-306-
TAhLE . PILOT SQUEEZOR TRIALS RESULTS
Feed slu ry content Flocculant 5, kc/tdG Weicht of Cyclo Product Product Cake thickner.rj rijr
Time
time, dryness to presa, nm
' Sludcc type onwnain
min Solids: Tar: Anionic Cationic
3 cakes,
kg sec. S
throughput
kg/m
s/100 ml SSw/w d.b.
A chargingmain, 25 20.7 23 2.36 70 64 72 5

highvolatile 307 52 65 128 9
blend " 315
3.00
50
70
68.5
67.5
138
93
9
9
ao for A 25 12.2 23 1.8 1.9 58 47 68 282 15

5.8 56 69 232 15
C chargingmain, 25 10.3 9.4 2.0 35 42 C-'-; 319 15

medium 4.0' 55 63 212 15
volatile blend .; 51 6J 216 15
54 63 6C V)k 15
5.0 80 67 140 15
D as Tor C 25 13.1 29.7 2.0 .0 - 45 6.7 230 15
E as for C 30 11.7 16.4 1.5 2.0 't.8 48 68.t 231 9

4.7 't? 68. 230 9
5.9 45 72.6 315 15
F chargingmain: 5.9 10 2.8 47 71 141 15

Cerchar, 1:1 (approx.) 3.6X 47 70 13"* 15
0 as F, but 1:2 8.7 5 2.0 30 48 59.6 124 5

(approx.)
H Cerchar nil 50 -
R industrial ~ ~ 11.'t ~ ~ 2.62 52 56 94 5
ohargingnain: 11.lt 2.56 8? 58 56 5
Cerchar, 2:J 13.0 53 42 64.5 274 15
"Weight of 2 cakes h e i g h t of 4 cakes

TABLE 5. PILO! ECOBELT TRIALS RESULTS
Slurry Tested CM HTCM HTCM/ HTCM/ Industrial Cerchar

only only Cerchar Cerchar CM/Cerchar* only
1:1 1:2
Total solids g/l 100 109 100 110.5 112 103.^

Flocculants AC1956 AC1956 AC1956 AC1956 AC351 AC351
Dosage kg/tonne d.s. 0.25 O.23 O.5O O.23 05 0.48
Draining Time 3 m. 1m.10s. 3m30s. 3m.20s. 7m.10s. 2m.30s.
Pressure bar 575 5.75 575 575 575 575
Cake moisture % 315 3O.O 33.I 37.1 36.5 41.6
Filtrate solids g/l 2.2 06 O.23 0.73 0.29
Estimated to be hOfc CM, 60& Cerchar

CM : Chargingmain slurry
HTCM : High Tar chargingmain slurry
Cerchar: Cerchar preheater slurry
LEGEND I. G a i - m a k t meter 2. W a i h - o i l regeneration 3. B enzoic scrubber 4. Ammonia rather S.Dctarrer
6. Exhauster 7. Coolen . Coal mixer 9. Coal crusher IO. - t oven
II.Rotln prchcatcr 12.Cerchar preheater 13. 17t oven 14.Coke weighbridge 15.Coke quencher
FIG. I. B RITISH CARB ONIZATION RESEARCH ASSOCIATION TEST PLANT FLOW DIAGRAM
To venturi
scrubber
Spray system
Charging main
Com-siae o
charge hole
Non test
Crushed coa charges
hopper
Calibrated
vibrator
feeder
Possible
,, clear liquor
Alternative
recirculation
crushed coal
feeds
Possible feed of
cerchar washer
slurry from
collection tank
To waste
Release to
waste OP to
filter bags
Agitator
Thick slurry
mixer Ejector
or pump
FIG.2. SOLIDS - RECOVERY SYSTEM

1 1 1 1 1 1 1 1 1 - 1 1 1 1 1 1
30
26
22
_ L _1 L
14 22 26 30 34 3 42 46 50
Tar concentration, / ( d r y basis)
FIG 3 PIPELINE CHARGING, PLOT OF TAR CONCENTRATION OF CARRYOVER AGAINST O N - M A I N TIME

1 The flocculator A Pressure plate
2 Draining table C Mobile press carriage
3 Lower filter cloth D| Lower drainage surface
4 Scraper &2 Upper drainage surface

5 Distributing roller Ml Upper membrane
6 Variable-speed drive roller M 2 Lower membrane
7 Upper filter cloth J Edge seal
Mobile pre
9 Wheel
10 Supporting frame
11 Flexible air connection
12 Pneumatic cylinder presi
13 Pneumatic cylinders traverse
14 Scraper
15 Filter cloth washing chambers
16 Pneumatically operated pinch rollers
FIG.4-THE SQUEEZOR
(Reproduced by kind permission of Deg remont - La ing Ltd.)
High - pressure
stage
Draining stage
/
Low - pressure stage
FIG. 5. THE HOESCH ECOBELT

DISCUSSION
Question - Chairman : What is the chance of implementation of the fines in

the process more special as the industry is reluctant to make
investment and in the field of "coal" high investment is required
to modernise and/or to open new coal mines and processes ?
Answer Systems of collection of slurry containing solids are already

available on most of the plants with preheating and may need
only some modifications and/or an addition of flocculents
The capital expenditure for the plant necessary to dewater
fines need not be excessive as the best choice can be made
using available technical information. The return on the
capital is very high once the fines are utilized and therefore
even in this difficult times worthwhile.
-314-
SESSION VI CATALYSIS
Chairman: J. UMIDO
Summary of Session VI
Catalytic fuel production using water and coal
FischerTropsch synthesis of hydrocarbons in the gasoil range
Valorisation des asphaltes par hydrognation trs haute

svrit sur catalyseur entran

315
SUMMARY OF SESSION VI
CATALYSIS
J. Limido, P.A. Pilavachi
It is now quite clear that whereas fossil hydrocarbons (crude oil and
natural gas) wiLL assure a large proportion of energy production during
several decades to come - coal, nuclear energy and renewable energy
resources, in the future - will take an increasing share.
This was the framework within which the various contractors grouped
together under the heading "CATALYSIS" described their aims in discussions
which were held at the meeting of 20 Hay 1982.
Their aim is to valorise well defined compounds, such as hydrocarbons and

oxygenated compounds which are used in specific fields and which emanate
from coal or petroleum residuals.
The first two projects - entrusted to the KATHOLIEKE UNIVERSITEIT, LEUVEN

and the UNIVERSITE DE LIEGE - have the same objectives : to achieve
decisive advances in applying different synthesis processes to a synthesis
mixture (CO2 + H2) which will yield firstly hydrocarbon cuts (petroleum
and gas-oil), and secondly higher alcohols or methanol (e.g. ethanol).
Fischer-Tropsch synthesis of hydrocarbons in the gas oil range - The

progress expected from the research work entrusted to the KATHOLIEKE
UNIVERSITEIT, LEUVEN is a substantial increase in selectivity; this will
allow a considerable improvement in the economy of presentLy-known
processes based on the FischeiTropsch reaction.
The first results obtained show that with a particular type of cataLyst it
is possible to achieve selective cuts, differing from the usual type of
product distribution obtained with the Schulz-Flory kinetics. The results
obtained in a batch reactor are to be confirmed on an on-line apparatus.
Catalytic fuel production using water and coal - The research undertaken
by the INSTITUT D E C H I M I E D E L ' U N I V E R S I T E D E L I E G E is more explorative in
nature and covers the development of new complex catalysts which can
replace conventional types of catalyst, the behaviour of which with regard
to the synthesis of oxygenated compounds are neither active or selective
enough. The first research results led to the selection of two new complex
catalysts likely to improve the gain in selectivity and further research
will be concentrated on them.
-316-
Hydrogenation of asphastic residuals - T he third project is operated
jointly by CERCHAR (Centre d'Etudes et de Recherches des Charbonnages de
France) and the INSTITUT FRANCAIS DU PETROLE who propose a new way of
transforming heavy oil residuals extracted from oil (asphalt) or from
primary hydrogenated coal into commercial products in the petroleum or
diesel-oil range.
Hitherto, conventional techniques have failed because of the unadaptabi lity
of present-day commercial catalysts and because of the resulting complexity
of the equipment. It is however reasonable to hope that a more rational
and more economic solution can be obtained by following the approaches
proposed.
317-
Catalytic fuel p r o d u c t i o n using w a t e r and coal,
Authors J. D r a p i e r , A.J. Hubert and Ph. Teyssi
Contract number EED.2.326.B ( R S)
Duration 30 mois 1 Mai 1981 - Octobre 1983
Head of Project Ph. T E Y S S I E , Universit de Lige
Cont rac t or Universit de Lige
Adres s U n i v e r s i t de Lige
Institut de Chimie
SART TILMAN 4000 LIEGE
Summary
The rese nt investig ation is devoted to the searc h for
nove 1 or imp rove d catalysts for the efficient and select ive
synt hes is of oxy genated che micals (> C2) direc tly from C 0 and
2 In this proj ect, the pr eparation of m e t h a n ol homolog ues
(eg e thanol , pro p a n o I s , but anols) is a particu larly att ractive
chai lenge be caus e these ale ohols would be inte res ting fu els
but could be als o easily co nverted to v a l u a b l e olefins ( e t h y l e -
n e , propyl en e, b u t enes) : such an achievement would be s i g n i f i -
cant in rela t ion with the roblem of the repla cement of p e t r o -
1 eum by coal . Preparat ion o f other oxygenated compounds and
of o lefins w oul d be also of i n t e r e s t . In the resent rep ort ,
we s ummariz e the preliminar y study of some mod i f i ed and o r i g i -
nal catalyti c sy stems (R h, Ru derivatives asso ciated to a
Lewi s acid o r to C u B r ; (Cob alt c o m p l e x e s ) , som e of them (par-
t i cularily C o ca talys ts) gi ving promising resu lts in the c o n -
text of oxyg enat ed product p r e p a r a t i o n .
The a ppar atus used f or the present study will be d e s -
crib ed brief ly t oge ther wit h a short discu ssion of the m e c h a -
nist ic aspee t s of the react ions .
-318-
Introduction
Since the impact of the petroleum crisis on the economy

of the western countries, important efforts have been devoted
to the search for novel energy supplies. The use of coal as a
fuel precursor is known since a very long time and received
much attention when problems of energy shortage were particu
larly accute .
The general conditions for considering the replacement
of petroleum by coal are again encountered in our countries.
However, two different approaches of the problem are
intensively investigated in the industry :
the more classical one consists in transforming coal
to CO and hydrogen (water gas) and use the synthesis gas for
the preparation of a hydrocarbons mixture analoguous to gaso
line.
the second approach consists in using synthesis gas
for the preparation of more valuable chemicals which up to now
are mainly available from petroleum or valuable biological
resources : as examples we have the preparation of olefins
(which is reasonably controlled at the present stage) and of
oxygen containing chemicals such as : methanol, ethanol, higher
homologues (propanol, C4 alcohols, ... ) acetic acid, esters,
ethers . . .
The problem of methanol preparation from synthesis gas
is already an important process at the industrial level.
The preparation of acetic acid from methanol is no more
a problem since the discovery of the Monsanto process and the
conversion of methanol to hydrocarbons (synthetic gasoline) is
also possible (Mobil process).
However the discovery of selective and economically
efficient processes for the direct synthesis from synthesis
gas (without the preparation of methanol as an intermediate
step) of more elaborated molecules such as ethanol, acetic
acid or of their homologues (propionic acid, butyric acid,
propanol, and tbuty1alcoho Is) remains an unsolved problem.
In fact there are several indications that these challen
ges will receive solutions in the next future : several patents
report on the preparation of ethanol from C0:H2 with reasona
ble select ivi t i es but relatively large amounts of very expen
sive catalysts (Rhodium) + are required in these processes (1,
2).
It is noteworthy that ethylene glycol production direc
tly from C0:H2 has been well studied by Union Carbide but the
process has not been developped further than the pilot stage
because of the economical drawback due to the use of the
expensive Rhodium catalyst and very high pressure.
The use of less expensive metals than Rhodium seems
therefore to be requisite for industrial application except if
the life and efficiency of rhodium based systems could be subs
tantially improved, the more important drawback of the existing
processes being however the lack of selectivity (the undesi
rable formation of hydrocarbons being the main side reaction).
Hoechst process uses 11 g of RhCl on 40 g of support

Cobalt catalysts may meet these requirements if their
selectivity could be improved (cobalt catalysts are in fact
efficient for the homologation reaction of methanol to ethanol
and of course much cheaper than rhodium based catalysts) .
I. Mechanistic aspects
The CO hydrognation mechanism is a challenging problem
for chemists and constitutes in fact the most remarkable in
tellectual ac hievement of the last few years, comparable in
fact to the s tudy of the Ziegler and metathesis reactions,
Howeve r, the problem of the understanding of CO hydro
genat ion is ot yet completely solved : several mechanisms
have been rec ognized to be acting under different conditions,
It is now rather well accepted that most of the Fischer
Tropsch relat ed processes are proceeding through the formation
of carbenoid intermediates, as example in the Olive's mecha
nism :
H
CO CH'
MH Cl MCH * MCH 2 OH
II I II
0 H 0
H
V H 2 0
MCH.OH * M=CH ethanol, ethene and homologues
H
Oxygenated products would result from the formation of
metal formyl species or more probably (for thermodynamic rea
sons according to some authors) as hydroxy carbenoids (4)
H OH
H OH OH
\ / - H' 2, 0" \ / 2H, \ / <?
2 CC CC- C -> C HC3
Il II II -, II H
M
M M M
The poly condens at ion of hydroxy carbenoids may explain

the formation of oxygenated products, the reduction of the mo
nomeric species giving methanol.
Dehydration of hydroxylated intermediates leads to car
benes and the formation of olefins can thus be explained.
H H .H CH,
II H 2 0
MC CH. M=C
II \
CH, CH,
H OH
But olefins such as ethylene can also result from elimina

tions :
MCH 2 CH 3 MH CH 2 =CH 2
Alcanes (eg ethane) result from hydrogenolysis of alkylmetal
bonds.
Insertion of CO into the m etalalkyl bonds has been pro
posed in other mechanistic propos als (eg : Olive's mechanism,
see above).
According to Pettit (5) un usual select ivi t ies such as
the formation of C3 and C4 olefin s would be the result of the
reaction of lower olefins with me thylene carbenoid (ethylene *
propylene * C 4 . . . ) . This author s tressed the role of bridged
methylene carbenoids in such reac tions.
However, the high selectiv ity for propene may be pro
bably best explained by the oligo merisation of three carbenoid
units on clusters presenting part icularly well defined struc
turai and dimensional features ex cept if a particularly strong
preferential reactivity of ethyle ne as compare to propylene
is pos tula ted.
Formaldehyde has however b een proposed as an interme
diate particularly in ethylene g lycol synthesis (6). In the
actinide series, the "side coordi nation" of CO has been pro
posed to explain the formation of oxycarbenes which dimerize
readily to give stable dimeric co mp1 exes .
II. Ins t rumen ta t ion :

Most of the efforts related to the first period of the
present contract have been devoted to the installation of pro
per high pressure equipment and to devise adequate techniques,
particularly to solve the intricate analytical problems.
Our experiments are run under two main sets of condi
tions :
1 . Static experiments :
In fact, most of our investigations are performed in
stainless steel autoclaves, a pressure of CO/H2 mixture being
applied at the beginning of the experiment. In fact, it is
necessary to study the catalytic activity of our catalysts
under a large scale of pressures as the efficiency and selec
tivity can be dramatically affected by this factor :
a) High pressure experiments (300 bars) are currently run at
the present stage in our laboratory in small autoclaves
(125250 ml) .
The autoclaves are heated with an electrical furnace and
mechanically shaken. (Prolabo system).
The original autoclaves have been modified to allow the
direct injection of air sensitive catalysts. Higher pressu
res (up to 1000 bars) are now available in our laboratory
by using a compressor (NOVA ) in association with an auto
clave of 1000 bars.
b) Low pressure autoclaves (up to 15 bars) are also available.
They have been constructed in the University.
c) A few experiments have been run in sealed glass tubes. This
technique was useful for studying very sensitive catalysts
under low pressure (1 bar) (eg Ti Br4 Ru or Rh carbonyl ).
The tube is heated while rotating in an electrical furnace.
motor
furnace
f rag'ile joint
The gases are analysed by GLC and MS after breaking the

fragile joint with a magnet,
d) This technique was used only in some exploratory experi
ments and is no more applied in our laboratory.
2. Dynamic experiments :
When the expected compounds are not very stable under
the operating conditions, it is of interest to remove them
as readily as possible from the reaction medium. We have there
fore built two devices A and for low pressure experiments
(glass reactors and traps), whereas a system suitable for wor
king under 100 bars is under construction (the whole apparatus
is made in stainless steel and the feed of gas will be assumed
by a mano stat.
The low pressure apparatus A , suitable to work at 1
1,5 bars, can be used for comparison of our catalysts with the
Ichikawa and Hoechst processes.
trap Feed
Reactor
><lBy pass
&
manometer I flowmeter
-322-
The low pressure apparatus is used for liquid phase
experiments with catalysts dissolved or in suspension in a
high boiling solvent.
The high pressure system will be used to run experiments
under the Hoechst's patent conditions : the general scheme will
be very similar to (A ) but the working pressure will be main
tained at 100 bars.
3. A nalytical techniques : The reactions run in the low pressi)

re circulating sys tem ar e followed directly by gas chromato
graphy. he li quid orti on collected in the cool trap is in
j ected in to th e chromato graph when the reaction is finished,
Th e rea c tion prod ucts of the experiments run in high
pres sures auto cl aves are analysed after completion of the reac
tion. The gas i s analyse d by GLC by using freon as a standard,
whereas t he li quid hase is condensed in a cool trap, the auto
clave bei ng he a t ed under vacuum during this operation,
Th e pro duc t s are identified by the GLCMS method,
By combin ing f our di ffer ent columns (see below), it is possi
ble to so lve s at isfa c torily the analytical problems encounte
red in th e pre sent s tudy
Qu anti tative anal ysis of the liquid fraction is perfor
med by us ing ropano 1 as a standard : the chromatogram of a
sample of the analys ed 1 iquid fraction is compared to the
chroma tog ram o f the same sample to which an known amount of
propano 1 has b een ad ded.
ANALYSED
APPARATUS Detector Column PRODUCTS
INTERSMAT P0RAPAK liquids

FID
IG 1 20 70220C(107min) gases
IDEM IDEM FFAP

60220C(10/min) liquids
VARAN SPHER0 CA RB H20, 02, N2,

CATHAR
3 700 IS0TH.50 0 (6 min)
CO, CH 4 , co 2
50150C(107min)
IDEM IDEM CHR0M0S0RB 102 alcohols

60220C(107min) water
III. Progress of work and results

Beside the solution of technical problems, the first
period of the present program was devoted to an exploratory
approach of novel catalytic systems for CO hydrognation with
particular attention centred on alcohols (eg ethanol) synthe
sis.
The early exploratory approach was based on low pressu
re experiments as the high pressure equipment was not fully
323
operative. Most of the first investigated catalysts appeared
to be very inefficient and most of the results were poorly
reproducible (eg. attemps with CuBrKBr and with M x (CO)y on
AI2O3T1CI4) .
More reproducible results were however obtained with
TiBr4 Ru(C0)i2 see below) but the best performances as far
as alcohols (and particularly ethanol) formation is concerned
are obtained with supported (expensive) rhodium catalysts.
However, promising results for oxygenated compounds
synthesis were also obtained by using reduced cobalt catalysts.
Summary of the preliminary experiments :
1 . The use o I . The use of fusedsaltswhere an acidic (Lewis acid) cocata

lyst is pres ent : according to Muetterties (8), such catalysts
lead to unus ual s elect ivi t ies in alkanes but the production
of oxygenate d compounds has not been reported with his systems
(Ruthenium, Rhodium and Iridium carbonyl in very strong Lewis
acid such as AICI3) : in fact the extremely acidic conditions
render the e xistence of oxygenated products extremely impro
bable.
Our approach consisted in using CuBr, a much weaker Lewis acid
than AICI3, as cuprous salts are known to coordinate efficient
ly CO.
The C uBr KBr eu tec ti c (mp 190C) was used in associa
tion with Ru 3 (CO)12 ( compo sition of the catalyst : CuBr

(3.IO 3 mole s) K Br (7. 6 103 m) R u 3 ( C O ) 1 2 (3.10 5 m ) .
The r eact ion wa s run in glass reactors (p = 12 atm)
and in autoc lave s ( = 101 5 atm, T=180200C).
The s yn thes is g as wa s a (1:3) mixture of CO and 2
We ha ve t hus ob serve d the formation of water, propene,
propane, met hyl and et hyl b romide, methanol, ethanol and
npropanol i some cas es .
Probi ems rel ated to autoclave corrosion (with formation
of potential cat alyt i c spec ies) and sublimation of Ru3(C0)]2
lead to cons ider able v ar iat ion of product compositions : hy
drocarbons ( C]C 5) are form ed in these cases,
A typ i cal exper iment s was performed under the following
condi t ions
CuBr 430 mgm CO/H2 = 2/1 (120 bars)
KBr 913 mgm = 200C
KI 60 mgm
t = 190 h
Ru 3 (CO),2 19
The liquid condensed in a cool trap after reaction contained

CH3OH (15%), C2H5OH (11), C3H7OH (42), acetic acid (23),
C2H5COOH (10) (no calibration, rel % ) .
An attempt without potassium iodide and based on Rho
dium catalyst afforded also oxygenated products :
CuBr 430 mgm C0/H 2 = 1/1 (150 bars)
KBr 9 I 3 mgm T = 200C
Rh6(C0)16 32 t = 190 h
324
The following mix ture of oxygenated products was
obtained : CH3OH (7,5%), C2H5OH (11), iso C3H7OH (?)(7)
n.C 3 H 7 OH (27), CH3COOH ( 40) , C2H5COOH (7,5) .
The results could not be reproduced when contact bet-
ween the catalyst and th e stainless steel autoclave was sup-
pressed by using a glass tube inside the reactor.
In some at temps, acetone, t-butanol, propene and an
unidentified product (C7 H]20) were obtained as the main com-
ponent s.
However, the lack of reproductibility of these results
led us to postulate the participation of active species formed
by reaction of the salts with the autoclaves,
Attempts performe d with a strong Lewis acid such as
TiBr4 and Ru3(CO))2 led to the formation of hydrocarbons as
main reaction products ( propane, isobutane, isopentane and
C6) together with haloge nated products (methyl, ethyl and
propyl bromides).
Table I
TiBr 4 -Ru 3 (CO)12 CO/H 2 =l/2(l bar) = 190e
Flask (IL) sealed tube (20 mL)

Hydrocarbons (relative %) (absolute %) *
24h Uh 7 days
Cl - 0 0, 6
CH 3 Br 1 1 - -
C
2 - 0 0,8
C 2 H 5 Br 12 - -
C 6 12 0,8
3
C 3 Br 2 - -
C
4 43 49 0, 9 (iso:0,5;n:0,4)
C 26 31 0,45 (neo?0,006;iso?0 25
5
. n?,0,20)
C
6 1 9 7 0,02
Ru 5 m gm 20
Ti 300 150
* absolute % calculated by using freon as a standard for

GLC determination.
-325-
In opposi tion to Muetterties' result s, we did no t ob-
served the simultaneous formation of methan e and ethane, the
C 4 and C5 hydrocarbons being predominant wh en the reaction was
run une er a pressure of 1.5 bar i n a glass (1L) apparatus.
The halogenated products are formed at the very begin-
nin g of th e reaction wh ereas the C4 and C5 fractions increase
wi th time. It seems therefore that halgena ted products could
be the precursors of hi gher alkanes in this case : the absence
of CH4 and C2H6 during the early stages of the reaction could
thus be ex plained by th e ready formation of the higher alkanes
from CH 3 Br and C2H5Br.
The use of Ir4(C0)]2 or Mn2(C0)]0 i nstead of RU3(CO)j 2
(in sea led tubes) leads to the preferential formation of C]
and C2 hyd ro carbons. (Table II) .
Table II
Rh 6 (CO)16:10 mgm Ir4(C0))2:20 Mn2 (CO) 10: 10

c
n H 2n+2 TiBr4 : 300 TiBr4 : 100 TiBr 4 : 200
1 0,25% (f 3 6
2 0,16 0,06 0,02
3 0, 6 0,9
4 0,25
5 0, 6
6 0,8
7 0,3
absolute % calculated by adding freon as an internal

standard after reaction.
The experiments performed in flask gave reproducible

results as far as the qualitative composition of the gas phase
is concerned whereas the attempts in sealed tubes were poorly
reproducible but as these TiBr4 based catalysts did not afford
oxygenated products, they did not fit therefore with the gene-
ral objective of the present program and were consequently no
more investigated.
However, further experiments (eg in the presence of
added alkyl bromides to the reaction medium to check their
eventual participation to alkanes formation) could be of inte-
rest and will be eventually studied latter.
-326-
2. Reduced Metal systems
Blanchard, Petit and Mortreux (9) publications on a
cob aIt (II) AIR3 catalyst generated in the presence of buta
dien e suggest the participat ion of Cobalt species similar to
tho se reported by Bnneman ( 10) for ace tyl enesni t ri 1 es coey
clot rimerizations to pyridin es rather than clusters as sugges
ted by the french authors.
We have therefore pre pared reduced metal catalysts by
trea ting the corresponding s alts by NaBH4 (in methanol or
ace tonitrile) in the presene e of ligands. The mixture was eva
pore d in vaccuo and used dir ectly for the experiments. The
salt s used for the catalyst preparation are : C0CI2. RuCl3,
NiCl 2, MnCl2 (Table III)
Table III : (experiments 1,2,3,4 of the first progress report)
Redu ced tr ansition metal sal ts as catalysts for CO hydrogena
tion (prel iminary experiment s ) . Solvent : CH3OH.
Catalyst Gas Li quid
CoCl 2 C Organic layer:C. C ,

4 > > C 2= W
C3H5>C5>CAH8>C5> Aqueous layer:traces of
C6>C2H4 CH 3 OH,C 2 HOH,npropanol,
nbutano1
RuCl, C 0rg.:C 4 C ] 3
rC6' C0
2
Aq.:CH 0H>nbutyrald(ehyde)>
npropionald , >acetald,
>npropanol>ethanol
NiCl 2 C0H 2 Org. :0

Aq. : traces of CH OH
CoCl 2
WW Org.:C 7 >C 6 >C 5 >C 4
+ MnCl
WC7 (+ unidentified products)
Aq.:CH 3 0H>butyrald
>propionald>acetald
>npropanol>ethanol
* Cn correspond to alkanes (n given), CnH.n refers to

olefins.
A typical attempt in acetonitrile as solvent afforded
the distribution of products reported in Table IV in the pre
sence of an added ligand.
Table IV
CoCl 2 (5mmoles), NaBH, (10 mmoles)
CH 3 CN 20ml, L : 10 mm oles
C0/H 1/2, 120 bars, V = 50 bars

= 200C, t = 62 hr
Gas Liquid *
CH 4 13 3 mgm CH0H
3%(relative)
C 2 H 5 OH
C 28 propanol s 33
2H6
C 51 butanols 15
3H8
C 32 pentanols ? 12
4H!0
+
C 20 XYZ 35
5 H 12
C 14
6H.4
other compounds than alcohols and unidentified products
In fact, the cobalt catalysts gave the best results as

far as alcohols are concerned and their op timali sa t ion is
presently studied.
3. Preformed reduced CO species :

The results obtained with reduced metal salts prompted
us to investigate preformed pure complexes presenting a low
oxidation state, our present research is in fact mainly cente
red on this aspect.
4. Supported catalysts :
Preliminary experiments : A series of exploratory experi
ments have been performed by dispersing the catalysts on diffe
rent supports.
a) Lewis acid containing systems :
1) The dispersion of RuCl (50 mgm) on AI2O3 (10 g) was treated
with T1CI4.
The CO hydrognation reaction was run at 180C under 10 bars
(C0:H 2 = 1:3).
The conversion is very low (0.01%) but the distribution of
hydrocarbons favours again the C3, C4 and C5(?) fractions.
Substantial amounts of olefins are present.
328
Table V
T1CI4 : 5 . l.O"4 mole T1CI4 : 2 ml

t = 24 h t = 72 h
Products % * % *
C- : propene 44 56
propane 2,5 9
C, : butenes 12
other C4 10
L
5 . ">.
13
unknown 41,5 and 12
Convers ion 0,01 0,01
% : relative peak area of each components calculated on the

sum of volatile components detected in the chromatogram
2) A dispersion of Ru^(CO)]2 on Al23 was treated with Ti

Cl4(5.10-4 moles) at 180 under 10 bars (CO : H2 = 3) .
Table VI
Products (%)
t = 24 h 144 h
C : propene 75 39
propane 10,5 30
C, : butenes 10,5 12,5

other C4 4 2,5
unknown 16
Conve rs ion 0, 1 0,3%
% see Table V.
These results are not very reproducible and corrosion

products of the autoclave seem again to be acting as catalysts.
Moreover , the absence of oxygenated products prompted us to
stop further investigations of these systems.
-329-
b) Complexes dispersed on oxides
The dispersion of transition metal complexes (eg deriva
tives of Rh, Ru, ...) on metal oxides (A l20 3 ,Zr02, TO2, S1O2
Si02~MnO) leads to different selecrtivities.
Some of these systems have been reported as displaying
high selectivities in oxygenated products (eg ethanol). We
are therefore studying the dispersion of the efficient cata
lysts on different supports.
However it appears that the use of static conditions
leads to low yield of oxygenated products. The use of a dyna
mic reactor is therefore a requirement.
The low pressure apparatus described in Fig.2 is there
fore used for such experiments. Of course the high pressure
device would be particularly useful.
1) Static experiments
a) Rh6(CO)i6 ( 0.05 g) was dispersed on La23 (10 g) according
to Ichikawa pro cedure ( 1) by evaporation of a THF solution
of the Rh c ompl ex on th e support, followed by heat treat
ment ( 180C in vaccuo) and reduction by hydrogen (200C).
The mi xtur e was heated at 220C during 90 h under an at
mo sphe re of CO : H 2 (1 : 3) (initial pressure at 25C=150 bars
end press ure = 80 bar s ) .
Produc ts th e gas ph as e contains a mixture of CO, H2,
CO2, araf finic and ole finie hydrocarbons
(CH4>C 3H8>C 4H10 > C 3 H 6 > C 2 H 6 > C 5 H 1 2 > C 6 H U ) .
h e liquid phas e (0,8 g) is an aqueous solution
of traces o f me thanol, ethanol, npropanol, nbutanol (and
possibly pe nt anol and hexanol ?)
b) RhCl 3 . 3H2O (0.62 g ) , MnCl24H20 (0.25g) were dispersed on
Zr02 ( 20 g ) .
One gr amme of t his catalyst was activated by reduction under
hydrog en at 200 C during one hour.
The CO H 2 (I :2) reaction was run at 200C during 60 h
( ini tial = 12 0 bars, end = 80 bars).
Produc ts : th e gas phas e is a mixture of CO, 2 and
satura ted hydro carbons CH4>C2H6>C3Hs>nC4H1o>i s OC4H10
no CO2 and olef ins were detected).
the liquid phase (0,2 g) is an aqueous solu
tion of traces of aldehydes (ace t aldehyde, propanal, iso
butanal) and alcohols (methanol, ethanol).
2) Dynamic experiments :
Experiments run in the low pressure circulator (Fig.2)
using Ichikawa catalyst afforded only hydrocarbons.
However the use of one of the catalysts disclosed in a
Hoechst patent (2) was efficient for CO hydrogenation
(CO = H2 = 1 : 2) even under atmospheric pressure, the pro
duct analysis being under investigation.
Catalyst : MgCl2 on S1O2 sintered at 950C was used as
support for RhCl3.
The e a t 200C in the
reactor and the COH2 (1:2) reaction was run during 24 hours,
The circulating device is now operating and will be exten
sively used for testing our catalysts(eg the cobalt based
330
sys terns) .
A comparison with experiments run with high pressure (100
bars or above) recycling system will be particularly in-
teres ting.
These catalysts afford reproducible results at least as far
as the distribution of oxygenated products is concerned and
they are therefore still investigated.
Analysis of results
The results obtained with the different classes of
studied catalysts show that very different products dis tribu-
tions are obtained according to the nature of the metal of the
catalyst and its oxidation state. The use of Lewis acid coca-
talysts has also a deep effect on the product formation. As
example, we observed in some cases the preferential formation
of C4 and C5 hydrocarbons without the simultaneous formation
of methane and ethane by us ing TiBr4 instead of the Muetterties
system (Ru3(CO)]2 - AICI3).
The use of CuBr-KBr-Ru system leads to the formation
of alcohols (Cj, C2 and C3) together with propane, propene
and alkyl bromides. In some cases, acetone and t-butanol or
carboxylic acids were even obtained but these results were not
reproducible.
Active species are in fact very probably formed by
corrosion of the stainless steel autoclaves as these results
could not be reproduced when the reactors were fitted with
glass protections.
The reduced metal systems show that the product distri-
bution depends strongly on the type of metal and the type of
added solvent, systems based on cobalt giving the more inte-
resting results at the present stage and are receiving much
attention together with modified supported rhodium catalysts.
Conclusions and future developments
In fact, our exploratory work has led us to the selec-
tion of one novel catalytic system which is susceptible to
lead to oxygenated products wi th good selectivity : in fact, large
amounts of such products relatively to hydrocarbons
were formed in some experiments with Co reduced species.
The problems of efficiency and improved selectivity in one
oxygenated derivative are still unsolved.
The future development of the present work is devoted
to the evaluation of reduced cobalt complexes under different
conditions (pressure, added ligands, solvents, . . . ) .
The role of, supports will be further considered mainly
in connection with Hoechst patent where Magnes ium containing
silicates appeared as particularly beneficial. In this appro-
ach, we shall try to decrease as much as possible (and ideally
to cancel completely) the use of the very expensive Rh salts
(Hoechst catalyst uses very large amount of this metal (llg/
AOg of catalyst)by replacing them by cobalt derivatives.
The use of rhodium associated with another metal (cobalt
at the present time) is therefore presently investigated under
different conditions in order to try to reach this objective.
The use of methanol synthesis catalyst (ZnO-CuO) in associa-
tion with a methanol homologation catalyst (eg Rh, Co) will be
also considered.
-331
References :
1. M. ICHIKA WA , Bul1.Chem.Soc.Japan, 5J_, 2273 (1978);
M. ICHIKA WA , . SHIKURA , . SEKIZA WA and . , Euro
pean Patent 22358, 03/07/80.
2. H.J. SCHMIDT, F. WUNDER, H.J. A RPE and E.I. LEUPOLD,

European Patent 4653, 31/03/79.
3. H.M. FEDER and J.W. RA THKE, Annals N.Y. A cad. Sciences , 4_5
(1980); C. MA STERS, Adv. Organomet. Chem., 17_, 61 (1979);
E.L. MUETTERTIES and J. STEIN, Chem. Reviews, 7_9, 479
(1979) .
4. R.B. A NDERSON, Catalysis, ed. P.H. EMMETT, Reinhold, New

York, Vol. 4, 29 (1956);
H.H. NIJS, P.A. JA COBS, J. Catal. bb_, 401 (1980).
5. R.C. BRA DY III, R. PETTIT, J. A mer. Chem. S o c , 102, 6181

(1980) .
6. B.D. DOMBEK, J. A mer. Chem. S o c , 102, 6855 (1980);

D.R. FA HEY, J. A mer.Chem.Soc., 103, 136 (1981);
J.F. KNIFTON, J.C.S. Chem. Comm. 188 (1981).
7. P.T. WOLCZA NSKI, J.E. BERCA W, A c e Chem. Res. J_3, 121

(1980) .
8. G.C. DEMITRA S and E.L. MUETTERTIES, J. A mer. Chem. S o c ,

29, 2796 (1977).
9. M. BLA NCHA RD, D. VA NHOUE, . M0RTREUX and F. PETIT, J. Chem.

S o c , Chem. Comm., 908 (1980).
10. H. BONNEMA NN, A ng. Chem. Int. Ed. Engl., 17, 505 (1978).
332
DISCUSSION
Question - Dr Jacobs, KUL : Concerning the CH4 formation on Manganese

carbonyl, do you have data to distinguish whether this is a
catalytic reaction or just a stoichiometric reaction : i.e.
just a consumption of the Ligand of the complex ?
Answer The amount of CH4 formed in this reaction is within the range
of a stoichiometric process and may include therefore a
consumption of the ligand.
Question - Dr Jacobs, KUL : For the selective synthesis of propane on

supported and promoted Ru dodecacarbonyl, have you an idea
about the carbon mass balance under these conditions : is
all the carbon entering the reactor as CO converted to
propylene or is an appreciable amount of it left on the
catalyst ?
Answer Due to the very small amount of catalysts used in the reaction
and the low conversion, a mass balance for carbon cannot be
evaluated.
Question - Dr Pilavachi, CEC : Why do you obtain a conversion as low as

0.01 in the Lewis supported case catalyst (table 1) ?
Answer The conversion is very low because the conditions are rather
smooth (low pressure). We believe also that products (eg
water) may inhibit the Lewis acid catalyst (hydrolysis of
TiCl 4 ).
-333-
FISCHER-TROFSCH SYNTHESIS OF HYDROCARBONS IN THE GASOIL RANGE
Authors P. STRUYF, P.A. JACOBS, J.B. UYTTERHOEVEN
Contract number EED-2-327-B
Duration 30 months January 1, 1981 - June 30, 1983
Head of project Prof. J.B. Uytterhoeven, K.U.Leuven
Contractor Katholieke Universiteit Leuven

Centrum voor Oppervlaktescheikunde en Collodale
Scheikunde
Address De Croylaan 42
B-3030 Leuven (Heverlee), Belgium
Summary
The well known Fischer-Tropsch synthesis has demonstrated its techni-

cal feasibility for the preparation of liquid fuels out of coal. The
major disadvantage of the presently operated process is its lack of
selectivity, which may be inherent to the chemical mechanism : a poly-
merization scheme according to Schulz-Flory kinetics. Therefore S.F.-
kinetics will have to be circumvented to improve selectivity.
The active metal of the catalyst is either ruthenium or cobalt. The
supports used are kieselguhr and microporous alumina. The catalysts
were calcinated in air and then reduced under hydrogen in the reactor.
Some catalysts definitely showed an anti-S.F.-behaviour. The highest
maximum that is reached with cobalt as well as with ruthenium is
situated at dodecane (carbon number 12).
It seems that the pore diameter of the support does not influence the
product distribution, while the degree of loading with the active
metal does. The tested industrial Fischer-Tropsch catalysts, with
kieselguhr as support, all showed a S.F.-behaviour.
-334-
1.1 Introduction : State of the art of selective synthesis of long chain
hydrocarbons
1.1.1 Constraints upon FischerTropsch (F.T.) chemistry
Typical synthesis reactions of hydrocarbons out of synthesis gas
(SYNGAS) are :
(2n+l) H + nCO W C H Q _,_, + nH.O (1)
2 2n+2 2
2nH, + nCO *- C H + H_0 (2)
2 2n 2
Two important side reactions are the Boudouard reaction, i.e. carbon
monoxide disproportionation, and the watergas shift (WGS) reaction :
2C0 C02 + C (3)
CO + H 2 0 * C02 + H2 (4)
The carbon generated in reaction (3) accumulates on the catalysts and con
sequently poisons its activity. The WGS reaction if it occurs as a side
reaction, will decrease the useful product yield on a carbon basis. If this
is aimed at, it will have to be suppressed during the F.T. hydrocarbon
synthesis reaction. Technologically however, there might exist reasons to
remove the oxygen contained in the reactants as carbon dioxide instead of
as steam : e.g. for reasons of catalyst stability (metal particles tend to
sinter very easily in a steamrich atmosphere).
The standard free energy changes per carbon atom for reaction (1) and
(2) can be calculated from available thermodynamic data in the temperature
range of interest (150450C). With regard to the selectivity of these
reactions, it follows that :
. paraffins are the preferred reaction products
. methane is the preferred alkane, since the stability of the paraffins
decrease with chain length
. the olefin selectivity is temperature dependent : light olefins are only
preferred at high reaction temperatures.
Therefore, the selective synthesis of high paraffins is thermodynamically
not possible and will rely totally upon the catalyst. In other words, this
type of product can only be formed on selective catalysts.
1.1.2 Kinetic limitations upon F.T. selectivity : the SchultzFlory (S.F.)
Most of the published F.T. product distributions can be fairly accura

tely described by the socalled S.F. polymerization scheme. In this scheme,
one carbon at the time is added to the growing carbon chain at the catalyst
surface. Mathematically, this is translated as follows :
W = ""1 (1) 2 (5)
in which W n is the product weight fraction of carbon number n, and the
chain growth probability. The latter quantity can be derived from a plot of
log(Wn/n) against n.
If this mechanism is of generally applicability, it will have severe
limitations upon the product selectivity :
. the chain growth probability is the only physical parameter which deter
mines selectivity. If the concentration of one component is determined,
all the others are fixed.
. there exists theoretical maxima for the selectivity of all products or
product groups. Using an optimal degree of polymerization, the maximum
selectivity can be calculated. This is done in Table I for gasoline and
diesel fuel. These data agree fairly well with the industrial data from
Sasol.
The main conclusion is that in order to obtain higher yields of gaso
line and/or diesel than those mentioned in Table I, catalysts will have to
be used which circumvent S.F.kinetics.
TABLE I. Maximum selectivity for F.T. products wj (%) and optimal growth
probability (cx::) according to S.F. kinetics and experimental data
(1).
Product Gasoline Diesel
C C C
6 12 13 Clf
w e x pr . 40.0 18.0

1
W calc. 43.5 20.9

0.80 0.87
1.1.3 NonSchulzFlory type FischerTropsch synthesis

1.1.3.1 Upgrading of fuel yields using proven technology
Product selectivities obtained on a large scale at Sasol using extrud
ed precipitated iron catalysts are given in Table II.
Their product composition and properties as fuel are shown in Table
III. It seems that the unrefined Synthol product has an attractive research
octane number. The latter can be increased by simple clay treating. For the
diesel fraction, the straightchain fixedbed materials has an attractive
cetane number.
Nevertheless, the F.T. synthesis as commercially applied now is very
unselective. Variation of the reactor conditions (as temperature, partial
pressure and contact time) only changes this to a minor extent. However,
the overall diesel and/or gasoline yields can be altered markedly by
further treatment using proven technology : e.g. oligomerization of light
olefins and wax hydrocracking (scheme I ) .
SCHEME I. Commercial upgrading of F.T. products.
Synthol fixedbed

F.T. Products

F.T. Products
Oligomerization of C_C
olefins + wax hydro
cracking
75 % 50 % 70 % diesel
diesel gasoline (cetane no. 70)
These fuels are very attractive, since nitrogen is absent in the products
and therefore they are less noxious engine fuels.
336
TABLE II. Product selectivities on precipitated iron catalysts (2).
Selectivity Commercial Commercial Pilot
fixed-bed fluidised recirculation plant
reactor (Synthol)
methane 2.0 10.0 14.

ethane 1.8 4.0 4.
ethylene 0.1 4.0 9.
propane 1.7 2.0 3.
propylene 2.7 12.0 13.
butnes 3.1 9.0 11 .
butane 1.9 2.0 2.
83.5 51.0 44.
others 3.2 6.0 0.
TABLE III. Product compositions on iron catalysts (2)

Composition gasoline (C5-C1]) diesel (C12-C18)
fixed-bed Synthol fixed-bed Synthol
olefins 32 57 25 73
paraffins (iso + n) 60 14 65 10
n-paraffins 57 8 61 6
cyclics 0 15 0 10
alcohols 7 6 6 4
ketones 0.6 6 1 2
acids 0.4 2 - 1
octane number
(lead free) 35 88
cetane number 65-70 50-55
1.1.3.2 Non-S.F.-products by direct synthesis

Recently, several reports appeared in literature which showed F.T.
product distributions deviating from the classic S.F.-behaviour. In the
C ] Q - C 2 O carbon number region mainly two groups have been showing this.
Table IV shows all the data pertinent to catalyst preparation, reac-
tor operation and catalytic behaviour. The experimental data taken from
the original work are given in Figures 1-3. Maxima in the product distri-
butions in the Cio-C7 carbon number fractions, are indicative of diesel
selectivity which with respect to S.F. behaviour is enhanced considerably.
These literature data seem to be very promising for those interested to
make high diesel yields out of CO/H2 mixtures in a single stage catalytic
process. It seem that catalytic systems may exist which are able to cir-
cumvent S.F.-kinetics during a % F.T. hydrocarbon synthesis reaction.
However, after careful examination of the results and operational
conditions, it remains unclear whether :
. it is possible to reproduce these catalyst formulations and obtain the
same selectivities.
. the material balance, mainly with respect to carbon, at the catalyst bed
exit corresponds to 100 %. In other words, it remains to be proven
whether no carbon is left in the reactor, or whether the selectivities
given in the publications are absolute.
-337-
TABLE IV. Literature data on anti-S.F. synthesis of hydrocarbons in the diesel range.
II III
Catalyst 2 % Co/alumina Co/kieselguhr 1 % Ru/alumina
Catalyst precursors "alumina SCS59 100 Co : 16 Th0 Y-A1 2 0 3

2 -1
90 m g ; porous 300 A 93 kieselguhr
"octacarbonyldicobalt 2 K2C03
preparation method impregnation in pentane precipitation
activation in H /200C in H2/400C

1 atm/4 hr
50 g Co.h/mol CO 210 v/v/hr
reaction temperature ('C) 200 197 241
H 2 /C0 2:1 1.9:1 2:1
type of reactor continuous flow continuous flow
CO conv. (%) 16 96
total pressure (bar) 1 11 3.1
anti-S.F. maximum yield at C., first max. at C . - C . ,

sec. max. at C.-C.
carbon balance not given not given not given
reference (3) (4) (5)
nature of products paraffinie paraffins + olefins -

0..
1 ..
-2 1111111111111111111
s m S LO isa
C\J
.arbori nur
Product distribution on a Fig. l.b. Schulz-Flory plot for a

2 % Co/alumina SCS59 2 % Co/alumina SCS59
catalyst (I). catalyst (I).
EXPERIMENTAL
\ SCHULZ -FLORY
/ /
I ^
--__
y
1
CARBON NUMBER
Fig. 2.a. Product distribution on a 1 % Fig. 2.b. Schulz-Flory plot

Ru/Al 2 0 3 catalyst at 3.1 bar for a 1 % Ru/Al 2 0 3
and 514 (III). catalyst (III).
-339-
CARBON NUMBER
Fig. 3. SchulzFlory plot for a sulfided alkalipromoted

precipitated cobaltbased catalyst (II).
r# _ iti J*L
-nnmmxun-i
-W
TER J
Fig. 4. Experimental setup.

F = flow regulator SV = 6port sampling valve
= pressure regulator IP = injection port
R catalytic reactor TC = thermostatted compartment
V = on/off valve FID = FID detector
S = splitter system TER = reporting terminal
CT = condensation trap M = makeup gas
C = capillary column
340
This research paper is an attempt to confirm the literature data, since it
aims to verify whether antiS.F. product distributions are possible or not.
If not, the diesel yield by direct synthesis will be limited and only
secondary catalytic treatment of primary formed F.T.products, using proven
technology (e.g. oligomerization of light olefins and wax hydrocracking)
will remain the only way to obtain' high diesel yield out of H/CO mixtures.
1.2 Description of apparatus and measuring equipment

1.2.1 Reactor_and_analyîcal_eguDment
The reactor is a thermostatted autoclave (volume 2 1) with one inlet
for the syngas C0/H2, one for hydrogen or helium, and one outlet for the
reaction products. After the outlet a condensation trap is placed, which
is connected to a sixport sampling valve in the gaschromatograph (HP
5880A).
The injection in the G.C. can be done online with a sixport sampling
valve or offline by manual injection. In both cases, the split injection
mode is used. In this mode the injected sample is vaporized, thoroughly
mixed to achieve homogeneity, and divided between the column inlet and a
vent line. The split ratio is 20:1 and the temperature of the injection
port is 350C. The capillary column is of stainless steel and 17 meters
long. The coated stationary phase, 0V101 (Liquid Methyl Silicone) has a
low bleeding rate in a wide temperature range. The other parameters are :
. oven temperature : initial value 35C
program rate 3C/min
final value 325C
. total time of the analysis : 120 min
. temperature of the flame ionization detector : 350C
. injection port temperature : 350C
. attenuation : 2.10
1.2.2 Reacion_p_rocedure
A glass tube containing the catalyst is placed in the autoclave,
which is then leakfree closed. The air is evacuated and replaced by hydro
gen for the reduction of the catalyst. After this pretreatment the reac
tion is started with the inlet of syngas. After 2 to 3 days the reaction
is stopped and all the gaseous hydrocarbons are condensed in a C.T., which
is cooled in liquid air. Flushing with helium ensures that most of the
hydrocarbons leave the reactor. Helium, hydrogen and carbon monoxide are
not condensed in liquid air and are vented. The next step is the evapora
tion of the gases in the C.T. at roomtemperature. When the pressure in C.T.
does not increase any longer, an online sample is taken and analysed in
the G.C. After evacuation of the gases, the C.T. is opened and the liquid
phase is manually injected in the G.C.
The used catalyst in the autoclave is extracted with benzene. This
extract is also manually injected in the G.C. The results of the analysis
are given in a SchulzFlory plot. The whole experimental setup is shown
in figure. 4.
1.3 Results
According to the literature, the active metal in catalysts exhibiting
antiS.F.behaviour is either ruthenium or cobalt. The supports used are
kieselguhr and microporous alumina (Spheralite). These aluminas are deli
vered by RhnePoulenc (Chimie Fine), and have the following properties :
341
TABLE V. Specifications of the aluminas used as support
Specification Presentation Surfac
Surface Average pore
area diameter (A)
(m2/g)
alumina I SAP 350 powder 10-120 um 350 25

alumina II SCS 100 balls 2-4 mm 100 200
alumina III SCS 9 balls 2-4 mm 8 3000
The used metal salts are Ru(NH ),C1 for Ru-catalysts and Co(N0)2.6H 0
for Co-catalysts.
The catalysts are prepared by :
1. Impregnation : the metal salt is dissolved in a minimal quantity of
water (equal or less than the amount of carrier); the carrier is added
to this solution with vigorous stirring; the catalyst is dried.
2. Precipitation : dissolved Na2C3 in boiling water and the carrier (in
powder) are simultaneously added with vigorous stirring to a boiling
solution of the metal salt : the product filtered; washed thoroughly
until free of sodium ions and dried at 110C.
3. Adsorption of Co-carbonyls : dicobaltoctacarbonyl dissolved in pentane
is added to the carrier under an inert atmosphere of nitrogen; the
carbonyl and pentane are evaporated and Co is left behind on the carrier.
This method does not need a pretreatment (reduction).
The catalysts are further calcinated on 400C, crushed and screened to a
suitable size. In the reactor, the reduction of the catalyst is effected
in hydrogen. The reduction temperature is raised from room temperature
to 350C during 1 hr and the temperature held at 350C for 2 hrs.
Temperature and pressure during the reaction are adapted to the activity
of the catalyst and therefore to the nature of the metal; while the used
C0/H 2 ratio is always 2/3.
The screened catalysts are shown in table VI.
1.4 Analysis of results and comments

A typical product distribution (8 % Ru/alumina I) and the experimen-
tal Schulz-Flory plot is given in figures 5 and 6. From the chromatographic
results in table VII, it can be concluded that some catalysts show an
anti-S.F. behaviour. The obtained results however must be carefully inter-
preted because the occurrence of some artefacts in the low carbon number
region as well as in the high carbon number region is possible.
The concentration found for the low carbon numbers up to octane is too
low because part of these hydrocarbons leave the condensation trap in the
vapor phase; this can be derived from the on-line chromatogram. Also the
high carbon numbers give too low amounts because part of the heavy hydro-
carbons does not leave the reactor. The chromatogram of the catalyst
extract shows which hydrocarbons partly stay in the reactor, but the
absolute amount remains unknown. If these amounts would be added to the
liquid phase, possibly the chromatogram will show a higher growth probabi-
lity for some catalysts. However maxima at dodecane (carbon number 12) are
real since decane and undecane do not leave the cold trap in the vapor
phase and on the other hand the chromatogram of the extract shows only
hydrocarbons with carbon numbers higher than eicosane (carbon number 20).
So, it can be concluded that there exist catalysts which circumvent
the Schulz-Flory kinetics, but the reaction conditions are different from
the ones reported in literature. Other indications for diesel selectivity
are the nature of products (more than 80 % is n-paraffinie) and the
-342-
TABLE VI. Reaction parameters and products of the tested catalysts.
Catalyst (a) Preparation Reaction Reaction Pressure drop Pressure of Weight of
method (c) temperature/ pressure/ during reaction gases in li.quids
C bar after 2 days/bar C.T./biar in C.T./g
12 % Co/alumina I 190 10 4 6 2.57

12 % Co/alumina III 190 20 15 16 2.12
45 % Co/alumina I 190 15 4 10 2.30
2 % Co/alumina III 190 15 low 7 0.51
16 % Co/kieselguhr 190 15 low 8 0.43
(0.91 % Th(N0 3 ) A - 0.57 % MgO -
13.22 % Na 2 C0 3 )
16 % Co/kieselguhr 190 15 15 13 4.65
(5.37 % Th(N0 3 ) 4 - 0.32 % K 2 CO 3 )
11 % Co/kieselguhr 190 15 9 10 3.16
(1.40 % Th(N0 3 ) 4 - 1.33 % MgO -
20.43 % Na 2 C0 3 )
8 % Ru/alumina I 190 15 low 9 0.6
8 % Ru/alumina II 190 15 low 8.8 0.2
8 % Ru/alumina III (b) 190 15 15 1 .8 0.3
8 % Ru/alumina III 190 15 13 7 3.10
8 % Ru/alumina III 241 15 13 15 1.98
2 % Ru/alumina I 190 15 2 5 0.22
2 % Ru/alumina III 241 15 low 5.5 0.70
Fe 3 0 A (5 % K 2 0) 190 15 6 0.5
(a) for alumina specifications see table V.

(b) this catalyst was not precalcinated
(c) = by precipitation; I = by impregnation; C with carbonyl.
14
+> EXPERIMENTAL
c 0 ..

o
L

Q_
10 U
10 \

-2
o

3
"
SCHULZFLORY
-3 ..
L
CL
-5 I III I I I I I I I I I I I | | | | nu-t
Carbon number o m c s m s i msi
c\j
i sin
c\i i i r )
m
Carbon number
Fig. 5. Experimental product distribution Fig. 6. Experimental SchulzFlory distribution
(8 % Ru/alumina I) together with the for a 8 % Ru/alumina I catalyst.
SchulzFlory distribution having a
maximum at C..
TABLE VII. Chromatographic results of the tested catalysts
Catalyst (a) Extract Maximum in Nature of Growth % c, 2 -c 18
the S.F.-plot products probability in liquid
of liquid phase of S.F.-plot
phase
12 % Co/alumina I 8 n-paraffinie 0.71 30.75

12 % Co/alumina III 9 n-paraffinie 0.63 23.64
45 % Co/alumina I 8 n-paraffinie 0.68 32.02
2 % Co/alumina III 12 n-paraffinie 0.62 55.80 anti-S.F.
16 % Co/kieselguhr / / / /
(0.91 % Th(N0 3 ) 4 - 0.57 % MgO -
13.22 % Na 2 C0 3 )
16 % Co/kieselguhr 8-9 n-paraffinie 0.67 29.90
(5.37 % Th(N0 3 ) 4 - 0.32 % CO )
11 % Co/kieselguhr n-paraffinie 0.69 28.90
(1.40 % Th(N0 3 ) 4 - 1.33 % MgO -
20.43 % Na 2 C0 3 )
8 % Ru/alumina I 12 n-paraffinie 0.72 57,.21 anti-S .F
8 % Ru/alumina II 12 n-paraffinie 0.71 69,.40 anti-S .F
8 % Ru/alumina III (b) 12 n-paraffinie 0.69 71..04 anti-S .F
8 % Ru/alumina III 9 n-paraffinie 0.70 41,.11
8 % Ru/alumina III 12 n-paraffinie 0.71 62,.70 anti-S .F
2 % Ru/alumina I 11 n-paraffinie 0.68 45,.50
2 % Ru/alumina III 9-10 n-paraffinie 0.67 42,.70
F e 3 0 4 (5 % K 2 0) 7-8 n-paraffinie 0.65 33.30
(a) for alumina specifications see table V.

(b) this catalyst was not precalcinated.
percentage of C^-Cjg in the liquid phase. Table VIII shows that the diesel
selectivity is doubled in comparison with the theoretical Schulz-Flory dis-
tribution with C.--C 10 maximised.
13 lo
TABLE VIII. Theoretical yields according to the S.F.-distribution with
C j 3C j o maximised and the experimental product distribution on
8 % Ru/alumina I.
Carbon number 5-12 13-18 19-21 22-30
theoretical yield (weight %) 36.89 20.98 7.52 13.69

experimental yield (weight %) 45.24 44.30 6.19 3.43
Also the percentage of C]2"C]g in table VII shows that the anti-S.F.
catalysts have a higher diesel selectivity than any other screened catalyst
which follows S.F.-kinetics.
Another conclusion is that the present results do not allow to make a
correlation between the pore size of the carrier and the obtained product
distribution. This in contrast with the loading of the active metal which
definitely has an influence on the product distribution : for the Co on
alumina catalysts the formation of heavier products increases with decreas-
ing metal loading, while for Ru on the same support the opposite is observ-
ed. The tested industrial Fischer-Tropsch catalysts, with kieselguhr as
support, all showed a S.F. behaviour.
1 .5 Conclusions
The well known Fischer-Tropsch synthesis has demonstrated its techni-
cal feasibility for the preparation of liquid fuels out of coal. The major
disadvantage of the presently operated process is its lack of selectivity,
which may be inherent to the chemical mechanism : a polymerization scheme
according to Schulz-Flory kinetics. To be applicable to European economic
conditions, catalysts with improved selectivity are a prerequisite and
therefore S.F.-kinetics will have to be circumvented, literature data show
that this is possible.
The aim of the work till now has been to verify whether these experi-
ments can be reproduced. With Ru and Co metal on mircoporous alumina this
has been possible. Definitely a maximum in the carbon number distribution
of the F.T. products is located around dodecane. However, the pore size
of the support, when varied over a wide range (25-3000 A) does not change
the product distribution, while the amount of metal on a given support
influences the product distribution : higher carbon numbers are formed
with decreasing Co loading, while for Ru the opposite is observed.
Therefore for future work, two standard catalysts (low Co loading on
alumina and high Ru loading on alumina) will be used and accurate carbon
mass balances will be determined in an on-line analysis up to C20 t 0
determine the absolute diesel yields of these catalysts. Afterwards
catalyst modification (use of promoting elements) and further physical
characterization will be done in order to understand the catalyst behaviour
and in order to be able to improve possibly these yields.
1.6 References
1. M.E. Dry, Ind. Eng. Chem. Prod. Res. Dev., _1_5, 282 (1976).
2. M.E. Dry, J.C. Hoogendoorn, Cat. Rev. Sei. Eng., 23, 265 (1981).
3. D. Vanhove, P. Mucambo, M. Blanchard, J.C.S. Chem. Comm., 605 (1979).
4. R.J. Madon, W.F. Taylor, Adv. Chem. Ser., J_7J.> 9 3 (1979).
5. R.J. Madon, J. Catalysis, 57, 183-186 (1979).
DISCUSSION
Question Mr. Jacquin, IFP : How are the cataLysts prepared ? How is
reduction effected before or during the reaction ?
Answer Generally, the catalysts are prepared by impregnation of the

metal salt (Ru(NH 3 ) 6 CL 3 for Rucatalysts and 0(0 3 ) 2 6 2 0
for Cocatalysts) on microporous aluminas. These aluminas
have narrow pore size distributions and average pore diameters
ranging from 25 8 to 3000 A. The catalysts are further
calcinated at 400C, crushed and screened to a suitable size.
The hydrogen reduction of the catalyst is effected in situ
in the reactor. The reduction temperature is raised from
room temperature to 350C in 1 hr and the temperature held
at 350C for 2 hrs. A fter this pretreatment the hydrogen
is evacuated from the reactor and the reaction is started.
Question Mr. Jacquin, IFP : Do you think there is a correlation

between the state of dispersion of the metal on the support
and the obtained selectivity ?
Answer Up to now, the state of dispersion of the metal on the carrier

has not yet been investigated. But for an equal loading of
the metal on aluminas with different pore sizes, always the
same product distribution was obtained.
Question Prof. Ketelaar, Expert : Did you find a different distribution

as a function of the reaction time ? Especially, did you
find a S.F. distribution after a short contact time with a
catalyst which finally gives an antiS.F. distribution ?
Answer A ll the experiments up to now were done in a batch reactor.

As the FischerTropsch synthesis is a slow reaction, it takes
a long time before steady state is reached. Therefore the
reaction is stopped only after 2 to 3 days. That is why there
are no experimental results available for short contact times.
But it is general knowledge that the product distribution in
transient conditions differs from the product distribution at
steady state.
Question Dr. Pilavachi, CEC : What is the reproducibility of the

experiments which gave an antiS.F. product distribution
in your laboratory ?
Answer This reproducibility is very good. A ll the experiments were

repeated several times and always identical product distri
butions were obtained.
347
Question - Dr. Pilavachi, CEC : When will you start experimentation in
the steady state equipment ?
Answer Probably at the beginning of June 1982 the first experiments

will be made in the new on-line reactor. In this type of
reactor samples containing hydrocarbons up to Cp-i can be
taken during the reaction. This makes it possible to follow
a change in the product distribution as a function of the
reaction time.
348-
VALORISATION DES ASPHALTES
PAR HYDROGENATION A TRES HAUTE SEVERITE
SUR CATALYSEUR ENTRAINE
Auteurs P. CHICHE, J. GAGET, P. GIULIANI et Y. JACQUIN
Contrat n EEB-1-122-F
Dure 2 ans - 1er janvier 1981 - 31 dcembre 1982
Responsable scientifique P. CHICHE, CERCHAR
Contractants : CERCHAR (Centre d'Etudes et Recherches des Charbonnages

de France) 33 rue de la Baume, 75008 PARIS
et B.P. 2, 60550 VERNEUIL-en-HALATTE
et IFP (Institut franais du Ptrole)

1 et 4 Avenue du Bois-Prau, B.P. 311,
92506 RUEIL-MALMAISON CEDEX
et B.P. 3, 69390 VERNAISON
(Centre d'tudes et de dveloppement industriel

de Solaize CEDI).
Rsum
L'volution du march des produits ptroliers se traduit en par-

ticulier par le problme de la valorisation des produits lourds, la
fois par le poids croissant de ces produits dans le bilan global de
l'industrie du raffinage et par la monte de la demande de distillate
moyens et lgers. La prsente recherche se propose d'tudier le trai-
tement des asphaltes rsultant du procd de dsasphaltage de 1'IFP
par hydrognation dans des conditions trs svres et en prsence de
catalyseurs entrans, dans une installation continue applique au
CERCHAR 1'hydroliqufaction des charbons, de faon obtenir des
produits plus lgers ou moins visqueux susceptibles d'tre repris par
les techniques classiques du raffinage.
Des travaux prliminaires ont permis de dterminer la nature et
le taux de solvant fluxant ncessaire pour rduire la viscosit une
valeur compatible avec le fonctionnement de l'installation, d'adapter
cette installation des charges plus visqueuses et plus riches en
soufre que celles qui y sont habituellement traites, et de slection-
ner l'chelle du laboratoire des systmes catalytiques solubles ou
en suspension paraissant de nature conduire des rsultats intres-
sants du point de vue de la dsuifuration, de la dmtallation et de
la conversion en fractions lgres.
Un premier essai rel d'hydrognation sous 200 bars d'hydrogne,
temprature variable sur un intervalle de 50C et en prsence d'un
compos organique soluble du molybdne, a fait apparatre, au-dessus
de 400C, un allgement trs sensible de la charge par conversion
ine grande partie de la fraction la plus lourde en produits lgers
moyens.
-349-
1. Introduction
Jusqu' ces dernires annes, les raffineries europennes parvenaient
satisfaire les besoins du march des produits ptroliers en fractionnant
le ptrole brut par distillation ventuellement complte d'un traitement
catalytique, aboutissant ainsi des fractions rpondant des spcifica
tions prcises des utilisateurs. L'adaptation de la production la demande
rsultait donc d'une part d'un approvisionnement judicieux en bruts, d'au
tre part, dans certains cas, d'un craquage catalytique transformant en
essences des excdents ventuels en distillate lourds. Les fractions non
distillables taient utilises comme combustibles industriels, essentiel
lement en France pour la production d'lectricit par EDF.
Cette structure de raffinage est schmatise figure 1.
Or l'volution du prix du ptrole et du contexte conomique europen
a entran d'importantes rpercussions sur l'industrie du raffinage : la
demande en produits ptroliers n'augmente plus aussi rapidement ou mme
diminue, l'adaptation par les enchanements de procds classiques aux
besoins d'un march profondment modifi devient de plus en plus difficile,
ainsi que la commercialisation des fractions lourdes dont la transformation
en fractions lgres rpondant la demande actuelle requerra la mise en
place de procds nouveaux.
Cette modification des schmas de raffinage est impose par la pers
pective d'un plafonnement ou d'une diminution de la consommation de pro
duits ptroliers et, plus encore, par le recours envisag des proportions
croissantes de ptroles lourds.
La transformation des fractions lourdes est rendue difficile par la
prsence de deux familles de constituants : mtaux nickel et vanadium
essentiellement et polymres de haute masse molculaire asphaltnes ,
qui provoquent une dtrioration rapide des catalyseurs de craquage et
dont il faut prvoir la sparation dans le cadre d'un schma de traitement
de ces fractions. Diffrentes solutions techniques sont envisages cet
effet, dont les deux suivantes sont proches de la ralisation industriens :
. craquage thermique direct de la fraction lourde, transforme en
hydrocarbures distillables et en coke que l'on peut brler ou gazifier :
c'est le procd qui s'organise autour d'une installation de cokfaction.
. prcipitation, par une fraction d'hydrocarbures lgers correspon
dant la coupe C (pentane), d'un asphalte qui contient la majorit des
mtaux et la totalit des asphaltnes : c'est la voie choisie par l'IFP,
qui conduit une fraction d'hydrocarbures lourds purifis susceptibles
d'tre soumis au craquage catalytique.
Le dveloppement de ce schma de raffinage, reprsent figure 2,
dpend des possibilits de valorisation de l'asphalte, dont trois types
taient jusqu'ici envisags :
. combustion directe analogue celle du charbon pulvris
. oxydation partielle l'oxygne conduisant la production de
l'hydrogne ncessaire au raffinage
. dilution par une partie du distillt moyen avec obtention d'un fuel
forte teneur en mtaux et en polluants comme le soufre et l'azote.
La recherche actuelle propose une solution nouvelle consistant en un
hydrotraitement svre temprature suprieure itOCC, pression sup
rieure 150 bars en prsence de systmes catalytiques entrans spcia
lement conus pour rsister la dsactivation par les mtaux et les
asphaltnes. On en attend la destruction des complexes organomtalliques
assurant la solubilit des mtaux, donc leur prcipitation sous forme de
sulfures, la conversion des asphaltnes en fractions moyennes et lgres
par dpolymrisation, hydrognation et craquage, et 1'hydrognolyse des
liaisons CS et CN avec formation d'hydrogne sulfur et d'ammoniac.
350
L'intrt d'un tel traitement et les possibilits de valorisation des
asphaltes qui en rsultent dpendront de l'importance de ces trois familles
de ractions dans les conditions adoptes.
La conjonction de l'exprience de l'IFP dans le domaine de la produc-
tion des asphaltes et des catalyseurs d'hydrotraitement de produits lourds
et de l'existence au CERCHAR d'une installation pilote continue d'hydroli-
qufaction du charbon a ainsi suscit la prsente recherche dont le drou-
lement comporte, outre la production d'asphalte, les trois phases princi-
pales suivantes :
. Choix d'un solvant appropri des asphaltes et prparation des
mlanges asphalte-solvant.
. Essais de slection de catalyseurs entrans, solubles ou en
suspension.
. Essais d'hydrognation en installation pilote continue et valua-
tion des produits .
2. Description des installations d'essais

2.1. Installation de dsasphaltage
L'IFP dispose, dans son Centre d'tude et de dveloppement industriel
de Solaize (CEDI), d'une installation pilote de dsasphaltage en continu
capable de produire les quantits d'asphalte ncessaires pour la ralisa-
tion des essais envisags. Le schma de principe en est reprsent sur la
figure 3 Le mlange asphalte-solvant est soutir la partie infrieure
de l'extracteur, le flux liquide tant rchauff de faon permettre de
rcuprer le solvant de dsasphaltage dans le strippeur d'asphalte.
L'asphalte liquide est soutir 200-250C.
2.2. Installation d'hydrognation

Le CERCHAR s'est quip, pour raliser 1'hydroliqufaction du charbon
dans le cadre d'une tude mene au sein du Groupe d'tude de la conversion
du charbon par hydrognation (GECH, constitu de Gaz de France, Charbonna-
ges de France, IFP et Centre national de la recherche scientifique), d'une
installation pilote continue d'un dbit de quelques kg/h et conue pour
pouvoir oprer jusqu' 500C sous une pression maximale de 400 bars d'hy-
drogne. Le schma en est indiqu sur la figure 4. L'installation se
compose successivement de la section de prparation et de mise sous
pression de la charge traiter, de la section des racteurs constitue
d'un prchauffeur tubulaire et de deux racteurs cylindriques identiques
de 10 1 de capacit, d'un sparateur chaud quip d'un systme de dtente
par vanne des produits condenss et d'un sparateur froid recueillant les
liquides lgers, d'une section d'puration des gaz et d'une section de
distillation des liquides soutirs la base du sparateur chaud, compose
d'une colonne atmosphrique et d'un flash sous vide et utilise, dans le
cas du traitement du charbon, pour sparer le solvant de recyclage. En
plus des mesures de dbit, de temprature et de pression diffrents
niveaux, un analyseur d'hydrocarbures totaux et un spectromtre de masse
permettent de suivre en continu la composition des gaz produits.
Un certain nombre de discussions entre les spcialistes du CERCHAR et
de l'IFP ont fait apparatre que le traitement des charges ptrolires ne
pourrait tre men bien dans des conditions satisfaisantes de fonction-
nement et de scurit qu'au prix de modifications justifies par les
caractristiques suivantes :
. viscosit leve des charges
. forte teneur en soufre ncessitant le lavage des effluents gazeux
et certaines prcautions lors du soutirage des liquides
- 351
. risques de formation de sulfure et de polysulfure d'ammonium
point de fusion lev dans la section des sparateurs.
C'est ainsi en particulier que, pour viter toute zone froide avant
l'entre dans la section des racteurs, on a install un nouveau bac
jaugeur simplifi plus compact et plus proche du premier racteur et
renforc le chauffage du circuit de malaxage. On a galement acclr la
mise en place de la dtente directe des produits condenss la base du
sparateur chaud par vanne asservie au niveau dans ce sparateur, mesur
par absorption d'un rayonnement Y , et intensifi le chauffage de la
jonction entre les sparateurs chaud et froid pour viter les risques de
bouchage par dpt de composs sulfurs d'ammonium.
Enfin la forte teneur en soufre des charges traiter, pouvant donner
lieu un dgagement d'hydrogne sulfur de 100 g/h, a impos l'installa-
tion sur le circuit des gaz de purge d'un systme de lavage l'hydroxyde
de sodium avec garnissage d'anneaux Raschig ainsi que, pour des raisons de
scurit, celle d'un appareil de surveillance de la concentration en
hydrogne sulfur et la mise disposition du personnel de masques
cartouche filtrante.
2.3 Appareillages pour essais de slection des systmes catalytiques

Trois types d'essais ont t raliss cet effet :
. essais en autoclave de 500 cm3 sur 30 g de charge avec pression
initiale de 100 bars
. essais en systme ouvert en racteur tubulaire de 200 cm (diamtre
16 mm, section libre 1,7 cm3 ) avec catalyseur en lit fixe en quantit de
80 cm5 (figure 5)
. essais en systme ouvert en racteur tubulaire identique, mais
vide.
3. Droulement de la recherche
3.1. Choix d'un solvant appropri des asphaltes et prparation du
mlange
L'asphalte obtenu la sortie de l'installation de dsasphaltage est
un solide de point de ramollissement voisin de 150C. Aussi doit-il tre
"flux" par un solvant capable de rduire sa viscosit pour pouvoir tre
mis en oeuvre dans l'installation d'hydrognation. Ce solvant doit tre
peu onreux, et fortement aromatique pour possder un bon pouvoir solvant
des fractions lourdes. Le choix s'est port sur un produit disponible en
grandes quantits et de faible valeur marchande : le gas-oil de craquage
catalytique, ou "light cycle oil"(LC0), sous-produit des units de cra-
quage catalytique, dont un exemple de caractristiques est indiqu dans le
tableau I.
Deux solutions pouvaient tre envisages pour prparer les produits
exprimenter :
. on pouvait imaginer de reprendre un asphalte et de le rchauffer
pour raliser la dilution par le solvant fluxant
. on pouvait profiter de la prsence d'asphalte liquide la sortie
du strippeur pour raliser en ligne la dilution, comme schmatis par la
ligne en pointill de la figure 3> bien que la production d'asphalte
flux dpende alors de la mise en service de l'unit de dsasphaltage.
La faible proportion de solvant utilis et la difficult pratique de
liqufier de grandes quantits d'asphalte ont nanmoins milit en faveur
de cette seconde solution pour la prparation des charges destine? au
traitement d'hydrognation.
-352-
Une premire charge a t prpare, constitue de 65 % d'asphalte
Safanya et 35 % de LCO en poids, dont les caractristiques sont rassem-
bles dans le tableau II. Mais un tel produit, encore solide temprature
ambiante, s'est rvl de viscosit trop leve :
860 cP 100C
465 cP nOC
203 cP 130C
100 cP 148C (valeur extrapole)
L'coulement par gravit devient normFl vers 140-150C, ce qui cor-

respond, avec les chauffe-fts classiques dont nous disposons, une
temprature de peau suprieure ou gale 200C, laissant craindre des
zones de surchauffe susceptibles de provoquer une modification du produit
de base.
Des mesures sur une seconde charge, fluxe galement 35 % de
produits lgers, ont confirm ces rsultats, conduisant par extrapolation
une viscosit de 100 cP 153C.
Aussi a-t-il t dcid de prparer une nouvelle charge "surfluxe"
de viscosit rduite de faon obtenir un coulement normal vers 100C.
Le calcul conduit cet gard aux estimations suivantes :
Temprature
Proportion de fluxant o la viscosit est de :
dans la charge
200 cP 100 cP
35 % 129C 148C
45 % 98C 115C
50 % 84C 100C
La charge n 2 a donc t prpare par mlange poids gal d'asphalte

Safanya et de LCO. Les caractristiques en sont rassembles dans le
tableau III.
Il a t dcid en outre, avant tout essai rel d'hydrognation, de
procder un essai "hydraulique" destin vrifier la circulation satis-
faisante dans toutes les parties de l'installation d'un produit de visco-
sit comparable celle de la charge n 2 : une charge a t constitue
cet effet partir de rsidus de distillation atmosphrique de brut Ratawi,
prsentant les caractristiques suivantes :
237 cP 71C
100 cP 86C (valeur interpole)
83 cP 95C
34 cP 111C
3.2. Essais de catalyseurs entrans, solubles ou en suspension

L'IFP poursuit depuis de nombreuses annes des programmes de recher-
che et de dveloppement dans le domaine des catalyseurs et des procds de
traitement des fractions lourdes du ptrole. Deux voies sont explores
simultanment : 1'hydrotraitement sur des catalyseurs htrognes adapts
aux traitements en lit fixe, mobile ou bouillonnant, et l'utilisation de
-353-
techniques nouvelles corame celles qui ont recours des catalyseurs solu
bles ou en suspension. Pour l'exprimentation sur les asphaltes, on a
slectionn dans chacun des programmes les systmes catalytiques qui ont
paru les plus prometteurs, et on les a soumis des essais d'estimation
l'chelle du laboratoire.
3.2.1. Systmes catalytiques drivs des tudes de catalyseurs

htrognes
Les essais de ce type ont t raliss avec des catalyseurs en
billes d'alumine de faible diamtre imprgns de sels de cobalt, de nickel
et de molybdne et mis en oeuvre en lit fixe dans un systme ouvert. La
charge passe en continu, dans les conditions choisies, sur le catalyseur
dont l'efficacit est juge d'aprs l'analyse des produits la sortie du
racteur.
La difficult de manipulation de l'asphalte, mme flux, a amen
oprer sur un ptrole moins visqueux, mais contenant les impurets
caractristiques des asphaltes. Les conditions opratoires et les rsul
tats obtenus sont indiqus dans le tableau IV.
Les essais dans l'installation pilote d'hydrognation auront
lieu avec un catalyseur de mme nature broy environ 40 u m de faon
tre mis en oeuvre en suspension dans la charge traiter. Il en rsultera
un accroissement d'efficacit d la rduction de granulometrie.
3.2.2. Systmes catalytiques drivs des tudes de mise en

oeuvre_de catalyseurs solubles ou en suspension
Les essais de qualification des systmes catalytiques apparais
sant le mieux adapts ont eu lieu dans la plupart des cas en autoclave
agit sur asphalte Safanya flux, l'efficacit tant estime d'aprs
l'hydrognation des asphaltnes, la destruction des complexes organomtal
liques du nickel et du vanadium (hydrodmtallation HDM), 1'hydrodsulfu
ration HDS et la formation de dpts insolubles.
Les catalyseurs dits "solubles" sont compltement dissous dans
la charge dans les conditions normales, mais les complexes organomtalli
ques qui les solubilisent sont dtruits l'approche des conditions de
raction, donnant ainsi naissance une "espce catalytique" forme in
situ l'tat trs divis.
Les rsultats les plus intressants ont t obtenus avec des
complexes organosolubles du molybdne ; quatre systmes catalytiques de ce
type ont t expriments, en comparaison d'un essai en l'absence de
catalyseur :
. un catalyseur classique d'hydrodsulfuration base de cobalt
et de molybdne dposs sur alumine, utilis en hydroraffinage sur des
charges lourdes exemptes d'asphaltnes
. deux catalyseurs organosolubles A et base de molybdne
. une association du catalyseur et de particules carbones
(cnosphres) .
Les rsultats sont rassembls dans le tableau V.
Des essais complmentaires en autoclave ont port sur l'influ
ence de la prsence de particules carbones, de la pression d'hydrogne et
de la teneur en molybdne.
Les rsultats relatifs la charge n 1 (tableau VI) confirment
que la prsence de particules carbones amliore la dmtallation et la
dsulfuration, tandis que la modification de la viscosit des produits
facilite la rcupration du molybdne, du nickel et du vanadium.
354
La rduction de la pression initiale d'hydrogne de 100 75,
puis 50 bars, correspondant respectivement 175, 125 et 75 bars dans
les conditions d'essai, a peu d'effet sur la dsuifuration, mais diminue
1'hydroconversion des asphaltnes et la dmtallation et accrot notable
ment le taux d'insolubles qui, 420C, passe de 1 2 puis 3 %.
Les essais prliminaires ayant montr l'absence de proportion
nalit stricte entre la teneur en mtal actif, en l'occurrence le molyb
dne, et le taux de conversion, contrairement ce qu'on observe avec les
catalyseurs classiques en lit fixe, on a cherch prciser ce point en
traitant la charge n 1 420C et sous une pression d'hydrogne de
175 bars cette temprature en prsence du catalyseur organosoluble
renfermant 500, 300 ou 100 ppm de molybdne, domaine des concentrations
envisages pour les essais en pilote. La baisse de la concentration en
mtal s'accompagne d'une augmentation du poids d'insolubles dans le
tolune et d'une diminution du taux de dsuifuration, tandis que la
dmtallation n'est gure affecte.
Quelques essais ont enfin t raliss sur un micropilote
continu constitu d'un racteur tubulaire de 200 cm , afin de dceler la
formation ventuelle de dpts dans diffrentes conditions opratoires.
Sur un rsidu de distillation atmosphrique de brut de Koweit, charge peu
visqueuse et facilement pompable, on a ainsi vrifi qu'un gradient de
temprature de 75 410C en 60 cm de racteur fonctionnant en "upflow"
avec une vitesse linaire de 0,06 cm/s correspondant aux conditions
d'coulement de l'installation pilote d'hydrognation ne provoquait pas
de cokage important sur la paroi du racteur en 48 h.
Dans le mme racteur et avec le mme gradient de temprature,
mais avec une vitesse linaire de 0,004 cm/s, on a ensuite trait la
charge n 1 lgrement dilue en prsence de 500 ppm de molybdne sous
forme de catalyseur organosoluble ou de catalyseur htrogne pulvru
lent de granulometrie 40200<Hm, et dans chaque cas sous alimentation
d'hydrogne ou d'un mlange constitu d'hydrogne 3 % d'hydrogne
sulfur.
On a pu observer l'absence de dpt notable au bout de 48 h et
d'accumulation du catalyseur pulvrulent mme cette trs faible vitesse
linaire, une amlioration de la dmtallation et de la dsulfuration en
prsence d'hydrogne sulfur, et l'hydrognation de 20 25 % de l'asphalte
en huile distillable, comme en tmoignent les courbes de distillation
simule dtermines par Chromatographie en phase gazeuse.
3.3. Essais dans l'installation pilote d'hydrognation

3.3.1. Essai "hydraulique"
L'essai "hydraulique" a t ralis avec succs en fvrier 1982
sur une charge constitue partir de rsidus de distillation atmosphri
que de brut Ratawi, avec une temprature de 350C dans les racteurs. Cela
confirme donc l'adaptation de l'installation au traitement de charges de
viscosit comparable.
3.3.2. Essai rel d'hydrognation

Le premier essai rel d'hydrognation a eu lieu sur la charge
n 2 dans les conditions opratoires suivantes :
Catalyseur : compos organique soluble du molybdne
355
Dbit de charge : 4 kg/h
Pression d'hydrogne : 200 bars
Temprature la sortie du prchauffeur : 300C
Temprature du premier racteur : infrieur 400C
Temprature du second racteur : quatre niveaux de tempra
ture ont t successivement
adopts, couvrant intervalle
de 50C, T r T 2 , T 3 et T 4
Un second essai diffrant seulement de l'essai cidessus par

l'addition la charge de 2 % de particules carbones a t entrepris
dans la semaine du 3 au 7 mai.
4. Analyse des rsultats et commentaires

4.1.Choix des catalyseurs
Les essais avec catalyseur htrogne (tableau IV) ont conduit des
taux de dmtallation suprieurs dans certains cas 80 % et de dsulfura
tion voisins de 60 %. Des essais raliss pour apprcier l'importance de
la granulometrie laissent prsager que le broyage 40 u m prvu pour
l'exprimentation dans l'installation pilote d'hydrognation devrait
amliorer l'efficacit observe au laboratoire.
Les rsultats obtenus avec les catalyseurs solubles (tableaux V et
VI) appellent en particulier les commentaires suivants :
. Le taux de conversion lev des asphaltnes lors de l'essai sans
catalyseur rsulte probablement de la prcipitation d'une partie des
asphaltnes contribuant l'insoluble finalement recueilli. Les asphal
tnes disparus de la phase liquide ne seraient, selon cette interprta
tion, pas valoriss, mais formeraient un rsidu solide.
. Le catalyseur de raffinage classique (tableau V colonne n" 2) se
rvle moins efficace que les autres systmes catalytiques expriments.
. Tous les essais impliquant des catalyseurs organosolubles base de
molybdne conduisent une conversion notable des asphaltnes en hydrocar
bures valorisables : les dpts observs en fin d'essai sont en effet peu
abondants.
. Les catalyseurs A et ont des activits comparables.
. L'association d'un catalyseur organosoluble et de particules
carbones (tableau V colonne n 6 et tableau VI) conduit une nette
amlioration de la dsulfuration et de la dmtallation.
. Bien que le rapport atomique H/C de la fraction asphaltnes soit
compris entre 1 et 1,1, on observe au cours des traitements catalytiques,
et spcialement en prsence de catalyseurs organosolubles, une hydrogna
tion de la fraction des asphaltnes rendue soluble dans l'huile ainsi
qu'une hydrognation de l'ensemble des hydrocarbures. L'essai 7 (tableau
V) montre clairement le pouvoir hydrognant des systmes base de molyb
dne, et les essais 3 6 indiquent que cette proprit est conserve en
prsence d'asphaltnes.
On rappellera enfin les conclusions tires des essais en racteur
tubulaire :
. des formations de coke ou des dpts de catalyseur pulvrulent ne
semblent pas craindre dans les conditions de fonctionnement de l'instal
lation pilote d'hydrognation.
. les taux de dmtallation et de dsulfuration sont amliors
lorsqu'on opre avec un mlange d'hydrogne et d'hydrogne sulfur.
356
4.2. Premier essai d'hydrognation en installation pilote
Le dpouillement de cet essai n'est encore que trs partiel, en
raison de son caractre rcent. Les principaux rsultats, provisoires,
actuellement disponibles sont les suivants :
Distillation simule
Temprature du second racteur

Charge
initiale Essai A Essai 1 Essai C Essai D
T T T T
i 2 I 3 4
< 240 C 5,85 % 7,03 % 10,71 % 1 14,26 % 18,74 %|

240350C 34,58 % 30,66 % 34,69 % %
39,17 41,46 %
| 350500C 13,44 % 15,05 % 18,66 % 1 21,11 % 21,95 % |
1 > 500 C 46,13 % 47,26 % 35,94 % 25,46 % 17,85 %
Teneur en H S des
gaz de sortie (vol. %) 1 2,4 % 2,6 % 2,7 %
Il est manifeste, comme on pouvait s'y attendre, que la temp

rature de 380C est insuffisante pour provoquer une conversion relle de
la charge traite dont en revanche la viscosit subit dj une rduction
sensible. Mais l'lvation de temprature audessus de 400C entrane un
enrichissement marqu en fractions lgres et moyennes au dtriment de la
fraction la plus lourde, ainsi qu'un accroissement du taux de dsulfura
tion. Il est noter toutefois que, pour l'essai D ( ) , la prsence de
"solides" autres que le catalyseur a t dcele dans 1'hydrognat, et l'on
peut penser que la poursuite du traitement dans de telles conditions aurait
conduit des risques de bouchage de l'installation.
Les essais ultrieurs auront en particulier pour objectif de
prciser, pour ce type de charge, la temprature optimale d'hydrognation,
qui semblerait se situer autour de 420425C.
5. Conclusion
Bien qu'il soit prmatur de vouloir tirer des.conclusions dfiniti
ves de l'exprimentation ralise, on peut nanmoins, titre provisoire,
souligner les points suivants :
. la ncessit de l'addition l'asphalte d'un solvant fluxant
abaissant la viscosit de la charge de faon en assurer une circulation
satisfaisante dans l'installation d'hydrognation ; un taux de l'ordre de
50 % s'est avr ncessaire pour satisfaire aux conditions de marche de
l'installation existante, mais on peut penser qu'une tude plus pousse de
certains circuits permettrait de rduire sensiblement cette valeur.
. la slection de diffrents systmes catalytiques entrans parais
sant, d'aprs les essais de laboratoire, de nature conduire des rsul
tats satisfaisants dans les conditions relles de traitement.
. l'adaptation de l'installation d'hydrognation au traitement des
charges d'asphalte flux dans la gamme des conditions opratoires envisa
ges.
. l'allgement notable de la charge d'asphalte Safanya flux par
50 % de "light cycle oil" aprs traitement 200 bars haute temprature
357
et en prsence d'un catalyseur organomtallique soluble base de molyb
dne.
Ces rsultats demandent naturellement tre complts, confirms et
tendus par la poursuite de l'exprimentation.
TABLEAU I
CARACTERISTIQUES DU SOLVANT SERVANT DE FLUXAN1r
Rfrence LCO 1669
Densit 20C 0 966

Indice rfraction 20 C 1 5621
Pt coulement 30
4,0
Viscosit 50C cSt) 2,15
Soufre % 3,34
Nickel ppm < 0,01
Vanadium ppm < 0,01
Azote total ppm 245
Pt clair vase clos C 75
Distillation A STM
Pt initial 160C
10 % 229C
50 % 264C
90 % 323C
95 % 336C
Pt final 341C
% distill 98
Rsidu 1,5
% C 87,6
% H 9,0
H/C atomique 1,23
358
TABLEAU I I
CARACTERISTIQUES DE LA CHARGE N 1
Rfrence S 1729
Origine Unit de dsasphaltage du CEDI
Composition 65 % asphalte SAFANYA
35 % fluxant (LCO S1669)
Densit 20C 1,063

Viscosit 100C 718 cSt
Viscosit 150C 90 cSt
Carbone Conradson 26, 1 %
Asphaltnes C 22,8 %
Nickel 75-80 ppm
Vanadium 225-270 ppm
Soufre 5,28 %
Pt coulement 27C
Pt clair 192C
Rapport atomique H/C 1,22

TABLEAU I I I
CARACTERISTIQUES DE LA CHARGE N 2
Rfrence : S 2279
Origine : Unit de dsasphaltage du CEDI
Composition 50 % asphalte SAFANYA
50 % fluxant (LCO S 2206)
Densit 20C 1,0565 g/cm3

Viscosit 100C 100 cP
Viscosit 150C 12 cP
Carbone Conradson 23,t %
Asphaltnes C 23,7 %
Nickel 52 ppm
Vanadium 167 ppm
Soufre 5,47 %
Distillation simule :
< 240C 5,85 %
240-350C 34,58 %
350-500C 13,44 %
> 500C : 46,13 %
-360-
TABLEAU IV
ESSAIS DE SELECTION DE CATALYSEUR DE TYPE HETEROGENE
Charge
Origine brut d'Athabasca Point d'coulement + 6C

Densit 15C 0,999 Distillation ASTM
Viscosit 50C 2150 cSt point initial 171C
Viscosit 100C 103 cSt 10 % en volume 301C
Carbone Conradson 12,8 % 50 % en volume 539C
Asphaltne au C 7,5 %
Nickel 81 ppm
Vanadium 220 ppm
Soufre 4,27 %
Conditions
I
Pression totale 100 bars
Temprature variabl de 360 420C
Dbit de charge 1 1 par heure et par litre de lit cata
lytique
Dbit d'hydrogne 850 1 TPN par litre de charge '
Rsultats
de 10 200 h
temprature C HDM % HDS %
360 45 45
380 55 51
400 68 54
420 88 62
aprs 390 h
400 65 45
tion
HDM = hydrodmtallation
HDS = hydrodsulfuraition
361
- TABLEAU V -
CONDITIONS ET RESULTATS DES ESSAIS SUR ASPHALTE SAFANYA FLUXE (CHARGE n 1)
Numro de repre 1 2 3 4 5 6 7
Nature de l'essai Thermique Catalyseur Catalyseur Catalyseur Catalyseur Catalyseur Catalyseur

sans classique A A B B B
catalyseur CoMo/Al20 organo- organo- organo- + 1 % de organo-
broy soluble soluble soluble particules soluble
sulfur carbones
Teneur en mtaux 0 500 ppm Mo 500 ppm Mo 200 ppm Mo 500 ppm Mo 500 ppm Mo 500 ppm Mo
Charge Asphalte Asphalte Asphalte Asphalte Asphalte Asphalte Fluxant seul

flux flux flux flux flux flux LC0
Conditions opratoires
Temprature (C) 411 121 119 418 418 418 418
Dure (heures) 2 2 2 2 2 2 2
Pression initiale (bars) 100 100 100 100 100 100 100
Rsultats
Insolubles au tolune
(% pds) 10 <0,3 <1 <0,3 <1 <0,3 < 1
HDS % 7 30,3 36 30 39,5 40,5 98
HDA % 27 16,7 36 34 40 35,5
HDV % 10 53 80 73 87 96,5
HDNi % 30 30 51 28 64 86,5
H/C atomique 1,2 1,22 1,33 1,29 1,30 1,33 1,44
(charge non traite (solvant
H/C = 1,22) non trait
1,23)
HDA taux de conversion des asphaltenes.
- TABLEAU VI -
CONDITIONS ET RESULTATS DES ESSAIS
SUR L'INFLUENCE DE PARTICULES CARBONEES
Charge : charge n 1
Conditions
Essai n 1 n 2 n 3
Mo ppm/charge 500 500 500
Additif type nant srie A srie
% pds/charge 0 1,0 1,0
20C (bars) 100 100 100
C 420 420 420
Rsultats
H DA % 49,7 40,3 51,8
HDV % 90,3 93,9 95
HDNi % 75,6 80 83,7
HDS % 48 - 54 59 - 65 59 - 65
Ajustement de s Ajustement det
proprit! par reparutions par
hydroralfinag era qua g
Brut d origine

X
varie
[ Fraction ^, M E
D , ' lgres f

1 Craquage catalytique 1
-' - *g,. 1' inter
A
Fractions
m dieu rea
X R
C
H
R
0
L
I
E
R
D. . : Distillation
atmosphrique
FIG. 1 Schema de raffinage conventionnel
Hydroraffinage
X
m
\ Fractions
- # ' lgres
Brul D
N^_-"
*"
.
| Fractions
moyennes X -
A

Craquage
D
R
Pmrtinnt my*nptf R
S catalytique -
roque*
* c
V
I.
H
I
, 0 A.S

R
. : Distillation atmosphrique
D.5.V.; Distillation sous vide
DAS = Cscspholtage au peritane
FIG. 2 Schema de raffinage avec conversion des produits ptroliers

J HUILE )5\!.\.
(SOLVANT J
&
ASPUAITE
FIG. 3 Desasphaltage au solvant
364
Hydrogene Events
Catalyseur
1
N2

X-1 2\
C1
Asphaltes
ND non dis illable Huiles

S = solides 200 - 500 "C
ALIMENTATION EN ASPHALTESCIRCUIT D'HYDROGENE REACTEURS SEPARATEURS EPURATEUR DISTILLATION
Ct Ft d'asphaltes flfl -
Rserves d'hydrogneR 100 _ Prchauffeur B1 _ Sparateur chaud B3 _ Laveur soude 12 _ Ballon de recette
C2 Bec jaugeur haute pression Racteur B2 _ Sparateur froid P3 _ Pompe de circulation b 13 Recette des lgers
R101
Cadres d'hydrogne RW2 _ Racteur B5 _ Vanne de dtente VS _ Dverseur rgulation BH
_ Recette
P1 _
P2 _
Pompe heute pression C3 -
Pompe haute pression Kl _Compresseur BIO _ Ballon du circuit BS _ Ballon de dtente de la pression
_ moyennesdes huiles
des disques de B7 _ Ballon de dtente B 1b
PS Pompe de garage VI _Vanne d'arrt
C4
_ Sas de rsidus
de secours rupture
Bolln du circuit '}* b
_ Colonne atmosphrique
V2 Vanne de rglage BU
de purge des V4 _ Vanne de dtente b Flash sous vide
Evaporateur sous vide
du dbit
V3 _ Vanne d'arrt de racteurs V5 _ Vanne de dtente PS _ Pompe doseuse
fa purge d'azote VS _ Vanne de purge P7 - Pompe de remise
de secours la pression
atmosphrique
VIO Vanne de dtente
FIG. 4 Schema de la boucle d'hydrognation des asphaltes

1 = Lignes "downflow" CATATEST "PRODUITS LOURDS"
2 = Lignes "upflow"
= Dbitmtre
R.R.Q)
:
^T
RTR.''
W o--.
N2 - ^ ^ - ^ - n ^ ^ . - U w y ^ U a M4a
IJ
D

1 t
Encein
inte rchauffe 0
)
? t
J
_Y i
Separateti
~ G/L.HP
-^-iJd-co
*L.,= .t
zM P. Tne- P.S. T
FIG. 5
DISCUSSION
Question - Prof. Uytterhoeven, KUL : Le cataLyseur C O M O / A L T O T a u n e

performance moins bonne que Les catalyseurs
soLubLes. Est-ce d un bouchage des pores ou une
raison chimique ?
Rponse Le cataLyseur CoMo/AlpO, auqueL vous faites mention (tabLeau V,

repre 2) est un cataLyseur conventionnel d'hydroraffinage.
IL a t conu pour Le traitement des hydrocarbures exempts
d'asphaLtnes et de rsines. Le diamtre moyen de sa porosit
est de 80 - 120 S. La mauvaise performance obtenue est
probabLement due L'inadaptation du cataLyseur La nature
de La charge et aux conditions opratoires utiLises. Dans
ces conditions, caractrises par une forte temprature,
c'est probabLement Le dpt de "coke" qui limite La perfor-
mance. En effet, on peut calculer que pour une dmtalLation
totale sur ce catalyseur 14 % M0O3, 3 % CoO, 83 % A L 2 0 3 ,
on obtiendrait un dpt de moins de 10 % de nickel et
vanadium. Cette valeur est faibLe et ne provoquerait pro-
bablement pas de bouchage des "bouches de pores".
En revanche, le catalyseur dont iL est fait mention tableau IV

est spcialement conu pour tre utilis en prsence
d'asphaltnes. Les rsultats rapports dans Le tableau
montrent bien qu'iL rsiste l'empoisonnement par le coke
et les dpts nickel et vanadium. Aprs 400 heures de fonc-
tionnement, on note une faibLe dsactivation qui affecte
surtout La fonction dsuifuration.
L'uti li sation de ce second catalyseur est donc envisage

dans un prochain essai.
Question - Dr. Jacobs, KUL : Vous avez mentionn une conversion notable
des asphaltnes en produits valori sables. Quel genre de
conversions chimiques se produisent sur ces catalyseurs dans
Les conditions de raction ? hydrognation, hydrocraquage ?
Rponse La complexit, du mlange d'hydrocarbures soumis La raction

rend difficile une rponse claire votre question. La
conversion mentionne est exprime en variation de la quantit
de produits dont le point d'bulLition est suprieur 500 C.
On mesure donc ainsi' une volution des poids molculaires.
Deux mcanismes diffrents ont t avancs pour expliquer Le
phnomne observ. L'un fait appel La nature des produits
Lourds et des asphaLtnes, l'autre d aux chercheurs de EXXON
fait appel au craquage thermique.
La teneur en soufre dans les asphaltnes et dans Les produits

lourds non asphaltniques est si leve qu'en moyenne, chaque
molcule possde un ou plusieurs atomes de soufre; L'limi-
nation de ceux-ci sous forme d'hydrogne sulfur peut donc
fragiliser La molcule et favoriser son craquage ou son hydro-
craquage. D'autre part, on admet couramment que les asphaltnes
dans des conditions d'hydrognation, sont dtruits suivant une
raction conscutive que L'on symbolise ainsi :
-367-
H H
2 2
AsphaLtnes > Rsines > Huile
(polymre (monomre (naphtnoaromati-
de feuillets polyaromatique que sans Ni et V)
polyaromatiques) plus hydrogn)
Les produits de la "dpolymrisation" des asphaltnes en rsines

au cours de l'hydrognation contiennent des fragments beaucoup
plus lgers que la charge initiale, ce qui expliquerait la
conversion obtenue.
Le second mcanisme dcrit par EXXON est spcifique des cata-

lyseurs base de molybdne soluble. Il suppose un craquage
thermique classique suivant un mcanisme de radicaux libres.
Le catalyseur dispers base de molybdne sulfur (MoS?)
interviendrait pour hydrogener les fractions lourdes qui, en
son absence, donnerait des polymres lourds et finalement du
coke.
Actuellement, nous ne disposons pas d'informations suffisantes

pour trancher entre ces deux explications.
Question - Dr. Pilavachi, CEC : Pouvez-vous expliquer pourquoi

1) la prsence de particules charbonneuses amliore la
dmtallation et la dsulfuration ?
2) la diminution de la pression d'hydrogne a peu d'effet
sur la dsulfuration mais diminue l'hydroconversion des
asphaltnes et la dmtallation tandis que la qualit
d'insolubles crot notablement ?
3) lorsque la concentration de molybdne dcrot, on observe
une augmentation du poids d'insolubles dans le tolune,
une diminution de la dsulfuration, mais la dmtallation
varie peu ?
Rponse 1) On ne connat pas exactement par quel mcanisme les parti-

cules charbonneuses amliorent la dmtallation et la
dsulfuration.
2) L'effet de la pression partielle d'hydrogne correspond bien

aux observations que l'on peut faire sur des systmes
catalytiques classiques. L'effet de la pression partielle
d'hydrogne sur la dsulfuration se traduit gnralement par
une augmentation quasi linaire dans le domaine des faibles
pressions, puis au-del d'une valeur variable suivant les
caractristiques du ptrole trait, la dsulfuration
n'augmente que faiblement lorsque la pression d'hydrogne
augmente au-del de ce seuil.
-368-
Les ractions qui conduisent La conversion des asphaLtnes
ne sont pas encore compltement lucides. On admet gn-
ralement que l'hydrognation des noyaux polyaromatiques
condenss est une tape importante. C'est probablement la
raison pour laquelle on observe exprimentalement une
augmentation de la conversion des asphaLtnes quand La
pression d'hydrogne augmente. Gnralement, La dmtal-
Lation et La conversion des asphaLtnes sont corrles.
En effet, La destruction de l'difice polymrique des
asphaLtnes permet aux complexes organomtalliques insrs
dans la particule d'asphaltne de venir ragir La surface
du catalyseur.
La formation d'insoLubles est attribue des ractions de

poLycondensation de noyaux aromatiques. Ces ractions sont
d'autant moins favorises que les conditions d'essai sont
pLus hydrognantes (pression d'hydrogne leve, faible
temprature).
3) En ce qui concerne L'effet de La diminution de La teneur en

molybdne sur la formation d'insolubles au toLune et La
dsuifuration, iL s'agit d'observations classiques qui ont
dj t dcrites sur de nombreux systmes de catalyseurs
htrognes. La faibLe sensibilit de la dmtaLLation
la teneur en catalyseur est plus tonnante. C'est proba-
blement un comportement que L'on doit attribuer La valeur
relativement leve de La temprature qui permettrait La
dpolymrisation et le craquage thermique des particules
d'asphaltnes en feuilLets lmentaires Librant les
complexes organomtalLiques jusqu'alors enferms dans la
structure du polymre. Ces complexes organomtalliques
ainsi dstabiliss seraient trs ractifs et de faibles
teneurs en catalyseur suffiraient produire des vitesses
de raction importantes. L'tape limitative dans Le cadre
de cette hypothse serait le craquage de L'asphaltne qui
dpend essentiellement de la temprature et du temps de
raction.
-369-
SESSION VII - MICROWAVES
Chairman: D.A. REAY
Summary of Session VII
Application of microwave heating and hot air for the continuous

production of plasticized PVC sheets
Microwave properties of some building materials in view of an

oven design
Development of a microwave applicator to dry thin structures -

First semi-annual report
371 -
SUMMARY OF SESSION VII
MICROWAVES
D.A. Reay, P.A. PiLavachi
Microwave energy has been used by industry for heating and drying proces-
ses for many years, but except for a few specialized applications, it has
not been generally introduced. Commercially available microwave machines
are applied to the continuous vulcanising of extruded (and other) rubber
components, (rubber being a bad heat conductor), and the food processing
industry, particularly in the United States, uses large microwave systems.
The reasons for the poor market penetration by this system are easy to
identify. To date, microwave systems have had high capital costs. It was
recently estimated by a manufacturer of such equipment that installed
machines of 10 kW or greater power currently cost between 1500 and
2000 per kW (2700 - 3600 ECU per kw). Running costs can also be high,
particularly when electricity tariffs are compared with the low prices of
natural gas which were common until comparatively recently. A conventional
microwave oven of 50 per cent efficiency overall would not compete on an
energy cost basis with a gas fired system of 25 per cent efficiency,
although of course many fossil fuel-heated systems have much lower
efficiencies. As discussed below, PVC treatment ovens may be only 5 per
cent efficient.
Microwave systems have benefits in other areas. They are inherently clean,
highly controllable, and convenient. Improved product quality and/or
production rate may result from the use of microwaves. Modifications may
of course be made to the microwave system itself to improve efficiency and
reduce the cost, and the economic benefits may well be acceptable when
microwave units are used in preference to very low efficiency units
operating on other fuels.
The contractors'meeting on microwaves considered three areas of application

where economic benefits might accrue. These included the treatment of
plastic sheeting, the production of construction materials (wood drying
and concrete curing) and the drying of coated paper.
Continuous production of plasticised PVC sheets - Wo r k at EUROFLOOR in

Luxembourg is directed at building and testing a pilot microwave oven for
treating PVC floor coverings. These are formed by heating a PVC paste,
which then solidifies. Normally this is carried out in a thermal oven at
temperatures between 175 C and 210 C at an efficiency of only 5 per cent.
Because of the low efficiency, the process seemed to be an ideal candidate
for microwaves.
The oven used has a power input of 6 kW and is designed to treat the pro-
duct at a rate of 20-30 metres/minute. A supplementary hot air source is
used to maintain the surface temperature of the sheet at a sufficiently
high level.
-372-
WhiLe the curing (gelling and foaming) of the PVC can be made to take place
rapidly in the oven, results to date have shown that local burning of the
material can occur very rapidly due to nonuniformity of the applied energy.
It was suggested that the material could influence the field in an unpredi
cable manner. It was also found that the hot air increased the sensitivity
of the material to microwaves, and the position and thickness of the sheet
also were important.
Work is under way to attempt to resolve these difficulties, and it has

already been shown that a uniform field effect with the applicator type
used for the PVC could be obtained when thick wet paper was treated.
Use of microwaves_for_ome_bu2ld2ng_maerials The RIJKSUNIVERSITEIT OF

GENT is studying the application of microwaves to the curing of concrete
and drying of wood materials typically used in the construction industry.
A number of stages are involved in the research programme, including
development of a 5 kW generator and applicator, construction of a laboratory
microwave oven, and the determination of the microwave and thermal proper
ties of the materials to be tested. The work to date at Gent has concentra
ted on the last aspect above, the determination of the appropriate proper
ties of the materials to be treated. This has also necessitated the
perfection of techniques affected by microwave absorption.
Properties of the material needed are the dielectric constant and the
permittivity. The permittivity of both bulk and sheet materials was
measured, the latter requiring the development of an automatic system.
All the relevant properties of cellular concrete have been successfully
measured, and fast curing experiments have been carried out. It was
found that intermittent microwave heating could be used to even out non
uniformities in temperature. Properties of wood venears have also been
determined.
An infrared recorder was used to determine surface temperatures of

material subjected to microwaves, while internal material temperatures
were measured using thermocouples. A computer model using microwave and
thermal performance equations may be used to predict the variation in
temperature in samples as a function of time and applied microwave power.
The work is now being applied to the design of a microwave oven, into
which a microprocessorbased control system will be integrated.
D
yD2 thin structures A second project in Belgium, located at the
KATHOLIEKE UNIVERSITEIT, LEUVEN is directed ultimately at the drying of
coated paper materials. It is claimed that energy savings of the order
of 70 per cent could be achieved if microwave systems were used instead
of conventional hot air drying.
The study at Leuven involves both theoretical and experimental components,

leading ultimately to a prototype system in which samples of coated paper
will be dried. Work reported at the meeting included the development of
a reflection coefficient meter, used for matching the load and the power
applied via the waveguide. Using an arrangement of probes, linked to a
373
microprocessor, matching may be based on an analysis of the reflection
coefficient, reflected and forward power and the average voltage of the
diodes (probes). If the mismatch can be eliminated, higher efficiencies
could be achieved.
In addition to this work, experiments are in hand to measure the microwave
properties of potential products.
All the application areas considered by the contractors are ones in which
it is believed that significant energy savings can be achieved, should the
use of microwave systems prove successful. As economic analyses are a
mandatory part of each project, the contractors will at a later stage
present information on the financial returns likely to be anticipated,
were the microwave process to be adopted. Much rests on the ability of
the researchers to minimize the capital cost of the plant while ensuring
operation at high efficiency.
374-
APPLICATION OF MICROWAVE HEATING AND HOT AIR
FOR THE CONTINUOUS PRODUCTION OF PLASTICIZED
PVC SHEETS.
Authors A. ROUSSEL, L HAAG
Contract number EEB-1-112-L
Duration 12 months
Head of project M. BOMBOIRE
Contractor EUROFLOOR S.A.
Address rue neuve

WILTZ, G.D. of Luxembourg
Summary
A microwave tunnel has been achieved for continuous

gelling and foaming plasticized PVC sheets. The poor
efficiency of the hot air pulsed heater for this use is
the reason of this choice The microwave tunnel is made
by 6 wave guides of 1 kw each. The plasticized PVC sheet
with blowing agent is foamed very quickly. But in spite
of addition of hot air, the nonuniform repartition of
energy did not permit to find the correct foaming of the
sheet.
-375-
1.1. Introduction
La fabrication de revtements de sol plastique est
ralise principalement par des enductions successives de
plastisols de Polychlorure de vinyle plastifi. Ces plasti-
sols sont des ptes obtenues par mlange sous agitation de
poudre de PVC, de plastifiant et d'adjuvants (stabilisants,
charges). Egalement on incorpore dans certaines couches des
agents porognes permettant l'obtention de mousses: ces
agents gonflants ont la proprit de se dcomposer et de li-
brer leur gaz des tempratures variant de 175 degrs C
21o degrs c La glification (durcissement) de ces couches
s'effectue dans des fours air chaud puis dont le rendement
actuel est de l'ordre de 5 % L'amlioration du systme de
chauffage devra permettre de faire de srieuses conomies
d'nergie.
Parmi les techniques existantes, le chauffage par micro-
ondes semble particulirement intressant. En effet, les
caractristiques lectriques du PVC correspondent aux exi-
gences du procd. De plus 1'chauffement se produisant au
sein de la matire, il devrait en rsulter une forte diminu-
tion des pertes de chaleur dues aux passages par des changeurS
(fluide-air, air-matire ..) et une rduction de l'encombre-
ment d. au volume norme d'air ncessaire pour assurer l'homo-
gnit du chauffage. D'autre part, un chauffement au sein
de la matire peut prsenter de gros avantages dans le cas
d'enduction moussable car il devrait favoriser la rgularit
de la cellurisation et les proprits mcaniques qui en d-
coulent.
Dans ce rapport nous rendrons compte pour un certain
type d'appareillage micro-ondes (guides fentes rayonnantes)
des effets de glification et d'expansion de feuilles de PVC
plastifies.
1.2. Appareillage
Des essais prliminaires ont montr que pour obtenir une
glification par micro-ondes de PVC dans des temps courts et
donc une lvation de temprature rapide, le matriel doit
permettre l'obtention de densit de champ lectrique impor-
tante pour une puissance minimum. De plus le choix de l'appa-
reillage est galement li la gomtrie particulire du
produit (feuille) et au mode de traitement (continu) de ce
produit. Le choix de 1'applicateur sera par consquent guid
par ses considrations prliminaires
Des essais d'orientation ont montr que 1'applicateur
de type guide fentes rayonnantes donne les meilleurs rsul-
tats. Dans ce type d'application, le guide fentes rayon-
nantes transversales semble le mieux adapt. Si nous excep-
tons la rgularit de chauffage, il permet une glification
des enductions PVC. Les caractristiques obtenues dans les
zones irradies semblent correspondre aux caractristiques des
produits obtenus par un procd de chauffage classique. Ce
dispositif permet une lvation de temprature rapide et l'ob-
tention de la glification dans des temps courts pour une
puissance faible
L'apparition du phnomne de glification et du fait
mme de l'intensit du champ lectrique est suivie trs rapi-
-376-
dement d'un phnomne de carbonisation: la diffrence de
temprature entre la glification et la carbonisation tant
faible. Ces rsultats ont suggr l'adjonction d'une unit de
chauffage classique par air puis pour temporiser l'emballe
ment de la raction.
Aussi le four tunnel ainsi ralis atil ces caract
ristiques suivantes:(voir figure 1)
largeur utile 1 m
6 guides d'ondes disposs transversalement
(puissance 1 kw chacun) possdant 23 fentes
dimension de chaque fente: 12,5 x mm
longueur d'onde = 12,2 cm ( f = 2^5o M H z )
extracteurs de vapeurs de plastifiants
admission d'air chaud entre les guides d'ondes avec
rgulateurs indpendants
bande transporteuse en tissu de verre garnie de PTFE
Le montage des 6 guides d'ondes identiques ont t d
cals l'un par rapport l'autre de telle manire que la
nappe de PVC traverse pendant son passage dans le four un
champ identique en tous ses points pris sur une mme droite
perpendiculaire au sens du dfilement du produit.
Le rglage du champ des microondes peut se faire par
la variation de 4 paramtres: (voir figure 2)
cote plafond : hauteur du rflecteur par rapport
au produit
cote guide : profondeur du guide par rapport
au produit
dcalage avant : position du guide par rapport la
largeur
cote du plongeur
d'accord : rglage du taux d'ondes station
naires (TOS).
1.3 Essais et rsultats

1.3.1. Choix du matriau d'essai
La matrialisation du phnomne de glification
se traduit par le passage d'une phase liquide en une
phase solide.
Toutefois on peut passer par une phase interm
diaire, dite ''prglifie'' puis effectuer sous micro
ondes le processus de glification: mais l la mat
rialisation de la glification est plus difficile
visualiser: en effectuant des essais de traction par
exemple.
Il nous a sembl plus pertinent d'effectuer des
essais avec enduction moussable car l le processus de
moussage (dcomposition d'un agent gonflant) se mat
rialise par une augmentation d'paisseur donc quelque
chose de bien visible et qui va nous permettre de situer
les impacts des ondes.
Aussi nous avons ralis diffrentes enductions
moussables de PVC plastifi (, mm) sur un film calan
dr PVC plastifi (0,5 m m ) . Industriellement l'expan
sion en prsence d'air chaud d'un tel produit se fait
I90 degrs C durant 2 minutes J o sec. conduisant une
augmentation d'paisseur de o,5 mm.
1.32 Reglages paramtres
Comme nous l'avons dit prcdemment, la tech
nique utilise ne permet pas l'obtention d'un champ
lectrique rgulier, h. premire tape de cette tude
nous a conduit dterminer la rpartition de ce champ
dans l'espace cr par le guide d'ondes et le rflecteur
puis le positionnement de la feuille de PVC dans cet
espace pour optimiser les rsultats de glification.
Pour ne pas multiplier inutilement les para
mtres variables, nous avons travaill tout d'abord
sans dfilement de produit et dure d'irradiation
constante. Le taux d'ondes stationnaires (TOS) est
rgl pour chaque exprience de manire obtenir un
rapport optimal.
Par le rglage du cote plafond et cote guide
nous avons pu constater une variation importante des
impacts et dterminer:
un optimum de cavit inter guide de 758 mm
pour laquelle on a un maximum d'impacts
(< 23 fentes)
une image des fentes des guides lorsque le
produit se trouve contre le guide cdire
la cote guide = o
des constatations prcdentes semblables mais
non identiques pour les 6 guides d'ondes
(htrognit des impacts).
Pour ces rglages ci dessus, effectus en fai
sant varier le temps d'exposition aux microondes, il
nous a t possible de dterminer:
une augmentation des impacts en fonction du
temps d'exposition
une htrognit des impacts indpendant du
guide d'onde cf photos 3^5 En effet on peut
observer certains impacts 15'' d'exposition,
invisibles 1 5 ' '. Ces variations sont
vraisemblablement attribuables l'imprcision
de la position et l'paisseur de la feuille
traiter.
1.33 Influence air chaud et position de la feuille
Nous avons cherch a amliorer l'homognit de
l'intensit du champ lectrique hauteur de chaque
fente rayonnant par l'apport d'air chaud et par la
position de la feuille dans la cavit.
Comme on pourra le constater sur les photos
6,7 et 8, l'apport d'air chaud amliore comme prvu la
sensibilit du matriau aux microondes. Sans la pr
sence d'air chaud on obtient des premiers impacts au
bout de 3o45 secondes. Avec air chaud (rglage llo
12o degrs C) on arrive au mme effet avec seulement
15 secondes de temps d'exposition auxnicroondes . Les
impacts apparaissent plus nombreux hauteur de chaque
fente mais subsistent malgr tout de nombreux points de
carbonisation.
De ce fait nous avons fait varier la position
de la feuille dans la cavit guide plafond pour tenter
avec air chaud de diluer le champ lectrique.
voir photos : photos distance
guide plafond
9 lo 11 6o mm 2o
12 13 Ik ko ko
Sans air chaud (photos 9 et 12) on perd une

grande partie de l'nergie et l'on obtient peu d'im
pacts mais avec air chaud le roduit traiter devient
plus sensible et visualise des impacts.(Les feuilles
ont t soumises un air chau d de llo12o degrs C
pendant 5 minutes)qui ne sont plus des images des
fentes, mais la rsultante des diffrents rebonds sur
les parois de la cavit et inte rfrences d'une fente
l'autre, d'o ces images 5 impacts.
L aussi on a amlior l'homognit de 1 ' ner
gie mais sans arriver la per fection.
1. 3^.Influence matires premires

On sait que la puissance dissipe sous forme
de chaleur dans un corps soumis un rayonnement
microondes est proportionelle son facteur de perte
E' '= E t g " avec E = cons tante dilectrique
tg<f = tangente de l'angle de
perte
Aussi si l'on examine les caractristiques d'uri
PVC plastifi 35 % de plastifiant DOP on a un fac
teur de perte E''=o,45, ce qui est tout de mme lo
fois suprieur la moyenne des matires plastiques
(y compris le PVC rigide) et l.ooo fois suprieur au
PTFE, mais le facteur de perte du PVC plastifi reste
lo fois infrieur celui de l'eau.
Une bonne part de l'absorption d'nergie serait
donc attribuable au plastifiant. C'est pourquoi,ne
conaissant pas le facteur de perte, nous avons pra
lablement effectu des essais en four microondes en
cavit multimodale sur diffrentes matires premires
susceptibles d'tre utilises: l'augmentation de
temprature est mesure aprs 1 minute sous 1 kw, on
obtient:
aucune lvation de temprature avec les PVC
poudre, carbonates de calcium, oxydes mtal
liques ( Zinc, titane), agent gonflant (azo
dicarbonamide )
5o degrs C pour le Dioctylphtalate, le Di
isononylphtalate
55 degrs C pour divers stabilisants
758o degrs C pour 1'thylglycol et un m
lange de diisobutylphtalate et de diisod
cylphtalate (4o=6o) (DIBP + DIDP)
115 degrs C pour le TXIB (2,2,4 trim
thyl 1,3 Pentanediol, diisobutyrate).
En consquence diffrentes formules ont t
essayes en fonction des rsultats prcdents et nous
avons incorpors les additifs suivants:
charge: minex 7 ( silice naturelle ) ko per
379
- charge: Alumine hydrate ko per
- carbon Black 2o per
- mlange DIBP et DIDP 60 per
- TXIB 2o per
Seule la formulation avec carbon black a mon-
tr une sensibilit accentue auxmicro-ondes ( temps
divis par 2 ) . Malheureusement si cette matire est
utilise frquemment dans le caoutchouc, il n'en est
rien pour nos produits car on ralise des impressions
en quadrichromie sur des supports moussables qui
doivent tre blancs.
Les autres additifs (mme le TXIB) n'ont pas
permis d'observer de diffrence de sensibilit aux
micro-ondes, ceci pour diffrents temps d'exposition,
avec et sans air chaud.
1.35.Dfilement du produit
En faisant defiler la feuille de PVC plastifie
avec prsence d'air chaud il est possible d'expanser
et de glifier la feuille.
Toutefois compte tenu des problmes d'htro-
gnit rencontr en statique, le rsultat est nga-
tif se traduisant par des stries de carbonisation du
matriau et des stries de matriau non expans.
A cela vient s'ajouter le fait que la rgula-
tion d'air chaud actuelle n'est pas uniforme ( batte-
ments +/- 15 degrs C ) ce qui pour effet d'aug-
menter cette htrognit.
1.4. Conclusions
Bien que des essais pralables en utilisant du carton
humide nous avaient permis de modliser notre appareillage et,
bien qu'il soit possible de glifier et d'expanser rapidement
du PVC plastifi avec ce type d'appareillage c'est--dire
comprenant des guides rayonnants, nous n'avons pas obtenu une
relle uniformisation du rayonnement.
Nous sommes amens penser que l'htrognit des
impacts est li:
- la configuration gomtrique des guides d'ondes: des
imperfections minimes entranant des interaction sur
les ondes
- au milieu brusquement ractionnel qu'est le PVC plas-
tifi avec des absorptions nergtiques alatoires ce-
ci malgr une recherche d'autres additifs, nous sup-
posons que le facteur de perte augmente brutalement
en fonction de la temprature ce qui ne fait qu'ac-
crotre cette absorption
- la position de la feuille de PVC et aux variations
de son paisseur et de son profil ( faibles ondula-
tions ) .
Figure 1 Four Tunnel
air chaud
plafond /
tapis f^MJ^JMUULP-UUL

i Li i 1 i U Li
] d D G
gnrateur microondes
Figure 2 Paramtres variables du Four Tunnel
coteplafond
plongeur d'accord
cote plongeur
guide d'onde d ' accord
vue en coupe
381
PHOTO PHOTO 3
PHOTO
microondes: ,"> sec m.icroondes : hH sec.
microondes: 1 5 sec,
cote plafond 7 mm cote plafond 7 mm
cote plafond 7o mm cote guide lo mm
cote guide lo mm cote guide lo mm
PHOTO PHOTO 7 PHOTO fe
Prchauffage 'j rain, 113 d.C. Prohavif f ago : 5 m i n . 115 d. Prchauffago : 5 min. 115 degrs C
mi ero-ondes 'l5 s e c . m i c r o - o n d e s : '}o s e c . micro-ondes : 15 sec.
cote plafond 7 o mm c o t e p l a f o n d : 7 o mm cote plafond: 7o mm
::
cote guido l o mtn cote guide : I o mm cote guide : lo mm -^: ' -
PHOTO lo PHOTO y
PHOTO li
Prchauffage: 5 min. 115 d.C. Sana Prchauffage
Prchauffage : 5 m i n . 1 1 5 cl. C . micro-ondes : 2 min.
micro-ondes : 1 m i n . TO S R C . micro-ondes : 1 min.
cote plafond: 6o mm cote plafond: 6o mm
cote plafond: Co mm cote gxiide : 2o mm
cote suide : 2 o mm cote guide : 2o mm
?HOT0 l'i rOTu i;
clianrrngo ") m i n . I I " d . C . i ' r ('; c h a u f.!N a g o "" t u t i\ , 11[ (KC. S a s # P r 6 c h a u Ca g o

iiiic.rt;~or\(.les i min* Nlcroondet I. t.LJ'l .
u l e r o mides .1 m i n . '} O .i c *
coto plafond 'i o mu c o t c; y 1 a t * (j a d i mm
;oto plafond ' o inni
c a t e L'iio o min C o t o L'IJ.JO ) mm
; o t o g u i. d o ; o mm
DISCUSSION
A considerable proportion of the discussion following this paper was

concerned with reasons for local burning on the sheets, samples of
which were brought to the meeting. Emphasis was also placed on the
need to determine the microwave properties of the materials, and
interest was shown in embossing techniques.
Question : What is meant by "blowing" ?
Answer An additive, when subjected to heat, decomposes to nitrogen,

forming bubbles within the material.
Question : How is embossing done ?
Answer An inhibitor is added to the sheet.

By adding zinc oxide, the local temperature decreases from
210C to 125C. When an inhibitor is printed on, no foaming
occurs locally.
Question : It is possible that the thickness of the sheet and choice

of PVC has an effect on the performance. Has this been
tested ?
Answer No.
Question : It is possible that the microwave properties are poor and

would always give problems. Do you know the properties
of the materials you want to treat ?
Answer Not all of them.
-386-
MICROWAVE PROPERTIES OF SOME BUILDING MATERIALS IN VIEW OF AN OVEN DESIGN
Authors M. De Pourcq, C. De Wagter, W. Van Loock
Contract number EE-B-1-135-B
Duration 1 January 1982 - 30 june 1983
Head of project Prof. J. Van Bladel, Laboratory of Electromagnetism

and Acoustics
Contractor State University of Ghent
Address Laboratory of Electromagnetism and Acoustics

St. Pietersnieuwstraat 41
B9000 Gent, Belgium
Summary
Permittivity measuring systems are presently being improved. The

slotted line system for sheet materials and the Von Hippel method
for bulk materials are briefly outlined. For maximum energy trans-
fer in microwave power applications a versatile computer simulation
has been developed. Measurement techniques for temperature distri-
butions within materials and at the surface are studied for process
control.
Results for cellular concrete and concrete are presented. Microwave
properties of some European woods are presented. A modular and
automatic microwave oven is being developed.
-387-
1.1 Introduction
The development of basic equipment and measuring systems was started
1979 under a National RD Energy Program contract on Microwave Power Ap
plications ' .
Successful microwave power application necessitates the full knowledge of
the microwave and thermal properties of the materials to be treated. To
investigate building materials the measuring systems were improved.
1.2 Permittivity measurement

1.2.1 Bulk methods
Nelson's computer program for the Von Hippel method was adapted for
accurate determination of * for bulk material. The basics of the measu
rement procedures are briefly given.
The voltage standing wave is determined of a section transmission line
which is at the end shortcircuited. The voltage standing wave pattern
is again determined of the same section but the short circuited end is
filled with the material under investigation, fig. 1. The difference
in pattern results in a transcendental equation from which the permittivi
ty is calculated. A transcendental equation has an infinite number of
solutions. In many cases a second measurement is needed with a different
sample length or the twoimpedance method is used. Combinations of these
methods give accurate results.
1.2.2 Slotted line systems

An automatic system for the measurement of the permittivity of sheet
materials has been developed.
First, the system was limited to narrow slots of 0.5 mm in Sband wave
guide. Slots of 5 mm have now been used to study veneers.
The propagation constant of the slab loaded waveguide section is
measured with an automatic network analyser, see fig. 2. Two measurement
modes are operative : a continuous modes and a discontinuous mode. In
the first method is determinated from the transmission coefficient
of the empty and loaded section. Reflections are neglected.
In the discontinuous mode Sparameters are measured for two sample
lengths.
The discontinuities at the edges of the sample are taken into account.
The Sparameters are measured with an accuracy of 0.2 dB on the amplitude
and Io on the phase.
In the continuous mode the relative enor on * is about 1% for thin
materials. In the discontinuous mode, however,the enor on * is larger
and attains 10%.
1.3 Computer simulation and measurement techniques for temperature distri

butions in layered media due to microwave absorption
Many building materials are multilayered structures. Microwave
power for heating or drying must result in uniform energy and temperature
distributions within the materials. To predict the distributions in a
particular application a computer program was developed.
The laminated structure is supposed to be illuminated by a plane wave with
normal incidence. A lso the related problem of bilateral irradiation
by coherent waves can be handled. The mathematical model takes into
account the two differential equations of electromagnetic waves and heat
transfer. For the numerical procedure, the method of finite differences
is used. Discretization of the continuous problem is accomplished by di
viding each layer, wherein the dielectric and thermal properties are con
stant, into a number of transverse slabs. The model is a good approxima
tion provided the thickness of each slab is sufficiently small.
First the distribution of absorbed microwave power density is calculated.
The power absorbed per unit area in each slab is computed using the trans
mission line equivalent of the slab and equals the difference of power
fluxes at the two ports. The resulting cascade of lossy transmission
lines is described by the ABCDmatrix formalism of network theory.
Secondly, the temperature profile, caused by absorbed microwave power and
eventual boundary sources, is calculated in the time domain. A bove
mentioned discretization process transforms the partial differential
equation, describing the onedimensional flow of heat in a bounded
region, into a system of coupled differential equations. The vector,
containing the temperature of each slab, is expanded as a linear combina
tion of the eigenvectors of the matrix associated with the system of
coupled differential equations. The respective eigenvalues and eigen
vectors are easily computed because the matrix is real, tridiagonal and
quasisymmetric. The time dependent coefficients of the expansion are
a function of the initial conditions of the expansion are a function of
the initial conditions and of the volume and boundary sources. A pplying
the method in an iterative way, allows treatment of the nonlinear case,
arising when the dielectric properties are temperature dependent.
Figure 3 gives an example of a triplelayered material. The dielectric
properties of the second layer are temperature dependent. The power
density, incident from the left, is 3W/cm2 at a frequency of 2.45GHz.
The smoothing effect, by diffusion of heat, of a waiting time without
power supply is depicted in the same figure.
The simulation is manmachine interactive and runs on a VAXcomputer.
So, the method allows optimization of the dielectric properties of cer
tain layers. The effects of environment temperature, forced convection
at the boundary surfaces and a rest time without microwave supply are
readily examined. The method allows also prediction of the temperature
profile versus time for a given sequence of incident microwave power.
1.4 Results
1.4.1 Concrete
Permittivity measurements on cellular concrete revealed that this
inhomogeneous structure is almost isotropic for microwaves. The dielec
tric constant is 2.7, the loss factor 0.38 and the loss tangent 0.14,
for a moisture content of 50%.
Experiments with microwave heating for fast curing have been carried
out according to results with concrete. It was found that about
0.15kwh/kg microwave power can cure in normal atmospheric condition.
A multiple mould made of Perspex and allowing standard strength test
has been constructed.
1.4.2 European Woods

Veneers of European woods as beech, ash and pine, having a moisture
content ranging from 9% to 12%, have been investigated. Bulk measure
ments were carried out using the Von Hippel method for comparison. The
results are summarised in table I. A s the reflection has been neglected,
the error on * increases with increasing sample thickness.
389
Veneer (Continuous mode) Blocks (V.Hippel)
Dir. of Mois
Sample thick fibres ture . ' " ' "
ness against content
(mm) ELfield (%)
Beech 0,9 J_ 9,4 1,97 0,18 2,39 0,24

0,9 _L 9.0 2,23 0,22 2,36 0,23
0,9 _L 10,0 2,09 0,24 2,43 0,26
0,9 _L 10,9 2,23 0,27 2,49 0,30
0,9 J 10,9 2,38 0,30 2,50 0,30
Ash 0,6 // 10,0 3,00 0,51 3,18 0,68

0,6 II 9,5 3,02 0,50 3,10 0,62
0,6 ii 9,2 2,99 0,50 3,05 0,59
0,6 II 8,8 3,11 0,54 2,99 0,54
European 0,7 il 12,2 3,12 0,75 3,10 0,71

pine 0,7 3,24 0,76
ii 11,4 2,97 0,63
0,7 li 12,0 3,01 0,68 3,07 0,69
0,7 11,3 3,22 0,74 2,95 0,62
II
TA.BLE I
Measurements of * of some European veneers at 3GHz
Results with E. P e y s k e n s , L a b . voor Biologie en T e c h n o l o g i e
van het hout .
1.4.3 Computer simulation and temperature measurements
For optimization of microwave power applications and verification of
theoretical models, it's necessary to measure temperature. A n infrared
thermometer is used for measuring surface temperatures. Figure 4 shows
such a recorded temperature profile after 60 seconds of illumination of
a triplelayered materialconsisting of eterspan^; woodfiber and rubber
by a horn antenna with 6W/cm in the center and at 2.45GHz. The structure
was previously transversely intersected to measure the surface temperature
of the crosssection. The Ultrakust type M202 infrared thermometer was
used. The discrepancy with the computer simulation is partially due to
the large resolution surface of the thermometer probe. On the other
hand, the thermal properties were not exactly known. For realtime
measuring of internal temperatures during microwave power supply,
NiCrNi thermocouples are used. The joint has a volume of only 0.2mm ,
while the wires have a diameter of 2 0.2mm. Since thermocouples are
sensitive to microwaves, they must be placed normally to the electric
field where possible, in order to avoid internal dissipation by Joule
losses. Only then, the probes are nonperturbing. In case of wrong
aligning, before temperature measurement power must be cut off during a
fixed rest time. A n automatic measuring setup is built around the IEEE
488 Bus. A tenchannel scanner connects sequentially the thermocouples
390
with a digital voltmeter. Each thermocouple is compensated by a cold joint
placed in smelting ice. In order to suppress disturbations it was necessa-
ry to extend the integration time of the digital voltmeter and to inter-
connect a low pass filter. Hence a compromise between accuracy and
measuring time is needed. A temperature measurement with a resolution of
0.1C takes about 1 second, setting time of the scanner and time needed
for data transfer included.
Fig. 5 is an example of a temperatures versus time measured by ten thermo-
couples during intermittent power supply.
1.5 Design and automatisation of microwave applicators

A microwave oven is being designed for drying. The concept is modu-
lar. Starting with the actual dimensions one module is optimized by
changing the dimensions of the loaded cavity. A special display is used
to monitor the field pattern and to show the temperature distribution.
The radiator is a slot antenna. The slots are optimized to have an even
temperature pattern. Stubs are used to adapt the antenna.
An automatic network analyser is used to optimise the oven.
Filter structures are being designed to prevent leakage of microwave powers
and to uncouple the oven modules.
A microprocessor and adapted interfaces are also being developed to control
the oven.
1.6 Conclusion
Some microwave properties have been investigated. The computer simu-
lation and temperature measurement technique is a powerful tool for
optmising microwave heating applications. The microwave oven is being
designed and we hope to report results in the near future.
REFERENCES
1
. "TTL controlled magnetron operation", M. De Pourcq, C. De Wagter,
W. Van Loock, Microwave Power Symposium, Toronto 1981, pp. 126-128.
. "Microwave heating of laminated materials", C. De Wagter, M. De Pourcq,
W. Van Loock, Microwave Power Symposium, Toronto 1981, pp. 225-227.
. "Computer Program for calculating dielectric properties of low a
high-loss materials from short-circuited waveguide measurements",
U.S. Department of Agriculture, ARS-NC-4, 1972.
CEx(z)]
max
KSL
Ejz)
Meting met lucht
'
max
^ KSL
\J ^s vy W \
min

. /2
d' "
L ?0 fel
. d
'
na Ag/2 = de "
a.
/
Meting met dielektricum
Fig. 1 VON HIPPEL method
PLOTTER
POP 11/34 TERMINAL
PRINTER
-t <r
STORAGE
IEEE 488 BUS

< ^ >
Fig. 2 Setup for thin
material
COUNTER GEN AD
HP apnr. CONVERTER
ETP J71
fas
NETWORK
ANALYSER ampi
392-
A: S miautts pou**-
no. -,
/fft /
, J Jo 3
3: 3/Ztt*ULti<lf>*K* *f**-
'A'
Fig. 3 Temperature p r o f i l e for the n o n l i n e a r case
I _
X
UITMKISTnzoi-^j
H-ij
I
&
j
dC
-3>&
-c
f.
Ti E C
Fig. 4 Measuring set up and

example of a recorded
temperature profile
393
Temp. C
50.-
flmocodpfe.
17. ti/X.e.ffCC*4,
Fig. 5 Recorded temperatures

DISCUSSION
Host of the discussion dealt with the measuring techniques outlined

by Mr. van Loock. It was also pointed out that wood drying, one of
the applications being studied at Gent, could be done with a high
energy efficiency using a heat pump dehumidifier (static dryer).
Question - Chairman : Could the infrared temperature measuring system

be used on a moving system ?
Answer You need a time constant, and have to be careful not to

loose information.
Question : Will the probes (temperature measuring) interfere with the

microwave field ?
Answer In the USA probes are sold which do not disturb an

electromagnetic field. (See also the paper - orientation
is important).
Question : An error of 10 per cent is quoted.

Do thinner sheets give a greater error ?
Answer The precision mode of operation takes a few minutes.

A continuous mode of operation would be useful for moving
sheets, but it would not be very accurate.
Question : What is the accuracy of measurement on static sheets ?
Answer It is difficult to say. Sheet material has an effect.
-395-
DEVELOPMENT OF A MICROWAVE APPLICATOR TO DRY THIN STRUCTURES.
FIRST SEMI-ANNUAL REPORT
Authors J.P. LUYPAERT, P. REUSENS, H. LEMMENS
Contract number EEB-1-137-B
Duration 01.01.1981 - 31.12.83.
Head of Project Prof. P.J. LUYPAERT
Contractor Katholieke Universiteit Leuven,

Dept. Elektrotechniek M.I.L.
Address : Kardinaal Mercierlaan, 94

B-3030 HEVERLEE - LEUVEN
Summary
This paper contains a review of the research made at the Division

Microwaves and Lasers of the Katholieke Universiteit Leuven in the
scope of the EC Energy Conservation R&D Programme, contract number
EEB-1-137-B. Results of the first six months of the contract period
are included.
In a first chapter a general outline of the research has been
given.
Chapter 2 describes an apparatus for the measurement of a re-
flection coefficient in a waveguide. This apparatus will be used
for matching the load with respect to the magnetron in operation.
The third chapter presents some results of the theoretical
study of slots in waveguides. Arrays of slots will be used for the
homogeneous irradiation of thin structures in a microwave applicator.
-396-
1. General outline
In the scope of the contract of the Laboratory of Microwaves of the
Katholieke Universiteit Leuven with the EEC, research is being carried
out concerning the development of microwave applicators to dry thin
structures. It considers three aspects :
1 General Study of microwave power systems : the magnetron, waveguides
with stubtuner, the transport of the microwave power from the waveguide
to the oven by means of antennas or slots, multimode cavities, ...
2 Analysis and improvement of the efficiency of the system by means of
recuperation of the magnetron heat, amelioration and scaling up of the
highvoltage generation and a system to measure the SWR in a waveguide
fed by a magnetron in operation.
3 Amelioration of the systems potentialities and facilitation of use and
control by means of a regular microwave source and an absolute power
measurement system of a magnetron inserted at the load.
During these first six months of the contract period the following
subj ects have been performed :
development of a reflection coefficientmeasurement apparatus,
theoretical study of slots in waveguides,
experiments on several products as to their microwave properties,
installation of a 12kW applicator.
2. Reflection coefficient meter for dynamic matching of an "applicator

magnetron" system.
2.1. Introduction
Matching of magnetrons is usually arranged by using an equivalent probe,
which replaces the magnetron. However, this approach has two main dis
advantages : first, the very expensive equipment set that goes along
with it and, second, matching can only be obtained for initial working
conditions (and not during operation).
For these reasons we looked out for a relatively cheaper micropro
cessorbased system.
2.2. System description

In order to introduce the system to the reader, we briefly describe the
features of the microprocessor based system.
F
ig 1. gives a block diagram of the system. A generator sends its
microwave power through a waveguide to an unknown load. Due to a mis
match a standing wave pattern will occur in the waveguide. The signals
detected by the three probes will enable the microprocessor to evaluate
the reflection coefficient, the reflected power, the forward power, the
average voltage of diode 3, 2 or 1.
2.3. Theoretical aspects

Three detection probes are spaced apart at distances of one eighth of a
wave length in the waveguide. A ssume the following situation: fig.2
Starting from v. = = v + + v~ = v + (1 + )
+
. ^ 3 1 (1 + r e _ 2 j s l )
+ jS 2js
v 3 = v e 2 (1 + r e 2)
397
and after some calculations, one can find that
2 S v 4)2-*'
4s3 / ( s3'
|r|x|r|+l=0 Or ITI = ;;
S
3
with the restriction
0 < < 1
= I eJa
(vl) 2 + (v3) 2
whereas sO
S3 = (si 2 s 2 2 ) 0 ' 5
_ (v2) 2 _ sO
S1
2 2
(vl) 2 (v3) 2
s2 =
2.4. Practical aspects

The probes
The probe forms the inner conductor of a coaxial structure.
Accordingly to their specifications, the diodes will derive a
dcsignal related, by those spec's, to the field. This dcsignal will
be treated properly by the microprocessor.
In order to eliminate high frequency leakage to the microprocessor a
simple coaxial LPfilter has been developed based upon the theory of
quarterwave impedance transformers.
Before measuring
Although the three diodes used are matched, a different behaviour
cannot completely be prohibited. To receive for either diode equal vol
tage output corresponding to a fixed field, the following test can de
crease the error made : (fig. 3) :
If the generator can dissipate his entire energy in the matched load,
there will be no reflection. The three probes are fixed together in a
small box of aluminium. While looking for the same readout, tuning of
each probe at the same place will be arranged by changing the penetration
depth of the probe into the waveguide.
During measurements
Frequency shifts between 2,44 and 2,46 GHz only affect the measu
rements for high reflection coefficient values. Depending on the phase
'a' of in the reference plane and the bandwidth of the frequency shift
the error made will vary between 0,1 % and 10 % if remains smaller
than 0.4.
About the same deviation error is expected, caused by the errors of
detection of the standing wave pattern.
398
2.5. Technical aspects
The microwave power meter consists of four blocks (Fig.4) :
- the 6802 microprocessor with clock and 128 bytes RAM memory,
- the monitor program which is stored in the EPROM (2716),
- the input unit,
- the displays (output unit).
The meter works fully automatically . The start is given by a power
up. At that moment the microprocessor finds the beginning of the program
at the addresses FFFE and FFFF. This program defines the PA's and PB's
of the PIA 6821 as output except for the PA 7 which is input. The
measuring period consists of 1062 samples of each diode. This important
amount of samples must and does guarantee a good average of the D.C. vol-
tage of the diode. With those averages the calculations are carried out.
During all this time (sampling and calculating) the displays are set
blank so the user can easily make the difference between two measurements.
The total measurement and calculation time is 245 msec. The results of
the calculation are visible on four 7 segments displays, so we have four
digits after the comma for the reflection coefficient. The fifth display
is used to indicate the output mode. There are six possible modes :
- mode 6 : the reflection coefficient
- mode 5 : the reflected power
- mode 4 : the forward power
- mode 3 : the average of diode 3
- mode 2 : the average of diode 2
- mode 1 : the average of diode 1.
On power up the meter stands automatically in mode 6. The changes of
the modes are effected by means of a switch which is coupled to the PA7
of the PIA. The diodes are placed on the waveguide so that the D.C. max.
voltage of each diode is IV. This analogue value is coupled to the bit
AD convertor 7574. The 8 bits give 256 steps between 0 V and 1 V. As
there is only one AD convertor we have to use an electronic switch to
connect the 3 diodes to the AD. This construction could mean a serious
time loss, as the diodes have to wait after each other, but the monitor
program was built up in a way that during a conversion this time is used
to execute some program steps.
The important chips such as the PIA, the AD convertor and the EPROM
are mastered by the microprocessor. To generate the chip selects we
used a decoder (74 LS 13'8).
The enable line and the VMA line together with the three highest
address lines are the input lines for the decoder. The addresses are :
E000 for the EPROM
C000 for the AD convertor
8000 for the PIA.
With the enable line we manage to distribute the 1 MHz working
frequency of the microprocessor so that there is no need for buffers
on the address and data bus.
3. Theoretical study of slots in waveguides
3.1. Actually used microwave power transport system

In our actually existing microwave applicators power is being transported
from the magnetron to the oven as shown of Fig. 5. The magnetron (1)
cooled by a ventilator (2) emits the power via a waveguide (3) into the
oven (4).
-399-
Matching from the oven with respect to the magnetron is possible by means
of a triple stub tuner (5). Small antennas (6) transport the microwaves
from the waveguide into the oven cavity.
The scope, position and dimensions of these antennas were determined
earlier experimentally in order to achieve frequency constant properties
in a small frequency band around 2.45 GHz.
The radiation pattern of this antenna is similar to that of a short
wire antenna near a conducting surface, i.e. with the mean lob parallel
to the oven wall. This implies some disadvantages :
- important amount of wall losses
- coupling of the different magnetrons
- impredictible field pattern in the oven
- important SWR due to the reflection at the side walls.
3.2. Theoretical description and sensitivity analysis

The impedance properties of small radiating slots in the broad side
of a rectangular waveguide were investigated by A.A. OLINER (IRE, AP,
January 1957). Formulas are deduced for the equivalent transmission
line network of a slot. These values are shown to be in good accordance
with measurement results.
The equivalent transmission line network of a transversal slot is
shown in Fig. 6. The slot is shown to be equivalent to a series
impedance _.
= (G + JjB )
s s s
where G and B resp. give the influence of the radiated, resp. stored
energy.
For extensive use of the slots in microwave applicators, a frequency
and tolerance sensitivity study is needed.
The formulas given by OLINER show that the frequency sensitivity is
small for wide slots.
The influence of the slotlength is discussed on Tabel I :
- The frequency sensitivity of the phase of Z s is the largest for
'resonating slots', i.e. slots with real Z s .
- The frequency sensitivity decreases for slots whose length differs
more from the resonating length.
According to Tabel I. the slots with length 40 mm and 70 mm are
preferable to those of 55 mm. Shorter slots are used for long arrays,
the longer slots are preferable for an array with only a few radiators.
The following subjects are still to be studied :
the effect of the rounding of the slotends
the optimal position of the slots to achieve a proposed radiation
pattern.
Generator . . . J A a. Unknown
(Magnetron! Load
Probes plus
Detection circuit

System Description
fig. 1
V +
v
-
toward toward
load generator
o
Transmission Line Model w i t h : s < = -4/o

5 =
2
fig. 2
f1 1 1
Matched
Generator
Load
fig. 3
DATA ^>
DISPLAY
CONTROL
MICRO P R O C E S S O R
El
INPUT

\

D2 ADC
SWITCH EPROM
D3
Fig. 4 Reflection coefficient Meter: Block D iagram
Fig. 5 Microwave Transport System
frT
Bs
Fig. 6 Equivalent Network Slot
-402-
TABLE I
Frequency sensitivity of the series impedance of a transverse slot.
zg = |zsl exp (j s) = 2 J 2 2
gs + bs g + vb
s s
b Y = characteristic
s= f admittance
Slotwidth = 10 mm
Waveguide : 86 43 mm.
L (mm) 40 45 50 55 60 65 70
|z 1 at 2.45 GHz 0,1997 0,3359 0,6327 1,1624 0,9484 0,6511 0,5104

1
s'
Freq. sensitivity (%)

. - '
2.47 2.43. 6,8 8,6 11 4 9 6,8 5,1
.s
2.45
at 2.45 GHz 82 75 60 16 + 36 +55 +62
s
Freq. sens. ()
I
,
j 1 *s(2.47) s(2.43)| 0,7 1,3 4,5 11 6 4 1,1
I
I
403
DISCUSSION
A number of observations were made. Mr. van Loock stated that his
research team had made measurements on 10 types of resins, and most of
them were found to be transparent to microwaves. The Chairman pointed
out that resin and its solvent couLd create severe fouling problems.
Question : Microwave equipment has a high capital cost. Have you

assessed this ?
Answer We estimate BF 80,000 (1800 ECU) per kW installed.

The problem is not so much with the capital cost of the
equipment, but with the maintenance of the magnetrons.
Question : What is the ratio of electricity to gas cost in Belgium ?
Answer We have calculated energy saving in prevulcanising of rubber.

A very significant reduction in energy consumption would
occur using microwaves because rubber is a poor heat
conductor.
-404-
SESSION VIII - ENERGY MANAGEMENT
Summary of Session VIII
An assessment of the energy needs of a heavy electrical

products factory with a view to their reduction
Economic study of the energy exchange between factories in

an industrial site
-405-
SUMMARY OF SESSION VIII
ENERGY MANAGEMENT
J. Carrasse, P.A. Pilavachi
In the energy-saving programme, the industrial sector is one of conside-

rable potential and a good number of R 8 D contracts have been concluded
in industries with high energy inputs such as iron and steel, cement,
ceramics, etc.
Besides this work on specific industrial processes, considerable attention

must be paid to a comprehensive analysis of energy needs and energy use in
complex factories or at complex industrial sites.
Two contracts are running in this sector. ACEC is analysing the energy
needs of a heavy electrical products factory consisting of more than 40
different buildings. S.C.K.-C.E.N, is undertaking an economic study of
the energy exchange between the ten factories of the industrial estate
of Mol-Donk.
Electromechanical factories - ACEC premises in Charleroi consist of 43

buildings spread over 2 km2; a major part consists of workshops of
various ages, designs and dimensions, housing different activities.
A comprehensive programme of action has been formulated as a means of

reducing expenditure on heating and on current consumption. This
composes six items :
1. General purpose actions, such as roof insulation, insulation of hot-

water piping, double-glazing of windowframes, installation of clock
and thermostat controls resulted in a continuous reduction of fuel
consumption. Combined maintenance and energy-saving actions are to be
continued.
2. Analysis of a particular heating system suggested more general conclu-

sions such as : avoidance of excessively complicated automatic heating
systems; establishment of standards for energy consumption for each
building and for monitoring purposes; formulation of maintenance
schedules ...
3. Development of a new micro-computer which controls the entire heating

system for the building. Input information includes outside tempera-
ture, wind and sun conditions, actual inside temperature, heat loss
factor for the building, working days and week-end days. Fuel savings
amounted to 15 to 30 %; in the near future, other buildings will be
provided with the same equipment.
-406-
4. Use of a computer program to calculate the heat losses of buildings of
any kind, taking into account the structural characteristics of the
building components, rate of change of air, exposure, degrees-days.
More accurate results are expected from a second, more complex version
of the program, which would include a better assessment of the rate of
air renewal and additional input data such as thermal inertia and
unsteady conditions.
5. Reduction of expenditure on electric current by : more detailed

investigation of the total peak demands, with a view to possible
evening-out; shifting daytime elements of the consumptions to night-
time; identifying and eliminating all unnecessary consumption,
particularly during peak time (e.g. lights to be extinguished, step-
down transformers not left in circuit).
6. Calculations in respect of the central boiler and associated hot-water

piping : In such complex factories it is important to know whether
transport of heat through an existing piping system to more or less
distant users causes an acceptable amount of energy waste or not.
Energy savings in some critical sectors of the factory may result in
lower overall efficiencies for the internal district heating systems.
DI2^_S2D2_D_i2liSS_2D_5!]_iny5li2i_Si - It is thought
that the use of practical models of the energy needs and uses at a site,
could show up considerable saving potentials for existing industrial sites.
These models could also make it possible to predict the types of industry
which have to be associated on a site to ensure optimum energy use.
S.C.K.-C.E.N, is developing such a model for the economic assessment of the

energy exchanges between the ten factories located at the Mol site.
The first step is to collect and classify energy data and to evaluate the
possible inter-relations between factories. A mathemacital model simulating
the industrial site consits of a network in which the nodes are representing
the energy sources or sinks for each factory and the links between the
nodes are the energy carriers for the possible interrelations between the
factories.
As a first step, S.C.K.-C.E.N, is undertaking a comprehensive survey on the

energy production possibilities as well as on energies that are wasted or
available in the different factories.
The collection of data is nearly finished and the computer input operation
is going on. Economic factors, costs and tarification factors will be
taken into account. Other criteria, such as reciprocal dependence between
two or more factories may be critical.
AN ASSESSMENT OF THE ENERGY NEEDS OF A HEAVY ELECTRICAL
PRODUCTS FACTORY WITH A VIEW TO THEIR RED UCTION
Author P. BERGER, Ingnieur Civil
Contract number EE-B-3-170-B
Duration 1st July 1980 - 30 June 1983
Head of project Dr Ir PH. KIRSCH
Contractor .C.E.C.
("Ateliers de Constructions Electriques de
Charleroi").
Address A.C.E.C.
.P. I*
6000 CHARLEROI
Summary
This paper outlines some problems encountered in a heavy electrical

products factory consisting of 43 different buildings, with regard
to heating and electricity consumption.
It deals with savings already gained or expected from general
purpose actions, investigations of a somewhat particular heating
system, development of a new micro-computer heating control device,
use of a program computation of the heat losses of buildings of
any kind.
Some steps to be taken in order to reduce the amount of expenses
for electricity consumption are also considered.
-408-
1 . Introduction
A.C.E.C. premises in Charleroi consist of 43 buildings spread over
2 sq. km approximately. Among them, there are three main and a few smal-
ler office buildings, the rest i.e. the major part of the built area being
workshops of various ages, shapes, designs and dimensions. Each of those
workshops is devoted to a specific kind of activity.
During the year 1981, 60 millions belgian francs have been spent to
heat the buildings, while the electricity expenses for lighting and for
running the tools ( in the most general sense) amounted to 140 millions
belgian francs.
As far as heating is concerned, 25 buildings are provided with their
own boiler or heating system, the remaining ones - and among them the
largest workshops - are provided with an array of heat exchangers con-
nected to an internal district heating system. This system consists of
a so called "central" boiler plant and of a hot water piping network.
Illustration : fig. 1.
2. General purpose actions

The maintenance services of ACEC have been energy saving minded since
a number of years, and some of their actions have been accordingly taken
with twofold purposes, i.e. repairing and insulating for instance.
Full detailed records of such actions would be out of place here,
just some of the most typical taken from 1979 to 1981 are mentioned :
Roofs - New insulation of :

2000 m2 on building 01 350 m2 on building 43
900 m2 on building 19 3000 m2 on building 42
550 m2 on building 23
Controls - New clock anfl thermostat controls of hot air blowers in a num-
ber of workshops.
Lighting - Replacement of old appliances by new more economical Sodium
types in the largest workshop halls.
Window frames - 120 old pieces replaced by double glazing types,
(buildings 23, 25, 26, 38).
Hot water pipings (of the district heating system) : new insulation on
2500 meters.
Plastic curtains placed on main large entrances of buildings 01 and 23
to control air drafts.
So far, energy savings resulting from the above mentioned actions
have not been checked on an individual basis, but it appears anyway on a
general level <&hat, for the whole plant, the average fuel consumptions
related to the degrees days have been decreasing over the last 3 years.
See fig. 2.
Conclusions :
1. Combined maintenance energy saving actions are to be continued in

the future.
2. Most advantageous energy saving actions are to be selected in the
future with the help of an infrared thermography equipment.
3. Two cases of improving heating systems
3.1. Case of the heating system of the main management building.
This 10 years old building is of a modern design as far as outside

appearance (all steel and glass structures) is concerned, but it has a
not too good heat transmission factor, = 3 W / m2, C.
It is provided with an all electric heating and air conditioning
system of a somewhat particular design (fig. 3.1.1.). Basically, heat
supplied by the electric network is accumulated in a battery of ACCUBLOC
units which in turn transfer the heat to a hot water tank connected at
will to the evaporator and to the condenser of a heat pump which can be
used either as a cooling or as a heating unit. Water to air heat ex
changers are provided to distribute either hot or cool air as required
in the building.
The functioning of this system had been satisfactory during the first
5/6 years of operation, and a standard of the electry consumption was thus
established for those years. But, as can be seen on fig. 3.1.2. an alar
ming increase of consumption appeared, namely for the year 1979. The case
was thoroughly investigated during the year 1981 and the following
misfunctionings were discovered :
1. Heat exchangers of the ACCUBLOCS dirty or defective; 50 % to be
replaced.
2. Some 3ways control valves untight with, as a consequence :
3. Unnecessary production of heat "corrected" by production of cold.
4. Unnecessary operation of air cooling units during cold season.
5. Consumption of electricity by ACCUBLOCS during night and day, instead
of during the night only.
Last but not least, it was realized that operating a heat pump on a
not free cold source is to be considered as a non economical functioning.
Accordingly, after defective equipments and control mis funetioning
had been duly repaired, heating is now supplied by the ACCUBLOC batteries
only without any additional heat supply from the heat pump.
The diagram of fig. 3.1.2. shows a satisfactory restore of the
standard energy consumption since the 6 last months of 1981. It must be
mentioned that improvements also result from a new lighting control system.
Conclusions :
1. Avoid too complicate ( unnecessarily so, at any rate) heating automatic

systems.
410
2. Avoid the use of a heat pump unless the heat from the cold source is
actually free.
3. Establish a standard of energy consumption for each building and carry
out a follow-up (this standard to be considered as a guideline).
4. Establish a maintenance schedule and check that it is carried out.
3.2. ENERGACEC.
ENERGACEC is a micro-computer control system designed and manufac-

tured by ACEC since 1981. It takes in charge the full command of the
heating of a building, taking into account all relevant data in order
to minimize the fuel consumption.
The following input informations are treated to this end by the
ENERGACEC system :
. outside temperature
. wind conditions
. sun conditions on main sides of building
. inside actual temperature
. working day or week-end day
. heat losses factor of the building
Such an equipment has been installed and tested on one of the main
offices building (nr 39) since the end of 1981.
In order to assess the actual savings of fuel resulting from the use
of the ENERGACEC, the consumed quantities in the nr 39 building have been
recorded for two reference periods, one before and one after the instal-
lation of the equipment. For comparison purposes, the same has been
done for another building, nr 38, similar in design and in dimensions
to the nr 39 building, but not equipped at the time with an ENERGACEC.
With due reference to the degrees-days of the comparison periods,
this method has positively shown that savings of fuel in building
nr 39 since the ENERGACEC has been in operation amounted to 15 to 30 %
according to conditions. This is illustrated by the diagram nr 3.2.1.
Conclusion :
The nr 38 building has been in turn provided with an ENERGACEC equipment,

and other buildings will also be provided in the near future.
4. AUDIT program.
AUDIT, a computer program developed since 1981 on behalf of ACEC,

aims at calculating the heat consumption of a given building over a
heating season. It is based on considerations in the line of the contents
of the DIN standard 4701. The computation makes use of the following
material :
. structure characteristics, i.e. surfaces and heat transfer coefficients
of the building components (walls, windows, doors, roofs, etc..)
. rate of airing
. orientation of the walls and roofs
. degrees-days.
A first version of the program relies on the assumption of steady
conditions. Although this first version is by now usable, it must be
mentioned that the assessment of the rate of air renewal is a guesswork
problem for most buildings; since the computed results are dependent of
this rate , they must be treated with due carefulness.
The fig. 4 shows as an example the printed output of the case
of a large dimensions workshop (building nr 1).
In the future, a second more complex version of the program is
intended, which will take into account additional input data, namely the
thermal inertia of the building and unsteady conditions.
Conclusion :
Although the results of the first version of the AUDIT program are not
to be neglected, more accurate results are to be expected from the more
sophisticated, more realistic approach of the second version.
5. An approach to restriction of electric energy expenses.
Owing to the tariff structure applying to high voltage supply, every

consumer is to some extent responsible for the average price of the kWh
charged in his monthly invoices.
As a rule, the total amount charged every month results from a
formula of the type ;
Amount = A.kW + B, kWh + B kWhK,
1 D 2
in which A, and B. are constants (more or less so, owing to variation
of various economic parameters)
kW denotes the maximum power (peak) occurred during the month
kWh denotes the total energy consumed during day hours
kWh denotes the total energy consumed during night hours.
For consumers of the size of ACEC, the consumed power is recorded
and averaged over every quarter of hour, and the daily or monthly consump
tion diagram can thus be drawn from these records.
As a guideline to reduction of the peak it is important to notice
the value of the monthlypeak and the moment when it most frequently
occurs. Moreover, for comparison and for speculation purposes, the
"monotonous" monthly diagram, i.e. the powers classified in decreasing
order, appears to be a more convenient tool than the regular common
power/time diagram.
Possible actions aiming to a reduction of expenses either with or
without reduction of the electric energy consumption are illustrated
in the accompanying sketches(fig. 5 ) , which are selfexplanatory.
The maximum monthly peak of power bears an important effect on the
amount of the electricity bill. Therefore, the knowledge gained so far
in ACEC in forecasting the likely value and time of the peak has resulted
in a weekly preventive procedure aiming at spreading out the individual
peaks of the largest workshops at different moments. This preventive
action results in savings estimated to some 4 % of the monthly electricity
expenses.
Conclusions :
Actions of the type described in fig. 5 are to be further promoted in the

future. More explicitly :
a. More detailed investigation of the total peak components, in view of
412
their possible spreading out.
b. Shift of day consumptions to night consumptions, in principle
wherever a permanent presence of manpower is not required.
c. Identifying and erasing all possible unnecessary consumptions,
particularly those occurring at the most likely time of the peak.
Some of such consumptions are obvious (lights on), some of them are
not (idle stepdown transformers for instance).
6. Central boiler plant and its associated hot water piping network.
The equipment of this plant consists of two dualfuel boilers of

7,5.106 kcal/h capacity each. Operated at their rated power they deliver
hot water in the pipings network with an average efficiency of 91/92 %.
But owing to losses occurring during transportation, only a part of
the heat leaving the boilers is actually delivered to the users.
Transportation heat losses are of two main kinds :
a. Losses due to heat conduction through the insulating material

(if any, which is not the case everywhere).
b. Losses occurring during shutdowns of the heating system, starting
from a hot condition, during weekends for instance.
The amount of the losses of the first kind can be represented by :
L = G . C, (U-r, - t/,,, ) (a) (for length L of pipe and per

I t r L . . .
time unit)
where G and C, represent the mass flow and the specific heat of the
transported water, and ( (7 - .) the temperature drop of the
water inside the piping over a length L.
As for the losses of the second kind, they are represented by :
L 2 = {% M.C
J ( P F o - &?t) (b) (for length L of pipe and
time lapse t)
where . M.C represents the sum of the masses of the water contents and
of the steel of the pipes, C their respective specific heats, and
( y . _ ) the temperature drop of the system at shutdown over some
lapse of time, starting at the beginning of shutdown.
An explicit view of the temperature differences (.t/v ~ Cpi) ana
( y ~ F ) is given bv fig. 6; bringing those differences into the
expressions (a) and (b) is to be the basis of a computation of the
losses of the hot water piping system over a specified lapse of time
including operation and shutdown periods.
Conclusion :
It is important to find an answer to the question whether transportation

of heat through an existing piping to more or less far away users does
or not cause an acceptable amount of wasted energy.
The necessary practical measurements of functioning parameters and
consequent computations will start this year as soon as the heating
system will be again in operation.
413
Fig 1
ACEC
Usines d e M a r c n e l le V
y ,\ &OlL3>iNS Caf'C
syyyy
CMT/t4 o/* PitNT
HORO
ô.fo'iwh
LJT FROID (\
K)M 3 /H \^J I
NOTE. FOUR LES 3 GROUPES
FRIGO I 1I.in.l_LS COM-
B lNAlSOkS FOSSB LES SONT
,80 KW / 16000 L V
fidati fo energy consomption decrease ACCUBLOC I B ACHE L .,
of 38 utlJgs as a reso/t "^T\ T'max 3fi'C /

ZONE EXTERNE
of central purpose acionS
Fig. 3.1.1 Circuit de production chaud/froid
Fig. 2
lo'iwh
fi/f / coosump t/ons o/
.to* ,**.
DU/icZ/ngS 38 an t/ 33
,frr t
(winter 81-X2J I Hm*/
7.*
1. /din q ' 3
(not provided witi, /t
TS.9
2. uilc/ing ne SS .St. i
Aim'/
(provi Jr d wit <3
HK6/tcec s/zje* /su-ffei)
fff. r
Cnerdo consumption (//purposes)
of huilc/ing r>' to i. /c/in q ~f

jtf.t
rf/.f
A
(js if not travideV wit/t dn
f*4G4cec since fsMtMr}
j r n A H J J / 1 s o - H x>
Fig. 3.1.2 Fig. 3.2.1

bijzonderheden
Fig. 3.2.2 Energacec apparatus
-417-
75:
m* IL M M C C=5ULTJ BUILDING Ci
vu i*Ein;> '*
oi n u l ' l i n i i it
UI L05SS (W/K) im*.
HH.L L ? j l S r t U * ' * >
F.jg>_ 4
il ' < * U J .H
aihi r t i B p - ~MH. _ ~ : * I T ~ T H W ' " SEPIE* "C'" idE~
.I'FSC/* i ___ sinet. ?. , '. 3 a ?" t . '; l'l'i si"_, **;_ uet. JO t. 304.
tosti's i?/**i_ t.g 1.2< .jz' l.i ili f,}; i,Jii ;i;g 24 i.274 l.ziv 1.9
t'jp rttei I.1S .M
t" HM < N i t l * e l
_ J; v
^ lt Hf. i l i o fe?* ''*t*_ IB ^' J e . ei?e ie.4 _ 1*.* 1. .
f.tUu-pt; (HMTtB tf s'i. ioifi. zsttS; 2 7 i s T ; ~" J s s i ; ~57 Sfti tj'f' iZ

B t' tDJT; ZS1Q19. JjSfl.
t i t l H C CMVL iiiiH/n<j ji.J '
iVontMu amount of ekctreitv expensa : . kW * ^.kW^+^.kW^
Reduction method ? 1
Initially : monthly a*i f3.3 HM
fiction peak reduction 1.3 MW

Results : } monthly eneryy : -.7i '/.
d monthly expenses : - 4.S7.
Reduction method " a
/nitjji'/y : SJ m e d'Jymm d o ye
Action : snift of .S ff y s//cr from
hours close to ind including th*
ptak time to other daily hours
Re S u/s monthly eneryy :o
monthly expenses : - f.S/.
Re doc Ion mei ho d />' 3
Initially : Ume diJynm <js oye
fiction : shift of .s H/ slice from

hours c/ose o and inc/vdiny the
ph ime to other hoars (nightj
/eScj/ts month/y energy : o
monh/y exfie ses : -1.8'/.
hours
Fig. 5
-418-
iTtmpirjtur*
A,L
(-0) / v > . ; e

vtside *f>*r* '* _
Q mj ss r/oW
I LtnstJ,
Cf: specif. f>* of fit/id
'.'-': '.'T/MSVLAT/VS .\">-.'.'..
FIOVJING riu

4T
'-.'()
r
ri
.-": ' :[:; _,
t* . ovtside net transfer
coefficient
ossss (,) = G.Cf. ( eFt)
/empenayre drop of Jto f/ow/no f/v/c/ oyer <f /engf/t

of in su /J fee/ i e
, Timpen tur*
k e .*,.'
fo-lj -V-4J
TT.L
4 fa
Ovtjlde^ _ *mertjr*_ ('\<
e-.Vmt I /eait, of bl\
: engl f>'f>6
A -..
^/i.c : /nasses of pipe
STf/U>y riu/ 4r &r
steel nd of inside fluiti
mu/tipJied y tAtir
Losses (s)=iZM.c){ar-o-0r) Speci fie heats
femperor* drop of <3 f>o S&t/y fJoicf inside /tyf/j L

of insulated 'fie over me /se t.
Fig. 6
-419-
DISCUSSION
Question - Dr. Roberts, Shirley Institute : In what way have you insulated
roofs and what payback periods do you find for these instal-
lations ?
Answer Host of the roof insulations carried out during the last
3 years in ACEC have consisted in spraying Polyurethane foams
on existing concrete vaults.
The payback time is estimated to 3 years, the breakdown of the
involved savings being 2/3 for reduction of heat losses and
1/3 for reduction of maintenance costs.
Question - Prsident
1. Mthode de simulation du chauffage du btiment 40 tenant
compte des caractristiques thermiques, y compris de l'inertie,
et des influences intrieures et extrieures alatoires :
intrt de la mise au point d'une mthode gnralisable.
2. Utilisation de l'Energacec : comment optimisez-vous l'intro-
duction des donnes propres au systme (Btiment + chauffage +
occupation + climat ..) que vous avez piloter ?
3. La chaufferie centrale (loin de ses lieux d'utilisation)

a-t-elle t remise en question dans sa gestion et mme dans
son existence ? Conclusions en tirer ? Consquences pour
l'exploitation de la chaufferie des conomies des changements
d'nergie raliss ?
4. Comment optimisez-vous diffrentes solutions d'amliorations
pouvant entrer en comptition suivant l'ordre de leur ralisation.
Exemple : Isolation et tanchit d'une toiture d'atelier,
compares amlioration du systme de chauffage (aerothermes,
panneaux rayonnants). Mthode gnrale ?
Rponse 1. Le programme de simulation permet d'valuer les conomies

d'nergie dans un btiment. Il intgre le phnomne d'inter-
mittence, et tient compte de diffrents scnarios de rgulation.
La modlisation du btiment prend en considration :
- les caractristiques des diffrentes couches de matriaux des
parois internes et externes
- les paramtres physiques de la construction dans son ensemble.
A partir de ce modle, on dduit les charges thermiques en
injectant les flux solaires et chaleurs sauvages dfinis par les
donnes climatiques locales.
Dans le cas d'application au btiment 40, certaines caract-
ristiques propres du btiment ont partiellement invalid les
valuations. Des campagnes de mesures dtailles ont t pra-
tiques sur le btiment de faon parfaire l'valuation
thorique. De plus, l'utilisation des valuations thoriques
s'est faite par comparaison relative entre diffrents scnarios
de rgulation (consommation 100% en situation avant analyse,
les diffrents scnarios donnant des rsultats rfrs ces
100%).
En conclusion, nous pensons la mthode thorique gnrale et
applicable n'importe quel type de btiment, mais nous
-420-
prconisons son emploi pour des valuations relatives et non
pour l'obtention de donnes absolues (bien que "l'talonnage"
de la mthode soit possible sur base de relevs exprimentaux).
2. L'Energacec
L'optimisation d'intermittence agit, tant sur la dtermination
de la dure de chauffage acclr avant occupation des locaux
que sur la dure d'anticipation la coupure du chauffage en fin
d'occupation.
En fait, par une connaissance sommaire de la zone considre
(inertie thermique des structures, coefficient de dperdition
volumique, taux de renouvellement d'air), il est possible de
prdterminer un rseau de courbes dfinissant la dure
d'anticipation de relance ou d'arrt en fonction de l'cart
entre la temprature extrieure et la consigne atteindre et
ce, en fonction du paramtre cart "Temprature de consigne
atteindre / temprature ambiante assure" ou de la chute
de temprature ambiante admise en fin d'occupation.
Si les courbes prdtermines s'avrent inexactes, le principe
autoadaptatif, assur par les mmoires du processeur, les
corrigera au fur et mesure des saisons.
En rsum l'optimisation d'intermittence est atteinte :
a. par l'introduction de quelques paramtres du btiment
b. par des mesures de temprature intrieure / extrieure
c. par comparaison des rsultats avec la prdtermination
du calculateur et autocorrection de celui-ci, en cas
d'cart.
Cette mthode implique videmment certaines prestations lors
des mises en services d'Energacec. Ces prestations tendent
se rduire avec l'exprience (constitution d'une banque de
donnes permettant d'tablir rapidement des similitudes avec
des btiments dj rencontrs).
En outre, un algorithme autoadaptatif total est actuellement
en cours de dveloppement.
3. La chaufferie dite centrale est mise en question dans le

principe mme du maintien de son existence. Le rglage des
chaudires installes (bi-combustibles) est automatis pour
ce qui concerne la combustion mais en revanche la distribution
de chaleur dans le rseau ne l'est pas.
Les investissements consentir simplement pour le maintien en
bon tat de marche des chaudires, d'une part, et pour
l'automatisation de la distribution de chaleur d'autre part,
sont lourds : de l'ordre de 36 millions de francs dans
l'immdiat, au moins.
De sorte que l'valuation des pertes de chaleur propres au
systme existant sera un lment dcisif de rponse la ques-
tion de son maintien ou de son remplacement par un systme de
chauffage dcentralis.
-421-
4. Notre point de vue est qu'il convient d'abord de raliser des
conomies bref dlai, en limitant les dperditions des bti-
ments, par exemple par des isolations de toiture, tant donn
que d'une part de telles transformations sont en gnral
ralisables dans des dlais assez courts, et que d'autre part
elles resteront profitables de toutes faons mme si ult-
rieurement on procde au remplacement du systme de chauffage
existant par un autre.
Il convient de mettre en balance les conomies susceptibles
de rsulter d'une amlioration d'isolation, et celles
attendre d'une modification du chauffage. Mais une telle
modification, tout au moins en ce qui concerne les grands
halls ateliers des ACEC, ne peut s'inscrire que dans une
stratgie plus longue porte, et implique des dlais de
ralisation nettement plus longs que de simples mesures de
renforcement d'isolation.
ECONOMIC STUD Y OF THE ENERGY EXCHANGE BETWEEN FACTORIES
IN AN IND USTRIAL SITE
Authors J.F. De Greef; A. Wouters

Contract number EE-B-3-171 (RS)
Duration 01.01.82 - 31.12.83.
Head of project . Wouters
Contractor C.E..-S.C..
Address Boeretang, 200
-2400 MOL
Belgium
Summary
A mathematical model is projected to predict on an economic basis

possible exchange of energy between factories of an industrial site.
Also the model is developed for predicting the type of new factories
to be installed in a given site.
-423-
Introduction
The aim of this study is to develop a mathematical model which

will be able to evaluate, for an industrial site grouping several but
different factories, the ability on an economical basis of inter-
changing energy between these factories. The purpose of such a
mathematical model is twofold. On one hand, the determination of the
economical interest of energy exchange between existing factories in
the industrial side; on the other hand, to predict the type of new
factories to be installed in a site for an optimal energy use.
To tackle the problem, several steps are considered. In a first
step, energy and economic parameters will be inventorized and classified
according to criteria of interest. A standard general purpose data
matrix will then be developed to be able to display for each factory
data in the same manner. By this way, factory energy and economic data
will be represented in the computer by individual files with an identical
structure. Also interrelations between factories will be represented
in the computer by means of a standard matrix including characteristics
of energy carrying and costs. The resulting mathematical model simulating
in this way the industrial site will be consequently a network where the
nodes are the factory data files and the links between nodes are the carry
data files of possible interrelations between factories. In a second step,
the mathematical model will be programmed on the computer and experienced
by means of the real data. Of course, the mathematical model has to be
kept as most general as possible. Therefore the model will be experienced
with the data collected from real industrial sites. Finally, in a third
step, the model will be applied for validation to an industrial site under
investigation.
The main characteristics of the factory energy data files will
include the different types of energy, such as electricity, oil, gas,
coal, steam and their amounts of installed power, consumption, waste,
balances and costs. The main characteristics of the energy carrying
files include different types of carrying such as cables, pipes and
their amounts of size, lengths and costs of installation between the
factories.
-424-
State of the work
This report describes the work performed during the f i r s t period

of the contract (01.01.1982-30.06.1982). The work consists in " f i e l d -
work" as well as computer preparation work. The following task, i . e .
the work to be done during the second period of h a l f a year, w i l l be
the "modlisation in general" by means of the information received from
the factories by the f i e l d - w o r k , and also already some cost-benefit
analyses of the computerised model with the help of the information of
two considered i n d u s t r i a l s i t e s . In order to make the work c l e a r l y
presented, i t w i l l be observed and described step by step.
Step one
As the responsible persons of the factories should be made sensible
to energy saving and thus to co-operation with the considered program,
a f i r s t v i s i t to the factories of the i n d u s t r i a l s i t e of Mol-Donk was
arranged.
The aim of the v i s i t was to explain the headlines of the research that
was going on f o r the program. Although i t was said to the people that
the study should not lead to the r e a l i z a t i o n of any interconnection bet-
ween the d i f f e r e n t factories of the s i t e , but "only" to a set-up of a
mathematical model permitting eventually the calculation of such an
interconnection f o r energy interchange, there was a general l i n e f o r
co-operation.
During the f i r s t v i s i t i t was t r i e d to have a clear i n s i g h t into energy
kind and quantity, used in the considered f a c t o r y . I t i s clear t h a t ,
where necessary, encountered waste of energy was shown, in order to reduce
such energy waste. As an example, in one of the v i s i t e d f a c t o r i e s , only
by changing the place of suction the a i r f o r the compressors from inside
to the outside of the b u i l d i n g s , i t was shown that a benefit of about
1,5 m i l l i o n of Belgian Francs per year could be r e a l i z e d .
Step tuo
Meanwhile an information inventory formulary was drafted, where
obtained information asked at the factories, was classified. The principal
idea of the formulary was to obtain the most convenient structure of the
data and carrying files of the model before programmation.
Analysis of the concept of the model is now going on.
-425-
ENERGY C O N S E R V A T I O N IN I N D U S T R Y -
APPLICATIONS AND TECHNIQUES
Proceedings of the Contractors' Meetings held in Brussels on
10 May, 17 June, 1, 7 and 22 October 1982
Edited by
H.EHRINGER.G. HOYAUX,and P. A. PILAVACHI
Directorate-General for Science, Research and Development,
In the framework of the European Community's second Energy

R&D Programme, 29 research projects on energy saving in
industrial applications are being carried out in the sub-programme,
Energy Conservation R&D. These projects have been subdivided
in eight contractor groups, which met in the course of 1982 to
discuss ongoing work. This book contains the papers presented
during these contractors' meetings together with their discussion,
subdivided into the following subjects: metallurgy; cement and
glass-ceramic industry; textile industry; food industry; chemical
industry; catalysis; microwaves; and energy management.


Energy Conservation in Industries. Applications and Techniques.

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Energy Conservation in Industries. Applications and Techniques.

Uploaded by

Copyright:

Available Formats

Energy R&D Programme

Energy Conservation in Industry -

D. Reidel Publishing Company

Energy Conservation in Industry

H. Ehringer, G. Hoyaux and P. A. Pilavachi

PARL EUROP. Biblioth.

Dordrecht / Boston / Lancaster

Organization of the Contractors' meetings by

Published by D. Reidel Publishing Company

Sold and distributed in the U.S.A . and Canada

In all other countries, sold and distributed

All Rights Reserved

The main objectives of the Community's Energy policy consist

In the framework of its strategy for scientific and technolo-

One of the most important short and medium term objectives of

In the European Community more than 35 % of the energy is

In order to enhance rational use of energy and energy conser-

The Commission has therefore concluded with industries, public

I should like to address my sincere thanks to all the authors

SESSION II : Cement and glass-ceramic industry

P. COROMPT CEA. GRENOBLE

R.B.M. HOLWEG T.N.O. FIBRE RESEARCH INST.

SESSION IV : Food Industry

L.W..H. DE VRIES TNO

SESSION V : Chemical Industry

M. SPOTO MONTEDIPE SPA.

J. DRAPIER UNIVERSITE DE LIEGE

SESSION VII : Microwaves

J.F. COURTOY EUROFLOOR

SESSION VIII : Energy Management

Dr. H. EHRINGER DG XII

Dr. J. CARRASSE expert SYNTEC-INGENIERIE FRANCE

Contractors meeting held on 22 October 1982

Economy storage in forged parts

Development of bainitic nodular iron for the construction of

Lubrication by slags of the continuous casting of steels

Economie d'nergie par mise en oeuvre de nouveaux cycles

Reduction in energy requirements for converting liquid aluminium

Composite metal jointing

Contractors meeting held on 1 October 1982

Summary of Session II 118

Etude par traceurs des paramtres de transfert du systme

Echangeur thermique cyclones, mise au point d'une mthode

Study of a treatment of oxygen steelmaking slag in view of its

Ceramic materials from molten blast-furnace slags by direct

SESSION III - TEXTILE INDUSTRY

Contractors meeting held on 1 October 1982

Summary of Session III 168

Industrial application of continuous wet-on-wet treatment of

Contractors meeting held on 17 June 1982

Summary of Session IV 198

Energy saving in the bakery by improvement of energy efficiency

Energy saving in the soybean extraction industry by reducing

Energy saving in edible oil processing plants by application of

SESSION V - CHEMICAL INDUSTRY

Contractors meeting held on 7 October 1982

Summary of Session V 238

Modification of cracking furnaces of existing plants to increase

Selective removal of HJ3 from gaseous mixtures containing C0 p

Energy conservation in the chlor-alkali industry

Thermal depolymerisation of waste tires by heavy oils

The recovery, treatment and utilization of solid bearing

Contractors meeting held on 10 May 1982