Extrusion Cooking
Craft or Science ?
DJ. vanZuiliche m
CENTBALE LANDBOUWCATALOGUS
0000 0664 4526 4^57 £ Promotoren: dr. ir.K . van 'tRiet , gewoon hoogleraar ind e levensmiddelenproceskunde aan de Landbouwuni- versiteitWageningen .
dr. ir. L.P.B.M. Janssen, gewoon hoogleraar in de technische scheikunde aan de Rijksuniversi- teitGroningen . DJ. van Zuilichem
Extrasion Cooking Craft or Science?
Proefschrift terverkrijgin gva nd egraa dva n doctor ind e landbouw-e nmilieuwetenschappen , opgeza gva nd erecto rmagnificus , dr.H.C .va nde r Plas, inhet openbaart everdedige n opwoensda g 29januar i 1992 desnamiddag st evie ruu r ind eaul a vand e Landbouwuniversiteit teWageningen . BiBLion-j, . XSDSDBOUWUNIVEi^^ji SEAGENINGEM
Thisthesi sha sbee n accomplishedat : Agricultural University Wageningen, Department ofFoo dScience , Foodan d BioprocessEngineerin g Group office: Biotechnion,Bomenwe g2 , P.O. box 8129,670 0E WWageningen , TheNetherlands .
Manuscript typing: J. Krauss/S.Nijweid e Photography: J.B. Freeke Illustrations: w. Stolp/M.Schimme l aan Ria Dirk Janneke Voorwoord
Ditproefschrif t isgegroepeer dron dee nzesta lwetenschappelijk e artikelen, waarvoor het onderzoek is uitgevoerd bij de Sectie Proceskunde van de Vakgroep Levensmiddelentechnologie te Wageningen. Het extrusie- onderzoek dateert uit de zeventiger jaren. Sinds 1973 is regelmatig gepubliceerd en gerapporteerd op symposia. Daarom is hoofdstuk 1 meer dan een publicatie, nl. een uitgebreide algemene inleiding, die de stand vanzake nbeschrijf t ophe tgebie dva nd e levensmiddelen-extrusietechniek. Tevensblijk t hieruit defilosofie achte rhe tonderzoe k sindsd eaanvan gi n de zeventiger jaren, toen is begonnen met de bestudering van een eenvoudigeenkelschroefextrude rva nNederlands emakelij .D ecomplexitei t vand eextrusietechnologi e zorgdevoo ree nlangzam emaa rgestadig egroe i van de Wageningse kennis op dit gebied, waarbij een omvangrijke groep collega'se nmedewerker snau wbetrokke ni se nzonde rwi edi t proefschrift nietto tstan dzo uzij n gekomen. Graagwi li khie rbedanke nWille mStolp , TheoJager , Ericva nde r Laan, end evel estudente n diebi j experimenten, discussies en presentaties onmisbaar waren. Voor hun wetenschappelijke kritieken en aanmoedigingen ben ik veel dank verschuldigd aan Henk Leniger, SolkeBrui ne n Klaasva n 't Riet, die mij achtereenvolgens ind e gelegenheid stelden dit onderzoek te verrichten. Buiten de LU, maar heel nabij in het vak als vriend en promotor is er Leon Janssen, die mij aanmoedigde om een deel van de Wageningse kennis te bundelen in dit proefschrift. Bij de vele contacten met hem bij de TU Delft en bij alle samenwerkingen metd ecollega' sva nd eRijksuniversitei t Groningen ise r altijd die prettige sfeer van collegialiteit zijn, die zo zeldzaam is. Van de extruderfabrikanten met wie is samengewerkt wil ik vooral APV- Baker noemen, die zeer genereus onze groep van medewerkers en studenten heeft ondersteund met apparatuur. Tenslottewi li kgraa gd emedewerker sva nd emechanisch ewerkplaats , de electrowerkplaats, en de servicegroep resp. onder leiding van Johan van de Goor, Guusva n Munster, Rink Boumae n Leo Herben bedanken voor de vlotte en deskundige hulp bij de vele modificaties en onderhouds- werkzaamheden aan de vele types extruders, waarop de metingen ten behoeve van dit proefschrift zijn verricht. Preface
Thisthesi s is written around sixpublication s for which research hasbee n undertaken at the Food and Bioprocess Engineering group of the Department of Food Science in Wageningen. The extrusion research in Wageningenoriginate d inth eseventies . Since 1973,result swer eregularl y published and presented at symposia. Because of this, chapter one has more to offer than just a publication; it is more of an extended general introduction which describes the state of the art of food extrusion. It also elucidates the idea behind the extrusion research program when started in the early seventies with a simple single screw extruder of Dutch origin. The know how in Wageningen progressed slowly but steady due to the complexity of extrusion technology. A considerable group of colleagues andco-worker swer eclosel yinvolve di nth eaccomplishmen to fthi sthesis . With pleasure, I would like to thank Willem Stolp, TheoJager , Eric van der Laan and the numerous students who were indispensable for the experiments, discussionsan dpresentations . Forth escientifi c criticisman d support I would like to express my gratitude to Professor H.A. Leniger, Professor S. Bruin and Professor K. van 't Riet whorespectivel y offered meth eopportunit yt oundertak ethi sresearch . Fromoutsid eth euniversity , I thankLeo nJansse nfo r beinga friend , colleague aswel la spromoto ran d for encouraging met o write this thesis. The numerous contacts with him and hisstaf f respectively atth eDelf t Technical University andno w atth e Rijksuniversiteit Groningen always lead to a cooperation on extrusion topicsi na pleasan tan drarel y found atmosphere. Awor do fspecia lthank s is reserved for APV-Baker, the extruder manufacturer who supported generouslyou rgrou po fco-worker san dstudent swit hextrude requipment . Also should be mentioned the great support of our mechanical, electrical andservic eworksho prespectivel y supervised byMr . J. vand eGoor , Mr. A. van Munster, Mr. R. Bouma and Mr. L. Herben. Their expert assistance wasver y much needed performing the many modifications and maintenance activities concerning the several extruder types used for the measurements resulting in this thesis. CONTENTS
CHAPTER 1 GENERAL INTRODUCTION TO EXTRUSION COOKING OF FOODAN DFEE D 1
1.1 Definition and scope 1 1.2 Historical development 3 1.3 Differences between food- and plasticating- polymers 6 1.4 Foodmeltin g 9 1.5 Rheological considerations 14 1.6 Residencetim edistributio n 18 1.7 Quality parameters 20 1.8 Influence of marketing interests on extrusion cooking 22
References 24
Notation 27
CHAPTER 2 RESIDENCE TIME DISTRIBUTIONS IN SINGLE-SCREW EXTRUSION COOKING 28
2.1 Introduction 29 2.2 Theory 30 2.3 Experimental 34 2.3.1 Measuring theresidenc e timedistributio n 35 2.4 Results 35 2.4.1 General 35 2.4.2 Soya 37 2.4.3 Maize 38 2.5 Conclusions 42
References 44
Notation 46 CHAPTER 3 MODELING OF THE AXIAL MIXING IN TWIN-SCREW EXTRUSION COOKING 47
3.1 Introduction 48 3.2 Theory 49 3.3 Materials and methods of measurement 51 3.4 Simulation model 53 3.5 Optimization method 56 3.6 Results 59 3.7 Conclusions 65
References 66
Notation 69
CHAPTER 4 CONFECTIONERY AND EXTRUSION TECHNOLOGY 71
4.1 Introduction 72 4.2 Problem description 73 4.3 Status quo of realised extrusion cooking pro cesses 76 4.4 Extrusion of starch 77 4.5 Extrusion of dry sucrose-crystals 78 4.6 Extrusion of sucrose-starch mixtures 83 4.6.1 Extrusion of sucrose-syrup mixtures 83 4.7 Dissolution time of crystals 88 4.8 Water content 92 4.9 Experimental results and conclusions 93
References 95
Notation 97
Appendix 4.1 98
CHAPTER 5 MODELLING OF HEAT TRANSFER IN A CO-ROTATING TWIN SCREW FOOD EXRUDER 101
5.1 Introduction 102 5.2 Thebasi cmode l fora singl e screwextrude r 103 5.3 Variousmodel s 106 5.3.1 Introduction 106 5.3.1 Dimensionlessnumber s 107 5.3.3 Heattransfe r 108 5.3.4 Theory ofTadmo ran d Klein 110 5.4 Boundary layertheory inth epum p zone 112 5.5 Extrudermode lb y Yacu 114 5.5.1 Solid conveying section 115 5.5.2 Meltpumpin g section 116 5.6 LUMode l 119 5.7 Theviscosit y 122 5.8 Material 123 5.9 Results 125 5.10 Discussion 126
References 130
Notation 133
Appendix5. 1 136
CHAPTER6 THE INFLUENCE OF A BARREL-VALVE ON THE DEGREE OFFIL LI NA CO-ROTATING TWINSCRE W EXTRUDER 138 6.1 Introduction 139 6.2 Theory 140 6.3 Experimental 141 6.4 Results 143 6.5 Discussion 145 6.6 Conclusions 149
References 151
Notation 153
CHAPTER 7 MODELLING OF THE CONVERSION OF CRACKED CORN BY TWIN-SCREW EXTRUSION-COOKING 154 7.1 Introduction 155 7.2 Theoretical background 155 7.3 Experimental 160 7.3.1 Setup 160 7.3.2 Methods and materials 160 7.3.3 Calculation 162 7.4 Results 163 7.5 Discussion 164 7.5.1 Extrusion (disclosure & liquefaction) 165 7.5.2 Saccharification 165 7.6 Proposed process 169 7.7 Conclusions 170
References 171
Notation 173
Appendix7. 1 174
CHAPTER 8 GENERAL DISCUSSION ON EXT 175 8.1 Gain in knowledge 175 8.2 Market 177 8.2.1 Petfood 179 8.2.2 Coextrusion 181 8.2.3 feed industry 185 8.2.4 Other food branches 185 8.2.5 Agrochemicals 186 8.3 Equipment manufacturers 187 8.4 Technology, final remarks 187
References 189 Extrusion Cooking,craf t or Science?
CHAPTER 1
GENERAL INTRODUCTION TOEXTRUSIO N COOKINGO FFOO DAN D FEED
1.1 Definition and scope It isno t longer ago than the seventieswhe n a food extruder wasdescribe d asa piec eo fequipment ,consistin g outo fa barre l and one or two screws.Th e function of the extruder was believed to receive an agricultural product, to transport it from a feed section to a screw metering section over a compression section. Withth ehel p of shearenergy ,exerte db yth erotatin g screw,an d additional heating by the barrel,th e foodmateria l isheate d to its melting point or plasticating-point. In this changed rheological status the food is conveyed under high pressure through adi e or a series of dies and the product expands to its finalshape . It is obvious that thismor e or less phenomenological definition is simple and not complete, since some important properties of the extruder as a food and feed unit operation are not yet mentioned. In the first place one should recognize the cooking extruder to belong to the family ofHTS T (HighTemperatur e Short Time)-equipment,wit h acapabilit y toperfor m cooking tasksunde r high pressure (van Zuilichem et al. 1975;va n Zuilichem et al. 1976). Thi s aspect may be explained forvulnerabl e food and feed as an advantageous process since small time span exposures to high temperatures will restrict unwanted denaturation effects on e.g. proteins, aminoacids, vitamins, starches and enzymes. However, one should realize that now physical technological aspects like heat transfer, mass transfer, momentum transfer, residence time and residence time distribution have a strong impacto nth e foodan d feedpropertie sdurin g extrusion cooking andca ndrasticall yinfluenc eth efina lproduc tquality .Secondl y we should realize that acookin g extruder isa proces s reactor (vanZuiliche m et al. 1979;Jansse n et al. 1980), inwhic h the designer has created the prerequisites in the presence of a certainscre wlay-out ,th eus eo fmixin gelements ,th eclearance s in the gaps, the installed motor power and barrel heating and cookingcapacity ,t ocontro l afoo dan d feed reaction.Thi sca n bea reaso n"i nitself" ,whe nonl ymas si stransferre d inwante d and unwanted reaction products due to heating, e.g. the denaturationo fprotein sunde rpresenc eo fwate ran dth eruptur e of starches,bot h affected by the combined effects of heat and shear. The reaction can also be provoked by the presence of a distinct biochemical or chemical component like an enzyme ora pH controlling agent.Whe n we consider the cooking extruder to bemor etha nwa smentione d originally,a thoroug h investigation of the different physical technological aspects is more than desirable.Thi s isth emotiv e forth e setu p of the chapters 1 to7 an dth eobjectiv e forthi sstudy . Incookin gextrusion ,i t isobviou stha tth estud yconcern sth eresidenc etim ebehaviour , which isdon e inth echapter s2 an d 3fo rsingl ean dtwi n screw extruder equipment. The transfer of heat and mass during extrusion-cookingi sreporte di nth echapter s4 an d5 .Th eeffec t ofequipmen tdesig no nreacto rperformanc eha salway sbee nver y strong. As an example of the outcome of a certain important designfeatur eo nth edegre eo ffil li na twin-scre wextrude ran d theresultin gunavoidabl eresidenc etim edistributio neffect si s reported inchapte r 6.Finall y thecapabilit y ofa nextrude rt o replacepartl ya conventiona lbioreactor ,i sdiscusse di nchapte r 7, in which a process is developed for starch disclosure in a cooking extruder, followed by a saccharification process in a conventionalpiec eo fequipment ,leadin gt oa calculabl eproces s with unique features in bioconversion. The above mentioned aspectsar eexplaine d further inth eparagraph s 1.2 to1.7 .
1.2Historica l development Extrusion cooking isnowaday sa foo dan d feedprocessin g and manufacturing tool. The significance of it should not be underestimated since the technology is relatively new and also during the last 40year s it ison e of the few successfulluni t operations for food in Europe that is still developing. Industrial extrusionprocesse showeve rdi dno toriginat e inth e food sectorbu t inth e rubber andplasti c polymer industry. In this branche there was a need for continuously operating manufacturing equipment in order tomee t the increasing demand forrubbe rhoses ,tubin g andsheet s forbuilding ,householding , farmingan dstartin gchemica lindustry .Betwee n1855-188 0singl e screwextruder s forrubbe rproduct swer edeveloped .Jus tbefor e
Fig.l. lCol d forming extruder 1900als osom eapplication s inth emeta lbranc hwer edescribed , e.g.th econtinuou sproductio n oftubin g outo f leadan dalumi niuman dsom eheat-formin gprocesse suse db yth estee lindustry , that canb e partly understood as extrusion processes. In those casessingl escre wextruder swer euse d asfrictio npumps ,whil e latth e sametim ea formingtas k isperforme d (Fig.1.1) .
Atfirs ti n193 5th eapplicatio no fsingl escre wextruder sfo r plasticating thermoplastic materials became more common as a competitor for hot rolling and shaping in hydraulic-press equipment.A plasticatin gsingl escre wextrude ri sgive ni nFig . 1.2,provide dwit ha typica lmeterin g screw,develope d forthi s Fig. l.2a Typical plasticating single screw extruder
feed compression metering section or tapered section section Fig 1.2b Typical plasticating screw application.
Inth emi dthirtie sw enotic eth e firstdevelopmen t oftwin - screwextruders ,bot hco-rotatin gan dcounte rrotating ,fo rfoo d products. Nearly inth e sameperiod , but shortly after, single screwextruder s inth epast a industrybecam ecommo nus e forth e production of spaghetti and macaroni-type products. In analogy withth echemica lpolyme rindustr yth esingl escre wequipmen twa s usedher eprimaril y asa frictio npum pan d itact smor eo rles s as continuously cold forming equipment, using conveying-type screws (Fig. 1.1).I t is remarkable that nowadays the common pasta products are still manufactured inth e same single screw extruder equipment with a length over diameter ratio (L/D) of approximately (6t o 7). Themai ndifferenc ewit hth ethirtie si s that since thatperio d much development work hasbee n done for screw and die-design andmuc h effort hasbee npu t into process control,suc ha ssophisticate dtemperatur econtro lfo rscre wan d barrelsections ,di etemperin gand ,th eapplicatio no fvacuu ma t the feed port. Finally scale up of the equipment hasbee n done froma poo rhundre d kiloshourl yproductio nu pt osevera l tons. It should be noticed that this progress in pasta extrusion technologymainl yha sbee ngaine di nth etim eperio dafte r1975 . In order to achieve this, combined research was performed by equipmentmanufacturer san dtechnologist sfro muniversities .Thi s hasle dt oth econstructio n ofhig hefficienc y screws,improve d screw and barrel-design, the development of good steeltypes, reduced wear, and better constructed downstream equipment such ascutter san ddryers .
The development of many different technologies seems to be catalysed by World War II, as is the extrusion cooking
solid compacting melt conveying melting conveying zone . zone zone Fig. 1.3 Food extruder fordirec t expanded products technology.I n194 6i nth eU Sth edevelopmen to fth esingl escre w extrudert ocoo kan dexpan dcorn -an drice-snack soccurre d (Fig. 1.3). Incombinatio nwit ha nattractiv e flavouringthi sproduc t type is still popular, and the way of producing snacks with singlescre wextrude requipmen t isi nprincipl e stillth esame . Awid evariet y ofextrude rdesign s isoffere d forthi spurpose . However it should be mentioned that the old method of cutting preshapedpiece so fdoug hou to fa shee twit hroller-cutter s is still inuse ,whic h isdu e toth e facttha t complicated shapes ofsnack slea dt over yexpensiv edie san ddie-head s forcookin g andformin gextruders .Her eth elac ko fknowledg eo fth ephysica l behaviour of atempere d dough and theunknow n relations of the transportphenomen a ofheat ,mas san dmomentu m towardsphysica l andphysico-chemica l propertieso fth e food inth eextrude rar e clearlynoticed .Althoug hmoder ncontro ltechnique soffe ra bi g help incontrollin gth emas s flow insingl escre wextruders ,i n alo to fcase si ti sa bi gadvantag et ous eextruder swit hbette r mixingan dmor estead ymas sflow ,the nsingl escre wequipmen ti s offering.
Inth emi d seventiesth eus eo ftwi nscre wextruder s forth e combinedproces so fcookin gan dformin go ffoo dproduct sha sbee n introduced, partly as an answer to the restrictions of single screw extruder equipment since twin screw extruders provide a more or less forced flow, and partly because they tend to be machinery with better results in scale up from the laboratory extruder types,i nus e forproduc tdevelopment .
1.3 Differences between food-and plasticating-polymers. Whenw e compare the effects of the polymers in thermoplastics and foodw enotic etha tnearl y allchemica l changes infoo dar e irreversible.A continued treatment after such an irreversible reaction ina nextrude rshoul db ea temperature ,tim ean d shear controlled process leading to a series of completely different functionalpropertie so fth eproduce d food (Table 1.1) Tablel. lCompariso n between thermoplastic- and food polymers
Plastics Food
1 Feed to the Single polymer Multiple solids, water and extruder oil
2 Composition Well defined Not well defined. Natural structure and biopolymers, starch, protein, molecular weight fibre, oil and water
3 Process Melting and Dough or melt-like formation forming. No with chemical change. chemical change. Irreversible continued Reversible treatment leads to wanted specific functional properties
4 Die forming Shape is subjected Subjected to extrudate swell to extrudate swell and possibly vapour pressure expansion
5 Biochemicals Use of fillers, Use of enzymes and e.g. starch biochemicals for food conversion
An answer nowadays of food familiar extruder equipment manufacturers is the design of process lines, where the extruder cooker is part of a complete line. Here the extruder is used as a single or twin screw-reactor, wheras the preheating/ preconditioning step is performed in an especially built preconditioner. The forming task of the extruder has been separated from the heating and shearing. The final shaping and forming has to be done in a second and well optimised post-die forming extruder, processing the food at a lower water level than in the first extruder (Fig. 1.4). In such a process line cooked Dry ingredients Mix
KA7WVI water lion Jd rjimT-n-jTri ™"f J I 11 11"!If l 17] steaitif\lm j Preconditioner
Forming Extruder Fig. 1.4 Extruder line for cooked and formed unexpanded food pellets andpreshape dbu tunexpande d foodpellet sca nb eproduced .Mos t benefiti sgo ti nhandlin gan dcontrollin gth eamount so fwate r inth edifferen t stageso fth eprocess .I nth epreconditioner -
Fig. 1.5 'Archimedes' Fig 1.6 Plasticating polymer conveying section melting model water, steam and possibly oil are injected. In the cooker extruder a special screw is operated avoiding steamleakage towardsth e feed zonean dthu shavin gmos tbenefi to fth eextr a steam input, for quicker cooking andhighe r throughput, caused byth elowe rviscosit y inth ecookin gzone .Afte ra ventin gste p at the die plate of the extruder a special forming extruder produces a homogeneous, precooked food material at the lowest possiblemoistur e content.Th eresul t isa typica l foodproces s linedifferin gver ymuc h froma comparabl eextrude r lineou to f thechemica l plasticating industry.
8 With the use of the extrusion equipment inproces s lines their tasksbecam emor especialised .Thi sencourage dth ecompariso no f extruder performance with that in the plastic industry, which promotesth etransformatio n ofth eextrusion-cookin g craft into a science, taylor made for the 'peculiar' properties of biopolymers. 1.4 Food melting In analogy with the chemical industry, handling plasticating polymers, a good understanding of the so called food melting phenomenai nth edifferen textrude rtype si sneede dfo requipmen t design.I ftha tknowledg ei savailabl eth eshearin gscre wsectio n inth esingl escre wextrude ro rth emeltin g sectiono fa se to f screwso fa twi nscre wextrude rca nb edesigned .Th e firstpar t of each screw is a conveying section, that brings the food material in the direction of the melting In a single screw extruderthi sconveyin gscre wi sdesigne da sa nArchimedes-scre w allowing a limited filling degree, 'resulting' in a solid bed thatonl y fillsu pth ebotto m ofth echanne lan dtha t ispushe d by the screw flight (Fig. 1.5).A s a result of the pumping towards a "pressure slope"th e food material iscompacted , the airi spresse d out,pressur e isdevelope d andth e foodmateria l isheate db yfrictio nan dshear .I nth eplasti cpolyme r industry themeltin gmode lo fMaddoc kan dVermeule n (1959)i suse d (Fig. 1.6),whic hinclude stha tth escre wfligh tpushe sa fas tmeltin g solid bed through the compression section of the screw (Fig. 1.2). This melting takes place over the melting distance, different in length, depending on the selected screw pitch, compression ratioan dbarre ltemperature .Th emeltin gmodel sb y Maddock and Vermeulen (1959) are almost invalid for food material,howeve rth emode ldescribe db yDekke ran dLind t (1976) (Fig. 1.7) isquit e well usable and describes more or lessth e typical way of foodmeltin g for anumbe r of foodapplications . Here the food melting only starts in the way as described by Maddock and Vermeulen (1959) in the first pitches, but imme diately aftertha t the solid bed is lifted up by themel t and atim econsumin gproces sstarts .Hereby ,th ehea tdevelope dove r thedecreasin g friction inth edirectio n of the die,cause d by Fig. 1.7 Meltingmode l by Dekker and Lindt adecreasin gviscosity ,ha st openetrat et oth esoli dbed ,whic h crumblesawa yan dfinall yi spresen ta sa thi nsoli dcrus ti nth e middleo fth escre wchannel .The ni tdepend so nth elengt ho fth e metering section ifth eavailabl e residencetim eallow st omel t downth eremainin gsoli dpart so fth efoo dbiopolyme rcompletely . If the aim of the extrusion cooking process in question is a simple denaturation of the food polymer, without further requirements of food texture, then the experience out of the chemicalpolyme rextrusio n fieldt oappl y special screwmeltin g parts is advisable, as the melting will be accelerated. In principleth eeffec to fthos emeltin gpart si sbase do nimprove d mixing. Thismixin g effect canb e based onparticl e distribution or on shear effects exerted on product particles. For distributive mixingth eeffect so fmixin g arebelieve d tob eproportiona l to thetota l sheargive nby :
dv dt (1) •/ dx
Whereasfo rdispersiv e (shear)mixin gth eeffec ti sproportiona l withth e shear stress (T):
T=J1 dv (2) dx
Thegrou po fdistributiv emixin gscrew sca nb edivide di nth epi n
10 mixing section, the Dulmage mixing section, the Saxton mixing section,th epineappl emixin ghead ,slotte dscre wflight san dth e cavity transfermixin g sectionrespectivel y (Fig.1.8) .Th epi n mixing section or avarian t ofthi s design isuse d for food in theBus sCo-kneade rwhic hi sa reciprocatin gscre wprovide dwit h pinso fspecia lfeature ,rotatin gi na barre lprovide dwit hpin s aswell . Whenw e recognize food expanders tob e food extruders thenmuc h use ismad e here of thepin-mixin g effect,sinc e the barrelso fexpande requipment ,buil tlik eth eorigina lAnderso n design,ar eprovide dwit hmixin g pins.Th eamoun to fmixin gan d shearenerg y issimpl y controlledb yvaryin gth enumbe ro fpin s (Fig.1.9) .Mixin ghead s liketh eDulmag edesig nan dth eSaxto n mixing section as such areno t used in food extrusion cooking, although some single screw extrudermanufacturer s have designs
a. Pin Mixing Section b. Dulmage Mixing Section
/ 0000000 \ -o-G-$~e««-e-o -V
c Saxton Mixing Section d. Pineapple Mixing Section a. Anderson Expander
f. Cavity Transfer Mixing Section h. Buss Kneader Fig 1.8 Distributive mixing Fig. 1.9 Mixing pin elements elements Anderson andBus s designs available, like e.g. special parts of Wenger single screw equipment, where some influence of the mentioned designs is recognizable. The pineapple mixing section or the most simple mixingtorped oha sa futur ei nfoo dextrusio ncookin gdu et oit s simplicity andeffectivity .Th edesign so fFig . 1.8.e and1.8. f areattractiv e infoo dextrusio nbecaus eo fthei rgentl emixin g action, which implies that mixing heads of this design can be used at the end of the metering sections of food extruders
11 withoutdisturbin g themass-flo w atthi s locationto omuch . In chapter4 i sdescribe d how solid saccharose crystals are succesfully meltedi na singl e screw extruder usinga torped o mixing head andho wheatin g anddissolutio n ofsaccharos e crystalsi nliqui dglucos esyru pi na twi nscre wextrude rca nb e described whena mixin g region consisting ofa pai r of slotted screw flights ispresent .
Both the designsi nFig . 1.8.e and Fig. 1.8.f are usedt o correctth eno ncomplet emeltin go ffoo dparticle si nth emeltin g partso fth eextrude rscrews .I ti slogi ctha tth eles sexpensiv e slotted screw flight ismostl yused ,especiall y inth emeterin g zone,providin gth efina lmixin gbefor eth efoo dmateria lenter s the die area, ata lo wleve l ofshear . Whenw enee d more
a Union Carbide Mixing Section b.Ega n Mixing Section
barrel valve barrel SfflS 53^5r c. Dray Mixing Section d Blister Mixing Ring
(- -s. sere* /screw . 1
barrel valve open barrel valve closed
e. Z-Shaped Fluted Mixing Section APV-Barrel Valve Fig. 1.10 Dispersive mixing Fig.1.1 1Adjustabl ebarrel - element valveb yD .Tod d (APV-Baker) dispersivemixin gt ob eexerte dt oth eproduct ,w eca nmak eus e ofth emixin gelements ,suc ha sth eUnio nCarbid emixin gsection , the Egan mixing section, the Dray mixing section, the Blister mixing ring or the Z-shaped fluted mixing section (Fig. 1.10). All these designs originate from the plastic industry, buit n foodextrusio n cooking onlyvariant so fth eBliste rmixin g ring areused ,o fwhic hth emos tremarkabl edesig n isth eon euse di n the Baker Perkins equipment, an adjustable barrel valve, constructed by D.Tod d (Fig. 1.11). Herea nadjustabl evalv eo f
12 which the position can be controlled from the topside of the barrel, ismounte d between two Blister rings,place d as apai r inth e screwset.Th e exact function isdescribe d inchapte r 6. It isexpecte d that theus e of such elements in food extrusion cooking will shorten themeltin g section of the extruder screw andb y thiswil l shorten thetota l length ofth e screw,du e to theabilit yo fth ebarre lvalv et oinfluenc eth edegre eo ffil l in the extruder and increasing the fully filled length as a result.Thi sallow sa bette rdissipatio no fth emoto rpowe rove r the friction ofth eproduct .
Inplasti cpolyme rextrusio nw ewil lals o findth e socalle d barrier flightextrude rscrew s (Fig. 1.12).A well-know ndesig n isth eon eb yMaillefe ra twhic hth emel ti sallowe dt oflo wove r the screw flight to its own melt channel. A variant on this design is the one by Ingen Housz which can be seen as an optimization of the separation ofmel t and solids on the screw forplasti cpolymers . Uptil lno wi nth efoo dan d feedindustr yther ei sno tmad ean y use of those barrier flight screws. In single screw food
^^OTXS a. Double Flight Screw a. Maillefer Screw ^5ES^> b. Variabele Pitch Screw ^k$m=tu±tdk[nb b. Ingen Housz Screw c. Venting Screw Fig 1.12 Barrier flight extruder Fig.1.1 3 Special screwsfo r screws single screw extruder extrusionon ewil lmak eus eo fdoubl e flighted screws,variabl e pitchscrew san di nrar ecase sa ventin gscre w (Fig.1.13) .Som e extruder manufacturers offer modular screw set-ups for single screw designs,makin g themmor e flexible than the solid screws out of Fig. 1.13, but it should be said that, especially when
13 aspectso f flexibility are inquestion ,th eselectio n ofa twi n screw extruder canb ejustified .
Inconclusio n itca nb esai dtha tther ei sa stron g impacto f plastic polymer extrusion on the field of food extrusion technology.A tfirs tther ei sth eavailabilit yo fwel ldevelope d and refined hardware, which includes extruder equipment and instrumentation & control systems. Of course they have to be developed further and/or sifted out to suit specific food applications.Secondl yther ei sth eavailabilit yo fpolyme rengi neeringproces sknow-how .Althoug hthi sknow-ho wi sver ylimited , even forthermoplasti c extrusion operations, itha s formed the basiso f foodextrusio n engineering analysis aswell .
1.5 Rheological considerations It has already been mentioned that usually non-Newtonian flow behaviour is to be expected in food extruders. A major complication is that chemical reactions also occur during the extrusion process (e.g. gelatinization of starch or starch- derivedmaterials ,denaturatio no fproteins ,Maillar dreactions) , whichstrongl yinfluenc eth eviscosit yfunction .Th erheologica l behaviouro fth eproduct ,whic h isrelevan tt oth emodellin gO f theextrusio nproces sa ssuch ,ha st ob edefine d directly after the extruder screw before expansion in order to prevent the influence of water losses, cavioles in the material and temperature effects due to the flashing process that occur as soon as the material is exposed to the (environmental) air. A convenient way would be to measure the pressure loss over capillaries with variable diameter and length. However, this methodha sa certai nlac ko faccuracy ,sinc epressur elosse sdu e to entry effects are superimposed on the pressure gradient inducedb yth eviscosit yo fth emateria l (Bagley1957 ). Moreover , since the macromolecules in the biopolymers introduce a viscoelasticeffect ,th ecapillar yentr yan dexi teffect scanno t beestablishe d easily fromtheoretica l considerations.
14 Inorde rt oovercom eth edifficultie smentione dabove ,severa l authors (van Zuilichem 1974,1980 ,vo n Lengerich, 1984)prefe r
driving power
slit viscometer
/iscometer housing— .. •—T
vanable slit ( 0.5 - 2.0
Fig 1.14 Slitviscosimete r a slitviscosimete r (Fig.1.14) .Her epressure sca nb emeasure d inside the channel in order to eliminate the entry and exit effects.
Itha st ob erealize d that ifth epressur egradient shav et o be measured at different slit geometries, and at different throughputs, the thermal history and the temperature of the material in the slit are changing. In order to determine the exactviscou sbehaviou ro fextrude d food,a correctio n forthes e twovariation sha st ob e introduced.
Iti swel lknow ntha twithi nnorma loperatin grange sstarche s andprotein-ric hmateria lar eshea rthinning .Thi sjustifie sth e use of a power law equation for the shear dependency of the viscosity:
(3) Tla=*ltl
15 in which rjg is the apparent viscosity, V stands for the shear ratean dn isth epowe r lawindex . Metzner (1959)ha sargue dtha t forchangin gtemperatur e effects this equation can be corrected by multiplying the power law effectan dth etemperatur edependency ,thu sgiving :
1 r\a=k' IYI""exp(-pAT ) (4)
Theviscosit ywil lals ob e influenced byth eprocessin g history of themateria l as itpasse sthroug h theextruder . In order to correct forthi schangin gtherma lhistor y itha st ob e realized thatinteraction sbetwee nth emolecule sgenerall y occurthroug h thebreakin gan dformatio no fhydroge nan dothe rphysico-chemica l bonds.Thi scross-linkin geffec ti sdependen to ntw omechanisms : atemperatur eeffec tdetermine sth efrequenc yo fbreakin gan dth e formationo fbond san da shea reffec tdetermine swhethe rth een d ofa bon dtha tbreak smeet sa "new "en do rwil lb ereattache dt o itsol dcounterpart .I fw eassum etha tthi slas teffec twil lno t bea limitin g factora ssoo na sth eactua lshea rrat e ishighe r thana critica lvalu ean dtha tth eshea rstres slevel swithi nth e extruder arehig h enough,the n theproces sma y be described by anArrheniu smodel ,giving , forth ereactio nconstant :
K(t)=k„exp--J£— (5) RT(t) where AE is activation energy, R the gas constant and T the absolutetemperature .Unde rth eassumptio ntha tth ecrosslinkin g processma y be described as a first order reaction, it iseas y tosho wtha t fromth egenera l reactionequation :
-f=^l-0 (6)
16 the following canb ederived :
I T (7) 1-C= ex p -fexp =r-r-Dt ^o J RT(t)
inwhic h Cdenote s the ratio between actual crosslinks and the maximum number of crosslinks that could be attained, and where Dt isa convective derivative accounting forth e facttha t the coordinate system isattache d toa materia l elementa si tmove s through the extruder. Therefore, the temperature, which is of course stationary at acertai n fixed position inth e extruder, willb ea functiono ftim e inth eLagrangia n frameo freferenc e chosen. This temperature history is determined by the actual positiono fth eelemen t inth eextruder ,a sha sbee nprove d for syntheticpolymer sb yJansse ne ta l (1975). Iti sexpecte d that this effect will cancel out within the measuring accuracy and thata noveral leffec tbase do nth emea nresidenc etim eT ma yb e chosen. In combination itma y now be stated that the apparent viscosity of the material as it leaves the extruder may be summarized byth e followingequation :
1 (8) r\a=k" IYI""exp(-PT )ex pj*exp -R £* Dt >0 /
Underth erestrictiv e assumption thatth econstant sk ,n , &an d AEar etemperatur eindependent ,i ti sobviou stha ta tleas tfou r different measurements have to be carried out in order to characterize themateria l properly!
Itappeare dtha tth epropose dprocedur eca nb eapplie dadequatel y tomaterial swit hsufficien thomogeneit ya sther ei se.g . inso y andpurifie d starches.
Onceth eapparen tviscosit y isknow n iti spossibl et oplo tth e value of a power number P* for different Reynolds numbersRe . This ispresente d for soy inFig . 1.15 and leads to avalu e of
17 Mapp Fig 1.15 Power-number vs -Re- for soy (van Zuilichem et al., 1980)
P*=10fo rth e single screw extruder in question (vanZuiliche m etal. ,1980) . Theexpressio nfo rth eapparen tviscosit y ineq .6 ca nno wb e a key factor in the calculation of the dissipated and convec- tionalheat ,a swil lb edescribe d inchapte r5 .
1.6 Residence time distribution The irreversible chemical reactions of food materials as mentioned in Table 1 make the time temperature history an important subject in extrusion-cooking. The residence time distributiongive ssom eclue st oth etim etemperatur ehistor yan d it is not surprising that there are more residence time distribution studies known for food,- than for plastic applications (Jager, 1991). Several studies by Wolf and White (1976), Janssen (1979), Pinto and Tadmor (1970), and Bigg and Middleman (1974) have been directed towards experimental verification of residence time distributions in extruders. Measurements performed with radioactive labelled material such asw Cu areshow nfo rmaiz egrit si nFig .1.1 6 andFig .1.17 .Th e measurementsyste mconsist so fa dua ldetector-se ti ncoincidenc e mode, allowing to measure accurate RTDs as described by van
18 spectrometer 1_channe l
anaLyser channel advonce 1024.channel generator supply sdntiJJgtion detector
magtape UV\AA>
spectrometer 1-channe l
-•time [S] response diagram
Fig l.1 6 Measurement set-up for residence time distributions (Bartelse tal. , 1982)
theory (n =1 )
N (kg/kg)
0 0.142 1.33 A 0.142 166 D 0142 2 0 • 020 1.33 • 0 20 20
Q5 1.5 2.0
Fig1.1 7F(t )curv efo rextrusio no fcor ngrit si nasingl escre w extruder (Bruine tal. , 1978)
Zuilichem atal . (1988). Itca nb esee n from Fig. 1.17that , within reasonable limits,th edistribution s inresidenc e times agree with thetheor y asi ti sderive d forsyntheti c polymer extrusion.Th esam eresul thold sfo rou rso yan dmaiz eextrusio n measurementsa sreporte di nchapte r2 .Ther ei ssom etendenc yfo r anearl ybreakthroug ha thighe rrotationa l speedswhe n compared with the theory. This wastake n from earlier work byva n
19 Zuilichem et al (1973) where measurements with a single screw extruder were reported. When evaluating these measurements careful,th econclusio nwa smad et oinvestigat ethi sproble monc e again but this time with a better measurement set up.Thi s is showni nth efollowin gchapter s2 an d3 fo rsingl ean dtwi nscre w extruders,wher ei nbot hcase sth esingl edetecto rfro mth ewor k in 1973 was replaced by a pair of detectors working in coincidence mode. This measuring set up allows very accurate measurements,givin greliabl eRTD-plot sfo rsingl ean dtwin-scre w extruders.
1.7 Quality parameters In conclusion, from the foregoing it can be stated that an extruder may be considered as a reactor in which temperature, mixing mechanism and residence time distribution are mainly responsible fora certai nviscosity .Qualit yparameter ssuc ha s thetextur ear eofte ndependen to nthi sviscosity .Th einfluenc e of various extruder variables like screw speed, die geometry, screwgeometr yan dbarre ltemperatur eo nth eproduce dqualit yi s describedb ynumerou sauthor sfo rman yproduct se.g .fo rso y and maize by van Zuilichem (1974,1975,1976,1977) and Bruin et al (1978). However, other extruder process variables like initial moisture content, the intentional presence of enzymes, the pH during extrusion,etc. ,als opla ya role .Althoug h avariet y of testmethod si savailabl e aversatil einstrumen tt omeasur eth e changeso fconsistenc ydurin gpastin gan dcookin go fbiopolymer s ishardl yavailable .A serie so fmeasurin gmethod si suse di nth e extrusioncookin gbranch .A compilatio no fthe mi sgive ni nFig . 1.18,fro mwhic hca nb esee ntha ta nextrude dan dcooke dproduc t isdescribe di npractic eb yit sbul kdensity ,it swate rsorption , its wet strength, its dry strength, its cooking loss and its viscosity behaviour after extrusion for which the Brabender viscosimetry producing amylographs may be chosen, which learn about the response of the extruded material on a controlled temperature,tim e function (Fig. 1.19). Only well trained people will understand those Brabender diagrams.Fo ra succesful ldescriptio no fpropertie so fstarche s
20 determination of product- propertie s
bulk density water sorption strength dry strength cooking-losses viscosity product wet product gelatination I weighing dry weighing dry area determination Kramer shear sample sample sharpy hammer press weighing dry torque sensor volume measurement water addition sample soaking C> for 30 min. l and drainage canning tor 10 min. I retorting weighing wet energy- torque sample consumption ! measurement drying l I strength I relation temp.-time $ weighing versuse.g .Brabende r calculation units calculation units l calculalation calculation units units (kg /m3) (percentage) calculation units ( kg per g product) (percentage) a % (kg cm" ) a**/. amylogram I I Fig. 1.18 Measuring methods for extruder product properties (Bruin et al., 1980) andprotein sw ewil lnee dadditiona lchemica ldat alik edextrose - equivalents, reaction rate constants and data describing the sensibility for enzymatic degradation. The task of the food engineer& technologis twil lb et oforecas tth erelatio nbetwee n thosepropertie san dthei rdependenc yo fth eextrude rvariables . Thereforei ti snecessar yt ogiv ea (semi)-quantitativeanalysi s of the extrusion cooking process of biopolymerswhic h can be done when adopting an engineering point of view, whereby the extruder isconsidere d tob ea processin g reactor.Althoug h the numbero fextrusio n applicationsjustif y anoptimisti cpoin to f view, more experimental verification is definitely needed focussedo nth eabov ementione dresidenc etim edistribution ,th e temperature distribution, the interrelation with mechanical settings likescre wcompositions ,restriction st o flow.An dth e useo fbio-chemical slik eenzymes .Enzyme sinfluenc edirectl yth e viscosityo fth ebiopolyme ri nth escre wchanne lo fth eextrude r reactora si sinvestigate d inchapte r7 wher ea nextrude rcooke r is used as a biochemical processreactor. After many of these efforts onewil l be ina positio n togiv e a fairanswe r toth e
21 time (min.) 103
95 95 29 95 95 29 temperature (•C) temperature (»C) Fig 1.19 Amylograms showing the influence of product moisture content and extrudervariable s (vanZuiliche m et al., 1982) question ifextrusio n cooking isa craf to ra science .
1.8 Influenceo fmarketin g interestso nextrusio n cooking In food production the most advanced extrusion processes are foundi nth esnac kindustry ,wher eth esale sar edominate db yth e combined effortso fth emarketing -an dth eproduct-development - departments.Th eproduc tdevelopmen tenginee rmostl yi ssomebod y who has grown up with the factory and who has a longstanding experience with the raw materials, commonly handled in the extruder-operations,an dwit hth eextrude requipmen t itself.H e will run trials with selected raw material compositions and flavours andwil l develop the dieswante d by themarketeer s by extensivetria lan derror ,an dfinall yth ene wproduc ti sthere . Sucha ma no rwoma n isa ke y factort oth equalificatio no fth e developedsnac kfo rth esupermarke tshelf .The yca nb edescribe d as specialists, although their knowledge of the fundamental phenomena in the extruder itself is very poor. Taking into
22 accounttha tthei rbackground sar ethos eo fa goo doperato rwit h feeling for the material, feeling for a certain extruder/die design and recognizing the possibilities of a flavour in combination with the new product,tha nw ewil l understand that a nearly impossible appeal isdon eo nthe m inmoder ntime slik e today, since there is the influence of fashion, trends, new exotic rawmaterials ,etc .Th emarke t expects ingenera l every half a year a new succesfull product in the supermarket, with invitingtaste ,color ,bite ,outlook ,shap eregularity ,lowe rfa t content, which means that the "specialist" in question should know much more about the transport phenomena in the extruder- equipment.
Inothe rbranche so fth efoo dindustr ya comparabl esituatio ni s found as in the snackfood branch where the extrusion cooking technology hasbee n introduced.Th econfectioner y industry,th e bakingindustry ,th epharmaceutica l industryan dth eanima lfee d industry canb ementioned . Here the extruder cooker still isa relatively newpiec eo fequipmen tan d specialists inthi s field expectth eextrude rcooke rt ob ea proces stoo lcapabl et ohel p theindustr yt odevelo pne wserie so fproducts .Fo rthi spurpos e one canmak e use of theuniqu e property of the extruder cooker to be a high temperature/pressure short residence time (HTST) pieceo fequipment ,capabl ee. g toreplac econventiona l process lines in itsown .
23 References
Bagley,E.B. , (1957). End corrections inth e capillary flow of
polyethylene.J . ADD.Physics .28 r 624.
Bartels, P.V., Janssen, L.P.B.M.,va n Zuilichem,D.J. , (1982). Residence time distribution in a single screw cooking extruder.
Bigg, D. M., Middleman, S. (1974). Mixing in a single screw extruder. A model for residence time distribution and strain. Ind.Eng .Chem .Fundam. . 13.66-71 .
Bruin, S., van Zuilichem, D.J., Stolp, W. (1978). A review of fundamental and engineering aspects of extrusion of biopolymers in a single screw extruder. J. Food Process Eng.. 2. 1-37.
Dekker, J. (1976). Verbesserte Schneckenkonstruktion fur das Extrudierenvo n Polypropylen.Kunststoffe . 66.130-135 .
Jager, T. (1991). Residence time distribution in twin-screw extrusion. Thesis of the Agricultural University of Wageningen.
Janssen, L.P.B.M. (1978). Twin Screw Extrusion. Elsevier Scientific Publishing Company,Amsterdam .
Janssen, L.P.B.M., Noomen, G.H., Smith, J.M. (1975). The temperature distribution across a single screw extruder channel. Plastics and Polymers.43 .135-140 .
Janssen, L.P.B.M., Smith, J.M. (1979). A comparison between singlean dtwi nscre wextruders .Pro cInt .Conf .o nPolvme r Extrusion.Plastic san dRubbe rInstitute ,London ,June ,91 -
24 99.
Janssen, L.P.B.M., Spoor, M.W., Hollander, R., Smith, J.M. (1979).Residenc etim edistributio n ina plasticatin g twin screw extruder.AIChE . 75,345-351 . von Lengerich, B.H. (1984). Entwicklung und Anwendung eines rechner unterstiitzten systemanalytischen Modells zur Extrusionvo nStarke nun dstarkehaltige nRohstoffen .Thesi s TU-Berlin. 1984F B 13,Nr .165 .
Maddock, B.H. (1959). Technical Papers. Vol. V. 5th Annual Technical Conference SPE.Ne wYork .
Menges,G. ,Klenk ,P . (1967).Aufschmelz -un d Platiziervorgange beim Verarbeiten von PVC hart Pulver auf einem Einschnecken-extruder. Kunststoffe. 57,598-603 .
Metzner, A.B. (1959). Flow behaviour of thermoplastics In: Processing ofThermoplasti cMaterials ,ed .E.C .Bernhardt . VanNostrand-Reinhold , NewYork .
Pinto, G., Tadmor, Z. (1970). Mixing and residence time distributioni nmel tscre wextruders .Polvm .Ena .Sci. .10 , 279-288.
Wolf, D., White, D.H. (1976). Experimental study of the residence time distribution in plasticating screw extruders.AIChE .22 .122-131 . vanZuilichem ,D.J . (1976).Theoretica laspect so fth eextrusio n of starch-based products in direct extrusion cooking process. Lecture Intern. Snack Seminar. German Confectionery Institute,Solingen . van Zuilichem, D.J., Bruin, S., Janssen, L.P.B.M., Stolp. W. (1980). Single screw extrusion of starch and protein rich
25 materials. In: Food Process Engineering Vol. 1: Fopd Processing Systems, ed. P. Linko,V . Malkki,J . Olkku, J. Larinkariels,Applie d science,London ,745-756 . van Zuilichem, D.J., Buisman, G., Stolp, W. (1974). Shear behaviour of extruded maize. IUFoST Conference. Madrid, September, 29-32. van Zuilichem, D.J., Jager, T., Stolp. W., de Swart, J.G. (1988).Residenc etim edistribution s inextrusio ncooking . Part III:Mathematica l modelling of the axial mixing ina conical, counter-rotating, twin-screw extruder processing maizegrits .j.Food.Eng. .8 109-127 . vanZuilichem ,D.J. ,Lamers ,G. , Stolp,W . (1975). Influenceo f processvariable so nqualit yo fextrude dmaiz egrits .Proc . 6thEurop .Svmp .Engineerin g and FoodQuality .Cambridge . van Zuilichem, D.J., de Swart, J.G., Buisman, G. (1973). Residencetim edistributio n ina nextruder .Lebensm .Wiss .
u. Technol..6 r184-188 . vanZuilichem ,D.J. ,Witham ,I. ,Stolp .W . (1977).Texturisatio n of soy flour with a single-screw extruder. Seminar on extrusioncookin g offoods .Centr ed ePerfectionnemen tde s Cadres Industries Agricoles etAlimentaires , (C.P.C.I.A.) Paris. vanZuilichem ,D.J. ,Stolp ,W. ,Jager ,T . (1982).Residenc etim e distributioni na singl escre wcookin gextruder .Lectur eat intern. Snack Seminar. Zentralfachschule der Deutschcn Susswarenwirtschaft. Solingen.
26 Notation
C Concentration [kgm31 Dt Convective derivative [-] k Consistency factor of power-law model [Ns'hn"21 k., Frequency factor [l/s] K(t) Reaction constant [l/s] n Flow behaviour index of power-law model P* Power number R Gas constant [kJmol"1K"1] Re Reynolds number T Temperature [K] v Velocity [m/s] x Channel depth [m] P Temperature correction constant AE Activation energy [kJ/mol; Y Shear [• Y Shear rate [l/s] fj Apparent viscosity [Nsm -2l fi Dynamic viscosity [Nsm -2n T Shear stress [Nm"2
27 CHAPTER2
RESIDENCE TIME DISTRIBUTIONS INSINGLE-SCRE W EXTRUSIONCOOKING .
ABSTRACT Residence time distributions were measured in a single screw extruder fed with maize grits and defatted soya flakes.^C u was useda sa tracer .Th einfluenc eo ffou rvariable sha sbee nstudied ; moisturecontent ,fee drate ,rotationa lspee do fth escre wan ddi e diameter.Rotationa lspee dan ddi ediamete rar eth emos t important variables affecting the mean residence time. The residence time distributions curves are analysed with amode l containing aplu g flow component and a number of CSTRs in cascade. For maize this model contains fourt ote nCSTRs .Th e fito fthi smodel ,however , can be improved. With the proteinaceous material soya,th emode l containstw oo rthre eCSTRs ,an da bette rfi ti sreached ,tha nwit h maize.Th eresidenc etim edistributio nmode li suse dt ocompar eth e measurementswit hmodel sdevelope d forplasti cmel textrusion .
THIS CHAPTERHA S BEEN PUBLISHEDAS : van Zuilichem, D.J., Jager,T. , Stolp,W. , and de Swart,J.G . (1988).Residenc eTim edistributio n inExtrusio nCooking , PartII :Single-scre wextruder sprocessin gmaiz ean dsoya . J. ofFoo d Ena. 7. 19-210
28 2.1 Introduction Cooking extruders are nowadays frequently described as "High TemperatureShor tTim ereactors" ,i nwhic ha rang eo fbiopolyme r reactions occur, such as gelatinization and cross linking of starch, browning and protein denaturation. The process time, together with factors such as shear and temperature, is very
-screw length
section screw-
Fig2. 1Simplifie d extruderschem e (vanZuiliche m et al., 1988)
importantt othes ereactions ,whic hmean stha tever y elemento f food should be subjected to more or less the same extrusion history. However in a single-screw extruder (fig. 2.1) the particles experience variations in extrusion history due to extruder geometry, energy input and rheological effects. This will result in a residence time distribution (RTD) which is dependento nsuc hengineerin gaspect sa sextrude rdesig nan dth e rheological behaviour of the materials extruded. For a good qualityproduc ti ti snecessar ytha tth ebiopolyme rreaction sar e controlledb yth eextrusio nconditions ,whic hrequire sknowledg e of mechanisms of mixing and flow behaviour. It is possible to calculate the RTD from defined velocity profiles, but, more importantly, the RTD can be measured by suitable experiments. Fromth eRTD ,informatio n aboutth edegre eo fmixing ,residenc e time expectancy of the fluid elements and the degree of uniformity of the stress exerted on the fluid elements during
29 their passage through the extruder can be gained. A theory suitable for predicting the RTD-curves and flow rates during extrusion of biopolymers in a single-screw extruder is not available at present, as most biopolymers undergo chemical reactions during extrusion-cooking, which significantly change therheolog y during extrusion (Bruine t al., 1978).
Earlier workwa s performed using ameasurin g system containing a single detector (van Zuilichem etal. , 1973). Residence tine distributions could be described by a model consisting of a serial combination of a plug flow and a cascade of CSTRs.Th e determination of the number of CSTRs required additional measurements.Thi smode lgive sn oinformatio no nth erheolog yo f theextrude dmaterials ,bu tshoul db eregarde da sa nintermediat e betweenrheologically-de fine dtheoretica lmodel san dmeasure dRT D curves.
2.2Theor y A single-screw extruder can be regarded as a continuous rectangularchannel ,forme d onthre eside sb yth escre w surface and on one side by the barrel. For modelling, this channel is usually regarded asa straigh t channelhavin gthre edimensions ,
Fig 2.2 Flow pattern in the channel of a single screw extruder (Janssen, 1978) with the screw stationary and the barrel wall moving. This is
30 demonstrated in fig. 2.2 which omits the barrel wall. The simplest profiles of thevelocit y (v) in the x-y plane of the extruder channel arise when the material is Newtonian with a temperature-independentviscosit yan dwhe nth eradia lcomponent s near the flights are neglected. This last assumption is reasonablewhe n thescre wchanne ldept h isles stha n 10%o f its width.Suc happroximat eprofile swer ederive dabou t3 0year sag o (Carleyan d Stub, 1962):
vx= U£(l-P+H) (1)
and:
vx= £^(2-3? ) (2)
Fig 2.3 Two dimensional flow profile for a constant viscosity Newtonian liquid (Bruine t al., 1978) in which U is the relative velocity of the barrel wall with respectt oth ecentr e lineo fth escrew ,P i sth erati obetwee n pressureflo wan ddra gflow ,an d$ i sth edimensionles sparticl e level within the channel inth e y-direction. Duet o the nature of the velocity profiles, a particle will circulate in the x,y-plane. Fig. 2.3 shows thevelocit y profiles inth e x-y and
31 y-z plane. The result is that the particles all flow down the channel ina helica l pathwit h ane tpositiv e component inth e z-direction (fig.2.2) .Th eshortes tpat hca nb efoun da t2/ 3o f the channel height. Pinto and Tadmor (1970) have analysed the RTD's resulting from these velocity profiles. The exit age distribution (E(t)) forNewtonia n liquidsthe nbecomes :
3 2 97tM)esin£>cos0(l-—K {(£-l)v/l+2£-3S } 3 (3) E(t) = 2L(6S2-45-l)Vl+25-3$2+3$3-l inwhic hN i sth escre wrotationa l speedan dQ th efligh tangle . This result is shown in figure2. 4togethe r with theF-diagra m for complete backmixing and for plug-flow. The mean residence time is equal to the hold-up divided by the volumetric output rate.Th eminimu mresidenc etim ei sapproximatel y3/ 4o fth emea n residence time.Th e shape of the F-curve is independent of the pressure built-up inth e extruder and of the rotational speed, only themea n residence timewil l be affected. TheRTD' s fora power-lawflui di na nisothermal ,single-scre wextrude rhav ebee n analysedb yBig gan dMiddlema n (1974). A sth etemperatur ean dth e viscosity ina food extruder are inhomogeneousth evalidit y of these RTD-curves for describing a food extruder are limited. Power-law behaviour, which is more appropriate to biopolymers thanNewtonia nbehaviou rwa sdescribe d bythe mas : y -D/2 im% 2 n inwhic h rji sth e coefficient of dynamic viscosity andM and n are power-law constants. When n=l the fluid is Newtonian. In figure2.4 apower-la wRT D(Big gan dMiddleman,1974 )i sincluded . The minimum residence time in this case of the power-law RTD-curves depends greatly on the values of the rheological coefficients.Sli pphenomen aa textrude rsurface sals oinfluenc e thevelocit yprofil e (Holslagan dIngen-housz ,1980) .Whe na thi n
32 F 1.0 Bigg and Middleman
Fig 2.4 Comparison of fourF-curve s (Bigge t al., 1974)
layero fwate ri spresen to nth ebarre lo rth escre wsurface ,th e food-material can slip over the surfaces. The barrel of a single-screw, food-extruder isusuall ygroove dt opreven tslip . Asth esurfac eo fth escre w ispolished ,sli p ismost likel yt o occur there. This is contrary to the assumption of a zero velocity at the barrel and screw surfaces made by Pinto and Tadmor (1970),an db yBig gan dMiddlema n (1974).A confirmatory indication of such slip was found by an increase in the differencebetwee nth emeasure d andpredicte d outputs,whe nth e rotationalspee dincrease s (vanZuiliche me tal. ,1983) .A mode l describing slip conditions and a non-isothermal, power-law viscosity conditionwa sconsidere d impracticable,a sth enumbe r ofvariable si sto olarge .Th eRT Dmeasurement sar edescribe db y a model consisting of a plug flow and a cascade of W perfect mixers,a sdescribe d ineq .5 :
E(R) (5) (W-l) ! inwhic hR i sa dimensionles stime ,calculate dfro mth eplug-flo w timet pan dth emea n residence time T,a sin :
33 R = <«> * The introductiono fthi splug-flo wtim e increasesth easymmetr y ofth eE-curve . 2.3 Experimental RTD measurements were carried out with a 48 mm diameter, single-screw, Almex Battenfeld-extruder using a screw with a compressionrati oo f3:1 .Aspect sinvestigate dwer eth einfluenc e of the feed rate,rotationa l speed, diediameter , and moisture content on the RTD-curves. During measurements all other variableswer ekep tconstant .Tw omaterial swer euse d :defatte d soya-flakesan dcoarsel ymille ddegerminate dmaiz egrits .I nbot h materialsth emea nsiz eo fth eparticle swa sabou t1 mm .Defatte d lance thermocouples ^EHESHBr \barrei temperature {°C\ 200 Fig 2.5 Temperature profiles maintained for these experiments (vanZuiliche m et al., 1977) 34 soya-flakescontai n56 %protei nan dth eextrude dproduc ti swel l known inth e pet-food industry. During extrusion of the soya a temperatureprofil ewa simpose dwit ha maximu mo f200° Can da di e temperature of 125"C (Fig. 2.5). The lance temperature approximatesth escre wtemperature .Th emaximu mpressur ewa s10 0 bar. Maize grits containing 85% starch and 8% protein were extruded adiabatically with amaximu m temperature of 200°Can d amaximu m pressure of 300bar . 2.3.1 Measuring theresidenc e time distribution Tomeasur e theRT D aradioactiv e tracertechnique ,wit hM Cu as atracer ,wa schose nbecaus eo fit saccuracy ,safet yan dchemica l inertness. The tracerwa s used inth e form of cuprous chloride mixed with some of the feed material. In earlier work (van Zuilicheme t al., 1973)a singl edetecto rwa sused ,reactin gt o all gamma quanta at the energy level characteristic of ^Cu, which also detects somewhat before and after the plane of detection.Fo rth epresen twor ka coincidenc edetecto rwa suse d asdescribe d byva n Zuilichem etal . (1988). 2.4 Results 2.4.1 General The influence of the rotational speed, die resistance and moisture content on themea n residence time,th edimensionles s minimum residence time,th e number of equivalent CSTRs and the so-called "non-Newtonian"characte ro fth eF-curve sar eshow ni n Tables 2.1, and 2.3. The number of CSTRs is used here as an indication of the amount of axial mixing in the extruder. The "non-Newtonian"characte ri sa measur eo fth edifferenc ebetwee n theRT Dcurv eo fa power-la wflui dan dtha to fa Newtonia nfluid . This difference is most prominent just after the minimum residence time (see fig. 2.4).Th e measured RTD curves are comparedwit ha Newtonia nRT Dcurve ,t oestimat eth edifferenc e betweenthe m interm so fth eare abetwee nth ecurve sjus tafte r 35 Table2. 1 influenceo f thedi eresistance ,moistur e contentan d rotational speedo nth eRT Dcurve sfo rsoy aan dmaiz eextrusio n Soya Maize Die Mois Rota Die Mois Rota resis ture tional resis ture tional tance content speed tance content speed Mean residence time r m — V A — V tminh m — V m — V Number of CSTRs — — — m V m Deviation from Pinto and — — — V V A Tadmor( 1970) model 'non-Newtonian' character — Nosignifican t effect, mMixe d effects. A Positiveeffec t ofvariables . V Inverse effect of variables. Table 2.2 Results for soya N Die diameter Moisture content No. of CSTRs W 'mi,,/* r (r.p.m.) (mm) (%) (-) (-) (s) 80 4-0 24-5 3 0-80 66 100 4-0 24-5 3 0-77 58 120 4-0 24-5 3 0-60 38 80 60 23-6 2 0-89 102 120 6-0 23-6 2 0-72 49 80 80 22-9 2 0-84 63 120 8-0 22-9 3 0-77 34 80 8-0 19-8 2 0-74 62 80 80 26-9 3 0-80 59 theminimu mresidenc etime .Th emeasure dRT Dcurve swer earrange d accordingt othei rrotationa lspeed ,di eresistanc ean dmoistur e content on their "Non-Newtonian" character. The qualitative effect of these variables is indicated in Table 2.1. Linear inversecorrelation sbetwee nmea nresidenc etim ean d rotational speed have been found for both soya and maize. Fig. 2.6 shows thiscorrelatio n fora non-Newtonia ndough .Wit hbot hmaiz ean d soya,th e influence ofth edie-resistanc e onth emea nresidenc e time issignifican t (seeTable s2. 2 and2.3) . The minimum residence time decreases faster than the mean 36 Table2. 3 Results formaiz e N Die diameter Moisture content No. ofCSTKs W 'min/l- T (r.p.m.) (mm) (%) (-) (-) (V 80 19 14-4 4 0-56 71 90 1-9 14-4 5 0-56 65 100 1-9 14-4 11 0-56 60 120 1-9 14-4 10 0-49 41 80 30 14-4 10 0-62 59 100 3-0 14-4 5 0-58 48 120 3-0 14-4 10 0-55 41 120 3-0 18-3 12 0-52 44 120 30 21-9 5 0-54 41 lis) 80- 60- 40- /* / maize 14% moisture 1.9 mm die / / 20- / / / / / / I i ' o- 0.2 0.4 06 O.f 1/N(s) Fig 2.6 The relation between themea n residence time Tan d the inverseo f therotationa l speed (vanZuiliche m et al., 1988) residencetim ewit h increasing rotational speed,a sca nb esee n inTable s2. 2 and2.3 ,wher eth erati oo fminimu mresidenc etim e 1 and mean residence time (tm]nT" ) decreases with increasing rotational speed ofth escre w(N) . 2.4.2 Soya The RTD-curves for soya in fig.2. 7 show a near-plug-flow behaviour,whic hresemble sth emode l ofPINT Oan dTADMOR .Thes e curvesca nb esimulate d reasonablywit hth eRT Dmode l ofeq .6 , usingtw oo rthre eperfec tmixer san da plug-flo w timewhic h is approximatelyequa lt oth eminimu mtime .Th enumbe ro fequivalen t CSTRs intabl e2. 2 variesslightl ybetwee ntw oan dthree ,whic h 37 F (t/l) 1.0- 0.8H 0.6 m 229% die x—, c°0ntent diam.8mm 0.4- 23.6 •• •• 6 •• .. 245 •• •• 4 •• 0.2- ~i—i—i—r—i—•—>—>—'—'—'—!—' ~ 0 0.6 1.0 2.0 \/l Fig2. 7 F-Diagram forsoy aextrusio n at8 0rp m (AllFigure san d tables;va n Zuilichem et al.,1988 ) isno tsignificant ,a sth etw obes tvalue swher etw oo rthre ei n allcases .Thi ssignifie stha tth eaxia lmixin g inth eextrude r is almost independent of the moisture content in the range 20-26%,o fdi ediamete ran do fth erotationa lspee do fth escrew . Anexplanatio nfo rthi sphenomeno nca nb efoun di na reactio no f the soya-proteins: soya forms a texture in the extruder (van Zuilicheme tal. ,1979 )whic hfavour splu gflow.Th erati oo fth e minimum andmea n residencetim e intabl e 2.2 isaffecte d byth e rotational speed and the die-resistance, which is at variance withth ePint oan dTadmo rmodel .Th emea n residencetim ewit ha 6 mm die is larger than with the 4 and 8 mm dies, while the effect on the ratio of minimum to mean residence times is unsystematic. The effect on the mean residence time might be caused by differences in the location and extent of the texturizing process, which can affect the hold-up in the extruder. 2.4.3 Maize The first aspect for investigation was the effect of the feed rateo nth eRTD .Remarkably ,bot hth emea nresidenc etim ean dth e minimum residence time decreased by only 5%whe n the feed rate changed from choked feeding to starved feeding. It is obvious 38 that this behaviour is provoked by the combination of acompressivescre wwit ha ver ysmal ldi ediameter .Thi sexplaine s adrasti cdecreas ei nhold-u po fth emaiz edoug h inth eextrude r withdecreasin g output (fig.2.8) ,an da decreas e inth esprea d hold-up 110'V) residence time spread (s) 300 20 die diam. 1.9mm moisture: 14.5 % r.p.m. 75 200- die diam. 1.9 mm moisture U.5 % 16- r.p.m. 75 100- \ 14- t 0 10 20 30 40 50 60 3 0 30 40 5.0 60 output (10 kg/s) output fia3 kg/s) Fig 2.8 Hold-up vs.outpu t of Pig. 2.9 Relationship between maize grits during extrusion thesprea do fth eresidenc etim e at different feed rates and the feed rate of the residence times (fig. 2.9).Th e general shape of the F-curves was not influenced by the feed-rate. Regarding the pressure in the extruder it can be seen in fig. 2.10 that the pressure drop across thedi e isproportiona l to the output.I n fig.2.1 1 thepressur e insideth eextrude r isplotted ,measure d atfou rposition salon gth escre wlength .Th eF-curve s formaiz e grits (figs. 2.12, 2.13 and 2.14) resemble the F-curves for non-Newtonian liquids predicted by Bigg and Middleman. The so-called "non-Newtonian" character is the difference, between thePint oan dTadmo rmode lan dth eRT Dmeasurement si nth eperio d oftim ebefor eth epeak-time ,whic hi sexpresse da sa narea .Th e "non-Newtonian"characte rbecome sprominen twhe nth e rotational screw speed increases,bu tdecrease s when themoistur e content of the maize dough increases. The fit of the maize RTD 39 px10 Pa 1 250- pressure probe.1 200— moisture:14. 6 % die diam. 1.9m m r.p.m. 75 160^ 1.0 3.0 50 outpu«ld3kg/s) 70 Fig. 2.10 Relationship between pressure before the die and throughput formaiz e grits pressure(10 Pa) full feeding 200- die diam. 2.4m m r.p.m.= 80 moisture 14.6% 3 100- full feed= 6.1xl0" kg/s starved feed= 3.4x10" kg/s starved feeding -V 200 300 400 500 600 screw lengthdO m) Fig.2.1 1Pressur ei nth escre wchanne lfo rmaiz eextrusio nwhe n feeding atmaximu m ratean dhal f themaximu m rate measurementswit hF-curve scalculate d fromeq .5 wa sno ta sgoo d aswit h soya.A systematic relationship to the number ofCSTR s could not be found (tables 2.1 and 2.3). The influence of increasing die-resistance on the mean residence time and the "non-Newtonian behaviour" are opposite to the influence of increasing rotational speed.A smaller die gives a longer mean residencetim ean da mor edistinctl y "non-Newtonian"character . 40 F (t/I 1.0- 0.8- 0.6- 0.4- moisture.-14.8% diediam.:3m m 18.3 0.2- 21.9 0 0 2.0 t/l Fig.2.1 2Th einfluenc eo fmoistur econten to fmaiz eo nth eshap e of theF-functio n at 80rp m F(t/L) 1.0- 0.8 mat'l:maize grits 0.6H die diam:1.9mm moisture %:14.8 0.4- •—"80 rpm. —- 90 •• 110 •• 0.2-H 120 •• 0 I y_I I I I i i i ' i i i 0 1.0 2.0 t/L Fig. 2.13 The influence of rpm and the shape of the F-function with die-diameter 1.9mm The moisture content affects the number of CSTRs and the "non-Newtonian" character, but not the mean or the minimum residencetime .I twa sexpecte dtha ta nincreas ei nth emoistur e content would reduce the viscosity and change the rheological constantsi neq .4 .Accordin gt oth epower-la wmode lo fBig gan d Middlemanthi swoul daffec tbot hth emea nan daverag e residence time.Als o the changes inth e number ofperfec t mixers without 41 F (t/t) m-i /^ /r 0.8- ( 0.6- mat'l maize grits die diam.3mm moisture %148 0.4- •—°80 r.p.m. 100 • 0.2- 120 - -^- 1 1 T s 1 1 1 i i i i i i i u () 0.6 1.0 2.0 t/L Fig.2.1 4 Theinfluenc e ofrp man dth eshap e ofth e F-functioon with die-diameter 3mm changes in the minimum residence time are atypical for this model. 2.5 Conclusions Therotationa lspee dan dth edi ediamete rar eth emos timportan t variables affecting themea n residence time.Th e influences of themoistur econten tan do fth efee drat ei ssmall .Compare dwit h maize,soy ashow sles saxia lmixing ,whic hcoul db ecause db yth e changei ntextur eo fsoy awithi nth eextruder .A sth etexturizin g of soya could affect theRT D it isrecommende d that theRT D of soya extrusion is studied in relation toproduc t quality. With maizegrit sa nincreas ei nth efee drat eo rth emoistur econten t decreases the spread of the RTD. Inthes e situations themode l ofBig gan dMiddlema n (1974)predict sconsiderabl echange si nth e minimum residence time, which was not found in the RTD. A possible, but farfetched hypothesis is that such changes are compensated forb y changes inth eRT D of the firstpar t of the extruder,wher eth epower-la wmode li sno tvali da sth emaiz ei s here still present in the form of grits.Th e moisture content should affect the viscosity which should change the mean residencetim eaccordin gt oth emodel so fPint oan dTadmo r(1970 ) ando fBig gan dMiddlema n (1974).Th emea nresidenc etim e isno t infac tfoun dt ob eaffecte db ychange si nth emoistur econtent , 42 while, withmaiz e grits,th ewidt h of theRTD-curve s decreases whenth emoistur econten tincreases ,withou tchangin gth eminimu m residence time. This can be explained by the less farfetched hypothesistha tth ethicknes so fth emaiz edoug hlayer ,o fwhic h thevelocit y isinfluence db y slip,decrease swhe nth emoistur e content increases, resulting in a decrease in axial mixing combinedwit ha decreas e inviscosity .Thi shypothesi s requires slip to be a dominant factor in the RTD of maize grits. With soya,texturizin gi sanothe rpossibl ehypothesis .Th eRTD-curve s of soya resemble the Pinto and Tadmor (1970) model and the RTD-curves of maize resemble the power-law rheology RTD-curves ofBig gan dMiddlema n (1974).Howeve rsom esignifican t features ofth emeasure dRTD-curve sar eno tpredicte db ythes emodel san d sothei rvalidit y forbiopolymer sar eno tproven . 43 References Bigg, D. Middleman, S. (1974). Mixing in a single screw extruder. A model for residence time distribution and strain. Ind.Enq .Chem .Fund . (13)66 . Bruin, S.,va n Zuilichem, D.J., Stolp,W. , (1978). A review of fundamental and engineering aspects of extrusion biopolymers in a single screw extruder. Journal of Food ProcessEngineering .V(2 ) (1)1 . Carley,J.F. , Strub,R.A. , (1962). Basicconcept so fextrusion . Ind.Eng .Chem. . (45)970 . Holslag, R.H., Ingen-Housz, J.F., (1980). Gleitphanomenen in einemExtruder .Lectur ea tZentra lFachschul ede rDeutsche n Susswarenwirtschaft.Solingen . Janssen, L.P.B.M. (1978). Twin Screw Extrusion. Elsevier Scientific Publishing Company,Amsterdam . Pinto, G., Tadmor, z., (1970). Mixing and residence time distributions in melt screw extruders. Polvm. Eng. Scit. (10)279 . van Zuilichem, D.J., Bruin, S., Janssen, L.P.B.M., Stolp,W. , (1980). Single-screw extrusion of starch and protein-rich materials. Proc. of 2nd Int. Congress on Eng. and Food. Helsinki,aug . 1979.745-56 . van Zuilichem, D.J., Reijnders, M.P., Stolp, W., (1983). Enkelschroefsextruders: berekeningswijzen, gedrag en prestaties. P.T. Procestechniek. 12,15-18 . 3. van Zuilichem, D.J., de Swart, J.G., Buisman, G., (1973). Residencetim edistributio n ina nextruder .Lebensm .-Wisa, . u. Technol.6(5 )184-188 . 44 van Zuilichem, D.J., Stolp, W., Witham, I., (1977). Texturizationo fsoy awit ha singl escre wextruder .Lectur e at; European seminar on extrusion cooking of foods. C.P.C.I.A..Pari s van Zuilichem, D.J., Jager, T., Stolp, W., de Swart, J.G., (1988). Residence time distribution in extrusion-cooking part 1:coincedenc edetection .J . FoodEng. .7 ,147-58 . 45 Notation 1 De Dispersion coefficient [m*s' ] E(t) Exit age distribution [s"1] F(t) Cumulative exit age distribution [-] r\ Dynamic viscosity [Nsm'2] n Power-law coefficient [-] N Rotational speed in revolutions per second [s"1] M Power-law constant [-] L Length of extruder screw [»] P Ratio between pressure flow and drag flow [-] Q Angle of screw flight [rad] R Dimensionless time [-] t„,„ Minimum residence time [S] mm t Plug flow time [$] U Relative velocity of barrel wall and screw [ms'1] v Fluid velocity [ms"1] W Number of perfect mixers [-] x,y,z Subscripts indicating relative directions [-] I Dimensionless particle height (y-direction) [-] T Mean residence time [s] 46 CHAPTER3 MODELLING OFTH EAXIA LMIXIN G INTWIN-SCRE W EXTRUSION COOKING ABSTRACT Amathematica lmode lwa sdevelope dt osimulat eth eaxia lmixin gi n a twin-screw extruder.A compariso n ismad ebetwee nth e residence time distributions (RTD) predicted by it and measurements on an actual counter-rotating, twin-screw extruder working on maize grits. The measured residence time distribution curves were characterisedb ythei raverag eresidenc etime san dPecle tnumbers . The model developed contains an infinite series of CSTR's (Continuous Stirred Tank Reactors), each one representing four screw chambers.Th e leakage flowsbetwee n the screw chambers are described in the model as backmixed flows. With this model, a fairlyclos esimulatio no fth emeasure dresidenc etim edistributio n could be made. Mixing in the compression zone proved to be independent of screwspee d inth erang e 40-80rpm .Pecle t numbers of26-3 4wer emeasure d forth etota llengt ho fth eextruder .A sfa r as biochemical reactions occuring mainly in the high temperature compression zone (120-150°C) are concerned, the upstream axial mixing at the low temperatures of the feed zone (60-100°C) may be neglected, in calculating an effective Peclet number for the extruder. This results incalculate d effective Pecletnumber so f 49t o 61,wher e asvalue so f 26-34wer emeasured . THISCHAPTE R HAS BEEN PUBLISHEDAS : van Zuilichem, D.J., Jager, T., Stolp, W., de Swart, J.G. (1988). Residenc eTim eDistribution si nExtrusio nCooking , Part IIIMathematica l Modelling of theAxia l Mixing ina Conical,counter-rotating ,Twin-Scre wExtrude rprocessin g MaizeGrits .J . ofFoo dEna., 8109-127 . 47 3.1 Introduction Althoughtwin-scre wextruder sar eknow nt ob eexcellen treactor s forprocesse ssuc ha sth eenzymi chydrolysi so fstarc h (Chouvel etal. , 1983), starchgelatinizatio n and liquefaction (Linkoe t al., 1983),formatio no famylose-lipi dcomplexe s (Merciere tal . 1980)an dethano lproductio n (Ben-Gerae tal. ,1983) , knowledge of reaction conditions in the reactor part of the extruder is incomplete.Thi si sillustrate db ya quotatio nfro mLink oe tal . (1983) :"Current uses of HTST extrusion cooking are basically physical, chemical and biochemical transformations, yet only recently has an extrusion cooker been studied as a reactor. Betterunderstandin g ofth ebasi cphenomen a takingplac e inth e high temperature, high pressure and shear environment of the reactor is necessary ". The phenomena mentioned are mainly studied inthre e fieldso fresearc h :engineering ,rheolog yan d biopolymer-chemistry.Engineerin gi sconcerne dwit hth egeometr y ofth ereacto ran dwit hcontro lan dmeasuremen t (Janssen,1978 ; van Zuilichem et al., 1983; Savolainen and Karling, 1985). Rheology coversal laspect so fmateria l flow (Launayan dLisch , 1983; Bush et al.,1984) . Transformations and modifications of starch,lipids ,sugars ,cellulos ean dprotein sar estudie di nth e fieldo fchemistr yan dbiochemistr y (vanZuiliche me tal. ,1983 ; van Zuilichem et al., 1985). Individual competence across all thesefield si srare ,whic hmake sth estud yo fcomple xphenomen a inth eextrusio no fbiopolymer sinterdisciplinary .Th ecombinin g of efforts in these fields is greatly helped by mathematical models which can describe the complex phenomena studied. Publications onmeasurement s of axial mixing in starchy doughs intwin-scre wextruder s (Olkkue tal. ,1980 ;Bouni ean dCheftel , 1986 ;Altomarean dGhossi ,1986 )d ono tcontai nan yaxia lmixin g modelwhic h isgenerall y accepted. Inth estud y reported herein a model was developed to simulate axial mixing in a twin-screw extruder. This model was used to translate residence time distribution curves, measured on a conical, counter-rotating, twin-screw extruder fed with maize grits, into mass-flows probably occurring inside the extruder. The refinement of this 48 model into amor e complex onewhic h can describe the extrusion process fora particula r rheology,seem spossible . 3.2 Theory Asth emateria lprocesse dpasse sthroug ha twin-scre wextruder , the chambers decrease in size whilst temperature and pressure increase.Leakag eflow sbetwee nchamber si nth ecompressio nzon e are caused by pressure gradients between the chambers. InFig . 3.1 the leakage flows are identified by the particular gaps throughwhic hthe ypass .I nthi swa ytetrahedron ,calender ,sid e and flight leakage flows can be distinguished (Janssen, 1978). Flight leakage is that over the flights of the screws. The calender leakage is that through the gap formed between the flight of one screw and thebotto m of the channel of the other screw: This material remains in the same screw channel. Side leakagei stha tthroug hth ega pbetwee nth eflank so fth eflight s of the two screws from the lower to the upper side of the intermeshing zone of the screws and the tetrahedron gap isth e onlyga pthroug hwhic hth emateria l istransporte ddirectl yfro m onescre wt oth eother .Thi sga pi sa resul to fth eangl eo fth e flightwall si nrelatio nt oth echanne lbotto mo fth escrews .I n the compression zone the pressure gradient is large enough to promote leakage but inth etranspor t zonen o leakage flowsar e transport direction Fig.3. 1Mode lo fa twi nscre wextrude rwit htw ostar tscrew .Q c, Qs,Q fan dQ tar e leakage flowsthroug h thecalender , flight and tetrahedron gaps respectively (Janssen, 1978) to be expected. Axial mixing in twin-screw extruders has been 49 studied mostly in relation to plastics or model fluids having rheological and other physical properties different from those ofstarches .On e importantdifferenc e isth ecompressibilit y of the material. When grits pass through the feed zone of a twin-screwextrude rthe ybecom ewater-solubl eb ygelatinization , which increases their compressibility. Albers (1976) gives specific densities of PVC as a function of temperature and pressure (Fig.3.2) .Th ePV Ci sheate d foron ehou rt oa certai n temperature and then compressed for one minute, after which P[g/cm3] 10 P[kg/cm 2] Fig 3.2 Density p of dryblend PVC (Halvic 227) related to pressure and temperature (Albers,1976 ) procedureth edensit yi smeasured .Measurin gth especifi cdensit y ofstarc hunde rhig htemperatur ean dpressur e is,however ,mor e complicated asth ephysica lan dchemica lpropertie so f starches will change rapidly when heated. As water functions in extrusion-cooking as a solvent for starch, it isclea r why the porosity of a starchy dough ina counter-rotating, twin-screw, extruderreadil ybecome s zeroa ta lowe rpressur etha ntha tfo r PVCgranules .Becaus eo fthis ,an dth eabsenc eo fleakag e flows inth etranspor tzone ,i tfollow stha tther ei sa rapi d increase indensit ywhe nth emaiz edoug henter sth ecompressio nzone .Fo r maizegrit sa maxima lspecifi cdensit yo fabou t135 0kgm" 3ca nb e 50 found in the compression zone. The specific density in the transport zone was evidently much less. The residence time distributionsar echaracterise db ythei rmea nresidenc etime san d Pecletnumber s (Pe).Th ePecle tnumbe rwhic hi sdiscusse db yva n Zuilicheme tal . (1988a), characterize sth esprea do fth eRT Dan d isdefine das : vl Pe - — (i, in which 1 is the extruder length, De the axial dispersion coefficient and v the axial velocity of the extruder chambers. The larger the value of Pe, the smaller the spread of the residence time. In order to calculate the residence time distributions from the axial mixing profiles, the mixing of leakageflow si nth echamber scanno tb eneglected .Whe na leakag e flowenter sa chamber ,an dmixe spoorl ywit hth econtent so fth e chamber, it can merge directly with a leakage flow which is leaving thatchamber .Thi swil l increaseth ePecle tnumber . 3.3 Materials andmethod s ofmeasuremen t The extruder used was a Cincinnati CM 45 conical counter-rotating,twin-screw ,extruder .Th escrew suse dwer eth e 1552typ e (seeTabl e3. 1an dFig .3.3 ).Th eextrude rwa sfe dwit h maize grits containing 25% moisture (wet weight) with a composition of 85% starch and 8% protein, on a dry basis. The sizedistributio n isgive n inTabl e3.2 .Th ediamete ro fth edi e was 20mm . The residence time distribution wasmeasure d by the coincidencetechniqu edescribe d byva n Zuilichem etal . (1988b) which measures thew Cu activity at the die outlet (Fig.3.4) . Half way along the screws, above the midpoints, two single detectorswer eplace deac hnorma lt oa screw .Bot hth esingl ean d thecoincidenc edetector sar edescribe d byva n Zuilichem etal . (1988a).Th etw o "midpoint"singl edetector shav ea symmetrica l sight angle, which leads to an erroneous Peclet number being measured,bu tthi serro ri ssmal l (0.4%)becaus eo fshieldin gb y 51 ^Axial denrcnce Spacer rings Proleclion against Clearance unvoluntary screw advance Fig. 3.3 Conical, counter-rotating, twin screw extruder (Cincinnatt). (vanZuiliche m et al., 1983) Table 3.1 Dimensions of screw set 1552 (van zuilichem et al., 1988) Length 100 m Calender gap 05 mm Flight gap 0-2 mm Maximum diameter 90 mm (one screw) Minimum diameter 45 mm (one screw) Volumetric compression ratio 2-1-1 Number of chambers per screw 42 Number of thread starts ~> Table3. 2Siz edistributio no fmaiz egrit suse d (Moistureconten t 13.1%,we t basis) (vanZuiliche m et al.,1988 ) Size (mm) Fraction (%) <0-50 0-8 0-50-0-60 1-7 0-60-0-70 9-2 0-70-0-85 26-5 0-85-1-00 54-2 > 1-00 7-5 Total 99-9 the screws and barrel. By van Zuilichem et al. (1988a) itwa s shown that measurements with the coincidence detector do not result indifferen t average residence times or Pecletnumbers . ThePecle tnumber swer edetermine d asdescribe db yTod d (1975), bycalculatin gth erati oo fth eresidenc etime sfo rwhic h16 %an d 64% of the tracer has passed the detector. This ratio was 52 input diagram response diagram Fig 3.4 Block diagram of the arrangement for measuring of residence timedistributiio n (vanZuiliche m et al., 1982) compared with that fora nRT Dmode lwhic h gives similar RTD's. ThePecle tnumbe rfo rth emode lwit hth esam erati owa sthe nuse d asth ePecle tnumbe ro fth eRT Dmeasurement .Th edisperse dplu g flowRT Dmodel ,describe d byLevenspie l (1972)an ddiscusse d in chapter 2wa suse d asRT Dmodel . 3.4 Simulation model In the simulation model each C-shaped chamber of the extruder screw-set is considered to be a CSTR. The contents of these chambersar echange db y leakage flows,represente d inth emode l asbackmixin gflows .A complet eextrude rsimulatio naccordin gt o theleakag eflo wequation so fJansse n (1978)shoul dcalculat eth e tetrahedron, calender, flight and side leakage flows, which requires a large number of assumptions.T ominimiz e the number of assumptions,th eblock-diagra m of Fig. 3.1 issimplifie d by summationo fth econtent so fcorrespondin gchamber so fdifferen t screws and summation of the contents of corresponding chambers ofdifferen tchannel so non escrew .Th esimplifie dmode lderive d fromth eblock-diagra m ofFig .3. 1ca nb efoun di nFig .3.5 .Th e leakage flowsbetwee n chambers inth e simplified model areth e 53 transport flow leakage flc w 2Qf 2QS C(j-D 20x C(j) C(j*i) „ at w Ij) W(>D Fig, 3.5 Leakage flowpatter nwhe n leakage flowsar e completely mixedwit h chamber contents (vanZuiliche m et al., 1988) summationo fth etetrahedro nleakag eflo w (Qt) , th esid eleakag e flow (Qs ), the flight leakage flow (Qf ), and the calender leakage flow (Qc ), Thus : W(j,t) = Qt+2Qs+2Qt+2Qc (2) inwhic hW(j,t )i sth eleakag eflo wleavin gchambe rj a ttim et . The CSTRs of themode l aredescribe d by a chambervolum eV(j) , achambe rtrace rconten t (C)an da locatio n (L)i nth esimulate d extruder. The location variable (L) can vary between L=0, corresponding to the first chamber in the feed zone of the extruder,an dL=l ,whic hcorrespond st oth e locationa tth edi e outlet.Th elocatio nvariabl eincrease swit htim et osimulat eth e transporto f screwchamber s inth eextruder .Th echambe rvolum e V(j) of a single CSTR from the simplified block-diagram of Fig 3.5,. is the total volume of the four chambers summed to form thissingl eCSTR .Th echambe rvolum e iscalculate d accordingt o equationso fJansse n (1978),bu tadapte d forth econica l aspect ofth e extruder used. The degree of fill,H(j ) isth e ratio of actual fillt omaximu m possible fill inth ej-t h chambero fth e extruder.Th emateria l feda ttim et= 0contribute st oth edegre e of fillo fal lchamber sdurin g itsresidenc e inth eextruder . The leakage flows in the compression zone of the extruder are simulated by two variables for each chamber of the simulation 54 model,a coefficient describing themagnitud e ofth eleakag e flowsA(j) ,an da mixin g coefficient G(j), which describesth e degreeo fmixin go fth eleakag e flowswit hth echambe rcontent . Theleakag e flowleavin gchambe rj a ttim et i scalculate das : W{j,t) =C[j,t)-A(j)-U-At-e (-MJ)-UAt) O) in whichU i sth erotationa l velocity ofth escrews . Without exponential termth etrace r content C(j,t) left afteron etur n ofth escre wi sdependen to fth etim e interval At,a sW(j,t )i s proportionalt oC(j,t) . Byth eintroductio no fth eexponentia l termthi sdependenc y disappears.I nth ecompressio nzon eo fth e modelth eleakag e flow leaving chamberj enter s chamber j-1.A fractionG ,o fthi sleakag eflo wmixe swit hth echambe rcontent s ofth e(j-l)t h chamberan dth eremainin g fraction (1-G)o fth e leakage flow, enters the (j-2)th chamber. This calculation procedurei srepeate dn time sunti lth eremainin gfractio n (1-G)n reaches the (j-n-l)th chamber with a zero leakage flow coefficient,wher ei tmixe scompletel ywit hth echambe rcontents . Themixin go fleakag eflow san dchambe rcontent si nth efee dzon e sectiono fth emode li sslightl ydifferen tfro mth emixin gmode l inth ecompressio n part.Her ea fractio nG o fth eleakag e flow leavingchambe rj ,mixe swit hth econten to fchambe r (j-1), whil e theremainin gfractio n (1-G)i sdirectl ytransporte dt oth efirs t chambero fth efee d zone.Afte reac htime-interva la ne wtrace r content (C)o fal lchamber si scalculated .Afte ra numbe ro fsuc h intervalscorrespondin gi ntota lt oon escre wrotation ,th ej-t h chamber becomesth e(j+l)t h chamber.Th edegre eo ffil lca nb e calculated fromth etota lamoun to ftrace rwhic hha sbee ninsid e the j-th chamber during the residence time distribution measurementas : H(j)= V±±l±fc(j.t)dt (4) V(j) Inwhic h C(j,t)i sth efractio no fth etrace rmateria lwhic hi s presenti nchambe rj a ttim et .Th esummatio no fC(j,t )i n time isproportiona lwit hth edegre eo ffil li nchambe rj .Th eminimu m 55 valueo fthi ssummatio ni s1.00 ,whic hoccur sfo ra zer oleakag e flowenterin g thischamber . Inthi s case thedegre e of fill in thischambe rca nb ecalculate db yth edegre eo ffil lo fth efirs t chamber under the feed hopper by the feed rate for a zero porosity, I, corrected for the differences in the chamber volumes,V(j ). Th eaverag eresidenc etim e (T) isdependen to nth e degreeo f fill inal l 21CSTR san dth e feed rate,a sin : j-ai EH{j) (5) T = JZI ru Theuppe rpar to feqn .5 give sth efille dvolum ei nth eextruder , whileth e lowerpar tgive sth evolumetri c feed rate ,Q . 3.5 Optimizationmetho d Simulation models with a large number of variables require an efficient search method in order to find an optimum or near-optimum solution with a modest computation time (Sisson, 1969). Sucha searc hmetho d canconsis to fa subdivisio n ofth e model into smaller sub-models, the optimization of these sub-models, and optimization of the complete model using the optimal solutions of the sub-model (Mesarovic, 1973). The proposed model here is subdivided into the feed zone, the transport zonean dth ecompressio n zone.Th ecomplet emode lha s seveninpu tvariables ,namely :th efee drat e (I),thre evariable s (S,,...S3)describin gth e leakageflo wprofil ei nth ecompressio n zone section of the model and three variables (S4...S6) describing theleakag e flowprofil e inth e feed zonesectio no f themode l (seeFig .3.6) .S. , andS 4ar eth evalue so fth eleakag e flow coefficient A, constant for all chambers in the zone described. The value of the leakage flow coefficient A in all otherchamber s inth emode l iszero .S 2an dS 5ar eth enumbe ro f chambers inth e zonesdescribed .s 3an dS 6ar eth evalue so fth e mixing coefficient G, constant for all chambers in the zones described. There are five output variables, 0, 05, for the model,whic har erespectively ,th egreates tdegre eo ffil lfoun d 56 feed rate :I leakage flow _ simulation filling degree of profile: S model jthCSTR:Hj Vo I CM F-curve E-cur ve E-curve F-curve centre detector centre detector die detector die detector 1 * 1 02 (x) (Pe) (Pe) Fig.3. 6Inpu tan doutpu tsimulatio nmode l (vanZuiliche me tal. , 1988) 0 fill degree I centre detector (R S2 start S, start —Ts4 V— leakage flow <5>- profile feeding search simulation optimal model Si T S3 start=0 "©4 yes & enlarge S3 unknown is interpolation possible of S, H e S,=1? i i^t\ if rr y\ 0t = Rt notation 0,S yes check accuracy 04= Rt? bounderies Fig.3. 7 Optimization procedure (vanZuiliche m et al.,1988 ) in the compression zone part of the model (O.,), the average residence times at two locations inth emode l corresponding to the locations of the two detectors (02,04), and the derived 57 Peclet-numbers (03,05). 02 and 03 are the "die outlet" output variables. Thevariable s 0.,....05correspon d to five variables R^.-.Rjmeasure d onth eactua lextruder .Tabl e 3.3 showswhic h inputvariable san doutpu tvariable sar edependen tan dwhic har e independent.Th efee dzon emode li suse dseparatel yb ysearchin g forvalue s ofth e input variables S4....S6a twhic h theoutpu t variables 02 and 03 are within the accuracy intervals ofth e measured variables R4 and Rj. For the optimization of the compressionzon emode li ti snecessar yt okno wthi ssolutio nfo r thefee dzon emodel .A rapi dstrategy ,describe di nFig .3.7 ,t o optimize the sub-models is based on feed-back, and uses some generalizationsan dassumptions . For each group of residence time distribution measurements a leakage flowprofil e forth efee d zonecoul d bedevelope d with onlyon evariabl e ,theleakag e flowcoefficien tS 4. Themaiz edoug hi nth ecompressio nzon ewa sassume dt ohav ezer o porosity,whic hmean stha tR^l.OO .Th eaverag eresidenc etim eO z and themaximu m degree of fill inth ecompressio n zone 0,ar e dependenta sca nb esee n fromeg .5 .Becaus eo fthi sdependenc e andth eassumption ,tha tR,= l, th enumbe ro fleakin gchamber sS 2 inth ecompressio n zone,whic h isa nintege rvariable ,ha son e value, forwhic h (R.,-0.,)i sminimal ,whil e 02=R2.Th enumbe ro f chambers S2ca nb eestimate d from a function F(v)derive d from eq. 5, which calculates the average residence time in the compression zonefro mth enumbe ro ffully-fille d chambers.Whe n S2 isknown ,a valu e forS 3 ischosen ,an dth evalu e ofs ,fo r which0 2=R2i ssearche dfor . A constant leakage flowcoefficien t (A)fo ral lchamber so fth e compression zone results ina degre e offil l lesstha n 1.00i n the chamber nextt oth edie .A constant degree offil l of1.0 0 for all chambers in the compression zone is possible, with anotheroptimizin gprocedure ,les sefficien ti ncalculatio ntime . This slow procedure only gives a small improvement in the accuracyo fth ePecle tnumbe ro fth emodel . 58 Table 3.3 Influence of input variables S and I on output variables (O-variables) (vanZuiliche m et al., 1988) O, O,andO, ()4 and O, I St S25 , s,s , s. x= Dependent o= Independent 3.6 Results The RTD measurements are labelled with the letter combination •exo' and a number, which indicates the sequence of the r. p.m.= 80 die= 2 0 mm 0.75 E-curves 0.50 0.25- 32 time (s. 32 time(s.) Fig 3.8 Extinction curvesmeasure d at the die and the midpoint positions (vanZuiliche m etal. ,1982 ) measurements.A nexampl eo fth emeasure d RTD'si sgive n inFig . 3.8. Combination of average residence times and Peclet numbers measured at the centre detector, could not be simulated by assuming the total mixing of leakage flows and chamber content (S6=0). The simulated average residence time and/or the Peclet number tended to be too high. A leakage flow profile was developed suitable for simulating the centre detector measurements 'exo(2)', 'exo(3)' and 'exo(4)'. In this leakage flowprofile ,al lleakag eflow sfro mchamber sfou ran dfiv emov e directlyt ochambe rone .Tabl e3. 4show sth edifference sbetwee n 59 themeasure d (R-variables)an dsimulate d (O-variables)RTD' st o besmall .Th emea nresidenc etim e (R4)i sinversel yproportiona l to the rotational speed of the screws.Fro m these measurements iti sno tclea rwhic hmai nvariabl eaffect sth evelocit yprofil e inth e feed zone,an dhenc eth ePecle tnumbe r (Rj). S, S2=5 chamber s I =0.22 Fig.3. 9Simulation sar evali d foroptima lvalue so fth eleakag e flowcoefficien t 8,an dth echambe rmixin gcoefficien t S3i nth e hatched areas of thecurv e forexperimen t 'exo2 ' The leakage flow profile variables in the feed zone were quantified so as to make the F-curve entering the compression zone match the measurements of the midpoint detector. The simulationo fth efee dzon ewa sno taffecte db yleakag eflow si n thecompressio nzone ,a sca nb esee nfro mTabl e3.3 .I nFig .3.9 , 3.10 and 3.11 all possible combinations of the coefficients related to the magnitude of leakage flow SI and the extent of chambermixin g S3ar egive n forthre emeasurements .Som evalue s from Fig. 3.9, 3.10 and 3.11 are shown inTabl e 3.5. The four lines in each figure are: the combinations of S,an d S3 which result in the simulated average residence time at the die detector (02=R-,), th euppe ran dlowe rboundarie so faccurac y for themaxima l degreeo f fill inth ecompressio n zone (O,)an d for the Peclet number (05). The hatched areas give all possible 60 D O4 = R(, S2 =4 chambers ° 0, =1.1 1 =0.2 0 A O1 =0.9 x 05 (Pe)=23 Fig. 3.10 Simulations are valid for optimal values of leakage flowcoefficien t S,an dth echambe rmixin g coefficientS 3i nth e hatched areaso f thecurv e forexperimen t 'exo3 1 combinations of S, and S3 if the number of chambers in the compressionsectio ni sinfinitel yvariabl einstea do fa ninteger . Inth ehatche d fieldsth e correctvalue s of the Peclet numbers can only be simulated within a limited range of the chamber mixingcoefficien tS 3.I nfou rou to fsi xmeasurements ,S 3range d from0. 0t o0. 1 (Fig.3.9 ,3.10 ).Th eothe rtw omeasurement sgav e a value of S3 ranging from 0.0 to 0.7 (Fig. 3.11). These intervalsar egreatl y influencedb y smallchange si nth ePecle t number. On the lines forwhic h O^Rj, where the mean residence timei skep ta ta constan tvalue ,th echambe rmixin gcoefficien t hasa suprisingl ysmal leffec to nth ePecle tnumber .A nincreas e inS 3fro m0. 0t o0. 4resulte d ina decreas ei nth ePecle tnumbe r from 29 to 25. Therefore such a determination of the chamber mixingcoefficien trequire sa ver yaccurat emeasuremen to fPecle t numbers.Whe nsimulation sar emad ewithou tbackmixin gi nth efee d zone,an dth ecompressio n zonei sdimensione d asgive n inTabl e 3.5,Pecle tnumber sincrease ,a sca nb esee ni nTabl e3.6 .Thes e Peclet numbers are large in comparison with those for other extrudertype shavin gcomparabl ehold-u pvolumes ,a sca nb esee n 61 a=R4 S2=5 chambers 0-,= 1.1 I =0.20 Oi = 0.9 0s(Pe) =2 3 Os(Pe') =29 Fig.3.1 1Simulation sar evali dfo roptima lvalue so fth eleakag e flowcoefficien tS 1an d thechambe rmixin gcoefficien t S3i nth e hatched areaso f thecurv e forexperimen t 'exo4 ' Table 3.4 (vanZuiliche m et al., 1988) Measurement of Mean Residence Time (R2)an d Peclet Number (/?,),Simulatio n Results for Mean Residence Time {02) and Peclet Number (O,), and Simulation Input Variable 54, of Mid-point Detector, asInfluence d by the Rotational Screw Speed U, Employinga Constant Feed-rate (/) Experiment U(s~') I R2(s) O, (s) Rj o, SA exo 2 0-67 0-22 18 18 7 7-4 0-41 exo 3 1-00 0-20 14 14 5 5-4 0-53 exo 4 1-33 0-20 9 9 8 7-4 0-41 Table 3.6 Peclet Number Measured at the Die Detector and Peclet Number asOutpu t Variable ofa Model Without Axial Mixingi nth e Feed Zone Experiment exo2 exo3 exo4 Peclet number measured, R^ 34 26 26 Peclet number simulated, 05 55 61 49 inTabl e3.7 .Th enearl ydouble dPecle tnumber si nTabl e3. 6giv e 62 Table 3.5 (vanZuiliche m et al., 1988) Leakage Flow Profile (5-Variables) in the Compression Zone Necessary to Simulate the Mean ResidenceTim e (R4)an d Peclet Number (/?5)Measure d atth e 'Die'Detector . The Calculated Peclet Number (Os)i sGive nfo r Three Valueso f theChambe r Mixing Coeffi cient 5, Experiment exo2 exo3 exo 4 Screw speed, U(s~') 0-67 100 1-33 Feed rate ratio, / 0-22 0-20 0-20 Average residence time measured, R4 (s) 45 31 23 Peclet number measured, R5 34 26 26 Accuracy rangeo f Rs 31-37 23-29 23-29 Fullyfille d chambers in compression 5 4 5 zoneo f simulation model, S2 S= 00 0 (leakageflow coefficient simulated Peclet number) 5, 0-59 0-61 0-62 o< 31 23 29 53= 0-0 5 (leakageflo w coefficient simulated Peclet number) •S. 0-58 0-58 0-59 0, 31 23 28 S,= 0-4 0 (leakageflo w coefficient simulated Peclet number) 5, " " 0-45 o, " " 25 "Ofcanno t be simulated within the accuracy range of Ry. information on the axial mixing at temperatures above the selected temperatures of the feed zone (60-100°C), ignoring mixing effects at temperatures below 100°C. In Fig. 3.12 the F-curves foraxia lmixin gabov eth eselecte dtemperature so fth e feed zone,an d aboveo ra t 150°C,ar egiven .Du et oth e absence ofmixin g inth e feed zonepar to fth emodel ,thes ecurve sten d to become more uniform. This uniformity is due to the equal mixing profiles of these threemeasurement s in the compression zone. The coefficient describing the magnitude of the leakage flowis ,assumin ga constan tvalu efo rth emixin gcoefficien tS 3, independent ofscre wvelocit y andth enumbe ro fchamber s inth e compression zone (see Table 3.5). Variations in the Peclet numbers measured at the die are largely due to variations in 63 Table 3.7 (vanZuiliche m et al., 1988) Peclet Numbers for Three Extruder Types Extruder type Peclet number r Feed rate (gs ') Hold-up volume (s) Single screw" 10-20 60 11 5x 10"4 Co-rotating, twin-screw'' 10-30 55 10 4x 10~4 Counter-rotating, twin-screw 26-34 30 25 6x 10"4 with axial mixing in the feed zone Counter-rotating, twin-screw 55-61 20 25 4x 1()-4 without axial mixing in the feed zone "Calculated from VanZuiliche m etal. (198S F(t) 1.0 -I screw type:155 2 A temp,s150*C D temps 100'C 0.75- o temp> 20 C 0.5 0.25- 25 SO ,. 75 100, , time xscre w speedI- ) Fig.3.1 2Calculate dF-curve sfo rth edi elocatio nfo rexperimen t •ex o 41 in which only the axial mixing in sections with a temperature aboveth eindicate d temperature is simulated leakage flows inth e feed zone. 64 3.7 Conclusions Theassumptio no fa constan tleakag eflo wcoefficien ti nfou ro r five chambers of the compression zonegive s apossibl e leakage flowprofil eo nwhic ha mode lsimulatio no fa workin gtwin-scre w extruder can be based, within the limits of accuracy of the measurements. In the model the effect of the coefficient describing themixin g of chamber contents and leakage flowso n the Peclet number proved to be minor. The coefficient for the mixing of chamber contents and leakage flows could not be determined,du et oth erestricte dnumbe ro fmeasurement san dth e small effect of this coefficient on the Peclet number. The averageresidenc etim ei nth efee dzon ei sinversel yproportiona l toth erotationa l speedo fth escrews .Whe na counter-rotating , twin-screw extruder isuse d asa reacto rfo rstarc h conversions occurring mainly in the compression zone of the extruder, the Pecletnumber smeasure dhav et ob ecorrecte dfo rth eaxia lmixin g in the feed zone, as this mixing is not relevant to the reactions.Thu sa measure d Pecletnumbe ro f2 6t o 34become sa n effective Pecletnumbe ro f 49t o61 . 65 References Albers, F., (1976). Beitrag zur Dimensionierung der Schnecfcen von einen Gegenlaufig kammenden Doppelschneckenextrudern. DissertationMontanuniversita t Leoben. Altomare,R.E. ,Ghossi ,P . (1986).A nanalysi so fresidenc etim e distribution patterns in a twin screw cooking extruder. Biotechnology Progress. 2 157-163. Ben-Gera,I. ,Rokey ,G.J. ,Smith ,O.B.(1983) .Extrusio n cooking ofgrain sfo rethano lproduction .J .Foo dEng. .2 ,177-187 . Bounie,D. ,Cheftel ,J.C . (1986).Th eus eo fTracer san dThenno - sensitive Molecules toAsses s Flow Behaviour and Severity ofHea tProcessin g DuringExtrusion-Cooking . In:Past aan d Extrusion Cooked Foods, ed. C. Mercier, C. Cantarelli. ElsevierApplie d Science Publishers,London ,148-158 . Bush,M. ,Milthorpe ,R.F. , Tanner,R.I . (1984). Finite element andboundar yelemen tmethod sfo rextrusio ncomputations .J . Non-Newtonia n FluidMech. . 16,37-51 . Chouvel, H., Chay, P.B., Cheftel, J.C. (1983). Enzymatic hydrolysis of starch and cereal flours at intermediate moisture contents in a continuous extrusion reactor. Lebensm.-Wiss.u-Technol.. 16,346-353 . Janssen, L.P.B.M. (1978). Twin screw Extrusion. Elsevier Scientific Publications,Amsterdam . Launay,B. ,Lisch ,J.M . (1983).Twi nscre wextrusio n cookingo f starches: flow behaviour of starch pastes expansion and mechanical properties of extrudates. J. Food Eng.. 2, 157-175. Levenspiel,0 . (1972). Chemical Reaction Engineering. 2nd Ed.. 66 JohnWile y &Sons ,Ne wYork . Linko, P., Linko, Y., Olkku, J. (1983). Extrusion cooking and bioconversions.J . Food Ena.. 2,243-257 . Mercier, C, Charboniere, R., Grebaut, J., de la Gueriviere, J.F. (1980). Formation of amylose-lipid complexes by twin screwextrusio ncookin go fmanio c starch.Cerea lChem. .5 7 (1)4-9 . Mesarovic,M.D . (1973). Onvertica l decomposition and modeling of large scale systems. In: Decomposition of large scale problems. ed. D.M. Himmelblau. North-Holland Publishing Company, Amsterdam, (American Elsevier Publishing Co., Inc.),323-340 . Olkku, J., Antila, J., Heikkinen, J., Linko, P. (1980). Residence time in a twin screw extruder. In: Food Proces EngineeringVol .1 :Foo dprocessin g Systems,ed .P .Linko , V. Malkki, J. Olkku, J. Larinkari. Applied Science Publishers,London ,791-794 . Sisson,R.L . (1969).Simulation :uses .Progres s inOptimizatio n Research, Vol. iii,ed . S. Arinofsky. John Wiley &Sons , New York, 17-70. Savolainen, A., Karling. R, (1985). The influence of screw geometry on the extrusion of soft PVC. J. APPI. Polvm. Sci.. 30,2095-2104 . Todd, D. (1975). Residence time distribution in twin screw extruders. Polym.Eng .Sc. . 15,437-442 . van Zuilichem, D.J., Bruin, S., Janssen, L.P.B.M., Stolp, W. (1980). Single screw extrusion of starch and protein rich materials. In: Food Proces Engineering Vol. 1: Food processing Systems, ed. P. Linko,V . Malkki,J . Olkku,J . 67 Larinkari.Applie d Science Publishers,London ,745-756 . van Zuilichem, D.J., Tempel, W.J., Stolp, W., van't Riet, K. (1985). Production of high boiled sugar confectionary by extrusion cooking ofsucrose .J . Food Ena.. 4, 37-51. van Zuilichem, D.J., Stolp, W., Janssen, L.P.B.M. (1983). Engineering aspects of single-and twin screw extrusion cooking ofbiopolymers .J .Foo d Ena.. 2,157-175 . van Zuilichem, D.J., Jager, T., Stolp, W., de Swart, 3.G. (1988a). Residenc etim edistribution si nextrusion-cooking . PartI :coincidenc edetection .J . Food Ena..7 ,147-158 . van Zuilichem, D.J., Jager, T., Stolp, W., de Swart, J.G. (1988b). Residenc etim edistribution si nextrusion-cooking . Part II:Single-scre w extruder processing maize and soya. J. Food Ena..7 ,197-210 . van Zuilichem, D.J., Stolp,W. ,Jager ,T. , (1982). Seminarkoc h undextrudiertechnieken ,Lectur eat :Zentralfachschul e der DeutschenSusswarenwirtschaft.Solinaen . 68 Notation A(j) Leakage flowcoefficien to fCST Rj [- C(j,t) Fractiono ftrace rconten t inCST Rj a ttim e p- 2 1 De Axial dispersion coefficient [m s" exo Series of residence time distribution measurements [- E Exitag edistributio n [s"1 F Cummulativeexi tag edistributio n [- G(j) Chambermixin g coefficient ofchambe r j [- H(j) Degreeo f fillo f jt hCST R [- I Ratioo f feed ratet omaximu m feed rate [- j CSTRnumbe r [- 1 Extruder length [m L Dimensionlessextrude r length [- 0,, Greatest filldegre e incompressio n zonemode l [- 02 Average residence time in reference to midpoint location inmode l [s 03 Peclet number in reference to midpoint location in model [- 04 Averageresidenc etim ei nreferenc et odi elocatio ni n model [s 05 Peclet number inreferenc e to die location inmode l [- Pe Pecletnumbe r [- Qc Dimensionless calender leakage flow [- Qf Dimensionless flight leakage flow [- Qs Dimensionless sideleakag e flow [- Qt Dimensionless terahedron leakage flow [- R, Degreeo f fill inth ecompressio n zone [- Rj Average residence time in reference to midpoint detectorpositio n [s R3 Peclet number in in reference to midpoiunt detector position [- R4 Average residence time in reference to die detector 69 position Rj Peclet number in reference to die detector position ["] S, Value of A in compression section of the model [-] 52 Number of CSTRs in the compression section of the model for which A>0 [-] 53 Chamber mixing coefficient G in the compression section of the model [-] 54 Value of A in the feed section of the model [-] 55 Number of CSTRs in the feed section of the model for which A>0 [-] 56 Chamber mixing coefficientG in the feed zone [-] t Time [s] U Rotational speed of the screws [s"1] v Mean axial velocity [is*1] V(j) Volume of jth CSTR [m3] W(j,t) Leakage flow out of CSTR j at time t [-] T Average residence time [s] At Time interval [s] 70 CHAPTER4 CONFECTIONERY AND EXTRUSION COOKING TECHNOLOGY ABSTRACT The capabilities of cooking extruder equipment in confectionery production is investigated. Sugar crystals are converted to a seedingfree clear melt using a single screw extruder. The production ofhigh-boile d sugarconfectioner y bymean so fa twin - screw extrusion-cooker (e-c) has been investigated and is described.Th eproduct sar ecrysta l freean duncoloured .Th ewate r content is low, even though no evacuation systems were used to removeexces swater .Th ehea ttransfe ri nth ee- cma yb ecalculate d by a stepwise procedure over the length of the extruder. The results indicatetha tth ee- c isa neffectiv ehea texchanger .Th e model developed enables the time needed for dissolution of the crystals to be estimated. It isconclude d that use of an e-c for thispurpos e isbot hpossibl e andattractive . THIS CHAPTER ISBASE DUPO NTH EPUBLICATIONS : van Zuilichem, D.J., Alblas, B., Reinders, P.M. and Stolp, (1983a). A comparative study of the operational characteristics of single and twin screw extrusion of biopolymers. InTherma l Processing and Quality ofFoods . ed. P. Zeuthen, J.C. Cheftel, C. Eriksson, M. Jul, H. Leniger, P. Linko, G. Varela, G. Vos, Elsevier Applied Science Publishers,London\Ne w York,1984 . vanZuilichem , D.J., Tempel,W.J. ,Stolp .W an dK .va n 'tRie t (1985). Production of high boiled sugar confectionery by extrusion cooking of sucrose:liquid glucose mixtures. J.FoodEnqnq .4 ,37-51 . Sugar Confectionery Manufacture, pp 311-330. Ed. Blackie/van NostrandREINHOLD/Ne wYor k1990 . 71 4.1 Introduction Among other applications, cooking extruders can be used to produce confectionery. As such extrusion cooking technology is nota ne wtechnology ,sinc epast aproduct sar ecommonl yextrude d and shaped since the thirties and sausage is produced for a hundred years on extruder-like equipment. The real novelty is that "cooking" extruder equipment is offered to the food industry, which means that a heating process is introduced in whicha controlle d food-chemical or-biochemica l reactiontake s place. The chemical polymer industry started up the use of extruder equipmentsinc eth efourtie swit hth es ocalle dmelting-extruder s basedo ndesig nconcept so fmixe rcompounders .Th efoo dindustr y discovered the cooking extruder since the fifties and used the equipment in the beginning for the production of cooked and expandedsnack so nbasi so fcorngrit san dothe rcereals .Mos to f theequipmen tuse dwer es ocalle dsingl escre wextrude rconcepts , whilst twin screw extruder equipment dates back since the sixties. A technological reason for the use of twin screw extruders is the need for equipment capable to handle high viscosities asthe yar eknow n inth econfectioner y branch ort o processlo wan dhig hviscositie si na distinc trecip ea tth esam e time in one piece of equipment. Although most of the confectionery-articlesar euniqu ei nit sow ni ti sclea rtha tth e basiccomponent sar esuga r(sucrose ), starc hsyrup s(treacle )an d starcho r flours (seeTabl e4.1) . 72 Table 4.1.Generalised composition of some confectioneries (van Zuilichem et al., 1985 Materials Liquorices Soft liquorice Cleargum s Winegum s Products sucrose X X X X syrups X X X ' X starch X X X X wheat flour X block liquorice X X caramel X NH4C1 X X additives X gelatines X X water X X Atth esam etim ewate r isalway spar to fth erecip ean dthi si s just the reason for the selection of extruder equipment as an innovativetoo li nthi sarea ,sinc edryin gcost sar edeterminin g fora grea tdea lth epric eo fth egoods .Cookin gextruder soffe r a possibility to process confectionery-goods at lower process moisturecondition scompare dwit hconventiona lprocessing ,whic h will leadt oattractiv esavings . 4.2 Problem description At the conventional confectionery-processes water is used to dissolve sucrose crystals atprese t temperatures,describe d by Honig (1953).Th eproduct sar ecaste di ncornstarc han dth ewate r isremove dt ocondition srequire dfo ra lon gshel flife .Fo rthi s operation careful and time consuming drying is necessary at temperaturesbetwee n 45-60°C.Th elo wdiffusivit y constants for water insugarlik ematerial sdetermin eth erat eo fdrying . Iti sclea rtha tthes emethod sar eenerg yconsuming .I twoul db e more attractive to develop a cooking method atwhic h thewate r 73 S3ESSS^>(!) ^^S3^s>e S33^HS3^0 Fig 4.1 Existing food extruder models (van Zuilichem et al., 1976) percentage isa sclos ea spossibl et oth emoistur e level ofth e endproduc t (seeTabl e4.2 ).Suc ha wis hma yb efullfille db yth e selection of an appropriate extruder for this purpose. The available extruder-designs can be divided in single and twin screwextruder s (seeFig .4.1) . 74 Table 4.2.Typica l shelflife/moistur e forsom e confectioneries Product Moisture Shelf life ( %) (inmonths ) chocolate bakery 1 6/8 milk 2 6/8 pure 1 6/8 toffees 6 a 7 8 hard boilings 2 a 3 12 agargel s 24 6/8 cream 12 a 13 6 whipped cream 6 8 fondant 12 10 winegum s 12 12 softgum s 22 6/8 fudge 7 5/6 gelatine gels 22 6/8 marshmallows 12 4 pastilles1 18 4 pastilles2 11 18 pectinegel s 22 6/8 tablets 1 18 turkish delight 20 5 liquorice 8/18 8/9 The twin screw extruder designs consist of corotating and counterrotatingdesign san dclosel yintermeshing ,s ocalle dself - wiping extruders.Th emos t important difference between single screwextruder san dclosel yintermeshin gtwi nscre wextruder si s that single screw extruders are so called "open channel" equipment, with no restriction prohibiting backflow of the material,wherea s the closely intermeshing twins consists of a series of C-shaped chambers,whic h pump themateria l from feed portt oth edi e (seeFig .4.2) . 75 Fig 4.2 C-shaped chamber oftwi nscre wextrude r (Janssen,1978 ) 4.3 Statusqu oo f realised extrusion cooking -processes Iti spossibl et oreplac eth econventiona lcontinuou scookin go f confectionery at high solids content by twin screw extrusion cooking,optionall y followedb ymouldin gan ddryin gi nstarc ho r other moulds. This may not be very spectacular as the post extruder process is still conventional but the aim is to save energyb ypreparin gth emel ta tlowe rmoistur econtents .Anothe r possibility isth eus e of extrusion cooking in the manufacture of gum confectionery cooking extrusion. Here forming-extrusion isalread y being used to produce liquorice,gums , chewing gum, caramels and marsh-mallows. These products have been well investigated by extruder manufacturers, mostly in combination with starch-producers. For example Staley haspatente d the use of the cold water swelling Miragel 463 for confectionery extrusion in US-patent no. 4567055. The patent describes the extrusiono fmaltos esyrup ,fructos esyru pan dMirage l46 3a t12 1 °C. Regular cold swelling starches are claimed not to lead to translucentproducts .I ti sobviou stha tproduc tdevelopmen t in this area is very well possible and necessary, but a gopd knowledgeo fth ebehaviou ro fth ebasi ccomponent s inextruder s should beunderstood . 76 4.4 Extrusion of starch The commercial attractive starch sources are: maize, potato, rice,manioc ,whea to rsago .I ti swel lknow ntha tstarche shav e a gelatinisation temperature between 60 and 80 'C. (See Table 4.3). Table 4.3 Gelatinisation of starches, (van Zuilichem at al., 1976) Basic starches Gelatinisationtemperatur e(°C ) Corn starch 62-72 Potato starch 56-66 Wheat starch 52-63 Rice starch 61-78 Tapioca 58-70 Sago 60-72 Thecompositio no fstarche sconsist si nmos tcase so famylos ean d amylopectini na rati o1 : 4 .A nexceptio ni sWaxy-maiz ea sshow n inth etabl e4.4 . Table 4.4 Starch composition (vanZuiliche m et al.,1976 ) Ratio (%) Amylose Amylopectin Corn starch 24-26 74-76 Potato starch 22-23 Approx.7 7 Tapioca 19-20 Approx.8 0 Sago 20-26 74-80 Waxymaiz e <1 >99 As shown in Figure 4.3 amylose consists of a chain of a-1,4 glucanelements .O nth eothe rhand ,amylopecti nshow sa branche d structure of a-l,4 glucan and a-1,6 elements. The aim of an extruder, when processing these polysaccharides, is to damage cellwall so fth estarc hparticle san dt oseperat ethe mmor eo r lessb y breaking or reducing the size of themolecula rchains . Forthi spurpos eth eproduc tapplicatio n isver y important.Fo r 77 H OH H OH H OH H OH H OH (n) Amylose.a 14 glucan (n) CH,OH CH2OH CH,OH H OH H OH H OH H OH _Tnl Amylopectin/* 1.4 and a 1.6 glucan Fig. 4.3 Molecular structure of amylose and amylopectin (Van Zuilichem etal. , 1982b) human consumption iti snecessary , forreason s ofdigestibilt y togelatiniz eth estarc halmos tcompletely , foranima l foodstuff this isno tth ecas ea sth eabilit y ofanimal s todiges t partly gelatinized starches,du et oenzym eactivity ,i smuc hbetter .Fo r confectionery products, however, thestarc h must be completely cooked. 4.5 Extrusion ofdr y sucrose-crystals Sucrose crystals areon eo fth emai n recipe-components used in the mixtures inth econfectionery-industry . Iti snecessar y to study themeltin g behaviour ofsucros e ina nextruder , as this CH2OH CH2OH Fig. 4.4Molecula r structure of sucrose (vanZuiliche m et al., 1982b) 78 stage offers the potential to save a considerable amount of energyuse d forth ecookin gstep .Sucros ei scompose do fglucos e and fructoseunits .Th e structural formula of sucrose isgive n inFigur e4.4 . Forapplicatio n inth econfectioner y industry sucroseshoul db e presenta sa clea rmel ta thig htemperature s (>16 0°C )tha tca n easily be cast in preformed maize-starch moulds, and that can easily bemixe dwit hothe rcomponents . The extruded productmus tmee t these requirements.A t the same time it is necessary to know the possible degree of caramellization caused by the high temperature levels in extruders. Furthermore one should also be aware of the possibilityo finversio no fsucrose ,whic hmean schemicall ytha t the dissaccharide is split up into the two monosaccharide components,glucos ean dfructose .Finall yon ewil lb eintereste d inth eviscosit y developed ina nextruder ,whic hwil l determine directly the capacity of the extruder equipment and the energy consumption. 1 through 6 thermocouples in the barrel thermocouplelance 7 through 12 thermocouples support and lock inside the screw Fig. 4.5 Thermocouple arrangement forscre w andbarre l 79 The extruder used by van Zuilichem et al. in 1983 was a 50m m singlescre wextrude roriginatin gfro mth eBattenfel ddesign .Th e screwsuse dwer eo fth ecompressin gtype ,wit hconstan tpitc han d compression ratio'so f 3 :1 , 2:1 and 1.15 :1 .Temperature s ofbarre lan dscre wwer emeasure dwit hthermocouples .Th escrew s usedwer ehollow .A thermocouple-lance ,supporte di nth enoseti p of the screw and at the rear of the extruder, were used to measureth etemperatur eprofil ei nth escre wsid e (SeeFig .4.5 ) =F ~r/////// nu thermocouple Fig. 4.6 Screwan dbarre l temperature profile forsuga rmeltin g (vanZuiliche m et al.,1982a ) The sugar was supplied to the extruder by a simple vibratory meteringdevice .At firs ta goo dtemperatur edistributio nha dt o be found in order to create a sufficient melting of the dry crystals.Th enecessar ytemperatur eprofile scoul db econtrolle d verywel lan da nexampl ei sshow ni nFi g4.6 .I tca nb esee ntha t thesuga ri ssheare do nth escre wsid eresultin g ina quit ehig h "lance"temperatur eprofil ei nconnectio nwit ha nextrude rbarre l profile going from ± 130 °C above ± 170 °C,t o 150 °C at the barrel outlet. This temperature profile was measured at 160 80 r.p.m., with a screw compression ratio of 2 : 1 and length/diameter ofth escrew ,1/ d= 12.Th edi euse dha d asli t of 8 mm. With this setup the process is unstable and the throughput is fluctuating. Amaximu mthroughpu to fmolte nsucros ei splotte di nFig .4. 7fo r screwcompressio nratio' s3 :1 ,2 :1 an d1.1 5 :1 a tth erang e 160 - 200 r.p.m. A maximum throughput of the melt of over 90 kg/hr proved to be possible at 200 r.p.m., using a 3 : 1 compression screw. The molten sugar exhibited an amorphous structurean dha dlos tit scrystallin enatur ewhe ni twa ssample d directly at the extruder die. After some time of storage the samples showed a slight recrystallisation due to very small seedingcrystals . throughput (x 10 kg/s) ,oc.r.= 3 22 c.r.=2 specific power consumption(kWh/kg) 16.5 0.3- .oc.r.= 1.1 5 0.2 5.5 0.1 V-"i 1 1— 160 180 200 r.p.m. 160 180 200 r.p.m. Fig.4. 7Maximu mthroughpu tfo r Fig. 4.8 Specific power sucrose (van Zuilichem et al., consumption for sucrose (van 1982b) Zuilichem etal. ,1982b ) Of interest isth eaverag especifi cpowe rconsume d necessary in theabov ementione dtrials .Thi s isshow n inFig .4. 8 forthre e differentscre wcompressio nratios .At compressio nratio so f3: 1 81 and 2:1 power consumption was low « 0,1 kWh/kg. The lower compressionrati oo f1,1 5: 1 neede da specifi cpowe rconsumptio n ofa tleas t0,1 5 kWh/kg.Whe nmoto rpowe ronl yi sconsidered, the figuresar emuc hlower ,i nth eregio no f (0,04- 0,05 )kWh/k gfo r screws with compression ratio's of 2 : 1 and 3 : 1. The difference in power is provided by the electric heaters that intermittently heated the barrel and the heat losses to the environment (SeeFig .4.9) . Themel tmus t beproduce d atth ehighe r r.p.m.value s used, in ordert oavoi dscalin gi nth edi ean drecrystallizatio no fmolte n sugar.Th ereducin gabilit yo fth esuga rafte rthi sdr yextrusio n was measured by the method of Nierle and Tegge and no considerable inversioncoul db edetermine da tth ecircumstance s mentioned. Some colourmeasurement s were also performed, using I—coolin g— I heaters— i barrel ////// //rm> screw - barrel | cooling 1 Theaters-] -^mp temperature(°C) mixing torpedo 210 170- feed:saccharose- spra y dried glucose syrup ratio: 1:1 die 4mm 130 c.r.= 3 ex.:1. 7 die diam.= 4 mm 1/0:12 r.p.m. =80 mat'I:maize grits l/d =16 -•20% sugar 90 50 8 9 Fig. 4.9 Temperature profiles Fig. 4.10 Temperature prof11* for the extrusion of a maize- for dry saccharose-glucose sugar mix (van Zuilichem et mixtures (van zuilichem et al., 1981) al., 1990) the standardized Icumsa method nr. 4, according to De Whalley 82 (1964). Clear melts without any recrystallisation are produced when a steep temperature profile is applied at a cooled screw providedwit ha s ocalle dmixin gtorpedo , (seeFig .4.10 ).Colou r ratingsar ebelo wIcums a50 .I nthi swa ya singl escre wextrude r can be used as a relatively economical melting tube for dry sucrosecrystals . 4.6 Extrusion of sucrose-starch mixtures In the confectionery industry many products, based on starchy materialslik erice ,whea tan dsometime scorn ,ar euse da scand y bar fillers.I n Fig. 4.10 atemperatur e profile isgive n fora single screw extruder processing acandyba r filler out of corn andsucros ei na rati o4:1 . Sucha produc tca nstil lb ehandle d by a single screw extruder without difficulties. If the percentageo fsucros erequired ,i shighe rthe n20% ,a twi nscre w extruderi sa goo dmachin efo rsuc ha product ,becaus eth eforce d flowneede d forsuc ha hig hviscosit ymixtur ei seasil yprovide d by the C-shaped chambers.A single screw extruder,however , is inthi scas ealread yproblematic . 4.6.1 Extrusion of sucrose-syrup mixtures Theextrusio ncookin go fthi styp eo fmixtur eform sth ebasi so f confectionery-products.Du et oth eunattractiv elo wviscosit yo f thestarch-syrup sa singl escre wextrude rwil lb eunsuitabl efo r such a job,bu t a closely intermeshingtwi n screw extruder can easilyb eused . In the conventional method of producing the melt from which hardboiled sweets are formed, water is added to the other ingredients ina cookin gvessel .Th eadvantag eo fa nextrusion - cooker (e-c) is that dissolution can be performed without the additiono fan ywater .Thu sb yusin ga ne- cth etime -an denergy - consuming step of evaporation of the excess water can be obviated.A twin-scre w e-c isabl et operfor mthi s function,a s itha sa larg ehea texchang ecapacit ycombine dwit ha favourabl e residence-timepattern .Th epositiv eforwar dtranspor tmechanis m together with the lateral mixing action of the screw result in a plugflow-like pattern combined with radial (interchamber) 83 mixing.Thi sresult si na well-controlled ,limite dresidence-tim e whilsta tth e sametim ea well-mixe d material isprocessed . Evidencefo rthi splugflo wcharacte ri sgive nb yJage r (1983)wh o defined S asth e ratioo f thetim e taken for 16%an d 84%o f an injectedtrace rt opas sa referenc epoin ta tth een do fth ee-c . Jagerfoun dtha ti na twin-scre we- cS tend st ob ei nth eregio n of 0.3-0.5, which isa n attractive value.Jage r also explained that the higher theviscosit y of thematerial ,th egreate r the variance of residence-time distribution, due to back-flow. Because the material used in the confectionery production described here has a relatively low viscosity at e-c temper atures,S coul db eeve nhighe r inthi scase ,whic h implieseve n more favourable conditions inth ee-c . Thesucros ecrystal sdissolv ei nth ewate rpresen ti nth eglucos e syrup. Inthi sproces s the time and temperature (andpressure , in combination with the temperature), determine whether the crystals will disappear and whether undesirable colour development will occur. A schematic representation is given below: Time ) f Crystal dissolution Temperature ) *» \ Off-colour development Pressure J I Water evaporation Theresidence-tim emus tno tb eles stha nth eminimu mtim eneede d to dissolve the crystals completely at the prevailing temperature. On the other hand, the temperature must be low enought oavoi d the formation ofoff-coloure dproducts . There is no melting of sugar crystals involved since working temperatures are in the region of 130-150 °C and the melting temperature of sucrose is 186 °C. Shear effects can also be neglected since Brinkman numbers are below 0.001, while Graets numbers are above 105,fo r the working points of the extruder used. Coloration of sugars or off-color development at high temperatures isdu e to caramelization. Unfortunately there are noexac tquantitativ edat aavailabl eo nth erelationshi pbetwee n 84 sugartemperatur ean ddegre eo fcoloration .Th ereaso n forthi s isthat ,onc eth e initiating reactionha sstarted ,a comple xo f condensation, fragmentation and dehydration reactions follows. The starting reaction isknown ,however .Shallenberge r &Birc h (1975) point out that colour formation begins with the 1,2- enolisation ofa suga rreducin g end (seeFig . 4.11). Since sucrosedoe sno tcontai na reducin gend ,th ecolourin go f themixtur e considered here ismos tprobabl y duet oth eglucos e syrup. An experiment with material in test-tubes heated in a temperature-controlled glycerinebath ,sustaine d thisview . Tubes filled with glucose syrup discoloured to twice the intensityattaine db ytube sfille dwit ha 1 : 1 sucrose :glucos e syrupmixture ,whic hindicate stha tth e50 %sucros ei nth elatte r tubesma yb eregarde da sa colou rdiluen tagent .Thes etest sals o showed thatcolou rdevelopmen t increases linearlywit htim ean d is following an Arrhenius model for the dependence on H-C=0 H-C-OH I - II g HO-C-H C-0 Fig. 4.11 Schematic representation of the 1,2 enolization of a sugar reducing-end (vanZuiliche m et al., 1985) temperature. This means that the temperature is especially important. The test tube experiment showed that the reaction mixtureshoul dno tb esubjecte dt otemperature shighe rtha n155° C forappreciabl etimes . Theimportan tquestio ntha tmus tb eanswere d iswhethe rth ebul k materialflowin gi na nextrude rwil lreac hth etemperatur eo fth e heated extruder wall. Van Zuilichem et al. (1983a)presente d a calculationmetho dfo rth emas stemperature ,whic hresult si nth e following equation: 0 5 0 85 Inwhic h Hall= cumulativ ehea tflo wint oth ebul kmateria love r length z(W), Q = mass density (kg m"3), k = mass thermal 1 1 1 1 1 conductivity (Js" m" K" ), C p= mas shea tcapacit y (Jkg' K' ), N= rotationa lspee do fscrew s (revolutionspe rsecond ,r.p.s.) , Dw = barrel diameter (m), Ds= screw diameter (m), TH = barrel temperature (K),T m= mas s temperature (K), Z = distance along heated lengtho fextrude r (m)wit hZ = 0a t feedentrance , T)= bulkviscosit y (Ns m" 2). This result needs to be corrected for the volumetric chamber filling degreeU whic h isdefine d asth epar t ofth evolum e of a C-shaped chambertha t isactuall y filledwit hmaterial . U= ^2— (2) 2nNVxp 1 in which Qm = mass flow (kg s" ), Vj= volume of one C-shaped chambera tth eentranc eo fth eextrude r (m3), n = numbe ro fscre w channels.Multiplicatio nb yth efacto rU i snecessar yan dca nb e done, although there is no linear relationship between the volumetric filling degree and the surface provided for heat exchange.However ,sinc eU isabou t 0.5 inth e feed section of theextruder ,th eeffectiv ehea texchang e surfacether e isals o about 0.5 of the total surface available, considering the circular geometry of the extruder. Thus, simple multiplication byU wil lhol d inthi scase .Includin gth eheatin g suppliedvi a thescrew s (vanZuiliche me tal. , 1983a)an dmultiplyin g byth e factorU ,th e following expression forth etota lhea t flowint o thebulk ,0 tot,ca nb e constructed: Equation (3)ca nb edifferentiate d with respect to Zt o obtain 86 *'tot =O mC^Tb (4) is used, in which Tb is the rise in temperature of the bulk over an arbitrary small distance Az along the extruder. For AZ = 1 cm, o • oifr't Arb = (5) Q*PP In this way the temperature development of the bulk along the extruder can be calculated for every centimeter, using the calculationschem eshow ni nFig .4.12 .I nthi sway ,a temperatur e profile of the bulk material in the extruder may be obtained (Fig. 4.13). Corresponding heat transfer coefficients, a , are foundt ob e inth eregio n of 1000W m" 2K' 1. The conclusion is that only at high throughputs and high rotational speeds thebul k temperature does not reach the wall temperature.I nothe rcase s itma yb econclude d thatth ebarre l temperature isreache d quickly.Th eextrude r isthu ssee nt ob e aneffectiv ehea texchanger . Thisconclusio n isconfirme db yth eexperiment sreporte dbelow . start T*(*K ) z-z+0.01 calculate new n.p.Cp N 0„ r.pm. x10"3kg/s 1 30 5.5 2 30 16.5 calculate 3 90 16.5 TW= A33" K z= 0.97:sto p 0.97 z< 0.97 Z(m ) Fig.4.1 2 Calculating schemefo r Fig. 4.13 Temperature temperature development (van profileso fbul ki nextrude r Zuilichem et al., 1985) (vanZuiliche me tal. ,1985 ) 87 4.6.2 Time:dissolutio n of crystals Thetim eneede dfo rcomplet edissolutio no fa sucros ecrysta lma y beexpresse das : t=t(dp0)T (6) inwhic hdp 0= particl ediamete ra tt = 0 m ,T = temperatur e (K). Following Levenspiel (1972), using a material balance and assumingtha tdiffusio no fsugarmolecule si sth erate-determinin g step: in which Ss = quantity of material in the particle (kg),d =• particlediamete r attim et (m), k c= mas stransfe rcoefficien t (m s"1),A C= concentration-differenc e driving force (kgm" 3),Q, , =materia l flux (kgm" 2s" 1). Thedrivin g forceAC ,th edifferenc e inth econcentratio no f sugara tth esurfac eo fth ecrysta l and inth ebulk ,i s AC = C'-Cbulk (8) inwhic hth econcentratio na tth ecrysta lsurfac ei stake nt ob e C* (thesaturatio n concentration atth eprevailin g temperature) and 0^^ isth econcentratio n ofsucros epresen t inth ewater . Followingth eassumptio n ofJansse n (1978)an dva nZuiliche me t al. (1983b) the bulk material at any linear position is consideredt ob eperfectl ymixe dradiall ys otha tC bulki sunifor m ina plan ea tan y z-location. For the solubility of sucrose in water the Hoynak and Bollenback (1966) formula isth ebes tavailable : 88 %solubility^ 25200 (9) 4oo-r T=ti nunit s (w/w)an dfo rt=t°C .Thi sformul ai sexpecte dt ob e quitereliabl ean daccurat e forth ehighe rtemperatur eregions . For the calculation of densities out of the percentages from eq.(9), the sensitive relation 9a isused . C*=100/(%SOi/l550+(100-%sol)/1000) <9a) Basedo nHonig' s (1953)dat aa solubilit ycoefficien ti suse di n ordert ocorrec t forth enon-idea l natureo fth esolution .I an solutiono f two sugars in water, the presenceo feac h sugar hindersth esolubilit y ofth eother .I ncas eo fmixture so f 1:1 sucrose and glucose,th e solubility coefficient is 0.85. Table 1ca nthu sb econstructed .I ti sals onecessar yt oestimat e C^, the finalconcentratio n ofsucros epresen t inth ewater .Fro am mass-balance ona glucose-solids-fre e basis, i.e. assuming50 % 3 sugar in 10%water ,c en=d 83.33 % or 1241.5k gm" . Ss iscalculate d fromth e relationship 3 Ss=QV=Q±nd p (10) fromwhic hdifferentiating . dS =Q±d 2 d(d ) (11) s 2 p p Now,k cca nb eestimate d by comparing the flowpatter n arounad spherical crystal with a Brian-Hales creep flow, following Sherwood. 89 Table4. 5Saturatio n concentrationso fsucros ei nwate r versus temperature (vanZuiliche me tal. , 1985) T (°C) C*(%) C*(kgm" 3) 150 84-83 1271.4 145 84-00 1254.8 144 83-67 1248.4 143 83-34 1241.6 et al. (1975). They proposed a relationship betweenth e dimensionless mass transfer number kd/2ID andth e flow characteristic,represente db yth ePecle tnumbe r(Pe) : i^E=(4.0+1.21P£" M <«) 2 ID inwhic hI Di sth ediffusio ncoefficien to fsucros ei nwate ra t thecorrespondin g temperature (m2/s) Thisexpressio nreduce st ok cd/ 2= 2 ,whe nth erelativ evelocit y between crystalsan dbul k material iszero ,whic hi sth ecas e assumedhere . Equation (7)ca nno wb ewritte nas : 1 Qndpd(dp) _i p ( C C ] %dl 2dt ~ dp ^ or dpd(dp)=*£(C'-Cbulk)dt (13) Inorde rt odetermin e theconcentratio n C,^^,us ei smad eo f 90 Cb»ik= C end-CeJ^\ (14) Equation (13)the nbecomes : C dpd(dp) =IE(c*-Cend) +_g e"fe dt (15) e * dp0 so t=f ^(d ^ (16) j3 i. ki+k2dP Inth eAppendi x4. 1 iti sshow ntha tth edissolutio ntim et thu s becomes: (17) t =A |ln(l+c '3 )+ln(c '3 -c'3 +l)+tan lv/3 v^J \^)) where QdpV3 A= -2 _i 3 3 54JZ5Ce ld(C'-Cend) and inwhic h C' = 2£l_ (19) c-cend To find data for ID , van der Lijn's (1976) results for maltose/watermixture sar eused .Thes edat aar eextrapolate dan d concentration effects are ignored (Table 4.6).Th e results of this calculation procedure for this particular combination is shown inTabl e4.7 . The conclusion from these calculations is that the two 91 Table 4.6 Diffusion coefficients formaltos ewate r systems (van Zuilicbem et al., 1985) 2 T(°C) Dfm s 150 2-5 x 10 145 1-9 x 10 144 1-8 x 10 143 1-7 x 10 Table 4.7 Calculated dissolution times for sugar crystals of different diameter at different temperatures (van Zuilichem et al., 1985) T(°C) t(s) t if dp0 = 0-0005 m t if dp0= 0-001 m a 2 150 2-43x \0 d p0 60-7s 243 s 145 4-97 x 1084o 124-2s 497 s 144 ,7-42x \0*dln 185-5s 742s H 143 64-00x \0 dlTn 1600-0s 6400 s 142-96 oc 4.8Wate r content When the product leaves the extruder its pressure drops immediately and water evaporates from it, reducing its temperature.Th etemperatur edro pmeasure d inthi swor k isfro m 150t o 80°C,correspondin g toa hea t losso f 3 5 1 Ar-Cp=(150-80 )C p=70- 2.5-10 =1.75-10 J kg' (20) Assumingtha teffectivel yal lthi shea tgoe si nevaporatin gwate r 92 from the product,wit ha laten thea to fevaporatio no f225 0k J kg'1,th ewate revaporate di s (1.75-105)/(2250-103)=7 8 g kg-1produc t (21) Theinitia lwate rconten ti s10 0g kg" 1 (i.e.10% )an dth e final water contenti sthu s 100-78= 2 2g kg" 1,o rsom e 2%,whic hi s thedesire d leveli nth eproduct . 4.9 Experimental resultsan dconclusion s Testshav e been performed witha Cincinnat i Milacron CM45 twin screwe-c . (Fora drawin go fconica lscrew ssimila rt othos euse d see Fig. 4.14).Th eresult so fth etes t are showni nTabl e4. 8 Tend* stn e temperatureo fth ebul kmateria ljus tbefor ei tleave s rAxial clearance Spacer rings tecHo n against unvoluntary screw advance Fig. 4.14 Exampleo fa conica l twin screw extruder (Cncinnati) (vanZuiliche me tal. , 1983) theextruder .Th efee den ddiamete ro fth e screwsi s9 0mm , and thedi een ddiamete ri s4 5mm .Th etota l lengtho fth e screws is97 0mm . The colour ofth eproduc twa smeasure d byth eICUMS A method number4 ,whic hessentiall ymeasure sth eligh textinctio na t42 0 nm. Water content wasmeasure d byth eKar l Fischer method. Presence ofcrystal s inth eproduc twa sdetecte d byus eo fa polarizingmicroscope .A temperatur eprofil eo f130-140-15 0mean s thatth ebarre ltemperatur ei s130' Ca tth efee dsectio nan dthe n risestoward sth een do fth eextrude rt o150°C . These tests confirm thatth ebul k material indeed attainsth e 93 Table 4.8 Results of extrusion tests at different temperatures, throughputs and rotational speeds (van Zuilichem et al., 19S5) x ^ban-el(°Q 4>m(kgh' ) Nfr.p.m.J TerKi Colour(IJ Water(%) Crystallinity 140 20 40 139 54 140 20 60 138-5 43 140 40 60 138 41 130 20 40 129 39 130 20 60 128-5 38 130 40 60 127-5 30 130-140-150 20 40 142 82 130-140-150 20 60 140 58 '130-140-150 40 60 136 47 walltemperature .Th eproduc tdi dno tcontai nan ysuga rcrystals , and there was little coloration. The water content of the product is low, considering that no special water-removal action was undertaken. If such was done, even lower values could be reached. It is clear that the important factors are: (1) for the dissolution of the crystals: - the barrel temperature - the size of the crystals at the entrance of the extruder (2) for the colour of the product: - the final bulk temperature attained fo):th e water content of the product - the final bulk temperature attained - any special provision for removal of water 94 References Honig, P. (1953). Principles of Sugar Technology. Elsevier, Amsterdam. Hoynak, P.X., Bollenback, G.N. (1966). This is Liquid Sugar. RefinersSyru p Inc.,Ne wYork . Jager,Th . (1983). Residence time distribution in a twin screw extruder. Internal Research Report, Agricultural University,Wageningen . Janssen, L.B.P.M. (1978). Twin-screw Extrusion. Elsevier, Amsterdam-New York. Levenspiel,0 . (1972).Chemica lReactio nEngineering .Wile yan d Sons,Ne wYork . Vande rLijn ,J . (1976).Simulatio no fhea tan dmas stransfe ri n spray drying. Dissertation, Agricultural University, Wageningen. Shallenberger, R.S., Birch, G.G. (1975). Sugar Chemistry. AVI Publishing Co.,Westport ,Connecticut . Sherwood, T.K., Pigford, R.L., Wilke, C.B. (1975). Mass transfer.McGra w Hill,Ne wYork . van Zuilichem, D.J., Alblas, B., Reinders, P.M., Stolp, W. (1983a). A comparative study of the operational characteristics of single and twin screw extruders. Proceedings Cost '91.Athen s (concluding seminar). van Zuilichem, D.J., Stolp, W., Janssen, L.B.P.M. (1983b). Engineering aspects of single and twin screw extrusion of 95 biopolymers.J . Food Enana.. 2,157-175 . van Zuilichem, D.J. ,Tempel , W.J., Stolp, W., K. van 't Riet (1985). Production of high boiled sugar confectionery by extrusion cooking of sucrose/liquid glucose mixtures. J. FoodEnana. .4 , 37-51. van Zuilichem, D.J., Stolp,W. , van 'tRiet , K., (1985). Twin screw rextrusion cooking of sugar doughs. I.C.E.F. IV.. Edmonton. van Zuilichem, D.J., Stolp,W. , (1976). Theoretical aspects of extrusion of starch based products in a direct extrusion cooking process. Lecture at; Zentral fachschule der DeutschenSusswarenwirtschaft .Solingen . van Zuilichem, D.J., Stolp,W. , (1982a). Survey of the present extrusioncookin gtechnique s inth efoo dan d confectionery industry. Lecture at: Zentralfachschule der Deutschen Susswarenwirtschaft.Solingen . van Zuilichem, D.J., Stolp, W., (1982b). Extrusion of sucrose crystals. Lecture at: Zentralfachschule der Deutschen Susswarenwirtschaft.Solingen . van Zuilichem,D.J. ,Stolp ,W. , (1981).Standortsbestimmun gun d Ausblicke iiber de Einsatz von Koch-Extrudern zur Herstellungvo nZuckerwaren .Lectur eat :Zentralfachschul e der DeutschenSusswarenwirtschaft .Solingen . 96 Notation Heat fluxpe rm extruder length [J/sm] Heat conductivity [J/smK] Heat capacity [J/kgK] P 3 P Density ofth emixtur e [kgm" ] n DynamicViscosit y [Nsm"2] K Barrel innerdiamete r [m] Rs Screw diameter [m] T« Barrelwall temperature [K] Productmass temperature [K] Q„ Throughput [kg/s] z Coordinate alongsidebarre l [m] m Numbero fextrude r screws [-] 3 Vin Volume ofC-shape dchambe r [m ] N Numbero f revolutions [S"1] *m Throughput [kg/s] AT Temperature riseo fproduc tmas s [K] AZ Small increaseo fcoordinat eZ [m] t Dissolution time [s] Ro Sucrosecrystaldiamete ra tZ= 0 [m] D Coefficient of diffusion [m2/s] 3 Cend Final concentration given asdensit y [kg/m ] C* Maximum solubility atgive n temperature [kg/m3] c Cend/(C*-Cend) [-] Xs Molecular fraction [-] T Finalproduc t temperature [K] %sol Solubility of sucrose inwate r [-] 97 Appendix4. 1 fo dp0d(dp0) = _1_ rofp0d(c?po) (16) 1+ —d'po 1 fa d a c2/3/t['3Jl + a3 (l+a)3 = (l +a ) X(l —a + a2) a _/fZid^daa , ff (ia + i) /3 da (l +a)(a 2-a +l ) c} U(l +a ) J(, a2-a+l) "4 + (i«+ i ) 2 /f-^da ^(2a- l)da (1+ a ) (a -a+l) /3 2 c{ U(l +a ) J a - a+1 + Ja2-a+1/ ^d a f|(2a-l)da /• = /3 kV3k J U(l+a) J a •a+ 1 f jda \ + J(a-f)2 +| / '__/ f-Jd« | rh2a-Dda t = 3 /3 2 k? k <} U(l +a ) J a a + 1 !•i•da + 2 ) (a-i)2 +l 1 3d a fg(2 a—1 )d a /3 *2/3*li U((1l + +a a )) JJ a2-a + 1 98 JV3.d(-^.(«-i)a) l>/5 + I" ) .(a-i)2 + l A/3 + |ln(a2-a + l) +iV3tan(-^(a-i)) ) _ d / C \"3) a p end dpo ^C — Cend c ^end c* c 2 pd p0.y/3 2 / 1/3 D 54DC , ld.(C'-Cend) *i — ~ (C — Cend) P •W'^ H pO D /C/Qndend \ Pu + ln(^^'2/3-— C'1/3+l) + tariff T^C'1'3 ^) +*) ; \>/3 ^Po V3 ' K is now found by using the boundary values, t =0,d p =d p0 + ,»(^c""-^)) 99 Soa t the desired final condition, whend p= 0 , t becomes: -$>^L2 3 1/3(1 „ 1 + 1„ 1 + tan t-±)\ 54DCe ^ d(C*-Cend) \ I x/3J +^^^L.)"(|n('+c,")+'ii(C'','"c'"+') from which: d 1/3 2/3 1/3 i=„„J,: hy*„3 ^lln(1/3 l +C' ) + ln(C' -C' + l) 54DC2i d(C*-CeiId) V +ta "^C'1/3-^)-tan(i)) 100 CHAPTER5 MODELLING OF HEAT TRANSFER IN A CO-ROTATING TWIN SCREW FOOD EXTRUDER ABSTRACT The main problem in modelling heat transfer in twin screw food extruders, is to include all variables in a complete analysis. However, the solutions found in the various models,concer nmainl yth emeterin gsectio nwhic hi sassume d to contain only molten material. To predict the heat transfer process for food processing in the feed and compression section is far more complex and well fitting models forthes e sectionshav eno tbee ndevelope dyet. Thehea ttransfe rmode lsuggeste dgives ,throug hcombinatio n of existing theories, an extension towards predicting the performance of foodextruders . Themode lgav epromisin gresults ,whe nteste do na AP VBake r Perkins twin screw food extruder. Wheat flour was used as rawmateria l toacquir eth edata . THISCHAPTE R ISBASE DUPO N THEPUBLICATIONS : vanZuilichem ,D.J. ,va nde rLaan ,E. ,Kuiper ,E . (1990).Th e development of a heat transfer model for twin screw extruders,J . ofFoo dEnqnq..11,187-207 . van Zuilichem, D.J., van der Laan, E. Janssen, L.P.B.M (1991). submitted forpublication . van Zuilichem, D.J. van der Laan, E., Stolp, W.(1991), Modelling of heattransfe r ina co-rotatin g twin screw extruder,Applie d Food Extrusion Science.Edt. Kokini , Ho, Karwe. 101 5.1Introductio n Extrusioncookin gprocesse snormall yinvolv emixing ,meltin gan d the application of intensive energy to the product at high pressureswithi na shor tperio do ftime .Thi sresult si nvaryin g degrees of starch modification,protei n denaturation and other physico-chemical reactions.Whe nth eproduc tleave sth edie ,th e temperature andpressur edro pabruptl y andth eproduc texpands . There are two major types of extruders: the single-screw and twin-screwextruder .Th elatte rtyp ei sdivide d intoco-rotatin g and counter rotating depending on how one screw rotates in relationt oth eother . Counter-rotating intermeshingtwin-scre wextruder sar epositiv e displacement-pumpswhic hfor mclose d *C-shapedchamber sbetwee n the screwswhil eminimizin g themixin g and thebackflo wdu et o pressure build-up. In co-rotating extruders, the material is transported steadily from one screw to the other. The flow mechanism can be described by a combination of dragflow and positivedisplacemen t causedb yth epushin gactio no fth escrew - set inth e intermeshing region.Th e co-rotating type generally works at a higher screw speed than the counter-rotating twin screwextruder . Althoughsophisticate dscre wconfiguration swit ha sman ya seigh t ornin escre wsection sexis tnowadays ,on eca nbasicall yidentif y threedifferen t screwsections : (1) the feed section; which ensures that sufficient solid material istransporte d intoth escrew ; (2) the compression section; inwhic h themateria l is heated and worked into a dough mass during passage through this section; (3) the metering section; in which the screw configuration feeds the die constantly with material. The viscous dissipationo fmechanica lenerg yi stypicall ylarg ei nthi s sections otha tth etemperatur eincrease srapidly .Th ehig h 102 1 shear rate in the screw also enhances internal mixing to produce temperature homogeneity inth eextruder . 5.2 Thebasi cmode l fora singl e screw extruder Iti sinterestin gt oloo ka tth eflo wpattern sinsid ea nextrude r as they provide an insight into the mechanism of mixing, and facilitateestimatio no fresidenc etim edistributions ,predictio n offlo wrates ,pressur edro pan dpowe rconsumptio n (Bruine tal . 1978). Fig. 5.1 givesa n outline sketch of the flow channel in 1. SECTION OF EXTRUDER WITH DEFINITION OF GEOMETRY rU2- 7T N D cosfl i SIMPLIFIED FLOW GEOMETRY (SECTION AA 1 ) V2° l)z= 71N D COS0 y = h Vz profile i FLOW PROFILE IN COORDINATE SYS T EM USED (SECT. B B1) Fig.5. 1Definitio no fgeometr yan dsimplifie d flowgeometr yfo r a single screw extruder (Bruin etal. ,1978 ) 103 a single screw extruder. For simplification, the barrel is assumed to be rotating around the stationary screw. The basic problem indescribin g flowpattern s insideth eextrude r istha t the flows in the compression and metering section are non- newtonian and non-isothermal.Therefor e the equation of motion aswel la sth eequatio no fenerg yshoul db esolve dt oobtai nth e exact flow patterns. The general procedure is indicated schematically in Fig. 5.2 The set of equations can only be solved with some approximations. The most important are the assumptions of steady state, negligible inertia and gravity forces and of fully developed incompressible fluid flow. With these assumptions the flow in a slit of height h and width W, with onewal lmovin gwit ha velocit y 7rNDan dagains ta pressur e gradientdp/d zi sdetermined .Th evelocit yo fth ebarre lwal lca n now be divided into two components: in the cross channel direction:U .= TrNDsin ean di nth echanne ldirectio nU = 7rDNcos© . I Geometry of extruder Barrel length; compression ;numbe r of flights; diameter, etc. "**l H Boundary conditions Rotation VrOat Heatflux at Net mass for v and T speed barrel wall barrel wall flow —•• IH Equation of motion 0 = Vp + V.x (Maddock.1959) t Constitutive equation x= -T|A ; T|= T|<1 , I );I =A:A ; I = Det(A) for stress tensorx 2 3 2 3 (Maddock,1959) e.g. power law :x =-{m|y[ l (A= A) ] |n"1}A Newtonian ;x i-^A IY Equation of energy 0=-v.VT-V.q -x:Vv (Maddock,1959) Constitutive equation • for heat flux q q=-XVT, X:f(T) (Maddock1959) ' ' Flow patterns; temperature distribution Fig.5. 2 Schematic approach forcalculatio n of flowpattern s in anextrude r (Bruin et al., 1978) The velocity distributions in the x and z-directions can be 104 ^ calculated seperately and combined to give the total flow profile.Th eflo wi nth ez-directio n isth eresul to ftw odrivin g forces: the drag caused by the z component of the relative velocity ofth ebarre l (Uz)an dth epressur egradien t inth ez - direction because of the gradual pressure build up in the z- direction. With these assumptions and simplifications the solution of the equation of motion and the energy equation is possible.Th emos t simplecas earise swhe nth emateria l behaves asa newtonia n fluidwit htemperature-independen tviscosit yan d I Equation of motion z.direction x.direction Ctefe.f- (tyz )[^ m 4ea 3z 3y * 0 ( 3x 3yl y T II Constitutive equation .1 dvz T -.11 "VX for stress tensor 3y (Newtonian liquid) II Boundary conditions for flow (from geo metry of extruder. see Fig. 3 ) UX =TID N sine Uz = nDN cose 2 2---J--P(-2-)(1--2-M ) = -2y-3£) Jz h h li _bL (3p_) P= K ( P=3) 2uUz dz ' 12 Energy equation (p= constant ) h-PCpvz§I-g^<§^(§^ Y Constitutive equation qy=-A for neat flux dy •SI Boundary conditions qy = qr y= 0 for heat transport v=n qy=qb T= T(y,z ) Temperature profile Fig 5.3 Calculation scheme for flow patterns in a single screw extruderfo ra nincompressibl eNewtonia nflui dwit hconstan thea t conductivity andviscosit y (Bruine t al., 1978) 105 whenth e velocitycomponent si nth ey-directio nnea rth eflight s are neglected. The latter assumption is a reasonable approximation for shallow channels (h/w<0.1). Fig. 5.3. illustrates how velocity profiles could be derived and successively thehea t flux atth e barrel surface and the screw root can be calculated. It is interesting to note that the equation of energy is fairly difficult to solve even for this simplerheologica lbehaviou rbecaus eo fth econvectio nterm .Onl y whenadiabati cextrusio ni sassume dan dtemperatur egradient si n they-directio nar eassume dt ob enegligible ,a relativel ysimpl e resultca nb e obtained. 5.3 Various models 5.3.1 Introduction Themai nproble m inmodellin g heat transfer ist o include all variablesi na complet eanalysis .Thi sappeare dt ob eimpractica l because ofth ecomputationa l effort involved. Inman ypractica l cases,however ,som eo fth evariable sca nb eneglecte d asbein g unimportant for the particular set of conditions under consideration. This means that it is possible to make some assumptions and to derive a solution which is useful under certainoperatin gconditions .Car eha st ob etaken ,however ,tha t themode lshoul dno tb euse dfo ra nunjustifie dextrapolatio nt o othercircumstances . The solutions found in the various models concern mainly the metering section which is assumed to contain only molten material.Thi sha sbee nsolve dfo ra complex ,thre edimensional , non-isothermal, non-newtonian flow of viscous temperature sensitiveflui d (Martine tal. , 1969).I ti sfa rmor ecomple xt o predict the heat transfer process in the feed and compression section andwell -fittin gmodel sfo rthes esection shav eno tye t been found. In practice, most food materials for extrusion show a non- 106 Newtonian rheological behaviour. Several authors have analyzed suchsituation s (Griffith1962 ,Pearso n196 6 ,Yanko v1978 ,Yac u (1985)).The yproduce dmodel swhic hgav ea bette runderstandin g of the extrusion of non-Newtonian materials. Griffith used a numericalanalysi st osolv eth edifferentia lequation sfo rfull y developed velocity and temperature profiles for single screw extrusion of power-law fluids.Pearso n solved the equations of motion and energy for transverse channel flow and a superim posed temperature profile for power-law fluids. Results were presented in terms of dimensionless parameters of output, pressurean dth eBrinkma nnumbers .Yanko vuse dfinite-differenc e techniques to solve the equations of motion for non-Newtonian fluids,wher eh e assumed that temperature did not change along the channel. One could, however,doubt the accuracy of these models because of the assumptions used in solving the several equations. This is especially true when the power-law approximation for the food behaviour was used. Yacu gave the mostthoroug hdescriptio no fnon-Newtonia nbehaviour .Wit h some alterations,thi sapproac hi spartl yincorporate d inth eL Umode l which isdiscusse d insectio n 5.6 5.3.2 Dimensionless numbers To describe properly the heat transfer in an extruder, is it useful to introduce several dimensionless numbers. With these numbers it is often possible to see whether particular assumptionsar ecorrect .Th efollowin gdimensionles snumber sar e most frequently used inhea ttransfe ranalyses : Peclet;Th erati oo fhea ttransfe rb yconvectio nt ohea ttransfe r byconduction ; Pe= ™ (1) a Nusselt: The ratio of the total heat transfer to the heat transferb y conduction; 107 Nu= -5 2 (2) Reynolds:th erati oo fth einerti aforce st oth eviscou sforces ; Re = MQ (3) Graetz:th erati oo fhea tconvectio n andhea tconduction . Ina n extruder this means the ratio between heat transferred by the food flux and the heat transferred through the barrel to the extrudate; Gz= -&- (4) UH2 Brinkman: The ratio of heat due to conversion of mechanical energyb yviscou sdissipatio nt oth ehea tdu et oconductiv ehea t transfer; Br= Ji£ i (5) A.AT 5.3.3 Heattransfe r coefficient Knowledgeo fth ehea ttransfe rencountere d infoo dextrusio n is essential forth escale-u pan ddesig no f foodextruder san dth e associated temperature control systems. Apart from the cal culation used inth epenetratio n theory (seesectio n 5.4) only a smallnumbe ro frelevan tpaper so nhea ttransfe r coefficients infoo dextrusio nhav ebee n found.Becaus eo fth e importanceo f thehea ttransfe rcoefficient ,tw opaper sar econsidere d inmor e detail.Mohame de tal . (1988)analyze dth eGreatz-Nussel tproble m fornon-newtonia nflo wwit hviscou sdissipatio nalon gth echanne l ofa singl escre wextruder .A theoretica lmode lwa sdevelope dt o studyth eeffect so fmateria lproperties ,geometry ,an doperatin g conditions on heat transfer coefficients in single screw food 108 extruders for both heating and cooling with constant barrel temperature.Mohame de tal .foun do na nAP VBake rMP F5 0extrude r thatfo rBrinkma nnumber sles stha n40 ,viscou sheatin g effects were dominant. For Greatz numbers less than 333 the viscous dissipationoverrule dth ecoolin gcapacit yo fth eextruder .The y solvednumericall yth eenerg yan dmomentu mequatio nt odetermin e the temperature andvelocit y profiles,whic h were then used to determine the Nusselt number. From the Nusselt number they calculated the heat transfer coefficient, which at the fluid barrel interfacewa sdefine das : a(T-T )=-X-^- dT (6) b dz Indimensionles sterm sthi sequatio nca nb ewritte nas : oir-(T.-Tjrl=i^-S (7a) H az which resultsin : Nu(l-T*) =4^ * <7b> az where (2V-T) _* (T0-Tb) H Using multiple regression, Levine& Rockwoo d (1986) founda dimensionless correlation for the heat transfer coefficient. Their findingswer ebase d onth emode lo fTadmo r& Klei n (1970) and they assumed that the viscosity could be described by the powerlaw .The yconclude dtha tth ehea ttransfe rcoefficien tca n be described asa dimensionles s equation having the following form: NuJllH^Uir)**=m (9) andmultipl e regressiongave : 109 Nu = 2.2Br0-19 <10> Thecorrelatio ncoefficien tobtaine dwa s0.88 .Th estandar derro r ofestimat ewa scalculate d as12.4 %o fth emea nvalue .Additio n of the dimensionless group containing the Peclet number gave virtuallyn oimprovemen ti nth epercentag eo fvarianc eexplained . Asa result ,th estatistica lsignificanc eo fth ecorrelatio nwa s actually reduced;therefor ethi s factorwa sno tinclude d inth e equation. 5.3.4 Theory ofTadmo r and Klein (1970) Using the qualitative description ofMaddoc k (1959)o f thetw o phaseflo wi nth emeltin gzone ,a mor ethoroug h investigationo f this zonewa s undertaken.Accordin g toMaddock , the solidpar ticles incontac twit h theho t surface of the barrel partially meltan dsmea ra fil mo fmolte npolyme rove rth ebarre lsurface . This film, and probably some particles, are dragged along the barrelsurface ,an dwhe nthe ymee tth eadvancin gflight ,the yar e mixed with previously molten material. The molten material collectsa ta nare aa tth epushin gfligh t (meltpool),wherea sth e forward portion of the channel is filled with solidparticles . The width of the solid bed, X, gradually decreases toward the outleto fth escrew .Th emeltin gproces send swhe nth esoli dbe d disappears. In developing a melting model for the metering section,th e following assumptionswer eused : (1) The most important melting mechanism takes place between theho tbarre l andth e solidbed ; (2) The influence of pressure gradients on the velocity profiles is negligible (highly viscous flow in very thin layers); (3) The solid bed issemi-infinite ; (4) Heatconvectio n isnegligible ; (5) Thevelocit y ofth e solid bed inth e channel direction is constant; (6) Screwan dbarre ltemperatur eremai nconstan ti nth emeltin g zone; 110 (7) Thechanne l depth inth emeltin g zone isconstant ; Based on the observations of Maddock, Tadmor & Klein (1970) proposed amode lwhic hprovide d theequation stha tar erequire d tocalculat eth elengt ho fth emeltin gzone .Thi slengt hdepend s on the physical properties of the polymer and operating con ditions.Th eonl ymeltin gmechanis m takesplac ebetwee nth eho t barrel andth esoli dbed .Th emolte nmas s istransporte d byth e movement of the barrel relative to the solid bed, until it reachesa scre wflight .Th eleadin gedg eo fth eadvancin g flight scrapes the melt of the barrel surface and forces it into a meltpool.Thi smeltpoo lneed sspac ean dmove sint oth esoli dbed ; thesoli dbe ddeform san dit swidt hdecreases .I ti sassume dtha t the meltfilm thickness profile does not change in the melting zonean dtha tth emeltfil mthicknes sequal sth efligh tclearance . After these assumptions and thosementione d before, analytical calculationsbecam epossible.I nth epast ,specia lattentio nwa s givent oth eassumptio no fa constan tmeltfil mthicknes sbecaus e ofth eimportan trol eo fthi sfil mthicknes so nth ehea tflu xan d therefore onth e calculated melting length.Shapir o (1973)an d Vermeulen et al. (1971) concluded that the meltfilm thickness cannotb econstan tan dwil lincrease .Thi sdecrease sth eheatflu x toth esoli dbe dand ,consequently ,increase sth emeltin glength . Is it important to note that the effect ofviscou s dissipation on the channel temperature distribution togetherwit h the flow and heat transfer in the downstream direction were completely neglected inTadmor' smodel . Over the years Tadmors model has been changed, improved and adjusted by himself andman y other authors,t o incorporate new developments and calculation methods with computers. A new discussionoriginate dfro mth eobservation so fDekke r(1976 ). Hi s findings refuted the remaining basic concept of Tadmor's model,namely that of the deforming solid bed and the existence ofa meltpool .Lind t (1976)analyse dthi ssituatio nan dpropose d another basic model. Tadmor's model and Lindt's model can be regardeda sth etw oextreme si nth efiel do fmeltin gzon emodel - Ill ling. Between these two,man y other models for melting inth e metering zone,o fvaryin g accuracy,hav ebee npublished . 5.4 Boundary layer theory inth epum p zone In1953 ,Jepso nconsidere dwha ti scalle dth e 'wipingeffec tfo r heattransfe ri nth epum pzone 1:Afte rth efligh to fth erotatin g screw has wiped a certain area on the inner barrel surface,a fresh layer of polymer becomes attached to the same region and remainsther e for,approximately , onerevolution .Th eamoun to f heat penetrating into this layer during that time by pure conductivehea ttransfe r isthe nremove dwit hth epolyme r layer y=z T=T0 wall y=0 T=Twaii Fig. 5.4 Boundary layer (Beek,1974 ) and is thought to be homogeneously distributed throughout the bulk of thepolyme r inth e screw channel.Thi s process of heat penetration during extrusion canofte nb e considered as anon - stationary penetration in a semi-infinite medium. The heat balance isgive nby : pCp (11) lTt - Hadx?2 dy2 dz2 or, in one dimension: 112 T?-gf Ifth epenetratio ndept ho fhea ti smuc hsmalle rtha nth eheigh t of the extruder channel, the solution of equation (11) is the wellknow n errorfunction : T.'z-?- —2-ex4—*—} <12> rTb /it 1,2jai) from which a time averaged heat transfer coefficient over one revolutiontim eca nb ederive d fora scre wwit hm threa tstart s and arevolutio n rateo fN RPS : IpCpmN Another approach is to consider the heat transfer to a semi infinite flowing medium for which the energy balance can be approximated by: x dx dy2 with boundary conditions: X < 0 T = TQ X>0 y-.oo T = T0 (14b) x>0 y = 0 T = TH Inorde rt osolv ethi sequatio n iti sassume d thatth evelocit y may be linearized over the thickness of the thermal boundary layer (Levequesolution )to : Vx= i-y (15) y=0 indicates the location on the barrel wall at uniform temperature T. 113 The temperature profile in the thermal boundary layer 1B approximated by a parabolic function: •H)" (16) where Sis the (unknown) boundary layer thickness. Introducing equations 15 and 16 in equation 14a gives after integration an expression for the boundary layer thickness: 6 =/= 36ax\- Pf f (17) and for the heat flux: -•= 2A(-^)" * I *= 2A/^* ^j P (") 5.5 Extruder model by Yacu The first simulation model of a twin-screw corotating food extruder, was developed by Yacu (1983). The extruder is divided into three main sections: - solid conveying zone - melt pumping zone - melt shearing zone The temperature and pressure profiles are predicted for each section separately. The following assumptions were made: (1) The rheology of the molten material is described by a non- Newtonian, non-isothermal viscosity model taking into account the effect ofmoistur e and fatcontent.Th e extruder is operating under steady state conditions. (2) Steady state behaviour and uniform conditions exist. 114 (3) The melt flow is highly viscous and in the laminar flow regime. (4) Gravity effectsar enegligibl e and ignored. (5) The screw isassume d tob eadiabatic . (6) The degree of cooking is assumed to be uniform over the cross-section. 5.5.1 Solid conveying section Twin screw extruders for food processing, operatemostl y under starved feedcondition san dth ethroughpu t isdetermine d byth e feeding unit,th e extruder's screw speed and torque. Therefore the screws in the feeding section are partially filled and no pressure isdeveloped . Thedissipatio n ofmechanica l energy is negligiblebecaus eo fthes econditions . Heat istransferre db y conduction fromth ebarre lt oth e foodmaterial . To simplify the analysis and taking into account that mixing withinth echanne l isreasonabl ygoo d inco-rotatin gextruders , thehea ttransfe ri sassume dt ob econtrolle doml yb yconvection . A pseudo heat transfer coefficient Us is used to replace the characteristicconductivit ypropert yo fth egranula rmateria li n the feed section. Constructinga hea tbalanc eacros sa nelemen tnorma lt oth eaxia l direction and solving forT wit h theboundar y condition atx=0 ,T=T f results inth eequations : QaCpsT+FUsA (Tb-T) dx = QmCps (T+dT) (20) fromwhic hfollows : -FU,AX (21) T = Tb-(Tb-Tf)e "*<>• Thehea t transfer coefficientU st oth epowde r inth efeedzone , forpractica lreason sestimate da sth ehea ttransfe rcoefficien t based onth epenetratio ntheory ,woul dgiv eirrealisti cvalues . Phase discontinuity and the existence of additional resistance 115 to heat transfer between the solid particles, are the main reasons forthis . 5.5.2 Meltpumpin g section Inthi ssectio nth echang eo fth emateria lfro msoli dfood-powde r to a fluid melt is assumed to take place abruptly. Martelli ^.bprfel^/X |6 Zijbeat between flightgap/barrel n> •® flight Zj) flight/ >' screwroot screwroot *--- (Zi) channel • flight flight transportdirection Zyheat between flights "-© -JLVUQgh t Fig.5. 5Leakag eflow si na co-rotatin gtwi nscre wextrude r (van Zuilichem et al., 1991) assumedtha tth eenerg ywa sdissipate dwithi nth echanne lan ddu e to the leakage flows in the various gaps (See Fig. 5.5).H e defined fourlocation swher eenerg ywa sconverted : - within thechannel : (22) 1 2h * inwhic h h isth e channel depth (m)an d D the equivalent twin 116 screw diameter (m). (see Figure 5.5fo r the different cross sections) - betweenth e flightti po fth e screwan dth einsid esurfac eo f thebarrel : %2D2 eC 2 Z2= b *m'»N (23) wherem *stan d fornumbe ro fscre wflights ,e i sth escre wfligh t tipwidt hi naxia ldirectio n (m)an dC estand sfo rth eequivalen t twin screwcircumferenc e (m). -betwee nth e flightti po fon escre wan dth ebotto m ofth e channel ofth eothe rscrew : 8u2j3e 2 Z3= i7?'n^ (24) in whichc give s the clearance between fligth tip and channel bottom oftw o opposite screws and Istand s for the distance betweenth e screw shafts (m). -betwee nth eflight so fopposit escrew sparalle lt oeac hother : Zi =j^h£^HL m^N2 (25) where a indicates the clearance between flightso fopposit e screwsparalle l toeac hother . The total energy converted perchanne l perscre w turnwa s therefore expressedas : 2 Zp =Z 1+Z2+Z3+Zt =ClpilpN (26) 117 Where C, canb e defined asth epumpin g section screw geometry- factor and isdescribe dwith : 4 2 2 2 3 2 2 2 2 _7i J>|j>tan(pp mJ n De Ce _8 n J e n I h/Tb -I ) ) i21\ Cl* 2h { 6 e 2a J Because the length of a screw channel per screw turn equals TrDtanip theaverag eamoun to fhea tgenerate dwithi na nelemen to f thicknessd xca ntherefor e beevaluate das : dZ-" C ^A^Sn^ (28> Theoveral lshea rrat eo nth eproduct ,whil epassin gthroug hth e pumpingzone ,includin gth eamoun ttake nu pb yleakag eflows ,ca n beestimate dby : y =^J/SP (29) VP Theviscosit y ofth eproduct ,u , isdescribe d byth epower-la w equation and is depending on shear rate(y), temperature(T), moisture content (MC)an d fatconten t(FC) : ni a iMC MC ) a FC _J AT |i=pH 0Y" e- > - ° e- * e * (30) Thisrheolog ymode li sdevelope d fora whea tstarch .Th evariou s indicesar edetermine db ymultipl eregressio nanalysi st ofi tth e rheologymodel . Constructinga hea tbalanc eacros sa nelemen ti nth emel tpumpin g zone, coupled with the boundary condition: T=Tm at X=Xm, gives therelation : Clpi e gcvmr+aA(r„-r)dx+ y * dx-oc^r+dr) (si) 118 r -|f =C 2p e-^ + C3p(Tb-T) (32) where C 2 C2p = ^^ (33) 2P wl?tani>J?.C_ and: C3P=^ (34) Thisfirs torder ,non-linea rdifferentia lequatio nca nb esolve d by integrating itnumericall y usingth eSimpso n 3/8 rule. Oneo fth eassumption smad eb y Yacu (1983),wa stha tth emomen t the mass enters the melt pumping section, the screws become completely filled. Inhi s calculation the mass is subjected to the total shear stress.However ,mos t twin screw extruders are starved fedwhic hresult s ina partiall y lowfillin gdegree .I t would therefore be reasonable to conclude that the shear the product receives islowe rtha nYac uassumes . 5.6 LOMode l The LUmode l ,develope d by Van Zuilichem et. al. (1983,1985 ) andlate rals oimprove db yVa nZuiliche met .al . (1990),ha sbee n extended andmad e suitable for computer calculations.A unique featureo fthi smode li stha ti tcalculate sth etota ltransferre d heat in the extruder for every position along the screw axis. Compared to the model of Yacu (1985), the LU model gives an approximation of the heat transferred from the barrel to the foodmaterial,base d onboundar y layer theory using the Leveque solution. The model presented here, is composed of two major parts. The first part calculates thehea t transferred from the barrelt oth eextrudate .Thi spar to fth ecalculatio nmetho d is based on the findings of Van Zuilichem et. al. (1983, 1985, 1990). Part two of the model calculates the heat generated by viscousdissipatio n (seeFig .5.6) .Wit hrespec tt othi smodel , 119 TOTAL HEAT PRODUCTION ccinvecfiVeVheat::S |heat by viscous dissipation; t barrelheater s screw setup surface area clearances heat exchange • screw speed coefficient • viscosity distributive • mixing dispersive mixing • degree of fill degree of fill ^ r residence time distribution Fig. 5.6 Division ofhea t development ina n extruder it isimportan tt onot eth e followingassumptions : - Inputparameter s ofth emode lare : * thetemperatur eprofil e imposed atth ebarrel . *torque . * feedflow. - Torque is converted into: pressure energy, phase transition energy and temperature increase.Hereb y isi sassume dthat : (1) Thetorqu ean dth ehea tconvectio n arelinearl y relatedt o the degree of fill, according to van Zuilichem et al. (1983) (2) The viscosity changes of the material are described by a non-Newtonian,temperature-tim edependen tpowe rla wmodel . (3) The amount of energy necessary to gelatinize the starchy food-product isnegligibl ecompare d toth eenerg y consumed torais eth etemperatur e ofth eproduct . (4) The meltflow is highly viscous and in the laminar flow regime. (5) Onlyth ene t flow inth eaxia ldirectio n isconsidered . Thecalculatio nschem e (Fig.5.7 )show sth estructur eo fL Umode l 120 6T / *w + C2pexpi -b,Tl) F 5X "^ CpQm . i. ST 1 F = * w 7^=C,p.,p(-b,T)F 5X cpom 1, , i , 0„E/P F = vm„ C,p »,N! •' . = K o.(TK -T). C2p= "D tan *pQmCp„, „ (360a-23) *p = »l.«»<'>iT> 360 ..^(JS^L)-"* 'p=»oVn1.«p'-a1(MC-MCo)W-j2FCW(-b1T)«p{oAx <-4E/RT(t))dtf „ 3/2 -1/2 ,p / ;«D N "7 " 46J1U/P) z S 2 2 3 2 2 2 2 t^O^D tan#p H,« D^Ce 8* I e i I h,/(D -I ). 2 X sh (T T ,+ k (MC_MC +x Fig. 5.7 Calculation scheme of LOmode l (van Zuilichem et al., 1990) with itsvariou s components.Base d onthi s calculation scheme, a computermodel ,followin ga stepwis eprocedure ,wa sdevelope d withth efollowin gprocess -an dmaterial-variable sfo rth einput : -Materia lvariables : * CP * initialviscosit y * melting temperature * density *moistur e content - Systemvariables : 121 * throughput * initialmateria l temperature * lengtho f extruder * screwgeometr y * temperatureprofil e Theoutpu to fthi scompute rprogra mi sa plo to fth etemperatur e of the food product versus the axial distance inth eextruder . It is also possible to plot the temperature increase, due to penetration ofhea t and dissipation,versu s the axial distance in the extruder. The calculated temperature values are also transcribed ontoa datafile . 5.7Th eviscosit y model The function of many operations in polysaccharide extrusion cooking is to rupture the starch particles and to gelatinize theircontent st oa certai nexten tfo rdifferen tapplication si n human food and animal feed. During this process considerable changes inth erheolog y ofth emateria l areoccurin gdu et oth e formationan dbreakag eo fge lbonds .Thes echange sca nhav ethei r impacto nprocessin gparameter san dstabilit y criteria.Jansse n and van Zuilichem (1980)suggeste d anequatio n especially for applications in food extrusion technology. Thismetho d isuse d in the LUmode l and describes the effects of the mechanism of gel-formation and -breakage on the viscosity as a function of shearrates ,fat-an dmoisture-content .The ycombine da powe rla w behaviourwit ha temperatur edependenc yan da nextr ater mfo rth e formationo fth enetwor kdependin g onth eactivatio n energy and the residence time. Now the apparent viscosity, np, can be writtenas : Ije^Dt] (35) { ! ) VLp=V01?e-* " -*> e-^e-^'e whereD ti sa convectiv ederivativ eaccountin g forth efac ttha t the coordinate system is attached to amateria l element as it moves through the extruder. The integration can be performed analytically as is shown in Appendix 5.1. With this model it became possible to explain certain instabilities as they occur 122 during theextrusio n ofstarches . 5.8 Material As a model food-material, standard biscuit flour (MENEBA Rotterdam)wa suse d (Table5.1) .Th eexperiment swer edon eo na Table 5.1 Physical constants of standard bisquit flour C 1740 (J/kgK) k0 0.303 (J/kgK) p 1285 (kgm*) T-.lt 348 (R) Moisture Content 15 (%) twinscre wco-rotatin gextruder ,typ eMPF-5 0AP VBake rwit ha 2 3 kWengine .A bi gadvantag eo fthi sextrude ri sth emodula rscre w design allowing to compose almost every desired screw configuration. Screw sections used are elements like: transportation elements and kneading elements e.g feed screw, paddles 30°/45V90° and single lead screws (Fig.5. 8 a,b ,c ). single lead screw paddle 145 I paddle 190 ) Fig.5. 8 Screw elementsuse d forthes eexpriment s 123 •4—barre l probe screw Fig. 5.9 Place ofth etherma l probe Heat issupplie d by9 heatin g elements located onth ebarre lo f the extruder. The experiments were done with two different feed paddle feed paddle feed paddle single Screw1 screw 90* screw 90* screw 90* lead feed paddle feed ai feed paddle -•s single Screw2 screw 30 screw a screw 30 aiai a m lead 0} OJt_ a. i/i Fig.5.1 0Lo wan dhig h shear screwprofile s (screw l:hig h shear, screw 2:lo w shear) temperature profiles imposed on the barrel. To determine the temperature ofth eproduct ,si xtherma l probes (seeApp . 1)wer e installed with the tip of the probe stuck 0.7 cm into the Barrel Section 1 2 3 4 5 6 7 8 9 Profile „A1" 40 80 100 110 120 130 140 150 160 Profile „B2" 40 60 80 100 100 120 120 130 130 Fig. 5.11 Temperature profiles (Profile A: 160°C, profile B: 130°C) 124 material.(Fig .5.9 ).Thi swa saccomplishe db yusin gspacer-scre w elements. It was assumed that there is no heat production by viscous dissipation at the tip of the thermocouple. To study the influenceo fviscou sdissipatio no nth eproduc ttemperature ,tw o screwtype swer eselected .I nfigur e5.1 0scre w1 ca nb eregarde d as a "high shearing" screw because of the relatively high percentage of90 °kneadin g elements,wherea sth escre w2 ca nb e regarded as a "low shearing" screw. The temperature profiles imposedo nth ebarre lar egive ni nFigur e5.11 .Th emeasurement s werecarrie d outwit hdifferen t screwspeeds ,moistur e contents andthroughputs . 5.9 Results It would go too far to present all the data acquired by measurement andprediction .Therefor eth egenera ltren dwil lb e discussed.Th escre wspeed suse dwer eresp .RP M50 ,150 ,200 ,25 0 and 300.At th elowe rRP Mnumber sa capacit yo f0.005 2kg/ swa s used,steadil y increasingtoward s0.01 6 kg/sa tth ehighes tRP M (Tabel5.1) .Th emoistur econtent sapplie d ranged from26.8 %t o 35.6%. The resulting temperature profiles calculated with the model andth ecorrespondin g thermocouplereading sar ecollecte d inTabe l 5.1. Forsom erun sth ereading san dth ecalculate d dataar egive n in the diagrams in Figure 5.12, 5.13, 5.14. figure 5.12 shows a temperatureplo t forscre wNo.2 ,th elo wshea rscrew ,extrudin g flour at 27%wate r contentwit h athroughpu t of 0.0061kg/ sa t a screw speed of 250 RPM. The imposed temperature profile was profileA .I nFig .5.1 3an d5.14 ,th etemperatur eplot sfo rscre w No. 1an dNo. 2ar egive nfo rresp .th etw odifferen ttemperatur e profilesA an dB ,extrudin g floura tresp .31.2 %an d24.4 %wate r content and throughput of 0.0061kg/ s at 250 RPM for both the runs. It can be seen clearly in Fig. 5.12, that there is a slight deviation incalculate d andmeasure dvalues, i nth eorde ro f1 0 °Cmaximu m atth emeterin g section.I nFig.5.1 3ther e isalmos t 125 Table 5.2 Comparison betweenmeasure d and calculated data SCREW 1. TEMPERATURE PROFILE A. RPM (KG/S) MC.(%) Tl T2 T3 T4 T5 T6 T end MEAS 56.3 77.1 105.3 117.6 134.3 149.2 156 50 0.0052 26.8 • MOD. 52.7 78.5 115.8 125.2 147.3 158.2 160.0 MEAS 55.9 73.6 109.2 121.8 139.5 150.4 159.0 150 0.0063 29.5 MOD. 47.9 70.3 117.8 127.0 149.2 159.9 161.6 MEAS 48.0 68.2 109.1 123.4 137.0 150.2 159.0 250 0.0072 35.6 MOD. 44.0 64.2 118.0 127.5 149.8 160.5 162.2 SCREW 2. TEMPERATURE PROFILE A. RPM (KG/S) MC.(%) Tl T2 T3 T4 T5 T6 T end MEAS 30.5 47.3 65.0 90.2 102.3 141.0 200 0.016 27.4 - MOD. 23.9 39.7 59.4 97.0 123.2 140.3 156.7 MEAS 29.8 47.2 61.9 86.6 101.6 143.0 300 0.0165 27.8 MOD. 23.8 39.1 58.4 90.4 127.4 145.3 159.2 nomeasurabl edeviation ,wherea s inFig .5.14 ,ther eagai n isa deviation smaller than 10 °C; giving a measured temperature slightly lowertha npredicted . 5.10 Discussion A studypoin t of themode l is,tha t it iscapabl et o calculate the resulting temperature profile of the product over the viscosity-history as the key rheological parameter. In the relation used for the viscosity (eq. 23) are taken up a restricted amounto fmateria lparameters ;lik emoistur econten t (MC%),fa tconten t (F)an dgelatinizatio nenerg yvalue s (AE).Fo r AE,MC %an dF ar etake naverag evalue sa sar eknow n forstarch . It isknow n that AE issomewha t dependent onth e shearrate and themoistur econten tan dwil ldeviat esomewha tt olowe rapparen t valuesdu et oth efac ttha tth echanne lo fa co-rotatin gextrude r is open and venting of water is continuously occuring via the feed port during operation. This might explain the higher 126 screw profile 2 7 I 8 | 9 temp.zones (1 thru 9) temperature ( C) 160 roduct temperature temp, measuring points -Q- penetration temperature isc. dissipation temperature mposed temp, profile 1 1 r 1 20 40 80 100 120 140 barrel length (cm) temp, profile A; moisture content 27%-, r.p.m. 250 ; throughput 0.0061 Kg/s Fig. 5.12 Producttemperature vs. axial distance screw profile 1 3 I 4 8 I 9 J— temp,zone s (1thr u 9I temperature ( C) 160H temp,measurin g points -r> •"' product temperature penetration temperature vise.dissipation temperature imposed temp, profile — T 1 r 1 80 100 120 140 barrel length (cm) temp, profileB; moisture content 31.2%; r.p.m.25 0 , throughput 0.0062 Kg/s Fig. 5.13 Product temperature vs.axia l distance deviations at the 160°C temperature profile and the high RPM 127 screw profile 2 T I 2 I 3 I 4 I 5 I 6 6 I 9 L temp.zones(1thru9) temperature ( C ) product temperature 160 5" 120— temp.measuring points-c^- visc.dissipatio n temperature penetration temperature imposed temp,profil e — 120 140 barrel length(cm ) femp.profil e A;moistur e content 24.4%,- r.p.m . 250; throughput 0.00602 Kg/s Fig. 5.14 Product temperature vs.axia l distance values,whic hwil lcaus emor eventin go fvapou ran ddeviatin gA E values.Anothe r facti stha tth ebiscuitflou rcontain sabou t9 % of protein,whea t gluten,whic h shows a different yield stress than the starch inth e flour,whic h againwil l be noticable at thehighe rRP Mvalue san dth edifferen ttemperatur eprofiles . Itwoul db ebette ri nthos ecase st ocalculat eth eweighte dshea r history of the constituents of the flour, inorde r to optimize therheologica l informationan dt oconside rsli pan d slip-stick phenomena inth eextruder ,especiall y inth emeterin g section. Itmus tb esai dtha tth ereading sfro mth ethermocouple s (Figure 5.9)a thig hscre wspeed san drelativel y lowthroughput sar eno t ideal because a low degree of fill in nearly all parts of the extruder can cause the effect that the thermocouples measure a mixtureo fmel ttemperatur ean dstea mtemperatur ei nth emeterin g section, instead ofa homogeneou sproduc ttemperature . 128 Withthi si nmin d itmus tb eadmitte dtha tther estil li sa lac k ofdat aan dther ei sth enessecit yt ostud yth efollowin gaspect s inth enea r future inmor edetail : l:The impact of the chosen rheology model on the predicted viscosity values. Other models have to be considered and tested onvalidity . 2:Measuring andmodellin gth eradia ltemperatur edistributio ni n thescre wwit h specialemphasi so nth emeterin g section in order to increase the accuracy of the predicted product temperature. 3:Studying the elongational flow of theproduc t before the die entrance and inth edi eitself . 129 References Beek,W.J. ,Fysisch eTechnolog ye nPolymee rverwerking ,ed .W.J . Beek,Polytechnisc htiidschrift ,Delf t Oct.1974 ,p 42-71. Bruin, S., van Zuilichem, D.J., Stolp W. (1978). A review offundamental and engineering aspects of extrusion of biopolymers in a single-screw extruder. J. Food Process Enq.. 2,1-33 . Dekker, J. (1976). Verbesserte Schneckenkonstruction fur das Extrudierenvo n Polypropylen.Kunststoffe . 66,130 . Griffith,R.M . (1962). Fullydevelope d flowi nscre wextruder's , theoretical and experimental, Ind. Enq. Chem. Fundam.. 2, 280-286. Jepson,C.H . (1953). Futureextrusio nstudies .Ind .Enq .Cheiq. f 45 (5)992 . Levine,L. ,Rockwood ,J . (1986).A correlatio nfo rhea ttransfe r coefficients in food extruders. Biotechnol. Progress. 2, 105-108. Lindt,T.J . (1976). A dynamic melting model fora single screw extruder.Polvm .Enq .Sci. .16 ,291 . Maddock, B.H. (1959). A visual analysis of flow and mixing in extruder screws.SP EJ. . 15,383-389 . Martelli, F.B. (1983). Twin-Screw Extruders. Van Nostrand- Reinhold,Ne wYork . Martin,B. ,Pearson ,J.R.A. ,Yates ,B . (1969).Numerica lstudie s of steady state extrusion process. Cambridge University, Department of Chemical Engineering Polymer Processing Research CenterRepor tNo .5 . 130 Pearson, J.R.A. (1966). Mechanical Principles of Polymer melt Processing.Trans .Soc .Rheol. . 13,357 . Shapiro, J. (1973). Melting in plasticating extruders. PhD Thesis,Cambridg eUniversity ,1971 . Shapiro, J., Halmos, A.L., Pearson, J.R.A. (1976). Melting in single screwextruders .Polymer . 17,905-917 . Tadmor, Z., Klein, I. (1970). Ena. Principles of Plasticating Extrusion.Va nNostrand-Reinhol dCompany ,Ne wYork ,185-90 . Vermeulen,J. , Scargo,P.G. , Beek,W . (1971). Themeltin g ofa crystalline polymer in a screw extruder. Chem. Eng.Sci. . 46, 1457-1463. Yacu, W.A. (1983). Modelling of a twin-screw co-rotating extruder.In :Therma l Processingan dQualit y ofFoods ,ed . P. Zeuthen, J.C. Cheftel, C. Eriksson. Elsevier Applied Science Publishers,London . Yankov, V.I. (1978). Hydrodynamics and heat transfer of apparatus for continuous high speed production of polymer solutions.Hea tTransfe rSovie tRes. . 10,67-71 . van Zuilichem, D.J., Alblas, B., Reinders, P.M., Stolp, W. (1983). A comparative study of the operational characte ristics of single and twin screw extruders. In: Thermal Processing and Quality of Foods, ed. P. Zeuthen, J.C. Cheftel,C .Eriksson .Elsevie rApplie d SciencePublishers , London. van Zuilichem, D.J., Tempel, W.J., Stolp, W., vanxt Riet, K. (1985). Production of high-boiled sugar confectionery by extrusion-cooking of sucrose: liquid glucose mixtures. J. Food Ena. 4, 37-51. 131 van Zuilichem, D.J., Jager, T., Spaans, E.J., De Ruigh, P. (1989). The Influence of a barrelvalve on the degree of fill in a co-rotating twin screw extruder. J. Food Enq.. 10,241-254 . van Zuilichem, D.J., Janssen, L.P.B.M. (1980). In: Rheoloqy vol.3. ed. G. Astarita, G. Marcus, L. Nicolais. Plenua Press,Ne wYor kan dLondon . 132 Notation Thermal diffusity [mz/s] Moisture coefficient ofviscosit y [-] Fatcoefficien t ofviscosit y [-] A Surfaceare a [m2] b Channelwidt h inaxia l direction [m] bi Temperature coefficiento fviscosit y [-] b2 Coefficient inhea t capacity [-] B Widtho frevers e cross-channel [m] Br Brinkmannumbe r C-] c General clearancewidt h [m] Ce Equivalenttwi nscre w interference [m] Heat capacity CP [J/kgK] °pn Heatcapacit y ofmolte nmatte r [J/kgK] Heatcapacit y of solidmatte r CPS [J/kgK] Heatcapacit y ofmolte nproduc t CPP [J/kgK] Heatcapacit y ofwate r CPW [J/kgK] d Thickness ofboundar y layer [m] D Barrel diameter [m] De Equivalent oftwi n screwdiamete r [m] Ds Diameter ofth e screwsa tZ= z [m] e Screw flightti pwidt h inaxia l direction [m] E Residence time [s] AE Gelatinization energy ofproduc t [J/mol] F Degree of fill [-] FC Fatconten t [%] G Deptho f reverse cross-channel [m] Gz Graetz number [-] h Channel Depth [m] H Characteristic height [m] I Distancebetwee n screw shafts [m] Secondan dthir dinvariant so fth erat eo fdeformatio n I2,I3 tensor [s"2,s" 3] Exponents inNussel tequatio naccordin gt oLevin ean d Rockwood [-] k3,k4 Coefficients intherma l conductivity [-] 133 K Twin screw circumference [m L Lengtho f extruder [m m Parameter inpower-la wmode l [Ns1^"2 m* Numbero f flights [- m, Numbero f reversecross-channel s [- MC Moistureconten t [% n, Power-law index [- N Screw speed inrevolution spe r second [s'1 Nu Nusselt number [- p Pressure [Pa P Ratiobetwee npressur e flowan ddra g flow [- Pe Pecletnumbe r [- 2 1 q Heat fluxvector ,component sq x,q,q z [Jm" s* Qm Feed rate [kg/s R Distance from screwaxi st obarre l [m Re Reynolds number [- s Width of flight [m t Time [s T Temperature [T U Relative velocity of barrel wall with respect to centerline ofscre w [m/s v Velocity vector,component sv x,v,v 2 [m/s V Channel volume [m3 w Width ofextrude r channel [m X Axial distance [n z Total penetration depth [m Z Viscous heat dissipation [N/«i a Heat transfer coefficient [Js^m^K'1 /3 Overlap angle between twoscrew s [- Y Shear rate [s"1 5 Flightclearanc e [m A Rateo fdeformatio ntenso r [s*1 e Clearancebetwee nfligh tti pan dchanne lbotto mo ftw o opposite screws [m 17 Non-Newtonian viscosity [Nsm'* 6 Angle of flight [- A, Thermal conductivity [J/smK 134 H Newtonianviscosit y [Nsm"2] p Density [kgm"3] a Clearance between flights of oppsite screws parallel toeac h other [m] *H' Heat fluxpe rmete r [J/m] 2 *H'' Heat fluxpe r squaremete r [Jm" ] Subscripts b Barrel f Feedmateria l h hydrodynamic max Maximum m Meltmateria l 0 Initial orreferenc e condition p Pumping section r Bottom of screw channel rs Reverse screw section s Solid (powder) T Thermal tot Total w Heat x,y,z Co-ordinate indices 135 Appendix Thetravelin g observerpar to fequatio n3 5ca nb ewritte nas : [ e~ bt+cdt (al) By substitutiono f " A7 b7 yu ,d tca nb eexpresse das : jDfc+ C u= -M.-ht+ c= M~ c) =Mt=^ | =-^fcfci - bt+c d(bt+u \ u J u2 dt = --^-du (a2) bu2 By substitutiono fequatio na 2i nequatio n al, itca nb efoun d that: nax -max AE „^ 2 J b J u • >2 *Tnin *Wi With: AE "min ^min + C And: AE "max = bt^+c "mx , , "W . AE**.f JLl du =A^f^Z_^l )+ AS. f J!-ldu (a4) Theintegra lo nth erigh than do fequatio na 4ca nb ewritte nas : ^ f JL!du= A£.fJ^du-A^.f JL!du (a5) £> "milJ l u b J "lainu b J "mau x 136 Forth eremainin g differential equation: f-^-du }jLldu =iog U—y-+_HL-_^i_+ (a?) J U 1! 2-2! 3-3! 137 CHAPTER6 THE INFLUENCE OF A BARREL-VALVE ON THE DEGREE OF FILL IN A C0R0TATIN6TWI N SCREW EXTRUDER ABSTRACT The influence ofa barre lvalv e onth edegre eo f fill ina corotating,twi nscre wextrude r (MPF-50)buil tb yAP VBake r was measured by opening the barrel and by residence time distribution measurements (RTD). The RTDs, which were measuredb ya radi otrace rtechnique ,wer echaracterise db y theirminimu m and average residence timesan d curvewidth . Measurements were made at a "low" and a "high" barrel temperatureprofile ,tw omas sflow san dthre escre wspeeds . The degree of fill and the minimum and average residence timeincrease dwhe nth evalv ei nth ebarre lwa sclosed .Whe n the rotational speed of thescre wdecrease d thehol d up in the extruder increased. The degree of fill during the measurements ranged from 30% to 56%, depending on the temperature,barrel-valv eposition ,rotationa lspee do fth e screws and the feed rate.Wit h a "low"temperatur e profile thewidt h ofth e curvewa s independent of thebarre l valve position. When the barrel valve was closed at "high" temperatures, only a minor increase in the curve width of theorde r oftwic eth emeasuremen t accuracywa s found. THIS CHAPTER ISBASE DUPO NTH E PUBLICATION: van Zuilichem, D.J., Jager, T., de Ruig, J.A.J., Spaans, E.J., (1989). The Influence of a Barrel-Valve on the Degree of Fill ina Corotatin g TwinScre w Extruder.J . Food Enq.. 10,241-254 . 138 6.1 Introduction In aco-rotatin g twin screw extruder, it isofte n difficult to make use of all the available barrel length.Approximatel y the first 75 % of the barrel length is being used mainly as a transport sectionwit ha lo wdegre eo f fillan dpressur ebuild up. Inth eremainin g2 5% amor eeffectiv eshea rinpu tan dhea t transfer arepossible .A chang e inth eproces svariable so fth e extruder such as the feed rate, rotational speed and barrel temperature,wil l have only amoderat e effect on the extrusion cooking process. Changes in the screw configuration such as reversed pitch sections, kneading paddles and other specially designedscre wsections ,ar ewidel yuse dt omodif yth eextrusio n cookingprocess .A greate rdegre eo f fillan dpressur ebuil du p result ina bette rextrude rperformance . Thereverse dpitc hsectio ni sth esyste mmos tofte nuse dt oshea r andhea tth eproduc t inth emiddl eregio no fth ebarrel .A mor e barrel barrel valve barrel barrel valve barrelvalv e open barrel valveclose d Fig 6.1 barrel-valve diagram (vanZuiliche m et al., 1989) recentcomponen ti sth ebarrel-valve .Th ebarrel-valve ,whic hi s shown in Fig. 6.1, should give better control of thedegre e of fill. Each screw is fitted with adis cwhic h almost completely blocksth ebarre lcross-section .Th etw odisc sar eoffse taxiall y becauseo fth eintermeshin ggeometry .A passagewa yi screate di n thesaddle-sectio no fth ebarrel ,whic hbypasse sthes efull-bor e restrictionso fth escre wi nwhic ha van evalv eo fspecia lshap e 139 is placed. A 90 ° rotation of this barrelvalve in either directionallow sth epassagewa yt ob efull yopene d (0° )o rfull y closed (90°) .Th e advantage of thisvalv e over reversed screw sections istha t its position canb e changed during operation. For the extrusion of plastics, Todd (1980) claims that this device influenceshol du pan dcompounding .Th e influenceo fth e barrel-valvepositio no nth edegre eo f fill isstudie db ymean s of residence time distribution (RTD) measurements and by inspectiondirectl yafte ropenin gth ehorizontall yspli tbarrel . In this case the RTDs and the degree of fill of a corotating, twin-screw extruder fed with maize gritswere measured with a radiotracer technique (van Zuilichem et al., 1988) at various screw speeds,mas s flowsan dbarrel-valv epositions . 6.2 Theory Levenspiel (1972)define d theE(t ) functionsuc htha tE(t)d ti s the fraction, at the exit, of the mass-flowwhic h has spent a time between t and (t+dt) in a system. The F(t)-functio n is obtained by integration togiv e acumulativ eRTD-functio n : Fit) = JE(t)dt (1) The average residence time, T (s), of the material in the extruder iscalculate das : T= ft£(t )d t (2) Thespecifi cfee drat e, I (-), i sdefine db yJage re tal .(1989 ) as: x =—e _ = _e_ (3) 2-Z-VU-p Q^ inwhic hQ (kg/s)i sth emas s flowwhic henter sth eextruder ,p 140 ±± Fig.6. 2 Geometryo fth escre wuse dfo rth eRT Dexperiment s (van Zuilichem et al., 1989) (kgm"3) is the density of feed material when it has a zero porosity, z isth e number of starts on one screw,V (m3)isth e chamber-volumeo f the first screw element in Fig. 6.2. U (1/s) is the rotational velocity of the screws and Qmax is the theoreticalmaximu mmassflo wfo rwhic hth echambe rvolume ,V ,i s fullyfilled . Thespecifi cfee drat eenable sth eeffect so fth e rotationalspee do fth escre wan dth emas sflo wQ t ob ecompared . Whenth eaverag eresidenc etim e Ti sknown ,th ehol dup ,H (m3), and thedegre eo f fill,D ,ca nb e calculated asfollows : H = l£ = DV^ (4) P "TOT inwhic h D isth edegre eo f fillan dV T0T isth etota lvolum e in theextruder .Th ecurv ewidt hca nb echaracterize db yth ePecle t number, the ratio of conveying and axial dispersion transport. Todd (1975)characterize dth ecurv ea sth erati oo fth etime sfo r 16% and 84%o f the tracer topas s thedetector .Thes eRTD sha d a simular shape to that of the axial dispersion model of Levenspiel.Th ePecle tnumbe rcoul db ecalculate dwit hthi smode l fromth emeasure d 16/84%ratio . Table 6.3 contains several conflicting relations demonstrating the complexity of the axial mixing in twin-screw extruder- cookers.Hold-u pan daverag eresidenc etim esee mt ob ea functio n ofa tleas tth emoistur econtent ,fee drate ,scre wspeed ,barre l temperature and screwgeometry . 6.3 Experimental The experimental extruder was a MPF-50 APV-Baker co-rotating, twin-screw extruderwit ha 'K-tron'twin-scre w feeder.Th efee d 141 materialwa smaiz egrit swit ha moistur econten to f23 %(w/w )£n d with adensit y at zeroporosit y of 1200kgm" 3.RTD s and degrees of fill were measured at three screw speeds (218,32 6 and 432 rpm) two mass flowrates (40 and 51 kgh"1) and two barrel temperatureprofile s (Table6.1) . A screw configuration with ten sections was used: a transport screw, the barrel-valve discs and eight sections in which the degree of fill has previously been measured (Fig. 6.2). It containedfiv edifferen telements :transpor tscrew ,singl estar t screw, self-wiping single start screw, barrel-valve discs and kneadingpaddles .Th ebarrel-valv e islocate d atth een do fth e fifth zone (Fig.6.2) .Th etota lreacto rvolum ei s10. 2dm 3.Th e RTDswer emeasure da sdescribe d byva nZuiliche me tal . (1988). A coincidence detector measures the "cu activity at the die- outlet.Maiz egrit s( 2g d.m. )soake di naqueou scuprou schlorid e (CuCl2) and dried to a moisture content of 12%wit h a copper content of 50 mg were used as tracer. The tracer was made Section number" Screw type Length Temperature profile (m) 'Low' 'High' m 1 Kneading paddles 0-088 30 51 2 Self wiping 0050 60 93 3 Kneading paddles 0050 60 93 4 Single-start 0050 90 121 5 Kneading paddles 0100 90 121 + barrel-valve plugs 6 Single lead 0-050 120 150 7 Kneading paddles 0062 120 150 8 Single-start 0050 120 150 "The section numbers refer to Fig.2 . Table6. 1Temperatur eprofiles ,th esectio nnumber srefe rt oFi g 6.2 (vanZuiliche m et al.,1989 ) radioactive by radiation in a gamma-plant. The detectors were shielded against background radiation by collimators having a slit type detection opening of 44x10mm . The minimum distance 142 betweenth ecollimato ran dth eho textrudat ewa s6 3mm .Wit hth e coincidence detector the average residence times and the curve shapes are not influenced by the measuring equipment (van Zuilichem et al. 1988). The average residence time was calculatedfro mequatio n (1)an dth edegre eo ffil lfro mequatio n (4). 6.4 Results The procedure forth e characterization of the curvewidt h used by Todd cannot be applied to the present RTDs, as they have a combination of a peak width and a tail which does not fit any curveo fth eaxia ldispersio nmode lo fLevenspiel .Th etime si n which 16 %an d 84%o f thetrace rhav e passed the detector give onlya n indication ofth esprea d ofth epea kwhic hneglect sth e shapeo fth etai l andwoul d result ina nunrepresentativ e curve spread. The ratio of the time in which 92%o f the tracer has passed and the average residence time is used here as an indicationo fth ecurv ewidth .Thi si smor erealisti ca si ttake s most of the tail into account. One experiment was repeated in order to study the reproducibility and the accuracy of the measurements.Th eaverag eresidenc etim ecoul db emeasure dtwic e and found tob ewithi n 0.1 s. Barrelvalve :ope n Fig. 6.3 gives the average degree of fill in the extruder at various feedrate san drotationa l screwspeed swhe nth ebarrel - valvei sopen .Whe nth erotationa lspee do fth escrew sincrease s thedegre eo f filldecreases .Wit ha "high "temperatur e profile anda maxima lrotationa lspee d (U=432RPM )th edegre eo ffil li s found to be independent of the feed rate. At the lower screw speeds,th e degree of fill decreases when mass flow increases. Thiseffec tca nb esee n inFig . 6.4,wher ea mas s flow increase of27 %fro m4 0kgh" 1t o5 1kgh" 1,result si na 23 %decreas e inth e degreeo ffil lfro m53 %t o41% .Wit hth elo wtemperatur eprofil e thedegre eo ffil lincrease swhe nth efee drat eincrease s (bya n increase inmas s flowo ra decreas e inth erotationa l speed). 143 degree of fill(%) 80- 70- 60 50 iO 30 20 10-. 12 3 4 5 6 7 8 zone number Fig. 6.3 Degree of fill in the co-rotating extruder (per zone) (vanZuiliche m et al, 1989) Whenth ebarre lvalv ei sopen ,th ecurv esprea dha sa maximu ma t adegre eo ffil lo f37 % Whenth emas sflo wincrease so rwhe nth e rotational speed of the screws decreases the curve spread decreases.Th eexperiment sa ta lo wbarre ltemperatur e resulted in only small differences in their degree of fill and curve spread. Barrelvalve :close d Theminimu man daverag eresidenc etim eincreas ewhe nth ebarrel - valve isclosed , (seeFig . 6.5) Thedegre eo f fill is13 %- 30 % greater than with an open barrel-valve. The degree of fill increaseswhe nth erotationa l speed ofth escrew sdecreases .A t the "high"temperatur e profile the degree of fill can decrease whenth emas s flow increases,whic h isdemonstrate d inFig . 6.6 byth emeasurement sa tU=21 8RPM ,o rremain sconstant ,whic h is demonstrated by the measurements at U=326 RPM. With a low temperature profile the degree of fill increases slightly when the feed rateincreases . 144 low temp. high temp. 0 (%) degree of fill 60 Fig. 6.4 Situation for aclose d barrel valve (van Zuilichem et al., 1989) The curve spread at the high temperature with a closed-barrel valve, depends on the screw speed and themas s flow rate.Whe n the degree of fill increases above 46 %, the curve spread decreases.A t low barrel temperatures the difference in spread of the measurements is less than the accuracy of the measurements. Thedegre eo f fill inth eextrude rvarie swit hrotationa l speed U and feed rate Q.A n illustration isshow n inFig . 6.6 fora n openbarrel-valve .Th epositio no fth ebarrel-valv eha da marke d effecto nth edegre eo ffil lo fsection s4 an d5 ,jus tbefor eth e barrel-valve. 6.5 Discussion InFig .6. 7 twomeasure dE(t)-curve swit hdegree so ffil lo f56 % and 46%ar ecompare dwit ha curv eo fAltomar ean dGhoss iwit ha 145 degreeo f fillo f 45%.Th ecurve swit ha degre e of fill of56 % and 46%ar ecomparabl e fort > T. Thedifference sbetwee nthes e curvesbefor eth eaverag eresidenc etim ear eo fmino r importance D(% )degre eo f fil o?| high temp. 60- O tmin t. U [-] A t Q= 51kg./h r 200- U=326r.p.m . 150- 100— ,opein , ... closed 432 U(r.p.m. ) r barrel valve position Fig. 6.5 Minimum and average Fig.6. 6 situationfo ra ope n residence timea sa functio no f barrel valve (van Zuilichem the barrel valve position (van et al.,1989 ) Zuilichem et al., 1989) forth ereacto rpropertie so fth eextruder .Whe n inth epresen t measurements the degree of fill increases from 46%t o 56%,th e curve width increases considerably. The curve widths of the measurementso fAltomar ean dGhoss iar eindependen to fth edegre e of fill.Th eextrusio ncondition so fth emeasurement s presented inthi spape r and that ofMoss o etal . (1981)an dAltomar e and Ghossi (1986) are not totally comparable but show some similarities in their results. In all these measurements the degreeo ffil lincrease swhe nth erotationa lspee do fth escrew s decreases.I nth emeasurement so fAltomar ean dGhoss ian dth eon e presented here,th e shape of the RTD-curves are independent of thefee drat ea tlo w (107 °C)barre ltemperatures .Moss o etal . andAltomar e and Ghossi find an increase inth e degree of fill 146 when the mass flow is raised. In the present measurements the degreeo ffil lca nincrease ,decreas eo rremai nconstan twhe nth e mass flow rate increases,dependin g on the screw speed, barrel temperaturean dbarrel-valv eposition ,whic hi srelevan tonl yt o this type of extruder. This feature is characteristic for the investigated combination ofextrude ran dbarrel-valve . Table 6.2 Extrusion conditions for RTD measurement in co- rotating, twin screw extrusion cooking (van Zuilichem et al., 1989) Aulhor(s) Bounie(1986) Altomare & Mosso el nl. Lim el at. Ghossi(l986) (1981) (1985) Extruder type Clextral BC-45 Werner & Pfleiderer Clextral BC-45 Clextral BC-45 ZSK - 57 Length (m) 0-6 IHI 0-6 0-6 L/D ratio 9:1 16:1 9:1 9:1 Die size (mm) 22-25 x 1-5 3-2-7-9 ? •) Material Wheat starch Rice flour 42% wheat starch wheat flour 20% maize starch 6% sodium caseinate 1l%soy a protein 20% saccharin 1°/..sal t Feed rate (kg h ') 29-7-333 68-2-227-3 26-50 20 Temperature CC) 165-200 79-177 192 9 Barrel moisture 15-45 10-28 13-21 25 content (%w) Reversed screw zone with and without with with with and without Viscous heat 300-460 350-480 ? i dissipation (kj kg"') The RTD and the degree of fill in co-rotating twin-screw extruderso fsimila rdesig nt otha tuse d inthi s investigation, has been measured by several investigators. They all used materials containing over 50% starch (Table 6.2) and they measured variations inth eminimu m residence time,th e average residencetime ,th ehol du pan dth ecurv esprea dwhe nchange si n the moisture content, the feedrate, the temperature and the rotational speed of the screws were made (Table 6.3).Bouni e (1986) found thatth e introduction of a reversed pitch section increases the curve-width , the average residence time, the viscousdissipatio nan dth eminimu mresidenc etime .Altomar ean d Ghossi (1986) have calculated degrees of fill, and found a maximum of 46%.At increasing feed rates,th e measurements of Mosso et al.(1981) and Altomare and Ghossi show an increase in thedegre e of fill. The average residence time and the degree of fill seem to be dependento na larg enumbe ro fvariable swhic hmake si tdifficul t 147 Table6. 3Averag eresidenc etime ,hold-u pvolum ean dcurv esprea d for co-rotating twin screw extruders vs. changes in moisture content,specifi cfee drate ,barre ltemperatur ean ddi ediamete r (vanZuilichem , 1989) Change in H Spread Bounie (1986)wit h —Moisture content from 15t o P P reversed screw 45% (w/w) section -Specific feed rate, constan l l Q,U changes from 110 to 70 rpm —Temperature inHTS T zone i I from 165 to200° C Bounie (1986)withou t -Moisture content from 15 to P P reversed screw 45% section -Specific feedrate , i P constant Q,U changes from 110 tp 70rp m Altomare &Ghoss i -Moisture content from 10t o P P (1986) 28% (w/w) -Specific feedrate , P P constant U,Q changes from 68 to 227k gh" 1 -Specific feedrate , constantQ ,U changes from 400 to 150rp m -Temperature inHTS T zone from 79 to 177°C -Die resistance Lim et al. (1985) -Specific feedrate , with reversed screw constant Q,U changes from section 120 to 60rp m Lim etal . (1985) -Specific feedrate , without reversed constantQ ,U changes from screw section 120 to 60rp m Mosso et al. (1981) -Moisture content from 13.4 to21.2 % (w/w) -Specific feedrate , constant U,Q changes from 26 to 50k gh" 1 -Specific feedrate , constant Q,U changes from 98 to 58rp m This s tud y -Temperature inHTS T zone from 120 to 150°C -Specific feedrate , constantU ,Q changes from 40 to 50k gh" 1 -Specific feedrate , constant Q,U changes from 432 to 218rp m —Barrel-valve resistance, open toclose d P An increase of thevariabl e increases thevalue . i Inverse effect. m Mixed effect - No significant effect toforecas tth edegre eo f fill,o rt odescrib e a 'standard'RT D 148 D(%) — measured curve 56 — measured curve 46 Altomare &Ghoss i 45 0.£ = APV-Boker MPF-50 I D= WSP ZSK57 L/D.16 t/X (-) 3 Fig. 6.7 RTD curves of co-rotating extruders (van Zuilichem et al., 1989) ina co-rotating ,twin-scre w extruder. 6.6 Conclusions Thebarrel-valv e providesa mean so fvaryin g thedegre eo ffil l andth eresidenc etim edurin gextrusion-cooking .Whe nth ebarrel - valve isclose d the increase inth edegre e of fill and average residencetim e is 16%t o30% . Atth elo wbarre ltemperatur eth eshap eo fth eRT Dcurv ewa sno t influenced by the barrel-valve position. The degree of fill increased slightlywhe nth efee drat e increased.A n increase in theeffectiv efee drat eeithe rb y increasingth emas s flowrat e orb ydecreasin g the screw speed gave similarresults . Withth ehighe rtemperatur eprofil eth edegre eo ffil lincrease d whenth e rotational speed ofth e screwsdecreased . An increase in the mass flow rate reduces the degree of fill when the rotationalspee do fth escrew swa slo w (218RPM ). At highe rscre w speedsth edegre eo ffil lwa salmos tindependen to fth emas sflo w 149 rate. Theshap eo fth eRT Dcurv ewa smainl ydependen to nth erotationa l speed, the mass flow rate and the degree of fill. When the barrel-valvewa sclose dth eincreas ei nth ecurv ewidt hwa sabou t twiceth eaccurac yo fth emeasurement .Whe nchange s inth emas s flowo rth erotationa lspee do fth escrew sincrease d thedegre e of fillabov e 37%fo rth eope nbarrel-valv e (orabov e45 %fo ra closed barrel-valve), the curve spread decreased. When the barrel-valvewa sope nth ecurv esprea ddecrease dwhe nth edegre e of filldecrease d below37% . A barrel-valve provided on a co-rotating twin-screw extruder makes possible a most efficient utilization of the available reaction volume of a particular screw combination by a better control of the degree of fill. This implies a minimal screw length fora give n extrusionprocess . 150 References Altomare,R.E. ,Ghoss i P. (1986).A nanalysi so fresidenc etim e distribution patterns in a twin screw cooking extruder. Biotechnology progress. 2,157-163 . Bounie, D. (1986). Etude de l'ecoulement et du melange axial dans un extrudeur bi-vis corotatif. These de l'Universite desScience s etTechnique sd uLanguedoc . Jager, T., van Zuilichem, D.J., Stolp, W., van't Riet, K. (1988). Residencetim edistribution s inextrusio ncooking , Part 4: Mathematical modelling of the axial mixing of a conical, counter-rotating, twin-screw extruder processing maizegrits .J . Food Eng..8 , 157-72. Jager, T., van Zuilichem, D.J., Stolp, W., van't Riet, K. (1989).Residenc etim edistribution s inextrusion-cooking , Part 5: the compression zone of a conical, counter-rotating,twin-scre wextrude rfe dwit hmaiz egrits . J. Food Eng..9 ,203-218 . Lim,J.K. , Wakamiya, S.,Noguchi ,A. , Lee,C.H . (1985). Effect ofrevers escre welemen to nth eresidenc etim edistributio n in twin-screw extruders.Korea n J. Food Sc. Technol..17 , 208-212. Levenspiel, O. (1972). Chemical reaction engineering. 2nd ed. JohnWile y and Sons.Ne wYork . Martelli, F.G. (1982) Twin Screw Extruders: A Basic Understanding.Va nNostrand-Reinhold , NewYork . Mosso, K., Jeunink I., Cheftel J.C. (1982). Temperature, pression et temps de sejourd'u nmelang e alimentaire dans 151 un cuiseur- extrudeur bi-vis. Ind. Aliment. Aaric.. 99 (1-2)5-18 . Todd, D.B. (1975). Residence time distribution in twin screw extruders.Polvm .Ena . Sc.. 15,437-442 . Todd, D.B. (1980). Energy control in twin-screw extruders, TechnicalPapers ,37t hAnnua lTechnica lConference ,Societ y of PlasticsEngineers ,Ne wOrleans ,220-221 . van Zuilichem, D.J. (1987). Extrusion-cooking in retrospect. Proceedings Koch-Extrusion 87, Zentral Fachschule der Deutschen Susswarenwirtschaft,Solingen ,BRD . van Zuilichem, D.J., Jager, T., Stolp W., de Swart, J.G., (1988a). Residenc etim edistribution si nextrusion-cooking . Part 1:Coincidenc e detection.J . Food Ena..7 , 147-58. 152 Notation D Degree of fill [-] H Hold-up [m3] I Specific feed rate [-] Q Feed rate [kgs"1] Q Maximal feed rate [kgs"1] max *• t Time [s] tx Passing time of x% of tracer along the detector [s] U Rotational speed of the screws [s"1] V Chamber volume [m3] 3 Vt Reactor volume [m ] z Number of thread starts on one screw [-] T Average residence time [s] p Specific density [kgm"3] 153 CHAPTER7 MODELLING OF THE CONVERSION OF CRACKED CORN BY TWIN-SCREW EXTRUSION-COOKING ABSTRACT Inth e search foralternativ e uses forby-product s of corn wet-milling, enzymatic conversion of cracked corn to fermentation substrates by means of twin-screw extrusion (tse) was investigated. The conversion was initiated with starchdisclosur ean dliquefactio ni ntw oextrusio noperati ons. Up to 0.5 %w/ w heat-stable a-amylasewa s added at a mass temperature of 120 °C. The extrudate was batch- saccharifiedusin g0.0 5t o1 % gluco-amylas ea ttemperature s ranging from 50t o7 0 *C,2 0t o4 0% dr ymatte ran dp H4.5 . The experiment was conducted according to the response surface analysis method, resulting in an empirical model whichdescribe sth ecours eo fth econversio n intodextrose - equivalent units during saccharification atvaryin g enzyme concentrations,temperature s anddry-matte rcontents . Resultssho wtha tth egluco-amylas e concentration isb yfa r the most important factor in the conversion. The plot obtained for gluco-amylase concentration versus conversion time isapproximatel y linear. THISCHAPTE R HAS BEEN PUBLISHEDAS : van Zuilichem, D.J., van Roekel, G.J., Stolp, W., van 't Riet, K. (1990). Modelling of theEnzymati c Conversion ofCracke dCor nb yTwi nScre wExtrusio nCooking .J .Foo d Enq.. 12.13-28 . 154 7.1 Introduction Beforewet millin go fcor ni sperforme d ina starc hplant ,i ti s necessaryt oseparat ecracke dcor nfro mwhol ecor ngrain sbefor e the corn entersth e steeping process,a sth epresenc e of free, soluble starch in the damaged grains and dust would cause the steeping tanks to slit up. Depending on quality, origin, transport circumstances and the number of transshipments, the separated cornrepresent s 2- 1 5% o fth etota lvolume . Thiscracke dcorn ,whic hha snearl yth esam ecompositio na swhol e corn, is used in cattle-feed. In this experiment the possibilities of converting cracked corn into a fermentation substratear einvestigated .Thi ssubstrat ewoul dfin dit sus ei n ethanol orothe r fermentations (Zhusman 1979;Cha ye tal .1984 ; Parke tal . 1987). Unlikei nth ewet-millin gprocess ,wher eglucos eca nb eobtaine d frompur estarch ,th econversio no fcor nstarc hint oit sdextros e units for fermentation purposes iscarrie d out inth e presence ofal lth eothe rcor ncomponents .Afte rfermentation ,th eethano l must be recovered from the fermentation product. The advantage istha tn oseparatio no rpurificatio nothe rtha nethano lrecover y isnecessary . The conversion investigated was carried out in a twin-screw extrusionprocess ,usin gamylolyti cenzymes .Th erequire d steps in breaking down raw, starchy material into glucose and oligosaccharideswer eprimaril yperforme db yextrusion-cooking . The following sections describe the processvariable s of major importancei nth econversio nan dho wt oquantif ythei reffec to n the contents of the fermentation substrate. By applying a statistical technique referred to as Surface Response Analysis [Cochran et al,1957], a model describing the course of the conversionb ymean so fth eDextros eEquivalen t (DE)valu ei nth e substrateha sbee n calculated. 7.2 Theoretical background The degradation of raw, starchy material into a substrate 155 consists of four distinct steps. First, the grain and cell structure -therigi dglute nmatri x inwhic hth estarc hgranule s areembedded -mus tb eruptured .Then ,th ecrystallin e structure of the granules must be broken to create access for chemical degradation.Thes etw ostep sar ereferre d toa sdisclosure ,an d require up to 1M J per kg of raw material. In the third step, onceth eamylos ean damylopecti nchain shav ebee nmad evulnerabl e andca nb edissolve di nwater ,gelatinisatio nwil loccur .At thi s timeth estarc hi sread yfo rliquefaction .Th ea-amylas eactivit y willb eabl et odivid eth emacromolecules ,thereb ybreakin gdow n themacromolecula rstructure .Fourt han d last isth econversio n of the remaining starch fractions (1t o 1000 glucose units in size) into glucose and oligosaccharides by means of gluco- amylolyticenzymes ,i.e .th esaccharification . The firstthre eo fth eabov ementione d stepsca nb ecarrie d out during extrusion cooking at high temperature, high shear, low moisture contents and in a short time (of the order of a few minutes) (VanZuiliche m et al. 1980;Chouve l et al. 1983). The saccharificatio n will take one to several tens of hours, depending on the desired yield of conversion. The extrusion process cannot complete the conversion, itmus tb e followed by a batch orcontinuou s saccharification step. Fig. 7.1 shows a model extruder configuration for performing disclosure, gelatinisation and liquefaction. In this investigation themode l extruder configuration is simulated by two sequential runsthroug h asingl eextruder . The first extruder section operates at ahig h temperature (150 °C)an d ahig h drymatte r content (70-85)% wet basis), causing a very high shear rate. In the second section the dry matter contenti slowere dt o (30-40)%b yaddin ga heat-stabl ea-amylaB e solution in water, starting liquefaction of the gelatinising starch.Th etemperatur ei nthi ssectio nshoul denabl ehig henzym e activity for a shortperio d of time.Th ehalf-lif e ofTermamy l 120La-amylas e is1 5minute sa t11 0 °C.Enzym einactivatio nwil l alsob e high,bu twithi n the short duration of activity needed 156 CRACKED CORN WATER WATER BUFFER BUFFER AMYLASE GLUCO-AMYLASE M/ \|/ DISCLOSURE GELATINISATION SACCHARI- &LIQUEFACTIO N FICATION Fig. 7.1 A model extruder configuration capable of disclosing, liquefyingan dinitialisin gsaccharificatio no fra wcracke dcor n (vanZuiliche m et al., 1990) thiswil lno talte rth eeffectivenes so fth eprocess .Heat-stabl e a-amylaseshav e an optimum activity at 120 'Cwhile ,unde r the givenconditions ,wate ractivit yfo renzym ereaction swithi nth e extruder is still close to 1 (Linko et al, 1983a). After liquefaction the starch degradation expressed in DEunit s is2 to5 .Th eeffectivenes so fgluco-amylas edurin g saccharification is higher when liquefaction results in a lower DE value (Marshall, 1980). In the third section the saccharification is started by injecting a gluco-amylase solution before the die. Thisi sdon et oachiev eadequat emixin gan dt ostar tth ereactio n ata relativel yhig hdr ymatte rconten t (25-35)% .Gluco-amylase , however,i smor eeffectiv ea tabou t6 0 °Can da tp H4.5 ,s othi s last section of the extruder must be cooled rapidly, and the injected enzyme solutionha st ob ebuffere d atp H 4.5. Thebarre l lenghto f sucha mode lextrude rmus tequa lo rexcee d 30diameter st ob eabl et operfor mth ethre estep s sequentially which israthe r impractical. At present, a useful model to succesfully describe the above- mentioned bioconversion isno t available.Th eMichaelis-Mente n kinetic model was developed to describe enzyme-catalysed reactions (Lehninger, 1970). However, it has not yet proved succesfuli ndescribin genzymati cconversio nprocesses ,disturbe d 157 by the presence of a variety of other constituents, i.e. proteins,fat san dcellulos e incorn . An empirical relation to fit the reaction rate r (the glucose production in kgm"3s"1) dependence on substrate- (Cs), water- (Cw), andenzym e (Ce)concentration s isa s follows (Reinikainen et al. 1986): r= k-(Cs)a-(Cw)b-(Ce)c U) This relationship can be simplified to a model describing a saccharification in terms of DE value (%w/w) during the conversion. Itwa s found, however,tha t the following equation fulfills the samepurpos ewit h an equal or even smaller error, (theD Evalu ei sgenerall yexpresse d inweigh tpercent so na dr y basis): DE= — (2) A+B-t wheret isth etim ean dA an dB ar econstants .Thi sequatio n is similart oth eproductio n equationwit ha secon d orderrate . Figure7. 2 illustratesth esignificanc eo fA an dB .Thi ssecon d model was chosen because of the practical use when fitting conversiondata .Wit hknow nD Evalue sa tcertai ntime sdurin gth e conversion, the following form of equation (2) can easily be fittedusin g linearregression : -£- =A+B-t (3) DE The constants A, (in (% w/w)"1.s+1) and B, (in % w/w)'1 characteriseth e initialconversio n ratean dth e final levelo f conversion, respectively, and as such describe the progress of a conversion. Usingsurfac erespons eanalysis ,th eeffect so frelevan tproces s variables on a conversion, characterised by A and B, can be quantified. 158 • Time[s ] Fig. 7.2 Typical plot of the progress of a saccharification reaction. The constants 1/A and 1/B characterise the initial conversion level and the final conversion level, respectively (vanZuiliche m et al., 1990) Previous work (Linko et al, 1983a,b) revealed that during extrusion the effect of temperature,dr y matter content,scre w geometry, feed rate and othervariable s showed certain optima, often dictated by the optimum conditions forus e ofa-amylase . In this investigation, therefore, the a-amylase dosage was selecteda sth eonl yvariabl ei nth eliquefactio nprocess .Othe r variableswer eselecte dfro mresult so finvestigation so nstarc h extrusion in the presence of amylolytic enzymes (Linko et al, 1980; Hakulin et al, 1983; Linko et al, 1983 a,b). During saccharification, the effect of the gluco-amylasedosag e iso f majoreconomi cinterest :i na saccharificatio nproces sth egluco - amylase cost largely determines the process cost. In order to confirm the suggested linearity of the reaction rate with the gluco-amylase concentration, the temperature and dry matter content of the substrate during saccharificationwer e selected asvariable st ob e investigated. Becauseo fth erelativel y long residence time in the saccharification, these variables have great influence onenzym eactivity ,an dthu so nproces scost . 159 7.3 Experimental 7.3.1 Setup The experiment was set up to determine the effect of the a- amylase concentration during liquefaction, and of the gluco- amylaseconcentration ,temperatur ean ddr ymatte rconten tdurin g saccharification.Th estatistica lmode luse sa mea nvariabl eX„ , and fourcode dvariable s forth evariable s inth eexperiment . Eacho fthes evariable swa sse ta tfiv edifferen t levels.Tabl e 7.1 showsth ecombination so fdifferen tlevel sfo reac hvariabl e used inth e 31conversio nexperiments . Thevariable swer ecode dusin gth e following relations: a-Amylase: X^ = 2.8log [a-amylase] +3 .7 (4) Gluco-amylase X, = 31og[gluco-amylase] +2 (5) Temperature X, r-60 <«> Drymatte rconten t: DM-30 (7) 7.3.2 Methods andmaterial s The extrusion part of the experiment could not beperforme d on onelong-barre lextruder .Therefor edisclosur ean d liquefaction were carried out sequentially in a Cincinnati Milacron CM 45 conical counter-rotating twin-screw extruder (screw type SK 1552/100B). Thedisclosur este pwa sth esam efo ral lexperiments ,s oa larg e quantity ofdisclose d cracked cornwa sprepare d at8 5% D M (1.5 h, 130 °C)an d 150 °C barrel temperature.A t 84 kgh"1, 83RP M 160 Table 7.1 Experimental design (vanZuiliche m et al., 1990) Exp. Coded variables Actual physical valuesb \r„ X„ X, x2 x, x4 fa-amj fg-amj Temp. DM (% w/w) (% w/w) CQ (% w/w) 1 [ — j — 1 — 1 — 1 003 01 55 25 2 -1 -1 -1 0-105 0-1 55 25 3 1 -1 -1 -1 003 0-47 55 25 4 -1 -1 0105 0-47 55 25 5 1 -1 -1 -1 0-03 01 65 25 6 -1 -1 0105 01 65 25 7 I -1 -1 003 0-47 65 25 8 -1 0105 0-47 65 25 9 I -1 -1 -1 003 01 55 35 10 -1 -1 0105 01 55 35 11 ] -1 -1 0-03 0-47 55 35 12 ] -1 0105 0-47 55 35 13 ] -1 -1 003 01 65 35 14 ] -1 0-105 . 01 65 35 15 1 - 1 003 0-47 65 35 16 J 0-105 0-47 65 35 17 1 -2 0 0 0 0 0-214 60 30 18 ] 2 0 0 0 0-262 0-214 60 30 19 1 0 -2 0 0 0-053 0-047 60 30 20 J 0 2 0 0 0-053 10 60 30 21 1 0 0 -2 0 0053 0-214 50 30 22 1 0 0 2 0 0053 0-214 70 30 23 1 0 0 0 -2 0053 0-214 60 20 24 1 0 0 0 2 0053 0-214 60 40 25 1 0 0 0 0 0053 0-214 60 30 26 1 0 0 0 0 0053 0-214 60 30 27 1 0 0 0 0 0-053 0-214 60 30 28 1 0 0 0 0 0-053 0-214 60 30 29 1 0 0 0 0 0-053 0-214 60 30 30 1 0 0 0 0 0-053 0-214 60 30 31 1 0 0 0 0 0053 0-214 60 30 "Centralcomposit e rotatable second-order designfo r four variables atfiv e levels, "cr-am: alpha-amylase,g-am :gluco-amylase . andapproximatel y 23k Wmechanica lpower ,a littl ewate r (4kgh " 1)wa sadde dt opreven tth eextrude rfro mplugging .Th eproduct , fractionated using a 8 x 4mm diameter die with cutter, was crispy, yellow and fluffy. These particles were ground in a Conduxpin-grinde r and stored atroo mtemperature . Liquefactionwa scarrie dou tusin g5 differen tconcentration so f 161 a-amylase (NOV OTermamy l 120L ).Th edisclose d meal, 85% DM , was extruded at amas stemperatur e of 120 °C.Abou t 0.1 m from thefeed-poin tth eenzym esolutio nwa sinjected .Thes esolution s wereprepare d soa st obrin gth eD Mconten tdow nt o 30% an dt o attain the correct a-amylase concentration. The solutions were buffered with a Sodium acetate/acetic acid bufferat p H 6,but liquefaction without buffer showed no variations from pH 6 greatertha n0. 5pH .Durin gliquefaction ,soli dmatte rthroughpu t was only 11 kgh"1wit h 21kgh" 1 enzyme-solution added at 32RP M andapprox .2 k Wmechanica lpower .Th esample swer ecollecte di n plasticboxe san d stored at 0°C . Saccharification was then carried out over 48 h in a shaking incubator. The 31 saccharification solutions were prepared by adding acetic acid to bring the pH to 4.5, adding water to correctt oth erigh tD M (DMdeterminatio n at 105 °Covernight) , blending for 1minut e and allowing the mixture to heat up for exactly 45 min before injection of the glucoamylase (Gist brocadesAmigas eGM ) Sampleswer e taken at 0, h, 1h, 3h, 8an d 48h after injection.Thes e 30m l sampleswer e frozen at -6 °C immediately. Problemsaros epreparin g experimentnr .2 4an dth e liquefied sample had tob evacuum-drie d forte nhour s up to 40 % DM. During saccharification DM increased 0.5 % at most. DE analyseswer ecarrie dou taccordin gt oNierle-Tegge ,precede db y a 10minute s 80% ethano l extraction. 7.3.3 Calculation The constantsA and Bwer e then calculated for each conversion usingth eGaus slinea rregressio n algorithm. Inregression ,t* 0 was taken to be 15 min before injection of the enzyme. This virtual starting point was taken to correct for the start of conversion during extrusion. Then for both A and B a surface responseanalysi swa sperformed .I nthi sanalysi sth ereciproca l values ofA and Bwer eused , asthe y appear inth e denominator ineqn .(2) . 162 RESULTS Table 7.2 showsth e resultso f the analysis.Th e columns '1/A' and '1/B'giv eth ecalculate d firstan dsecon dorde reffect sfo r eachvariable ,an d all interaction effects.Significanc e (90% confidencelevel )wa sdesignate db yth eeffec tbein ggreate rtha n the corresponding standard error. The columns 'S' show the significance of each effect(•- •denotin g insignificance). The Table7. 2Result so fsurfac erespons eanalyse s (vanZuiliche me t al., 1990) Effect HA .S" 1/B .V First-order: Mean K 137-1 67-7 a-Amylase b, 28-5 0-3 — Gluco-amylase b, 83-8 3-0 Temperature b, 23-3 -2-0 Dry matter b, -19-7 -0-1 — Second-order: (a-Amylase) bu - 10-9 1-3 (Gluco-amylase) b„ 17-2 -01 — (Temperature) bu 2-5 — 0-7 (Dry matter) b^ 18-3 -0-1 — Interactions: cc-Amy./Gluco br. 34-0 -1-4 ct-Amy./Temp. bt, 24-6 - 0-4 '— a-Amy./Dry mat. *M -3-4 — 0-8 Gluco/Temp. A„ 16-8 1-6 Gluco/Dry mat. />24 -9-9 -01 — Temp./Dry mat. *34 -9-9 0-3 — Means of squares: ./' ./' First-order 52 548-1 17-7 80-6 221 Second-order , 5 830-1 2-5 15-1 41 Exp. error 230-9 1-7 Lack of fit 2 827-5 3-5 Standard errors (i, j= 1.....4): Per observation 14-07 1-49 s.e. bj 2-87 0-25 s.e. bu 2-60 0-23 s.e. b„ 3-52 0-30 *f values for 'firstorder 'an d 'second order1 are statistical 163 valuesexpressin g thevalidit y ofth eanalyse s forbot h1/ Aan d 1/B.Th ecritica lvalu ea t9 0% confidenc eleve lar e2.6 1 (first order effect) and 2.32 (second order effect). Table 7.2 shows that first and second order effects are significant for both analysesa t9 0% confidenc elevel . Fordetail sconcernin gth estatistica lanalysis ,se eCochra nan d COX (1957). The effects in Table 7.2 are constants in the following equations, with which a conversion of cracked corn into a fermentationsubstrat eca nb edescribed .Onl ysignifican teffect s aretake n intoaccount . 1/A = 137.1 +' 28.5 X, + 83.8 X2 + 23.3 X3 - 19.7 X4 (8) 2 2 2 - 10.9 X, + 17.2 X2 +18.3 X4 + 34.0 X,X2 +24.6 X,X3 + 16.8 X2X3 - 9.9 X2X4 - 9.9 XjX4 1/B = 67.7 + 3.0 X2 - 2.0 X3 (9) 2 2 + 1.3 X2 - 0.7 X3 - 1.4 X,X2 +0.8 X,X4 +1.6 X2X3 With< (8) and (9) 1/A and 1/B can be found for a certain combination ofvariables .Wit hknow nA an d B,th eprogres so fa conversion, is described by equation (2). For instance a conversionwit hal lvariable sse tt o0 ;0.05 3% Termamyl , 0.214 % Amigase, 60 °Ca t 30% DS ,i sdescribe d by: DE = (10) 0.00729+0.0148t Wheret isexpresse d inhour san d DE isexpresse d in (%w/w). 7.5 Discussion 164 7.5.1 Extrusion (disclosure & liquefaction) Inextrusio ntw omajo rsource so ferro rar erecognised .Th efirs t is in the grinding of the expanded and cooled extrudate after disclosure.Th e exposure of the disclosed material to air,th e cooling down after disclosure and the reheating during li quefaction probably cause formation of some structure in the disclosed extrudatewhic h isagai ndestroye d bygrinding .Also , exposuret ooxyge ni sknow nt oinhibi tnativ ea-amylas eactivity . Second,storag eo fth eliquefie dextrudat ea t0 " Cdi dno tfull y stopenzymati c activity. DE determinations immediately after extrusion and just before saccharification varied from 0 to 7 % w/w in one experiment. Because of the orthogonal setup of the experiment the small significance of the a-amylase effect does not affect the value ofth eothe rthre eeffects . Figure7.3(a )show sa fa rfro mlinea reffec to fa-amylas e on1/ A (thecontinuou s drawn line incorporates both first and second order effects, the dotted line shows the first order effect only). This agrees with the general theory that the use of a- amylasedurin gliquefactio nfollowe db ysacchari ficatio nprevent s retrogradation (the formation of inhomogeneties during liquefaction)rathe rtha nactuall yinitialisin gconversion .Usin g higha-amylas econcentration sdurin gliquefactio ndoe sno tresul t inhighe r initialconversio n rates1/A . Figure7.5(a )show sth e effecto fa-amylas e concentration onth e finalconversio n level 1/B.Th e derived relationship is significant,bu t stayswithi n a range of four DE-units, which is in the same range as the combined erroro f regression (1%)an d DE-determination (2.1%). Figure7.4(a )show sa decreasin g activity ofgluco-amylas ewit h increasing a-amylase concentrations at low gluco-amylase concentrations. 7.5.2 Saccharification The gluco-amylase effect on 1/A (Fig. 7.3(b)) obviously is partially a second-order effect.Th edecreas e in the effect at concentrationso f0. 5% an dupward scorrespond st oth emanufactu - 165 1/A(1/s) — amylase 300- optiamyl — amylase 1st order effect amigase — 1st order effect 200 100 ® 0.00 0.10 0.20 0.30 0.00 0.40 0.80 1.20 alfa-amylase (%w/w) gluco-amylase (%w/w) 1/A (1/s) 300 — temperature 1/A (1/s) •DM --- 1st order effect 300- • 1st order effect 200 200- 100: ® 65 70 20 30 40 temp.C O DM content (% w/w) Fig 7.3 The effects of the four conversion variables on the initialconversio nrat e1/A .Al lscale sar elinea r (vanZuiliche n et al., 1990) rer'sdata .A t low concentrations the effect canb e considered linear.Fig .7.3(c )show sa linea rincreasin gtemperatur eeffect . Thisca nb eexplaine db yth efac ttha tthi si sth eeffec to nth e initial conversion rate 1/A, where no inactivation has yet occured. 166 5? 1.000- 70 0.470- •0.2140- « 60- 0100 55 0.047- 0.001 0.03 0.053 0.105 0.262 0.047 0.100 0.214 0.470 1.000 alfa amylase concentration (%w/w) gluco amylase concentration (%w/w) Fig. 7.4a The combined effects Fig. 7.4b The combined ofa-amylas ean dglucoamylas eo n effects of glucoamylase and theinitia lconversio nrat e1/A . temperature on the initial Bothscale sar elogarithmi c (van conversion rate 1/A. The Zuilichem et al., 1990) horizontal scale is logarithmic (van Zuilichem et al.,1990 ) The significant effect of dry matter content (Fig. 7.3(d)) is decliningsteadily .Th einsignifican tincreas ea tD Mgreate rtha n 35% isdu et oth evacuu mdryin gbefor e incubation inexperimen t no.24 . Figure 7.4b clearly shows the combined linear effect of temperature and gluco-amylase on 1/A. Inth e linear regression calculations,A was found asth e intercept.Thi s intercept was relatively small, causing an estimated error of 30% (mean regressionerro rfo r1/ Ai nth e3 1regressions) .Nevertheles sth e total firstorde reffec t issignificant .Th etota l secondorde r effect issignifican t at 0.87 confidencelevel . The gluco-amylase effect on the final conversion level (Fig. 7.5(b)) shows a declining slope. The temperature effect (Fig. 7.5(c))show ssever egluco-amylas e inactivation overth eperio d of saccharification. The dry matter content (Fig. 7.5(d))doe s notaffec tth e final conversionlevel . Figure 7.6(a), the combined effect of gluco-amylase and a- amylase,show stha ta thig ha-amylas econcentration sth eeffec t ofgluco-amylas e decreases,du et oth e loweractivit y ofgluco amylase on more degraded starch. Figure 7.6(b) shows that the 167 1/B (-) VB(-) - amylase — amylase 80 i amigase 80- optiamyl -1st order effect ---1st order effect 75 75 70 70 ® 65 75- 60 60 1 1 1 1 0.00 0.10 0.20 0 30 0.00 0.25 0.50 0.75 1.00 amylase (%w/w) gluco-amylase(%w/w) 1/B (- 1/B 80- temperature -DM 80- •linea r effect 75 75- ® 70- 70H 6 5- 6 5- 60 60 50 55 60 65 70 20 25 30 35 40 temperature ( C) DMconten t (%w/w) Fig. 7.5 The effects of the four conversion variables on the finalconversio nleve l1/B .Al lscale sar elinea r (vanZuiliche m et al., 1990) inactivation ofglucoamylas e over aperio d of time (denoted by theleve l1/B )i smor esever ea tlo wglucoamylas econcentrations . B was calculated from the slope in linear regression. The accuracy oftha tcalculatio nwa sapproximatel y 1%. 168 * 60 0047 0.100 0.214 0.470 1000 0.001 0.03 0.053 0.105 0.262 gluco amylase concentration [%w/w] alfa amylase concentration (%w/w} Fig. 7.6aTh e combined effect Fig. 7.6b The combined efect of a-amylase and glucoamylase of glucoamylase and on the final conversion level temperature on the final 1/B. Both scales are conversion level 1/B. The logarithmic (van zuilichem et horizontal scale is al.,1990 ) logarithmic (van Zuilichem et al., 1990) 7.6 Proposed process Figure 7.7 shows a proposed process design for cracked corn conversion,followe db ya cell-recycl efermentatio nprocess .Th e cracked corn is extruded and saccharified in a continuous reactor. A separating operation (i.e.sieving ) removes solids from the dextrose mass before it enters the fermentor. The biomass in the fermentation product is separated by centrifugation and recycled into the fermentor. The remaining fluid isdestilled ,producin gethanol . Cracked Ethanol Corn ^W/4AAAA%- 4> A Q Screen Centrifuge Fig7. 7A propose dproces sfo rconversio nan dfermentatio no fra w cracked corn (vanZuiliche m etal. ,1990 ) The estimated cost of this simplified process, with capital 169 depreciationove rthre eyears ,i sapproximatel y$ 130, -pe rtonn e substrate for the conversion process. This results in an attractive rawmateria lcos t forth e fermentationprocess . 7.7 Conclusions The described conversion wasprove n to be possible.Th e use of one continuous extrusion process instead of the two steps describedwoul dhoweve r improveth eprocess . Conversiono fcracke dcor n (70%o fD Mstarch )int oa fermentatio n substrateusin g extrusionunde rth egive ncircumstance s (0.053% Termamyl a-amylase during extrusion, 0.214% Amigase gluoo- amylase,6 0 °Can d 30%D Mdurin g saccharification) results ina DE4 0substrat ewithi n 1h ,an d DE6 0i napproximatel y 4h .Th e total process cost for such a DE 40 substrate (including depreciation,excludin gra wmateria lcost )woul db earoun d$ 15, - pertonn esubstrate . Thepresente dmetho do finvestigatio nprovide da usefu ltoo lfo r calculating andoptimisin g sucha conversio nprocess . 170 References Chay, P.B., Chouvel, H., Cheftel,J.C. , Ghommidh, C.,Navarro , J.M. (1984). Extrusion ofhydrolyse d starch and flourst o fermentation substrates for ethanol production. Lebensm. Wiss. &Technol. . 17,257-267 . Chouvel, H., Chay, P.B., Cheftel, J.C. (1983). Enzymatic hydrolysis of starch and cereal flours at intermediate moisture content in a continuous extrusion-reactor. Lebensm.Wiss .& Technol. . 16,346-353 . Cochran,G.W. ,Cox ,G.M . (1957).Experimenta lDesigns .2nd .ed. . Wiley,Ne wYork . Hakulin, S., Linko, Y.Y., Linko, P., Seiler, K., Seibel, W. (1983). Enzymatic conversion of starch intwin-scre w HTST extruder.Starch . 35,411-414 . Lehninger, A.L. (1970). Biochemistry; The molecular basis of cell structure and function. Worth Publishers, Inc., New York,153-157 . Linko,P. ,Linko ,Y.Y. ,Olkku ,J . (1983a).Extrusio ncookin gan d bioconversions.J . Food Eng.. 2,243-257 . Linko, P., Hakulin, S., Linko,Y.Y . (1983b). Extrusion cooking of barley starch for production of glucose syrup and ethanol.J . CerealSci. , 1,275-284 . Linko, Y.Y., Vuorinen, H., Olkku, J., Linko, P. (1980). The effect ofHTS T extrusion on retention of cereal a-amylase activity and on enzymatic hydrolysation of barley starch. In:Foo dproces sEngineerin gVol .II :Enzym eEngineerin gi n Food Processing, ed. P. Linko, L. Larinkari. Applied Science Publishers,London ,210-223 . 171 Marshall, J.J. (1980). Some aspects of Glucoamylase action relevant to the enzymatic production of dextrose from starch. In: Food process Engineering Vol. II; Enzyme Engineering in Food Processing. ed. P. Linko, J. Larinkari.Applie d Science Publishers,London ,224-236 . Park, Yong K., Sato, H.H., San Martin, E.M., Ciacco, C.F. (1987). Production ofethano l fromextrude d cornstarc hb y a non-conventional fermentation method. Biotechnology letters. 9, 143-146. Reinikainen, P., Suortti,T. , Olkku,J. , Malkki,Y. , Linko,P . (1986). Extrusion cooking in enzymatic liquefaction of wheat starch.Starch .38 ,20-26 . Zhusman, A.I. (1979). Extrusion of starch products from maizeprocessing waste at the Korenovskii experimental factory.Sakharnav aPromvshlennost . 12,33-35 . van Zuilichem, D.J., Bruin, S., Janssen, L.P.B.M., Stolp, W. (1980). Singlescre wextrusio no fstarch -an dprotein-ric h materials. In: Food process Engineering Vol. I: Food processingsystems ,ed .P .Linko ,Y. ,Malkki ,J. , Olkku,J . Larinkari.Applie d SciencePublishers ,London ,745-756 . van Zuilichem, D.J.,va n Roekel,G.J. , Stolp,W. ,va n 'tRiet , K. (1990).Modellin go fth eEnzymati cConversio no fCracke d Cornb yTwi nScre wExtrusio nCooking .J .Foo dEng. .12 .13 - 28. 172 Notation r Reaction rateglucos e [kgm3s"1] C Substrate concentration [kgrn'3] s [kgm'3] C Water concentration H [kgrn"3] Ce Enzym concentration [s-1] k Reaction rateconstan t [%w/w] DE Dextrose equivalent [(%w/w)'1s] A Constant [(%w/w)"1] B Constant XQ-X4 Coded variables [%] X1 Concentrationa-amylas e [%] X2 Concentrationglucoamylas e [•C] X3 Temperature [%] X4 Drymatte r content 173 Appendix 7.1 dC. kd U> dt t C = 1-DE = ± — = A+Bt-Bt = A. 1 (2) 3 B B A+Bt B(A+Bt) B A+Bt dC * - A. -1 .B= -_A_ (3) dt B {A+Bt)2 (A+Bt)'- 2 kC s = k-^- i (i) B2 (A+Bt)2 Substitutiono fequation s 4an d 3i n1 gives : k = * (5) A which suggests that the kinetics of the reaction are second order. 174 CHAPTER 8 GENERAL DISCUSSION ON EXTRUSION-COOKING 8.1 Gain inknowledg e As explained in chapter 1 the so called HTST processes are prefered for processing of food because of their ability to maintain upmost the original food quality. High temperatures combined with attractive short processing times (HTST) can be realised inmos textrude rtypes,bu tth erealisatio n atth esam e timeo fa goo d residencetim e distribution (RTD)i sno talway s possible. In some casesth e chosen extruder design is limiting ,i nothe rcase sth especification so fth ewante dproduc tma yb e toonarrow .Ho wth ehardwar eca nb elimitin gwil lb eunderstoo d whena nextruder-use rha st odecid ebetwee na singl escre wo ra twin screw extruder with respect to the RTD. As explained in chapter 2 for single screw extruders and inchapte r 3 fortwi n screw extrudersbot h types of equipment havepossibilitie s but especiallyth etai li nth eRTD-curv efo ra singl escre wextrude r may be an unacceptable feature in combination with the unavoidablesli pphenomena .A goo dexampl ei sth epreparatio no f some baby food products and dietetic food with narrow specifications in cooking extruders. Here it is extremely importanttha teac h foodparticl e istreate d almost inth esam e way, with respect to the amount of heat, shear energy and residence time. Ifth e food iscontainin g starchy material, it isunacceptabl ei fnon-gelatinized ,nativ estarc hparticle swil l be detected in the food. The philosophy of the extruder manufacturerma ycaus esuc ha defec ta swil lb eexplained .Trial s performed with Clextral corotating extruders inth e beginning ofth eeighthie s learnedtha talway sa certai namoun to fnativ e starch particles arrived at the die. This was caused by the "houseconcept "o fth escre wlayou ta swa sadvise db yClextral , where at the tip of the screw set always a so called counter- 175 pitch element ismounte d inorde r tobrin g thewante d pressure level and to create at lastth e shear andmixing . This concept worksou tquit eacceptabl e forproduc tspecification swhic har e nots odefinitel y seta si sfo rbab yfoo dan ddieteti cpurposes . The clearances for screw and barrel and the more or less open screwchanne lpat hstil lallo wsom eparticle st otrave l fastan d untreated to the reverse screw element which again will allow someparticle s to arrive atth e die innativ e state.Th e sane trials, performed with Baker Perkins equipment in Raleigh (U.S.A.), where the corotating extruder was provided with a barrelvalve ,a sdescribe di nchapte r6 ,learne dtha tth eproduc t washomogeneousl y treated andn ountreate d individualscoul db e traced.At th esam etim eth eextr aadvantag ei scashed ,tha tth e variance in RTD will not increase so much by using the barrelvalvea si sth ecas ewhe nth econcep to fth ereverse dpitc h element isused . In comparison with thebeginnin g of the eighties knowledge has gained considerably about the effects of changed screw lay-out for single and twin screw cooking extruders on the RTD.At th e same time the real effects of rooted prejudices advertised by some cooking extruder manufacturers how special screw elements shouldb euse dar einvestigate d inth echapter s2/ 3an d6 an dar e evaluated furtherb yJage r inhi sthesi s (1992). Data are available now for a number of cooking extruders, at differentworkin gpoint san d forvariou s rawmaterials ,e. g soy and some starches, which allow to predict the expected over- treatment for a calculable part of the product, using simple equations.Fo rthes edenaturation sth eseverit yca nb eestimate d usingequations ,constructe di nanalog ywit hth eArrheniu smodel , likee. gi sdon ei nchapte r4 describin gth ecoloratio no fstarc h syrup and sucrose mixtures. Recently, comparable calculations havebee nmad efo rth eanimal-fee dbranc hwher eth einactivatio n ofant inutritiona lfactor s (ANF)i nlegum esee ddurin gextrusio n cooking could be described successfully (van der Poel et al., 1991a, in press). Fora saf eus eo fra wmaterial si nthi sbranch ,lik efeedin gpea s 176 fromdifferen torigin ,broa dbeans ,fiel dbean san dothe rpulses , theextrude rcooke rwil lb ea nappreciate dproces stool ,whe nth e cooking effects canb epredicted , as isstate d byva n der Poel et al., (1991b). How important a gain in knowledge is reached by developing a goodfittin ghea ttransfe rmode lfo rcookin gextruder sa si sdon e inchapte r 4an d 5,i sdemonstrate d byth e Arrheniusmodel ,i n generalwritte nas : Ak=k e RT -T U> At " In which K is reaction rate constant (1/s), T is absolute temperatuur (K),R isth ega sconstan tan dA E isth e activation energy (kJ/mol).AL/A ti sth ederivativ eo fe. gnutritiona lvalu e to time. This means that for each residence time step At,th e losso fAN Fpresen t inth eextruder ,a tth eproduc t temperature T now can be calculated, assuming a certain thermohomogeneity overth e extruder channel cross section after a residence time t from the feed port.A s isdemonstrate d inchapte r 5ther e is some inaccuracy between measured and predicted temperatures of thecookin gextrude ra tth ehighe rthroughputs .Thi si sa subjec t whichwil lb epar to fth eresearc hprogra m inorde rt ooptimiz e thecalculatio n procedure assoo na spossible . Themos tstrikin ggai ni nknowledg eabou textrusio ncookin gsinc e theeightie s is found inth etoolin g offered inth echapter s1 through7 ,leadin gtowar dcalculation swher ecombine deffort so f heat, residence time, product moisture and screw layout on product quality now can be performed. At the same time this knowledge leads to aprocedur e of product development, ifmor e dataar ecollecte dan dstore di nexper tsystem sfo rth edifferen t food branches.Th e collection of such data isa craft,th eus e ofth edat awit hhel po fth escienc elead st oa mor efull-grown - extrusion-cooking technology. 8.2 Market Whenconsiderin gth eresult so fth einvestigate dsubject son eca n conclude that the understanding of the behaviour of extruder 177 cookersha sincreased .I nsevera lcase si tha dle dt oa nincreas e ofth enumbe ro fapplication so fextrude rcooker si nth ebranche s of food industry but mostly market questions have given a big push in technology development. Janssen and van Zuilichem forecasted in 1976 at the Dutch Machevo-fair in Utrecht a six fold increase of the number of food extrusion cooking applicationsove ra perio do f1 5year s (Fig.8.1) .Th efigur eo f 1990,foresee ni n1976 ,tell stha tth eextrude rsnac kmarke tan d pasta food market would not grow so much,wherea s at the same feed stoc \5^ ^ \ 30 16 / pet \\ pas a food ^A 15\ -^-reform prognoses 1990 Fig.8. 1Histor y and forecasto fextrusion-cookin g applications inwester n Europe, 1975-1990 (van Zuilichem et al., 1978) time the market for instant products and convenience products likepetfoo dwoul dexpan dheavily .Havin gsom eresult savailabl e forth eextrude d foodan dfee dmarke ti n199 0i tca nb econducte d thatth eforecas tfo re.g .th epetfoo d industryha seve nbee nt o 178 pessimistic. 8.2.1 Petfood In Europe the market for extruded petfood is still steadily growing. In comparison with the figure of 1975 the number of applicationsi smor etha na tenfol dan dth eextrude dpetfoo dha s ^vjear 1975 1980 1985 1990 sales in billion 1.5 2.4 3-5 4.4 dollars 1975 1980 1985 1990 year Fig. 8.2 Expected growth for the petfood market (vanZuiliche m et al., 1978) becomea sever ecompetito ro fth ecanned ,s ocalle d"wet "petfoo d (Fig.8.2) .A sa genera l figureca nb esai dtha tabou t55-60 %o f allpetfoo d isno wa nextrude ddr yproduct .Th epetfoo d industry tries to produce well shaped,mult i coloured "fancy"products , facing, with a time delay of approximately twenty years, identical technological problems as are known in the snack industry,whic har eno tonl yrestricte d todi edesign .Th etim e foronl y simpleexpande d drypetfoo dha spasse db yan dth elac k of knowledge what really happens in extruder-cookers becomes obvious.Manufacturers ,wh oca naffor dt obu ysophisticate dtwin - screwextruder sca ncom eu pmor eeasil yt oth eexpectation sse t by the marketeers and the customers, making use of the flexibility of the twin-screw extruder equipment.Th e users of singlescre wextrude requipmen twil lhav emor eproblem sa sthe y are condemned to the so called bulk-market, which is big, but 179 with smaller profits. This means that single screw extruder equipment of themos t economical design will be chosen forth e production (seechapte r1 Fig . 1.9.a),a stha nth edepreciatio n ofth elowe r investmentwil lallo wbette rprofits .However ,th e more important now will be the knowledge of the impact of processvariable s likeresidenc etime ,hea ttransfer ,viscosit y behaviour,mixing ,etc. ,o nproduc tproperties .Anothe r factqr, whichstart splayin ga nimportan trol ei sth escal eo fproductio n per unit. In petfood production as well as in fish feed productionan danima lfee dproductio nhourl ycapacitie so f(5-10 ) metric tons will be quite normal (Veenendaal, 1990). At the momentonl yth eexpander-extrude requipmen tmentione di nchapte r 1 (Fig.1.9.a )wil lb ecapabl et operfor mthi sjob ,an dt obrin g the wanted capacities,whic h means that we have to accept the limitedperformance s ofth eexpander-extrude r equipment. The limitationsmentione d concern respectively: A greater dependence of flow patterns and residence time distribution ofmateria lproperties ,includin gmoisture . - Lesspositiv edisplacement . Lesshea ttransfe rhomogeneity . Less flexibility, interm so ffixe d screw length/geometry. Difficult toscal eup . Somewhatdependen t ofth ecraftsmanshi p ofth eoperator . However, we have to admit that considerable advantages can be gainedwhic htur nth escal efo rindustria lapplication seasily , like: Easymaintenance . Low sparepar tcosts . Suitable for steam injection. Acceptablemixin g bymixin g pins. Especially theus eo fstea m injection iso fparticula rvalu ea s it increases the production per unit considerably due to the combined influence of controllable heat input and decreased viscosity. It overcomes the limited heat transfer from the barrel to the 180 bulk.O f course the use of the amount of steam inth e expander islimite d since onewil l avoid toohig hwate r contents inth e productan dto olo wviscosities .I ti sremarkabl et onotic etha t thisbranc ho ffoo dindustr yha sstarte di nth epas twit hsingle - screw extruders, that has made quite often use of twin-screw extruders in the seventies and the eighties. They are now, however,bac kagai nwit hth emos tsimpl edesig no fa singl escre w extruder possible;th e expander extruder.A s stated before the mainreaso nfo rthi si sth esimpl equestion :"Wh opay sth esecon d screw?", but the increasing know how of extrusion cooking processes has made the use of the expander extruder possible, sinceno ww eunderstan dmuc hbette rth ecombine deffect so fhea t transfer,shear ,mixin g andviscosit y behaviour. 8.2.2 Coextrusion Onth eothe rhan dthi sbette runderstandin gallow st odevelo pth e more complicated products as they are produced in the snack industry,mostl ywit htwi nscre wextruder s (SeeFi g8. 3an d8.4 ). Inth eyear s198 5t o198 7s ocalle dcoextrude dproduct swer ever y popular inth eU.S.A ,disappeare d fromth emarket ,bu tar ebac k againove rthere ,mad eb yM& Man dstar tt ocom ei nEurop e (Bass, 1986). One extruder Fig.8. 3Co-extrusio n filledbisquit s (demoAPV-Baker ,Interpac k 1984;va n Zuilichem et al., 1987) 181 twinscfew extruder closedsea m (eagle) opensea m (frito-lay) cutting Fig.8. 4Alternativ efo rco-extrusion :post-die-hea dfille dsnac k (van Zuilichem et al., 1987) should notunderestimat eth ecomplexit y ofthi sprocess ,reaso n todiscus sher e someo fth eproblems . Coextrusioni sa redesigne dproces sfoun di nth ebaker yan dcon fectionery industry. The coextruded product is mostly of the tubular type, in which the outer component is produced by a cooker-extruder and the inner component is a pumpable product which will not flow freely at ambient temperatures. The two components are extruded simultaneously to produce a continuous filled tube,whic h islate rcu tt osize . Earlier products on the English market, like Cadbury's "Criss Cross"an dMar s "Cornquistos"consiste d outo fa nextrude d tube and a "post-filled" center. In the U.S.A. however Herschi produced a liquorice type product, which is die-filled with a dried fruitcontainin g chewable innerfilling . For each product development a number of problems have to be solved; including die design; recipe-restrictions;pumpin g the illing; post extruder processing and the cutting, which is explainedbelow . Diedesign ;Th ebes tarrangemen t ist obrin gth ecente r filling over 90 degs. in the extruded outer tube, giving the center a more restrictive path but allowing the outer tube to be 182 consistent. The outer tubing is mostly forced through a restriction formed by anumbe r of small holes orb y an annular ring.I ti sdesirabl et ohav eth ecente rentr ypoin ta sfa rawa y fromth e finaldi ea spossible ,i npractic ea distanc eo fabou t 40mm .A n imporatantproble m ismanufacturin g adie-arrangemen t which will maintain the accurate gaps between center die -and outerparts ,durin g thedi elifetime . Size restrictions: When the minimum dimension of the outer annulus is about 1m m thickness and the wall-thickness of the center tubing is also chosen 1 mm , the restriction will be clear. The outer shell easily expands to twice the diameter of the die annulus, giving way to a certain pumping area cross section.I tdoe sno tappea rt ob epractica lt owor kwit ha rati o greatertha n 10,givin g aminimu m insidediamete r ofa finished product of about 10 mm, whilst the outside is about 14 mm. Reduction of the expansion canb e achieved by changing the raw material properties and the product moisture,Multipl e outlets are perfectly possible as demonstrated by Demo APV-Baker at Interpack Fair Dusseldorf BRD in 1984, where 10 tubes were extruded simultaneously. Recipe restrictions: Crisp extruded wheat based outers can be directly produced in a twin screw extruder at 5% moisture. A problem forth ecente r fillingwil lb eth eequilibriu m moisture content which will ultimately exist, but enough knowledge is available fromth ebiscuit ,wafe ran dconfectioner y industryt o developattractiv ecente r formulations forcheese ,mea tpastes, onion/nuts,caramel ,gums , jelliesan d sugar/fatcreams . Density-influence: If a round tube is extruded and filled the weight of the center will be 2 or 3 times that of the outer, whichmake sth ecente rt odominate .Thi sca nb eovercom eby : alteration of the cross section after extrusion from circular tubing toa flattened crosssection , addingfinel ygroun dextrude dan dexpande dstarch ymateria l toth ecente r filling, inorde rt o reduceth edensity . 183 aeration of the center filling, which brings the density back for a straight cream from 1,2/1,3 towards 0,7/0,8 kg.dm" 3. Center pumping: Usually each tube,produce d byth e twin screw extruder is provided with itsow n metering-pump, in order to match theweight s inth etw ostream s exactly.Th etemperature , at which the cream ispumpe d depends on several factors,ver y viscouscream smus tb epreheated ,bu tnormall ypumpin ga tambien t temperatures should be preferred. For such a cream a typical temperature curvewoul dbe : hopper 22° C pump outlet24° C dieentr y 25°C dieexi t50° C postdi e (60S)65° C Postprocessing :Severa ltechnologica lsolution sar ei nus elik e rolling, theus eo fsecondar y dies, cooling provisions, drying and toasting. Itshoul db ementione d thatmos tproduct s havea memory for the original die shape and more than one roller passage isneede d fora nalteratio no fth ecros ssection . Sincelon gsecondar ydie sar eimpractica lt ooperate ,thes edie s areonl y inus efo route r recipestha t setquickly . Those dies mustb eadjustable , inmuc h thesam ewa yshapin g jaws areuse d forflat-brea dproduction .Sinc eth eoute rtub ecool sfaste rtha n the inner center, the outer tubing acts as an effective insulator. Nevertheless, in practice 2% moisture loss canb e reachedb ycooling . For drying purposeswar m airi ssuitabl e andwil l help togiv e a final texture toth eoute rtube .Finall y theproduct s inth e toasting sectionshoul db erotate d inorde rt ocreat ea nevenl y distributed colouran dtaste . Cutting:I fclose dproduc tend sar eneede dth ecutting/crimpin g devicemus tb eclos et oth edie .I fa symmetrica lcrim pi sneede d apai ro fcuttin groll si sessential .Whe nth eproduc tend-plane s 184 aret ob eopen ,allowin gth ecente rt ob eseen ,th etub emus tb e setprio rt oth ecutting ,an da tth esam etim eth ecente r cream mustb ecoo l enoughno tt o flowout . Iti sdifficul tt ose eho wth efamil yo fcoextrude dproduct swil l develop but product makers will handle this technology as a strongtoo li nth efuture .On eha st oadmi ttha tthi scoextrusio n inth estag ewher e iti sdescribe d here,i smor ea craf ttha na science. 8.2.3 Feed industry Asi smentione dabov eth eexpectatio n isconfirme dtha tth efee d industry has discovered cooking extruders for part of their production, especially for weaning feed. The extruder will be used asa preconditione r inthei rpelle t lines,jus tbefor eth e press. It is a wise decision using the extruder in this way as the processing reactor, in which unwanted ANF can be inactivated, whereasth econventiona lshapin gan deconomica lforming-equipmen t likepelletizin gmill swil lb emaintained .Fo rth efee d industry thegai ni nknowledge ,describe d above,ca nb eo fimmediat ehel p to solve some ANF-problems and allow to "engineer" product quality. Also here the marginal profits in this industry will exclude at a first glance the costly twin screw extruders and willfavo rsingl escre wequipment ,a si sth ecas ei nth epe tfoo d industry. This tendency also counts for the fish feed industry where the gross margins are small and a good balance must be found in caloric value of the products and the total process costs, includingth edryin gcosts .Thi swil ldirec tthi sbranc h aswel l toth eus eo f expander equipment 8.2.4 Other Food-branches In the remaining branches of the food industry, mentioned by Janssenan dva n Zuilichem in1975 ,th ecookin g extruderwil lb e used as a processing reactor capable to produce products with certain unique rheological properties, such as watersorption, watersolubility, wateractivity or a certain consistency. This countsfo rth epast aindustr ya swel la sfo rth econvenienc efoo d 185 branch, where people understand that an extruder cooker is capable to "engineer" Theological properties. A survey of products, made by extruder cookers is given in Table 8.1. For thispurpos eth eextrude rcooke ri suse dt ocreat ea produc twit h a complicated texture and structure, which cannot be produced with conventional equipment.Her e thedi e is an important part of the hardware and the die construction mostly is very vulnerablean dcrucia lfo rth een dresul ta si se. gth ecas efo r instantproducts . 8.2.5 Agrochemicals Themarketshar efo rextrusio ncookin gapplication si sno tlimite d to foodalone ,sinc eth eextrude rcooker ,a sa proces sreactor . EXTRUSION PROCESSES V — ANIMAL FEED J Q Cattle & Pigfeed ] [[Fish-feeds Agro -Chemicals Pasta (2types) Animal protein Fertilizers Partial precooked Precooked flours (starch end „Tortilla " flours Sausages Vegetables/soups cereals/seeds protein)wit h different bulk Enzyme engineered Bakery products Scrap meat/fish BagaSSe and other densities feed Confectianaries Blood protein Cellulosic material Complete gelati -candies Milk protein Cocoa waste mat'l -liquorices nized cereals -caramels Germ extrusion Sucrose based Yeast wastes -chocolate Co-extruded chemicals _, , .-portly cooked Breaking up of cellulose meat/starch Spices Starches!orgelQtlf11ze d Protein hydrolysates Soft cheese Maltodextnns Spent grains (Brewery) Extruded pulses Vegetable protein Pulp for paper Cereal snacks soya/cottonseed/rape seed production Health foods Pet foods Proteins chemically modified Breakfast cereals Oilseed meals Flat breads/cookies Health foods Crosslinked starches Brann extrusion Lactose based Crumb chemicals Instant cooked foods Bio-deqradable Potato products plastics Baby foods (C.S.M./W. SB. ) Fig. 8.5 Survey of products made by extrusion cooking (van Zuilicheme t al., 1987) provest ob ea goo dpiec eo fequipmen tfo rs ocalle dagrificatio n purposes.I nth esurve y ofFig .8. 5 appearsa thir dgrou pagro - chemicals,rangin ge.g . fromchemicall ymodifie d starches,use d inth epape r industry aspape rcoating ,t owoo d chipsan dplan t fibre extrusion. The number of applications in this field is stillincreasing .Th eknowledg ehoweve rt odevelo psuc hprocesse s islimited , sinceth ethoroug hdescription s ofextrude rcooker s asprocessin g reactors arescarc eavailable . 186 8.3 Equipment manufacturers Many expectations in extrusion-cooking have not come up,quit e someequipmen tmanufacturer shav ehel d outgrea thope st othei r customers,promisin g irrealisticthroughputs ,produc tqualitie s and savingswhe nusin g extrudercookers .Especiall y thoseamon g the manufacturers, who had none or too little experience with food,non-newtonia nfoo dpropertie san dwit hth efoo dmarke twer e badadvisors . Not all of them are toblame ,esp .thos e companies specialised in food equipment for which extruder cookers only are one interestamongs tothers ,sinc ethe yar ever ymuc hfamilia rt oth e difficult behaviour and sometimes remarkable rheological properties of foodproducts . Most failures have been made by companies which are used to construct equipment for the plastic polymer industry. They underestimated the effect of the presence of water and oil as normalconstituent si nagricultura lproduc ean dha dn oknowledg e about the behaviour of not well defined bio-polymers in food, sucha sdifferen t starchesan dproteins . Withth eknowledg epresente d above,th eequipmen tmanufacturer s will be able to advise their clients more properly about technological questions related toproduc t development. On the otherhan dthe y should equipthemselve swit hth eknowledg etha t isavailabl e now and use it forth edevelopmen t of better food extruders.Tim eha spasse d inwhic ha food-extrude r lay-outwa s made by a design staff, educated in mechanics and chemical polymer science. When extrusion cooking operations will be performedmor eefficien ti nth efuture ,th eexistin gdesig nteam s will have to includewel l educated food process engineerswit h a substantial amount ofscienc e inthei rluggage . 8.4 Technology, Final Remarks RTD General knowledge isavailabl e how axial mixing works out ina 187 RTD.Thi si squit eexac tfo rth eco -an dcounter-rotating ,twin - screwextruders .Thi sknowledg ei sstil lincomplet efo rdifferen t designso fsingl escre wextruders ,lik eexpande requipment .Thi s should be studied further. Heat transfer Knowledgei savailabl eno wabou tth ecalculatio no fth ehea tflo w insingl ean dtwin-scre wextruders .Her eth epeculia rregio nnea r the barrel inside diameter appears to be a limitation in the understanding of the radial mixing and the flow patterns in differentkind so fscre wgeometries .Thi sfundamenta l knowledge willhel pt ooptimiz eth ehea ttransfe rcalculations .Als othi s isa subjec t for furtherstudies . Biotechnology,consequentl y mixing It is common to use extruder equipment quite successfully for dispersive mixing applications in the plastic and rubber industry,howeve rth estud yi nchapte r7 show sdirectl ytha tth e restrictions inusin gextruder sa sreactor s in food/biochemical operationsar eo fa differen tnature . Integration When an attempt is made to integrate models of a RTD, heat transfer and mixing in one extruder model the predictions on extruderperformanc eca nb eimproved .Thi swil lno tdiminis hth e "craft" part in extrusion cooking as the complexity of such a modelwil lguarante eth equestio n "crafto rscience "t oremain , butno w forth eus eo fthi smodel . 188 References Bass, C.P. Co-extrusion: The problem and some of the solutions.Pastaan dextrusio n cooked foods,proceeding so f symposium held inMila n 25-26march ,1985 ,pl59-166 . Poel,A.F.B .va n der, Blonk,J. , Huisman,J. , denHartog ,L.A . Steamprocessin g methods forpheasolu svulgaru sbeans . Livestockproductio n science.1991a ,p305-319 . Poel, A.F.B. van der, Doorenbos, J., Huisman, J., Boer, H., Evaluation of Techniques to determine the protein digestibility inhea t processed beans forpiglets ,Anima l feed science and technology. 1991b,p331-341 . Veenendaal,J . Extrusion in the compound feed industry. Moritz Schaeffer Detmold,Advance s inFee dtechnology . 1990,^ - 63. 189