D Meier - Project Co-ordinator, BFH-Institute for Wood Chemistry, DE MI Crespo Rodriguez - University of Santiago de Compostela, SP AV Bridgwater - Aston University, UK P Nakos - Sapemus - ARI, DE-GR D di Blasi - Naples University, IT W Prins - Twente University, NL J Alonso - Union Fenosa, SP, M Samolada - CPERI, GR Contract JOR CT95-0081 PUBLISHABLE FINAL REPORT January 1996 to June 1998 Research funded in part by THE EUROPEAN COMMISSION in the framework of the Non Nuclear Energy Programme Joule III 1 Abstract The project covered five different research areas: 1. catalytic pyrolysis and improvement of pilot plants, 2. liquid modification, 3. incorporation of bio-oil into wood adhesives, 4. modelling, and 5. development of analytical methods. Work The main part of the work comprised the use and evaluation of catalysts in fast pyrolysis aimed at improving the quality of the liquids in terms of stability. More than 50 catalysts were tested in screen experiments using micro- scale and bench-scale fast pyrolysis apparatus. Four modes of operation were used: co-feeding the catalyst together with the biomass, replacing the sand fluid bed with a catalytic one (in bed), including catalysts in the primary reactor (ex bed) and close-coupling catalysts in a secondary reactor. Commercial FCC catalyst was also tested in a circulating fluidized bed and a rotating cone reactor showing that the catalyst was too active leading to reduced oil and increased gas yields. The experiments revealed that thermally expanded slate, a cheap construction material, gave the best quality liquids as measured by initial viscosity, stability and water content. Therefore, larger scale experiments were conducted with this type of material using lab-scale and pilot plant scale fluid bed reactors and a rotating cone reactor. The results show that the results from the laboratory could be transferred. However, difficulties in the fluid bed reactors were observed due to the different fluidization behaviour of slate. Substantial technical improvements of the pilot plants with fluid bed reactor and rotating cone reactor could be made in the course of the project. Both oxidants and anti-oxidants have been used for liquid modification. The addition of a 10% wt / wt aqueous solution phenol was found to result in the best quality pyrolysis liquid. This chemically modified pyrolysis liquid was more stable and had a lower water content than regular pyrolysis liquid. A process has been tested for the use of biomass pyrolysis liquid product as a partial substitute for formaldehyde-based resins in the production of composite panel. The technology was proven at the lab scale. The pyrolysis oil can be used as such to replace up to 20% of the UF/PF binder in composite board production with no significant impairment of the board strength properties. A new experimental system has been developed to investigate the fast pyrolysis kinetics of solid fuels, in the absence of heat and mass transfer limitations. A model has been formulated for the fluid-bed flash pyrolysis of biofuels with and without the presence of catalyst particles. A closed-form solution has been obtained for the tar profiles along a spherical catalyst particle, as a result of intra- and extra-particle diffusion and chemical reactions. The dynamics of tar vapors as they enter a packed bed reactor have been mathematically described. Furthermore, a model has been developed for the ablative pyrolysis of wood particles. Finally, an existing kinetic model was adopted to predict the distribution of primary products in fluidized bed and rotating cone reactors. Several analytical methods have been modified and developed. The most important ones are the determination of carbonyl groups by oximation, the evaluation of whole liquids by FTIR, and the assessment of stability by use of the Aston Viscosity Index. Conclusions The use of catalysts in fast pyrolysis is a suitable measure to improve the stability of liquids with respect to viscosity and water content. The understanding of the catalysts behaviour in fast pyrolysis is not yet understood. Selectivity, reactivity and specificity have still to be investigated. Models for various fast pyrolysis reactor configurations were developed and allow the prediction of the yields of the three main products oil, gas, and char. However, the performance of the models should be further improved. One of the most important result from the project was the addition of crude pyrolysis liquids to adhesives to substitute expensive phenol in resin formulations. Exploitation intentions As both the fluid bed reactor and the rotating cone reactor were improved within this project the commercialization of these technologies is being intended, basically to further improve the plants and to produce more oil for energetic and chemical applications. The use of bio.oil in adhesives will be further persued in future EU-projects as short term commercial solutions can be expected. 2 1 Partnership Table 1 The partners and names / addresses in the EC project JOR CT95-0081 Name Address Contact Person BFH, Institute for Wood Chemistry D-21027 Hamburg, GER D. Meier tel.: +49 40 73962-517 fax: +49 40 73962-502 email: dmeier@aixh0501.holz.uni- hamburg.de Aston University Birmingham B47ET, UK A.V. Bridgwater tel.: + 44 121 359 3611 fax: + 44 121-359 6814 e-mail: A.V.Bridgwater@aston.ac.uk Twente University P.O.Box 217 7500 AE Enschede, NL W. Prins tel.: +31 53 4892891 fax: + 31 53 489 4774 e-mail: W.Prins@ct.utwente.nl Sapemus Chemie GmbH / Adhesives Research Institute Ltd. Sofouli 88 GR-55131 Kalamaria Thessaloniki, GR P. Nakos tel.: + 30 31 424 167 fax: + 30 31 424 149 e-mail: panakos@ari.gr CPERI P.O.Box 361 57001 Thermi Thessaloniki, GR M. Samolada tel.: + 30 31 498 146 fax: + 30 31 498 180 e-mail: samolada@alexandros.cperi.forth. gr Union Fenosa Ingeneria Piquer 7 28039 Madrid, SP J. Alonso tel.: + 34 91 384 71 00 fax: + 34 91 384 71 99 e-mail: gen@ufisa.es Napoli University P.le V. Tecchio, 80125 Napoli, IT C. di Blasi tel.: + 39 081 768-2232 fax: + 39 081 2391 800 e-mail: diblasi@unina.it University Santiago de Compostela Avda. de las Ciencias s/n ES-15706 Santiago de Compostela, SP I. Cresp tel.: + 34 981 56 31 00 fax: + 34 981 59 50 12 e-mail: eqisacre@usc.es 3 2 Objectives The use of pyrolysis liquid, has been limited by a belief that the quality is inadequate for commercial applications. Pyrolysis liquid, known as bio-oil, is relatively unstable and changes with time. Bio-oil can have a high viscosity, is acidic and can separate into two or three phases under certain circumstances. These negative characteristics increase with high temperature, long storage times and exposure to air and light. Therefore, the project main objective is to explore catalytic and chemical modification of the fast pyrolysis process and the pyrolysis liquids it produces, to improve the liquid properties and characteristics and thus make them more amenable to utilisation. The results from laboratory studies should be transfered to two existing pilot plants in of which the operation had to be improved. Two other objectives were formulated: (1) modelling of fast pyrolysis processes and reactors and (2) the application of whole bio-oil in adhesive formulation for wood panels. 3 Technical Description 3.1 Catalytic Pyrolysis and I mprovement of Pilot Plants Experiments with catalysts were performed at different scales and with different reactors: 3.1.1 Micro scale pyrolysis Catalysts were tested in a commercial analytical pyrolysis equipment (CDS Pyroprobe) which consists of a controller and a resistively heated platinum coil probe. The probe is directly coupled to a gas chromatograph which is equipped either with a mass spectrometer or a flame ionization detector. After pyrolysis under well defined conditions the vapors are directly swept onto the cold column and separated, identified and if possible qunatified. The resulting chromatograms were compared against blank runs to assess the catalytic activity. Both in-bed and ex-bed modes were tested with the micro-scale unit which is shown in Figure 1. Interface 200C Pyrolysis tube with heating coil Carrier gas: He Split injector to Capillar GC (GC/MS) inert glass wool Liner (wood) sample and Catalyst in heating coil Figure 1 Schematic of the micro-scale pyrolysis unit for catalyst sreeening 4 3.1.2 Bench-scale pyrolysis For these experiments mainly small fluid-bed reactors were used with a capacity of 100- 200 g/h. Most of the bench-scale runs were conducted with a commercial fluid-bed reactor consisting of a continuous fedding system for small particles, the reactor with cyclone and charpot, cooling device and electrostatic precipitator to collect non-condensables aerosols.Four modes of operation were used: co-feeding the catalyst together with the biomass, replacing the sand fluid bed with a catalytic one (in bed), including catalysts in the primary reactor (ex bed) and close-coupling catalysts in a secondary reactor. A typical reactor configuration is presented in Figure 2. Nitrogen Biomass FEEDER FURNACE CYCLONE AND CHAR POT WATER CONDENSER ELECTROSTATI C PRECI PI TATOR COTTON WOOL FILTER Product Gas to Meter, Sampling and Vent Mai n Liquid Product Water In Wat er Out FLUI D BED REACTOR Aqueous Secondary Product DRY I CE CONDENSER Figure 2 Typical fluid bed reactor configuration for bench scale experiments 3.1.3 Lab-scale pyrolysis These experiments were performed in larger fluidized bed reactors with a capacity of 1 and 5 kg/h. A drawing with the description of the latter is presented in Figure 3. 1 2 3 4 5 6 7 8 9 10 11 12 1 silo 2 vibration conveyor 3 screw feeder 4 fluidized bed reactor 5 cyclone 6 heat exchanger 7 intensive cooler 8 electrostatic precipitator 9 flair 10 compressor 11 gas preheater 1 12 gas preheater 2 13 overflow container charcoal gas oil gas circulation 500 C 13 Figure 3 Typical fluidized bed plant (5kg/h) for lab-scale experiments 5 3.1.4 Pilot plant scale pyrolysis Two pilot plants, each with a capacity of processing 200 kg/h of biomass, were used for these experiments: a fluid bed reactor and a rotating cone reactor. A picture of the rotating cone is presented in Figure 4. Figure 4 Photograph of the pilot plant rotating cone: the inner space with the rotating cone 3.2 Liquid modification Liquid modification has been done by mixing various solvents and other reagents to the pyrolysis liquid. This is an easy procedure and requires no specific technical apparatus. Furthermore, the formation of methyl-aryl-ethers (MAE) has been tested with phenol as model compound to give anisol in a catalytic test unit shown in Figure 5. Steam unit 200 N 2 TIC TIC TIC TI FIC N 2 PI Vent Vent Volumetric cylinder G.C. Analysis 7 8 9 3 5 6 TIC P.C. Simulated Distillation Leco Coke Analysis 1 2 F 2 F 1 FIC Gases G.C. Liquids 101 104 103 102 102 FIC 102 101 1.Liquid Feedstock 2.Syringe Pump 3.Feed Injector 4.Steam Injector 5.Catalyst Bed 6.Reactor Furnace 7.Cold Liquid Bath 8.Liquid Product Receiver 9.Gas Collection System PI 101 PI Vent 4 101 BIOMASS UNIT 100 BIOMASS UNIT 100 Figure 5 Fixed bed reactor system for chemical modification experiments 6 3.3 I ncorporation of bio-oil into wood adhesives A literature search in the field of utilisation of the biomass pyrolysis oil (bio-oil) in the production of resins for the wood panel industry, was carried out. The references found were limited and relatively new. In any case, a positive trend for the application of pyrolytic oil in such a field was identified. In a preliminary series of lab scale trials, pyrolysis oil produced from eucalyptus biomass was tested for its standard properties (e.g. pH value, solids content) and potential to substitute or upgrade the performance of Urea-Formaldehyde (UF) resin in the production of particleboards. To optimise the parameters of bio-oil application, the incorporation of pine bio-oil into a UF resin during synthesis was tried out. The best resin formulation thus obtained was applied in lab scale particleboard production. 3.4 Modelling For the assessment of the computer models, an experimental system has been designed and constructed which allows biomass pyrolysis to be carried out under fast heating conditions. Weight loss and temperature of the sample are continuously recorded and these data are used to formulate semi-global reaction mechanisms and to estimate the related kinetic constants. In order to compare the data obtained from the innovative experimental system here developed, a first set of measurements has been carried out for cellulose and, successively, a second set of tests has been conducted for beech wood, the feedstock of interest in this project. The system combines the characteristics of pyrolysis tube reactors and the classic thermoanalyzer systems. A schematic drawing is presented in Figure 6. Reflectors Lamps Heated zone Quartz reactor A radiative heater sample precision balance thermocouple B gas out 15 cm quartz reactor 2.5 cm 2.6 cm Figure 6 Schematic picture of the cross section of the radiant furnace (A) and of the reactor (B) 3.5 Development of analytical methods Well-known techniques have been introduced, to analyse the oil on basis of its physical (e.g. density, viscosity) and chemical (pH, CHN, and H 2 O contents) properties. Additional methods were developed to investigate the oils composition with respect to main compounds like acids, aldehydes and ketones, as well as many others. These parameters will be used in comparing the oils derived at different operating conditions (reactor type, operating temperature, feedstock rate etc.) or in the oils derived from various catalysts. In Table 2.1 the various analysis techniques applied are summarised. 7 Table 2 Analysis methods used materials Analyses wood moisture content pyrolysis gas overall volume with gas meter separation and quantification by gas chromatography pyrolysis oil Fourier transform infrared spectroscopy (FTIR)* UV-spectroscopy* gas chromatography-flame ionization detection (GC-FID) identification of single products by gas chromatography-mass spectrometry (GC-MS) elemental analysis water determination by Karl-Fischer titration acid number by titration* methanol determination by head space GC (HS GC)* viscosities with capillary and falling ball viscosimeters* determination of carbonyl groups by oximation* pour point ash content ethanol insolubles char Elemental analysis Methods with an asterisk were newly developed. 4 Results and Conclusions 4.1 Catalytic pyrolysis In total 71 catalysts (single or mixtures) were tested in over 250 test runs using the analytical pyrolysis-GC/MS and GC/FID system at IWC and a micro fixed bed reactor system at CPERI. Conclusions of the screening tests were drawn from the gaschromatograms and from the spectra of the MS. Screening of the catalysts intended to be used in catalytic pyrolysis, aims to: reduce the number of catalysts used in larger experimentation, ascertain which would be the best catalyst to use, indicate which chemicals should be analysed for in the product pyrolysis liquids. Several categories of catalysts were included in the experiments: 1. zeolites 2. FCC 3. Silica 4. Alumina 5. Iron/Chromium 6. Sodium 7. Magnesium 8. Cerium 8 The following conclusion can be drawn from the results: analytical pyrolysis in combination with GC/FID or MS is a valuable tool for screening experiments. A qualitative result can be obtained and the basic feature of a catalyst can be observed. However, quantitative results cannot be obtained. Therefore, tests at larger scale are necessary for a better evaluation of catalysts. Furthermore, the screening method only monitors the volatile products whereas larger scale experiments with complete oil collection can lead to more substantial results. cracking is observed mainly for carbohydrate derived compounds. Lignin derived compounds are more stable against catalysts. the amount of gas is increased. the number of arising components is reduced. the amount of some components is reduced (e.g. acetic acid). no distinct differences could be seen between In-bed and Ex-bed experiments. the results from pyrolysis GC-MS which were not clear enough to evaluate the effect of a catalyst Y-zeolite is expected to give best oil quality as a wider variety of chemicals are produced HZSM-5 completely converts carbonyls into furans and hydrocarbons Iron/chromium gives less water formation zeolites mostly decrease the amount of carbohydrate derived vapours Zinc, magnesium and cerium and showed almost no effect on the vapours composition Catalysts have been incorporated into the fast pyrolysis process in four modes of operation: co-feeding catalyst with biomass; replacing the inert sand fluid bed with a catalytic one; including catalysts close-coupled in this primary reactor and close-coupling catalysts in a secondary reactor. The following results were obtained with catalytic pyrolysis in bench scale: The slate fluid bed gave the best quality pyrolysis liquid as measured by initial viscosity, stability and water content. The maximum organic yield using slate was similar to that using inert sand and found to be at 475C, lower than that of the original pyrolysis process when using inert sand as the fluidising medium. (Aston) Experiments with co-feeding and in-bed gave the highest oil yields and lowest viscosities. A mordonite zeolitic catalyst yielded more than 80 % of an oil and exhibited at the same time a reasonable viscosity of 15 cSt (BFH-IWC). Furthermore, the mordonite oil contained only 1.7 mol/kg of carbonyl groups whereas the other pyrolysis liquids were in the range of 3-5 (BFH-IWC). However, due to more cracking the water content of the oils increased to 30 %. This corroborates also the findings from CPERI who found that a zeolite type catalyst converted most of the carbonly groups but leads also to more water formation. This is a positive effect as this type of functional groups is prone to further chemical reactions which may lead to the undesired instability. (BFH-IWC). Also a commercial Y-zeolite gave high oil yields and low viscosities when fed in the co-feed mode (BFH-IWC). Thermally expanded slate, first discovered and used at BFH-IWC, gave a low water content around 10 % and yields in the range of 74 %. The carbonyl content is around 4 mole/kg. Slate 9 is a very inexpensive natural material and must not necessarily recovered. The particle size can be controlled and this makes it a good candidate for a catalysts in fast pyrolysis as this technology should not require expensive upgrading methods (BFH-IWC). With alumina, magnesium and sodium the oil yields decreased drastically to 54-63 % and were not considered as suitable catalysts (BFH-IWC). Catalytic experiments with nickel, slate, Octavision, and -alumina were performed in the fluid bed at conditions where the non-catalysed runs showed optimum values for the oil yield, while using beech wood as the feedstock. The yield trends of catalytic pyrolysis were in line with those of the non-catalysed runs, although the absolute values differed of course significantly. The decrease in the oil yield was mainly due to the increase in the non- condensable gases (Twente). Oil yields for the -alumina catalyst were also notably lower than for the non-catalysed runs, while the gas yields were somewhat higher. Furthermore, the water content of the produced oil was much higher (the water content of the CT oil was over 65 wt.%) (Twente). For the nickel and the slate materials, yields and water contents were comparable with those of the sand-derived oils. However, the viscosity of the condensation train oil is significantly lower. It should be noticed also that the oils from the electrostatic precipitator are in this case less viscous. Unfortunately the viscosity of the EP oil could not be measured (Twente) The Octavision FCC catalyst has been shown to deactivate quite rapidly during the experiment. After a sufficient long period, the catalyst seems to become almost inert towards the cracking reaction, resulting in a bio-oil type resembling the sand derived oil. The oil derived from the slate is similar to the oil derived from sand. Slate offers a good starting point as the catalyst to be used for the RCR experiments (Twente). It was found that the liquid yield is reduced in all catalytic runs made with a fixed bed reactor. No major differences in the distribution of product yields was detected by changing the operation mode, in-bed versus ex-bed (CPERI). Comparing in-bed and ex-bed GC/MS results, it is evident that in-bed catalysis produces less heavier compounds than ex-bed An unexpected lower water content of the liquid was measured in slate pyrolysis. This result is reasonable, considering the fact that in the in-bed mode the primary pyrolysis vapours are in a direct contact with the catalyst, while in the ex- bed mode spend some time under thermal cracking conditions, where secondary reactions and polymerisation might occur. Coke is enhanced in the ex-bed mode (CPERI). Catalytic biomass pyrolysis runs in the fixed bed reactor gave comparable trends concerning the catalysts evaluation with those obtained using the synthetic bio-oil mixture of model compounds (CPERI). Screening at CPERI was performed with a solution of model. The results showed that HZSM- 5 had the largest effect on the product distribution. All carbonyl groups of the synthetic bio- oil were destroyed and converted to furans and hydrocarbons. It can be concluded that the model compounds approach is a useful and easy approach for catalysts evaluation, without any interference of any faults coming from the pyrolysis reaction (CPERI). 10 Lab-scale and pilot plant scale experiments with slate gave some difficulties in fluidization due to the lower density of slate compared with sand. More research should be done to determine optimum conditions for slate as fluidizing material. However, the oils obtained were comparable to those produced at bench scale. (BDH-IWC, FENOSA). FCC and slate were tested as catalysts in the rotating cone. Again, the results were quite similar to those obtained in the catalytic fluid bed experiments. The catalytic effect of FCC appeared to be much too strong, while the slate had a very moderate effect. While applying the latter catalyst, a less viscous oil was obtained with a slightly lower amount of polymeric components. A noteworthy result for FCC was the strongly improved mixing behaviour, which could be observed from the fact that almost no char was collected in the cyclone after the reactor (Twente). 4.2 Catalyst regeneration Heating in a furnace was applied to burn off the coke deposits. Scanning Electron Microscope (SEM) was used to view catalysts used, and the char or coke produced from running the pyrolysis apparatus. The final location in the apparatus of the catalyst following pyrolysis may also be ascertained by examining product char samples for catalyst. The SEM technique can also be used to evaluate the degree of regeneration achieved (Aston). It is recommended that the optimum time for regeneration is less than twenty-four hours, to avoid damage to the surface, and greater than one hour (in the region of two hours) to remove all the coke (Aston). When thermally expanded slate is used as the fluidising medium, coke appears to build up around the pores rather than blocking or covering them. The surface is therefore more smooth after regeneration when the coke has been burned off (Aston). Coke deposited on the catalyst was burned (600 o C for 10 hr) and the regenerated catalyst was characterised in terms of BET surface area and XRD analysis. Mesoporous materials are not stable catalysts, because they are very sensitive to the presence of water in the reaction system. They become more amorphous after pretreatment reaction and regeneration (CPERI). The obtained product yields using the regenerated catalysts are comparable to those obtained using the fresh samples. XRD results indicate that the same phases are present in the regenerated catalysts and the fresh samples, while BET support the fact of the complete regeneration of the catalysts CPERI). Characterisation of both the FCC type catalyst and the olefin plus used in catalytic biomass pyrolysis in the FCC unit indicate that the catalyst remains unchanged after regeneration in the regenerator reactor of the unit. Catalysts regeneration can be effectively realised after their use in catalytic biomass pyrolysis (CPERI). All catalysts after the experiments were black and the particles covered with coke. This coke formation leeds to a deactivation after a certain run time. A change of oil quality in our lab- scale runs could not be detected (BFH-IWC). Regeneration was done in a funace at 650C over night. After the treatment the black colour disappeared as the coke was completely oxidized. In conclusion it can be stated that simple 11 oxidation can be used to clean the catalyst from coke. The subsequent use of regenerated catalyst did not lead to a decease in oil yield (BFH-IWC). After each catalytic flash pyrolysis experiment, the catalyst material was regenerated by feeding air to the fluidised bed. In this way the carbon deposited on the catalyst could be burnt off. For three of the applied catalysts this procedure appeared to have no influence on the activity in any following pyrolysis experiment. However particle sintering was observed for Nickel, which resulted in an irreversible deactivation of the catalyst (Twente). 4.3 Liquid modification Solvents such as alcohols, acetone, ethyl acetate, phenol, and glycerol were added. The addition of all solvents that were able to be measured, produced more stable liquids compared to the blank runs (Aston). Water also produces a comparatively stable liquid, but this is done with detriment to the final water content. Unlike water, phenol addition actually reduces the water content of the liquid compared to the blank pyrolysis liquid (Aston). Chemicals have also been added to the pyrolysis liquid, after production, to try to remove the source of polymerisation, as an alternative method of stabilisation. Both oxidants and anti- oxidants have been used. The addition of a 10wt.% aqueous solution phenol was found to result in the best quality pyrolysis liquid, measured by water content and the Aston Viscosity Index AVI (Aston). The catalytic methylation of biomass pyrolysis vapours was studied to investigate the chance of the selective direct conversion of the phenolic compounds to methyl aryl ethers (MAEs), either inside the pyrolysis reactor or just at the reactor exit prior the liquid products condensation. Unfortunately, the procedure indicated that it cannot be realized in a direct step, because of the high reactor exit temperature. It is believed that entirely new catalysts, than those usually applied to the methylation of phenolic compounds, should be developed CPERI). 4.4 Production of Adhesives The following results were obtained from project research work (SAPEMUS): A positive trend for the application of pyrolytic oil in the production of resins for the wood panel industry was identified through a literature survey. Lab particleboards were produced by substituting up to 25% of the UF binder with eucalyptus or pine bio-oil. Lab plywood was produced by substituting up to 20% of the PF binder with pine bio-oil. A synthesis procedure was modified in the lab for the production of PF resins (resoles) by partial substitution of the phenol needed with bio-oil. The performance of these resins was tested in the production of lab particleboards. A synthesis procedure was modified in the lab for the production of UF resins by partial substitution of the resin solids with bio-oil. The performance of these resins was tested in the production of lab particleboards. 12 Lab particleboards were produced by substituting 20% of the UF binder with pine bio-oil and without adding any resin curing agent. The bio-oil samples of UNION FENOSA, Inst. for Wood Chemistry and Univ. Of TWENTE were incorporated in PF resins during synthesis. The efficiency of the resins was tested by measuring standard properties and by producing lab scale particleboards.
From the results the following conclusions were thus drawn: The pyrolysis oil can be used as such to replace up to 20% of the UF/PF binder in composite board production with no significant impairment of the board strength properties. The partial substitution of resin raw materials by bio-oil in UF/PF systems provides sufficient reactive binders for particleboard production. The use of bio-oil as a hardener for UF systems is not considered advantageous. The UF/PF resins substituted/modified by bio-oil have low formaldehyde emission, irrespective of the high formaldehyde content used. The performance of a specific bio-oil sample depends on the wood raw material and the pyrolysis conditions applied. Summarising, the use of the bio-oil as a UF/PF resin replacement in the production of composite panels was proven possible, however the price of such a material should be attractive and the supply guaranteed, to initiate industrial application. The performance of the bio-oil resins could be improved if the aldehyde and/or phenolic fraction of the oil was increased. The dark colour of bio-oil, however, prevents its use in the production of resins for decorative applications (e.g. decorative paper/veneer impregnation). 4.5 Modelling The predictions of several semi-global mechanisms, which take into account the formation of char, tar and gas, have been compared for the primary pyrolysis of cellulose and wood, under chemical reaction control. The cellulose mechanisms examined are: 1) the Broido mechanism with two different sets of kinetic constants, the Shafizadeh mechanism and a modified Broido-Shafizadeh mechanism with two sets of kinetic constants. For wood, a mechanism which combines the contributions of cellulose, hemicellulose (xylan) and lignin is considered, where char and volatiles (tar and gas) formations compete. The widely used three-reaction mechanism, leading to the formation of char, tar and gas, is also examined, with four sets of kinetic constants (including those estimated in this project). Product yields and conversion times are evaluated as the temperature and the heating rate are varied (Naples). It has been found that, while differences in the simulations of cellulose pyrolysis are only quantitative, both qualitative and quantitative differences are established for wood. Furthermore, only the kinetics determined in this project can describe correctly the dependence of product yields on temperature under fast heating (Naples). A model has been formulated for the fluid-bed flash pyrolysis of biofuels with and without the presence of catalyst particles. The hydrodynamics of the bed are described by empirical relations, widely used in the literature (Naples). 13 A detailed particle model was coupled with the reactor model to take into account intra- particle processes. The models describe changes in the mass concentration of biomass, char, gas and tar, due to primary particle degradation and secondary extra-particle reactions (thermal cracking and catalytic conversion), and temperature. The effects have been investigated of the volumetric inert gas flow rate, the biomass mass flow rate and the furnace temperature on the product yields. It is shown that a maximum in the tar yields exists for intermediate reaction temperatures and sufficiently short volatile residence times (Naples). Simulation results reproduce well the experimental observation for cellulose and, when the kinetic mechanism formulated in this project is used, also for wood. Simulations indicate that intra-particle secondary reactions are negligible, though spatial temperature gradients can be high at the most severe conditions (Naples). A closed-form solution has been obtained for the tar profiles along a spherical catalyst particle, as a result of intra- and extra-particle diffusion and chemical reactions (Naples). AQ model has been developed which describes the combustion of coke deposited on the surface of catalyst particles under fluidized-bed conditions, based on the assumption of a continuous perfectly stirred reactor. Numerical simulations have been made by varying the thermal and fluid-dynamic conditions of the bed as well as the rate of catalyst supply. They indicate that oxygen diffusion across the catalyst particle plays a controlling role (Naples). The dynamics of tar vapors as they enter a packed bed reactor have been mathematically described. The reaction rate shows the usual Arrhenius dependence on temperature and is first order in the mass concentration of tar (Naples). A model has been developed for the ablative pyrolysis of wood particles. Pyrolysis is described by a semi-global, multi-step mechanism, consisting of solid depolymerization, followed by two competing pathways leading to vapor (liquid) formation (high temperatures) or to gas and char formation (low temperatures). Depolymerized wood is a molten phase intermediate. Only for temperatures above 823K and heat transfer coefficients above 0.05 cal/cm 2 sK the process presents the typical features of ablative pyrolysis. For these conditions, product yields show only a weak dependence on the wall temperature. Indeed, there is almost complete conversion to volatiles (80-87 percent on a dry wood basis) (Naples). The thickness of the molten layer, which remains constant for the whole duration of the conversion, becomes successively thicker as the wall temperature is decreased, with a corresponding decrease in the ablation rate. The predictions of ablative pyrolysis show good quantitative agreement with single particle and reactor experiments in relation to ablation rate, thickness of the molten layer and product distribution (Naples) A reactor model has been developed including the primary wood decomposition reactions as well as the cracking of primary and secondary tars. It allows the calculation of the product distribution as a function of the operation temperature and the hot vapour residence time. Primary kinetics are shown to determine the ultimate product distribution (Twente) 14 The maximum oil yield obtainable appears to depend only on the pyrolysis temperature, because heat transfer limitations and intra-particle cracking of the produced vapours do not occur for the fine wood particle considered (Twente). Cracking of the hot vapours elsewhere in the reactor and its connection to the condensation section may occur depending on their residence time in the hot sections. If the degree of cracking remains limited to below 25%, the type of gas mixing (ranging from plug flow to ideally mixed) becomes irrelevant. In that case a detailed knowledge of the flow pattern in the transfer tubes and cyclones is not required; the only parameter needed is the average vapour residence time (Twente). The predictions of a simplified model are in reasonable agreement with the measured data for sand and slate as a heat carrier or catalyst material. Any observed deviations can be explained by the uncertainties in the average hot vapour residence times (Twente). It has been learned from the modelling that in future flash pyrolysis operation, the residence times should be reduced to below one second to approach the maximal possible oil yield. On the other hand, the temperature in the transfer line to the condensation section should be well selected. Too high temperatures will induce excessive cracking of the desired products, while too low temperatures may cause undesired condensation and clogging. The optimal temperature for heat tracing must be determined experimentally, and will usually be found around 675 K (Twente). 4.6 Development of analysis methods Several methods were applied, adopted, improved and developed: Gases: In total 16 different gases were analyzed using two methods: 1. Micro- GC CP 2002 (CHROMPACK) with two TCDs on Hayesep and mol sieve columns and 2. GC 439 A, CHROMPACK, with FID and alumina PLOT column coated with KCl. All components were quantified based on calibration mixtures. CO, CO 2 , methane, nitrogen, hydrogen and other gases up to C 2 were determined with the TCD. Other gases were analyzed in a GC equipped with FID. Methane (CH 4 ), ethane (C 2 H 6 ), ethene (C 2 H 4 ), ethine (C 2 H 2 ), propane (C 3 H 8 ), propene (C 3 H 6 ), n - butane (C 4 H 10 ), iso butane (C 4 H 10 ), as well as cis -, trans - and iso - butene (C 3 H 8 ) were qunatified based appropriate calibration mixtures (BFH-IWC). Liquids: Liquid samples were qualitatively measured on a GC/MS system. A special mass spectra library was used and further expanded within the project for the identification of compounds. For quantification 41 different mixtures were prepared with the most important components. The individual compounds were calibrated and determinded using the internal standard method (BFH-IWC). From the chromatograms peaks were grouped into the following categories: 1. carbohydrate derived products 2. lignin derived products 3. carbonly products 4. hydroxyl products 5. furane products 6. acetic acid 15 7. levoglucosan Thus, the volatile part from all oils could be directly compared in order to monitor changes caused by catalysts. Up to 40 % of the pyrolysis oil could be separated by GC and three quaters from that part could be qunatified (BFH-IWC). FTIR Spectroscopy: All FTIR spectra were recorded on a BIO-RAD Typ FTS-40 instrument with absrbance technique. The liquid samples were placed between two Irtran II plates (ZnSe) with a thickness layer of ca. 20 m. A scanning rate of 32 was chosen and a spectral resolution of 4. The selected measuring range was between 4000 and 700 cm 1 wavenumbers. With the help of te BIORAD software all spectra were baseline corrected and normalized to the vibrations of teh aromatic ring whih is between 1520 and 1500 cm 1 (BFH-IWC). A new evaluation method for pyrolysis liquids using FTIR spectra was developed. The spectra were subdivided into 11 sectors of which 7 are in the fingerprint region. Division was based on different signals coming from various finctional groups of the oil constituents. The integrated areas of the sectors were compared beween the different oils in order to monitor changes caused by pyrolysis conditions (BFH-IWC). It is an advantage of the FTIR method, that no sample preparation is needed. It is very quick and during the analyis, the whole oil and not a fraction is being measured. Problems arise with the interpretation of the spectra. A skilled person is needed for spectra manipulation and to draw the right conclusions (BFH-IWC). UV Spectrometry:UV spectrometry was applied with whole pyrolysis mixtures. The information content of the spectra is not very high. The differences between the oils were not very pronounced as mostly the absobance of lignin is monitored by UV. It seems that the chemical nature of the pyrolytic lignin, which is dissolved in the pyrolysis liquid, does not change very much. By comparing the absorbance at 279 nm differences of carbonyl contents can be measured (BFH-IWC). Carbonyl Groups: The quantitative determination of the carbonyl groups was performed through oximation. The carbonyl groups are oximated through hydroxylaminhydrochlorid (HAHCl).The addition of pyridin ensures te complete oximation. After the reaction the free acid in the form of pyridinhydrochlorid is determined through titratin with NaOH.and is a direct measure for the amount of reacted carbonyl groups (first transition point of the titraton curve). The unreacted rest of hydroxylaminhydrochlorid can be determined through titration with NaOH (BFH-IWC). The method is working correctly as controlled with standards. However, skilled persons are needed and some further research work is needed. The results from the wet chemical determination do not in all cases correspond with those measured by FTIR and GC (BFH- IWC). Capillary viscometry was used to establish the Aston Viscosity Index (AVI). It uses raised temperaturesfor 24 h to accelerate ageing of the pyrolysis liquid and to therefore asses liquid stability. It was developed following a study of the effect of storage temperature on pyrolysis liquid viscosity. Three temperatures, 25, 50 and 75 C were used to store pyrolysis liquid for viscosity and water content testing. It was found that pyrolysis liquid was most stable with respect to viscosity when stored at 25 C. This temperature was chosen as the reference 16 temperature at which all liquid would be tested. A direct comparison could therefore be made between liquids stored at different temperatures. The lower the number is, the more stable the pyrolysis liquid. A zero result indicates no viscosity change. Its advantage is that only 48 hours is required for a test, rather than prolonged ageing tests (Aston). 4.7 I mprovement of pilot plant The fast pyrolysis plant located in Meirama, Galicia, Spain and operated by FENOSA was improved. The following technical changes have been made: New mechanical feeding system to the pyrolyser based on the first design of the system installed during the operation. Cleaning and substitution of some filters, with new implementation of screen filters before the cyclone. Modifications of the collection system, including cleaning and checking the shock cooler. Modification of the lower flange of the reactorin order to avoid leaking and clogging. Modification of the drain system with the incorporation of new tanks in order to improve the recovery of different oil fractions at the condensing or draining points. 17 5 Exploitation plans and anticipated benefits The project led to new results both in a scientific area and technical area. The results concerning the scientific area are mainly derived from the work packages catalyst screening, modelling, and analysis methods, namely FTIR, GC and the Aston Viscosity Index. These results have been and will be published in scientific journals and presented at international conferences. The whole scientific community may use the results as new tools for the characterization of bio-oils. Thanks to the PyNe Network the results can easily be distributed to all members within Europe and the IEA members in USA and Canada. The technical progress is mainly due to the discovery of slate which is an interesting, inexpensive material with moderate catalytic activity, so that overall yields are not affected. It is planned in future work to perform more tests with larger scale experiments. The most important finding was the suitabiliy of pyrolysis liquids to be a substitute for phenol in resins used as adhesives in panel manufacture. First results were already published as a contribution to the PyNe Newsletter. There is a lot of potential in this field and further research is required. The use of bio-oil in adhesives will provide the wood panel industry with more environmentally friendly and less costly resins.