Ethanol Port Injection and Dual-fuel

Combustion in a Common-rail Diesel Engine

Srinivas Padala

A thesis in fulfilment of the requirements for the degree of

Doctor of Philosophy

School of Mechanical and Manufacturing Engineering

University of New South Wales

February 2013
 PLEASEM'E
Tllf UNIVERSITY OF NEW SOUTI1 WALE~
Thesls!Dissert:)fion Shec l

Surrmmc ar ~JMI.Iy name P11dnta

'"'' ro.rne- Srit\lvn

ll,r)Q(("VI;'Itll)n ror clf!q!t'C ~ g!yen wt lhe U!\l'tll•tMV cak!nct.:r PhD

S.:t'IOOI S~1aoll)( M~ctt."nleal Mid M l'lnuf3euuln g Eng11\eetftl!g Faculty. Focutlv or Eno1Me:r1'!g

Jl\1~ Etll•lnol Pon fnftt~tfon ond OunJ·futl C~bu:a1ion In a

Commo(l-rnll Drnnl Efl pino

O f:lp!Y.A'\<1 '0 a Cat'IVM!li»\M ll~:t':~ ot l'li'I{J fltMII"-01 ~ M iparlt.fgnltiOn !!"9ifw tM atudy dt!mu!'llii iUitr.r;: •t CIOiWI\ial of uiJI.!IMOt ui!IIU!I(I'n "' ot
Ol~tl Of19")tt u''w!o ~Mucl Q7nf)(;Jtool'1 11n11f"!JY wtl~;~m l!!ll'!anai IS mjedod ~nlo Ike ll'llnbe !'!Uinlflllti ""d ltiB~SP! Is Cli18'CII-; t"tC'dPd tnlfl ~~~
CM'Iti'I"~:IIM c:n.nm~~P lhlfl mnm: ltloeu" otlht• "lXlV tt me -.lfrot.t o' elhl.mol port 'u!!lii•W (PFl) Jplt'lys on llu.:ti·I'Uat eornbusbon ;tttd L'lms~l,.,.,
F"'5tiv Qf:l."ttl!" r1 !"t"OCI'll1 'W1d •pa~.tmt Otii.,.,..Dti'"'H~nt ot I'I!W'd F'FI SPf"A\'S were sll.died t.,..ng t-.'-e·111tA!IerlnQ lind t'l~ ~ lft!OIQ!ng
!'!lthniQIIGS ~mtonli.m nu.c-Ma@tld 111jtlei!Cn f ttt" f'neiii)I.U!t11eM WO\S also perftm\ed The tnnu.noM 'ol ft•..:.'ll fiOW•tiUil. 11\je<hOM dVtttl)Ofl i\nd
ambleot ;lr Q~.fk,w 01~ oJ pullblll\f lflltifiUI' M' ~ toiiOtt IU U~t41'1d etl'\a"'->1 PA SC)IIY c.l".ara(.(\141+'$ lf\AI litO '~MI tO aU!(!I(Mbol6 tMIJine1
F.ltQnQ( JQra~ d1tl &lao 1J11chfd for 'I' All~ Pfl po11U011J 10 txiWYI~ ll'te polflnl)al offetl Ol drooiciS·lWb WUffile:!IOII OM W&ll N~t f)Q Wtt" ~~
d!-~r u"d"'f'Stll'ltllt\iJ en ~JMnol PFI soravs d!.t11·fual engtMI ~fllf\inl!.t worG conduttld lOt wr~ ~Qf'\01 iW\e!QV ,~ .;~J;Id 9Ft PO"\JoN; 11 .s
IQUnd tf\al '~ etf~ Clf PFI ~on Qlqbnl r••lll'fiOmellS.SUdl at If! cylinder ?NS11fiif01 a~n\ hit:.! nittaH: mN. Gnd fn«NW' flrt~l~!,.! p!OSt.Url' ,,_
rnUth t~ ~.:g~•ue""' ttlltl'l the ~r&CJ N l'!lhannl el1tlf9Y ftaUian tiowu\'UI, thD: miUlm'lg r.,.,lt show.~> nHtD,utnb" dtftor~ .such tlllfl l'ie =FI *"'on
clor..ol to tl\e W'llaltc:~ Vlf!\~J tf!111Jif!"IJ!\ 1f1'4 h!g~ tt~nof ttnergy fmt:tiDn !han thai or tl)t (url!'t!r ~~m ptr,lllon Reduc:ecJ wall watllf!Q d'IC IO
surfnce bof\lng occ;vrri"9 QOI/W I'll( i'·"ll'o"i!' ~--~ SUS19HIIl0 as il po~ C:IIUSI! Whitt~ U GOI'lS!SlOf'll ~11'! :10'- ~ c.rlft!O't moM~ leSe <011\d 6Afi
iQwef onbun"ll rtyr.l1'01;81rbon "'"..•lion• 011'1~ IMveat~t~ Cor YMnJs. f!IN.no1 et~etg)' h•cJ~M-S was Qlio coru:1ucltd From tne 111\-t~,_
preMUre meaSJJrerr.eni.Jr 1! a toufld 1h~tl !lie trttMStd tiJ,tilnal ene~gy fhlcllon lncreate:so ~ ~ engtM erf~t!nnc)' vP to 10T\. until •11:5 llmliOO oy
m!Sflliflq rhe retUI!t ... comr.~~tad to <lieHJ""'-'Y OI)Ct3~ w•ll\ 'l&"ffr'9 ~lio't tim4!lOS •~"~ orccr to ~n wnemet IM lncrca'ed tllic.Jencv ~~

IdiW to It~ a:JJntNolac)n ch;r...ing Olti'I\CJI!Wed CO.,~II:M\ ~IW Wtth fut bu~ ol' ~ Ft.lt'lt'* Milo~ oi l"o Otl1~ rt~~ ~ INI
Ia ihfl ptlmtlry CDUih): rot U\a "*~Y g~ Gy Jd\llndng I he diew' ~ 11mlna • Is tovl'ld lti.Jt lllfl m.~ · ·mtlfll e;lh.~ f~lon ~fl ,_ •'\Nnrl~
to ?0"11. "llhovt tne mb;.Mng prtlltlcm WI W mtfeo3'Se. lf'lnft~ o;Judt tm~JIOI'It t$ alto Ob~wd wt~ ~.,,.._ '' ql.«''IOtt on l•w ~~~~~!)tea al
b11Attrw.
Ublii.JI~ t!fha.1Cii In, rftet.d eng!ttt rlo"""ver ''~IDle ,.-nol.e sn!IMins Nl!! m~ al•dha,..OI t!nllfVY rniJO gf l()'llt 01 h!ghcl w-aocrnnq ltwl
O(lll~611ort onn~ Mtttlllions" much M~ eomcan!d'NIIh convMba(ISI ciM!siM t.ottlbUJtiOtt

1
Dc<llllr~llon rclatlng to dl•pQ.litrQ(' o-f p rqp•c;t lha,slsldl ue,t:aUon

1Mrt.lbV 9f!!nl ta tha lhJ!o:~tvor N..v Su!Att ttl.:ilet ClbQOVf!tf tt.. r~tt! 10 :,ucnl>(e ~ 10 rn~e &n~thtlle M\v ~hesis OiOIS.Sert3liOn W\ ~Of 1n
JJmW In IMI Uf!IYemll .. ~tiM If\ fill~ ot nJCCIIll ftOW.OI ttere~e1 known. SUtJiel:l to tM- p!:OYISIQflot ofthl! CoPf'llght Ad 1966 I~'" etl!
p~rcy nghl$ Auc:n as pe!MI ''llhls. l s.IJQ !Urutn me (tgtlllo V$ei!'t fututewoP:.J tsvc.h-a$1.lr1Jc.let or bQoll:f') '"'Of
oort ot lhlslhesls ordlss~i1(1('1n
I
I ~ l1Ul!'Knle Un!VetSity Mladllms lo trw I~ )50 WOJdO'l~r.td of IS\)' ~~In DfMtlrrnllnO ~ds Sfllf"t''l;1tloroll1t1'!"5 iS <tpgll~.,nle 10 !jof(l"lf1Jt
I!)MSh only)

.Jvl . .~~. "'"" . .. _Q.!::~. . ··-· ,y Zv 1.1.

. LS /11ft..~
1'ff sill"''""' (r ,_, q,,...v~r~l.."' ~1!.)
Th6 Unl'Mtlilly n!CognlH$ 1h&J ttwra rna~ be ~~ ~~~ ftoQ~InQ rt~l'tOOf\5 Of'l eQOj'"§ Ot QOMtllQns on 1M ~,lot
or
rosir'lc:tbn farll penod of up to 2 'fl!lll• m~l bo mJI(to '"w1t1ng Rct~\le$1 S.IOt a ~et ponod rosltooctiOn maY 1M! ~Joefed I'I'II!X~PI~~t!
c•reum!!Uinces and W ufre ll\\1 1.'1\'41 or thu [)(u:ln ar GI3dua1e R'esl!fl<ltdt

' <OR OFFICE USE ONLV

THIS SIIEET IS TO BE GLUED TO THE INSIDE FRONT COVER Of f HE TJIESIS

ORIGI NAL tn STA f'EMENT

•1 ha!reb~ Jednn: 1ht11 'h'"' ..ohnw··~·ion 1~ Ill)' U\\11 wnrk nnd to lh~ h~'' t'l m~· kni"IWic.Jge 11 cttnlllll1

uu nmtc:tt.t1"' pl'\•\.hltt.,l) publl8htd Ot \HHH.'tl b~ JUmlhOt ptNlln "r -..u~o; tmUJDI rtuf'l\.l11l~'~f1'" ,,,

rnut~nDf '"'111ch h1" t.: b\.'1.'1'1 -n~:..-c:plt'CI f11r- tile- ..S\Hird of .1oy ul.her dtg~ ur l.hplconnl dl UN~\\ IIf 1.111~
•\tlu:r ~..S\Ic:ltlun:.l Jn:-'HIUII('IA. ~>\(11:f11 when: due ac:kJ\OY.Iedgcmcm '' nlildc In th.: thc.st,. ·\II)
cou~r•hullon uuul-.: h\ lhL· rL""tUU\'h hy ofhcr.s. \\·1111 whnm I hn\'~ WL1fi«.••l .u I rNSW 1tf d'c!-Wiu:n; . '"

nphcl1l~ ~,~~OII\\It;tl',!l"tltn lite tht:'-l." 1 at~,, (J.cc1n~ tl~u th-= intciiC\.'1trul c•,nh:ut uf th1.;:the"''""'"lh..:
r'Fc~luct ,,, "'" tm•n \\mk. 1:'\Cl'piiO thr cxtcn1 U1m ll~~l;o~l nm:e (rmn ullwn. m lh~ l'loJICct'~ dL~U!II nnd
tfln~•:runn \'lr in ~'> lc.. rn..-..+:nuumn and llnsmttt1c c!>.pcc.li.~llln l'i nct.nrm·t~r:tlllL-d

~1-" \,-'- . v•M

-~~~~~~ I.J

ii
iii
Acknowledgements

This work would not have been completed without the help of many wonderful people.
First, I would like express my sincere gratitude to my supervisor Dr. Shawn Kook. Without his
competence and generosity it would have been difficult to assemble this small piece of work. I am
truly grateful to him for keeping me focussed and giving me clear defined goals to accomplish the
task. I am also thankful to my co-supervisor Associate Prof. Evatt Hawkes for continually providing
the comments on my research work throughout my time at UNSW.

Without having good people around, it would be difficult to have a pleasant time in the
research laboratories. In that aspect, I am lucky enough to be surrounded by wonderful co-
researchers. Particularly, experimentalists from Engine Research Group deserve a huge credit for
completion of this thesis. I acknowledge Bryan’s help, who offered his hand in fixing the engine
number of times and played a key role in making the engine keep running. We collectively learned
a lot about engine hardware while re-building the engine. Thanks to Alvin, Yongming, Josh, Minh
and Shaun for creating a friendly and enjoyable working lab which always worth going.

On a personnel note, I take this opportunity to say thanks to my Sydney brother Anand, and
friends Arnov, Rizu and Sukesh, because of whom my life at Sydney is worth cherishing. Special
thanks extended to Obulesu for helping me to settle down during my initial days in Sydney.

Many thanks go to my family for their unconditional support and love. My appreciation
cannot be expressed in words for my father who sacrificed many little things to provide me the best
he could do. Finally, would like to thank my fiancée Bharathi for her love during the hard times of
the study and having the patience to be in a long distance relationship.

I would like to acknowledge the financial support provided by Australian Research Council
via the Linkage scheme partnered with Haltech, a Sydney-based company specialised in engine
management systems. Tuition fee, and living expenses of my study were supported by UNSW’s
University International Postgraduate Award (UIPA).

iv
Abstract

Opposed to a conventional approach of using ethanol in a spark-ignition engine, this study

demonstrates a potential of ethanol utilization in a diesel engine using dual-fuel combustion strategy
where ethanol is injected into the intake manifold and diesel is directly injected into the combustion
chamber. The main focus of this study is the effect of ethanol port fuel injector (PFI) sprays on
dual-fuel combustion and emissions. Firstly, details of temporal and spatial development of ethanol
PFI sprays were studied using Mie-scattering and high-speed shadowgraph imaging techniques.
Momentum flux-based injection rate measurement was also performed. The influences of fuel flow-
rate, injection duration, and ambient air cross-flow are of particular interest in an effort to
understand ethanol PFI spray characteristics that are relevant to automobile engines. Ethanol sprays
are also studied for various PFI positions to examine the potential effect of droplets-airflow
interaction and wall wetting. With the clear understanding on ethanol PFI sprays, dual-fuel engine
experiments were conducted for various ethanol energy ratios and PFI positions. It is found that the
effect of PFI position on global phenomena such as in-cylinder pressure, apparent heat release rate
and mean effective pressure is much less significant than the effect of ethanol energy fraction.
However, the misfiring limit shows measurable difference such that the PFI position closer to the
intake valves results in 10% higher ethanol energy fraction than that of the further upstream
position. Reduced wall-wetting due to surface boiling occurring on the hot valve seat is suggested
as a possible cause, which is consistent with 30% lower carbon monoxide and 64% lower unburnt
hydrocarbon emissions. Detailed investigation for various ethanol energy fractions was also
conducted. From the in-cylinder pressure measurements, it is found that the increased ethanol
energy fraction increases the engine efficiency up to 10% until it is limited by misfiring. The results
are compared to diesel-only operation with varying injection timings in order to explain whether the
increased efficiency is due to the combustion phasing or improved combustion associated with fast
burning of ethanol. Further analysis of the data reveals that the latter is the primary cause for the
efficiency gain. By advancing the diesel injection timing, it is found that the maximum ethanol
fraction can be extended to 70% without the misfiring problem but 20% increase in nitrogen oxide
emissions is also observed, which raises a question on the advantages of utilizing ethanol in a diesel
engine. However, negligible smoke emissions are measured at ethanol energy ratio of 20% or
higher suggesting that optimization of these emissions is much easier compared with conventional
diesel combustion.

v
List of Publications

Journal Papers

Padala, S., Woo, C., Kook, S., and Hawkes, E.R., “Ethanol utilisation in a diesel engine using dual-
fuelling technology”, Fuel 109:597–607, 2013.
(Chapter 5)

Padala, S., Le, M.K., Kook, S., and Hawkes, E.R., “Imaging diagnostics of ethanol port-fuel-
injection sprays for automobile engine applications”, Applied Thermal Engineering 52(1):24–37,
2013.
(Chapter 4)

Padala, S., Kook, S., and Hawkes, E.R., “Effect of ethanol and ambient pressure on port-fuel-
injection sprays in an optically-accessible intake chamber”, Atomization and Sprays 21(5):427–445,
2011.
(Chapter 4)

In Preparation:

Padala, S., Kook, S., and Hawkes, E.R., “Effect of ethanol port-fuel-injector position on dual-fuel
combustion in an automotive-size diesel engine”, submitted for a special edition of Energy and
Fuels, 2013.
(Chapter 5)

Conference Proceedings

Padala, S., Kook, S., and Hawkes, E.R., “Effect of ethanol port-fuel-injector position on dual-fuel
combustion in an automotive-size diesel engine”, to be presented at 4th (2013) Sino-Australian
Symposium on Advanced Coal and Biomass Utilisation Technologies, Wuhan, China, 9–11
December 2013.

Padala, S., Woo, C., Kook, S., and Hawkes, E.R., “Performance improvement of compression
ignition engine by ethanol and diesel dual-fuelling”, 18th Australasian Fluid Mechanics
Conference, Launceston, Tasmania, December 2012.

Padala, S., Bao, Y., Kook, S., and Hawkes, E.R., “Effects of cross flow on ethanol port fuel
injection (PFI) sprays”, Australian Combustion Symposium 2011, Newcastle, Australia, December
2011.

Padala, S., Kook, S., and Hawkes, E.R., “Influence of injection duration on ethanol port fuel
injection (PFI) sprays”, The 15th Annual Conference on Liquid Atomization and Spray Systems -
Asia, Kenting, Taiwan, October 2011.

Padala, S., Kook, S., and Hawkes, E.R., “Ethanol spray penetration and droplet diameter at boosted
intake conditions”, 23rd Annual Conference on Liquid Atomization and Spray Systems, CA, USA,
May 2011.

vi
Poster Presentation

Padala, S., Kook, S., and Hawkes, E.R.,, “Ethanol spray characteristics at simulated intake manifold
conditions of automotive engines”, IEEE Technologies of the Future-2010, UNSW, Sydney,
November 2010.

Also, the following manuscripts were co-authored during PhD candidature, but not part of this
thesis.

Bao, Y., Padala, S., Kook, S., and Hawkes, E.R., “Influence of injection pressure on gasoline and
ethanol spray penetration in a spark-ignition direct-injection fuelling system”, 18th Australasian
Fluid Mechanics Conference, Launceston, Tasmania, December 2012.

Kook, S., Le, M.K., Padala, S., and Hawkes, E.R., “Z-type Schlieren setup and its application to
high-speed imaging of gasoline sprays”, SAE Technical Paper 2011-01-1981.

vii
Contents

Acknowledgements ............................................................................................................................. iv
Abstract ................................................................................................................................................ v
Contents ........................................................................................................................................... viii
List of Tables ...................................................................................................................................... xi
List of Figures ...................................................................................................................................xii
Abbreviations................................................................................................................................... xvii

Introduction ........................................................................................................................................1

Literature Review ..............................................................................................................................4

2.1 Ethanol as Transportation Fuel...................................................................................................4

2.2 Ethanol in Compression Ignition Engines: Dual-fuel Combustion ............................................5

2.3 PFI Spray Characterization and its Impact on Engine Combustion .........................................10

Experimental Facilities ....................................................................................................................12

3.1 PFI Spray Test Rig ...................................................................................................................12

3.1.1 Optical spray Chamber ......................................................................................................12

3.1.2 Air Management System ...................................................................................................14

3.1.3 Fuel Injection System ........................................................................................................16

3.1.4 Control and Data Acquisition Systems ..............................................................................16

3.2 Spray Visualization Techniques ...............................................................................................17

3.2.1 Mie-scattering ....................................................................................................................17

3.2.2 Image Processing: Mie-scattering......................................................................................19

3.2.3 Shadowgraph Imaging .......................................................................................................25

3.2.4 Image Processing: Shadowgraph Imaging .........................................................................27

3.3 Momentum Flux Measurement ................................................................................................28

3.3.1 Data Processing: Momentum Flux ....................................................................................30

viii
 3.4 Engine Test Facility..................................................................................................................32

3.4.1 In-cylinder Pressure and Emissions Measurement ............................................................35

3.4.2 Data Acquisition and Controlling System .........................................................................37

3.4.3 Data Processing .................................................................................................................37

Effect of Injection and Ambient Conditions on PFI Sprays ........................................................41

4.1 Momentum Flux and Injection Rate Measurements ................................................................42

4.2 Ethanol Spray Penetration and Droplet Diameter Results........................................................44

4.3 Effect of Fuels ..........................................................................................................................49

4.4 Effect of Injector Flow-rate ......................................................................................................56

4.5 Effect of Injection Duration .....................................................................................................60

4.6 Effect of Cross-flow .................................................................................................................66

4.7 Effect of Ambient Pressure ......................................................................................................72

4.8 Summary of PFI spray results ..................................................................................................76

Dual-fuel Combustion ......................................................................................................................77

5.1 Effect of PFI Injector Position on Dual-fuel Combustion ........................................................79

5.2 Effect of Ethanol Premixing on Dual-fuel Engine Efficiency ..................................................87

5.3 Effect of Diesel Injection Timing on Dual-fuel Combustion ...................................................91

5.4 Summary of the results from dual-fuelling engine tests ...........................................................97

Conclusions and Future Works ......................................................................................................99

6.1 Conclusions ..............................................................................................................................99

6.2 Future Works ..........................................................................................................................102

References .......................................................................................................................................104

Appendix .........................................................................................................................................114

A1 Software controllers: LabVIEW .............................................................................................114

A2 Injection Rate Calculation ......................................................................................................116

ix
A3 Dynamometer Torque-Speed Map .........................................................................................118

A4 Fuel Flow Rate Measurement .................................................................................................119

A6 Combustion Parameter Calculations ......................................................................................121

x
List of Tables

Table 2.1: Summary of the ethanol-diesel dual fuelling studies 8

Table 3.1: Camera settings: Mie-scattering setup 19
Table 3.2: Camera settings: Shadowgraph imaging setup 26
Table 3.3: Engine specifications 36
Table 4.1: Fuel properties 42
Table 4.2: PFI injectors used for investigation of the injector flow-rate study 56
Table 4.3: Summary of the results from PFI spray studies 76
Table 5.1: Engine operating conditions 77
Table 5.2: Port fuel injector (PFI) positions relative to the intake valves 80
Table 5.3: Summary of all the results from dual-fuelling experiments 98

xi
List of Figures

Figure 3.1: Spray chamber design and stress analysis [Sherwood Design 2009] 13
Figure 3.2: Optical spray chamber mounted on a breadboard table 14
Figure 3.3: Anemometer probe placed inside the optical spray chamber for
cross-flow air velocity measurement 15
Figure 3.4: Mie-scattering setup with strobe light, optical fibre cable and CCD
18
camera
Figure 3.5: Demonstration of Mie-scattering imaging background correction
20
procure
Figure 3.6: Effect of threshold level on the mean droplet diameter 21
Figure 3.7: Boundary detection of fuel droplets and calculation of spray tip
penetration. Also shown is near-nozzle region which masks the spray
22
and omitted during the mean droplet diameter calculation
Figure 3.8: Magnified view for selected spray areas a ~ c in Fig. 3.7. For the
presentation purpose, the colour is inverted so that the liquid droplets
are black and the background is white. Each boxed area has 31 by 31
pixels corresponding to a 2 mm by 2 mm square box in Fig. 3.7 23

Figure 3.9: Axial slices of 5 mm height (red colour) used to draw axial profiles
of number of droplets and mean droplet diameter later in Figs. 4.4,
4.5, 4.11, 4.18, 4.23 and 4.26 24
Figure 3.10: The error range calculated for various numbers of individual images
25
(test conditions selected: ethanol, 3ms aSOI, 0 bar, 3ms pulse width)
Figure 3.11: Schematic layout the Z-type shadowgraph imaging setup 26
Figure 3.12: Processing of high-speed shadowgraph images showing temporal
evolution of ethanol sprays between 1 ms and 4 ms after the start of
27
injection (aSOI)
Figure 3.13: Spray momentum flux (force) measurement setup 28
Figure 3.14: Image of anvil when it was mounted on force transducer 29
Figure 3.15: Force transducer with adaptor installed in the spray chamber 29
Figure 3.16: Force profile of the spray measurement. Also shown is the PFI
injector driver signal 30
Figure 3.17: The variation of error range for start of injection (SOI) calculated
31
from various numbers of individual injections
Figure 3.18: Injection rate calculated from the momentum flux profile. Also
shown is the PFI injector driver signal 32
Figure 3.19: Photo of engine mounted on the engine bed in test-cell 33
Figure 3.20: Schematic diagram of the dual-fuelling engine setup 34

xii
Figure 3.21: The variation of CoV of IMEP calculated from various numbers of
38
individual cycles
Figure 3.22: Example of instantaneous in-cylinder pressure traces from 100 firing
cycles and ensemble averaged pressure trace 39
Figure 3.23: An example of measured in-cylinder pressure and calculated
apparent heat release rate 40

Figure 4.1: Momentum flux and mass flow-rate of the PFI injector for ethanol
and gasoline fuels 43
Figure 4.2: Evolution of ethanol spray at various time aSOI. The timing at which
the image was taken is shown at the top of each image 44
Figure 4.3: Spray tip penetration (top), and mean droplet diameter and number
of droplets (bottom) against time aSOI for ethanol fuel 45
Figure 4.4: Axial profile of droplet number along the injector axis for various
times aSOI for ethanol sprays 46
Figure 4.5: Axial profile of mean droplet diameter for various times aSOI (top)
and droplet size distribution at 4 ms aSOI for different axial
locations (bottom) 48

Figure 4.6: Spray tip penetration against time aSOI for gasoline, E50, and
49
ethanol
Figure 4.7: Mean droplet diameter versus time aSOI for gasoline, E50, and
ethanol (top) and number of droplets plot for the same conditions
(bottom) 50

Figure 4.8: The rectangular window shown is to be used to discuss the droplet
size detection-limit in Fig. 4.9 52
Figure 4.9: Images showing a selected spray region from the standard image
(high detection-limit) of Fig. 4.8 (top) and the same region with low
detection-limit (middle) for ethanol sprays. The bar graph at the
bottom shows droplet size distribution for both standard and low
detection-limit images of gasoline and ethanol sprays. Solid and
dashed lines are drawn for cumulative number of droplets for ethanol
and gasoline, respectively. Data was accumulated for 25 injections 53
and the camera was triggered at 4 ms aSOI

Figure 4.10: Close-up view of standard spray images (droplet diameter detection-
limit: 72.2 μm) for gasoline, E50 and ethanol at 4 ms aSOI 54
Figure 4.11: Axial profile of mean droplet diameter along the injector axis at 4 ms
aSOI for gasoline, E50, and ethanol. The Mie-scattered spray images
similar to Fig. 4.2 are used to obtain 5 mm spray slices in the axial
direction (definition given in Fig. 3.9) 55

Figure 4.12: Momentum flux and mass flow-rate of the injectors used in this
study. Ethanol fuel was used 57

xiii
Figure 4.13: Temporal evolution of Mie-scattered sprays for the low flow-rate
injector (top) and high flow-rate injector (bottom). Ethanol fuel was
used 58

Figure 4.14: Number of droplets (top) and mean droplet diameter (bottom)
against time aSOI for low and high flow-rate injectors for ethanol
and gasoline 59

Figure 4.15: Temporal evolution of Mie-scattered ethanol sprays for the long
injection (top) and short injection (bottom) case. The low flow-rate
injector was used. The actual end-of-injection occurs at 2.4 ms aSOI
for the short injection case while the injection continues up until 5
ms aSOI for the long injection condition 61
Figure 4.16: Shadowgraph (liquid/vapour) boundaries overlaid on binary images
of Mie-scattered (liquid) ethanol droplets for the long injection (top)
and short injection case (bottom). The low flow-rate injector was
used. The end-of-injection occurs at 2.4 ms aSOI for the short
injection case while the injection continues up until 5 ms aSOI for
the long injection condition 62

Figure 4.17: Number of droplets (top) and mean droplet diameter (bottom) for the
long and short injection cases for ethanol and gasoline. The low
flow-rate injector was used 64
Figure 4.18: Axial profiles of droplets number (top) and mean droplet diameter
(bottom) at different times aSOI for the long and short injection
cases. The data were obtained from Mie-scattered spray images in
Fig. 4.15 and 5 mm spray slices in the axial direction were used as
shown in Fig. 3.9 65

Figure 4.19: Mie-scattering image boundaries of ethanol sprays (top row),

shadowgraph boundaries overlaid on raw images (middle row) and
shadowgraph boundaries overlaid on Mie-scattering images (bottom
row) under the influence of cross-flow. The low flow-rate injector
and short injection duration were used. The end-of-injection occurs
67
at 2.4 ms aSOI

Figure 4.20: Distance from the nozzle to rightmost shadowgraph boundaries (i.e.
radial vapour tip penetration) for quiescent and cross-flow
conditions. The low flow-rate injector, short injection duration and
ethanol fuel were used. The end-of-injection occurs at 2.4 ms aSOI 68

Figure 4.21: Number of droplets (top) and mean droplet diameter (bottom)
against time aSOI at quiescent and cross-flow conditions for ethanol
and gasoline. The low flow-rate injector and short injection duration
were used. The end-of-injection occurs at 2.4 ms aSOI 69

Figure 4.22: Droplet size distribution for the quiescent and cross-flow sprays at 4
ms aSOI for ethanol. The low flow-rate injector and short injection
duration were used 70

xiv
Figure 4.23: Axial profiles of number of droplets (top) and mean droplet diameter
(bottom) at different times aSOI for quiescent and cross-flow
conditions for ethanol sprays. The quiescent spray data for the short-
injection case from Fig. 4.18 is plotted again for comparison
purpose. The Mie-scattered spray images in Fig. 4.19 are used to
obtain 5 mm spray slices in the axial direction (definition given in
Fig. 3.9). The end-of-injection occurs at 2.4 ms aSOI. 71

Figure 4.24: Ethanol spray images at 4ms aSOI for various ambient pressures 72
Figure 4.25: Ethanol spray tip penetration against time aSOI for various ambient
pressures 73
Figure 4.26: Axial profile of ethanol spray droplets number along the injector axis
at 4ms aSOI for various ambient pressures 74
Figure 4.27: Total number of ethanol sprays droplets against time aSOI for
various ambient pressures. 75
Figure 4.28: Mean droplet diameter of ethanol spray against time aSOI for
various ambient pressures 75
Figure 5.1: Diesel and ethanol injection masses for various ethanol energy
fractions. The total fuel energy per cycle was held constant at 1098 J 78
Figure 5.2: Illustration of the intake ports, manifold and port fuel injectors at
two different port fuel injector (PFI) positions 79
Figure 5.3: Figure 5.3: Effect of ethanol energy variation on in-cylinder pressure
traces (top) and apparent heat release rate traces (bottom). The left
column shows results for PFI position A and the right column shows
the results for PFI position B 81
Figure 5.4: Indicated mean effective pressure (IMEP, at top-left) and coefficient
of variation (CoV) of IMEP (bottom-left), ignition delay (top-right),
83
CA50 (bottom-right) for various ethanol energy fractions.
Figure 5.5: Effect of ethanol energy fraction and PFI position on carbon
monoxide (CO), oxides of nitrogen (NOx), unburnt hydrocarbon
(HC) and exhaust gas opacity (i.e. smoke level) emissions 85
Figure 5.6: Combustion efficiency calculated using measured HC and CO
emissions for both PFI positions tested 87
Figure 5.7: Diesel and ethanol injection masses for various ethanol energy
88
fractions. The total fuel energy per cycle was held constant at 1098 J
Figure 5.8: Indicated mean effective pressure against CA50 for dual-fuelling and
diesel-only operation. For dual-fuelling cases, PFI injector was
positioned at position A 89
Figure 5.9: Burn duration for various stages of combustion at different ethanol
energy fractions (PFI injector was positioned at position A) 91

xv
Figure 5.10: In-cylinder pressures (top) and apparent heat release rates (bottom)
for various ethanol energy fractions. The diesel injection timings
were varied from -8 to 0° CA aTDC as noted at the bottom-left
corner of each aHRR plot (PFI injector was positioned at position A) 92
Figure 5.11: IMEP and CoV of IMEP for various ethanol fractions and diesel
injection timings (PFI injector was positioned at position A) 93
Figure 5.12: Ignition delay, CA50 and burn duration for the data as in Fig. 5.9
(PFI injector was positioned at position A) 94
Figure 5.13: CA50 (at top)and CA10 (at bottom) values calculated from for the
individual cycles along with mean values for the same conditions as 96
of Figs. 5.10 and 5.11
Figure 5.14: Carbon monoxide (CO), unburnt hydrocarbon (HC), nitrogen oxides
(NOx) and exhaust gas opacity (i.e. smoke level) emissions for all
diesel injection timings tested (PFI injector was positioned at
97
position A)
Figure A1.1: Block diagram (at top) and front panel (at bottom) of the developed
controller for Mie-scattering imaging. 114

Figure A1.2: Block diagram (at top) and front panel (at bottom) of the developed
115
controller for injection rate measurements.
Figure A2.1: Force profile of the spray measurement. Also shown, area integrated
profile, PFI driver signal, averaged momentum flux ‫ܯ‬ሶ௙ and actual 117
duration (Δt)
Figure A3.1: Operation range of Froude Hoffmann AG-30HS eddy current
118
dynamometer
Figure A4.1: Diesel mass per injection measured using optical engine as well as
injection rate meter at different injection durations (130 MPa 119
injection pressure).
Figure A5.1: Demonstration of the effect of cut-off frequency on the apparent heat
release rate traces. The cut-off frequency is annotated on the top left
120
of each plot
Figure A6.1: Calculated Equivalence ratio for various ethanol energy fractions
122
The total fuel energy per cycle was held constant at 1098 J.

xvi
Abbreviations

aHRR Apparent Heat Release Rate

aSOI After Start of Injection
CA Crank Angle
CA aTDC Crank Angle after TDC
CCD Charge Coupled Device
CMOS Complementary Metal Oxide Semiconductor
CI Compression Ignition
CNG Compressed Natural Gas
CO Carbon Monoxide
CoV Coefficient of Variation
DAQ Data Acquisition
DI Direct Injection
EGR Exhaust Gas Recirculation
FEA Finite Element Analysis
GDI Gasoline Direct Injection
HC Hydro Carbon
HCCI Homogeneous Charge Compression Ignition
HFR Hydraulic Flow Rate
IMEP Indicated Mean Effective Pressure
LPM Litres Per Minute
LPG Liquefied Petroleum Gas
NOx Nitrogen Oxides
PDA Phase Doppler Anemometry
PFI Port Fuel Injection
SCCI Stratified Charge Compression Ignition
SI Spark Ignition
SMD Sauter Mean Diameter
TDC Top Dead Centre
TL Threshold Level
TTL Transistor Transistor Logic
UNSW University of New South Wales

xvii
Chapter 1

Introduction

According to Australia Biofuels Annual Report 2009 [Darby & Pettrie 2009], total
consumption of petroleum products in Australia for 2009 is estimated at 58,119 million litres and a
6 % increase (up to 61,463 million litres) is projected by 2014. This example indicates a steady
increase in fuel demand for the future. A bigger issue is that the two most popular fossil fuels,
gasoline and diesel consumption account for a significant portion of greenhouse gas emissions. For
instance, about 14% of the total greenhouse gas emissions in Australia are originating from
combustion of these conventional fuels in transportation vehicles [Australia’s Emissions Projections
2010]. Considering limited fossil fuel resources, alternative fuel feed stocks have to be found and
the question of how to use those properly in an engine should be resolved. More importantly,
stringent emission regulations must be met when alternative fuels are used in an engine.
Accordingly, industry and academic researchers look towards alternate energy sources which have
lower environmental impact and higher efficiency. Examples are hydrogen [White et al. 2006],
compressed natural gas (CNG) [Ahmad et al. 2005], liquefied petroleum gas (LPG) [Bayraktar &
Durgun 2005] and bioethanol. Over other alternative fuels, ethanol has lucrative benefit as it can be
derived from domestically produced agricultural (e.g. sugar cane and corn) or cellulosic sources.
Also, ethanol can be used in existing gasoline engines without a major hardware replacement;
especially for bio-ethanol fuel blended with gasoline [Cooney et al. 2005]. Moreover, engine
knocking is less problematic for ethanol due to its higher octane rating. As a result, ethanol has
penetrated into the market and in some countries (e.g. Australia and USA), a blend of 10% ethanol

1
and 90% gasoline (E10) is widely available. In Brazil, most new cars are capable of a selective
utilization of gasoline or neat ethanol (i.e. bi-fuel engine). Many fuel injector manufacturers
produce E85-compatible models and car companies offer flex-fuel vehicles that auto-detect E85 and
adjust the engine control.

However, the current approach, i.e. ethanol blended with gasoline and burned in
conventional SI engines is not optimal for the following reasons: First, ethanol has a much higher
resistance to knock, and hence could be burned in higher compression ratio engines resulting in
higher efficiency; Second, to avoid phase separation, the ethanol must not contain water [Demirbas
2000], which costs a lot more to produce than ethanol containing small amounts of water; Third,
ethanol has a high heat of vaporization and a low vapour pressure compared to gasoline, leading to
poor vaporization and hence ignition problems especially during cold-start [Chen et al. 2011].
These issues can be addressed by using the ethanol in a dual-fuelling diesel engine where the
ethanol is delivered in the intake manifold and diesel is directly injected into the cylinder using two
separate fuel injection systems [Splitter et al. 2011, Ogawa et al. 2010, Tsang et al. 2010, Rodríguez
et al. 2009, Ekholm et al. 2008, Himabindu et al. 2008, Lu et al. 2008a, Lu et al. 2008b, He et al.
2004, Abu-Qudais et al. 2000]. The higher compression ratio of the diesel engines offers efficiency
advantages by exploiting knock resistance of the ethanol. Also, hydrated ethanol can be directly
used in diesel engines without any blending and hence water in ethanol would not be an issue.
During the warm-up period, diesel-only combustion can be used to avoid the cold-start issue. In
addition, ethanol is an oxygenated fuel and is likely to burn in a lean premixed mode, therefore
helping to reduce harmful soot emissions that are problematic in conventional diesel engines [Chen
et al. 2007].

Ethanol-only combustion in a diesel engine has great potential to achieve high-efficiency

low-emissions combustion regimes such as homogenous-charge compression-ignition (HCCI) and
stratified-charge compression-ignition (SCCI) [Maurya & Agarwal 2011, Sjoberg & Dec 2010, Ma
et al. 2008]. The drawbacks of these combustion regimes are limited operation range due to an
excessive pressure rise rate at high-load conditions and poor combustion efficiency at low-load
conditions [Dec 2009]. More importantly, the start of combustion does not rely either on injection
timing as in CI engines or on spark timing as in SI engines but instead relies on chemical reaction
rates that are highly sensitive to temperature and mixture fluctuations and hence hard to control. In
a dual-fuelling engine, combustion of a premixed ethanol-air mixture starts with the in-cylinder
injection of diesel and therefore the diesel injection timing controls the combustion phasing.

2
Accordingly, the dual-fuelling of ethanol and diesel is a practical solution to utilize the ethanol in a
diesel engine.

The objective of this thesis is to advance the overall understanding of dual-fuelling

compression ignition engine running on ethanol and diesel fuels. More specifically, how ethanol
spray would behave inside the intake manifold of an engine and the effects of ethanol premixing on
dual-fuelling engine performance. To this end, the following topics were studied in great details:

1. Ethanol spray behaviour in simulated engine intake conditions with specific aims to
x design a spray chamber with optical access to simulate the engine intake airflow
conditions;
x design a momentum flux measurement setup to measure the momentum flux from
PFI sprays;
x understand ethanol spray characteristics such as spray tip penetration, droplet
diameter and droplet number and how differently ethanol behaves with respect to
conventional fuels such as gasoline; and
x understand the differences in spray structures due to change in various injection and
ambient conditions such as injection flow-rate, cross-flow velocity, and steady vs.
transient sprays.
2. Dual-fuel combustion performance in a compression ignition engine, with specific aims to
x develop an engine working on ethanol diesel dual-fuelling with necessary controls
and data acquisition systems;
x understand the relation between PFI spray behaviour and engine performance by
testing at different positions of PFI injector; and
x explore the ethanol-diesel dual-fuelling to extend the dual-fuelling operation limit.

3
Chapter 2

Literature Review

2.1 Ethanol as Transportation Fuel

Ethanol (ethyl alcohol) is an alcohol made by fermenting and distilling various feed stocks.
Ethanol can be produced from:

x biomass via the fermentation of sugar derived from grain starches of many crops including
wheat wastes and sugarcane [Lin & Tanaka 2005]
x biomass via utilization of the lignocellulosic fraction of crops [Lin & Tanaka 2005] or
x petroleum and natural gas via an ethylene intermediate step [Roptary & Webb 2003].

Ethanol is isomeric with di-methyl-ether (DME) and both ethanol and DME can be
expressed by the chemical formula C2H6O. Ethanol has been known as a fuel for many decades.
Bio-ethanol was used as a transportation fuel in reciprocating combustion engines a century back as
early as 1894 [Demirbas 2007]. It is well known that early spark-ignition (SI) engines were
expected to use ethanol made from biomass, however gasoline emerged as a dominant
transportation fuel because of its abundance at that time [DiPardo 2012]. Numerous studies have
been done in recent years evaluating the life cycle impacts of bio-ethanol, and despite debates in
economical and environmental impacts of using ethanol as a transportation fuel, ethanol is always
included in any studies of fuel supply perspectives for upcoming decades [Blottnitz & Curran
2007].

4
 Some of the properties associated with ethanol make it alternative for the conventional
fossil fuels. The auto-ignition temperature of the ethanol is higher than the gasoline suggesting
advantages in storage and transportation [Hsieha et al. 2002]. The heat of vaporization is higher
than that of gasoline and diesel which means improvement in volumetric efficiency due to charge
cooling effect in PFI engines. Also, ethanol has higher octane number than conventional gasoline.
The Research Octane Number (RON) and Motor Octane Number (MON) of the ethanol are 106 and
89 respectively, while RON and MON of the gasoline are 90 and 83 respectively [Wyman &
Hinman 1990]. Higher octane number of ethanol allows the engine to run at higher compression
ratios without knocking. Also, ethanol’s shorter burn time, broader flammability limits, and higher
flame speeds than gasoline [Wyman 1996] lead to theoretical efficiency advantages over gasoline.

However, ethanol has its own drawbacks. Disadvantages of bio-ethanol include 36% lower
energy density than gasoline, corrosiveness, lower vapour pressure, cold-start problem, miscibility
with water, and toxicity due to aldehyde emissions [Hansen et al. 2005, Poulopoulos et al. 2001].

2.2 Ethanol in Compression Ignition Engines: Dual-fuel Combustion

One might suggest ethanol utilization in compression-ignition engines can be best practiced
by kinetics-controlled combustion regimes such as homogenous-charge compression-ignition
(HCCI) combustion. In HCCI combustion engines, a single fuel is injected directly into the
combustion chamber very early in the intake stroke [Helmantel & Denbratt 2004] or in the intake
manifold [Maurya & Agarwal 2008, Megaritisa et al. 2007, Yap et al. 2005, Christensen et al.
1998]. The combustion is initiated by compression heating without any assistance of spark or
additional fuel injection. By this way, the fuel is premixed with the surrounding air for long time
before ignition and avoids the over rich zones limiting soot formation. Also, lean fuelling rates and
absence of high temperature diffusion flame suppress thermal NO generation. However these
combustion regimes have inherent problems to overcome such as:

• Auto-ignition of fuels [Christensen et al. 1999]: There are challenges to achieve the
auto-ignition of ethanol in CI engines. The high temperatures required for the auto-
ignition of the fuel demands a very high compression ratio or preheated air system.
For instance, the compression ratio can be as high as 22 when the intake air
temperature is 40°C, which is costly and causes significant nitrogen oxides formation.
With a lower compression ratio of 17, which is widely used in conventional CI
engines, the intake air must be heated to unrealistically high temperature of 120°C.
• Lack of combustion control method [Dec 2009, Nakano et al. 2000]: The kinetics-

5
 controlled combustion lacks active control of start of combustion, which imposes a
significant challenge to achieve seamless transition between different engine operating
conditions.
• Limited operating range [Dec 2009, Hyvönen et al. 2003]: The spontaneous ignition
of mixture results in a high rate of pressure rise, which limits high-load engine
operations. The low-load engine operation is also limited by too lean mixture
condition to sustain combustion, as otherwise misfiring would occur.
• Cold-start issues [Shi et al. 2006]: If liquid fuel is used, evaporation of fuels during
the cold-start phase becomes an issue due to increased wall wetting.

Various methods have been proposed to control the combustion in HCCI engines such as
variable valve timing [Milovanovic et al. 2004], variable compression ratio [Hyvönen et al. 2003],
variable exhaust gas recirculation (EGR) [Bhave et al. 2005]. However, none of these methods have
been fully successful in practical applications and remain in a concept stage.

To overcome challenges posed by HCCI combustion using a single-fuel, dual-fuel

combustion has been suggested. In dual-fuelling diesel engines, one fuel is injected into the intake
manifold to form a premixed charge as in many kinetics-controlled combustion regimes. However,
the start of combustion is not initiated by auto-ignition of the premixed charge but by near top-dead-
center injection of second fuel as in conventional diesel engines. This second fuel injection avoids
the need of large compression ratios or preheated air systems. Hence, the current generation engines
with compression ratios in the range of 15 to 18 [Splitter et al. 2011, Surawski et al. 2012, Ekholm
et al. 2008, Lu et al. 2008a, Lu et al. 2008b, He et al. 2004] can be easily modified to run on dual-
fuelling.

In dual-fuelling, two fuels are selected such that the premixed air-fuel mixture does not
auto-ignite prior to the secondary injection to avoid excessively high rates of pressure rise [Karim
1980]. The secondary injection fuel should have high reactivity such as diesel so that the
combustion starts after a short ignition delay. Such in-cylinder blending of two different fuels with
different auto-ignition characteristics offers the advantage of tailoring the auto-ignition timing
[Kokjohn et al. 2011]. For instance, the variation of proportion of two fuels and diesel injection
timing can control the combustion phasing to avoid a high rate of pressure rise during the premixed
combustion. Moreover, the engine can run only on diesel injection to resolve the wall-wetting issue
during the warm-up period and misfiring problem at low-load conditions.

6
 Many have explored the potential of dual-fuel diesel combustion using various high octane
fuels as the premixed fuel. These include natural gas [Srinivasan et al. 2012, Srinivasan et al. 2006,
Krishnan et al. 2002, Karim 1980], liquefied petroleum gas [Jian et al. 2001] and hydrogen
[Saravanan et al. 2008]. Natural gas dual-fuel combustion is widely investigated in heavy-duty
engines, owing to its increasing availability [Werpy et al. 2010]. It was reported that the dual-fuel
combustion showed increased efficiency and reduced smoke and NOx emissions compared to
conventional diesel-only operation [Werpy et al. 2010, Srinivasan et al. 2006]. The liquefied
petroleum gas (LPG) and hydrogen studies demonstrated the same benefits of increased efficiency
and reduced emissions [Saravanan et al. 2008, Jian et al. 2001].

One might expect similar advantages for the dual-fuel combustion of ethanol and diesel.
Indeed, many have studied the dual-fuelling of ethanol and diesel in various engines [Splitter et al.
2011, Ogawa et al. 2010, Tsang et al. 2010, Rodríguez et al. 2009, Ekholm et al. 2008, Himabindu
et al. 2008, Lu et al. 2008a, Lu et al. 2008b, He et al. 2004, Abu-Qudais et al. 2000]. It is interesting
to note that these studies deliver mixed conclusions on the maximum ethanol fraction for successful
engine operation. Some reported the engine could not run at over 40 % ethanol fraction by energy
[Goldsworthy et al. 2013, Zhang et al. 2011, Himabindu et al. 2008, Lu et al. 2008a, Lu et al.
2008b, Chaplin et al. 1987] or 20% by engine load taken by ethanol [Tsang et al. 2010] due to
engine knocking. It may be presumed that the increased premixed combustion associated with
increased ethanol fraction caused a drastic rise of in-cylinder pressure. By contrast, others claim
ethanol energy fractions of 60% or higher [Splitter et al. 2011, Ekholm et al. 2008, He et al. 2004]
are achievable. It was explained that the combustion phasing becomes over-retarded with increasing
ethanol fraction [Tsang et al. 2010], which results in decreased peak cylinder pressure; however, the
combustion phasing can be moved forward by advancing diesel injection timing and achieve up to
97% ethanol fraction [Ekholm et al. 2008]. It may be understood that both the increased rate of the
pressure rise and retarded combustion phasing need to be considered to determine the maximum
ethanol fraction. However, no detailed discussions are found in the literature in this context.

Engine-out emissions have been main discussion points in the previous studies and there is
a consensus that the unburnt hydrocarbon (HC) and carbon monoxide (CO) emissions increase
[Abu-Qudais et al. 2000, Himabindu et al. 2008, Tsang et al. 2010, Rodríguez et al. 2009] or the
combustion efficiency decreases [Ekholm et al. 2008] with increasing ethanol fraction. It was
explained that some ethanol is trapped inside the crevice resulting in increased HC emissions
[Himabindu et al. 2008]. Almost all studies report benefits of ethanol substituting diesel in reducing
smoke emissions because ethanol is oxygenated fuel and the ethanol-air charge is well mixed. The

7
trends in NOx emissions however are mixed. For instance, NOx emissions decrease with increasing
ethanol fraction when the ethanol fraction is limited at 20% (or lower) [Himabindu et al. 2008,
Tsang, et al. 2010, Rodríguez et al., 2009] or the combustion phasing is retarded [Lu et al. 2008a,
Lu et al. 2008b, He et al. 2004, Ekholm et al. 2008]. However, when a wide range of diesel
injection timings were tested, the NOx emission was found to increase with increasing ethanol
fraction [Ogawa et al. 2010]. This trend relates to the increased peak heat release rate for advanced
diesel injection timings, which leads to higher maximum temperatures and thus increased thermal
NOx formation. Not only are the conclusions for the maximum ethanol fraction and NOx emissions
mixed but also an important question on the engine efficiency remains unanswered in the literature.
In a dual-fuelling scenario, the thermodynamic efficiency of the engine can either be increased or
decreased relative to a diesel-only baseline. Assuming relatively good combustion efficiency and
roughly the same heat loss in both scenarios, this mainly depends on the timing (i.e. combustion
phasing) [Splitter et al. 2011, Ogawa et al. 2010, Ekholm et al. 2008] and duration of the
combustion event. Timing is controllable by diesel injection but duration depends on the prevailing
combustion mode and may be shorter or longer than the diesel-only burn duration.

The key findings of the previous studies on ethanol-diesel dual-fuel combustion is

summarized in Table 2.1

Table 2.1: Summary of the ethanol-diesel dual fuelling studies

Authors, Engine Specifications Key Findings

Journal, Year
Goldsworthy, L., Six cylinder 5.9 L
Experimental Compression ratio 17.2
Thermal and Fluid Speed 1800 rpm
Science, 2013. Common rail injection
system
Surawski et al., Four cylinder 4.2 L
Energy Conversion Compression ratio 15.5
and Management, Speed 1700 rpm
2012. No information on
injection system
Mancaruso et al., Single cylinder optical
Proceeding of engine 0.52 L
Combustion Institute, Compression ratio 16.1
Italian Section, 2011. Speed 1500 rpm
Common rail injection
system
Splitter et al., SAE Single cylinder 2.5 L
2011-01-0363, 2011. Compression ratio 16
Speed 1300 rpm
Common rail injection
Ethanol up to 34% by energy.
With ethanol fraction increase:
Increased thermal efficiency.
Decreased NOx emissions.
Increased CO emissions and exhaust opacity.
Ethanol up to 40% by energy.
With ethanol fraction increase:
Increased HC and CO emissions.
Decreased NOx emissions and particulate matter.
No information on amount of ethanol.
Advancing the diesel injection increased the IMEP.
Two different injection timings -3°CA and -6° CA
aTDC were tested.
No comparisons made with baseline diesel.
Named as “Reactivity controlled compression
ignition engine”.
E85 up to 78% by energy with EGR.
16.5 bar IMEP was achieved with E85 fuel as

8
 system
Zhang et al., Four cylinder 4.3 L
Atmospheric 1800 rpm
Environment, 2011. Compression ratio 19
No information on
injection system
Ogawa et al., SAE Single cylinder 0.8 L
2010-01-0866, 2010. Compression ratio 16.5
Speed 1500 rpm
Common rail injection
system
Tsang et al., Energy Four cylinder 4.4 L
and Fuels, 2010. Compression ratio 19
Speed 1800 rpm
Inline fuel injection pump
Rodríguez-Fernández Single cylinder 0.77 L
et al., SAE 2009-01- Speed 1500 rpm
1853, 2009. Compression ratio 15.5
Mechanical fuel injection
system
Ekholm et al., SAE Six cylinder 12 L
2008-01-0033, 2008. Compression ratio 18.5
Speed 1450 rpm
Unit injector
Himabindu et al., Single cylinder 0.66 L
SAE 2008-01-2609, Compression ratio 17.5
2008. Speed 1500 rpm
Mechanical injection
system
Lu et al., Energy and Single cylinder 0.78L
Fuels, 2008. Compression ratio 18.5
Speed 1800 rpm
No information on
injection system
9
premixed fuel with 40% EGR.
With ethanol fraction increase:
Lower NOx, and soot emissions but increased HC
and CO emissions.
Ethanol up to 20% by energy.
Ethanol addition increases the engine efficiency at
low loads but decreased the thermal efficiency at
high loads.
With ethanol fraction increase:
Decreased CO and smoke emissions
Increased NOx emissions
Ethanol up to 20% by energy.
Fuel injection was adjusted to maintain ignition at
TDC.
Higher injection pressures helped in decreasing
smoke but increases NOx emissions.
Higher pressure rise rates at increased ethanol
fractions.
The combination of premixed ethanol and EGR can
achieve smokeless combustion and low NOx
emissions.
Ethanol fraction was defined as of % of load taken
by ethanol.
Maximum 20% of load could run by ethanol.
With ethanol fraction increase:
Increased ignition delay.
Thermal efficiency, NOx emissions, and smoke
were decreased.
Ethanol up to 15% by mass.
With ethanol fraction increase:
Lower and retarded in-cylinder pressure.
Higher HC, CO emissions and lower smoke
emissions, Lower NOx emissions
No change in thermal efficiency.
Ethanol up to 97% by energy.
Continuous of change of ethanol and diesel
quantities to maintain peak PRR.
Achieved 20 bar IMEP with dual-fuelling by proper
control strategies.
With ethanol fraction increase:
Increased HC and CO emissions.
No benefits of using dual-fuelling at mid-load.
High thermal efficiencies at high load conditions.
Ethanol up to 20% by energy.
Peak in-cylinder pressure decreased and retarded.
With ethanol fraction increase:
NOx decreased with EGR.
CO emissions depended on the load.
Low Temperature Reactions (LTR) were observed
with ethanol.
Ethanol up to 34% by energy.
With ethanol fraction increase:
Increased HC and CO levels.
Lower NOx, smoke and emissions.
Similar efficiency.
Increased and retarded pressure.
 Leahey et al., SAE Single cylinder 0.61 L
2007-01-4011, 2007. Compression ratio 16.5
Speed 900 rpm
Inline fuel injection pump.
He et al., SAE 2004- Two cylinder 1.75 L
01-0094, 2004. Compression ratio 17.1
Speed 1540 rpm
Electronically controlled
injection
Abu-Qudais et al., Single cylinder 0.58 L
Energy Conservation Compression ratio 18
Management, 2000. 1000~2000 rpm
Noguchi et al., Three cylinder 2.5 L
Bioresource Compression ratio 17.4
Technology, 1996. Speed 800~1800 rpm
Mechanical injection
system
Chaplin et al., Four cylinder 2.8 L
Transactions of the Compression ratio 16.1
American Society of Speed 800~1800 rpm
Agricultural Mechanical injection
Engineers, 1987 system
Ethanol up to 33% by volume.
Heated intake air.
With ethanol fraction increase:
Lower combustion duration.
Decreased smoke and NOx emissions.
Ethanol up to 97% (by energy) with EGR.
Peak in-cylinder increased till E60 (60% ethanol
energy) and then started decreasing.
With ethanol fraction increase:
NOx emissions were almost constant.
Ethanol up to 20% by energy.
With ethanol fraction increase:
Increased efficiency.
Higher HC, CO emissions and lower smoke
emissions.
Addition of ethanol over the diesel only running.
Knocking at higher ethanol rates.
Retarded diesel injection helped in lowering peak
pressure rise rates.
Ethanol up to 40% by energy.
Increased ethanol substitution with advanced
injection.
Efficiency depends on the operating load of the
engine.

2.3 PFI Spray Characterization and its Impact on Engine Combustion

Since dual-fuel combustion requires an intake fuel injection, spray characterization in the
intake environment is one of the important parameters for successful engine operation. There are
many studies reporting that the characteristics of intake sprays and their impact on in-cylinder
combustion and emission formation [Merola et al. 2010, Kato et al. 2007, McGee et al. 2000,
Daniels & Evers 1994]. For example, it was reported that the position at which the spray is injected
has a significance influence on the unburnt hydrocarbon emissions [Merola et al. 2010, Kato et al.
2007, Kim et al. 2005, McGee et al. 2000, Meyer & Heywood 1999, Cheng et al. 1993]. On one
hand, the injector positioned further upstream of the intake valves would cause longer time for
interaction between the fuel droplets and ambient airflow, which improves break-up [Moon et al.
2007]. On the other hand, the fuel injector located closer to the intake valves deliver better engine
stability and lowered unburnt hydrocarbon emissions because fuel sprays hitting the hot surface of
intake valves lead to enhanced vaporization of fuel droplets and reduced wall wetting [Kato et al.
2007]. These confounding effects raise an interesting question on the influence of PFI location on
dual-fuel combustion.

10
 The PFI location is not the only parameter that impacts the engine combustion. For
instance, Zhao and Lai [Zhao & Lai 1995] pointed out the importance of spray tip penetration,
droplet diameter, and spray angle when PFI system is used in an engine. The major factors which
influence these spray characteristics are identified as injection pressure, injector geometry and fuel
properties. For instance, with increasing injection pressure, the spray tip penetration increases due
to the higher momentum delivered by the injector [Anand et al. 2010a, Zhao et al. 1995, Lai et al.
1994]. Mean droplet diameter decreases with increasing injection pressure because increased fuel
jet velocity and turbulent kinetic energy enhances the liquid atomization [Dumouchel et al. 2005].
Since it is believed that smaller droplets help reduce the unburned hydrocarbon emissions [Heneina
& Tagomori 1999] and delivers better engine stability, high injection pressure is preferred. This also
has the advantage of high fuelling rate. However, very high injection pressure can lead to excessive
spray tip penetration and can create unwanted fuel dispersion problems which offset the advantages
of enhanced atomization [Nuglisch & Nally 1997, Zhao & Lai 1995]. Nozzle geometrical
parameters such as hole size, number of holes, and hole alignment with injector axis also play an
important role in atomization of the fuel sprays [Aoki et al. 2005, Dumouchel et al. 2005, Ren &
Sayar 2001, Tani et al. 1999]. Once again, the high turbulence helps reduce droplet size and
therefore a complex trajectory for the fuel path, which induces the secondary flow, is commonly
included in modern injectors [Tani et al. 1999]. Also, the injector design plays a critical role in
delivering the required momentum flux for the subsequent break-up processes to occur. Recently it
was suggested that the measurement of momentum flux generated by the injector can greatly help to
understand spray characteristics [Payri et al. 2005]. However, no data is available for sprays at PFI
conditions.

Fuel properties also make a significant impact on droplets break-up and evaporation
[Williams & Beckwith 1994]. The fuel viscosity, surface tension and density dictate the Weber
number of the spray, a good indicator of break-up. Ethanol has significantly different properties
from conventional gasoline with which existing fuel injection systems are developed. Many have
studied the spray differences between ethanol and gasoline using a gasoline direct injector (GDI)
[Matsumoto et al. 2010, Oh et al. 2010, Wang et al. 2005]. However, the findings from direct-
injection spray studies are not directly relevant to PFI sprays since both ambient pressure,
temperature and fuel-rail pressure are an order of magnitude different.

11
Chapter 3

Experimental Facilities

This chapter details the experimental setup and facilities used for spray visualization and
engine performance and emissions testing.

3.1 PFI Spray Test Rig

To investigate the port fuel injection spray characteristics, a spray visualization facility was
designed and built in the Engine Research Laboratory, UNSW. The facility includes an optical
spray chamber, visualization system, injection rate meter and control systems as follows:
x a spray chamber with optical access to visualize the spray structures
x air management system and fuel injection system
x Mie-scattering setup to visualize the liquid spray information
x shadowgraph setup for detection of the vapour phase boundaries
x momentum flux measurement system
x control and data acquisition systems

3.1.1 Optical spray Chamber

The optical spray chamber was designed to withstand an internal pressure of 200 kPa gauge
(to simulate the turbo charged intake pressure conditions) at a max temperature of 100°C meeting
Australian Standards Procedure AS1210. The design has been outsourced to an engineering

12
 Figure 3.1: Spray chamber design and stress analysis [Sherwood Design 2009]

consultant who carried out the detailed design including stress calculations considering the
boundary conditions [Sherwood Design 2009]. Figure 3.1 gives the results from the finite element
analysis (FEA), which shows the peak stress as 22.9 MPa which is well under limit of allowable
material stress of 94 MPa. Therefore the operating pressure of the optical chamber was limited by
the strength of glass windows used in this study. After fabrication of the chamber, it was
hydraulically tested and certified to work under design pressure. Separate fixtures were designed
and fabricated to mount the spray chamber on the breadboard (PB151515, Thorlabs) which was
fixed on the optical table (SDR 90120, Thorlabs). Fig. 3.2 shows the spray chamber when it was
mounted on the optical table.

Spray chamber was provided with an optical access on both longitudinal faces as well as
bottom face by specially machined boro-silicate (BK7) glasses. The thickness of the glass was
calculated using equation 3.1 [ispoptics.com, Rudnitzki 2005]

ܲ‫ܨכܣכ‬
‫ݐ‬ൌඨ ሺ͵Ǥͳሻ
͵ǤͶͺ ‫ܯ כ‬

Where t = Thickness of the glass

P = Pressure

13
 Figure 3.2: Optical spray chamber mounted on a breadboard table

A = Unsupported area
F = Safety factor
M = Modulus of rupture of glass

Considering the imperfections (pores, internal cracks etc.) that can occur in the glass
manufacturing, a large factor-of-safety of 10 was used and calculated 20 mm as thickness of the
glass. The side optical windows with a field of view of 170 mm by 70 mm enable the imaging of
spray penetrations for a time long enough to study both initial transient and the quasi-steady period
of injection.

3.1.2 Air Management System

Spray visualization was conducted at various ambient conditions including atmospheric

pressure, boosted pressure conditions, and cross-flow-influenced conditions. The spray chamber
pressure was monitored using a pressure sensor (PX219, Omega Engineering Inc.), with pressure
regulator upstream was controlling the desired pressure to attain in the spray chamber. The pressure
sensor was directly connected to the data acquisition system. An external air compressor with a high

14
flow rate of 350 LPM was connected to the spray chamber to control the ambient air pressure. The
spray chamber combined with the air compressor facilitated a continuous flow of air, which ensures
the removal of the fuel vapour after the fuel injection event. This not only ensured the same ambient
air conditions for different fuel injections but also prevented window contamination due to liquid
fuel droplets. Another advantage of the constant-flow chamber was a faster imaging interval
compared to a constant-volume-type chamber, as all the fuel droplets from the previous injection
were vented out continuously. A flow control device (FMA5444, Omega Engineering Inc.) was
placed between the compressor and the spray chamber, which continuously monitor and control the
flow-rate of air. The fuel vapour from the injections gets filtered in the coalescing filter leaving the
fresh gases into ambient atmosphere. Care was taken that continuous flow-rate of air not to create
any considerable axial moment to impact the spray characteristics. A mean air flow velocity was
estimated at less than 1 m/s, which was an order of magnitude lower than a typical PFI spray
penetration velocity (e.g. 18 m/s). It always ensures the fuel is injected into the quiescent ambient
condition.

To create ambient airflow for cross-flow studies, an external air blower (ON025-201,
Aerovent Australia Pty Ltd.) was used. The velocity profile of the cross-flow of air inside the spray
chamber i.e. the velocity across the different points along the injector axis in the spray chamber was

Anemometer Probe

Figure 3.3: Anemometer probe placed inside the optical spray chamber for cross-flow air
velocity measurement

15
measured using hotwire anemometer (VelociCALC 8345, TSI Incorporated). The measuring probe
was traversed along the injector axis to get the velocities at different points. Figure 3.3 shows the
arrangement of the measuring probe when inserted through the injector along the injector axis.

3.1.3 Fuel Injection System

Fuel injection system consists of a fuel tank, low pressure fuel pump (Bosch 0 580 254
044), fuel filter, custom-made fuel rail [Michael 2010, Lim 2011], fuel injector and pressure
regulator. Conventional fuel pump and injectors commonly used in passenger cars were selected.
The fuel pump was powered by 12V external power supply. The fuel pump and regulator were used
to maintain a constant pressure difference of 300 kPa across the fuel injector: excess fuel from the
regulator was sent back to the fuel tank. For the boosted intake conditions, the pressure regulator
automatically increases the fuel injection pressure so that the pressure difference across the injector
was fixed. A custom-made fuel rail was designed and machined using aluminium block to hold the
injector onto the spray chamber. The injector was a 6-hole dual-spray injector (Bosch MPI Model
EV-6, B 280 431 127-07), commonly used in passenger cars. The mass flow-rate of injector was
261 g/min (specification by Bosch: tested at 300 kPa of injection pressure using n-heptane). The
injector spray angle was 70°. For the injection flow-rate study, another injector with the same
specifications but much higher flow-rate was selected.

3.1.4 Control and Data Acquisition Systems

The timing and pulse width of the injector were controlled by in-house developed control
programme. It consists of two key elements

1. Software which triggers the injector drivers

2. Universal injector driver

The software was programmed in National Instruments LabVIEW (Version 9). NI PCI
6120 data control card was used to generate the necessary signals to control the injectors. This card
has 16-bit resolution and 8 analog output signals which were used to control different units. A
universal injector driver with a PFI module (Drivven Standalone system) was used in conjunction
with logics generated in LabVIEW programme. The pulse-widths and the delay between the
injection signals were individually controlled to achieve the desired sequence of injection events.
The snapshots of the develop LabVIEW controllers are given in Appendix A1.

16
 A number of parameters were recorded during each injection event using National
Instruments NI PCI 6221 multifunction data acquisition (DAQ) card. The DAQ card has 16-bit
resolution, 16 analog input, and 2 analog output channels with a maximum sampling rate of 1 MHz.
This DAQ card was integrated with control programme and programmed to save the data on the
computer on the user’s command. This DAQ system was used to monitor the chamber pressure
during spray visualization experiments and record the momentum flux data.

3.2 Spray Visualization Techniques

Spray visualization was achieved using Mie-scattering and focused shadowgraph imaging.
Mie-scattering imaging was performed for the identification of liquid droplet boundaries while
shadowgraph imaging was performed for detection of vapour phase boundaries.

3.2.1 Mie-scattering

In the principle of Mie-scattering, the amount of light scattered by liquid droplets is

proportional to the droplet size and intensity of illuminating source. The main advantage of this
imaging technique is a simpler setup than other optical/laser-based diagnostics [Suzuki et al. 1994].
Despite its wide acceptance [Fang & Lee 2011, Kook et al. 2009, Park et al. 2002], however, one
might argue that the Mie-scattering imaging has certain limitations to measure fuel sprays. For
example, a recent study conducted by a group of injector suppliers pointed out that Mie-scattering
was not recommended for PFI spray measurements [Hung et al. 2008]. It was suggested that a
mechanical patternator, which measures a pattern of liquid-phase fuel footprint at some distance
from the nozzle, be a better option to measure global parameters (e.g. spreading angle). Still, there
are many occasions that Mie-scattering diagnostics offer a great benefit over the mechanical
patternator. For example, spray images can provide detailed information of the spray tip
penetration, droplet diameters, and the droplet number. An axial profile of mean droplet diameter
and droplet numbers at certain timing can also be obtained, which will clarify spray break-up and
evaporation processes. Such measurements are not possible in the mechanical patternator because it
interferes with the sprays. For instance, if the spray tip penetration is not limited by break-up and
droplet momentum but by evaporation of droplets, the liquid footprint left in the patternator may
include liquefied fuel, which was originally in the vapour-phase and condensed during the
measurement. One major concern of imaging diagnostics and a good reason to promote the
mechanical patternator is line of sight integration and droplet overlapping. This is because the line
of sight integrated images can give inaccurate results for dense spray regions where droplet
overlapping becomes significant. For the PFI sprays in this study, the droplet distribution was dilute

17
enough to detect individual droplets sufficiently far from the nozzle, although the near-nozzle
droplets were very dense and hence were excluded in the image processing. Understanding pros and
cons clearly, Mie-scattering imaging was performed to measure ethanol/gasoline sprays in this
study.

Fig. 3.4 illustrates the Mie-scattering setup along with the optical chamber. A strobe light
(PerkinElmer X400) with 8 ~ 13 μs duration and 12 lumen-sec was used to freeze the spray motion
and illuminate liquid droplets. An optical fibre cable was used to place the light source as close as
possible to the spray chamber in an attempt to obtain the maximum illumination. Various layouts
the fibre cable locations were attempted to get bright as well as uniform illumination. It was found
that a location upstream of the chamber performs the best.

The scattered light from the droplets was detected by a CCD camera (Pike 421B) equipped
with a 16-bit sensor (65532 maximum image counts) and a maximum resolution of 2048-by-2048
pixels. In this study, a 1600-by-1200 pixels resolution (“standard setup”) was used since it was
sufficient for imaging the entire spray region. For this setup, a detection-limit of the equivalent
droplet diameter was measured as 72.2 μm per pixel. This detection-limit was likely enough to
characterize the droplet size of the PFI sprays. The lens used for the standard setup was a 50 mm
Nikon Nikkor lens. Aperture opening was optimized for the amount of light and depth of field, and

Control
and Data
Acquisition Pressure
Regulator
Injector Fuel
Driver Tank
Injector
Fuel Rail

Strobe
Light
Optical
Fibre Cable

Camera

Figure 3.4: Mie-scattering setup with strobe light, optical fibre cable and CCD camera

18
 Table 3.1: Camera settings: Mie-scattering setup

Camera model Pike 421B

Image resolution 1600x1200
f-number 8
Exposure time 70 μs

as a result an opening of f/8 was selected. The camera settings are summarized in Table 3.1.

While Mie-scattered liquid-phase spray imaging was performed, one side of the glass
window on the other side was covered with a black sheet to avoid unnecessary reflections from the
other end of the chamber. The camera, strobe light and optical fibre cable were positioned and fixed
on the breadboard using opto-mechanical mounts. With this optical setup, single-shot imaging was
performed at a fixed time after the start of injection (aSOI). The same imaging was repeated for 25
different injections to obtain ensemble-averaged data and uncertainty. The images were saved onto
the computer attached to the spray chamber facility. The SmartView programme [Allied Vision
Technologies 2010] that comes with camera was used for recording and saving the images. All
images were saved in RAW format so that no information would be lost.

3.2.2 Image Processing: Mie-scattering

The Mie-scattering images were post-processed using an in-house-developed Matlab code.

Prior to the fuel injection and spray imaging, a scale image and background image were taken. The
image processing began with a background correction. An example of the back ground correction
procedure is shown in Fig. 3.5. The image at the top-left shows the background image prior to spray
injection. The image at the top-right shows Mie-scattered spray signals with reflections from the
window. This background noise was eliminated by subtracting the background image as shown in
the image at the bottom-left of Fig. 3.5. The background-corrected images were then converted to
binary images using a threshold-based boundary-detection technique. Specifically, the pixels with
higher intensity than the threshold limit were identified as liquid droplets. An example of the binary
image is shown at the bottom-right of Fig. 3.5. These binary images were used to detect the droplet
boundaries.

For conversion of the background corrected images to binary images, the selection of the
threshold level plays a crucial role. In threshold based methods, the pixels with intensities higher

19
 Background Image Spray Image Prior to Background Correction

Spray Image After Background Correction Binary Image

Figure 3.5: Demonstration of Mie-scattering imaging background correction procedure

than a specified threshold are recognized as pixels with image count 1 and all others with an image
count zero. These pixels with an image count 1 are identified as liquid droplets and others are
treated as background. Therefore, the selection of this threshold level is an important factor to
determine whether a pixel contains a signal from the liquid droplets or not. While the threshold-
based boundary detection was arbitrary by its nature, a systematic approach was formulated to
select the threshold level.

At the top of Fig. 3.6, the binary images obtained using different threshold levels (TL) are
shown. Below the images, a plot of the mean droplet diameter versus the threshold level is also
shown. A clear trend seen from Fig. 3.6 is that as the threshold level increases, the mean diameter
(determined in the image processing) increases, attains a peak, and then decreases. This trend was
due to a trade-off between noise-like small pixels and large droplets. For example, the image
corresponding to the threshold level of 0.05 is very noisy and exhibits small pixels (including single

20
 TL=0.05 TL=0.1 TL=0.25

Otsu’s
Threshold
Level

Figure 3.6: Effect of threshold level on the mean droplet diameter

pixels) that are falsely identified as liquid droplets. A selection of threshold level of 0.1, by contrast,
filters out this noise and shows the highest mean droplet diameter. If the threshold level is further
increased to 0.25, many large droplets, including those near the nozzle are ignored and hence the
mean diameter decreases.

One of the threshold-based methods that are widely accepted is Otsu’s model [Gauding et
al. 2009, Moon et al. 2007]. Otsu’s model [Otsu 1979] selects the threshold level that minimizes the
intra-class variance of the black and white pixels; hence this threshold is a more reasonable choice
than an arbitrary choice of the threshold value [Kook et al. 2011]. One of the built-in functions in
the Matlab software, “graythresh” [Matlab 2009] was developed based on Otsu’s method. This
function calculates the measure of the spread of pixel levels on each side of the threshold value for

21
all the possible threshold values, and then selects the threshold level that generates the minimum
variance within the foreground and background pixels. When Otsu’s method was used for the
sprays in the present study, the mean diameter was rather lower than the peak value as shown in
Fig. 3.6. After assessing plots similar to this figure, the threshold level that gave the maximum
mean droplet diameter for each operating condition was selected. This was well justified as the
selected threshold was optimized in that it is not too low to include noise-like signals and at the
same time not too high to miss out real droplets. This threshold level selected was constant for
different fuels and different ambient conditions tested enabling the comparison of spray
characteristics.

Once the boundaries of liquid droplets were detected, a scaling factor (number of
pixels/cm) was applied to the liquid boundaries detected from the binary images. Then, each droplet
area was used to determine the number of droplets and an equivalent diameter. The averaged value
of equivalent diameter for all droplets in an image was defined as the mean droplet diameter (i.e.
d10 or D[1,0]) and calculated using equation 3.2.
Near-nozzle
region

b
a

Figure 3.7: Boundary detection of fuel droplets and calculation of spray tip penetration. Also
shown is near-nozzle region which masks the omitted spray during the mean droplet diameter
calculation

22
 Ͷ‫ܣ‬௜
σ௜ୀ௡ ට
௜ୀଵ ߨ
ሾͳǡͲሿ ൌ ሺ͵Ǥʹሻ
݊

Where D[1,0] = Mean droplet diameter

Ai = Area of ith droplet
n= Total number of droplets

Fig. 3.7 shows the original spray image with detected liquid droplets boundaries (green)
from binary images. While calculating the mean droplet diameter; it was found that the near nozzle-
region was too dense to distinguish individual droplets. Therefore, up to 15 mm from the nozzle
was ignored for the calculation of mean droplet diameter and number of droplets.

One concern for the spray imaging was image resolution that is limited by the camera
sensor as well as selected field of view. It is simply the higher the better but to capture all spray
plumes in a single image, the image resolution was set to 1600 by 1200 pixel resolution (“standard
resolution”); which corresponds to minimum detectable diameter of 72.2 μm. Figure 3.8 shows how
individual droplets were detected and processed for mean droplet diameter calculation at this
resolution. Various spray regions are selected for the presentation purpose with a variation in axial
location and spray plume as annotated by squares as shown in Fig. 3.7. Figure 3.8 shows some large
liquid droplets occupying multiple pixels as well as many small droplets that correspond to only the

(a) (b) (c)

Figure 3.8: Magnified view for selected spray areas a ~ c in Fig. 3.7. For the presentation
purpose, the colour is inverted so that the liquid droplets are black and the background is white.
Each boxed area has 31 by 31 pixels corresponding to a 2 mm by 2 mm square box in Fig. 3.7

23
 Figure 3.9: Axial slices of 5 mm height (red colour) used to draw axial profiles of number of
droplets and mean droplet diameter later in Figs. 4.4, 4.5, 4.11, 4.18, 4.23 and 4.26

size of a single pixel. It should be noted that the Mie-scattering signal would still be detected
although an actual droplet might be smaller than one pixel resolution and count as one pixel
resulting in over-estimation of the overall droplet size.

Along with mean droplet diameter and the number of droplets, spray tip penetration was
found using Mie-scattering images. The spray tip penetration was defined as the distance between
the nozzle and the liquid droplet which was farthest downstream, as shown in Fig. 3.7. In addition
to the total number of droplets and mean diameter per spray image, the spray image was divided
into 5 mm high slices at different axial stations as shown in Fig. 3.9, and processed individually to
obtain an axial profile of the number of droplets and the mean diameter.

To determine the uncertainty in spray parameters from injection-to-injection variations, an

analysis was performed to quantify these variations. Various numbers of images were imaged and
analysed to calculate the error range of the mean droplet diameter using the following equation 3.3
[Antonius 2004].

ଵǤଽ଺σ
݁ൌ ሺ͵Ǥ͵ሻ
ξ୬

24
 Figure 3.10: The error range calculated for various numbers of individual images (test
conditions selected: ethanol, 3 ms aSOI, 0 bar, 3 ms pulse width)

Where e= Error range

σ= Standard deviation
n= Number of images used to calculate ‘e’

The calculated error range with different number of images is shown in Fig. 3.10. After 25
images, there was no significant change in the calculated error range and hence it was decided to
use 25 images to calculate mean parameters at any time aSOI.

3.2.3 Shadowgraph Imaging

To get the information on transient behaviour as well as the vapour boundary of fuel sprays,
high-speed shadowgraph imaging was also performed [Kook et al. 2011]. A z-type focused
shadowgraph setup was built around the spray chamber for this purpose. Figure 3.11 illustrates this
setup. On the bottom-left corner of Fig. 3.11, the nozzle hole configuration is given (how it affects
the spray field of view will be discussed later in section 4.5).

A continuous xenon-arc lamp (LS150, ABET Technologies) was used as a light source: a
reflector was placed behind the lamp to obtain uniform and high-intensity light, an f/1 condensing
lens was used to focus the light, and a 1-mm-aperture circular slit was placed to spatially-filter the

25
 Parabolic
Mirror (f/4.1) Reflector
Xe-Arc lamp
Condensing
Lens (f/1)
Circular Slit

18°

Glass
Windows Parabolic
Mirror
(f/4.1)
Nozzle Configuration Air
(Bottom View) Blower

Figure 3.11: Schematic layout the Z-type shadowgraph imaging setup

light and eliminate irregular boundaries. The light rays coming from the slit were directed towards
the parabolic lens (aperture of f/4.1 with a focal length of 444.5 mm) and collimated through the
test section. Another similar parabolic mirror was used to refocus the light rays from test section
onto the image sensor of a camera.

To capture the transient behaviour of the fast moving spray, a camera with very high
acquisition rate is needed. A high-speed CMOS camera (Phantom v7.3) was used for the high-speed
visualization. The camera is capable of delivering 190,476 frames per second but this acquisition
rate comes with very limited field of view. To capture the entire spray field, a resolution of 512 by

Table 3.2: Camera settings: Shadowgraph imaging setup

High-speed camera Vision Research Phantom

Frame rate (frames/s) 15,019
Image resolution 512x384
f-number 2
Exposure time 2 μs

26
384 pixels was selected at a framing rate of 15,019 frames per second i.e. 66.6 μs interval between
each frame. The exposure time was fixed at 2 μs to avoid saturation of the bright background as
well as blurring of the fast-moving sprays.

A 50 mm Nikon Nikkor f/1.4 D lens was used with an f-number of 2 selected for
shadowgraph imaging. The camera, light source, circular slit, and parabolic mirrors were positioned
and fixed on optical breadboard using various opto-mechanical mounts. The camera settings for
shadowgraph imaging are summarized in Table 3.2. The high-speed camera was in synchronization
with the injector using LabVIEW control programme.

3.2.4 Image Processing: Shadowgraph Imaging

Background (BG) BG Corrected Image Raw Image with

Raw Image Corrected Image with Spray Boundary Spray Boundary

Figure 3.12: Processing of high-speed shadowgraph images showing temporal evolution of

ethanol sprays between 1 ms and 4 ms after the start of injection (aSOI)

27
 Shadowgraph movie processing is similar to Mie-scattering image processing that starts
with back ground correction. The high-speed shadowgraph movies of sprays were background
corrected using a frame prior to the fuel injection. An example of temporal evolution of typical PFI
sprays is shown in Fig. 3.12. An intensity offset was also added so that the images were shown with
more contrast to highlight features of the spray. Given at the bottom-right of each image is the time
aSOI (the procedure to find the time at which the injection starts is described later in section 3.3.1).
The first column of Fig. 3.12 shows the raw images and the second column shows the background
corrected images. These background corrected images were converted to binary images using
threshold based boundary technique. Different threshold levels were tried; however it was found
that the threshold level has very little impact on spray boundary detection. It was because of a clear
distinction between background and the spray boundaries, and the gradient between the adjacent
pixels near the boundaries of the sprays are sharp. Under these circumstances, the threshold level
had minimum impact in identifying the spray boundaries. Therefore, boundary detection was
performed using Otsu’s algorithm for the threshold selection [Otsu 1979]. The “graythresh”
function available in Matlab image processing software is used for this purpose. This method
chooses the threshold to minimize the intra-class variance of the black and white pixels and hence is
usually a more reasonable choice than an arbitrary choice of the threshold value. The detected
boundaries are overlaid on background corrected spray images and original spray images in 3rd and
4th columns of Fig. 3.12 respectively. These boundaries revealed the vapour-phase spray regions
[Kook et al. 2011], which can complement the liquid boundaries from Mie-scattering images.

3.3 Momentum Flux Measurement

Data
Acquisition
Fuelling and Control
System System
Fuel Injector

Force Sensor

Adaptor

Chamber

Figure 3.13: Spray momentum flux (force) measurement setup

28
 By measuring the force exerted by spray, it is possible to measure the momentum flux and
injection rate delivered by the injector [Naber & Siebers 1996]. Figure 3.13 shows the experimental
schematic of the momentum flux measurement setup. Spray emanating from the injector is
impinged onto a force sensor which measures the momentum flux directly. A force transducer
(Kistler type 9217A) was used to measure the force exerted by the spray and the charge signals
from the force transducer were amplified to voltage values in charge amplifier (Kistler 5015A
charge meter). The charge amplifier was connected to data acquisition system to record the
momentum flux profile from the sensor. Also, given the amount of fuel mass injected per injection,

Anvil Mounted
on Force Transducer

Figure 3.14: Image of anvil when it was mounted on force transducer

Figure 3.15: Force transducer with adaptor installed in the spray chamber

29
the injection rate (mass, g/s) was calculated. To avoid thermal drift in the signals [Gehmlich 2009],
an adaptor with an anvil was designed [Liu 2010]. The anvil was installed to avoid the direct
contact of the fuel with the sensor as well as to increase the fuel contact area as shown in Fig. 3.14.

Figure 3.15 shows the force transducer when it was fixed with the adaptor and installed in
the spray chamber during the measurements. An opto-mechanical fixture (Edmund Optics Inc.) was
used to hold the adaptor and to mount within the spray chamber. The force sensor was positioned
vertically so that the spray was injected normal to the sensor. The sensor was placed very close to
the nozzle (4.1 mm downstream) to ensure that the area of spray impingement was less than the
sensor facial area so that all spray droplets were measured.

3.3.1 Data Processing: Momentum Flux

The momentum flux data for ethanol together with an injection trigger signal are shown in
Fig. 3.16. The injection trigger signal was a square-shaped electronic pulse created in the control
system, which drove a solenoid valve of the PFI injector. To account for injection-to-injection
variation and experimental uncertainty, 30 repetitive injections were measured and ensemble-
averaged. Similar to Fig. 3.10, a sensitivity study was performed to select the number of injections
for ensemble averaging. To this end, the start of injection was selected to assess the cycle-to-cycle

1.73 ms

Figure 3.16: Force profile of the spray measurement. Also shown is the PFI injector driver
signal

30
 Figure 3.17: The variation of error range for start of injection (SOI) calculated from various
numbers of individual injections

variations. The error range of the start of injection was calculated using an equation similar to 3.3
for various numbers of fuel injections as shown in Fig. 3.17. The figure shows that after 30
injections, there was no significant variation in the error range of start of injection. Therefore, 30
injections were implemented for each operating condition of this study for ensemble averaging. An
ensemble averaged momentum flux is shown as a dashed line in Fig. 3.16. In the averaged signal,
there are very high frequency fluctuations present throughout the injection. This momentum flux
trace required a numerical smoothing due to high frequency noise, particularly in the early stage of
fuel injection. The profile was smoothed using the “Moving Average” method [Matlab 2009] with
the window size of 50 points (corresponding to 0.42 ms). It was found that the effect of smoothing
was minimal on calculated force or injection rate. It however affected the start of injection and
therefore raw data was used for calculation of the start of injection. The smoothed momentum flux
profile is shown as a solid line in Fig. 3.16.

The first noticeable feature of Fig. 3.16 is a delay between the electronic trigger and the rise
of momentum flux. This was due to the time required for the solenoid valve to lift the needle
against the spring force [Passarini & Nakajima 2003]. Also, as mentioned above, there was a gap
between the nozzle and force sensor, which further increased the delay. As a result, the total delay
time was measured about 1.73 ms regardless of fuels or injectors used in this study. By subtracting
the time required for the spray to travel the gap between the injector and the sensor, the actual start

31
 Actual Start of Injection
Determined from Spray Movie

Solenoid
Delay

Delay Due to
Nozzle-sensor Gap

Figure 3.18: Injection rate calculated from the momentum flux profile. Also shown is the PFI
injector driver signal

of injection after triggering was obtained. It was found that a solenoid delay for the injectors used
was 1.5 ms, which was consistent with the observations from the high speed movies. Hence, this
time was referred as actual start of injection. In the following sections, all the spray events are
referred with respect to actual start of injection rather than the electronic triggering time.

From the momentum flux data the injection rate (mass flow-rate, g/s) can be calculated
using the mass per injection [Naber & Siebers 1996]. The procedure used to calculate injection rate
is given in Appendix A2. An example of the calculated injection rate (g/s) is shown in Fig. 3.18.
The time axis also updated with the actual start of injection.

3.4 Engine Test Facility

A new test-cell was built in the Engine Research Laboratory, UNSW for the dual-fuelling
engine performance and emissions tests. A single-cylinder diesel engine, eddy current
dynamometer, intake air system, exhaust system, emission analysers, and engine cooling system
were built. Figure 3.19 shows the photo of engine test-cell with major components listed.

32
 A single-cylinder, automotive-size, naturally aspirated diesel engine was used to study
ethanol and diesel dual-fuelling. The engine was modified from the 4 cylinder version of 2-litre
diesel engine and entire engine crank case was rebuilt to suit for the single cylinder application.
Since a single-cylinder engine was used; pressure fluctuations in the intake and exhaust pipes were
identified as a potential issue. To minimize them, large-volume (60 times the displacement volume

Exhaust Surge Tank Intake Surge Tank

TDC Encoder

EC Dynamometer

PFI Fuel Injection System

Engine Coolant
Temperature Controller

Figure 3.19: Photo of engine mounted on the engine bed in test-cell

33
 Exhaust
Smoke Gas Sampling
Op a c i t y % Diesel
Common-rail
Exhaust Intake
Opacimeter Injector
Surge Surge Ethanol
Tank Tank Port Fuel Injector

Control and
Data Acquisition
NOx y y y ppm HC y y y ppm System
Pressure
NO y y y ppm CO yyy % Sensor
NO x Analyser HC/CO Analyser

Coolant Temperature
Controller

Figure 3.20: Schematic diagram of the dual-fuelling engine setup

of the engine) surge tanks were placed in both intake and exhaust sides. Figure 3.19shows the surge
tanks mounted on the aluminium profiles. The engine has 83 mm bore and 92 mm stroke with a
compression ratio of 17.7. The combustion chamber is formed by a flat fire-deck with two intake
valves and two exhaust valves, a cylinder liner, and a piston with cylindrical bowl. Experimental
setup of the engine is illustrated in Fig. 3.20.

The engine is coupled to an eddy current dynamometer (Froude Hoffmann AG-30HS) by a

drive shaft with universal joints on both ends. The dynamometer used has a rating of maximum
power of 30 kW, and a maximum torque of 95 Nm between the speeds of 2000 to 3000 rpm (the
torque-speed map of the dynamometer is given in Appendix A3). The engine delivers a maximum
torque of 30 Nm at 2000 rpm, hence this dynamometer can well control the speed and torque of the
engine within the measurement accuracies. The heat generated from engine and dynamometer was
dissipated by circulating cooling water through two separate water circuits. A separate water pump
circulates the coolant to the dynamometer; hot return coolant from the dynamometer flow through a
heat exchanger where it dissipates the heat to external flowing water. The engine coolant
temperature was controlled by a coolant temperature controller (ThermalCare Aquatherm RA092).

34
The temperature of the coolant circulating through engine head and crank case can be fixed at set
temperature with an accuracy of ± 1°C.

The engine specifications are summarized in Table 3.3. Diesel fuel injection system
comprises of a low pressure electric pump, cam driven high pressure fuel pump, common-rail and a
solenoid-type 7-hole injector. The injector was centrally mounted at the engine head. Each hole has
a nominal outlet diameter of 134 μm and is hydro-eroded and tapered with a K-factor of 1.5 (K =
(inlet diameter – outlet diameter)/10), according to the manufacturer’s specification. The nozzle has
a discharge coefficient of 0.86 and hydraulic flow rate (HFR) of 400 cc measured using 10 MPa
pressure drop for 30-second injection duration. The included angle between holes is 150o. To
operate the engine in a dual-fuelling mode, ethanol injection system was installed in the intake
manifold. The same PFI injector that was used for spray studies was fitted (i.e. Bosch EV6 with 6
holes running at 300 kPa injection pressure). The location of the injector in the intake system was
varied; one at a distance of 11 cm from the intake valves targeting the engine intake ports, and the
other 32 cm from intake valves and at the elbow connecting the manifold and the intake surge tank.

Prior to running the engine tests, the diesel injector was calibrated for the fuel flow rate
measurements. The calibration data i.e. amount of fuel injected per injection was conducted in a test
bench but at the common-rail pressure of 130 MPa and using the same injector. Later, this
calibration was further validated using the injection rate measurement as shown in Appendix A4.
For ethanol injection, the calibration data provided by the injector manufacturer [Bosch 2012] was
used due to the challenge of measuring the injected mass of volatile ethanol using a port-fuel
injector.

3.4.1 In-cylinder Pressure and Emissions Measurement

The crankshaft position was measured by an incremental optical encoder connected to the
crankshaft on a mounting fixture. The TDC position was pegged based on the location of the peak
cylinder pressure measured during the cranking. Figure 3.19 shows the mounted encoder. This
encoder sends the signals through two channels. One channel gives a transistor-transistor logic
(TTL) pulse for every revolution and second channel gives a total of 1800 pulses per revolution i.e.
one pulse for every 0.2°. These signals are fed to injection controller for accurately timing the
injection signals to be sent for ethanol and diesel injectors.

35
 Table 3.3: Engine specifications

Displacement 497.8 cc
Bore 83 mm
Stroke 92 mm
Compression ratio 17.7
Number of valves 2 intake and 2 exhaust
Piston Cylindrical bowl
7-hole Bosch common-rail
Nominal diameter: 134 μm
Diesel direct-injection injector K-factor: 1.5
Discharge coefficient: 0.86
HFR: 400cc for 30s
Included angle: 150°
6-hole Bosch EV6
Ethanol port-fuel injector Flow-rate: 261 g/min
Spray angle: 70°

In-cylinder pressure measurements were carried out using a piezo-electric pressure

transducer (Kistler 6056A). The pressure sensor has a glow plug type adaptor so that was installed
using an existing hole of the cylinder head. The charge signal from the sensor was taken to charge
amplifier (Kistler 5015A) which converts the charge (coulombs) to voltage (volts) to get the in-
cylinder pressure in MPa/voltage. Intake air temperature was monitored by a K-type thermocouple
inserted in the intake manifold of the engine.

The exhaust gases were sampled after the exhaust surge tank in the tailpipe for the analysis
of HC, CO, NOx and smoke emissions. The HC and CO emissions were measured using a non-
dispersive infrared analyser (Horiba MEXA 584L). NOx emissions were obtained using a
chemiluminescence-type NOx analyser (Ecotech 9841AS). HC/CO and NOx analysers were
calibrated at manufacturer’s site prior to the measurement. An opacimeter (Horiba MEXA 130S)
was used to measure the exhaust smoke level. This instrument was calibrated prior to the every set
of experiments.

36
3.4.2 Data Acquisition and Controlling System

Dynamometer was controlled by Texcel V4 (Froude Consine, UK) dynamometer controller.

Dynamometer was calibrated frequently with known weights to correct for any errors between
actual torque and indicated torque levels. This dynamometer is capable of maintaining the engine
either in constant speed mode or constant torque mode. During the all sets of experiments presented
in this thesis the constant speed operation was used.

The diesel injection timing, duration and pressure were controlled independently using an
electronic injection controller (Zenobalti injection and pressure control valve drivers). This
controller facilitates three individual channels to control different injectors/devices. First channel
was used for control of in-cylinder diesel injector. Second channel was used to drive the Drivven
PFI injector driver which controls injection timing and duration of PFI injector. The referencing
signal for timing of injection events was taken from TDC encoder.

A data acquisition system (MC USB-1616HS-BNC) was used facilitating simultaneous

acquisition of 16 digital analog inputs at a frequency of 100 kHz. This system simultaneously
records the data from crank shaft position sensor, engine pressure sensor, and the signals from
injection controller. A real time Matlab GUI (graphical user interface) programme was used for
synchronizing the injection controller and data acquisition system. Matlab programme allows the
operator to monitor the pressure trends and indicated mean effective pressure (IMEP) values in real
time and facilitates recording and saving the data.

3.4.3 Data Processing

The data acquisition system records the data from the in-cylinder pressure sensor in volts.
The voltage signal is converted to physical units in Matlab script. Since the piezo-electric pressure
transducer was used, the measured pressure is relative pressure (pressure differential) and needs to
be pegged to obtain an absolute pressure. The pressure during intake stroke i.e. between the crank
angles (CA) of 90° to 135° after intake TDC was averaged and taken as a reference pressure at 100
kPa, and the offset was added to the entire pressure trace to get the absolute pressure. The indicated
work was calculated from each cycle by calculating area under pressure curve using the equation
3.3. IMEP from each cycle is calculated by dividing the indicated work by swept volume, using
equation 3.4.

37
 ଷ଺଴ι
†‹ƒ…–‡†‘” ൌ න  ‫ † כ‬ሺ͵Ǥ͵ሻ
ିଷ଺଴ι

†‹ƒ…–‡†‘”
 ൌ ሺ͵ǤͶሻ
™‡’–‘Ž—‡

Even though all the experiments were conducted at controlled operating/boundary

conditions, there were inherent cycle-to-cycle variations in measured pressure traces. These are due
to various reasons such as random fluctuations in equivalence ratio, unsteady nature of turbulent
flows [Daily 1987], variations in the injection rate and injection pressure between different cycles
[Zhong et al. 2003]. To account for these variations, it is apparent to collect the data for number of
cycles for the same set of operating conditions. To quantify the cycle-to-cycle variations, the
coefficient of variation of IMEP (CoV of IMEP) was calculated using the following equation 3.5
[Heywood 1988].

ටͳ σ୧ୀଵ ሺ୧ െ ୟ୴୥ ሻଶ

 ୧ୀ୬
‘୍୑୉୔ ൌ ሺ͵Ǥͷሻ
ୟ୴୥

Where CoVIMEP= coefficient of variation of IMEP

n=number of cycles
IMEPi=IMEP of ith cycle

Figure 3.21: The variation of CoV of IMEP calculated from various numbers of individual
cycles

38
 IMEPavg=Average IMEP of ‘n’ cycles

Various numbers of cycles (n) were tested to determine the enough number of cycles for
CoV of IMEP calculation. Figure 3.21 shows that if 80 cycles or higher are used the fluctuations of
CoV of IMEP is not significant. Throughout the study, 100 cycles were used for data analysis.
Figure 3.22 shows an example of pressure traces from 100 instantaneous consecutive firing cycles
drawn against crank angle degrees. An ensemble averaged pressure trace was calculated from the
individual pressure cycles representing that particular engine operating condition. The ensemble
averaged pressure trace calculated from the 100 pressure traces is shown as a dotted line in Fig.
3.22. This averaged pressure trace represents the actual engine operating condition.

From the averaged in cylinder pressure trace, apparent heat release rate (aHRR) was
calculated from one dimensional heat release rate model using the equation 3.6 [Heywood 1988].

݀ܳ௔௣௣ ߛ ܸ݀ ͳ ݀ܲ
ൌ൬ ൰‫݌‬ ൅൬ ൰ ܸ ሺ͵Ǥ͸ሻ
݀ߠ ߛ െ ͳ ݀ߠ ߛ െ ͳ ݀ߠ

where Qapp is the apparent heat release, θ is the crank angle degree, γ is the specific heat ratio, P is
the measured in-cylinder pressure at a given θ, and V is the volume of combustion chamber at angle

Figure 3.22: Example of instantaneous in-cylinder pressure traces from 100 firing cycles and
ensemble averaged pressure trace

39
 Figure 3.23: An example of measured in-cylinder pressure and calculated apparent heat release
rate

θ. The ratio of specific heats γ is assumed as 1.35 [Heywood 1988]. The heat transfer to the walls
was ignored in calculating the heat release rate. The calculated heat release rates are very noisy
particularly after the start of combustion. This is because of the pressure gradient terms involved in
heat release rate calculation (equation 3.6). These traces require further filtering to draw meaningful
conclusions. A second order Butterworth filter with normalized cut-off frequency of 0.1 is applied
(more details on the selection of cut-off frequency are given in Appendix A5). Figure 3.23 shows an
example of averaged in-cylinder pressure trace (top) and the filtered apparent heat release rate
(bottom).

From the aHRR plots, the ignition delay was calculated which is defined as the time
difference between the start of diesel injection (SOI) and the crank angle of 10% heat release (i.e.
CA10). The burn duration for the initial stage of the combustion was calculated by reading the
difference between CA10 and the crank angle of 50% heat release (CA50). Similarly, a difference
between CA50 and CA90 was calculated for the late-cycle burn duration.

40
Chapter 4

Effect of Injection and Ambient

Conditions on PFI Sprays

In this chapter, the influence of different injection and ambient conditions on ethanol PFI
spray characteristics is presented. A detailed investigation of ethanol spray characteristics such as
spray tip penetration, total number of droplets, droplet diameter, axial distribution of droplet
number and droplet diameter is conducted. It is followed by results from experiments varying
different injection and ambient conditions. The details of the impact of the injection and ambient
conditions on sprays are discussed with ethanol data but at the same time, any differences in spray
behaviour associated with fuel properties are highlighted by comparing the ethanol and gasoline
sprays.

A list of the fuel physical properties is given in Table 4.1. Compared to gasoline, ethanol
has lower vapour pressure and higher density, surface tension, flash point, and heat of vaporization.
Also notable is that gasoline shows a range of boiling points reflecting that it is a multi-component
fuel. It is expected that some components (likely high-carbon components) had much higher boiling
point than ethanol.

41
 Table 4.1: Fuel properties

Gasoline Ethanol
Vapour pressure (kPa) 45-90 18
Density (kg/m3) 720 794
Viscosity (Pa-s) 0.00042 0.0012
Surface tension (N/m) 0.0189 0.02205
Flash point (°C) -43 15
Heat of vaporization (kJ/kg) 310 904
Boiling point (°C) 30-200 78.5

4.1 Momentum Flux and Injection Rate Measurements

The momentum flux was measured for gasoline and ethanol sprays using the experimental
facility as described in section 3.3. An ensemble averaged momentum flux is shown as a dashed
line in Fig. 4.1 for gasoline and ethanol. The smoothed momentum flux profile is shown as a solid
line. The momentum flux shows a linear increase soon after the start of injection, which lasts for
about 0.3-ms. Soon after this linear increase, the momentum flux is near constant until the end of
injection (i.e. a quasi-steady period of injection existed). Similar to the start of injection, there is
also a delay for the end-of-injection. The down-edge of the trigger signal occurred about 1 ms prior
to the decline of the momentum flux. However, this delay was shorter than the start-of-injection
delay because the closing of the solenoid valve was assisted by the spring [Passarini & Nakajima
2003]. Overall, a top-hat-shaped profile was obtained for the PFI injector used in this study.

An interesting observation from Fig. 4.1 is an initial spike seen right after the start of
injection. This markedly high momentum flux is much clearer in unsmoothed, raw signals. This is
likely due to an overshoot in the force sensor, which is evidenced by a large fluctuation in the
signal. However, there was also a possibility that the fuelling rate was higher during the initial
transient and this larger mass of fuel hit the sensor immediately after the opening of the injector.
Considering the small volume inside the nozzle where fuel is stored prior to the injection, the higher
fuelling rate during the needle opening should contribute to the high spike of the momentum flux.
How this initial spike in the momentum flux impacts the spray penetration and droplet size will be
discussed in the following sections. After this initial transient, however, little difference was found
in the momentum flux between gasoline and ethanol. As the momentum flux is constant for both

42
 Solenoid
delay

Delay due to nozzle-sensor gap

Figure 4.1: Momentum flux and mass flow-rate of the PFI injector for ethanol and gasoline fuels

fuels, any differences found in spray structures to be discussed will therefore be primarily caused by
fuel physical properties rather than momentum-driven variations.

In Fig. 4.1, below the momentum flux plot, mass flow-rate is also shown. An expected
trend is found that at the same injection pressure, ethanol shows a higher mass flow-rate than
gasoline due to the higher fuel density. Roughly 9% higher injection rate was measured for ethanol,
which corresponded to 9% higher fuel density. It is notable that the ramp-up, ramp-down and actual
duration of the injection rate show no difference between two fuels but lower injection rate for
gasoline. More details about the differences in spray structures with change in fuels will be
discussed later in section 4.3.

43
4.2 Ethanol Spray Penetration and Droplet Diameter Results

Typical PFI spray images of this study are shown in Fig. 4.2 for ethanol fuel. Shown in the
top-right of each image is time after the start of injection (aSOI). A long injection of 8 ms was
chosen so that the end of injection occurs after the spray tip leaves the camera field of view. This
represents visualisation of a quasi-steady spray i.e. the injection period of a constant injection rate.
The same injector that was used in momentum flux results of previous section was used. As
mentioned earlier, the injector was equipped with a 6-hole nozzle and to obtain maximum number
of droplets in the camera view-field, it was oriented such that three sprays are seen in each image.
Notable was that, despite line-of-sight integrated images, individual droplets were discernable. This
enabled detection of the boundaries of the liquid droplets, as overlaid onto the binary images in Fig.
4.2. As mentioned previously, a problem in the near-nozzle region was that the droplets were too
dense to process. Therefore, the spray tip penetration and droplet diameter were determined only
downstream of this region, excluding the near-nozzle sprays.

Spray tip penetrations obtained from processed spray images are shown in Fig. 4.3 for
various times aSOI. An error range including injection-to-injection variation and experimental
uncertainty is also given. Figure. 4.3 shows that the spray tip penetration increases linearly with
increasing time aSOI. Note that the period 2 to 5 ms aSOI corresponds to the quasi-steady period of
injection and therefore a linear increase in the tip penetration is expected. Below the spray tip
penetration, the mean droplet diameter along with the total number of droplets is also shown in Fig.
4.3. The droplet diameters found in this study were higher than the data available in the literature.
For example, one study reported about 100 μm of sauter mean diameter (SMD) from Phase Doppler

2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Figure 4.2: Evolution of ethanol spray at various time aSOI. The timing at which the image was
taken is shown at the top of each image

44
 Number of
Droplets

Mean
Droplet Diameter

Figure 4.3: Spray tip penetration (top), mean droplet diameter and number of droplets (bottom)
against time aSOI for ethanol fuel

Anemometry (PDA) measurements [Zhao et al. 1995]. There is another study suggesting 120-160
μm of SMD [Anand et al. 2010b] from shadowgraph images. Considering arithmetic mean diameter
of this study was typically 125 μm, the droplet size measured was higher than the previous studies.
This might be due to differences in diagnostics but there are also variations in operating conditions.
For instance, Zhao et al. [1995] used various injection pressures (250-800 kPa) and a pintle-type
nozzle as opposed to the multi-plate orifice in the present study. Considering these differences, the
droplet diameter measured in the present study appears to be reasonably close to those measured in
prior works.

In Fig. 4.3, the mean droplet diameter shows a decreasing trend with increasing time aSOI.
By contrast, the total number of droplets increases while the mean droplet size decreases. There
were two causes for this increase of total droplet numbers: one was an increase of the region of

45
space occupied by the spray and the other was breakdown of droplets (i.e. atomization). While the
continued injection of fuel increased the spray region and the total number of droplets, a breakdown
of big droplets into small droplets would also increase the droplet numbers. Therefore, whether this
breakdown is due to the secondary (droplet) break-up and atomization or not would be the first
question to ask. One consideration is that, in low Weber number flows, it is known that the
aerodynamic interaction between the surrounding gas and the flowing liquid is very low and
therefore the secondary break-up is very weak compared to initial break-up [Dumouchel et al.
2005]. Hence, in the present ethanol sprays, it is believed that the droplet break-up played only a
minor role in the increase of the number of droplets. In other words, the increase of the total droplet
numbers is primarily due to the addition of droplets from continued fuel injection. With no droplet
break-up and atomization, however, the decreasing mean droplet diameter is not explained.
Obviously, the mean diameter should not decrease but stay constant if no other causes exist.

To discuss further details of ethanol droplets, an axial distribution of the droplet numbers
was determined by taking a 5 mm slice of sprays along the injector axis similar to the Fig. 3.9.
Figure 4.4 shows these results for each time aSOI of 3, 4, and 5 ms. An immediate conclusion from
Fig. 4.4 is that with increasing time aSOI, the number of droplets increases. Throughout the axial
locations, sprays at 5 ms aSOI show higher droplet numbers than those at earlier times aSOI,
consistent with the total number of droplets in Fig. 4.3. Once again, the spray length is longer for

Figure 4.4: Axial profile of droplet number along the injector axis for various times aSOI for
ethanol sprays

46
later times aSOI and therefore the integration of this axial profile (i.e. total droplet numbers) must
increase. However, the figure shows that the number of droplets increases with increasing time
aSOI at any fixed axial distance from the nozzle. For example, at 40 mm from the nozzle, the
droplet numbers at 3 ms aSOI is about 200. At the same location but 1 ms later (4 ms aSOI), the
number of droplets is doubled. The mass flow-rate of the fuel was fixed during this injection period.
Therefore, this was strong evidence that the droplet break-up did occur. The figure also supports the
possibility that liquid droplets evaporated while penetrating downstream. At fixed time aSOI, the
number of droplets decreases with increasing distance from the nozzle, which should be explained
by evaporation of small droplets. Since Mie-scattering diagnostics only detects liquid-phase sprays,
evaporation of small droplets would result in decreased number of droplets. Clearly, both droplet
break-up and evaporation impacted the ethanol PFI sprays.

Figure 4.5 shows the axial profile of the mean droplet diameter for the same times aSOI of
Fig. 4.4. The figure shows that at 3 ms aSOI when the axial spray penetration is about 42 mm, the
mean droplet diameter is fairly constant up to 30 mm from the nozzle. Further downstream,
however, a rapid increase in the mean diameter was observed which continued to the spray tip. This
trend was also found from the other times aSOI. The evaporation of small droplets likely caused
this trend. Specifically, if very small droplets contributing to a low mean diameter in the upstream
region evaporated first while they travelled across the chamber, the mean diameter in the
downstream region should increase.

Further investigation of droplet diameter distribution at a fixed distance from the nozzle
supported this idea as shown below the mean diameter plot in Fig. 4.5. This plot shows the droplet
size distribution at 5-mm-slice locations of 32.5 mm and 47.5 mm from the nozzle at fixed 4 ms
aSOI. To illustrate, circle symbols are provided on the mean droplet diameter trace for each of these
locations. In the size distribution plot, it is clear that a much higher number of small droplets exist
for the 5 mm slice at 32.5 mm downstream compared to 47.5 mm. This means that indeed, small
droplets evaporated earlier than big droplets and therefore the population of droplets is skewed
towards the larger droplet sizes. The result was increased mean droplet diameter with increasing
distance from the nozzle. Back to the mean diameter plot in Fig. 4.5, another interesting finding is
that the high mean diameter at the spray tip continues from 3 to 5 ms aSOI. For example, the peak
value in the mean droplet diameter was 0.147 mm at 3 ms aSOI and it gradually increased to 0.153
mm at 5 ms aSOI. A likely cause for this behaviour was the start-of-injection transient. In Fig. 4.1, a
markedly high momentum flux (and mass flow-rate) was measured during the initial transient.
Therefore, the droplets injected in the early stage had higher momentum and mass than those

47
 32.5mm away from nozzle

47.5mm away from nozzle

Figure 4.5: Axial profile of mean droplet diameter for various times aSOI (top) and droplet size
distribution at 4 ms aSOI for different axial locations (bottom)

injected during the quasi-steady period of injection. This potentially made the liquid droplets
survive through break-up or evaporation while travelling downstream. Indeed, Fig. 4.5 shows that
the high mean diameter at the spray head region shifts downstream as the injection continues while
no significant change in the mean diameter is found. This is also consistent with a linear increase in
the spray tip penetration (Fig. 4.3) where no deceleration is observed despite evaporation of
droplets.

Finally, the decreasing mean droplet diameter with increasing time aSOI (Fig. 4.3) is also
clarified using an observation from Fig. 4.5. For instance, from 3 to 5 ms aSOI, not only the high

48
diameter region shifts downstream but also the upstream region with a near-constant and the low
diameter region extends as the spray penetration continues. In fact, this “upstream” droplet diameter
of about 0.12 mm at 3 ms aSOI declined to 0.116 mm at 5 ms aSOI because droplets injected after
the initial transient (or during the quasi-steady period) experienced significant break-up. Certainly
there was no decrease of droplet size at the spray head region; however, an extension of the low
diameter region in the upstream resulted in the decreasing mean droplet diameter with increasing
time aSOI.

4.3 Effect of Fuels

While the ethanol spray results in the previous section provide valuable information to
further improve PFI sprays in ethanol-compatible SI engines, how ethanol sprays perform
differently from gasoline sprays will also be important to accelerate the market penetration. To
address this issue, the same imaging diagnostics were conducted for gasoline and a 50% blend of
gasoline and ethanol (E50).

Firstly, the spray tip penetrations for gasoline, ethanol, and E50 are shown in Fig. 4.6. A
well-expected trend is observed: all fuels tested show a linear increase in the spray tip penetration
with increasing time aSOI. However, the penetrations are different between fuels. While gasoline
and E50 have similar tip penetration, ethanol shows lower penetration than the others for the entire

Figure 4.6: Spray tip penetration against time aSOI for gasoline, E50, and ethanol

49
injection duration. Note that the momentum flux was found to be the same for all fuels (Fig. 4.1)
and therefore the penetration should be the same if the momentum transfer solely drove the spray
penetration. One might expect this discrepancy could be explained by different evaporation rates for
different fuels. The E50 is known to have similar vapour pressure with gasoline [Pumphrey et al.
2000] and therefore the same tip penetration was expected. However, ethanol should have higher
liquid penetration due to its lower vapour pressure than gasoline and E50. Despite these
considerations, ethanol shows lower tip penetration than gasoline and E50.

Previous studies [Gao et al. 2007, Park et al. 2009, Wang et al. 2005] suggested that a cause

Figure 4.7: Mean droplet diameter versus time aSOI for gasoline, E50, and ethanol (top) and
number of droplets plot for the same conditions (bottom)

50
for similar observations in gasoline direct-injection (GDI) sprays could be due to the multi-
component nature of conventional fuels. Gasoline is a mixture of number of different hydrocarbons
and its evaporation rate varies as the evaporation progresses [Perry & Gee 1995]. It is likely that the
higher vapour pressure of gasoline was mainly contributed by its lower carbon-number components.
These light components would evaporate faster than heavy molecules. By contrast, ethanol was a
single-component fuel, which could evaporate earlier than the heavy components of gasoline. The
tip penetration of E50 supports this view in that heavy molecules included in E50 dominate the
spray tip penetration and hence the same penetration distance with gasoline.

The mean droplet diameter and number of droplets also show a similar behaviour. Fig. 4.7
shows that the mean droplet diameter decreases and the droplet number increases with increasing
time aSOI for all fuels investigated. This was expected due to the same reasons found in Fig. 4.3.
However, increasing the ethanol content caused a decrease in the mean droplet diameter and an
increase in the droplet number at any fixed time aSOI.

One may argue that a smaller ethanol droplet size than gasoline might not be due to fuel
effects but a potential artefact in imaging technique. For example, there could be an issue of line-of-
sight integrated droplets that behaved differently for gasoline and ethanol. Also, extremely small
droplets that were below the detection-limit (minimum detectable diameter of 72.2 μm) could have
caused an opposite trend. To further clarify these issues, a small window of 15 mm by 5 mm shown
as a rectangle in Fig. 4.8 was selected for droplet processing. An additional Mie-scattering imaging
was performed that had this small window as a full view-field (i.e. 1532-by-512 pixels for the
selected region) enabling a lower detection-limit of minimum detectable diameter 11.2 μm.

As mentioned earlier, the lens used for the standard setup was a 50 mm Nikon Nikkor lens
and an aperture opening of f/8 was used. When the low detection-limit imaging was performed, the
camera was moved closer to the spray chamber and used a 105 mm Nikon Nikkor lens with f/5.6.
This reduced the view field and therefore only a portion of the spray region was imaged resulting in
higher pixel resolution per spray region (i.e. a lower detection-limit). Also, this would decrease the
depth of field so that only a slice of the droplet group to be photographed, reducing the chance of
droplet overlapping in the images. The results are shown in Fig. 4.9.

On the top, the extracted small window from the full spray image (i.e. the standard image)
is presented in conjunction with the low-detection-limit image in the middle for ethanol sprays. The
droplet diameter detection-limit is given at the top-left of each image. On the bottom, the droplet
size distribution from an ensemble average of 25 different injections is plotted for two different

51
 Figure 4.8: The rectangular window shown is to be used to discuss the droplet size detection-
limit in Fig. 4.9

detection limits and for both ethanol and gasoline. Also shown are cumulative plots of the droplet
diameter.

Figure 4.9 shows that the low detection-limit image includes very small droplets that cannot
be seen in the standard image, as expected. In the standard image with a higher detection-limit,
these smaller droplets would have been fitted into the 70 μm bin. Note that the cumulative droplet
number is different for different detection-limits. If all small droplets detected in the low-detection-
limit image were also detected in the standard image but simply misinterpreted their diameter, the
cumulative count must be the same. However, the results indicated that a different number of
droplets was detected depending on the detection-limit. This was due to variations in focal-length of
the lens. In other words, a higher focal length was used for the low detection-limit image, which
decreased the depth of field and hence visualized a narrower region of the spray. Therefore, care
should be taken to use an absolute value of the mean droplet diameter obtained from the Mie-
scattered spray images. However, this did not necessarily mean that the small droplet diameter trend
for ethanol was a misinterpretation.

52
 Droplet Diameter
Detection Limit: 72.2μm

Droplet Diameter
Detection Limit: 11.2μm

Ethanol:
High Detection Limit

Gasoline:
High Detection Limit

Ethanol:
Low Detection Limit

Gasoline:
Low Detection Limit

Figure 4.9: Images showing a selected spray region from the standard image (high detection-
limit) of Fig. 4.8 (top) and the same region with low detection-limit (middle) for ethanol sprays.
The bar graph at the bottom shows droplet size distribution for both standard and low detection-
limit images of gasoline and ethanol sprays. Solid and dashed lines are drawn for cumulative
number of droplets for ethanol and gasoline, respectively. Data was accumulated for 25
injections and the camera was triggered at 4 ms aSOI

53
 The results in Fig. 4.9 show that the droplet diameter of ethanol is smaller and the number
of droplets is larger than that of gasoline even in the data from the low detection-limit images. In
fact, the accumulated droplet number plots in Fig. 4.9 shows the droplet numbers for ethanol are
higher than that of gasoline during the entire range of droplet diameters. Higher number of droplets
seen in ethanol low detection image is because of better sensing of smaller droplets which is
otherwise not possible to see in standard image. Therefore, it is demonstrated that the interesting
observation in the ethanol droplet size was due to fuel effect than experimental discrepancy that
could be caused in Mie-scattering imaging. Furthermore, note that to study the spatial variation of
the fuel droplets, it is imperative to image full sprays in a single shot. Therefore, at the expense of
the detection-limit, full-spray standard images are used for following discussions.

To further clarify the smaller droplet diameter for ethanol, a close-up view of the sprays
near the nozzle was reproduced from the standard images as shown in Fig. 4.10. The images were
taken at 4 ms aSOI, a long time after the initial transient, to avoid large droplets with markedly high
momentum flux that resulted in very high droplet diameters at the spray tip (Fig. 4.5). Once again,
the close-up imaging of the near-nozzle sprays reveals that ethanol has smaller droplets (better
initial break-up) compared to gasoline or E50. Similar results were obtained by other researchers
[Tian et al. 2010, Park et al. 2002], where gasoline sprays show higher droplet diameter than
ethanol near the direct-injection injector nozzle. Consistent with this visual inspection, an axial
profile of the mean droplet diameter in Fig. 4.11 shows distinctively smaller near-nozzle droplets
for ethanol.

Figure 4.10: Close-up view of standard spray images (droplet diameter detection-limit: 72.2 μm)
for gasoline, E50 and ethanol at 4 ms aSOI

54
 Figure 4.11: Axial profile of mean droplet diameter along the injector axis at 4 ms aSOI for
gasoline, E50, and ethanol. The Mie-scattered spray images similar to Fig. 4.2 are used to obtain
5 mm spray slices in the axial direction (definition given in Fig. 3.9)

For example, at 17.5 mm downstream of the nozzle, ethanol shows only 0.12 mm in the
mean droplet diameter, which stays nearly constant up to 40 mm from the nozzle. By contrast,
gasoline shows about 0.135 mm (12.5% higher) in the mean diameter at the same location and then
decreases with increasing axial distance from the nozzle. This decrease was well explained by the
droplet break-up and atomization as in Fig. 4.5. The mean diameter of gasoline droplets eventually
reaches a diameter equivalent to ethanol and thereafter increases at the same rate due to the fuels
injected during the initial transient, again similar to Fig. 4.5. No significant differences between
fuels are seen (or all are within the uncertainty range) for this downstream spray region. Therefore,
Fig. 4.11 in conjunction with images in Fig. 4.10 suggests that the lower mean diameter of ethanol
than that of gasoline was due to initial break-up that was higher for ethanol. Once again, this was
likely because heavy molecules in gasoline cause lower initial break-up compared with ethanol.

Another interesting finding from Fig. 4.11 is that E50 shows the same mean droplet
diameter as that of gasoline at 17.5 mm from the nozzle. If physical properties of gasoline/ethanol
blend vary with the blending ratio (by volume), the mean diameter for E50 at this location should be
in the middle of gasoline and ethanol. Instead, presence of heavy gasoline components in E50
resulted in the same mean droplet diameter. This result is also consistent with the previous

55
discussion regarding the effect of heavy molecules on the tip penetration, mean droplet diameter,
and number of droplets.

4.4 Effect of Injector Flow-rate

Although a direct comparison between ethanol and gasoline sprays provided valuable
information, previous experiments were performed at limited injection conditions. For example, the
nozzle-hole size was fixed although a smaller nozzle-hole size would result in smaller fuel droplets
near the nozzle, which would impact droplets evaporation process significantly. To understand the
effect of nozzle hole diameter on the PFI sprays, two different injectors with different flow-rates
were investigated. An additional injector (high flow-rate injector) was selected with higher flow-
rate than the injector used for previous studies (sections 4.1~4.3). The selected injectors have the
same spray angle but different flow rates facilitated by different nozzle diameters. The
specifications of the injectors used in this study are given in Table 4.2.

Prior to the study of spray images, injection rate was measured for injectors with two
different nozzle sizes as shown in Fig. 4.12. Averaged signals are shown as dotted lines and
numerically smoothed momentum flux is shown as solid lines in Fig. 4.12, similar to the Fig. 4.1.
Below the momentum flux plot, the injection rate is also shown.

As expected, a higher momentum flux and injection rate was measured for the high flow-
rate injector than the low flow-rate injector. By contrast, the same start of injection was observed
since this timing was determined primarily by the solenoid valve response, which was the same for
both injectors. Once the initial ramp-up completes, this trend continues for rest of the fuel injection
period. While the result simply confirmed the manufacturer’s flow-rate specification, an interesting

Table 4.2: PFI injectors used for investigation of the injector flow-rate study

Injector Low flow-rate injector High flow-rate injector

Injector model B 280 431 127-07 B 280 431 129-03
Injector flow-rate 261 g/min 364 g/min
Number of holes 6 6
Spray angle 70° 70°
Injection pressure 300 kPa 300 kPa

56
 Solenoid
delay

Delay due to
nozzle-sensor gap

Figure 4.12: Momentum flux and mass flow-rate of the injectors used in this study. Ethanol fuel
was used

trend was observed during the end-of-injection transient. In Fig. 4.12, it is clearly seen that the
actual injection duration for the high flow-rate injector is longer than the low flow-rate injector for
the same electronic pulse width. Considering both the ramp-up and ramp-down of the injection rate
were very similar between the injectors, it was a delay between the electronic signal and actual start
of ramp-down of the injection rate to cause the longer injection duration for the high flow-rate
injector. It was likely that the injector needle took longer time to respond in the higher flow-rate
environments. Upon the deactivation of solenoid, the spring attempts to push down the needle to
close the nozzle holes. Inside the nozzle, the flow momentum is higher due to higher flow-rate,
which created higher resistance against the needle closing. This extended delay in the needle
closing results in longer injection duration for the high flow-rate injector.

57
 2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Low Flow-rate Injector

2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

High Flow-rate Injector

Figure 4.13: Temporal evolution of Mie-scattered sprays for the low flow-rate injector (top) and
high flow-rate injector (bottom). Ethanol fuel was used

Figure 4.13 shows Mie-scattering images of the temporal development of the ethanol spray
for the low flow-rate and high flow-rate injectors at the image top and bottom, respectively. From
visual inspection, it is seen that the high flow-rate nozzle leads to higher number of droplets than
the low flow-rate nozzle at any time aSOI. This is well reflected in the number of droplets graph in
Fig. 4.14 where data for the high and low flow-rate injectors are compared for both ethanol and
gasoline. The ethanol and gasoline results from low-flow-rate injector experiments shown in Fig.
4.7 are re-plotted here for comparison purpose.

Figure. 4.14 shows that regardless of the fuel and injector flow-rate, the number of droplets
increases as the injection continues. This was due to a supply of new droplets as well as droplet
break-up, as explained in section 4.2. As a result, higher number of droplets was measured for the
high flow-rate injector at any fixed time aSOI and the difference in the number of droplets between
the low flow-rate and high flow-rate injector increases as the injection continues. The same trends
are observed for both ethanol and gasoline sprays with roughly the same difference between two
different flow-rate injectors. When ethanol and gasoline sprays are compared using the same
injector, the ethanol sprays show higher number of droplets than the gasoline sprays. Previously in
section 4.3, it was explained that the light components of gasoline droplets evaporate at a faster rate
than ethanol droplets, which would result in less number of droplets as observed in Fig. 4.14. Also,

58
 Figure 4.14: Number of droplets (top) and mean droplet diameter (bottom) against time aSOI
for low and high flow-rate injectors for ethanol and gasoline

the droplet break-up of heavy components of gasoline is lower than that of ethanol, which would
decrease the number of droplets of gasoline, as described earlier in section 4.3.

At the bottom of Fig. 4.14, the calculated mean droplet diameter is shown. Firstly for
ethanol sprays, a moderate decrease of the mean droplet diameter is observed for both flow-rate
cases, consistent with the expected process of droplet break-up. While this decreasing trend was the
same for both injectors, the high flow-rate injector shows higher droplet diameter at any time aSOI.
This was expected, as the larger nozzle would have bigger droplets initially which would persist for
longer time while they travel downstream. The result simply confirms the well-known fact that a

59
smaller nozzle diameter helps reduce fuel droplet sizes as a result of a smaller characteristic length-
scales dominating primary atomization.

The higher mean droplet diameter for the high flow-rate injector is also repeated for
gasoline fuel; however, the differences appear to be very small. This means that the droplet size of
gasoline is less sensitive to variations in the primary break-up. As discussed previously (Fig.4.10
and 4.11), much smaller droplets of ethanol are imaged near the nozzle (Fig.4.10) implying that the
break-up of ethanol sprays is much higher than gasoline sprays right after the fuel is introduced to
the ambient gas. This near-nozzle break-up supports that the primary break-up does not make a
significant impact on gasoline droplet size as it does on ethanol droplet size. It also explains a
marked difference in the mean droplet diameter between two fuels observed at earlier time aSOI in
Fig. 4.14 (bottom). The figure also shows that the decreasing rate of the mean droplet diameter with
increasing time aSOI is higher for the gasoline data than that for the ethanol data. It was surprising
as the number of droplets (Fig. 4.14 top) showed no such difference depending on fuel types. The
higher decreasing rate of the mean droplet diameter for gasoline sprays was the result of droplets
break-up. For ethanol sprays, the most of break-up processes ended near the nozzle and therefore
the impact of droplets break-up on the mean droplet diameter at later time aSOI were not
significant. However, the gasoline sprays underwent significant break-up of large droplets that
survived near-nozzle primary break-up as the injection continued and sprays penetrated further
downstream. This higher decreasing rate of droplet diameter did not change the fact that the ethanol
has lower mean droplet diameter, similar to the low flow-rate injector data in Fig. 4.14. However,
for the high flow-rate injector, the higher decreasing rate of gasoline results in nearly the same
mean droplet diameter between two fuels at later time aSOI (e.g. 4 and 5 ms aSOI). In other words,
the significance of increased mean droplet diameter due to the increased injector flow-rate was
found to be higher than the higher break-up of ethanol. This finding motivated further investigation
of the fuel injection parameters in addition to the fuel variation.

4.5 Effect of Injection Duration

In all the results discussed in earlier sections, the injection duration was fixed such that the
quasi-steady period of injection dominated the overall injection process. If the fuel injection
duration was short (i.e. the injection ends early), a transient period of fuel injection (after the end of
injection) would dominate the overall spray development. During this end-of-injection transient, the
injection momentum transfer declines and eventually stops which might result in significantly
different droplet characteristics compared with those of quasi-steady sprays. For instance, Hung et

60
al. [2004] performed Mie-scattering imaging and laser diffraction droplet measurements to study
the end-of-injection behaviour of n-heptane-fuelled PFI sprays and reported smaller droplet size
than that of the quasi-steady sprays.

In this section, an injection case with a small injection pulse width is compared to a long
steady-injection case to examine details of the transient behaviour of ethanol sprays. Specifically,
this comparison allows differentiation of the spray behaviours with presence of the injection
momentum (long injection case) and in absence of injection momentum (short injection case). A 3-
ms pulse width was selected for the short injection case, which was short enough to image the entire
end-of-injection transient for a given camera field of view. On the other hand, an 8 ms pulse width
was used for the long injection case (similar to the previous studies) representing the steady-state
spray.

Typical temporal evolutions of Mie-scattered ethanol sprays for the long and short
injections are shown on the top and bottom of Fig. 4.15, respectively. On the top of each image, a
time stamp after the start of injection is given. For the long injection, the injection momentum

2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Long Injection
2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Short Injection

Figure 4.15: Temporal evolution of Mie-scattered ethanol sprays for the long injection (top) and
short injection (bottom) case. The low flow-rate injector was used. The actual end-of-injection
occurs at 2.4 ms aSOI for the short injection case while the injection continues up until 5 ms
aSOI for the long injection condition

61
continues to drive the spray penetration for all imaging timings presented in Fig. 4.15. However, for
the short injection, the injection momentum was cut off at 2.4 ms aSOI, and few droplets were
found near the nozzle.

In addition to the single-shot Mie-scattering images, a high-speed shadowgraph movie of

ethanol sprays was obtained to capture the transient spray behaviours as well as vapour-phase
boundaries. The vapour boundaries of the ethanol sprays are overlaid onto the Mie-scattered images
at different time aSOIs as shown in Fig. 4.16. The green lines indicate the vapour boundaries from
the shadowgraph movies, while the white regions show the Mie-scattered liquid droplets. There are
some mismatches between the Mie-scattering and shadowgraph imaging that must be considered.
The injectors used were dual-spray type that deliver the fuel in two separate plumes with three
sprays in each plume (this nozzle hole configuration is shown at the bottom-left of Fig. 3.11). For
the shadowgraph imaging, the nozzle was oriented such that two separate spray plumes were
visualized. By contrast, three separate spray plumes were visualized in the Mie-scattering imaging
(Fig. 4.15). This difference was caused by a fixed position of camera and strobe light for the Mie-
scattering imaging. If the same nozzle orientation were attained for two different imaging

2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Long Injection

2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Short Injection

Figure 4.16: Shadowgraph (liquid/vapour) boundaries overlaid on binary images of Mie-

scattered (liquid) ethanol droplets for the long injection (top) and short injection case (bottom).
The low flow-rate injector was used. The end-of-injection occurs at 2.4 ms aSOI for the short
injection case while the injection continues up until 5 ms aSOI for the long injection condition

62
diagnostics, the Mie-scattering camera and strobe light would interfere with the shadowgraph setup.

Despite the difference in nozzle orientation, Fig. 4.16 provides valuable information for the
following discussions, which cannot be obtained from Mie-scattering or shadowgraph images alone.
Two different injection durations appear to have similar liquid and vapour regions as shown in
images taken at 2 ms aSOI. However, from 3 ms aSOI, long and short injection cases begin to
exhibit a difference. For the long-injection case, the vapour-phase spray boundaries overlap with
liquid regions except the near nozzle region where dense liquid droplets are present. While droplets
evaporation likely occurs within the spray regions, outer boundaries do not show clear separation
between the liquid and vapour. For the short injection, however, a marked difference is seen with no
or very few of liquid droplets visualized near the nozzle. For example, at 4 ms aSOI or after, no
liquid regions are found but vapour up to 30 mm from the nozzle. The same imaging and image
post-processing of Figs. 4.15 and 4.16 are repeated using gasoline for comparison purposes.

How this evaporation rate affects the liquid droplet characteristics is demonstrated in Fig.
4.17 for both ethanol and gasoline sprays. The figure shows the number of droplets and the mean
droplet diameter calculated from the Mie-scattering images for both long and short injection cases.
It is found that up to 3 ms aSOI, the number of droplets increases for both injection durations
simply because of continued fuel injection and droplets break-up. However, a noticeable separation
is seen after 3 ms aSOI that the number of droplets decreases for the short-injection while the long-
injection case shows increasing number of droplets. This was because a supply of new droplets
from fuel injection was ceased in short injection. The mean droplet diameter also shows a
separation at 3 ms aSOI. For instance, the decreasing trend of the mean droplet diameter accelerates
for the short injection case. In addition to the droplet break-up, the result for the short injection
suggests an important role of droplets evaporation on the number of droplets and mean droplet
diameter. One may conclude that the break-up is the only cause for the decreasing mean diameter
trend; however, the decreasing number of droplets after the end of injection conflicts with this
explanation. Therefore, it is suggested that droplet evaporation is another important effect on PFI
spray characteristics. That is, the decreasing number of droplets after the end of injection is a result
of droplet evaporation, consistent with visual observations from Fig. 4.16. The same trends are
observed for both ethanol and gasoline sprays with consistently higher number of droplets and
lower mean droplet diameter for ethanol sprays due to the same reason explained for Fig. 4.14. The
significance of injection duration variation on sprays appears to be less than that of fuel variations.

63
 Figure 4.17: Number of droplets (top) and mean droplet diameter (bottom) for the long and
short injection cases for ethanol and gasoline. The low flow-rate injector was used

To further discuss the transient PFI spray behaviours, axial profiles of the number of
droplets and mean droplet diameter were drawn similar to Figs. 4.4 and 4.5 respectively. Since the
impact of the injection duration on fuel sprays did not show a difference depending on fuel types, an
example is selected for ethanol. Figure 4.18 shows the results of this analysis. In the top row, an
axial profile of the number of droplets is plotted, while the bottom row gives the mean droplet
diameter profile. The imaging timing for each profile is given at the bottom-right of each plot. The
figure shows that, for the long injection case, the number of droplets decreases with increasing
distance from the nozzle at any time aSOI. This was clear evidence that the evaporation occurred as
droplets penetrate downstream, i.e. disappearance of the Mie-scattered liquid droplets in the image

64
 2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Figure 4.18: Axial profiles of droplets number (top) and mean droplet diameter (bottom) at
different times aSOI for the long and short injection cases. The data were obtained from Mie-
scattered spray images in Fig. 4.15 and 5 mm spray slices in the axial direction were used as
shown in Fig. 3.9

due to the evaporation. The same trend is shown for the short injection case at 2 and 3 ms aSOI
during which ethanol fuel injection continues. After 3 ms aSOI, however, a very low number of
droplets were measured near the nozzle as the ethanol fuel injection stopped. For example, at 4 ms
aSOI, the droplet number increases up until 35 mm from the nozzle, hits the maxima, and beyond
the maxima follows the same profile of the long injection case. The same trend was also observed at
5 ms aSOI with the maxima further downstream at 50 mm from the nozzle. Once again, this
analysis suggests that with no ethanol supply after the end of injection, droplets evaporation occurs
from the near-nozzle region, which then propagates downstream. In other words, the near-nozzle
droplets are more dilute and can easily mix with ambient air while less droplet interaction occurs,
similar to Ref. [Wong et al. 1992]. This near-nozzle evaporation clearly differentiates the short-
injection spray from the long, quasi-steady spray.

The same trend is also observed in the mean droplet diameter profiles given in the bottom
plots of Fig. 4.18. The mean droplet diameter increases with increasing distance from the nozzle as
small droplets evaporate (and disappear in the Mie-scattering images) leaving only large droplets.
An interesting observation from Fig. 4.18 is that at up to 3 ms aSOI, the mean droplet diameters are
the same for both short and long injections. However, upstream sprays show a lower droplet
diameter at 4 and 5 ms aSOI. Similar to the number of droplets discussion for the short injection,

65
the fuel injection stopped at 3 ms aSOI that resulted in no addition of large droplets from the nozzle
hole. This caused lower mean droplet diameter near the nozzle. Again, this is consistent with
shadowgraph image boundaries in Fig. 4.16 where the near-nozzle region is dominated by vapour-
phase fuel with no liquid droplets evident in Mie-scattering images.

In summary, the number of droplets and mean droplet diameter in Figs. 4.17 and 4.18
together with Mie-scattering and shadowgraph images in Fig. 4.16 indicate that the sprays behave
the same for both short and long injection durations for the steady period of fuel injection.
However, once the injection ends for the short injection case, no supply of large droplets and faster
evaporation of near-nozzle droplets result in lower number of droplets and mean diameter in the
near-nozzle region, which then propagate downstream and eventually affect entire sprays. These
effects on fuel sprays are found to be the same for both ethanol and gasoline with no measurable
variations in the significance.

4.6 Effect of Cross-flow

While findings from previous sections are relevant to automobile engines, it should be
noted that the measurements were performed at quiescent ambient conditions. In an actual engine,
sprays injected into the intake manifold or port are under the strong influence of air cross-flow
regardless of its position and angle of attack. To understand how this cross-flow affects PFI sprays,
the spray characteristics were further investigated for the selected low flow-rate injector and short
injection conditions.

Figure 4.19 shows the temporal evolution of ethanol sprays under the influence of cross-
flow. In the top row, boundaries from Mie-scattered spray images are shown, which can be
compared to the short-injection case in Fig. 4.15 for the quiescent ambient condition. On the
leftmost image taken at 2 ms aSOI, the velocity profile of the cross-flow is superimposed.

The mean velocity of the ambient air cross-flow was measured at 12 m/s with only 0.5 m/s
variations across the chamber. In the middle row, shadowgraph image boundaries under the
influence of cross-flow are overlaid on raw shadowgraph images. These shadowgraph (i.e.
liquid/vapour) boundaries are overlaid on the Mie-scattered spray images and presented in the
bottom row, similar to the short-injection case at the quiescent ambient condition as in Fig. 4.16.
The sprays under the cross-flow influence would deflect toward the airflow direction compared to
the sprays at a quiescent ambient condition simply due to the radial momentum transfer. However,
this is not the case for the images taken at 2 ms aSOI. The Mie-scattered liquid droplets appear not

66
 2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Figure 4.19: Mie-scattering image boundaries of ethanol sprays (top row), shadowgraph
boundaries overlaid on raw images (middle row) and shadowgraph boundaries overlaid on Mie-
scattering images (bottom row) under the influence of cross-flow. The low flow-rate injector
and short injection duration were used. The end-of-injection occurs at 2.4 ms aSOI

to be affected by the cross-flow but show similar structure to the sprays at quiescent conditions in
Fig. 4.16. This is likely due to an order of magnitude higher injection momentum. The fuel injection
velocity of 20 m/s is only 8 m/s higher than the cross-flow air velocity; however, the momentum of
fuel droplets is much higher than that of the cross-flow air because the droplet density is much
higher than the air density. In other words, the axial momentum during the injection period is more
prominent than the radial momentum transfer from the cross-flow air. By contrast, the Mie-
scattered liquid droplets after the end of injection (3 ~ 5 ms aSOI) show the influence of cross-flow.

67
 End of
Injection

Figure 4.20: Distance from the nozzle to rightmost shadowgraph boundaries (i.e. radial vapour
tip penetration) for quiescent and cross-flow conditions. The low flow-rate injector, short
injection duration and ethanol fuel were used. The end-of-injection occurs at 2.4 ms aSOI

The liquid droplets at the left of the nozzle are less dense than those at the right. Moreover, the
right-side sprays are shifted toward the cross-flow direction. It is clear that the radial momentum
transfer affects the liquid droplets once the injection ends as there is no high injection momentum
present. The shadowgraph images are consistent with the Mie-scattering images. The shadowgraph
image boundaries tend to shift to the right when affected by the cross-flow only after the end of
injection. This trend is seen clearly in the radial tip penetrations defined by the distance from the
nozzle to the rightmost shadowgraph image boundaries as shown in Fig. 4.20. The radial tip
penetrations are same for the quiescent and cross-flow sprays up to the end of injection at 2.4 ms
aSOI. At the end of injection, the radial tip penetration for the cross-flow sprays departs from the
quiescent sprays and then the difference continues to increase as the sprays penetrate further
downstream.

Another interesting trend observed from Fig. 4.19 (the bottom row) is that the left-side
shadowgraph image boundaries are still highly populated even under the influence of the cross-
flow, unlike the liquid droplets. Some shadowgraph boundaries overlap with Mie-scattering
regions; however, the shadowgraph (liquid/vapour) boundaries occupy larger areas than the Mie-
scattered liquid droplets. This suggests that the liquid droplets at the left interact with the coming

68
airflow and evaporated at a higher rate than the right-side liquid droplets. This is a good example
that the air cross-flow can enhance evaporation of ethanol droplets after the end of injection.

Characteristics of ethanol liquid droplets at the cross-flow condition are discussed in greater
detail by calculating the number of droplets and mean droplet diameter as shown in Fig. 4.21. The
data for cross-flow-influenced sprays is compared to quiescent sprays. For comparison purposes,
the data for gasoline sprays is also shown, similar to Figs. 4.14 and 4.17. The figure shows that for
ethanol at 2 ms aSOI, the number of droplets for the cross-flow condition is higher than that for the
quiescent condition. However, this is reversed after 3 ms aSOI (i.e. after the end of injection): the

Figure 4.21: Number of droplets (top) and mean droplet diameter (bottom) against time aSOI at
quiescent and cross-flow conditions for ethanol and gasoline. The low flow-rate injector and
short injection duration were used. The end-of-injection occurs at 2.4 ms aSOI

69
number of droplets is lower for the cross-flow condition. The higher number of droplets for the
cross-flow condition that occurs during the injection can be explained by higher droplets break-up.
This is because Weber number that estimates likelihood of the droplet break-up increases when the
cross-flow is present. However, the higher Weber number does not appear to increase the number of
droplets for the cross-flow condition after the end of injection: the number of droplets is lower for
the cross-flow condition for the data measured at 3~5 ms aSOI. This trend is explained by the
evaporation of ethanol droplets. Once the injection stops, it is likely that the evaporation rate
increases and hence liquid droplets quickly disappear in the Mie-scattering images. This is
consistent with the high population of shadowgraph boundaries but low Mie-scattering signals in
the left side of sprays as mentioned previously. If evaporation of fuel droplets plays a role in the
number of droplets, one might expect that variations in evaporation characteristics of ethanol and
gasoline would cause a difference under the influence of ambient cross flow. However, figure 4.21
shows that the impact of cross flow on the number of droplets appears to be similar regardless of the
fuel type.

In contrast to the trend observed for the number of droplets, the mean diameter of the liquid
droplets is lower for the cross-flow condition regardless of whether it is measured before or after the
end of injection. Since higher droplet break-up is expected associated with higher Weber number,
the observed trend is expected. Higher evaporation rate for the cross-flow sprays would also

Figure 4.22: Droplet size distribution for the quiescent and cross-flow sprays at 4 ms aSOI for
ethanol. The low flow-rate injector and short injection duration were used

70
decrease the mean droplet diameter as the liquid droplets shrink in size. Indeed, droplet size
distributions calculated from the images at 4ms aSOI for ethanol shown in Fig. 4.22 confirms that
the cross-flow sprays have a lower number of droplets over the entire size range. In addition, no
fuels effect is found in Fig. 4.21 such that the decrease of the droplet diameter is nearly the same for
both ethanol and gasoline with about 15 ~ 25 μm reduction depending on time aSOI. Simply, the
expected trends of higher number of droplets and lower droplet diameter for ethanol sprays are
observed similar to Fig. 4.21.

The previous discussion raises another interesting question: where does the higher
evaporation occur after the end of injection for the cross-flow sprays? It might be the head region of
sprays where fuel droplets interact with fresh air at the spray boundary. The near-nozzle area might
also experience increased evaporation since the dense spray region does not exist after the end of
injection.

To answer this question, an axial profile of the number of droplets and mean droplet
diameter is plotted (similar to Fig. 4.18) and is shown in Fig. 4.23. Since the same impact of cross
flow on sprays is found regardless of fuel types (see Fig. 4.21), the ethanol sprays are selected for

2ms aSOI 3ms aSOI 4ms aSOI 5ms aSOI

Figure 4.23: Axial profiles of number of droplets (top) and mean droplet diameter (bottom) at
different times aSOI for quiescent and cross-flow conditions for ethanol sprays. The quiescent
spray data for the short-injection case from Fig. 4.18 is plotted again for comparison purpose.
The Mie-scattered spray images in Fig. 4.19 are used to obtain 5 mm spray slices in the axial
direction (definition given in Fig. 3.9). The end-of-injection occurs at 2.4 ms aSOI

71
this analysis. The same Mie-scattering images in Fig. 4.19 are used to obtain the number of droplets
and mean droplet diameter for each 5 mm spray slice from the nozzle. The short-injection spray
data from Fig. 4.18 is plotted again to compare the sprays at quiescent ambient air conditions to the
sprays under the influence of the cross-flow air. One can draw an immediate conclusion from Fig.
4.23: there is no specific spray region with higher or lower number of droplets and mean droplet
diameter but the same trend is found for the entire spray region. The number of droplets for the
cross-flow sprays before the end of injection at 2 ms aSOI is higher regardless of the axial distance
from the nozzle. Right after the end of injection at 3 ms aSOI, the number of droplets is nearly the
same for both cases and thereafter the cross-flow sprays show lower number of droplets at any fixed
axial distance from the nozzle. The trend of lower mean droplet diameter for the cross-flow sprays
does not change with the axial distance from the nozzle at any time aSOI.

4.7 Effect of Ambient Pressure

The previous PFI spray data shown for ethanol and gasoline fuels were taken at
atmospheric ambient pressure. In this section, the results from experiments at higher ambient
pressures are presented. With increasing popularity of the intake air boosting (e.g. turbocharger,
supercharger, or twin turbo) in gasoline-fuelled engines, the ethanol sprays at higher ambient
conditions were tested. Ethanol fuel and the low-flow-rate injector were selected to investigate the
effect of ambient pressure.

Typical spray images of ethanol at 4 ms aSOI for various ambient pressures (0, 100 and 200

0 kPa 100 kPa 200 kPa

Figure 4.24: Ethanol spray images at 4ms aSOI for various ambient pressures

72
kPa gauge) are shown in Fig. 4.24. A well-expected trend is observed that at higher ambient
pressure, droplets travel at a slower velocity and hence the spray tip penetration is lower at a fixed
time aSOI. Details are shown in Fig. 4.24 such that more fuel droplets are packed in a smaller
region at higher ambient pressures and denser sprays are essentially created. While the droplet
distribution was still dilute enough for image processing, the sprays became denser as the ambient
pressure increased.

The abovementioned trends in spray images are well reflected in the spray tip penetration as
shown in Fig. 4.25. The spray tip penetration is lower for higher ambient pressure. While varying
the ambient pressure, same pressure differential of 300 kPa was maintained across the injector.
Therefore, the injection velocity and momentum were fixed. This means that increased drag in the
ambient air for high ambient pressure lowered the droplet’s velocity and the spray penetration rate.
Similar results can be found in previous studies conducted for DI injectors [Gao et al. 2005, Wang
et al. 2005].

Fig. 4.26 shows an axial profile of the droplet number for all ambient pressures investigated
at a fixed 4 ms aSOI. Similar to Fig. 4.4, a 5 mm slice of sprays along the injector axis was
processed to obtain a plot of the droplet number versus distance from the nozzle. It is well known
that higher ambient pressure (and so higher density) leads to better break-up because of increased
Weber number. This causes a higher number of droplets at increased ambient pressures. Fig. 4.26

Figure 4.25: Ethanol spray tip penetration against time aSOI for various ambient pressures

73
 Figure 4.26: Axial profile of ethanol spray droplets number along the injector axis at 4ms aSOI
for various ambient pressures

shows the expected result that more droplets are measured for higher ambient pressures. However,
this does not necessarily mean that the break-up is enhanced because the tip penetration decreases
due to increased drag. For instance, the increased number of droplets at a fixed distance from the
nozzle could be a result of dense sprays rather than the increased break-up. Therefore, how the
integration of the droplet numbers in Fig. 4.27 changes for different ambient pressures must be
investigated.

In Fig. 4.27, the total number of droplets for various times is shown to answer this question.
The figure shows that the total droplet number does not vary significantly and there is no monotonic
behaviour found. Therefore, the higher number of droplets observed in the upstream region (Fig.
4.26) was not due to break-up of droplets but denser distribution of droplets within the spray region
at higher ambient pressures. That is, the same number of droplets exists within a smaller spray
region at higher ambient pressures. Another important consideration for ethanol sprays when
increasing the ambient pressure is the evaporation rate.

At higher ambient pressures, the fuel saturation temperature increases, this suppresses the
evaporation of liquid droplets. Indeed in Fig. 4.28, the mean droplet diameter for various times
aSOI shows an increasing trend with increasing ambient pressure. Since no significant effects of
break-up on ethanol spray droplets were observed with increasing ambient pressure, this increase

74
should be due to a different evaporation rate. Therefore, it was likely that lower evaporation rate for
higher ambient pressure (i.e. higher saturation temperature) increased the mean droplet diameter,
similar to the droplet evaporation results reported in references [Morin et al. 2004, Nomuraa et al.
1996]

Figure 4.27: Total number of ethanol sprays droplets against time aSOI for various ambient
pressures

Figure 4.28: Mean droplet diameter of ethanol spray against time aSOI for various ambient
pressures

75
4.8 Summary of PFI spray results

The following table summarizes the results of spray studies in Chapter 4. These results will
be referenced in the following sections.

Table 4.3: Summary of the results from PFI spray studies

Parameter Spray Tip Droplet Mean Droplet

Penetration (mm) Number Diameter (μm)
at 5ms aSOI
Fuel Type Ethanol 81.3 4992 119
(low flow rate, long E50 82.3 4620 125
injection, quiescent, 0
kPa ambient pressure) Gasoline 82.4 3706 126
Injector Flow Rate High Flow Rate 81.4 7141 129
(ethanol, long injection,
quiescent, 0 kPa Low Flow Rate 81.3 4992 119
ambient pressure)
Injector Duration Long Injection 81.4 4092 119
(ethanol, low flow rate,
quiescent, 0 kPa Short Injection 80.97 2236 110
ambient pressure)
Ambient Flow Cross-flow 85.3 1655 92
(Ethanol, low flow rate,
short injection, 0 kPa Quiescent 80.97 2236 110
ambient pressure)
Ambient Pressure 0 kPa 81.3 4992 119
(ethanol, low flow rate, 100 kPa 71.5 5490 122
long injection,
quiescent) 200 kPa 59.8 5356 124

76
Chapter 5

Dual-fuel Combustion

Dual-fuel engine tests were conducted at fixed operating conditions as listed in Table 5.1.
The engine was run at a constant speed of 2000 rpm at which the maximum torque is measured for
the base engine. The coolant temperature was fixed at 90°C and the intake air temperature was held
constant around 35°C. The engine was started in a diesel-only mode during the engine warm-up
period. The total fuel energy per engine cycle was fixed at 1098 J for ethanol and diesel combined.

Table 5.1: Engine operating conditions

Engine speed 2000 rpm

Coolant temperature 90°C
Intake air temperature 35°C
Total energy input per cycle 1098 J
Brake mean effective pressure ~450 kPa
Indicated mean effective pressure ~900 kPa
Ethanol injection pressure 300 kPa
Ethanol injection timing Intake TDC
Diesel injection pressure 130 MPa
Diesel injection timing -13~1°CA aTDC

77
This fuel energy is equivalent to an overall equivalence ratio of ~0.64 at 2000 rpm. For the baseline
operating condition of the diesel-only combustion, this fuel energy corresponds to brake mean
effective pressure (BMEP) of around 450 kPa and represents a medium load condition. The
indicated mean effective pressure (IMEP) was about 900 kPa, suggesting the frictional mean
effective pressure (FMEP) of about 450 kPa. These effective pressures varied depending on the
fraction of ethanol and diesel injection timing. The PFI injector used was low flow-rate injector,
ethanol injection pressure was held constant at 300 kPa, same as that of injection system used in
spray visualization studies (chapter 4). The injection timing was fixed at the intake TDC to allow a
long time for mixture preparation in the intake manifold and to avoid the HC emission increase
associated with later start of injection [Bianchi et al. 2006, Alkidas 1994]. The diesel injection
pressure was set at 130 MPa and the injection timing was varied between 1°CA and -13°CA after
TDC (°CA aTDC).

For dual-fuelling the ratio of ethanol energy to the total fuel energy was varied, while the
total fuel energy was fixed. The ethanol fraction Ex was defined as:

mሶ ethanol *CVethanol
Ethanol fraction Ex= ሺ1ሻ
mሶ ethanol *CVethanol +mሶ diesel *CVdiesel

Figure 5.1: Diesel and ethanol injection masses for various ethanol energy fractions. The
total fuel energy per cycle was held constant at 1098 J

78
where mሶ is the fuel mass flow-rate and CV stands for the calorific value (heat of combustion).
Figure 5.1 shows the diesel and ethanol masses for various ethanol energy fractions used in the
present study. To increase Ex, the diesel mass per engine cycle was decreased by reducing the
injection duration while ethanol mass was increased by extending the injection duration such that
total energy was fixed at 1098J. Since the total energy is fixed, increasing the ethanol energy
fraction means increased fuel mass considering lower calorific value of ethanol than diesel.
Therefore, actual air-fuel ratio decreases with increasing ethanol fraction. However, a
stoichiometric air-fuel ratio of ethanol is 9, significantly lower than diesel’s 14.5. The decrease in
both the stoichiometric air-fuel ratio and actual air-fuel ratio with increasing ethanol energy fraction
resulted in negligible variations in the overall equivalence ratio. For example, by varying the
ethanol fraction from 0% to 80%, the equivalence ratio changed from 0.65 to 0.63. Hence, it can be
considered that the tests were performed at a fixed total energy as well as fixed overall equivalence
ratio. Details of this calculation are given in Appendix A6.

5.1 Effect of PFI Injector Position on Dual-fuel Combustion

The influence of port fuel injector (PFI) location on dual-fuel combustion was investigated
in an automotive-size, single-cylinder diesel engine. Two different ethanol PFI locations were
selected: one closer to the hot intake valves with sprays hitting the hot intake ports and therefore
enhanced droplet vaporization (i.e. less wall wetting) is expected despite shorter time for airflow-
droplets interaction meaning less break-up, and the other further upstream of the intake valves with

PFI Position A PFI Position B

Intake Ports

Intake Manifold

Figure 5.2: Illustration of the intake ports, manifold and port fuel injectors at two different port
fuel injector (PFI) positions

79
 Table 5.2 Port fuel injector (PFI) positions relative to the intake valves

PFI position A PFI position B

Distance between the injector 115 mm 325 mm
and intake valves

Estimated residence time of 3.8 ms 7.5 ms

ethanol droplets in the intake

increased time for airflow-droplets interaction potentially resulting in enhanced break-up but
limited chance for hot valve surface vaporization of ethanol droplets (i.e. more wall wetting).
Therefore, questions on the significance of ethanol break-up and hot surface vaporization on dual-
fuel engine combustion can be addressed.

Figure 5.2 shows the top view and side view of the intake manifold with two different PFI
positions, namely PFI position A and B. The injector at position A is located 115 mm from the
intake valves and ethanol sprays aim at the intake ports. The ethanol droplets would stay in the
intake ports under the influence of airflow before they hit on the port surface and flow into the
cylinder. Using sprays penetration data (Fig. 4.3) and the distance between the injector and valves,
the residence time of ethanol droplets in the intake was estimated at 3.8 ms as listed in Table 5.2.

In contrast to position A, the injector at position B was placed further upstream of the
intake valves (325 mm) and the ethanol sprays were delivered into the intake manifold connecting
the intake ports and intake tank. This position would allow 7.5 ms of droplets residence time in the
intake, allowing more interaction between ethanol droplets and intake airflow. In section 4.6, it was
proved that this additional interaction of ethanol spray with cross-flow of air results in lower droplet
diameter (Figs. 4.21 and 4.22). It is worth to note that some researchers [Nassiri et al. 2007, McGee
et al. 2000] suggested vaporization of liquid droplets on the hot surface of the intake valves play a
significant role in the intake charge preparation. This is because the intake valves are directly
exposed to combustion gases and therefore the surface temperature is high enough to cause surface
boiling of liquid fuel droplets. Due to its shorter distance to the intake valves, PFI position A would
have higher chance for ethanol droplets hitting the hot intake port than that of position B. In
addition to the residence time for the droplets-airflow interaction, the surface boiling would be
considered as an important parameter in the following sections. The diesel injection timing was set

80
at -3°CA aTDC for these tests, which was selected as a baseline condition from preliminary tests of
diesel injection timing (will be described in section 5.2).

Figure 5.3 shows measured in-cylinder pressure and apparent heat release rate for various
crank angle degrees for different ethanol energy fractions (E%) tested at two PFI positions A and B
at left and right sides respectively. The first noticeable trend from Fig. 5.3 is that the impact of
ethanol fraction on these global phenomena is more significant than the influence of PFI position. In
fact, in-cylinder pressure and aHRR traces for two different PFI positions at any fixed ethanol
fraction are nearly identical. By contrast, variations in ethanol energy fraction cause measurable
differences in in-cylinder pressure and aHRR.

Figure 5.3: Effect of ethanol energy variation on in-cylinder pressure traces (top) and apparent
heat release rate traces (bottom). The left column shows results for PFI position A and the right
column shows the results for PFI position B

81
 For a fixed PFI position, figure 5.3 shows that an increase in the ethanol fraction leads to a
decrease in the in-cylinder pressures near TDC. This happens because of the vaporization cooling of
the ethanol [Ogawa et al. 2010, Tsang et al. 2010, Ekholm et al. 2008]. Also, higher ethanol fraction
leads to a decreased specific heat ratio (γ) than that of lower or zero ethanol fractions. Both cause
decreased TDC pressure and temperature. Using these measured in-cylinder pressure traces, the
apparent heat release rates (aHRR) are calculated and shown at the bottom of Fig. 5.3. A clear trend
is observed from the figure that with increasing ethanol fraction, the start of combustion is delayed,
the combustion phasing is retarded, and the peak aHRR increases. This was because the lower
ambient pressure and temperature extended the diesel ignition delay period. Moreover, the ambient
gas entraining into the diesel jet is not air-only but an ethanol-air mixture and therefore oxygen
concentration would be lower compared with the pure compressed air. This ambient gas dilution
(lower oxygen concentration) further extends the ignition delay [Kook et al. 2005, Pickett et al.
2005]. As the ignition delay increases, the timing of the pressure rise shifts to the right and the peak
aHRR appears at later timings. The increased aHRR is proposed to be due to two reasons. First, as
ethanol fraction is increased, the proportion of burning occurring in a rapid premixed combustion
mode is increased. Second, the increased ignition delay also allows extended time for pre-
combustion mixing of the diesel fuel, which also results in an increased peak aHRR [Sahoo & Das
2009]. This increasing trend in peak aHRR continues until the misfiring occurs due to over-retarded
combustion phasing. For example, markedly low peak aHRR is observed for E50 for PFI position A
and E35 for PFI position B. This misfiring limits the maximum ethanol fraction for a given engine
operating condition.

How this combustion translates to the power output is quantified by the indicated mean
effective pressure (IMEP) and indicated thermal efficiency as shown in the top-left corner of Fig.
5.4. As mentioned previously, the total energy input was fixed for all test points, and hence the
indicated thermal efficiency follows the same trend of IMEP. Also shown at the bottom is the
coefficient of variation of IMEP (CoV) for various ethanol energy fractions at two different PFI
injector positions tested. The CoV of IMEP is a good indicator of engine stability. In the present
study, if the CoV of IMEP was higher than 5%, it was considered as misfiring [Weall et al. 2012].

The similarity observed for two PFI positions but measurable differences for ethanol
fraction variations seen in Fig. 5.3 are repeated again in Fig. 5.4. Between PFI positions A and B,
the only difference is found in the misfiring limit. As mentioned previously, the engine misfires
when ethanol fraction is 50% for PFI position A due to over-retarded combustion phasing. Indeed,
the CoV of IMEP at this condition exceeds 16%, high enough to be considered as misfiring. By

82
 5% Misfire Limit

Figure 5.4: Indicated mean effective pressure (IMEP, at top-left) and coefficient of variation
(CoV) of IMEP (bottom-left), ignition delay (top-right), CA50 (bottom-right) for various
ethanol energy fractions

contrast, the measured CoV of IMEP is lower than 4% at lower ethanol energy fractions, which is
lower than the acceptable fluctuation of 5%. An interesting finding is that for PFI position B, the
misfiring (i.e. high CoV of IMEP) occurs at lower ethanol fraction of 35% than that for PFI position
A. This means that PFI position A performed better in making ethanol premixed charge than PFI
position B so that higher ethanol energy fraction was achievable.

This result was against the expectation because PFI position B would have longer time
available for the droplets-airflow interaction. One possible explanation is the wall wetting in the
intake system. As mentioned previously, PFI position A is closer to the intake valves and therefore
has a higher chance that ethanol droplets hit on the hot valve seat and vaporize, resulting in reduced
wall wetting [Nassiri et al. 2007, McGee et al. 2000]. Longer time available for the droplets-airflow
interaction for PFI position B might decrease sizes of ethanol droplets as seen in Fig. 4.21;

83
however, increased chance of those liquid droplets wetting the wall and decreased possibility of hot
surface boiling might result in higher wall wetting than PFI position A. The significance of this wall
wetting might not be high at low ethanol fraction conditions and therefore IMEP, CoV of IMEP
appear to be very similar for different PFI positions; however, as the ethanol fraction increases
further, the impact became significant, resulting in a different misfiring limit. The same trend is
observed in the ignition delay and combustion phasing. As shown in the top-right corner of Fig. 5.4,
the ignition delay is similar for both PFI positions up to 30% ethanol fraction; however, at higher
ethanol fraction, the ignition delay is higher for PFI position B due to increased wall wetting, which
in turn causes over retarded combustion phasings as shown in the bottom-right corner of Fig. 5.4.
As a result, the misfiring limit was found at lower ethanol fraction for PFI position B.

The importance of wall wetting will be further supported by the measured engine-out
emissions in the following discussions. The effect of PFI position and ethanol energy fraction on
CO, NOx, HC and smoke emissions are shown in Fig. 5.5. The first noticeable trend from these
emissions is that between PFI positions A and B, the overall trend with increasing ethanol fraction
is very similar. The figure shows that at a fixed injector position, CO emission decreases up to
certain ethanol fraction. This was a result of the increased combustion temperature with increasing
ethanol fraction as was expected from the increasing pressure trend in Fig. 5.3. The NOx emission
shown next to the CO emission also increases with increasing ethanol fraction suggesting increasing
combustion temperature. However, figure 5.5 shows that the CO emission eventually starts to
increase as the ethanol fraction increases further. This was not expected because pressure as well as
apparent heat release rate continues to increase implying continued increase of the combustion
temperature.

There are two causes that can result in increased CO emissions. Firstly, the CO to CO2
conversion would freeze at higher ethanol fractions. From figure 5.3, it is seen that except for the
misfiring cases, the combustion ends approximately at the same crank angle position i.e. about
17°CA aTDC for both PFI locations A and B. Also, as the ethanol fraction increases, the
combustion phasing is further retarded, which means bulk of the combustion occurs during the
expansion stroke. At these retarded combustion phasings, the cylinder expansion rate could happen
faster than chemical reaction rates which cause a drop in temperature effectively freezing the CO
kinetics due to absence of hydroxyl radicals [Sjoberg & Dec 2005] at this low temperature
conditions. Secondly, even before the cylinder expansion, the quenching of the premixed ethanol
combustion can occur at the wall. It is well known that in HCCI engines [Kong & Reitz 2003,
Bhave et al. 2006], local temperatures at the wall can be insufficient to lead to proper conversion of

84
 Figure 5.5: Effect of ethanol energy fraction and PFI position on carbon monoxide (CO), oxides
of nitrogen (NOx), unburnt hydrocarbon (HC) and exhaust gas opacity (i.e. smoke level)
emissions

CO to CO2, leading to high CO emissions. For very high ethanol fractions (E50 for PFI position A
and E35 for PFI position B), the CO emission shows the highest level for a given diesel injection
timing. This was due to misfiring where the combustion cannot self-sustain as it occurs too late at
the expansion stroke.

The HC emissions show a different trend to the CO emissions. Figure 5.5 shows that HC
emission is proportional to the ethanol energy fraction. The increased HC emission is expected for
those over-retarded and hence misfiring conditions; however, the increasing trend of the HC
emission at a low ethanol fraction range (e.g. 0~40% for PFI position A) is a surprise considering
the decreasing CO emission and increasing combustion temperature. There should be additional

85
cause for the HC increase other than the quenching of the premixed ethanol combustion at the wall.
Previous studies [Tsang et al. 2010, Himabindu et al. 2008] reported the same issue of increased HC
emission for increased ethanol fraction although the causes were not clarified. Recently, Colban et
al. [2007] suggested that the bulk gas quenching due to low combustion temperature is not the only
source of HC emissions of diesel engines but a significant portion is originated from the mixture
trapped inside the crevice. Therefore, it was likely that some ethanol mixture or liquid droplets were
trapped inside the crevice during the intake and compression strokes prior to the diesel injection
[Himabindu et al. 2008], which increased the HC emission even at the low ethanol fraction range.
From the measured HC and CO emissions, the combustion efficiency was estimated following the
same approach as in Ref. [Kook et al. 2005]. Extent of unburnt HC and CO emissions in exhaust
gases indicates the combustion inefficiency and it measures the completeness of the combustion
reactions. The details of this calculation are given in Appendix A6. As shown in Fig. 5.6, the
combustion efficiency of both PFI positions follows a reversed trend of CO emissions with a drastic
increase of HC emissions at high ethanol fraction attributing to a marked decrease in combustion
efficiency. However, it should be noted that the combustion efficiency is estimated higher than 97%
except the misfiring conditions, which suggests that the variations of combustion efficiency would
not make significant impact on dual-fuel combustion.

Figure 5.5 shows more significant problem of NOx emission when high ethanol fraction is
applied. Except in the misfiring conditions, the NOx emission increases due to increased combustion
temperature via thermal Zel’dovich mechanism. An important finding from Fig. 5.5 is that the
smoke shows negligible level. It was expected because the premixed combustion dominates the
overall dual-fuel combustion. Also, the presence of additional oxygen molecule in ethanol would
lead to lower in-cylinder soot formation. For all tested conditions, the opacity of the exhaust gases
is presented for the smoke emissions. Except the misfiring conditions which show very high opacity
due to unburnt hydrocarbons (i.e. “white smoke”), the smoke emission for ethanol dual-fuel
combustion shows much lower value than that for diesel-only combustion. When the ethanol
fraction is higher than 20%, the opacity level is below 3%. This is a great benefit for the combustion
optimization of the dual-fuelling engine over the conventional diesel combustion because the
smoke-NOx trade-off does not exist.

An interesting observation in Fig.5.5 is that CO and HC emissions are higher for PFI
position B compared to PFI position A. As mentioned previously, the overall trends appear to be the
same but the values differ significantly, suggesting additional source of CO and HC increase other
than the reaction quenching near the wall and ethanol charge trapped inside the crevice volume. As

86
 Figure 5.6: Combustion efficiency calculated using measured HC and CO emissions for both
PFI positions tested

mentioned previously, PFI positions A and B have differences in the intake charge preparation
influenced by the droplets-airflow interaction and the hot surface boiling. The higher CO and HC
emissions measured for PFI position B suggested that the reduced wall wetting of PFI position A
outperformed the smaller droplets of PFI position B. Lower CO and HC emissions for PFI position
A are also consistent with the extended misfiring limit seen in Fig. 5.4.

5.2 Effect of Ethanol Premixing on Dual-fuel Engine Efficiency

Although the focus of the previous section was the effect of PFI position on dual-fuel
combustion, it is worth to note that Fig. 5.4. shows increasing IMEP with increasing ethanol energy
fraction until it decreases again due to misfire. For tested conditions, a maximum of 5% IMEP
increase was observed and since the total energy was fixed, this meant 5% efficiency gain. One
might simply think that increasing apparent heat release rate seen in Fig. 5.3 explains this efficiency
gain. However, it is well known that IMEP depends not only on how much of heat energy is
released from the combustion but also at what timing in respect to the crank angle position the heat
energy is released [Bittle et al. 2012]. Higher heat release rate would certainly increase the power
output but if the combustion occurs during the later part of expansion stroke, the resulting power
might not be as high as the lower heat release occurring near TDC. This timing factor is termed
combustion phasing. Since Fig. 5.3 shows an increasing trend of the apparent heat release rate but

87
 Figure 5.7: Effect of diesel injection timing on in-cylinder pressure (top) and apparent heat
release rate (bottom). Results are from diesel-only operation (E0). The -3oCA aTDC injection
data is as in Fig. 5.3 for E0

retarded combustion phasing, it is unclear whether the increased IMEP was a result of the increased
heat release rate or properly positioned combustion phasing.

To address this question, experiments were conducted with diesel-only operation (E0) at
various injection timings that can simulate the combustion phasings of the dual-fuelling cases.
Figure 5.7 shows the in-cylinder pressure traces (top) and apparent heat release rates (bottom) for
various diesel injection timings. A trend is clear that advancing the diesel injection timing results in
earlier start of combustion and higher peak pressure. The higher peak pressure with advanced diesel
injection was observed because the drastic increase of in-cylinder pressure associated with the main

88
combustion occurred at closer to TDC where the ambient pressure and temperature were high.
During the experiments, it was found that the diesel injection timing earlier than -13°CA aTDC
would likely cause severe knocking, which limited the diesel injection timing advancement. On the
contrary, misfiring was observed for later than 1°CA aTDC injection and so no further attempts
were made. Figure 5.7 shows that aHRR decreases for advanced diesel injection timing, opposite to
the trend in the peak in-cylinder pressure. This was due to the decreased ignition delay and hence
less premixed combustion [Choi & Reitz 1999], consistent with the aHRR trend in Fig. 5.3. This
diesel-only dataset is useful to discuss the combustion phasing effect on IMEP because the range of
combustion phasings in Fig. 5.7 is much wider than that in Fig. 5.3.

The combustion phasing is best characterised by measuring a crank angle position of 50%
of total heat release (i.e. CA50) [Srinivasan et al. 2012, Olsson et al. 2001]. Therefore, CA50 for
both dual-fuelling and diesel-only operations was calculated and its correspondence with IMEP is
plotted in Fig. 5.8. As combustion parameters did not show significant difference between two PFI
positions (Fig. 5.3 and 5.4), position A was selected which gives larger ethanol operation range. In
the figure 5.8, the diesel injection timings or ethanol energy fractions are annotated for each

E40
-13oCA aTDC
E30
-8oCA aTDC
E50
E0, -3oCA aTDC
E20

0° CA aTDC

1oCA aTDC

Ethanol energy ratio variation at f ixed

diesel injection timing of -3°CA aTD C
Diesel injection timing variation at diesel-
only condition (E0)

Figure 5.8: Indicated mean effective pressure against CA50 for dual-fuelling and diesel-only
operation. For dual-fuelling cases, PFI injector was positioned at position A

89
operating condition. The first noticeable trend in Fig. 5.8 is that IMEP increases as the injection
timing advances for the diesel-only case. This explains that the advanced combustion phasing can
outperform the decreased aHRR in determining IMEP. In other words, if the timing of the
combustion is well positioned such that the main combustion occurs near TDC, the IMEP can be
higher despite lower aHRR. This finding however cannot explain the increased IMEP with
increasing ethanol fraction for the dual-fuelling case because figure 5.3 showed the timing of the
pressure rise moves away from TDC. Figure 5.8 also suggests that the combustion phasing was not
a cause for the efficiency gain of the ethanol and diesel dual-fuelling. In fact, the combustion
phasing does not vary much for the dual-fuelling cases except the misfiring E50. Therefore, it can
be concluded that the increased IMEP for high ethanol fractions does not relate to the combustion
phasing but is simply due to the increased rate of heat release.

Another interesting finding from Fig. 5.8 is that the dual-fuelling delivers higher IMEP than
diesel-only operation for the same combustion phasing of 15°CA aTDC. A quick comparison
between Fig. 5.7 and Fig. 5.8 does not seem to suggest that the peak aHRR have a marked
difference; however, a conclusion cannot be drawn until detailed analysis of the aHRR is
conducted. One way to attest the aHRR for each combustion stages of ignition delay, premixed
combustion, and late-cycle combustion is calculating the burn durations. For a fixed combustion
phasing, shorter burn duration corresponds to a higher average rate of heat release resulting in
increased power output.

Figure 5.9 shows various burn durations calculated from the apparent heat release rates of
all ethanol energy fractions tested in Fig. 5.8. The CA10-SOI corresponds to the ignition delay that
increases with increasing ethanol fraction as was discussed previously. Of high interest is the
CA50-CA10, the initial burn duration where the heat release is driven mostly by the combustion of
both diesel and ethanol in a pre-mixed mode. It is noticeable that this premixed burn duration
decreases with increasing ethanol fraction (except misfiring E50), inversely proportional to the
ignition delay trend. The late-cycle combustion (i.e. CA90-CA50) decreases even more strongly
with increasing ethanol fraction. The much stronger decrease is explained considering the different
modes of diesel combustion occurring in the different stages. In the initial stage, the combustion of
diesel is mainly premixed, and therefore the duration of this stage is not strongly affected by the
shift of burning to the premixed ethanol-air. In contrast, in the late stage, the diesel combustion
becomes limited by relatively slow mixing process. Therefore, substitution of the mixing-limited
diesel combustion with kinetically limited ethanol-air combustion leads to relatively larger changes
in the burn duration. In summary, increasing ethanol fraction leads to moderate reductions of the

90
 Figure 5.9: Burn duration for various stages of combustion at different ethanol energy fractions
(PFI injector was positioned at position A)

early phase combustion duration and strong reductions of the late phase combustion duration, which
leads to greater IMEP, up to the point of misfire where the bulk of ethanol-air fraction does not
reach sufficient temperature to fully burn.

5.3 Effect of Diesel Injection Timing on Dual-fuel Combustion

Previously in Fig. 5.8, it was found that properly positioned combustion phasing could
increase the IMEP. However, the combustion phasing for various ethanol fractions did not show
measurable variations. Therefore, it was motivated that the diesel injection timing should be further
advanced to achieve even higher efficiency. Also the maximum ethanol fraction was limited at
lower than 50% due to over-retarded combustion phasing and thereby misfiring. Advanced diesel
injection might bring the combustion phasing forward and therefore the maximum ethanol fraction
could be increased.

91
 Figure 5.10 shows the in-cylinder pressure and apparent heat release rate traces for various
ethanol energy fractions when tested at PFI position A and at three different diesel injection timings
of -8, -3 and 0°CA aTDC from left to right. The -3°CA aTDC results in Fig. 5.3 are shown again in
the middle for the purpose of direct comparison. In selecting diesel injection timings, the baseline
diesel operations which exhibited safe and stable operations in Fig. 5.7 are considered. Figure 5.10
shows the same trend as in Fig. 5.3 that the TDC pressure decreases, timing of the pressure rise is
slightly retarded evidencing increased ignition delay period, and the peak aHRR increases and
occurs at later timing as the ethanol fraction increases. The misfiring conditions are also found
when the combustion phasing is over-retarded. It is notable that compared to -3°CA aTDC injection
case, the advanced diesel injection at -8°CA aTDC has much higher in-cylinder pressure at a fixed
ethanol fraction while the in-cylinder pressure is much lower for the retarded injection at 0°CA
aTDC. Another noticeable trend is that the pressure rise occurs at more advanced timing as the
diesel injection timing is advanced. For -8°CA aTDC injection, the pressure rise happens right after
TDC suggesting very high power output. The aHRR shows the opposite trend that the advanced

-8°CA aTDC -3°CA aTDC 0°CA aTDC

Figure 5.10: In-cylinder pressures (top) and apparent heat release rates (bottom) for various
ethanol energy fractions. The diesel injection timings were varied from -8 to 0° CA aTDC as
noted at the bottom-left corner of each aHRR plot (PFI injector was positioned at position A)

92
diesel injection results in lower peak aHRR due to shortened ignition delay.

An interesting finding from Fig. 5.10 is that the maximum ethanol fraction is extended as
the injection timing is advanced. As expected, the combustion phasing moves forward for the
diesel-only operation and therefore the over-retarded combustion phasing due to ethanol
substitution could be avoided although higher ethanol fraction is applied. If the diesel injection
occurred later at 0°CA aTDC, the ethanol injection made no impact as evidenced by almost
identical in-cylinder pressure traces for E0 and E10. For higher ethanol fractions, the combustion
phasings were simply over-retarded and the combustion became unstable. These data were further
analysed to acquire the IMEP and CoV of IMEP as shown in Fig. 5.11. The IMEP trend is
consistent with the in-cylinder pressure shown in Fig. 5.10 that the IMEP increases with increasing
ethanol fraction for -8 and -3°CA aTDC injections. The 0°CA aTDC injection was an exception as
there is no efficiency gain due to over-retarded combustion phasing. Apparently, the advanced
diesel injection results in higher IMEP at fixed ethanol energy fraction as the pressure rise is placed
near TDC. When the misfire limit is set at 5% CoV of IMEP, the maximum ethanol fraction can be
extended from 15% to 70% by advancing the diesel injection timing from 0 to -8°CA aTDC. This
result clearly demonstrates that the diesel injection timing should be advanced to increase the

5% Misfire limit

Figure 5.11: IMEP and CoV of IMEP for various ethanol fractions and diesel injection timings
(PFI injector was positioned at position A)

93
maximum ethanol fraction when the operation is limited by misfiring. For the tested conditions of
this study, the maximum ratio of the ethanol energy to the total fuel energy was 70%.

Figure 5.12 shows the ignition delay, CA50 and burn duration for the same conditions of
Fig. 5.11. An expected trend is found that the ignition delay increases with increasing ethanol
fraction and decreases with advancing diesel injection timing. The CA50 appears to be very similar
for various ethanol fractions but advances as the diesel injection occurs at an earlier timing. This
advanced combustion phasing should play an important role in increasing the IMEP (or efficiency).
Unlike the ignition delay and CA50, however, the burn duration trend raises a question. The
decreasing burn duration with increasing ethanol fraction is consistent for all diesel injection
timings, similar to Fig. 5.9. Since the burn duration decreases while CA50 is fixed, the increasing
IMEP with increasing ethanol fraction was explained by fast burning of the premixed charge, which
is also consistent with Figs. 5.8 and 5.9.

However, figure 5.12 shows that at a fixed ethanol fraction, the burn duration is longer for

Figure 5.12: Ignition delay, CA50 and burn duration for the data as in Fig. 5.10 (PFI injector
was positioned at position A)

94
advanced diesel injection despite the fact that higher IMEP is measured. The result suggests that
both the burn duration and CA50 should be considered to explain the IMEP trend.

Previously in Fig. 5.8, it was demonstrated that advanced CA50 can increase the IMEP as
the main combustion is positioned closer to TDC. It was likely that the impact of advanced CA50
on IMEP was more significant than that of increased burn duration. Once again, the advanced diesel
injection helps achieve high IMEP and for the tested conditions of this study, 15% increase in IMEP
(or efficiency as the total fuel energy was fixed) was measured when 70% ethanol fraction was
achieved by injection diesel fuel at advanced -8°CA aTDC. Further to note, the control of diesel
injection timing and its impact on combustion suggests a great advantage of ethanol and diesel dual-
fuelling over other kinetics-controlled combustion regimes such as HCCI. The ability to control
combustion conditions using diesel injection timing will ensure the optimum running of engines
that respond promptly to fast-changing engine speed and load conditions.

From Fig. 5.12, it is observed that at any particular diesel injection timing, very high
ethanol fraction led to over retarded combustion phasing and thereby misfiring. However, when
comparing different diesel injection timing cases, it was found that combustion phasing is not the
only cause for the misfiring. For example, two circles in CA50 plot of Fig. 5.12 illustrate the same
CA50 for 0°CA aTDC and -3°CA aTDC observed at 0% and 50% ethanol energy fraction,
respectively. The burn duration appear to be the same for these cases. Despite the same CA50 and
burn duration, -3°CA aTDC injection at 50% ethanol fraction represents misfiring condition
whereas 0°CA aTDC injection at 0% ethanol fraction falls into a normal operation condition. To
further discuss this interesting finding, the combustion phasings (CA50) from all individual
pressure cycles along with the mean values are plotted in Fig. 5.13 for the same data of Fig. 5.12.
The variation in CA50 shown at the figure top demonstrates that even though the mean combustion
phasing was the same for selected conditions, the error range was much higher for -3°CA aTDC at
50% ethanol fraction. Such high fluctuations in the combustion phasing are characteristics of
misfiring conditions. Furthermore, the plot of CA10 (i.e. the start of combustion) shows increased
fluctuations for -3°CA aTDC at 50% ethanol fraction, consistent with CA50 results. Therefore, it
can be concluded that both CA50 and fluctuations of CA50 should be considered to explain very
high CoV IMEP of misfiring conditions.

Figure 5.14 shows how the ethanol fraction affects the exhaust CO, NOx, HC and smoke
emissions. The PFI position A results from Fig. 5.5 at -3° CA aTDC are replotted for comparison
purposes. Figure 5.13 shows the same trend as in Fig. 5.5, for all the injection timings tested. CO

95
 Figure 5.13: CA50 (at top) and CA10 (at bottom) values calculated from for the individual
cycles along with mean values for the same conditions as of Figs. 5.10 and 5.11

emissions decreases first and then increases, HC emissions continuously increases with increase in
ethanol fraction. NOx emissions increases and smoke decreases except for misfiring cases.

For fixed ethanol energy fraction, figure 5.13 shows that the advanced diesel injection
decreases both the HC and CO emissions as the combustion temperature increases. It is promising
because the advanced injection achieves the highest ethanol fraction as well as the highest
efficiency as discussed previously. It is notable that for -8oCA aTDC diesel injection, NOx emission
increases 50% from E0 to E60. Previously in Fig. 5.11, about 10% of IMEP increase was observed
and therefore the NOx increase outperforms the power or efficiency gain. Therefore, the
optimization of efficiency, NOx, and HC/CO emissions are required for successful operation of the

96
 Figure 5.14: Carbon monoxide (CO), unburnt hydrocarbon (HC), nitrogen oxides (NOx) and
exhaust gas opacity (i.e. smoke level) emissions for all diesel injection timings tested (PFI
injector was positioned at position A)

dual-fuelling engine. When the ethanol fraction is higher than 20%, the opacity level is below 3%,
which is equivalent to less than 0.6 filtered smoke number (FSN) [Lapuerta et al. 2005]. This is a
great benefit for the combustion optimization of the dual-fuelling engine over the conventional
diesel combustion because the smoke-NOx trade-off does not exist. One can reduce NOx emission
effectively using a well-known technique of exhaust gas recirculation while still taking great
advantages of increased efficiency associated with the ethanol dual-fuel combustion.

5.4 Summary of the results from dual-fuelling engine tests

The following table summarizes the key results from engine tests.

97
 Table 5.3: Summary of all the results from dual-fuelling experiments

Parameter IMEP CoV of HC CO NOx Opacity

(kPa) IMEP (g/kW- (g/kW- (g/kW- (%)
(%) H) H) H)
PFI Position PFI Position A E0 872 2.71 0.68 17.98 3.43 22
E20 868 2.57 0.76 10.45 3.53 3.2
E30 900 3.05 0.84 10.54 3.64 2.4
E40 913 2.98 0.97 14 4.63 2.2
E50 880 16.55 2.23 22.04 3.84 61.5
PFI Position B E0 872 2.71 0.68 17.98 3.43 22
E20 877 3.44 1.22 11.28 3.31 0.12
E30 908 2.03 1.38 13.63 3.65 0.15
E35 906 14.35 2.85 109.3 1.78 2.9
Injection 0° CA aTDC E0 864 2.69 0.98 20.04 3.19 22
Timing E10 808 2.1 1.12 19.39 3.43 6.4
E15 808 3.55 1.21 18.89 3.57 5.1
E20 818 15.77 1.61 20.68 3.58 64.1
-3° CA aTDC E0 872 2.71 0.68 17.98 3.43 22
E20 868 2.57 0.76 10.45 3.53 3.2
E30 900 3.05 0.84 10.54 3.64 2.4
E40 913 2.98 0.97 14 4.63 2.2
E50 880 16.55 2.23 22.04 3.84 61.5
-8° CA aTDC E0 897 1.45 0.49 16.55 6.03 16.2
E20 912 1.18 0.62 9.95 6.29 3.8
E30 936 1.33 0.77 9.25 6.63 2.4
E40 974 1.96 0.84 8.47 6.70 1.4
E50 1000 2.02 0.91 11.96 6.87 1.2
E60 1014 1.96 0.99 13.82 7.14 0.95
E70 986 2.98 1.46 25.52 6.86 24.5
E80 970 6.79 13.74 829 4.22 90

98
Chapter 6

Conclusions and Future Works

6.1 Conclusions

An experimental study was conducted to investigate ethanol PFI sprays in simulated intake
manifold injection conditions. A separate spray visualization facility was built for this purpose.
Mie-scattering and high-speed shadowgraph imaging were performed to study the effects of
injection rate, injection duration, cross-flow, and ambient pressure on the ethanol and gasoline PFI
sprays. Momentum flux measurements were also performed for all injectors tested. Many important
spray parameters including spray tip penetration, mean droplet diameter, and number of droplets
were determined from the Mie-scattered liquid spray images. Vapour boundaries obtained from the
shadowgraph movies were used to discuss evaporation processes.

The main conclusions drawn from the spray visualization studies are summarized as

x From typical PFI spray images of ethanol, it was found that the mean droplet diameter
decreased with increasing time aSOI primarily due to droplet break-up occurring for a
steady period of fuel injection. It was also found that the total number of droplets increased
by the addition of droplets from continued fuel injection as well as the droplet break-up and
atomization.

x Detailed analysis was performed for sprays at a fixed time aSOI and it was found that the
number of droplets decreased and the mean diameter increased with increasing distance
from the nozzle. The droplet size distribution results clarified that small droplets evaporated
earlier than big droplets and therefore the mean diameter increased with increasing distance

99
 from the nozzle. Clearly, both droplet break-up and evaporation impacted the ethanol PFI
sprays.

x When different fuels such as ethanol, E50 and gasoline were compared, an unexpected
trend was observed such that ethanol had 1.1 mm lower penetration than gasoline and E50
while the latter two had similar tip penetrations. It was explained that although ethanol had
higher viscosity and density than gasoline, droplets of heavy gasoline components
atomized/evaporated at a slower rate than ethanol droplets. The lower mean droplet
diameter and higher number of droplets for ethanol were also explained by the effect of
heavy molecules contained in gasoline.

x The actual injection duration was 0.5 ms longer for the high flow-rate injector for the same
electronic pulse width. It is due to the slower response of the needle closing associated with
increased flow resistance inside the nozzle for the high flow-rate injector. The mean droplet
diameter was explained to be higher for the higher flow-rate injector simply because of the
larger nozzle-hole size.

x The transient sprays had a similar number of droplets and mean droplet diameter to those of
the quasi-steady sprays prior to the end of injection. However, after the end of injection, the
number of droplets and mean droplet diameter were less for the short injection sprays. A
decrease of 10 μm in mean droplet diameter was observed for transient sprays compared to
quasi-steady sprays at 5 ms aSOI. This was explained by the higher evaporation rate near
the nozzle, which then propagated downstream after the end of injection. This was
consistent with the shadowgraph images that showed larger vapour regions compared to the
regions of liquid droplets.

x Sprays at quiescent and cross-flow ambient air conditions penetrated at a similar rate during
the fuel injection period because the injection momentum was dominant over the radial
momentum transfer from the cross-flow air. However after the end of injection, the cross-
flow shifted the spray toward the airflow direction and both the number of droplets and
mean diameter were lower than those of the quiescent sprays. A 16% decrease in mean
droplet diameter was measured for cross-flow conditions. This was explained by increased
evaporation rates which were expected larger for two reasons. First, evaporation rates were
directly increased since the relative velocity was larger. Second, since relative velocity was
larger, droplet break-up was expected to occur more quickly, leading to smaller droplets

100
 which then had greater evaporation rates due to larger surface area compared with the
parent droplet.

x For ambient pressure variation, a well-expected trend was observed that at higher ambient
pressure, droplets travel at a slower velocity due to an increased drag and hence lower spray
tip penetration results at fixed time aSOI. A 25% decrease in spray tip penetration was
observed at 200 kPa ambient pressure compared to atmospheric pressure. The sprays
became denser as the ambient pressure increased resulting in higher number of droplets at a
fixed distance from the nozzle. Also, increased mean droplet diameter was measured
because of the lower evaporation rate for higher ambient pressure. However, no marked
change in the total number of droplets was found implying there was no significant
difference in droplet break-up.

Dual-fuel combustion using ethanol port injection and diesel in-cylinder direct injection
was investigated in a single-cylinder, automotive-size diesel engine. The influence of ethanol port
injector position on dual-fuel combustion and emissions was clarified. The effect of ethanol energy
fraction on the engine efficiency was discussed. Later, the diesel injection timing was varied to
attest its ability to optimize the combustion phasing and extend the maximum ethanol substitution.
Major findings from the engine performance and emissions testing can be summarized as follows:

x Global phenomena such as in-cylinder pressure and apparent heat release rate do not show
variations for two different ethanol port injector positions tested in the present study.
However, the misfiring occurs at 50% ethanol energy fraction for the port injector located
closer to the valves, whereas further upstream location resulted in misfiring at 35% ethanol
energy fraction. This implies another parameter that plays a significant role other than the
ethanol droplet size.

x It is suggested that the injector closer to the intake valves can take great advantages of hot
surface boiling of ethanol droplets and hence increased vaporization and reduced wall
wetting. Measured engine-out emissions of unburnt hydrocarbon and carbon monoxide are
consistent with this explanation, showing lower values when the injector is placed closer to
the intake valves.

x With increasing ethanol supply via port injection, the engine power output and efficiency
increase until they are limited by over-retarded combustion phasing leading to misfiring. It

101
 is suggested that increased efficiency is not a result of the combustion phasing but increased
rate of heat release associated with faster burning of the premixed ethanol-air-diesel
mixture. For a given conditions of this study, a maximum of 10% efficiency gain was
achieved with 60% ethanol energy fraction.

x At a fixed ethanol energy fraction, advancing the diesel injection timing decreases the
ignition delay and increases in-cylinder pressure due to combustion occurring at higher
ambient pressure and temperature. Advancing the injection timing from 0°CA aTDC to -
8°CA aTDC resulted in extending the misfire limit from E20 to E80.

x With increasing ethanol energy fraction, an increase of HC, CO and NO x emissions is

observed. At the maximum ethanol fraction tested in this study, an increase of 30% in CO
emissions, 20% increase in NOx emissions and 200% increase in HC emissions were
observed compared to diesel-only operation. However, these increased gas emissions could
be controlled easily because the smoke is negligible if more than 20% ethanol fraction is
applied for a given operating condition of this study

6.2 Future Works

Listed below are suggestions for future investigations that can provide further details of findings of
this study:

x This thesis addressed the PFI spray evaporation characteristics at intake manifold
conditions with ethanol as primary fuel at various injection and ambient conditions. The
presented experimental studies were conducted at room temperature, which is lower than
the actual engine intake manifold temperatures. While the essence of the spray
characteristics would not change significantly, it would be worthwhile to confirm them at
more realistic conditions such as swirl flow, intake manifold heating, valve movements etc.,
if major investment on a new spray chamber rig can be made. From my experience
throughout this study, about half a million dollars would be required to achieve this goal.

x The present study was limited to downstream of the spray (15 mm away from nozzle tip)
due to the difficulties in imaging the near-nozzle high dense sprays. By creating a planar
illumination using laser based optical techniques, it would be possible to investigate the
ligament break-up more closely and this study can further help in complete understanding
of the PFI sprays.

102
x It is proved that the spray preparation plays a key role in subsequent dual-fuelling engine
combustion. This thesis addressed the importance of PFI injector location on charge
preparation in the intake manifold. During the experiments, it was observed that in addition
to the droplet characteristics, manifold/wall wetting plays a crucial role in successful
operation of engine. More detailed experiments on wall wetting phenomenon both in the
intake manifold and intake valve seat would help better understand the impact of intake
sprays on engine performance and engine-out emissions, particularly unburnt hydrocarbon.
Once again, this requires significant investment on a new experimental rig to create such
realistic environment.

x The dual-fuel combustion operation range is limited by high CoV of IMEP at large ethanol
fractions due to over-retarded combustion phasing. This could be overcome by advancing
the diesel injection timing, the advance angle was limited to -8° CA aTDC and beyond this
the engine operation was limited by maximum in-cylinder pressures. However by
employing an additional EGR, it may be possible to extend ethanol operation range and it is
expected to have positive impact on NOx emissions. This would be a worth of another
Ph.D. study.

x Once engine performance and emissions testing is completed, a few operating conditions
might be selected for optical diagnostics. Since an optical diesel engine is readily available
at the UNSW Engine Research Lab, visualisation of in-cylinder processes is suggested to
clarify details of mixture formation of two fuels and ignition mechanism.

103
References

Abu-Qudais, M., Haddad, O., and Qudaisat, M., “The effect of alcohol fumigation on diesel engine
performance and emissions”, Energy Conversion and Management 41(4):389–399, 2000.

Allied Vision Technologies GmbH, “SmartView: Operating AVT cameras with SmartView:
V2.3.1”,
http://www.alliedvisiontec.com/fileadmin/content/PDF/Software/AVT_software/AVT_software_stu
ff/FirePackage_SmartView_V2.3.1_en.pdf, Sep. 2012.

Alkidas, A.C., “The effects of fuel preparation on hydrocarbon emissions of a S.I. engine operating
under steady-state conditions”, SAE Technical Paper 941959, 1994.

Ahmad, I.N., Babu, M., and Ramesh, A., “Experimental investigations of different parameters
affecting the performance of a CNG-Diesel dual fuel engine”, SAE Technical Paper 2005-01-3767,
2005.

Anand, T.N.C., Deshmukh, D., Madan, M.A., and Ravikrishna, R.V., “Optical characterization of
PFI gasoline sprays: Effect of injection pressure”, SAE Technical Paper 2010-32-0067, 2010a.

Anand, T.N.C., Deshmukh, D., Madan, M.A., and Ravikrishna, R.V., “Laser-based spatio-temporal
characterization of port fuel injection (PFI) sprays”, International Journal of Spray and Combustion
Dynamics 2(2):125–150, 2010b.

Aoki, F., Nakase, E.S., and Oumura, H., “Spray analysis of port fuel injector”, SAE Technical
Paper 2005-01-1154, 2005.

Antonius, R., “Interpreting quantitative Data with SPSS”, Sage Publications Ltd, California, 2004.

Australia’s emissions projections 2010, Department of Climate Change and Energy Efficiency
2010, DCCEE, Canberra, ACT. http://www.climatechange.gov.au/

Bayraktar, H., and Durgun, O., “Investigating the effects of LPG on spark ignition engine
combustion and performance”, Energy Conversion and Management 46(13–14):2317–2333, 2005.

Bianchi, G.M., Brusianim, F., Postrioti, L., Grimaldi, C.N., DiPalma, S., Matteucci, L., Marcacci,
M., and Carmignani, L., “CFD analysis of injection timing influence on mixture preparation in PFI
motorcycle engine”, SAE Technical Paper 2006-32-0022, 2006.

104
Bhave, A., Kraft, M., Montorsib, L., and Maussc, F., “Sources of CO emissions in an HCCI engine:
A numerical analysis”, Combustion and Flame 144(3):634–37, 2006.

Bhave, A., Kraft, M., Mauss, F., Oakley, A., and Zhao, H., “Evaluating the EGR-AFR operating
range of a HCCI engine”, SAE Technical Paper 2005-01-0161, 2005.

Bittle, J., Zheng, J., Xue, X., and Jacobs, T., “Cylinder-to-cylinder variation sources in diesel low
temperature combustion and the influence they have on emissions”, presented at Spring Technical
Meeting of the Central States Section of the Combustion Institute, Dayton, Ohio, USA, April 22–
24, 2012.

Blottnitz, H., and Curran, M.A., “A Review of assessments conducted on bio-ethanol as a

transportation fuel from a net energy, greenhouse gas, and environmental life cycle perspective”,
Journal of Cleaner Production 15(7):607–619, 2007.

Botero, M.L., Huang, Y., Zhu, D.L., Molina, A., Law, C.K., “Droplet combustion of ethanol, diesel,
castor oil biodiesel, and their mixtures”,7th US National Combustion Meeting of the Combustion
Institute, Atlanta, GA, USA, March 20–23, 2011.

Bosch Injection Valve EV 6, http://www.bosch-

motorsport.de/pdf/components/injection_valves/ev_6.pdf, Sep. 2012.

Chaplin, J., Janius, R.B., “Ethanol fumigation of a compression-ignition engine using advanced
injection of diesel fuel”, Transactions of the American Society of Agricultural Engineers 30
(3):610–614, 1987.

Chen, R.H., Chiang, L.B., Chen, C.N., and Lin, T.H., “Cold-start emissions of an SI engine using
ethanol gasoline blended fuel”, Applied Thermal Engineering 31(8–9):1463–1467, 2011.

Chen, H., Shi-Jin, S., and Wang, J., “Study on combustion characteristics and PM emission of
diesel engines using ester-ethanol-diesel blended fuels”, Proceedings of the Combustion Institute
31(2):2981–2989, 2007.

Cheng, W., Hamrin, D., Heywood, J., Hochgreb, S., Min, K., and Norris M., “An overview of
hydrocarbon emissions mechanisms in spark-ignition engines”, SAE Technical Paper 932708,
1993.

Choi, C.Y., and Reitz, R.D., “An experimental study on the effects of oxygenated fuel blends and
multiple injection strategies on DI diesel engine emissions”, Fuel 78(11):1303–17, 1999.

Christensen, M., Johansson, B., Amnéus, P., and Mauss, F., “Supercharged homogeneous charge
compression ignition”, SAE Technical Paper 980787, 1998.

Christensen, M., Hultqvist, A., and Johansson, B., “Demonstrating the multi fuel capability of a
homogeneous charge compression ignition engine with variable compression ratio”, SAE Technical
Paper 1999-01-3679, 1999.

Cooney, C.P., Yeliana, Worm, J.J., and Naber, J.D., “Combustion characterization in an internal
combustion engine with ethanol-gasoline blended fuels varying compression ratios and ignition
timing”, Energy & Fuels 23(5):2319–2324, 2005.

Colban, W.F., Miles P.C., and Oh, S., “On the cyclic variability and sources of unburned

105
hydrocarbon emissions in low temperature diesel combustion Systems”. SAE Technical Paper
2007-01-1837, 2007.

Daily J. W., “Cycle-to-cycle variations: A chaotic process?”, Combustion Science and Technology
88 (57):149–102, 1987.

Daniels, C. and Evers, L., "The influence of mixture preparation on a warm 1.9L Ford engine", SAE
Technical Paper 940444, 1994.

Darby, M., and Pettrie, G., “Australia biofuels annual report 2009”,
http://gain.fas.usda.gov/Recent%20GAIN%20Publications/General%20Report_Canberra_Australia
_6-1-2009.pdf, Aug. 2012.

Dec, J.E., “Advanced compression-ignition engines—Understanding the in-cylinder processes”,

Proceedings of the Combustion Institute 32(2):2727–2742, 2009.

Demirbas, A., “Fuel alternatives to gasoline”, Energy Sources, Part B: Economics, Planning, and
Policy 2(3):311–320, 2007.

Demirbas, A., “Conversion of biomass using glycerin to liquid fuel for blending gasoline as
alternative engine fuel”, Energy Conversion and Management 41(16):1741–1748, 2000.

DiPardo, J., “Outlook for biomass ethanol production and demand”, Energy Information
Administration, U.S. Department of Energy, http://webapp1.dlib.indiana.edu/cgi-
bin/virtcdlib/index.cgi/4265704/FID2/pdf/Multi/biomass.pdf, Dec. 2012.

Dumouchel, C., Cousin, J., and Triballier, E., “On the role of the liquid flow characteristics on low
Weber-number atomization processes”, Experiments in Fluids 38(5):637–647, 2005.

Ekholm, K., Karlsson, M., Tunestål, P., Johansson, R., Johansson, B., and Strandh, P., “Ethanol-
diesel fumigation in a multi-cylinder engine”, SAE Technical Paper 2008-01-0033, 2008.

Fang, T., and Lee, C.F., “Low sooting combustion of narrow-angle wall-guided sprays in an HSDI
diesel engine with retarded injection timings”, Fuel 90(4):1449–1456, 2011.

Gauding, M., Pawlowski, A., Felsch, C., Sonntag, B., Kenner, R., and Peters, N., “Fundamental
investigation of diesel spray-spray interaction for cluster-hole nozzles”, presented at 11th Triennial
International Annual Conference on Liquid Atomization and Spray Systems-ICLASS 2009, Vail,
Colorado USA, July, 2009.

Gao, J., Jiang, D., and Huang, Z., “Spray properties of alternative fuels: A comparative analysis of
ethanol–gasoline blends and gasoline”, Fuel 86:1645–1650, 2007.

Gao, J., Jiang, D., Huang, Z., and Wei, Q., “Characteristics of non evaporating free sprays of a
high-pressure swirl injector under various ambient and injection pressures”, Energy & Fuels
19:1906–1910, 2005.

Gehmlich, R.K., “Rate-of-injection measurements for modern diesel injectors”, Personnel

communication with Dr. Shawn Kook, 2009.

Goldsworthy, L.,“Fumigation of a heavy duty common rail marine diesel engine with ethanol–water
mixtures”, Experimental Thermal and Fluid Science47:48–59, 2013.

106
Hansen, A.C., Zhang, Q., and Lyne, P.W.L., “Ethanol-diesel fuel blends-A review”, Bioresource
Technology 96(3):277–285, 2005.

He, B.Q., Wang, J., Shuai, S., and Yan, X., “Homogeneous charge combustion and emissions of
ethanol ignited by pilot diesel on diesel engines”, SAE Technical Paper 2004-01-0094, 2004.

Helmantel, A., and Denbratt, I., “HCCI operation of a passenger car common rail DI diesel engine
with early injection of conventional diesel fuel”, SAE Technical Paper 2004-01-0935, 2004.

Heneina, N.A., and Tagomori, M.K., “Cold-start hydrocarbon emissions in port-injected gasoline
engines”, Progress in Energy and Combustion Science 25:563–593, 1999.

Heywood, J.B. “Internal combustion engine fundamentals”, first ed., McGraw-Hill, New York,
1988.

Himabindu, M., Sathyanarayanan, G., Thomson, S., and Mahalakshmi, N.V., “Effect of premixed
charge of ethanol and exhaust gas recirculation on the performance and emission characteristics in
an ethanol-diesel fueled HCDC engine-An experimental investigation”, SAE Technical Paper 2008-
01-2609, 2008.

Hsieha, W., Chenb, R., Wub, T., and Lin, T., “Engine performance and pollutant emission of an SI
engine using ethanol-gasoline blended fuels”, Atmospheric Environment 36(3):403–410, 2002.

Hung, D.L.S., Harrington, D.L., Gandhi, A.H., Markle, L.E., Parrish, S.E., Shakal, J.S., Sayar, H.,
Cummings, DS., and Kramer, JL., “Gasoline fuel injector spray measurement and characterization -
A new SAE J2715 recommended practice”, SAE Technical Paper 2008-01-1068, 2008.

Hung, D.L.S., Humphrey, W.A., Markle, L.E., Chmiel, D.M., Ospina, C.A., and Brado, F.E., “A
novel transient drop sizing technique for investigating the role of gasoline injector sprays in fuel
mixture preparation”, SAE Technical Paper 2004-01-1349, 2004.

Hyvönen, J., Haraldsson, G., and Johansson, B., “Operating range in a multi cylinder HCCI Engine
using variable compression ratio”, SAE Technical Paper 2003-01-1829, 2003.

Jian, D., Xiaohong, G., Gesheng, L., and Xintang, Z., “Study on diesel-LPG dual fuel engines”,
SAE Technical Paper 2001-01-3679, 2001.

Karim, G.A., “A review of combustion processes in the dual fuel engine—The gas diesel engine”,
Progress in Energy and Combustion Science 6(3):277–285, 1980.

Kato, S., Hayashida, T., and Iida, M., “The influence of port fuel injection on combustion of a small
displacement engine for motorcycle”, SAE Technical Paper 2007-32-0009, 2007.

Kim, H., Yoon, S., and Lai, M.C., “Study of correlation between wetted fuel footprints on
combustion chamber walls and UBHC in engine start processes”, International Journal of
Automotive Technology 6(5):437–444, 2005.

Kokjohn, S.L., Splitter, D.A., Hanson, R.M., Reitz, R.D., Manente, M., and Johansson, B.,
“Modeling charge preparation and combustion in diesel fuel, ethanol, and dual-fuel PCCI engines”,
Atomization and Sprays 21(2):107–119, 2011.

Kong, S., and Reitz, R.D., “Numerical study of premixed HCCI engine combustion and its

107
sensitivity to computational mesh and model uncertainties”, Combustion Theory and Modelling
7(2):417–433, 2003.

Kook, S., Le, M.K., Padala, S., and Hawkes, E.R., “Z-type Schlieren setup and its application to
high-speed imaging of gasoline sprays”, SAE Technical Paper 2011-01-1981, 2011.

Kook, S., Pickett, L.M., Musculus, M.P.B., and Gehmlich, R.K., “Influence of diesel injection
parameters on end-of-injection liquid length recession”, SAE International Journal of Engines
2(1):1194–210, 2009. SAE Technical Paper 2009-01-1356, 2009.

Kook, S., Bae, C., Miles, P., Choi, D., and Pickett. L.M., "The Influence of charge dilution and
injection timing on low-temperature diesel combustion and emissions", SAE International Journal
of Fuels and Lubricants 114(4):1575–95, 2005. SAE Technical Paper 2005-01-3837, 2005.

Krishnan, S.R., Biruduganti, M., Mo,Y., Bell, S.R., and Midkiff, K.C., “Performance and heat
release analysis of a pilot-ignited natural gas engines”, International Journal of Engine Research
3(3):171–184, 2002.

Lai, M.C., Zhao, F.Q., Amer, A.A., and Chue, T.H., “An Experimental and analytical investigation
of the spray structure from automotive port injectors”, SAE Technical Paper 941873, 1994.

Lapuerta, M., Martos, F.J., and Cárdenas, M.D., “Determination of light extinction efficiency of
diesel soot from smoke opacity measurements”, Measurement Science and Technology
16(10):2048–2055, 2005.

Lim, S.L., “Intake spray imaging for ethanol-fuelled cars”, UG thesis, Engine Research Laboratory,
University of New South Wales, 2011.

Lin, Y., and Tanaka, S., “Ethanol fermentation from biomass resources: current state and
prospects”, Applied Microbiology and Biotechnology 69(6):627–642, 2005.

Liu, Y., “Rate of injection measurement for modern car injectors”, UG thesis, Engine Research
Laboratory, University of New South Wales, 2010.

Lu, X., Ji, L., Ma, J., and Huang, Z., “Improved NOx and smoke emission characteristics of a
biodiesel fueled engine with the port fuel injection of various premixed fuels”, Energy & Fuels
22(6):3798–3805, 2008a.

Lu, X., Ji, L., Ma, J., and Huang, Z., “Simultaneous reduction of NOx emission and smoke opacity
of biodiesel-fueled engines by port injection of ethanol”, Fuel 87(7):1289–1296, 2008b.

Ma, J., Lü, X., Ji, L., and Huang, Z., “Evaluation of SCCI potentials in comparison to HCCI and
conventional DICI combustion using n-heptane”, Energy & Fuels 22(2):954–60, 2008.

Mancaruso, E., Vaglieco, B.M., “Ethanol and diesel fuel in Euro5 single cylinder research engine”,
24th Meeting of the Italian Section of the Combustion Institute, Rome, Italy, October 24–26, 2011.

MATLAB version 7.9.0.529, Natick, Massachusetts: The MathWorks Inc., 2010.

Matsumoto, A., Moore, W.R., Lai, M.C., Zheng, Y., Foster, M., Xie, X.B., Yen, D., Confer, K., and
Hopkins, E., “Spray characterization of ethanol gasoline blends and comparison to a CFD model for
a gasoline direct injector”, SAE Technical Paper 2010-01-0601, 2010.

108
Maurya, R.K., and Agarwal, A.K., “Experimental study of combustion and emission characteristics
of ethanol fuelled port injected homogeneous charge compression ignition (HCCI) combustion
engine”, Applied Energy 88(4):1169–1180, 2011.

Maurya, R.K., and Agarwal, A.K., “Combustion and emission behavior of ethanol fuelled
homogeneous charge compression ignition (HCCI) engine”, SAE Technical Paper 2008-28-0064,
2008.

McGee, J., Curtis, E., Russ, S., and Lavoie, G., “The effects of port fuel injection timing and
targeting on fuel preparation relative to a pre-vaporized system”, SAE Technical Paper 2000-01-
2834, 2000.

Megaritisa, A., Yap, S., and Wyszynskic, M.L., “Effect of water blending on bioethanol HCCI
combustion with forced induction and residual gas trapping”, Energy 32(12):2396–2400, 2007.

Merola, S.S., Sementa, P., Tornatore, C., and Vaglieco, B.M., “Effect of the fuel injection strategy
on the combustion process in a PFI boosted spark-ignition engine”, Energy 35(2):1094–1100, 2010.

Meyer, R., and Heywood, J.B., “Effect of engine and fuel variables on liquid fuel transport into the
cylinder in port-injected SI engines”, SAE Technical Paper 1999-01-0563, 1999.

Michael, D., “The effect of changing fuel injector pulse-width on fuel spray development in a
simulated intake pipe”, UG thesis, Engine Research Laboratory, University of New South Wales,
2010.

Milovanovic, N., Chen, R., and Turner, J., “Influence of the variable valve timing strategy on the
control of a homogeneous charge compression (HCCI) engine”, SAE Technical Paper 2004-01-
1899, 2004.

Moon, S., Bae, C., Choi, J., and Abo-Serie, E., “The influence of airflow on fuel spray
characteristics from a slit injector”, Fuel 86(3):400–409, 2007.

Morin, C., Chauveau, C., Dagaut, P., Gökalp, I., and Cathonnet, M., “Vaporization and oxidation of
liquid fuel droplets at high temperature and high pressure: application to n-alkanes and vegetable oil
methyl esters”, Combustion Science and Technology 176(4):499–529, 2004.

Naber, J.D., and Siebers, D.L., “Effects of gas density and vaporization on penetration and
dispersion of diesel sprays”, SAE Technical Paper 960034, 1996.

Nakano, M., Mandokoro, Y., Kubo1, S., and Yamazaki, S., “Effects of exhaust gas recirculation in
homogeneous charge compression ignition engines”, International Journal of Engine Research
1(3):269–279, 2000.

Nassiri, T.A., Nozari, A.L., and Etghani, M., “Numerical study of the effect of fuel spray velocity
and targeting on fuel evaporation rate in a port fuel injection engine using KIVA3-V2 simulation
code”, presented at 2007 Fall Conference of the ASME Internal Combustion Engine Division
ICEF07, Charleston, South Carolina, USA, October 14–17, 2007.

Noguchi, N., Terao, H., Sakata, C., “Performance improvement by control of flow rates and diesel
injection timing on dual-fuel engine with ethanol”, Bioresource Technology 56:35–39, 1996

Nomuraa, H., Ujiiea, Y., Rathb, H.J., Satoc, J., and Konod, M., “Experimental study on high-

109
pressure droplet evaporation using microgravity conditions”, Proceedings of Combustion Institute
26 (1):1267–1273, 1996.

Nuglisch, H.J., and Nally, S., “Atomization versus targeting for port-fuel injectors - An
unresolvable conflict?”, International Journal of Fluid Mechanics Research 24(1–3):117–127,
1997.

Ogawa, H., Setiapraja, H., and Nakamura, T., “Improvements to premixed diesel combustion with
ignition inhibitor effects of premixed ethanol by intake port injection”, SAE Technical Paper 2010-
01-0866, 2010.

Oh, H., Bae, C., and Min, K., Spray and combustion characteristics of ethanol blended gasoline in a
spray guided DISI engine under lean stratified operation, SAE Technical Paper 2010-01-2152,
2010.

Olsson, J., Tunestål, P., and Johansson, B., “Closed-loop control of an HCCI engine”, SAE
Technical Paper 2001-01-1031, 2001.

Otsu, N., “A threshold selection method from gray-level histograms”, IEEE Transactions on
Systems Man and Cybernetics 9(1):62–66, 1979.

Park, S.H., Kim, H.J., Suh, H.K., and Lee, C.S., “Atomization and spray characteristics of
bioethanol and bioethanol blended gasoline fuel injected through a direct injection gasoline
injector”, International Journal of Heat and Fluid Flow 30(6):1183–1192, 2009.

Park, S., Cho, H., Yoon, C., and Min, K., “Measurement of droplet size distribution of gasoline
direct injection spray by droplet generator and planar image technique”, Measurement Science and
Technology 13(6):859–864, 2002.

Passarini, L.C., and Nakajima, P.R., “Development of a high-speed solenoid valve: An investigation
of the importance of the armature mass on the dynamic response”, Journal of the Brazilian Society
of Mechanical Sciences and Engineering 25(4):329–335, 2003.

Payri, R., García, J.M., Salvador, F.J., and Gimeno, J., “Using spray momentum flux measurements
to understand the influence of diesel nozzle geometry on spray characteristics”, Fuel 84(5):551–
561, 2005.

Perry, R., and Gee, I.L., “Vehicle emissions in relation to fuel composition”, Science of the Total
Environment 169(1–3):149–156, 1995.

Pickett, L.M., Siebers, D.L., and Cherian, I., “Relationship between ignition processes and the lift-
off length of diesel fuel jets”, SAE Technical Paper 2005-01-3843, 2005.

Poulopoulos, S.G., Samaras, D.P., and Philippopoulos, C.J., “Regulated and unregulated emissions
from an internal combustion engine operating on ethanol-containing fuels”, Atmospheric
Environment 35(26):4399–4406, 2001.

Pumphrey, J.A., Brand, J.I., and Scheller, W.A., “Vapour pressure measurements and predictions
for alcohol–gasoline blends”, Fuel 79(11):1405–1411, 2000.

Ren, W., and Sayar, H., “Influence of nozzle geometry on spray atomization and shape for port fuel
injector”, SAE Technical Paper 2001-01-0608, 2001.

110
Rodríguez-Fernández, J., Tsolakis, A., Theinnoi, K., Snowball, J., Sawtell, A., and York, A.P.E.,
“Engine performance and emissions from dual fuelled engine with in-cylinder injected diesel fuels
and in-port injected bioethanol”, SAE Technical Paper 2009-01-1853, 2009.

Roptary, M., and Webb, R., “Fuel ethanol-Background and policy issues”, Australian Parliamentary
Library,
http://www.aph.gov.au/About_Parliament/Parliamentary_Departments/Parliamentary_Library/Publi
cations_Archive/CIB/cib0203/03cib12, Oct. 2012.

Rudnitzki, R.M., “Optical spray patternation of gasoline fuel injectors”, M.S. thesis, Engine
Research Centre, University of Wisconsin, Madison, 2005.

Rusly, A.M., Kook, S., Hawkes, E.R., and Zhang, R., “Effect of pilot injection on diesel knock in a
small-bore optical engine”, Proceedings of the ASME 2012 Internal Combustion Engine Division
Spring Technical Conference, Torino, Italy, May 2012.

Sahoo, P.K., and Das, L.M., “Combustion analysis of Jatropha, Karanja and Polanga based
biodiesel as fuel in a diesel Engine”, Fuel 88(6):994–99, 2009.

Saravanan, N., Nagarajan, G., Sanjay, G., Dhanasekaran, C., and Kalaiselvan, K.M., “Combustion
analysis on a DI diesel engine with hydrogen in dual fuel mode”, Fuel 87(17–18):3591–3599, 2008.

Saydut, A., Kafadar, A.B., Tonbul, Y., Kaya, C., Aydin, F., and Hamamci, C., “Comparison of the
biodiesel quality produced from refined sunflower (Helianthus annuus L) oil and waste cooking oil”
Energy, Exploration & Exploitation 28(6):499–512, 2010.

Sherwood Design and Engineering Pty Ltd., “University of New South Wales: Ethanol injection
spray chamber vessel design”, Internal document of Engine Research Laboratory, UNSW, 2009.

Shi, L., Cui, Y., Deng, K., Peng, H., and Chen, Y., “Study of low emission homogeneous charge
compression ignition (HCCI) engine using combined internal and external exhaust gas recirculation
(EGR)”, Energy 31(14):2665–2676, 2006.

Singh, S.P., and Singh, D., “Biodiesel production through the use of different sources and
characterization of oils and their esters as the substitute of diesel: A review”, Renewable and
Sustainable Energy Reviews 14:200–216, 2010.

Sjoberg, M., and Dec, J.E., “Ethanol auto ignition characteristics and HCCI performance for wide
ranges of engine speed, load and boost”, SAE Technical Paper 2010-01-0338, 2010.

Sjoberg, M., Dec, J.E., “An investigation into lowest acceptable combustion temperatures for
hydrocarbon fuels in HCCI engines”,Proceedings of the Combustion Institute 30:2719–2726, 2005.

Song, J., Zello, V., Boehman, A.L., and Waller, F.J., “Comparison of the impact of intake oxygen
enrichment and fuel oxygenation on diesel combustion and emissions” Energy & Fuels 18(5):1282–
1290, 2004.

Splitter, D., Hanson, H., Kokjohn, S., and Reitz, R., “Reactivity controlled compression ignition
(RCCI) heavy-duty engine operation at mid and high-loads with conventional and alternative fuels”,
SAE Technical Paper 2011-01-0363, 2011.

Srinivasan, K.K., Krishnan, S.R., and Qi, Y., “Cyclic combustion variations in dual fuel partially

111
premixed pilot-ignited natural gas engines”, presented at 2012 Internal Combustion Engine Division
Spring Technical Conference ICES2012, Torino, Italy, May 6–9, 2012.

Srinivasan, K.K., Krishnan, S.R., and Midkiff, K.C, “Improving low load combustion, stability, and
emissions in pilot-ignited natural gas engines”, Proceedings of Institute of Mechanical Engineering,
Part D: Journal of Automobile Engineering 220:229–239, 2006.

Suchithra, T.G., “Bio-oil production through fast pyrolysis and upgrading to “Green” transportation
fuels”, Ph.D Thesis, Auburn University, Alabama, 2012.

Surawski, N.C., Ristovski Z.D., Brown, R.J., and Situ, R., “Gaseous and particle emissions from an
ethanol fumigated compression ignition engine”, Energy Conversion and Management 54: 145–
151, 2012.

Suzuki, M., Nishida, K., and Hiroyasu, H., “Imaging of drop and vapour clouds in an evaporating
fuel spray by ultraviolet and visible lasers”, Particle & Particle Systems Characterization
11(3):241–249, 1994.

Tani, Y., Mori, Y., and Mochizuki, K., “Multiple-hole nozzle atomization for SI engines”, SAE
Technical Paper 1999-01-0564, 1999.

Tian, G., Li, H., Xu, H., Li, Y., and Mohan Raj, S., “Spray characteristics study of DMF using
phase doppler particle analyser”, SAE Technical Paper 2010-01-1505, 2010.

Tsang, K.S., Zhang, Z.H., Cheung, C.S., and Chan, T.L., “Reducing emissions of a diesel engine
using fumigation ethanol and a diesel oxidation catalyst”, Energy & Fuels 24(11):6156–6165, 2010.

Wang, X., Gao, J., Jiang, D., Huang, Z., and Chen, W., “Spray characteristics of high-pressure swirl
injector fueled with methanol and ethanol”, Energy & Fuels 19(6):2394–2401, 2005.

Weall, A., Szybist, J,P., and Edwards, K.D., “HCCI load expansion opportunities using a fully
variable HVA research engine to guide development of a production intent cam-based VVA engine:
The low load limit”, SAE Technical Paper 2012-01-1134, 2012.

Werpy, M.R., Santini, D., and Burnham, A., “Natural gas vehicles: Status, barriers, and
opportunities”, Center for Transportation Research, Argonne National Laboratory, Argonne, IL.
Report no. ANL/ESD/10-4, 2010, http://www.afdc.energy.gov/pdfs/anl_esd_10-4.pdf, Nov. 2012.

White, C.M., Steeper, R.R., and Lutz, A.E., “The hydrogen-fueled internal combustion engine: A
technical review”, International Journal of Hydrogen Energy 31(10):1292–1305, 2006.

Williams, P., and Beckwith, P., “Correlation between the liquid fuel properties density, viscosity
and surface tension and the drop sizes produced by an SI engine pintle-type port fuel injector”, SAE
Technical Paper 941864, 1994.

Wong, S.C., and Cheng, J.C., “Evaporation of non dilute and dilute mono disperse droplet clouds”,
International Journal of Heat and Mass Transfer 35(10):2403–2411, 1992.

Woo, C., Padala, S., Kook, S., Tupufia, S., and Rogers, P., "Effect of fuel molecular structure on
global in-cylinder phenomena and engine-out emissions of a common-rail diesel engine ", In
preparation, 2013.

112
Wyman, C.E., and Hinman N.D., “Ethanol: Fundamentals of production from renewable feedstocks
and use as a transportation fuel”, Applied Biochemistry and Biotechnology 24-25(1):735–753, 1990.

Wyman, C.E., “Handbook on bioethanol: Production and utilization”, first ed., Tayor & Francis,
Washington, 1996.

Yap, D., Megaritis, A., and Wyszynski, M.L., “An experimental study of bioethanol HCCI”,
Combustion Science and Technology 177(11):2039–2068, 2005.

Zhang, Z.H., Tsang, K.S., Cheung, C.S., Chana, T.L., and Yao, C.D., “Effect of fumigation
methanol and ethanol on the gaseous and particulate emissions of a direct-injection diesel engine”
Atmospheric Environment 45 2001–2008, 2011.

Zhao, F.Q., and Lai, M.C., “The spray characteristics of automotive port fuel injection-A critical
reviews”, SAE Technical Paper 950506, 1995.

Zhao, F.Q., Yoo, J.H., and Lai, M.C., “The spray characteristics of dual-stream port fuel injectors
for applications to 4-valve gasoline engines”, SAE Technical Paper 952487, 1995.

Zhong, L., Singh, I.P., Han, J., Lai, M.C., Henein, N.A., and Bryzik, W., “Effect of cycle-to-cycle
variation in the injection pressure in a common rail diesel injection system on engine performance”,
SAE Technical Paper 2003-01-0699, 2003.

http://www.ispoptics.com/designofpressurewindow.htm, accessed on 10/09/2012

113
Appendix

A1 Software controllers: LabVIEW

Fig. A1.1 gives the snap shots of the algorithm diagram (at top), and the front panel (at
bottom) of the controller developed in LabVIEW for the Mie-scattering imaging experiments.
Through the front panel, the operator can adjust the injector pulse width, camera gate width, and
delay between the injector and camera events. Also it allows monitoring the air flow rates in cross-
flow experiments and high ambient pressure experiments.

Fig. A1.1: Block diagram (at top) and front panel (at bottom) of the developed controller for Mie-scattering
imaging

114
 Fig. A1.2 gives the snap shots of the algorithm diagram (at top), and the front panel (at
bottom) of the controller developed for the injection rate measurements experiments. In addition to
the functions that were performed by the injector controller described using Fig. A1.1, it
additionally triggers the force sensor and saves the force data to the data acquisition system in the
injection rate measurement experiments.

Fig. A1.2: Block diagram (at top) and front panel (at bottom) of the developed controller for injection rate
measurements

115
A2 Injection Rate Calculation

The procedure to calculate injection rate from the measured momentum flux is described in
this section. The injection rate can be calculated provided the mass per injection and coefficient of
discharge (Cd) of nozzle is known [Naber & Siebers 1996]. The mass per injection was calculated
from the injector manufacturer specifications [Bosch 2012]. Cd value was approximated from
Bernoulli’s relation knowing the mass flow-rate of injector and pressure difference across the
injector. The injection rate can be calculated from the following equation A1.1

‫ܯ‬௙ሶ
݉ሶ௙ ൌ ሺ‫ͳܣ‬Ǥͳሻ
ܷ ‫ܥ כ‬௏

Where ݉௙ሶ = Injection rate (g/s)

‫ܯ‬ሶ௙ = Measured momentum flux (N)
U = Spray velocity (m/s)
Cv= Coefficient of velocity

The spray velocity is calculated from area of nozzle and volume flow-rate specifications
from the manufacturer. The coefficient of velocity (Cv) is defined as the ratio of Coefficient of
discharge (Cd) and Coefficient of area contraction (Ca)

ୢ
୴ ൌ ሺͳǤʹሻ
ୟ

Knowing the mass per injection, density of the fuel (ρ), Cd*A value should be calculated
from the mass conservation from the following relation

Mass per injection = ‫׬‬௧ ‫ܥ‬ௗ ‫ ݐ݀ כ ܷ כ ߩ כ ܣ כ‬ൎ ‫ܥ‬ௗ ‫ כ ܷ כ ߩ כ ܣ כ‬ο‫ݐ‬ሺ‫ͳܣ‬Ǥ͵ሻ

೔೙ೕ

Then, Cv valve can be calculated from the following equation

‫ܥ‬ௗ ‫ܯ‬௙ሶ
‫ܥ‬௩ ൌ ൌ ሺ‫ͳܣ‬ǤͶሻ
‫ܥ‬௔ ʹ ‫ܥ כ ܣ כ‬ௗ ‫ כ‬οܲ

Where ‫ܯ‬ሶ௙ = Average momentum flux

A = Area of injector nozzle
ΔP= Pressure difference across the injector

116
Average momentum flux ‫ܯ‬ሶ௙ can be obtained by approximating the momentum flux profile as a top
hat profile and dividing the area under momentum flux curve by actual injection duration Δt, as
shown in Fig. A2.1.

Δt

Figure A2.1: Force profile of the spray measurement. Also shown, area integrated profile, PFI
driver signal, averaged momentum flux ‫ܯ‬ሶ௙  and actual injection duration (Δt)

117
A3 Dynamometer Torque-Speed Map

Figure A3.1 shows the maximum torque of the EC dynamometer used in this study for
various rotational speeds. All experiments in the present study were conducted at a fixed speed of
2000 rpm at which the dynamometer can handle up to 95 Nm torque and 24 kW power. The torque
and brake power of this study ranges around 22 Nm and 5 kW, suggesting roughly 22% of the
dynamometer capacity was used.

Figure A3.1: Operation range of Froude Hoffmann AG-30HS eddy current dynamometer

118
A4 Fuel Flow Rate Measurement

Prior to performing the engine experiments, the fuel flow rate was measured using
an optical engine. The Engine Research Laboratory, UNSW has an optical engine which features
the same injection system as that of the engine used for the present thesis. This engine allows
collecting the fuel in a separate collection-bowl and the amount of fuel collected in number of
injections was separately weighed and mass per injection was measured. More details about the
optical engine facilities can be found in Ref. [Rusly et al. 2012]. To validate the injected mass
measurements, additional experiments were carried out and measured the injection rate using
injection rate meter [Woo et al., 2013]. The results of the same are shown in Fig. A4.1. At baseline
condition (at 650 μs, 1300 bar) the difference in these measurements was less than 2%, with a
maximum error of 6% at smaller injection durations. With these errors, the calculated indicated
efficiencies can differ by a maximum error of 6% error.

For the calculation of air-fuel ratio, the air-flow rate was not measured but estimated based
on the assumption of 100% volumetric efficiency. Since the diesel engine was running with wide
open throttle and the injected mass into the intake port is negligible, it is fair to assume that full
displacement volume of air was breathed throughout the experiments.

Figure A4.1: Diesel mass per injection measured using optical engine as well as injection rate
meter at different injection durations (130 MPa injection pressure).

119
A5 Apparent Heat Release Rate Filtering

Cut-off Cut-off
Frequency 20 KHz Frequency 10 KHz

Cut-off Cut-off
Frequency 5 KHz Frequency 2 KHz

Figure A5.1: Demonstration of the effect of cut-off frequency on the apparent heat release rate
traces. The cut-off frequency is annotated on the top left of each plot

Apparent heat release rates are calculated from in-cylinder pressure traces. Large
fluctuations are observed in calculated apparent heat release rates even if in-cylinder pressure traces
are ensemble averaged. Therefore, aHRR traces requires a filtering of raw data. A second order
Butterworth filter is used for filtering the high frequency signals from the aHRR plot [Olsson et al.
2001, Song et al. 2004]. The one-way-forward filtering results a shift in the signal in forward
direction, leading to an offset in combustion phasing calculations. Therefore, a forward-and-back
ward filtering is applied to remove the shift from the filtered signal. Various cut-off frequencies are
tried until the best method is selected. How the cut-off frequency affects the aHRR signal is shown
for a wide range of cut-off frequencies (2 ~ 20 kHz) in Fig. A5.1. As cut-off frequency decreases, it
filters off the high frequency signal resulting in smoothed curves. However, the peak heat release
rate also decreases with decreasing cut-off frequency. It is important to select a cut-off frequency,
which filters off the high frequency noise, yet captures the overall traces. The aHRR traces are
optimized when the cut-off frequency of 5 kHz is selected in this study.

120
A6 Combustion Parameter Calculations

A6.1 Calculation of Overall Equivalence Ratio:

The complete combustion or stoichiometric combustion of ethanol and diesel can be

represented by the following equation A6.1, with three exhaust species CO2, H2O and N2.

଴Ǥ଻ଽ
ߙ‫ܥ‬௫ ‫ܪ‬௬ ൅ ߚ‫ܥ‬ଶ ‫ ܱ ଺ܪ‬൅ ߛ ቀܱଶ ൅ ଴Ǥଶଵ ‫ܰ כ‬ଶ ቁ ՜ ܽ‫ܱܥ‬ଶ ൅ ݀‫ܪ‬ଶ ܱ ൅ ݁ܰଶ (A6.1)

Following the atomic balances

Carbon Balance ߙ‫ ݔ‬൅ ʹߚ ൌ ܽ

Hydrogen Balance ߙ‫ ݕ‬൅ ͸ߚ ൌ ʹ݀
ఈ
݀ ൌ ‫ ݕ‬൅ ͵ߚ
ଶ

Oxygen Balance ߚ ൅ ʹߛ ൌ ʹܽ ൅ ݀
Nitrogen Balance ͵Ǥ͹͸ߛ ൌ ݁
Solving the above equations gives results in
ௗ ఉ
ߛൌܽ൅ଶെଶ

Mass of Air ൌ ߛ ‫ כ‬ሺ͵ʹ ൅ ͵Ǥ͹͸ ‫ʹ כ‬ͺሻ

ఈ ଷ ఉ
ൌ ቀߙ‫ ݔ‬൅ ʹߚ ൅ ସ ‫ ݕ‬൅ ଶ ߚ െ ଶ ቁ ‫ כ‬ሺ͵ʹ ൅ ͵Ǥ͹͸ ‫ʹ כ‬ͺሻ
ఈ
ൌ ቀߙ‫ ݔ‬൅ ͵ߚ ൅ ସ ‫ݕ‬ቁ ‫ כ‬ሺͳ͵͹Ǥʹͺሻ

Fuel Mass ൌ ߙ ‫ כ‬ሺ‫ ʹͳ כ ݔ‬൅ ‫ͳ כ ݕ‬ሻ ൅ ߚ ‫ כ‬Ͷ͸

ൌ ߙሺͳʹ‫ ݔ‬൅ ‫ݕ‬ሻ ൅ Ͷ͸ߚ
ߙ
‫ܣ‬ ቀߙ‫ ݔ‬൅ ͵ߚ ൅ Ͷ ‫ݕ‬ቁ ‫ כ‬ሺͳ͵͹Ǥʹͺሻ
൬ ൰ ൌ ሺ‫ܣ‬͸Ǥʹሻ
‫ ܨ‬௦௧ ߙሺͳʹ‫ ݔ‬൅ ‫ݕ‬ሻ ൅ Ͷ͸ߚ

In the above equations, the values of x and y play a key role in finding the proper
stoichiometric air fuel ratio for any given dual-fuel combustion. The chemical components in the
diesel range from range from C10H20 to C15H28 with an average chemical formula of C12H23 (Botero
et al. 2011, Saydut et al. 2010, Singh & Singh 2010, Suchitra 2012). Hence, C12H23 was taken as the
formula of diesel for the calculation of overall equivalence ratio. An example calculation for the
overall equivalence ratio was shown below for E30 case:

121
Diesel Molecular weight = 12*12+23*1=167
Ethanol Molecular weight = 46
Diesel injection quantity for E30: 18.45 mg = 18.45 e-3/167 =0.000110491 mol
Ethanol injection quantity for E30: 12.2 6 mg = 12.26e-3/46 = 0.000266539 mol
α= 0.000110491
β=0.000266539
x=12
y=23
Substituting these values in AFRst equation A6.2 results in
Ƚ
 ቀȽš ൅ ͵Ⱦ ൅ Ͷ ›ቁ ‫ כ‬ሺͳ͵͹Ǥʹͺሻ
൬ ൰ ൌ ൌ ͳʹǤ͵Ͷ
ୱ୲ Ƚሺͳʹš ൅ ›ሻ ൅ Ͷ͸Ⱦ

୅ ସଽ଼ୣି଺‫כ‬ଵଵ଼଴
Actual AFR ቀ ቁ ൌ ൌ ͳͻǤͳ͵
୊ ୟୡ୲୳ୟ୪ ଵ଼ǤସହୣିଷାଵଶǤଶ଺ୣିଷ
ఽ
ቀ ቁ ଵଶǤଶଷ
ూ ౩౪
Equivalence ratio ɔൌ ఽ ൌ ଵଽǤଵଷ ൌ ͲǤ͸Ͷͷ
ቀ ቁ
ూ ౗ౙ౪౫౗ౢ

Fig. A6.1 shows the calculated equivalence ratios from the measured fuel masses for
various ethanol energy fractions.

Figure A6.1: Calculated equivalence ratio for various ethanol energy fractions. The total fuel
energy per cycle was held constant at 1098 J

122
A6.2 Calculation of Combustion Efficiency

Combustion efficiency is defined as

οு
K௖௢௠௕ ൌ οுೃ೎ (A6.3)
ೃ

Where

K௖௢௠௕ =Combustion efficiency

ο‫ܪ‬ோ௖ = Heat of reaction from stoichiometric combustion
ο‫ܪ‬ோ = Heat of reaction from actual (or incomplete) combustion

Heat of Reaction (οࡴࢉࡾ ) from Stoichiometric combustion:

The complete combustion or stoichiometric combustion of ethanol and diesel fuels with
oxygen can be represented by the following equation A6.4, with only two exhaust species CO2, and
H2O.

ߙ‫ܥ‬௫ ‫ܪ‬௬ ൅ ߚ‫ܥ‬ଶ ‫ ܱ ଺ܪ‬൅ ‫ܱܣ‬ଶ ՜ ܽ‫ܱܥ‬ଶ ൅ ݀‫ܪ‬ଶ ܱ (A6.4)

The atomic balances

Carbon Balance ܽ ൌ ߙ‫ ݔ‬൅ ʹߚ

ఈ௬ା଺ఉ ௬
Hydrogen Balance ݀ൌ ଶ
ൌ ଶ ‫ ߙ כ‬൅ ͵ߚ

Oxygen Balance ߚ ൅ ʹ‫ ܣ‬ൌ ʹܽ ൅ ݀

ͳ ‫ݕ‬
‫ ܣ‬ൌ ሺʹܽ ൅ ݀ െ ߚሻ ൌ ߙ ቀ‫ ݔ‬൅ ቁ ൅ ͵ߚ
ʹ Ͷ

Heat of reaction for the complete combustion is represented by equation A6.5

ο‫ܪ‬ோ௖ ൌ σ ο‫ܪ‬ோǡ௣௥௢ௗ െ σ ο‫ܪ‬ோǡ௥௘௔௖௧ (A6.5)
Where
σ ο‫ܪ‬ோǡ௣௥௢ௗ = Sum of enthalpy of formations of the products
σ ο‫ܪ‬ோǡ௥௘௖௔௧ = Sum of enthalpy of formations of the reactants

௢ ௢ ௢ ௢
ο‫ܪ‬ோ௖ ൌ ൣܽ ‫ כ‬ο݄௙ǡ஼ைమ
൅ ݀ ‫ כ‬ο݄௙ǡுమை
൧ െ ቂߙ ‫ כ‬ο݄௙ǡ஼ೣ ு೤
൅ ߚ ‫ כ‬ο݄௙ǡ஼మ ஼ுల ை
ቃ

123
 ௢ ‫ݕ‬ ௢ ௢ ௢
ο‫ܪ‬ோ௖ ൌ ቂሺߙ‫ ݔ‬൅ ʹߚሻ ‫ כ‬ο݄௙ǡ஼ை ൅ ቀ ‫ ߙ כ‬൅ ͵ߚቁ ‫ כ‬ο݄௙ǡுమை
ቃ െ ቂߙ ‫ כ‬ο݄௙ǡ஼ೣ ு೤
൅ ߚ ‫ כ‬ο݄௙ǡ஼మ ஼ுల ை
ቃ
మ ʹ
Where

௢
ο݄௙ǡ௜ = Enthalpy of formation of ith element.

In the above equation Ș, ș can be calculated from the fuel flow rates. Enthalpy of formation of all
elements are known quantities and x, y depends on the fuel type.ο‫ܪ‬ோ௖ can be calculated for any
given Diesel/Ethanol percentage.

Heat of Reaction (οࡴࡾ ) from actual (or incomplete) combustion:

The actual or incomplete combustion of ethanol and diesel fuels with oxygen can be
represented by the following equation A6.6, with the exhaust species CO2, H2O, CO and HC
emissions. The emission analyser (Horiba MEXA 584L) gives hexane (C6H14) equivalent of
hydrocarbons emissions. Hence, unburnt hydrocarbon elements can be treated as hexane equivalent.

For incomplete combustion,

ߙ‫ܥ‬௫ ‫ܪ‬௬ ൅ ߚ‫ܥ‬ଶ ‫ ܱ ଺ܪ‬൅ ‫ܱܣ‬ଶ ՜ ܽ‫ܱܥ‬ଶ ൅ ܾ‫ ܱܥ‬൅ ܿ‫ܪ ଺ܥ‬ଵସ ൅ ݀‫ܪ‬ଶ ܱ (A6.6)

Carbon Balance ܽ ൌ ߙ‫ ݔ‬൅ ʹߚ െ ܾ െ ͸ܿ

ఈ௬ା଺ఉିଵସ௖ ௬
Hydrogen Balance ݀ൌ ൌ ଶ ‫ ߙ כ‬൅ ͵ߚ െ ͹ܿ
ଶ

Oxygen Balance ߚ ൅ ʹ‫ ܣ‬ൌ ʹܽ ൅ ܾ ൅ ݀

ͳ ‫ݕ‬ ܾ
‫ ܣ‬ൌ ሺʹܽ ൅ ܾ ൅ ݀ െ ߚሻ ൌ ߙ ቀ‫ ݔ‬൅ ቁ ൅ ͵ߚ െ െ ͻǤͷܿ
ʹ Ͷ ʹ

௢ ௢
ο‫ܪ‬ோǡ௥௘௔௖௧ ൌ ቂߙ ‫ כ‬ο݄௙ǡ஼ೣ ு೤
൅ ߚ ‫ כ‬ο݄௙ǡ஼మ ஼ுల ை
ቃ(A6.7)

௢ ௢ ௢ ௢
ο‫ܪ‬ோǡ௣௥௢ௗ ൌ ൣܽ ‫ כ‬ο݄௙ǡ஼ைమ
൅ ܾ ‫ כ‬ο݄௙ǡ஼ை ൅ ܿ ‫ כ‬ο݄௙ǡ஼ల ுభర
൅ ݀ ‫ כ‬ο݄௙ǡுమை
൧ (A6.8)
௢ ௢ ௢ ௬
ο‫ܪ‬ோǡ௣௥௢ௗ ൌ ቂሺߙ‫ ݔ‬൅ ʹߚ െ ܾ െ ͸ܿሻ ‫ כ‬ο݄௙ǡ஼ைమ
൅ ܾ ‫ כ‬ο݄௙ǡ஼ை ൅ ܿ ‫ כ‬ο݄௙ǡ஼ల ுభర
൅ ቀଶ ‫ ߙ כ‬൅ ͵ߚ െ ͹ܿቁ ‫כ‬
௢
ο݄௙ǡுమை
ቃ (A6.9)

ο‫ܪ‬ோ ൌ σ ο‫ܪ‬ோǡ௣௥௢ௗ െ σ ο‫ܪ‬ோǡ௥௘௔௖௧ (A6.10)

124
In the above equation α, β can be calculated from the fuel flow rates. Enthalpy of formation of all
elements are known quantities and x, y depends on the fuel type. The number of moles of b and c
can be obtained from the measured emission values.

ܾ ൌ ܺ஼ை ‫ܰ כ‬௧௢௧ǡ௣௥௢ௗ ܿ ൌ ܺ஼ల ுభర ‫ܰ כ‬௧௢௧ǡ௣௥௢ௗ 

Where ܺ஼ை = volume fraction of carbon monoxide

ܺ஼ల ுభర ൌ Volume fraction of unburnt hydrocarbons

ܰ௧௢௧ǡ௣௥௢ௗ =total number of moles in exhaust gases (including that of nitrogen which we did not
include in reaction equation)
௬
Number of reactant mole‫ ݏ‬σ ݊௥௘௖௔௧ : ߙ ൅ ߚ ൅ ‫ ܣ‬ൌ ߙ ൅ ߚ ൅ ߙ ቀ‫ ݔ‬൅ ସ ቁ ൅ ͵ߚ

‫ݕ‬ ܾ
ൌ ߙ ቀͳ ൅ ‫ ݔ‬൅ ቁ ൅ Ͷߚ െ െ ͻǤͷܿ
Ͷ ʹ
Number of product molesσ ݊௣௥௢ௗ : ܽ൅ܾ൅ܿ൅݀
‫ݕ‬
ൌ ሺߙ‫ ݔ‬൅ ʹߚ െ ܾ െ ͸ܿሻ ൅ ܾ ൅ ܿ ൅ ቀ ‫ ߙ כ‬൅ ͵ߚ െ ͹ܿቁ
ʹ
‫ݕ‬
ൌ ቀ‫ ݔ‬൅ ቁ ߙ ൅ ͷߚ െ ͳʹܿ
ʹ
ȟ݉‫ ݏ݈݁݋‬ൌ ෍ ݊௥௘௖௔௧ െ ෍ ݊௣௥௢ௗ

‫ݕ‬ ܾ
ο݉‫ ݏ݈݁݋‬ൌ ቀͳ െ ቁ ߙ െ ߚ െ ൅ ʹǤͷܿ
Ͷ ʹ
Total number of moles per cycle including airൌ ሺߙ ൅ ߚ ൅ ܰ௔௜௥ ሻ
ܰ௧௢௧ǡ௣௥௢ௗ ൌ ሺߙ ൅ ߚ ൅ ܰ௔௜௥ ሻ െ ο݉‫ݏ݈݁݋‬

‫ݕ‬ ܾ
ൌ ሺߙ ൅ ߚ ൅ ܰ௔௜௥ ሻ െ ൭ቀͳ െ ቁ ߙ െ ߚ െ ൅ ʹǤͷܿ൱
Ͷ ʹ
‫ݕ‬ ܾ
ൌ ܰ௔௜௥ ൅ ߙ ൅ ʹߚ ൅ െ ʹǤͷܿ
Ͷ ʹ
ܾ ܿ
ܺ஼ை ൌ ǡܺ஼ల ுభర ൌ ǡ
ܰ௧௢௧ǡ௣௥௢ௗ ܰ௧௢௧ǡ௣௥௢ௗ
ܾ ൌ ܺ஼ை ‫ܰ כ‬௧௢௧ǡ௣௥௢ௗ ܿ ൌ ܺ஼ల ுభర ‫ܰ כ‬௧௢௧ǡ௣௥௢ௗ 
‫ݕ‬ ܾ
ܾ ൌ ܺ஼ை ‫ כ‬൬ܰ௔௜௥ ൅ ߙ ൅ ʹߚ ൅ െ ʹǤͷܿ൰
Ͷ ʹ

125
 ‫ݕ‬ ܾ
ܿ ൌ ܺ஼ల ுభర ‫ כ‬൬ܰ௔௜௥ ൅ ߙ ൅ ʹߚ ൅ െ ʹǤͷܿ൰
Ͷ ʹ
Rearranging the equations yields
‫ݕ‬ ܺ஼ை
ቀܰ௔௜௥ ൅ ߙ ൅ ʹߚቁ ܺ஼ை ൌ ൬ͳ െ ൰ ‫ ܾ כ‬൅ ʹǤͷܺ஼ை ‫ܿ כ‬
Ͷ ʹ
‫ݕ‬ ܺ஼ ு
ቀܰ௔௜௥ ൅ ߙ ൅ ʹߚቁ ܺ஼ల ுభర ൌ ൬െ ల భర ൰ ‫ ܾ כ‬൅ ൫ͳ ൅ ʹǤͷܺ஼ల ுభర ൯ ‫ܿ כ‬
Ͷ ʹ
These are two linear equations in the form of
R1 = C1*b + C2*c (A6.11)
R2 = C3*b+C4*c (A6.12)
Where
‫ݕ‬
ܴͳ ൌ ቀܰ௔௜௥ ൅ ߙ ൅ ʹߚቁ ܺ஼ை
Ͷ
ܺ஼ை
‫ ͳܥ‬ൌ ൬ͳ െ ൰
ʹ
‫ ʹܥ‬ൌ ʹǤͷܺ஼ை
‫ݕ‬
ܴʹ ൌ ቀܰ௔௜௥ ൅ ߙ ൅ ʹߚቁ ܺ஼ల ுభర
Ͷ
ܺ஼ ு
‫ ͵ܥ‬ൌ ൬െ ల భర ൰
ʹ
‫ܥ‬Ͷ ൌ ൫ͳ ൅ ʹǤͷܺ஼ల ுభర ൯

R1, C1, C2, R2, C3, C4 are constants and depends on α, β, XCO,  ஼ల ுభర . The two equations A6.11,
and A6.12 can be solved to get b and c. Later b and c can be substituted in equations A6.7 to A6.10
to get ο‫ܪ‬ோ and combustion efficiency can be calculated using equation A6.3

Example Case:

E30 (70% energy from diesel+30% energy from ethanol) (-3aTDC diesel injection, PFI position A)
Diesel mass: 18.45 mg
Ethanol mass: 12.2 mg
CO: 0.23 %
HC: 60 ppm
Enthalpy of formations for various elements
Element name οࢎ࢕ࢌ Enthalpy of
formation (kJ/mol)
†
Diesel (C12H23) -319

126
Ethanol (C2H6O)* -276.2
*
Oxygen (O2) 0
Carbon dioxide (CO2) * -393.5
Carbon monoxide (CO) * -110.5
*
Hexane (C6H14) -167.1
*
Water (H2O) -241.8
(†http://www.nfcrc.uci.edu/2/Activities/ResearchSummaries/DynamicModeling/Fuel_Processors//,
*
http://chemeo.com/)

x=12
y=23
݉ௗ௜௘௦௘௟ ͳͺǤͶͷ݁ െ ͵
ߙൌ ൌ ൌ ͲǤͲͲͲͳͳ݉‫݈݋‬
‫ܯ‬ௗ௜௘௦௘௟ ͳ͸͹
݉௘௧௛௔௡௢௟ ͳʹǤʹ݁ െ ͵
ߚൌ ൌ ൌ ͲǤͲͲͲʹ͸݉‫݈݋‬
‫ܯ‬௘௧௛௔௡௢௟ Ͷ͸
ͳͲǤʹ͵
ܺ஼ை ൌ ൌ ͲǤͲͲʹ͵
ͳͲͲ
͸Ͳ
ܺ஼ల ுభర ൌ ൌ ͲǤͲͲͲͲ͸
ͳ݁͸
Ͷͻͺ
݉௔௜௥ ͳͲͲͲ ‫ͳ כ‬Ǥͳͺ݁͵
ܰ௔௜௥ ൌ ൌ ൌ ͲǤͲʹͲͶ݉‫݈݋‬
‫ܯ‬௔௜௥ ʹͺǤͺͶ

‫ݕ‬ ʹ͵
ܴͳ ൌ ቀܰ௔௜௥ ൅ ߙ ൅ ʹߚቁ ܺ஼ை ൌ ൬ͲǤͲʹͲͶ ൅ ‫Ͳ כ‬ǤͲͲͲͲͺ͹ ൅ ʹ ‫Ͳ כ‬ǤͲͲͲʹ͸൰ ‫Ͳ כ‬ǤͲͲʹ͵
Ͷ ʹ
ൌ ͶǤͻͷ͵݁ െ ͷ
ܺ஼ை ͲǤͲͲʹ͵
ܿͳ ൌ ൬ͳ െ ൰ ൌ ൬ͳ െ ൰ ൌ ͲǤͻͻͺͻ
ʹ ʹ
‫ ʹܥ‬ൌ ʹǤͷܺ஼ை ‫ ܿ כ‬ൌ ʹǤͷ ‫Ͳ כ‬ǤͲͲʹ͵ ൌ ͲǤͲͲͷͺ
‫ݕ‬ ʹ͵
ܴʹ ൌ ቀܰ௔௜௥ ൅ ߙ ൅ ʹߚቁ ܺ஼ల ுభర ൌ ൬ͲǤͲʹͲͶ ൅ ‫Ͳ כ‬ǤͲͲͲͲͺ͹ ൅ ʹ ‫Ͳ כ‬ǤͲͲͲʹ͸൰ ൌ ͳǤʹͻʹ݁ െ ͸
Ͷ Ͷ
ܺ஼ ு ͲǤͲͲͲͲ͸
‫ ͵ܥ‬ൌ ൬െ ల భర ൰ ൌ ൬െ ൰ ൌ െ͵݁ െ ͷ
ʹ ʹ
‫ܥ‬Ͷ ൌ ൫ͳ ൅ ʹǤͷܺ஼ల ுభర ൯ ൌ ሺͳ ൅ ʹǤͷ ‫Ͳ כ‬ǤͲͲͲͲ͸͸ሻ ൌ ͳǤͲͲͲʹ
Solving for b and c
b=4.958e-5
c=1.293e-6

127
Substituting the known values of x,y, α , β ,b,c,ο݄௙௢ values in equation A5.5 gives
ο‫ܪ‬ோ ൌ σ ο‫ܪ‬ோǡ௣௥௢ௗ െ σ ο‫ܪ‬ோǡ௥௘௔௖௧ ൌ െ ͳǤͳͲͶ݇‫ܬ‬
ο‫ܪ‬ோ௖ ൌ െ ‫ͳ כ‬Ǥͳʹ͵݇‫ܬ‬

ο‫ܪ‬ோ െͳǤͳͲ͵
K௖௢௠௕ ൌ ൌ ൌ Ǥͻͺ͵ ൌ ͻͺǤ͵Ψ
ο‫ܪ‬ோ௖ െͳǤͳʹʹ

128