Elﬁn.”

. \vwn

 

_:..n
.3.) 7:... 2. .2; 1......) 75.4,: 1.1.1.
.313 uéﬁin

i. razegﬁu: .

er C. It: 3. 33.1! i.
. v: . f“... ...~\IP.X:I.....\¢.¥.
3.)»; X51». . v.

c .x l :3... :3 air 334:1
.1 . .. 5‘35. 5....EIIEGE5H at!
.25.... w 3.1;

9i.fl.1t...(i: xi
l:llf!fl.(: .21..

 

 

 

735 9136477

 

meme/w smeuwensmLemmas '-| LIBRARY ~
lllllllillllllljlgllllllllﬂllslllﬂlllﬂllllljlll 3 Michigan state
University

 

 

"'1

 

 

This is to certify that the

dissertation entitled

Starch Liquefaction by Thermostable Alpha—Amylase

During Reactive Extrusion

presented by
Vanee Komolprasert
has been accepted towards fulﬁllment

of the requirements for

Ph. D . degree in Agricultural Engineering

 

/ ' 04%;
Major ﬁfessor Z] I

Date 12—11-1989

 

MSU is an Afﬁrmative A ction/Equal Opportunity Institution 0-12771

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATE DUE DATE DUE DATE DUE

iii? i 3 _

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Afﬁrmative Action/Equal Opportunity Institution

 

 

STARCH LIQUEFACT ION BY THERMOSTABLE ALPHA-AMYLASE
DURING REACTIVE EXTRUSION

By

Vanee Komolprasert

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Agricultural Engineering

Department of Agricultural Engineering

1989

ABSTRACT

STARCH LIQUEFACTION BY THERMOSTABLE ALPHA-AMYLASE
DURING REACTIVE EXTRUSION

By

Vance Komolprascrt

An MPF-SOD Baker-Perkins twin screw extruder was usd as a bioreactor for
liquefaction of pre—gelatinized corn starch by thermostable a -amylase. A
one-dimensional axial dispersion model was used to predict the extent of liquefaction,
and was evaluated using data from several experimental extrusion runs.

An initial rate study at low starch concentration (1-8% w/w) showed that the
kinetics of a—amylase follows the classical Michaelis-Menten model. At a starch
concentration of 40%, a continuous rate study was used. Results from this study showed
that the enzyme kinetics could be described by a modiﬁed ﬁrst order equation with a rate
constant of 0.033 min'1 at 95°C and a pH of 6.0.

The effect of shear on enzyme activity was studied in both a Haake viscometer and
the extruder. While loss of activity was found to depend signiﬁcantly on strain history
alone in the viscometer, loss of activity in the extruder depended on the combination of
strain history, shear rate and speciﬁc energy consumption (SEC). The loss of enzyme
activity in the extruder was small (20%), possibly due to the shielding effect of soy
polysaccharide (SP8) which was used as a carrier for the enzyme.

Corn syrup (Newtonian), 7% SP8 in honey and 2.5% methocel in water were used
to determine the effect of rheological behavior on the dispersion of ﬂuids during
extrusion. The extruder was modeled as a tube, using the hydraulic diameter as a
characteristic dimension. The Taylor-Aris dispersion model for laminar tube ﬂow was

used to describe the effect of molecular diffusion and velocity proﬁle on the dispersion

coefﬁcient. Results showed that the Taylor-Ads model is applicable to the SPS-honcy
mixture and to corn syrup. Effect of velocity was dominant in comparison to that of
molecular diffusion.

The extent of starch liquefaction depended signiﬁcantly on moisture content,
residence time and enzyme dosage. Dispersion numbers were calculated from residence
time distribution (RTD) data and used in the axial dispersion model. The dispersion
number decreased with increased length of reaction zone but increased with increased
screw rpm. The axial dispersion equation was solved analytically and compared to
experimental data, with very good agreement. The maximum fractional conversion was

0.3 at a residence time of 11 minutes.

iv
ACKNOWLEDGMENTS

I would like to express my sincere gratitude and appreciation to Dr. Robert Y.
Ofoli, my major professor, for giving me ﬁnancial support throughout my program, and
enabling me to pursue my degree. Many thanks for his guidance, encouragement and for
all his hard work in the preparation of this dissertation. Without his guidance and
support, this would not have been possible.

Thanks to Dr. K. Jayaraman (Associate Professor of Chemical Engineering) for his
suggestions in experimental design and analyses, and for helping me learning more about
viscoelasticity. Thanks to Dr. D. Miller (Associate Professor of Chemical Engineering)
who helped me extend my knowledge of kinetics and for his assistance in analyzing the
kinetic data. Thanks also to Dr. K. Berglund (Associate Professor of Chemical
Engineering and Agricultural Engineering) and Dr. Bruce R. Harte (Associate Professor
of the School of Packaging) for their advice and willingness to serve on my guidance
commitee.

Thanks to Dr. Badal Saha (Senior Scientist of the Michigan Biotechnology
Institute) for his suggestions on analytical procedures. Thanks to Steve Hull, John
Keenan and Drew Van-Norman for their help in all the extrusion runs, and to all my
friends for their help in making my study possible.

Special thanks to my mother in Thailand for her love, support and encourangement
throughout my study. Thanks to the Gingas family for their love and care, and for
making me feel at home during my stay in Michigan.

Support for this work from the Research Excellence Fund (REF) of the State of
Michigan is gratefully acknowledged.

 

TABLE OF CONTENTS
LIST OF TABLES ............................................................................................. viii
LIST OF FIGURES ............................................................................................ x
1 INTRODUCTION .............................................................................................. 1
2 OBJECTIVES ..................................................................................................... 3
3 LITERATURE REVIEW ................................................................................... 4
3.1 The Nature of Starch ................................................................................... 4
3.1.1 Starch components and structure ........................................................ 4
3.1.2 Starch gelatinization ........................................................................... 6
3. 2 Amylolytic Enzymes ................................................................................... 7
3. 3 Processes for Producing Sweeteners ........................................................... 9
3 .4 Enzyme Kinetics ......................................................................................... 10
3.5 Legal Aspects of Enzyme Uses ................................................................... 12
3. 6 Overview of Food Extruders ....................................................................... 13
3 .7 Extrusion Processing ................................................................................... 15
3.7.1 Mixing in extruders ............................................................................. 18
3.7.2 Process instability ............................................................................... 20
3. 7. 3 Application of polymer extrusion theory to food extrusion ............... 21
3. 8 Rheological Changes During Extrusion ...................................................... 21
3. 9 Ideal and Non-ideal Reactors ...................................................................... 22
3.9.1 Residence time distribution ............................................................... 23
3. 9. 2 Extent of chemical reaction ............................................................... 24
3. 9. 3 Residence time distribution in single and twin screw extruders ......... 25
3.10 Starch Hydrolysis in Twin Screw Extruders ............................................. 27
3.10. 1 Liquefaction ...................................................................................... 29
3.10.2 Enzyme inactivation in the extruder ................................................. 29
3.11 Nomenclature ............................................................................................ 30
3.12 References ................................................................................................. 32
4 STARCH HYDROLYSIS KINETICS OF THERMOSTABLE
ALPHA-AMYLASE .............................................................................................. 40
4.1 Abstract ....................................................................................................... 40
4.2 Introduction and Background ...................................................................... 40
4.3 Theory ......................................................................................................... 43
4.4 Materials and Methods ................................................................................ 46
4.4.1 Initial rate study at low starch concentrations .................................... 46
4.4.2 Continuous rate study at high starch concentrations .......................... 46

vi

4.5 Results ........................................................................................................ 47
4.5.1 Initial rate study at low starch concentrations .................................... 47
4.5.2 Continuous rate study at high starch concentration ............................ 48
4.6 Discussion ................................................................................................... 60
4.7 Conclusions ................................................................................................. 61
4.8 Nomenclature .............................................................................................. 62
4.9 References ................................................................................................... 64
5 EFFECT OF SHEAR ON ENZYMATIC ACTIVITY DURING REACI'IVE
EXTRUSION ......................................................................................................... 67
5.1 Abstract ....................................................................................................... 67
5.2 Introduction and Background ...................................................................... 67
5.3 Theory ......................................................................................................... 69
5.4 Materials and Methods ................................................................................ 71
5.4.1 Haake viscometer experiments ........................................................... 71
5.4.2 Extruder experiments .......................................................................... 72
5.4.3 Determination of the effect of shear in the TSE ................................. 74
5.5 Results and Discussion ................................................................................ 75
5.5.1 Shear deactivation in the batch system ............................................... 75
5.5.2 Shear deactivation in twin screw extruders ........................................ 79
5.6 Conclusions ................................................................................................. 81
5.7 Nomenclature .............................................................................................. 82
5.8 References ................................................................................................... 84
6 EFFECT OF MATERIAL RHEOLOGY ON THE DISPERSION OF
FLUIDS IN EXTRUDERS .................................................................................... 86
6.1 Abstract ....................................................................................................... 86
6.2 Introduction and Background ...................................................................... 86
6.2.1 Diffusion and Dispersion ................................................................... 89
6.3 Theoretical Development ............................................................................ 93
6.4 Materials and Methods ................................................................................ 96
6.4.1 Rheological behavior .......................................................................... 98
6.5 Results and Discussion ................................................................................ 98
6.5.1 Rheological data ................................................................................. 98
6.5.2 Extrusion data ..................................................................................... 104
6.6 Conclusions ................................................................................................. 109
6.7 Nomenclature .............................................................................................. 110
6.8 References ................................................................................................... 112
7 A DISPERSION MODEL FOR PREDICTIN G THE EXTENT OF
LIQUEFACI‘ION BY ALPHA-AMYLASE IN REACTIVE EXTRUSION ........ 114
7.1 Abstract ....................................................................................................... 114
7.2 Introduction and Background ....................... 114
7.3 Model Development .................................................................................... 117
7.4 Materials and Methods ................................................................................ 119
7.4.1 Determination of residence time distribution (RTD) .......................... 127
7.4.2 Estimation of dispersion coefﬁcients .................................................. 129
7.4.3 Sample preparation and analysis for reducing sugar .......................... 131

 

vii

7.4.4 Estimation of the apparent viscosity of starch in the extruder ............ 132
7.4.5 Rheological data ................................................................................. 134
7.5 Results and Discussion ................................................................................ 135
7.5.1 Estimation of dispersion numbers and dispersion coefﬁcients ........... 135
7.5.2 Use of the dispersion model to predicr RTD ...................................... 143
7.5.3 Extent of starch liquefaction in the extruder ....................................... 147
7.5.4 Relationship between power number and Reynolds number for
mixed screws .............................................................................................. 149
7.5.5 Effect of rheology on extent of reaction ............................................. 153
7.5.6 Dispersion model for predicting starch hydrolysis in reactive
extrusion ...................................................................................................... 154
7.6 Conclusions ................................................................................................. 160
7.7 Nomenclature .............................................................................................. 161
7.8 References ................................................................................................... 164
8 CONCLUSIONS ................................................................................................ 167
9 RECOMMENDATIONS FOR FURTHER RESEARCH .................................. 169
10 APPENDICES .................................................................................................. 170
Appendix 1. Shear deactivation data from Haake viscometer ..................... 170
Appendix 2. Analysis of variance (SAS program) of effect of strain
history on % residual enzyme activity from Haake viscometer data ............ 172
Appendix 3. Shear deactivation data from extrusion runs ............................ 173
Appendix 4. Analysis of variance (SAS program) of effect of strain
history on enzyme activity during extrusion ................................................. 174
Appendix 5. Analysis of variance (SAS program) of effect of strain
history, shear rate and SEC on enzyme activity during extrusion ................. 175
Appendix 6. Extrusion data for corn syrup, 7% SP8 in honey and 2.5%
Methocel ........................................................................................................ 176
Appendix 7. Paddle conﬁgurations in the twin screw extruder .................... 177

Appendix 8. Extrusion data on starch liquefaction (extrusion runs 1-20) 178
Appendix 9. Analysis of variance (SAS program) of mean residence time

(RTD) for extrusion runs 1-12 ....................................................................... 179
Appendix 10. Analysis of variance (SAS program) of dispersion number
(DI) for extrusion runs 1-12 .......................................................................... 180
Appendix 11. Analysis of variance (SAS program) of dispersion number
(DI) for extrusion runs 13-20 ........................................................................ 181
Appendix 12. Percent reducing sugar and DE from extrusion runs 1-20 ..... 182

Appendix 13. Analysis of variance (SAS program) of effect of moisture
content, screw rpm and enzyme level on % reducing sugar (RS) and % DE
for extrusion runs 1-12 .................................................................................. 185
Appendix 14. Analysis of variance (SAS program) of effect of method of
gaggle addition on % reducing sugar (RS) and % DE for extrusion runs

- .............................................................................................................. 186

viii

LIST OF TABLES
Table 3.1. Gelatinization temperatures of several native starches. ....................... 7
Table 3.2. Proﬁles of industrial sugar produced from starch. ............................... 10
Table 4.1. Relative reducing power of various sugars. ......................................... 41
Table 4.2. Stability of Termamyl 60L and 120L in terms of enzyme half-er in
minutes for DE values in the range of 0-12. ........................................................... 42
Table 4.3. Kinetic models from the literature. ....................................................... 45
Table 4.4. Percent of unreacted starch, reducing sugars DE, and fractional
conversion (ﬁrst replicate of continuous rate study). ............................................. 50
Table 4.5. Percent of unreacted starch, reducing sugars DE, and fractional
conversion (second replicate of continuous rate study). ........................................ 51
Table 5.1. Experimental conditions used in the Haake viscometer. ...................... 72
Table 5.2. Residual enzymatic activity obtained from the Haake viscometer. ..... 76
Table 5.3. Power and Reynolds numbers for 30% SPS in water. ......................... 80

Table 5.4. Values of apparent viscosity, shear rate, mean residence time, strain
history, speciﬁc energy consumption, ratio of mass ﬂow rate to screw rpm and
% residual enzyme activity. .................................................................................... 80

Table 6.1. Rheological models of the ﬂuids used, derived from steady shear
experiments. ............................................................................................................ 99

Table 6.2. Zero shear viscosity and equilibrium creep compliance of corn syrup,
7% SPS in honey and 2.5 % methocel in water. .................................................... 104

Table 6.3. Apparent viscosity and estimated average shear rate of corn syrup,
7% SPS in honey and 2.5% methocel, at various extrusion conditions. ................ 105

Table 6.4. Dispersion number, dispersion coefﬁcient and molecular diffusion
coefﬁcient of corn syrup, 7% SPS in honey and 2.5% methocel. .......................... 105

Table 6.5. Axial and radial Peclet number, and dimensionless time of corn
syrup, 7% SPS in honey and 2.5% methocel. ......................................................... 109

Table 7.1. Experimental design for extrusion runs 1-12. ...................................... 123
Table 7.2. Experimental conditions for extrusion runs 13-20. .............................. 123
Table 7.3. Mean residence time, dispersion number and dispersion coefﬁcient

from extrusion runs 1-12. ....................................................................................... 135
Table 7 .4. Mean residence time, dispersion number and dispersion coefﬁcient

from extrusion runs 13-20. ..................................................................................... 141
Table 7.5. Mean velocity and dispersion coefﬁcient at various zone lengths

during extrusion run 20. ......................................................................................... 142
Table 7 .6. Values of "b" (Eq. 7.31) and standard errors for various extrusion

runs. ........................................................................................................................ 145
Table 7.7. Percent reducing sugar and DE (extrusion runs 1-12). ........................ 148
Table 7.8. Percent of un-reacted starch, reducing sugar and DE, and fractional
conversion (from extrusion runs 13-20). ................................................................ 148
Table 7.9. Consistency coefﬁcients for corn syrup at various temperatures. ........ 149
Table 7.10. Power and Reynolds numbers for corn syrup. ................................... 151

Table 7.11. Normalized viscosity indices of starch extrudates from extrusion
runs 1-12. ................................................................................................................ 153

Table 7.12. Normalized viscosity indices of starch extrudates from extrusion
runs 13-20. .............................................................................................................. 154

Table 7.13. Observed and predicted fractional conversions (extrusion runs
13-20). .................................................................................................................... 158

LIST OF FIGURES

Figure 4.1 Lineweaver-Burk plot of reciprocal rate of product formation versus
reciprocal initial starch concentration. ................................................................... 49
Figure 4.2 Fractional conversion (X) versus residence time. ................................ 52
Figure 4.3 Dextrose equivalent (DE) and starch concentration ([8]) versus

residence time. ........................................................................................................ 54
Figure 4.4 Rate of starch consumption versus modiﬁed starch concentration. ..... 56
Figure 4.5 Rate of product formation versus modiﬁed starch concentration. ....... 58
Figure 4.6 Rate of product formation versus rate of starch consumption. ............ 59

Figure 5.1 Percent residual activity of or—amylase versus strain history (data
obtained from a mixer viscometer). ........................................................................ 77

Figure 5.2 Percent residual activity vs. strain history of rennet, catalase,
carboxypeptidase (Charm and Wong, 1970) and or—amylase (this study). ............ 78

Figure 6.1 Contribution of convective transfer and molecular diffusion to
dispersion in laminar ﬂow (Butt, 1980). ................................................................ 91

Figure 6.2 Schematic of the setup of the MPF-50D Baker-Perkins twin screw
extruder at 15 L/D. ................................................................................................. 97

Figure 6.3 Storage modulus, loss modulus and complex viscosity versus
frequency for corn syrup at 23.4°C. ........................................................................ 100

Figure 6.4 Storage modulus, loss modulus and complex viscosity versus
frequency for 7% SPS in honey at 246°C. ............................................................. 101

Figure 6.5 Storage modulus, loss modulus and complex viscosity versus
frequency for 2.5% Methocel at 228°C. ................................................................ 102

Figure 6.6 Velocity proﬁle in laminar tube ﬂow of a Newtonian ﬂuid (n=1),
non-Newtonian ﬂuid (n<1), and in plug ﬂow (n=0). ............................................. 106

Figure 6.7 Dispersion coefﬁcient versus mean velocity and molecular diffusion
coefﬁcient (based on Eq. 6.13). .............................................................................. 108

Figure 7.1 Layout of the Baker-Perkins MPF-50D twin screw extruder. ............. 120
Figure 7.2 Dimensions of screw elements used in the MPF-50D Baker-Perkins

twin screw extruder. ............................................................................................... 121
Figure 7.3 Extruder setup for experiments 1-12 showing entry points of enzyme
solution (B) and tracer dyes. ................................................................................... 124
Figure 7.4 Extruder setup for experiments 13-16 showing entry points of

enzyme solution (B) and tracer dyes. ..................................................................... 125
Figure 7.5 Extruder setup for experiments 17-20 showing entry points of

enzyme solution (B). ............................................................................................... 126
Figure 7.6 Extruder setup for experiments 20.1, 20.2 and 20.3 showing entry

points of dye solution. ............................................................................................ 128
Figure 7.7 Schematic setup of the MPF-50D Baker-Perkins twin screw extruder

for corn syrup experiments. .................................................................................... 133
Figure 7.8 Comes from experiments 1, 2, 5, 6, 9 and 10 at 60% moisture

content. ................................................................................................................... 136
Figure 7.9 C-curves from experiments 3, 4, 7, 8, 11 and 12 at 50% moisture

content. ................................................................................................................... 137
Figure 7.10 C-curves (redness versus residence time) from experiments 13-16

(mean residence times: 2.2, 4.7, 7.6 and 10.7 min, respectively). .......................... 139
Figure 7.11 C-curves (redness versus residence time) from experiments 17 ~20

(mean residence times: 4.6, 5.7, 10.3 and 10.4 min, respectively) .......................... 140
Figure 7.12 Experimental and calculated F functions versus time. ....................... 146
Figure 7.13 Viscosity versus inverse temperature for corn syrup. ........................ 150

Figure 7.14 Power number versus Reynolds number for corn syrup (Newtonian
standard). ................................................................................................................ 152

Figure 7.15 Predicted fractional conversion versus normalized length of reactor
from experiments 13-16. ........................................................................................ 156

Figure 7.16 Predicted fractional conversion versus normalized length of reactor
from experiments 17-20. ........................................................................................ 157

Figure 7.17 Predicted and observed fractional conversion versus normalized
length of reactor from experiments 13-20. ............................................................. 159

1 INTRODUCTION

Since twin screw extruders provide high productivity and versatibility, they have
become popular and have replaced the single screw extruder in many processes. The
twin screw extruder improves mixing and heat transfer, therefore it is good for high
temperature short time (HTST) processes. Use of the twin screw extruder as a bioreactor
is a novel approach, and has induced much interest. In the last decade, most of the
studies have focused on hydrolysis of starch both with and without enzymes.

Heat and mechanical energy cause degradation of starch, resulting in reduced
viscosity and increased solubility. Solubility improves texture and increases digestibility
for human nutrition. Without addition of enzymes, solubility is limited and depends
highly on extrusion conditions. However, use of an or -amylase accelerates viscosity
reduction and increases solubility, preventing the starch from retrogradation. Many
investigators have used extrusion for the pretreatment and/or liquefaction of substrates
for subsequent sacchariﬁcation and fermentation. This process converts starch, a cheap

material, to more value products such as glucose, maltose and alcohol.

Although (Jr-amylase has been used in liquefaction for many years, the process

usually takes place in a batch system where the residence time can be increased to obtain
high conversion rates. In contrast, the extrusion process generally occurs over relatively
short residence times, therefore a high conversion is difﬁcult to obtain in a single
passage. However, the process may be used to partially saccharify starch, and enhance
product sweetness without addition of external sugars.

The use of enzymes in twin screw extruders for liquefaction and/or sacchariﬁcation

has a lot of potential. However, its acceptance and utility has been slow because of lack

of understanding of the extrusion process during enzymatic reaction, and lack of
information on appropriate quantitative models for predicting the extent of starch

conversion.

2 OBJECTIVES

To realize the potential of using twin screw extruders as bioreactors, an accurate
model for predicting the extent of reaction is required for process design and control.
Because of its versatility and combination of possible operating conditions, a model for
enzyme reaction kinetics in the extruder must of necessity be quite general.
Unfortunately, such a generalized model would be difﬁcult to develop. However, if the
extruder is set up with a screw conﬁguration to induce a ﬂow pattern near plug ﬂow, then
a one-dimensional axial dispersion model could be used.

The primary goal of this study was to develop and verify an axial dispersion model
for predicting the extent of starch liquefaction by thermostable (rt-amylase in a
co-rotating twin screw extruder. To develop an appropriate model, however, it is
necessary to develop an understanding of the interactions between enzyme kinetics,
material rheology and ﬂow characteristics in the extruder.

Therefore, to achieve the primary objective above, the following speciﬁc activities

were undertaken:
1. Determination of the kinetics of starch hydrolysis by a—amylase and the use of

the kinetic information for predicting the extent of starch conversion in an extruder being
used as a bioreactor;

2. Investigation of the inﬂuence of extrusion shear on enzymatic activity during
reactive extrusion;

3. Assessment of the effect of material rheological behavior on degree of ﬂuid
dispersion, measured in terms of a dispersion coefﬁcient; and

4. Development of a methodology to determine the dispersion coefﬁcient and the

use of the coefﬁcients in a model for predicting the extent of starch liquefaction.

3 LITERATURE REVIEW
3.1 The Nature of Starch

3.1.1 Starch components and structure

Starch is a major source of reserved carbohydrate found in plant cells as large
granules. These granules are arranged either in concentric layers (as in cereal grains) or
in eccentric layers (as in potato tubers). Starch is an a-glucan that consists of two distinct
components: amylose and amylopectin.

The structure of amylose is not well deﬁned (Hood, 1982); it was ﬁrst thought to
be a linear glucose polymer linked by (Jr-1,4 linkages until it was proved to be nonlinear,
based on chemical evidence reported by Williams (1968), and Wolfrom and Khadem
(1965). Banks and Greenwood (1975) concluded that amylose is made up of a mixture of
linear molecules with or-l,4 linkages and branched molecules with a-1,6 linkages.

Pure amylose can be separated from amylopectin by two methods (Hood, 1982):

1. Partial gelatinization, which releases low molecular weight linear amylose from
the granule.

2. Solubilization of granular starch in dimethylsulfoxide or alkali, followed by
adding thymol to form an insoluble complex with amylose.

The purity of the isolated amylose is measured by iodine binding capacity (IBC).
The IBC of pure amylose is 20% while that of amylopectin is much smaller. The iodine
complex is associated with a helical structure of the starch chains. Dextrins containing
less than six D- glucose residues do not produce the characteristic colored complex
(Solomon, 1978). Amylose has a molecular weight of 1.5x105-106, depending on its
native source. In the linear chain, the average length is 500-2000 glucose units. Amylose

with 250-300 units can form a helix.

Amylose does not easily dissolve in water, or does so only to a limited extent. In
the form of its butanol complex, amylose disperses in aqueous solution; its stability
depends on pH, concentration, molecular size and the presence of electrolytes (Solomon,
1978). Molecular conformations of concentrated amylose are responsible for
retrogradation.

Amylopectin is a branched structure containing 94-96% a-1,4 and 4-6% a—1,6

linkages (Hood, 1982). The average chain length is 20-26 glucose units. Intensive
studies on the structure of amylopectin have been conducted using a combination of
enzymatic methods and gel permeation chromatographic methods. The molecular
arrangement is deﬁned conventionally in terms of the degree of polymerization (DP).
The molecular weight of amylopectin is in the 107-10g range. Within the granule,
amylase may associate in an orderly manner with the linear portion of amylopectin in the
crystalline region of the amylopectin.

Amylopectins dissolve in water, and neutral aqueous solutions are extremely
stable, with little tendency towards retrogradation. Amylopectin forms a red color with
iodine in contrast to the blue color obtained with amylose. The low binding power as
indicated by the red color is due to the branch points which hinder helix formation of
amylopectin molecules (Solomon, 197 8).

The relative amount of amylose and amy10pectin in the starch granule is in the
range of one to three. Granules can vary in size and shape depending on their biological
origin. Under polarized light, intact granules exhibit a well-deﬁned birefringence pattern
with a dark cross which is an indication of the highly organized structure of the granules.

The starch molecules are radially oriented within the granule. The highly organized

 

 

structure is a result of the double helices formed between the outer chains of amylopectin
which contribute to the crystalline properties of starch granules (French, 1982), and are
responsible for the birefringent characteristic.

In nature, amylose and amylopectin are combined with other cellular components.
Amylose forms a complex with fatty acids, phospholipids and others found in cereal
starches. The formation of amylose-lipid complexes in starch alters the gelatinization
process. Starches vary in chemical and physical properties depending on their native

SOUICCS.

3.1.2 Starch gelatinization

The structure of the starch granule is signiﬁcantly changed when heated in water.
In the presence of excess water and at temperatures above the gelatinization threshhold,
gelatinization occurs and causes an increase in viscosity. Gelatinization causes starch to
solubilize in water under applied heat and can be described by non-equilibrium
thermodynamics in three processes (Blanshard, 1979):

1. Diffusion: water penetrates into the starch granule and interacts with amylose,
primarily in the amorphous region.

2. Swelling: the granule swells and expands its volume.

3. Melting or hydration: with progressive hydration, the granule size increases
under restricted movement. This causes the viscosity to increase up to the point where
the internal hydrogen bonds break and the granule ruptures. This results in tearing and
destruction of the crystalline regions, facilitating enzymatic attack on the unbound starch

molecules during starch hydrolysis.

The temperature required to produce complete gelatinization varies, depending on
the source of the starch. Using X-ray difﬁ'action and differential scanning calorimetry
thermograms, the gelatinization temperature of different native starches can be obtained
(Table 3.1).

 

 

 

 

 

Table 3.1. Gelatinization temperatures of several native starches.
Gelatinization temperature (°C)
Method:DSC Methodzmicroscope

Type of starch Onset Peak End Onset Peak End
Corn 65 7 1 77 65 69 76
Waxy com 65 72 80 64 70 78
Wheat 52 59 65 55 61 66
Rye 49 54 61 51 54 58
Oat 52 58 64 54 58 61
Rice 70 76 82 72 75 79
Potato 61 65 71 58 64 68
Tapioca 63 68 79 64 69 80
Anew root 67 75 85 69 76 84

 

 

 

 

 

 

 

 

 

Source: Reichelt (1983)

3.2 Amylolytic Enzymes

Amylolytic enzymes are used as biocatalysts to convert starches to sweeteners via
hydrolysis. These enzymes were originally called diastases but were later renamed
amylases. They can be classiﬁed into three groups (Norman, 1981): endo-amylases,
exo-amylases and debranching enzymes. The speciﬁc actions of these enzymes are
described by Boyce (1986).

Endo amylases are generally a—amylases which hydrolyze or— 1,4 glucosidic
bonds in amylose, amylopectin and related polysaccharides, and produce
oligosaccharides of varying chain lengths. They hydrolyze the bonds located in the inner

region of the substrate, which results in decreasing the viscosity of starch solutions.

\J

a -amylases attack the amylose fraction in two steps. Initially, a completely rapid

degradation of amylose occurs and produces maltose and maltotriose by a random attack.
The second step is slow hydrolysis of the oligosaccharides to form glucose and maltose
as the ﬁnal products. or —amylases also attack amylopectin and form glucose, maltose
and a series of tat-limit dextrins (oligosaccharides of DP > 4, containing a— 1, 6 branch
points). Malta -amylases can act on both insoluble and soluble starch to produce

dextrins and maltotrioses (Marc and Engasser, 1983).

a -amylases are derived from animal, plant and microbial sources, and have

different action patterns in starch hydrolysis. Microbial or —amylases are thermostable and
are used in liquefaction. In contrast, fungal or-amylases are thermolabile and are used

during sacchariﬁcation.

Exo-amylases consist of B —amylases and glucoamylases. These enzymes
hydrolyze or - 1, 4 glycosidic bonds in amylose, amylopectin and related polysaccharides.
[i -amylases hydrolyze the a- 1, 4 glycosidic bonds in starch by inverting the
conﬁguration at the 01 position of glucose from a. to [3, thereby the B-preﬁx.

B -amylases act at the non-reducing ends of the outer chains of starch and gradually
remove maltose units. However, they cannot by-pass the or - 1, 6 glycosidic branch
points.

Glucoamylases act in a manner similar to B -amylases, but hydrolyze or— 1, 6

glycosidic bonds at a slower rate. These enzymes remove glucose from the end of the
non-reducing chain in a stepwise manner. The decrease in viscosity induced by
exo-amylases is slower than that of endo-amylases.

Debranching enzymes are generally isoamylase and pullulanase. They hydrolyze
or - l, 6 linkages at branched points and produce glucose and maltose. Pullulanase

hydrolyzes linear and branched substrates when the side chain joined by the or- 1, 6
linkage consists of “at least two glucose units in length, but isoamylase requires at least
three units (Abdullah et al., 1965). If the branch point joins with one glucose molecule,
the branch point cannot be degraded by an enzyme. Amylo-1,6-glucoamy1ase can
hydrolyze the branch points, but only on a premodiﬁed substrate.

3.3 Processes for Producing Sweeteners

In general, sweeteners can be produced from starch through the three processes of
gelatinization, liquefaction and sacchariﬁcation. Starch must ﬁrst be gelatinized to
release amylose and amylopectin from the granule to facilitate enzymatic modiﬁcation.

Liquefaction dextrinizes starch by thinning or decreasing the viscosity of the starch
slurry. The traditional liquefaction agent used to be acid. The starch slurry (industrially,
25-40% dry solid, w/w) was acidiﬁed to pH 1.5-2.0 and heated to 140-155°C for 5-10
min to complete gelatinization and to produce a liqueﬁed product. These products were
impure because of undesirable compounds and color. Therefore, acid/enzyme and then
enzyme systems were introduced. The degree of starch degradation may be quantiﬁed in
terms of the dextrose equivalent (DE), where the reducing power of the liqueﬁed product
is compared with pure dextrose.

(rt-amylases derived from Bacillus subtilis can operate at temperatures of 85-87°C

and at 90-95°C for short periods of time. The more heat stable or -amylase derived from
Bacillus Iichemformis was developed and used at temperatures above 95°C and at
105-110°C for a short time. These thermostable enzymes require a cofactor, calcium ion,

to maintain their functionality during the process.

 

The liquefaction process can be carried out in either batch reactors or continuously

10

stirred tank reactors using a single enzyme or dual enzyme. The required ﬁnal DE in

liquefaction is 12-15 and can be up to 40. The residence time for these enzymatic

reactions varies, depending on enzyme dosage and starch concentration, but is generally

in the range of 1-2 hours. Starch hydrolysates with DB of 3-20 produced by bacterial

a-amylases are maltodextrins.

The liqueﬁed products (oligosaccharides) become substrates for enzymatic

sacchariﬁcation and are further hydrolyzed by two enzymes, glucoamylase and fungal

(rt-amylase. These enzymes can be used individually or in combination depending on

the sugar proﬁles required. Some industrial sweeteners are given in Table 3.2.

 

 

 

 

 

 

 

 

 

Table 3.2. Profile of industrial sugars produced from starch.

Product Glucose Maltose High High glucose Iso-
proﬁle Syrup maltose + maltose glucose
Dextrose

equivalent 96-98 40-45 48-55 56-68 98
Glucose 95-97 16-20 2-9 22-35 52
Maltose 1-2 41-44 48-55 40-48 -
Fructose - - - - 42
Isomaltose 0.5-2 - - - -
Maltotriose - - 15-16 - -

 

Source: Reichelt (1981)

3.4 Enzyme Kinetics

 

An enzyme is a biocatalyst used to accelerate the rate of chemical reactions by
lowering the required activation energy (barrier) without itself being consumed.
Enzymatic reactions are a factor of 10" faster than nonenzymatic reactions (Segel, 1975).
The enzyme possesses the active site where a speciﬁc substrate can bind. The active site

is a speciﬁc and precise spatial arrangement of amino acid residues (R-groups) that can

 

 

11

interact with complimentary groups on the substrate. The enzyme-substrate interaction is
analogous to a template or a lock-key model, which has successfully been used to explain
the majority of speciﬁcity patterns exhibited by enzymes (Segel, 1975).

Enzymes are high molecular weight proteins but the active site occupies only a
very small fraction. In fact, only few amino acid residues participate in substrate binding
or catalysis. The other amino acid residues maintain the protein in its tertiary structure
through the forces of electrostatic interactions, hydrogen bonds, disulﬁde bonds,
hydrophobic interactions and dipole—dipole interactions.

The ﬁrst rate equation for enzymatic reactions was derived in 1903 by Henri
(Segel, 1975). The overall reaction was described as:

k.
— -> k,
E+S ES E+P (3.1)
<— — — -—>
k 1

Henri’s equation was based on the following assumptions:

- The enzyme (E) is a catalyst.

- The enzyme reacts rapidly with substrate (S) to form an enzyme-substrate
complex (ES) in the transition state.

- Only a single E8 is formed and breaks down directly to free enzyme and product
(P).

- E, S and ES are at equilibrium, and the reverse reaction is much faster than the
rate of product formation.

- [S] is much larger than [B], therefore [ES] does not signiﬁcantly change.

- The product formation step is the limiting step.

- The rate or velocity is measured during the early stage of the reaction when the

reverse reaction is not signiﬁcant.

12

Henri’s equation can be expressed as:

[S
1 + Elf.) (3.2)

K

v=K

and was later modiﬁed to obtain the Michaelis-Menten (M-M) equation (Segel, 1975):

= kplE].[S] _ ths1

K.+[S] " K.+[Sl (3'3)

This equation was derived using an equilibrium expression for the enzyme and
substrate. The rapid equilibrium approach is the simplest and most direct method to
derive a kinetic expression without knowledge of the magnitude of the rate constants. In
many cases, both the rapid equilibrium and steady state approaches give the same form of
kinetic expression but the constants are deﬁned differently. However, the steady state
approach can lead to an equation too complicated for practical use.

The rate of reaction of glucoamylase derived from Schwanm’omyces castellir’ acting
on starch can be modeled by the M-M kinetics based on an initial rate study (Pasari et al.,
1988). The M-M equation can be used for malt [3 -amylases (Marc and Engasser, 1983).

3.5 Legal Aspects of Enzyme Uses

The majority of industrial enzymes are hydrolases and are used as additives or
processing aids in the baking (Fox, 1982) and other industries. The Codex Committee of
Food Additives has published a list of enzymes derived from plant, animal and microbial
sources, limiting their application in food products. In the United States, enzymes are
considered secondary direct food additives and are "Generally Recognized As Safe"
(GRAS) when manufactured with sound processing practices. A list of enzymes derived
from microorganisms is published in the Code of Federal Regulation (CFR). The
organisms used for enzymes and applied in the food industry must be known or shown to

 

13

be nonpathogenic and not produce toxic substances (Anon., 1987). The amount of
enzymes used depends on the type and source of the enzyme. For amyloglucosidase
derived from Rhizopus niveus, a level not to exceed 0.1% by weight of the gelatinized
starch is used. The use of enzymes from other microbial sources is limited by the "Food
Additive Amendment" (Anon., 1987).

The amount of residual enzyme allowed in the food product is not mentioned in the
GRAS list. However, if the enzyme remains active in the food product, it can cause
excessive starch hydrolysis with subsequent deterioration of product quality and a
decrease in shelf life.

3.6 Overview of Food Extruders

Extrusion is a widely used thermal process of industrial interest because it provides
a combination of three operations in one equipment: transporting, mixing and forming.
Food extrusion processes provide many advantages (Harper, 1978), among which are
high productivity in a single processing step and versatibility to handle a number of food
products under processing conditions such as high temperature short time (HTST). A
typical HTST extrusion process also occurs at high shear and high pressure (Linko ct al.,
1980).

Single-screw extruders (SSE) were initially used to provide mixing and
pressurization of high moisture dough through a die, to form desired shapes of food
products such as macaroni, ready-to-eat (RTE) cereals and puffed snacks (Harper, 1985).
When twin screw extruders (TSE) were introduced, they gained a lot of interest due to
expanded operational capabilities and broader ranges of application. The TSE improves

heat transfer and mixing and its self-wiping design helps eliminate stagnant zones.

14

A TSE can be differentiated by the direction of screw rotation (co- or
counter-rotating). Co-rotating and counter-rotating intermeshing twin screw extruders
provide different power consumption, RTD and mixing performances (Rauwendaal,
1981). The speciﬁc energy consumption (SEC) of counter-rotating extruders depends on
the shear stress while that of co-rotating extruders depends on the shear strain.
Co-rotating extruders provide a broad RTD and better distributive mixing capabilities due
to the high shear rate. On the other hand, counter-rotating extruders give a narrow RTD
and better dispersive mixing due to local, high shear stress in the intermeshing region.

The co-rotating TSE is more popular for food processing than the counter-rotating
unit because it provides a larger capacity and better mixing within the chambers, and can
be operated at a higher screw speed. In the co—rotating TSE, various screw sections are
used to perform different functions during the extrusion process. Screw sections with
fully intermeshing and self-wiping screws are desired for sticky or difﬁcult-to-convey
materials, to avoid surging. Mechanical energy dissipation and mixing can be increased
in TSE by using kneading disks such as two-lobed elements. If the screw section is fully
ﬁlled, increased mixing, better heat transfer and more viscous dissipation will occur.
These conditions allow a TSE to act as a heater rather than a cooler (Harper, 1985). The
fully filled condition is obtained by placing restrictions such as reversed pitch screws to
build up pressure. The pressure proﬁle in a TSE depends on the screw proﬁle and other
process variables (Valle et al., 1987).

Screw elements which are generally used in twin screw food extruders are
single-lobe, double-lobe and kneading discs or paddles. A single-lobe screw is used as a
discharge screw to build up pressure against back pressure at the die. It also provides

conveying, though not as well as the double-lobe screw, which is used as a feed screw.

 

 

 

 

15

Kneading discs are used to perform different processing tasks depending on the number
of discs and the staggering angle between the discs. They can provide good mixing and

generate interparticulate frictional heat to cook the product.

3.7 Extrusion Processing

Although food extrusion has been used to produce a variety of food products, the
process has not been well characterized because of complexities resulting from many
factors. These include composition of feed ingredients, moisture content, feed rate, screw
conﬁguration, screw speed, jacket temperature and die geometry. These factors affect the
viscosity of the extruded material, shear rate, throughput, die pressure, power
requirements, residence time, total strain, product temperature and product quality
(Harper, 198 1).

The theory for food extrusion has developed on the basis of polymer extrusion
models for ﬂuid and heat transport (Harper, 1981). However, unlike polymers which are
relatively homogeneous and singly extruded, food materials are heterogeneous and
consist of many ingredients. The interaction between these ingredients and water activity
leads to complex reactions such as starch gelatinization and protein denaturation. These
reactions cause irreversible changes in extruded materials which are quite opposite to the
reversible melting and irreversible polymerization reactions encountered during
thermoplastic extrusion.

The extrusion process consists of three basic mechanisms: feed conveying, melting
and metering. Although, most of the process models were developed for single screw

extruders, their important concepts are applicable to twin screw extruders.

 

16

The conveying rate is governed by the dynamic coefﬁcient of friction on the barrel
(f,) and screw (f.) and local pressure (Rauwendaal, 1986). If f, increases relative to f,, the
conveying rate will increase because the frictional force on the barrel induces a driving
force on the feed bed. If the local pressure increases, the conveying rate will decrease
because a positive pressure gradient creates a frictional force on the screw that retracts
the feed bed. The maximum forward transport will occur when the frictional force on the
screw is zero. If fb approaches f, under a relatively small pressure gradient, the retracting
force will dominate and there will be no forward movement of material. If there is a
positive pressure gradient, the ﬂow backs up toward the feed port The extruder is
unstable when the two frictional coefﬁcients are equivalent or when the frictional
coefﬁcients change due to variations in temperature.

In the melting zone, materials are cooked by heat generated from viscous
dissipation and heat supplied ﬁ'om the barrel. In a fully-ﬁlled TSE, power consumption
is proportional to the screw speed raised to an exponent of 2.0; the exponent is 1.9 for
partially-ﬁlled screws (Secor, 1986). The melting rate can be predicted if the material
temperature proﬁle is known.

The metering section acts as a pump to transport the material towards the die. The
throughput is governed by drag and pressure ﬂows (Rauwendaal, 1986) and can be
described by the expression:

Q = Qd-Qp GA)

01'

 

Q = PW” (1WD COS¢)_PW”3(§)

2 12h (3'5)

 

17

The throughput decreases if pressure ﬂow increases due to positive pressure
gradient and when the channel depth (H) increases. The output rate may also decrease
due to leakage through clearances between the ﬂight and the barrel. A corrected
throughput is obtained when the effect of leakage is included through correction factors.

One-dimensional ﬂow of Newtonian ﬂuids can be determined easily when the
operation occurs at steady state under isothermal, fully developed, laminar ﬂow
conditions in the absence of slip. In general, the ﬂuids are also assumed to be
incompressible and at constant density and viscosity; and gravity and inertial forces are
neglected. For one-dimensional ﬂow of a power law ﬂuid under conditions similar to
those for Newtonian ﬂuids above, the throughput in dimensionless form (Rauwendaal,
1986) is:

Q 2n
a = a: =1+2n

 

(3.6)

When cross channel ﬂow occurs, two-dimensional ﬂow results. If the helix angle
is zero, there is no cross channel ﬂow and the result of two-dimensional analysis will be
equivalent to the one-dimensional analysis (Rauwendaal, 1986). In the presence of cross
channel ﬂow, the polymer is subjected to a higher shear rate, which affects the viscosity
and, therefore, the ﬂow rate (Rauwendaal, 1986). Two-dimensional analysis shows that
throughput-pressure gradient relationships are different from those of one-dimensional
analysis when n is less than 0.8. If two-dimensional analysis is required, numerical
techniques can be used. However, complicated computation can be avoided by
incorporating correction factors for non-Newtonian ﬂuids in the Newtonian model as

follows (Rauwendaal, 1986):

 

Q = [4+n\pWH(nNDcos¢)_( 1 \PWHsig') (3.7)

10 j 2 1+2n} 4n

18

 

where u = m('~€,°°")'—l.

When Eq. 3.7 is used with n in a range of 0.3 to 1.0, the result is less than 10%
different from the two-dimensional analysis (Rauwendaal, 1986).

Food doughs are highly pseudoplastic. Therefore, drag and pressure ﬂows become
dependent and the predicted ﬂow rate can be under-estimated when the pressure gradient
is positive, and over-estimated when it is negative (Harper, 1981). Wyman (1975)
developed an axial velocity proﬁle for shallow channels in a TSE in a manner similar to
that of a SSE and found that pressure ﬂow induced ﬂow toward the die rather than away
from the die as in a SSE. The model did not include leakage between screws.

3.7.1 Mixing in extruders

Mixing is a process that reduces non-uniformity in composition, basically by
physical motion of ingredients. Convective motion normally occurs in laminar ﬂow of
viscous ﬂuids as a result of shear ﬂow or elongational ﬂow.

The mixing zone in extruders starts from the transition zone and ends at the die.
Mixing in the extruder is not uniform for all material elements (Rauwendaal, 1986)
because the element with a longer residence time has a greater mixing history. Therefore,
the mixing history of the material is difﬁcult to determine. The mixing process is
analyzed by determining the velocity proﬁles in the screw channel, and using these to
determine the local deformations in the ﬂuid. However, only the case of simpliﬁed ﬂow
of isothermal Newtonian ﬂuids is easily solved.

The degree of mixing has been described by using a residence time distribution
function. The RTD provides information about mixing in the axial direction but not in
the transverse direction which governs laminar mixing of viscous ﬂuids (Bigg, 1975).

Dispersion mechanisms were important in single-screw and twin screw compounding

 

19

extruders (Eise et al., 1983). Because ﬂuid elements in the extruder are subjected to
various amounts of strain which are directly related to mixing, a strain distribution
function (SDF) can be calculated in a manner similar to the RTD. The SDF for a
continuous mixer, f(y)dy, is deﬁned as the fraction of exiting ﬂow that experienced a
strain between 7 and 7+ (17. This represents the probability that an entering ﬂuid element
will exit with that strain. The cumulative SDF is expressed by (Tadmor and Klein,
1970):

M) = 17mm (3.8)

The F(y) is the fraction of exiting ﬂow with strain less than or equal to y. The

mean strain of the exiting stream is calculated by (Tadmor and Klein, 1970):
i = L mm (3.9)

Pinto and Tadmor (1970) developed the weighted average total strain (WATS) to
describe mixing in non-homogeneous ﬂow ﬁelds in single screw extruders. The WATS

is expressed as:
i. = framed: (3.10)
1.

WATS is a theoretical concept that is difﬁcult to measure experimentally. The
efﬁciency of mixing is hard to determine because of the lack of a unifying mixing theory
(Ottino and Chella, 1983). The degree of mixing should be determined based on the
physical process, the ﬂuid mechanics and the compatibility of the components being
mixed (Ottino and Chella, 1983).

 

20

It is important to understand how the materials are fed into the extruder and what
the mixing mechanisms are. In a SSE, mixing is induced by shear (rotational) ﬂow; in a
TSE, the mechanism is elongational (irrotational) ﬂow. Rotational ﬂow causes stretching
of materials without breakup while irrotational ﬂow causes breakup of the material
matrix, which enhances mixing. The viscosity and elasticity (viscoelasticity) of the
components govern the mode of dispersion. The lower viscosity component may
encapsulate the higher viscosity component. The degree of mixing also increases with
increased interfacial areas. Mohr et a1. (1957) reported that interfacial areas increased
with increased shear strain. Laminar mixing can be analyzed if the exact ﬂow patterns
are known. The velocity proﬁle in the screw mixing zone is normally complex, making
quantitative analysis of the mixing section difﬁcult.

3.7.2 Process instability

Process instabilities result from large periodic variations in process and product
variables. The extrusion process can undergo stable oscillations for long periods of time,
in spite of constant feed rates. The oscillations are of the same order as or longer than the
extruder residence time. The oscillations involve the entire extrusion process and are
typically caused by a ﬂuctuation in feed moisture content (Roberts and Guy, 1986). As
moisture decreases, viscosity increases and the ﬂow through the die decreases while the
die pressure increases. The material behind the die is then subjected to shear for a longer
period. This results in viscosity reduction, and the material ﬂows through the die quickly
while the die pressure decreases. The inconsistent ﬂow through the die, alternately
having high and low viscosity, may continue until a stable oscillation is produced or may

be damped out to approach a steady state.

 

21

Roberts and Guy (1987) have observed that extrusion cooking of expanded wheat
ﬂour products exhibited two steady states in processes carried out under the same
independent process variables. The two equilibrium states were explained as resulting
from moisture movement within the screw system, and were governed by the degree of
ﬁll, torque, pressure and mass temperature. The ﬁrst state occurred at high torque and
low die pressure, and with low density products. A good steam barrier was formed under
the ﬁlled screw in this state. If the steam escaped and backed up to the feed section to
moisten the feed, less energy input resulted. The ﬁrst state was then shifted to the second
state, which occurred at low torque and high die pressure, and with high density products.

3.7.3 Application of polymer extrusion theory to food extrusion

The simpliﬁed extrusion theory for isothermal, Newtonian ﬂow has been
successfully applied to some food systems to measure effects of changing screw
geometry and extruder operating conditions (Harper, 1981). Using soy ﬂour data from
Mustakas et al. (1970), Fricke et al. (1977) found that the output rate was unaffected by
pressure ﬂow because drag ﬂow dominated. The ﬂow was proportional to screw speed at
low screw speeds but not at high screw speeds. Jasberg et al. (1979) employed the plug
ﬂow model of Darnell and M01 (1956) to predict the output rate of soy dough at low
temperatures (30-50°C), and found agreement between predicted and experimental
results. The output rate was signiﬁcantly affected by moisture and screw speed but not
by die size. They also reported that increasing ﬂow rate did not affect the back pressure

from the die, contradicting the results from plastic extrusion.

3.8 Rheological Changes During Extrusion

Changes in rheological properties of food materials during extrusion has been

investigated by Cervone and Harper (1978), Jao et al. (1978), Luxenburg et al. (1985),

22

Mackey (1989), Morgan et al. (1989) and Remsen and Clark (1978). Jao et a1. (1978)
reported that the apparent viscosity of defatted soy dough was affected by moisture, shear
rate and temperature during extrusion. Luxenburg et al. (1985) used DSC to follow the
enthalpy change of soy ﬂour dough during extrusion cooking. They found that the
rheological change was due to conformation of protein molecules rather than protein
crosslinking.

Most viscosity models for food extrusion have been developed empirically without
incorporating temperature-time history and/or strain history effects. Remsen and Clark
(1978) combined temperature-time history in a model for soy ﬂour suspensions (70-75%
moisture), but the model exponentially approached inﬁnity at large temperature-time
histories. In contrast, Morgan et al. (1989) have recently developed a model that
approaches a ﬁnite value for temperature-time histories and shear rates encountered in
extrusion. The model adequately predicted effects of shear rate, moisture content,
temperature and temperature-time history on apparent viscosity of defatted soy ﬂour
dough undergoing heat induced protein denaturation. The model showed good agreement
with experimental data for a wide range of conditions. Mackey (1989) has also shown
that the effects of temperature-time and strain histories are important and should be

incorporated in models predicting apparent viscosity of starch doughs during extrusion.

3.9 Ideal and Non-ideal Reactors

Ideal reactors consist of batch reactors, continuously-stirred tank reactors (CSTR)
and plug ﬂow reactors (PFR). Ideal ﬂow patterns refer to complete backmixing (in batch
and CSTR) and zero backmixing (in PFR). In real reactors, ﬂow patterns fall somewhere

23

between the two ideal ﬂows. The deviation from ideality results from channelling of
ﬂuids, recycling of ﬂuids or creation of stagnant regions (Levenspiel, 1972), all of which
can occur in food extruders.

Non-ideal ﬂow causes problems in scale-up and design because the magnitude of
the nonideality is not always predictable. To predict the ﬂow behavior of ﬂuids in the
reactor, it is necessary to determine the length of time an individual element spends while
passing through the reactor, resulting in a residence time distribution of the ﬂowing ﬂuid.
Fogler (1987) stated that non-ideal reactors can be characterized by residence time
distribution (RTD), degree of mixing and the ﬂow model.

3.9.1 Residence time distribution

The RTD provides information on the type of mixing taking place within the
reactor. The residence time in batch and plug ﬂow reactors is the same for all ﬂuid
elements, while that in a CSTR varies. The RTD can be directly determined by
stimulus-response techniques. The stimulus is generally a tracer input into the ﬂuid
entering the reactor and the response is a time record of the tracer leaving the reactor.
Any material can be used as a tracer if it can be detected and does not disturb the ﬂow of
interest. The distribution of the time required by ﬂuids leaving the reactor is called the
exit age distribution, E, or the RTD function (Levenspiel, 1972) where E is deﬁned as:

(2(t)

E(t) -
J; C (t)dt

(3.11)

In normalized form,

fund: = 1 (3.12)

24

The C curve is a plot of concentration against the residence time. The F curve is a
plot of fraction of tracer concentration against the residence time. The F parameter is
deﬁned by:

(3(t)
C.

 

F(r) = (3.13)

The RTD from different reactors can be compared using three moments: the mean

residence time (i), the variance (0") and the skewness (s3), expressed respectively as:

i = f-tE(t)dt (3.14)

o2 = fo-azaoya (3.15)
and

33 = f'a—Ifauya (3.16)

3.9.2 Extent of chemical reaction

In a linear system, the stimulus-response experiment at steady state is linear in
concentration, and the output tracer concentration is directly proportional to the input
tracer concentration. If the reaction is ﬁrst order, the extent of reaction can be
determined directly from the reaction rate expression and tracer information (Fogler,

1987; Levenspiel, 1972) which can be expressed as:
E, = f C,(t)E(t)dt (3.17)

For a nonlinear process, where the reaction is not ﬁrst order, the ﬂow model must
be incorporated into the analysis. Models for predicting extent of reaction from RTD
data can be developed from a number of adjustable parameters (Fogler, 1987).

Segregation and maximum mixcdness models are zero adjustable parameters. CSTRs in

25

series and the dispersion model have one parameter: the number of reactors or the
dispersion number. A combination of ideal reactors may involve two or more
parameters.

3.9.3 Residence time distribution in single and twin screw extruders

The ﬂow pattern in an extruder has been investigated using RTD data by many
researchers (Altomare and Ghossi, 1985; Colonna et al., 1983; Davidson et al., 1983;
Valle etal., 1987; Lidor and Tadmor, 1976; Olkku et al., 1979; Todd, 1975; van
Zuilichem et al., 1973; Wolf and Resnick, 1963; Wolf and White, 1976; and Wolf et al.,
1986). The RTD curve in an extruder has a long tail due to material held in dead spaces.

Wolf and Resnick (1963) proposed an F-function to describe the RTD for real

systems:

F(t) = 1- [5(7) (3.18)
The parameter 8 is a measure of mixing efﬁciency. It is equal to unity for perfect

mixing and approaches inﬁnity for pure plug ﬂow. It is greater than unity if the system
consists of dead spaces. An error in determining mean residence time can cause 8 to be
larger or smaller than unity. The parameter 8 is a measure of the phase shift in the
system. Both parameters are determined from a plot of 1n (l-F) against t/t' which yields a
straight line.

Davidson et al. (1983) studied the RTD of wheat starch in a SSE and ﬁtted their
data with two ﬂow models developed by Wolf and Resnick (1963) and Levich et al.
(1967). They found that the Wolf-Resnick multistage model ﬁt the tail portion well
while the Levich model ﬁt the initial portion. The fraction of material held up in the dead
space was 5-20%. The remainder passed directly through with an RTD close to plug
ﬂow. They found that the dispersion number was larger than 0.01.

26

Based on the RTD data of PVC polymers, Wolf et al. (1986) reported that the ﬂow
pattern in a counter-rotating twin screw extruder was near plug ﬂow. The F(9) function

was similar to that for solid conveying (plug ﬂow). The normalized 0 was deﬁned as:

6 =

(3.19)

The effect of process variables on RTD in a twin screw extruder has been
extensively examined by Altomare and Ghossi (1986). The process variables studied
were screw proﬁle, feed rate, screw speed, moisture content, barrel temperature and die
size. They found that the mean residence time decreased with increased screw speed and
feed rate, but decreased with decreased moisture content. The screw inducing higher
shear resulted in a broader and longer RTD. Using various die diameters, they found that
pressure had little effect on the RTD under a given process condition. A normalized
delay time (ratio of time that tracer ﬁrst appears in the extrudate to average residence
time) was about 0.5. Larger values for a single screw extruder were reported by
Davidson et al. (1983). The delay time for Newtonian ﬂuids in a SSE was 0.75 (Bigg and
Middleman, 1974). Because the dough viscosity is high, ﬂow in the TSE may approach
that of laminar pipe ﬂow. Altomare and Ghossi (1986) found that their data correlated
well with the model of Wolf and White (1976) which combined complete mixing and
plug ﬂow.

The temperature and pressure proﬁles in a TSE are affected by screw proﬁle, barrel
temperature, feed rate, moisture content, screw speed and die size (Valle et al., 1987).
Among these variables, moisture content and feed rate have the greatest effect on
pressure and temperature proﬁles. Temperature and pressure decreased with increased
moisture content. The ratio of feed rate to screw speed (Q/N) affected the degree of ﬁll

in the screw channel. The degree of ﬁll increased with increased feed rate. An increase

27

in screw speed led to increased shear, decreased viscosity and decreased pressure.

3.10 Starch Hydrolysis in Twin Screw Extruders

The use of an HTST extrusion cooker as a biochemical reactor for enzymatic
reactions has been extensively investigated by Linko et al. (1979), Linko et al. (1983a,
1983b), Linko et al. (1984), Reinikainen et al. (1986) and Chouvel et al. (1983).

Thermomechanical modiﬁcation of starch structure during extrusion cooking has
been observed by Christensen (1987), Davidson et al. (1984), Diosady et al. (1985),
Linko et al. (1983a), Linko et al. (1980), Linko et al. (1981), Mercier (1979), Mercier and
Feillet (1975), Gomez and Aguilera (1983), Colonna et al. (1984), Reinikainen et al.
(1986), and Darnoko and Artz (1988). Thermal and mechanical energy input caused
cleavages of hydrogen bonds between amylopectins and resulted in reduced viscosity and
increased solubility (Mercier and Feillet, 1975 ). Mercier (1979) found that extrusion
cooking solubilizes cereal starches to a certain extent, which increases digestibility for
human nutrition. The amount of soluble starch increased with decreased feed moisture
contents, but no formation of maltodextrins was observed (Mercier and Feillet, 197 5 ).
Christensen (1987) demonstrated that extrusion improved the texture and palatability of
an extruded dog food and also increased extruder throughput when or -amylase was used
during extrusion of corn grits.

Extrusion of starch is sufﬁcient pretreatment for subsequent enzymatic
sacchariﬁcation to obtain DE values of up to 98 (Linko etal., 1980). Extrusion cooking
causes thermomechanical destruction of the starch structure and allows thermostable
(Jr-amylase to break down the starch molecules (Linko et al., 1983a). Cassava starch was

partially hydrolyzed in the TSE prior to adding saccharifying enzyme to continue

28

hydrolysis of starch to glucose (Darnoko and Artz, 1988). The degree of starch
hydrolysis increased at low pH and high temperature. The glucose concentration in the
extruded starch after sacchariﬁcation was higher than that in the acid-liquiﬁed starch.

Pretreatment of substrates can be thermomechanical without addition of enzyme or
with enzyme added in a single pass of HTST-extrusion cooking. The combined effect of
thermomechanical and enzymatic starch liquefaction in HTST-extrusion was investigated
by Linko et al. (1983a) and Chay et al. (1984) who used Termamyl 60L or—amylase, and
by Hakulin et al. (1983) who used Termamyl 120L a-amylase. Both enzymes were
extremely thermostable under extrusion conditions, but were almost completely
inactivated at 160°C (Linko et al., 1983a, 1983b).

Hydrolyzed wheat starch at DE values of 25—30 was obtained using 0.9% w/w
Termamyl 120L 0t —amy1ase in a twin screw extruder at 120°C, 1500 g/min feed rate and
250 rpm screw speed (Hakulin et al., 1983). It was found that feeding premixed
starch-water slurry into the extruder gave higher DE values than feeding starch and water
separately. The addition of enzyme to materials in the extruder either before or after
starch gelatinization did not signiﬁcantly affect DE values after the subsequent
sacchariﬁcation process. The addition of both or —amylase and glucoamylase into the
extruder accelerated the sacchariﬁcation process after extrusion.

Both starchy and cellulosic materials can be pregelatinized and preliqueﬁed by
HTST-extrusion cooking, which leads to a reduction of reaction time for subsequent
sacchariﬁcation and ethanol fermentation (Linko et al., 1984). Pretreated barley starch
was sacchariﬁed by glucoamylase at 60°C and a pH of 4.5 (Linko et al., 1979) and was
used for ethanol fermentation by yeast at 60°C (Linko et al., 1983b).

29

./ 3.10.1 Liquefaction

The effect of pH, temperature, moisture content, enzyme to substrate ratio, and
calcium ions on extent of enzymatic liquefaction of pregelatinized corn starch in a TSE
was investigated by Chouvel et al. (1983). The screw proﬁle consisted of both
forwarding and reversing screw elements. Bacillus lichenifomris Termamyl 60L
or -amylase (Novo Industries, France) was used. They observed that the optimum pH
was 6.0. The extent of hydrolysis increased with increased moisture content from 40% to
70%, and with increased enzyme to substrate ratio from 5 to 30 ml/kg starch. The
addition of calcium ions resulted in a higher heat tolerance of enzymes during a
temperature rise from 100 to 145°C.

Continuous liquefaction of pregelatinized starch was carried out at relatively high
dry solid contents (up to 60%) in a short period of time, using an a—amylase dosage
higher than the usual Novo process (Chouvel et al., 1983). The product proﬁle from the
hydrolysis contained less than 1% glucose and relatively high proportions of maltotriose,
maltopentaose and maltohexaose. The authors also examined the combination of
gelatinization and liquefaction of raw corn starch in the TSE. About 20-30% of
glycosidic bonds were hydrolyzed during gelatinization and liquefaction.

J 3.10.2 Enzyme inactivation in the extruder

Extrusion cooking may be used to inactivate enzymes such as cereal a—amylase,
lipase, lipoxidase, peroxidase and urease present in food materials (Linko et al., 1983a).
These enzymes interact with food constituents and cause deterioration. However, some
residual enzymes such as or -amylase may remain even after extrusion cooking (Linko et

al., 1980; Linko et al., 1983a, 1983b; and Chouvel et al., 1983), especially when exposed

to short residence times in the extruder (Linko et al., 1980).

30

The highest residual or -amy1ase activities in TSE were obtained at short residence

times, relatively high moisture content and low temperature (Linko et al., 1980).

Residual enzyme activity was small under experimental conditions of moisture contents

less than 30%, with relatively complete inactivation at 25% moisture content. Above

125°C, enzyme deactivation was pronounced, however some residual enzyme activity

was detected in samples extruded at 140°C (Linko et al., 1980).

3.11 Nomenclature

C.
C(t)

C.
D

Input concentration of tracer, mole/l

Output concentration of tracer, mole/l

Mean concentration of reactant in the exit stream, mole/l

Concentration of reactant at any time, mole/l

Barrel diameter, m

Total enzyme concentration, ml of enzyme/ml of solution

Exit age distribution, s'1

Dynamic coefﬁcient of friction on the screw surface, dimensionless
Dynamic coefﬁcient of friction on the barrel surface, dimensionless
Channel depth, m

Rate constants in Michaelis-Menten model, s'1

Dissociation constant, (mole/ll1 (Eq. 3.2)

Constant for a particular enzyme, s’1 (Eq. 3.2)

Consistency coefﬁcient, Pas"

Power law index, dimensionless

Screw speed, RPS

<<
‘2

N

31

Local pressure, Pa
Number of parallel ﬂights, dimensionless

Net ﬂow rate, m3/h

Pure drag ﬂow rate, m3/h
Pure pressure ﬂow rate, m°/h

Dimensionless volumetric ﬂow rate (Eq. 3.6)

Substrate concentration, mole/l
Skewness, s3

Residence time, 5

Mean residence time, s

Initial velocity (Eq. 3.2), mole/l/s
Maximum reaction rate, mole/Us
Screw width, m

Down channel or axial distance, 111

Greek symbols
Coefﬁcient, dimensionless (Eq. 3.18)

System phase shift (Eq. 3.18)

Effective viscosity, Pa.s

Minimum strain, m/m

Mean strain, m/m

Strain undergone by a ﬂuid element at time t, m/m

Newtonian viscosity, Pa.s

32

0’ Variance, s2

4’ Screw helix angle, rad.

0 Normalized time, dimensionless
3.12 References

Abdullah, M., Cattley, B.J., Lee, E.Y.C., Robyt, J., Wallenfels, K. and Whelan, WJ.
1965. The mechanism of carbohydrase action. XI. pullulanase, an enzyme speciﬁc
for the hydrolysis of or- 1,6 bonds in amylaceous oligo-and polysaccharides.
Cereal Chem. 43:111.

Altomare, RE. and Ghossi, P. 1986. An analysis of residence time distribution patterns
in a twin screw cooking extruder. Biotech. Prog. 2(3):157.

Anon. 1987. Enzyme preparations and microorganisms in subpart B, part 173 of
Secondary direct food additives permitted in food for human consumption. Food
and Drug Administration, HHS, 21 CFR Ch.1 (4-1-1987 edition).

Banks, W. and Greenwood, CT. 1975. Sm!) and Its Components. Halsted Press, New
York.

Bigg, D. 1975. On mixing in polymer ﬂow systems. Polym. Eng. Sci. 15(9):684.

Bigg, D. and Middleman, S. 1974. Mixing in a screw extruder: a model for residence
time distribution and strain. Ind. Eng. Chem. Fund. 13:66.

Blanshard, J.M.V. 1979. Physicochemical aspects of starch gelatinization. In:
2212mm in Foods. Blanshard, J.M.V. and Mitchell, J.R. (Editors).
Butterworth, London.

Boyce, C.O.L. 1986. Novo’s Handbook of Practical Biotechnology. A publication of
Novo Industries A/S Enzymes Division, Bagsvaerd, Denmark.

33

Cervone, NW. and Harper, J .M 1978. Viscosity of an intermediate moisture dough. J.
Food Proc. Eng. 2:83.

Chay, P.B., Chouvel, H., Cheftel, J.C., Ghommidh, C. and Navarro, J .M. 1984.
Extrusion-hydrolyzed starch and ﬂours as fermentation substrates for ethanol
production. Lebensm-Wiss. u.-Technol. 17:257.

Chouvel, H., Chay, RB. and 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(6):346.

Christensen, RM. 1987. Application of enzymes in food extrusion process. Paper
presented at the 7th. World congress of Food Science and Technology, Sept 28th -
Oct 2nd., Singapore.

Colonna, P., Doublier, J .L., Melcion, J .P., de Monredon and Mercier, C. 1984. Extrusion
cooking and drum drying of wheat starch. 1. Physical and macromolecular
modiﬁcations. Cereal Chem. 61(6):538.

Colonna, P., Melcion, J.P., Vergnes, B. and Mercier, C. 1983. Flow, mixing and
residence time distribution of maize starch within a twin-screw extruder with a
longitudinally-split barrel. J. Cereal Sci. 1:115.

Darnell, W.H. and M01, E.A.J. 1956. Solid conveying in extruders. SPE J. 12:20.

Damoko and Artz, W.E. 1988. Twin-screw extrusion as a continuous pretreatment
process for the enzymatic hydrolysis of cassava. J. Food Sci. 53(6):1792.

Davidson, V.J., Paton, D., Diosady, LL. and Spratt, W.A. 1983. Residence time
distribution of wheat starch in a single screw extruder. J. Food Sci. 48:1157.

Davidson, V.J., Paton, D., Diosady, LL. and Larocque, G. 1984. Degradation of wheat
starch in a single screw extruder: characteristics of extruded starch polymers. J.
Food Sci. 49:454.

34

Diosady, L.L., Paton, D., Rosen, N ., Rubin, LJ. and Athanassoulias, C. 1985.
Degradation of wheat starch in a single screw extruder: Mechano-kinetic
breakdown of cooked starch. J. Food Sci. 50:1697.

Eise, K., Curry, J. and Nangeroni, J .F. 1983. Compounding extruders for improved
polyblends. Polym. Eng. Sci. 23(11):642.

French, D. 1982. Physical and chemical organization of starch granules. In: Starch
Chemistry and Ieehnelegy, 2nd ed. Whistler, R.L., Paschall, BF. and BeMiller,
J.N. (Editors). Academic Press, New York.

Fogler, HS. 1987. Elements ef Chemieal Reaenen Engineering. Prentice-Hall, New
jersey.

Fox, PF. and Mulvihill, D.M. 1982. Enzyme in wheat, ﬂour, and bread. In Pomeranz:
Anyanees in Cereal Science and mnnelegy. Amer. Assoc. Cereal Chem., St. Paul,
Minesota.

Fricke, A.L., Clark, J .P., and Mason, TR 1977. Cooking and drying of fortiﬁed cereal
foods: Extruder design. Chem. Eng. Prog. Symp. Ser. 73(163):134.

Gomez, M.H. and Aguilera, J .M. 1983. Changes in starch fraction during extrusion
cooking of corn. J. Food Sci. 48:378.

Hakulin, S., Linko, Y.Y., Linko, P., Sciler, K., Seibel, W. and Detmold. 1983.
Enzymatic conversion of starch in twin-screw HTST-extruder. Starch/Starke
35(12):41 1.

Harper, J .M. 1978. Extrusion processing of food. Food Technol. 32(7):67.

Harper, J .M. 1981. Exnmjen ef Bends. Volume 1. CRC Press, Inc. Boca Raton, Florida.

Harper, J .M. 1985. Processing characteristics of food extruders. In: Feed Engineering
and Rmeess Annlieanen. Volume 2. Unit Operation. Le Maguer, M. and Jelen, P.
(Editors), Elesevier Applied Science Publishers, London and New York.

35

Hood, LF. 1982. Current concepts of starch structure. In: M W. IFI‘
Basic Symposium Series. Lineback, DR. and Inglett, GE. (Editors). AVI
Publishing, Connecticut.

Jao, Y.C., Chen, A.H., Lewandowski, D. and Irwin, W.E. 197 8. Engineering analysis of
soy dough rheology in extrusion. J. Food Proc. Eng. 2:97.

Jasberg, B.K., Mustakas, GO and Bagley, EB. 1979. Extrusion of defatted soy ﬂakes -
model of a plug ﬂow process. J. Rheology 23(4) :437.

Levenspiel, O. 1972. Qhemieal Reaen’en Engineering. 2nd edition. John Wiley & Sons,
New York.

Levich, V.G., Markin, V.S. and Chismadzhev, V.A. 1967. On hydrodynamic mixing in a
model of a porous medium with stagnant zones. Chem. Eng. Sci. 22:1367.

Lidor, G. and Tadmor, Z. 1976. Theoretical analysis of residence time distribution
functions and strain distribution functions in plasticating screw extruders. Polym.
Eng. Sci. 16(6):450.

Linko, P., Hakulin, S. and Linko, Y.Y. 1984. HTST-extrusion cooking in ethanol
production from starchy materials. Enzyme Microb. Technol. 6:457.

Linko, P., Linko, Y.Y. and Olkku, J. 1983a. Extrusion cooking and bioconversions. J.
Food Eng. 2:243.

Linko, P., Hakulin, S. and Linko, Y.Y. 1983b. Extrusion cooking of barley starch for the
production of glucose syrup and ethanol. J. Cereal Sci. 1:275.

Linko, Y.Y., Lindroos, A. and Linko, P. 1979. Soluble and immobilized enzyme
technology in bioconversion of barley starch. Enzyme Microb. Technol. 1:273.

Linko, P., Colonna, P. and Mercier, C. 1981. High temperature short time extrusion

cooking. In Pomeranz: Adxanees in Cereal Seienee and Ieehnelegy. Amer. Assoc.
Cereal Chem., St. Paul, Minesota.

36

Linko, Y.Y., Vuorinen, H., Olkku, J. and Linko, P. 1980. The effect of HTST-extrusion
on retention of cereal a-amylase activity and on enzymatic hydrolysis of barley
starch. IanQQdecessEnaineetina.EnmneEnain§erinsinEmdmssine
Linko, P. and Larinkari, J. (Editors). Volume 2. Applied Science Publishers,
London.

Luxenburg, L.A., Baird, DC. and Joseph, EC. 1985. Background studies in the
modeling of extrusion cooking processes for soy ﬂour doughs. Biotech. Prog.
1(1):33.

Marc, A. and Engassar, J.M. 1983. A kinetic model of starch hydrolysis by or-and

B -amylases during mashing. Biotech. Bioeng. 25:481.

Mercier, C. and Feillet, P. 1975. Modiﬁcation of carbohydrate components by
extrusion-cooking of cereal products. Cereal Chem. 52(1):283.

Mercier, C. 1979. Structure and digestibility alterations of cereal starches by twin-screw
extrusion cooking. J. Food Proc. Eng. 1:795.

Mackey, KL. 1989. A generalized viscosity model for the cooking extrusion of starch
based products. Ph.D. Dissertation, Food Science and Human Nutrition, Michigan
State University, East Lansing, Michigan.

Mohr, W.D., Saxton, R.L. and Jepoon, CH. 1957. Theory of mixing in the single screw
extruders. Ind. Eng. Chem. 49:1857.

Morgan, R.G., Steffe, J .F. and Ofoli, R.Y. 1989. A generalized viscosity model for
extrusion of protein doughs. J. Food Proc. Eng. 11:55.

Mustakas, G.C., Albrecht, W.J., Bookwalter, G.N., McGhee, JE, Kwolek, W.F., and
Grifﬁn, EL. 1970. Extruder-processing to improve nutritional quality, ﬂavor and
keeping quality of full-fat soy ﬂour. Food Technol. 24: 1290.

37

Norman, BE. 1981. New developments in starch technology. In: Enzymes and Engd
Messing Birch, G.G., Blakebrough, N. and Parker, K.J. (Editors). Applied
Science Publishers, London.

Olkku, J ., Antila, J ., Heikkinen, J. and Linko, P. 1979. Residence time distribution in a
twin screw extruder. In: Feed Precess Eng'meering. Linko, P., Malkki, Y., Olkku, J.
and Larinkari, J. (Editors). Volume 1. Applied Science Publishers, London.

Ottino, J .M. and Chella, R. 1983. Laminar mixing of polymeric liquids: a brief review
and recent theoretical developments. Polym. Eng. Sci. 23(7):357.

Pasari, A.B., Korus, RA. and Heimsch, RC. 1988. Kinetics of the amylase system of
schwanniomyces castellii. Enzyme Microb. Technol. 10:156.

Pinto, G. and Tadmor, Z. 1970. Mixing and residence time distribution in melt screw
extruders. Polym. Eng. Sci. 10:270.

Rauwendaal, C]. 1981. Analysis and experimental evaluation of twin screw extruders.
Polym. Eng. Sci. 21(16):1092.

Rauwendaal, C.J. 1986. Eelymex Exnnsien. Hanser Publishers, New York.

Reichelt, JR. 1983. Starch. In: Inﬁnsnial W, The ADDIEEIIQB MW in
mm. Godfrey, T. and Reichelt, J. (Editors), Nature Press, New York.

Reinikainen, P., Suortti, T., Olkku, J ., Malkki, Y. and Linko, P. 1986. Extrusion cooking
in enzymatic liquefaction of wheat starch. Starch 38:20.

Remsen, CH. and Clark, J .P. 1978. A viscosity model for a cooking dough. J. Food
Proc. Eng. 2:39.

Roberts, SA. and Guy, R.C.E. 1986. Instabilities in an extrusion-cooker: a simple
model. J. Food Eng. 5:7.

Roberts, SA. and Guy, R.C.E. 1987. Metastable states in a food extrusion cooker. J.
Food Eng. 6:103.

38

Secor, RM. 1986. Power consumption of partially ﬁlled twin screw extruders. Polym.
Eng. Sci. 26(14):969.
313861.111. 1975. Enzyme Kineu'cs Behaxicr and Analysis of Rapid Earrﬂihrium and

Steady State Em 5x51521113. John Wiley & Sons, New York.
Solomon, B. 1978. Starch hydrolysis by immobilized enzymes, industrial applications.

In: Adxmes 'm Bjeehemjeal Eng'megmg. Ghose, T.K., Fiechter, A. and
Blakebrough, N. (Editors). Springer-Verlag, New York.

Tadmor, Z. and Klein, 1. 1970. Engineering Principles ef Elnsneamjg Emsjen. Van
N ostrand Reinhold, New York.

Todd, DB. 1975. Residence time distribution in twin-screw extruders. Polym Eng. Sci.
15(6):437.

Valle, D.G., Tayeb, J. and Melcion, J .P. 1987. Relationship of extrusion variables with
pressure and temperature during twin screw extrusion cooking of starch. J. Food
Eng. 6:423.

van Zuilichem, D.J., de Swart, J .G. and Buisman, G. 1973. Residence time distributions
in an extruder. Lebensm-Wiss. u. Technol. 6(5): 184.

Williams, J .M. 1968. The chemical evidence for the structure of starch. In: Staten and Its
W, 4th ed. Radley, J .A. (Editor), Chapman and Hall, London.

Wolf, D. and White, DB. 197 6. Experimental study of the residence time distribution in
plasticating screw extruders. AIChE J. 22: 122.

Wolf, D. and Resnick, W. 1963. Residence time distribution in real systems. Ind. Eng.
Chem. Fund. 2:287.

Wolf, D., Holin, N. and White, DH. 1986. Residence time distribution in a commercial
twin-screw extruder. Polym. Eng. Sci. 26(9):640.

39

Wolfrom, ML. and Khadem, HE. 1965. Chemical evidence for the structure of starch.
In: Starch: Chemistry and Ieehnelegy. Whistler, R.L. and Paschall, E.F. (Editors).
Volume 1. Academic Press, New York.

Wyman, CE. 1975. Theoretical model for intermeshing twin screw extruders: axial
velocity proﬁle for shallow channels. Polym. Eng. Sci. 15(8):606.

4o

4 STARCH HYDROLYSIS KINETICS OF THERMOSTABLE
ALPHA-AMYLASE
4.1 Abstract

Reaction kinetics for thermostable a -amylase were determined by two

experiments: an initial rate study at low starch concentrations (1-8% w/w), and a
continuous rate study at a high starch concentration of 40%. The Michaelis-Menten
model was used to describe the kinetics at low starch concentrations, and a modiﬁed ﬁrst
order model was used at high concentration. There appeared to be no substrate inhibition
at the high starch concentration. The rate of product formation (in terms of DE values)

was found to be proportional to the square root of the rate of starch consumption.

4.2 Introduction and Background

Reaction kinetics are important in reactor design and analyses. They are used to
predict the extent of substrate conversion under a given set of conditions and can be used
to evaluate reactor performance. Reaction kinetics can be determined from experiments
using a batch or continuous system.

Theoretically, the activity of or -amylase could be characterized by the number of

bonds broken in a D-glucan per unit time. Since this is difﬁcult to measure
experimentally, lumped methods for determining the number of reducing groups are used
instead. However, these lumped methods are not related to stoichiometry (Greenwood
and Milne, 1986). Equimolar quantities of maltodextrins do not have equal reducing
power to an alkaline 3,5 dinitrosalicylate reagent (DN S), which is widely used to track
the rate of reaction. Also, there is a problem with thermal degradation of the alkaline

reagents which contain ferricyanide or copper. Therefore, the lumped methods cannot be

41

used to determine absolute values of maximum velocities or kinetic constants. Iodine
stain-ability of whole starch and viscometric methods cannot be used to estimate the rate
of bond scission, either. It appears that no suitable methods for the absolute
determination of maximum velocity or kinetic constants has been found for a -amy1ase
(Greenwood and Milne, 1968).

The relative reducing power of low molecular weight sugars is presented in Table

4.1.

 

 

 

 

Table 4.1. Relative reducing power of various sugars.

Type of sugar Theoretical Observed
Monosaccharide 100.0 100.0
Disaccharide 52.6 58.0
Trisaccharide 35.7 39.5
Tetrasaccharide 27.0 29.8
Pentasaccharide 21.7 24.2
Hexasaccharide 18.2 20.8

 

 

 

 

Source: Anon., 1984.

There are many factors which inﬂuence the activity of or —amylase. In extrusion,

effects of high temperature, pressure and shear on enzyme activity are of primary
concern, besides pH and the presence of Ca ions. For example, the optimum temperature
for a—amylase varies depending on the source. The optimum temperature for B.
licheniformis is 92°C (110°C maximum) whereas that for B. amyloliquefaciens is 70°C
(SS-90°C maximum).

 

42

Optimum pH for a -amylase also varies with enzyme sources. The optimum pH
ranges are 6.0-7 .0 for mamalian enzymes, 5.9-6.0 for B. subtilis, 4.8-5.8 for A. oryzae and
4.7-5 .4 for barley-malt. B. licheniformis or -amylase is active over a much wider pH
range than B. amyloliquefaciens (Norman, 1981).

a-amylase requires calcium ion to stabilize its activity. The enzyme from B.
licheniformis at 70°C requires about 3-4 ppm (Norman, 1981). The saturation level of
calcium for this enzyme is about 150 ppm. In a starch slurry, Termamyl (rt-amylase is
stabilized in the presence of 50-70 ppm of Ca++ (Anon., 1985a). Table 4.2 shows the
stability of Termamyl 60L and 120L at different Ca“ levels, pH and temperatures.

 

 

 

 

 

 

 

 

 

 

Table 4.2. Stability of Termamyl 60L and 120L in terms of enzyme half life in
minutes for DE values in the range of 0-12.

93°C 98°C 103°C 107°C
Ca++ 70 ppm
pH 6.5 1500 400 100 40
pH 6.0 600 200 75 20
pH 5.5 300 75 25 10
Ca4+ 20 ppm
pH 6.5 450 125 40 10
pH 6.0 250 75 20 5
pH 5.5 100 25 5 2
Ca” 5 ppm
pH 6.5 150 40 10 4
pH 6.0 75 20 5 2

 

 

Source: Anon., 1985a.

The objectives of this study were to determine the kinetics of a -amylase and to

develop a kinetic model to predict the extent of starch conversion in an extruder used as a

bioreactor.

43

4.3 Theory

Using electron microscopy, particle size analysis, DSC, and x-ray diffractometry,
Colonna et al. (1988) found that Bacillus subtilis a-amylase attacked starch from one
granule to another, while acid hydrolysis occurs at all granules simultaneously. The

enzyme attacks amylose and amylopectin equally. The mechanism of starch hydrolysis is

a two-stage process:
Sam
’9
S+E""’E—s '9 E+LS (4.1)
<——— ——->
k—r
3mm
1‘3
—_ k
LS+E TE-Ls ‘ E+SS (4.2)
<—-- -—-—>
k-a

The reaction was assumed to be quasi-steady-state, leading to the expression:

 

dS dSS k,ES.S,
-— = — = lqES = (4.3)
dt dt 534B,

This rate expression is applicable to short term hydrolysis only, because of product

inhibition during long term processes.
The kinetics of a -amylase from different sources have been studied by several
researchers (I-Iakkarainen and Linko, 1985; Henriksnas and Lovgren, 1978; Marc and

Engasser, 1983; Pasari etal., 1988; Reinikainen et al., 1986; Rollings and Thompson,
1984; Steverson et al., 1984; and Yankov et al., 1986). The results of their studies are

 

 

 

44

summarized in Table 4.3. The reaction kinetics of a —amylase depends on the source of
the enzyme and the chain length of the starch. Enzymes from different sources have
different optimum reaction conditions (temperature, pH and cofactor levels).

Using gel chromatography, Henriksnas and L6vgren (1978) found that dextrins
were formed within four ranges of molecular weights (MW): <5x10‘, 5x10‘-10’,
105-5x10s and >5x105. Most high molecular weight products were obtained at short
reaction times. The authors used a lumped ﬁrst order reaction for the dextrinization. The
lumped ﬁrst order model provides good simulation of the experimental results but does
not intrinsically describe the enzyme kinetics. Marc and Engasser (1983), and Rollings
and Thompson (1984) developed the kinetics of (Jr—amylase on the basis of resultant
product proﬁles. Incorporating the effects of starch dis solubility and thermal deactivation
of enzymes, Marc and Engasser (1983) showed that the theoretical results agreed with
experimental data.

Using an aqueous size exclusion chromatographic (SEC) method, Rollings and
Thompson (1984) found that the reaction products consisted of four MW groups:
1.5x103, 2x10“, 5x105 and 10'. Their results were consistent with those reported by
Henriksnas and Lovgren (1978). Rollings and Thompson (1984) derived the rate of
formation for each constituent based on the molecular weight of the possible
transformation between groups using a pseudo kinetic constant, and enzyme and
constituent concentrations. The results showed good agreement between predictions and
experimental data because the SEC analysis is more sensitive than reducing sugar

measurement.

 

45

 

Table 4.3. Kinetic models from the literature.

 

 

 

 

 

 

 

 

 

 

Source of Experimental Kinetic model Reference
(1 —amylase conditions
Schwannio 1% starch at 32°C, Michaelis-Menten with Pasari et
-myces pH 5.0 product inhibition: al. (1988)
c I II" = 55] = 7%“
as e u v a, m“ ‘ +%1
Saccharo- 0.1-0.9% starch at Michaelis-Menten with Steverson
mycopsis 32°C, pH 4.8 product inhibition. et al.
ﬁbuliger (1984)
Malt malt starch at First order reaction for each Marc and
0t -amylase 40-70°C product component. First Engasser
order enzyme deactivation. (1983)
Termamyl 60L 35-55% wheat N on-linear model: Reinikainen
starch at 80°C, in a r = f(C,,C,,,C,) et al. (1986)
batch system
Termamyl 60L 30-50% wheat Irreversible ﬁrst order Hakkarainen
starch at 90- 120°C reaction. & Linko
in a batch system (1985)
Thermo- 1% corn starch at if- : f(k, [5.], arm) Rollings &
stable 60-65°C, pH 6.0, Thompson
Termamyl batch process (1984)
Bacillus 1% wheat ﬂour Consecutive ﬁrst Henriksnas &
subtilis order reaction vagren
(1978)
Bacillus 6.6-40% starch at Michaelis-Menten Yankov et
licheni- 100°C and pH 7.0 al. (1986)

formis

 

 

 

 

 

 

46

4.4 Materials and Methods

Pre-gelatinized corn starch (23% amylose and 77% amylopectin) (American
Maize, Indiana) was used to study the kinetics of Termamyl 120L and 240L (rt—amylase
(N ovo Industries, Connecticut). The reaction kinetics of the enzyme were investigated by
an initial rate approach at dilute starch concentrations and by a continuous rate at high
starch concentrations. The activity of Termamyl 120L (30,000 units/ml) and 240L
(80,000 units/ml) were determined by the methods described by Saba et al. (1987).

4.4.1 Initial rate study at low starch concentrations
The initial rate of starch hydrolysis by Termamyl 120 L or —amylase was

determined in a test tube used as a batch reactor. Pre-gelatinized corn starch at
concentrations of 1%, 2%, 4% and 8% (w/w) in 50 mM acetate buffer at a pH of 6.0 were
prepared and placed in test tubes. An enzyme solution containing 18 units of Termamyl
120L per ml was prepared using the same buffer. A reaction mixture containing 1 g
starch solution and 0.1 ml of the enzyme solution was incubated in an ethylene glycol
bath at 100°C for 5 min. The reaction was stopped by adding 0.3 ml of 0.1 M NaOH and
cooled to room temperature. The sample was diluted to 10 g using distilled water and
centrifuged at 15,000 rpm for 10 min. The supernatant was analyzed for total reducing
sugar, using the DNS method (Saha et al., 1987).

4.4.2 Continuous rate study at high starch concentrations

An MPF-SOD Baker-Perkins twin screw food extruder was used to carry out starch
liquefaction by or -amylase. Two replicate experiments were conducted at 60% (w/w)
moisture, 95°C and a pH of 6.0, using 45 units (activity/g starch) of enzymes. Termamyl
240L or—amylase was diluted in a 30 mM acetate buffer solution (pH 6.0) before mixing

with starch in the extruder. After the system reached steady state, the residence time was

 

47

measured using a tracer method. About 0.3-0.5 g of a dye ball containing starch and an
erythrosine solution in water was dropped through the feed Opening. Extrudate collection
began when a slight change in color was observed, and continued at intervals of 158 until
the dye was no longer visible. The intensity of dye in each sample was measured against
a base containing no dye using a CR-200 Chroma Meter (Minolta Camera, Japan). The
mean residence time was calculated to be about 10 min by the ﬁrst moment formulation.

After RTD measurement, six extrudates were collected in liquid nitrogen to freeze
the reaction, and then placed in a freezer at -17°C for subsequent sugar and starch
analyses. To obtain the complete reaction proﬁle, about 400 g of extrudate was collected
in a PE-laminated ﬁbrous sausage casing (Union Carbide, OH) 5.08 cm in diameter. The
casing was sealed and immediately immersed in a circulating water bath which was
maintained at 95°C to continue the reaction. Periodically, the casing was opened and
stirred before approximately 610 g of liqueﬁed product was withdrawn and dipped in
liquid nitrogen. This was continued for 25-30 hours until the reaction reached
equilibrium.

Reducing sugars were analyzed using DNS and DE methods. The DNS procedure
used was developed by Bernfeld (1955) and modiﬁed by Saha et a1. (1987). The DE
method used was developed by Dygert et al. (1965). The amount of unreacted starch was
analyzed by the UV method as described by Boehringer (Anon. 1987).

4.5 Results
4.5.1 Initial rate study at low starch concentrations

Results from the initial rate experiments showed that the Michaelis-Menten kinetic
model could be used for data analysis. Using the Lineweaver-Burk plot (Figure 4.1) and
regression analysis (R2 of 0.986 at P<.05), Vm (1.1334 mg/g of extrudate/min) and K1,,

48

(17.34 mg/g of extrudate) were calculated.

4.5.2 Continuous rate study at high starch concentration

Tables 4.4 and 4.5 show experimental data obtained from two extrusion runs and

the fractional conversion (X) at each residence time, calculated by:

_ [S]

X =1[Sl.

 

(4.4)

Results at reaction times less than 10 minutes were obtained from separate
experiments (discussed in Chapter 7) which were conducted under identical conditions of
moisture content, temperature, pH and enzyme level.

Results from the DNS and DE methods were different in magnitude, although both
increased with increasing reaction times. Between %RS and %DE, %DE values were
more consistent with respect to fractional conversion. In fact, the DE method is more
accurate since it can be used to measure the amount of reducing sugars in a very narrow
range of 0-150 micrograms. Since the results from the two experiments were fairly
consistent (Figure 4.2), they were pooled together and analyzed to formulate a rate

expression.

 

.cBEbCoocoo 50.6% BEE 609.302

mnmco> cozoctou. Enoch Co 30.. 608900.. .6 #03 xcnmlce>oozocj :V 829...

Tam“—

NTO 09.0 00.0 00.0 .v0.0 No.0 00.0 No.0] t.400! 00.0!
rthLL-n—nthFbLLbLL-lrLPb unngbh-.—thl—

 

49

‘0.
,.J

O.
N

O

L”.
(\l

 

[IIII‘ITIIIIIIIIITIITIIIIIIIIII

50

 

Table 4.4. Percent of unreacted starch, reducing sugars, DE, and fractional
conversion (first replicate of continuous rate study).

 

 

 

 

 

 

 

Residence time % Unreacted % Total %DE Fractional
(min) starch reducing conversion
sugar (RS)

0.0 38.8 0.0 0.0 0.00
2.2 34.3 8.5 3.3 0.12
4.7 34.4 10.0 3.4 0.11
7.5 30.8 12.3 4.7 0.21
10.7 27.6 15.1 6.3 0.29
43.9 9.3 32.4 15.6 0.76
48.9 9.1 33.0 17.1 0.77
61.9 8.1 36.8 19.5 0.79
65.9 7.1 36.9 19.5 0.82
73.9 5.9 41.5 22.0 0.85
78.9 5.2 39.8 21.6 0.87
86.9 5.1 41.0 23.0 0.87
102.9 4.4 44.0 24.9 0.89
116.9 4.1 44.3 26.3 0.90
132.9 3.2 46.4 28.0 0.92
161.9 3.3 48.9 30.7 0.92
190.9 2.8 50.7 32.0 0.93
1500.0 2.3 53.1 41.8 0.94

 

 

51

 

Table 4.5. Percent of unreacted starch, reducing sugars, DE, and fractional

conversion (second replicate of continuous rate study).

 

 

 

 

 

 

 

Residence time % Unreacted % Total % DE Fractional
(min) starch reducing conversion
sugar (RS)

0.0 42.8 0.0 0.0 0.00
10.4 31.4 20.4 7.1 0.27
20.4 24.3 23.4 9.3 0.43
27.9 19.5 25.0 9.1 0.54
31.4 17.4 25.1 10.2 0.59
37.9 13.2 28.9 12.5 0.69
41.4 11.7 28.3 12.4 0.73
48.9 7.5 32.5 13.3 0.83
57.9 5.8 34.1 15.4 0.87
68.4 5.0 36.6 17.7 0.88
77.4 4.9 38.0 18.9 0.89
82.9 4.7 38.4 20.6 0.89
91.9 4.4 38.8 20.6 0.90

110.9 4.1 41.8 21.5 0.90
141.9 3.6 45.3 26.0 0.92
171.9 3.8 47.3 28.0 0.91
299.9 3.5 54.3 32.6 0.92
401.9 3.2 54.9 35.4 0.93
586.9 3.0 54.9 37.4 0.93
1201.9 3.3 57.1 39.7 0.92
1333.4 2.8 63.4 41.0 0.94
1880.9 2.7 64.0 42.0 0.94

 

 

52

.oE: 026068 mzmco> CO co_muo>coo 650500.; New 830E

EE .oEﬁ oocoEmom

000— 00F 0— F
— L..PP#.— _ ruPhbb L P —_~_hr_ L h 0.0

N EoEtoaxm l:
_ EoEtoaxm ll

.. to
Ad
Ad.
Lao
70.0
190
Inc

10.0

 

10.0

 

10.—

sseluoisuewlp ‘uogsreAuoo 10110110on

 

53

Since the SEC method was used in this study, the liqueﬁed product proﬁle could
not be determined Therefore, the rigorous approach which uses the probability of
producing products by starch hydrolysis could not be used to derive the intrinsic kinetics
for Termamyl (Jr-amylase. However, if the rate expression is used as a tool to predict the
extent of starch hydrolysis in the extruder at a given condition, then the kinetics can be
empirically developed. I

A plot of starch concentration and DE values versus residence time shows that a
rapid reduction in starch concentration and a rapid increase in DE values occurred within
the ﬁrst 2 hours, at similar but inverse rates (Figure 4.3). However, the DE values
increased at a higher rate than the reduction of starch concentration after 2 hours. This is
due to the action of the enzymes which randomly attack starch molecules rapidly in the
early stages to produce oligosaccharides (DP>3). As the reaction continues, the enzymes
attack the oligosaccharides to produce shorter chain molecules which have higher
reducing power. When the reaction is complete (X = 1), only glucose is obtained.
However, this condition never occurs because starch molecules contain branched chains
which or -amylase cannot break down. When the reaction reaches equilibrium, the
products are a mixture of low DP molecules and dextrins. From the experimental results,
the equilibrium DE was estimated at 42, which is consistent with what has been reported

in the literature.

.H- ‘_._..A._.

 

54

.GES COCOUECL WDmL®>

2m: cozobcoocoo 588% 95 $9 E20228 mmobxoo me. 959;

SE US: oocoEmmm

 

 

 

 

OOON 000—. OONF 00w
— u n h — h a P )— ~ P h h n
D D a
t a
]
S
[
O
J
G
3
O IIIIII .111
0|. lllll lolllrplrlrllltll

engimoag n mg .i
Ewieemoivaxmmxm n E l

 

 

55

Empirical relationships for starch concentration and the DE value as a function of

reaction time were developed using regression analyses (R2 = 0.99). They are,

respectively:
[S] = 37.5 exp(—0.034t)+2.81 (4.5)
l
DE = (4.6)

 

(2.02 + 0.023t)

By taking the derivative of Eq. 4.6, the rate of starch consumption was calculated
and plotted against starch concentration. The rate decreased with decreased starch
concentration; however, this behavior did not follow the Michaelis-Menten model.
Instead, the proﬁle approached a ﬁrst order reaction for most starch concentrations except
those lower than 3%, where the rate was close to zero. Therefore, the rate of
consumption was plotted against a modiﬁed starch concentration (Figure 4.4), [S]’,
deﬁned by:

[S]’ = [S] -0.06[S], (4.7)

where 0.06 is the fractional conversion of unreacted starch at equilibrium. This deﬁnition
allowed the plot to be extended through the origin.

56

.cowobcoocoo 50.6% ooEooE mzmco> cozaEzmcoo £83m *0 30m :4 050E

$84.3

Que OM O_N _

iP/[S]P-

 

A.Hmmooimmvnno. u :53? .l. .

 

57

An empirical ﬁrst order model was developed using regression analysis (R2 = 0.99)
and can be expressed by the equation:

_ gm
r‘ — dt

0.033[S]’ (4-8)
The rate constant was 0.033 g/g/min. The rate of product formation in terms of DE

values was calculated in the same manner and is plotted against the modiﬁed starch

concentration (Figure 4.5). Using regression analysis (R2 = 0.99), the rate of product

formation was developed and may be expressed by:

,p = “3‘51 = 0.191exp(0.026[S]’)-0.206exp(—0.415[S]’) (4.9)

 

Since Figures 4.4 and 4.5 could not be superimposed, the rate of product formation
was not equal to the rate of starch consumption. Figure 4.6 shows a plot between the two
rates. The relationship between the two was developed using regression analysis (R2 =
0.99) and may be expressed by:

r = 0.425(-r,)°“" (4-10)

P
On the whole, the rate of product formation is proportional to the square root of the

rate of starch consumption.

58

.cozobcoocoo

£83m ooEooE mzmco> cozoEcoa 00460.5 *0 30m man 839...

.3

0% Oh ON or O
_ c _ . _ _ 0.0

 

 

23m_4.10388...pageaxm:3. u aims? l
10.0

3.9/[30k

59

.cochEzmcoo 588% *0 BE mamco> Conestoa 3,260.5 U6 Bow. 0% Boom

“REP.

0.0 #0 N0 0.0
_ _ u t _ L 0.0

Lvé N; 0.? 0.0
T L

 

 

Eartogmmmenvtmmx n e305?) l.

tp/[30]p

4.6 Discussion

Within the range of low starch concentrations (up to 8%), the rate expression for
starch hydrolysis by thermostable Termamyl or —amylase can be represented by the
classical Michaelis-Menten model. These results are in agreement with those reported by
Yankov et a1. (1986), who found that the model was valid for up to about 25% starch
concentration. In this study, viscosity increased with increased pre-gelatinized starch
concentration, therefore experiments at the high concentrations were difﬁcult to conduct.

At high starch concentration, the reaction has been known to become substrate
inhibited (Yankov et al., 1986). This may be due to network formation by starch
molecules which inhibited attack by enzymes. Macroscopically, a system containing
soluble enzyme in a starch-water mixture is homogeneous, but microscopically it may be
considered as segregated where enzyme molecules are dispersed in the starch-water
matrix. At low starch concentrations, starch and enzyme molecules move freely to
contact each other, therefore reaction takes place. On the other hand, at high starch
concentrations, the molecules cannot move freely due to viscous effects, therefore the
rate of reaction decreases.

This viscous effect is analogous to the effect of mass transfer, which may be
described in terms of the mass transfer coefﬁcient. In suspensions of small particles, the

mass transfer coefﬁcient can be determined (Geankoplis, 1983) by:

 

 

2D 12 0.31 (pa - Pp)llc8 0.333
= . l
where
Sc = u“ (4.12)

9:0 12

61

9.96x10'16T
D12 = W (4.13)

It is obvious that increasing viscosity results in decreasing kL and, therefore, the
rate of mass transfer. However, the rate of mass transfer may be improved if convection
compensates for the effect of limited mass diffusion.

The rate of reaction in this study, however, did not decrease at high starch
concentrations (Figure 4.4). Therefore, it appears that substrate inhibition was negligible,
and that the results can be used to represent the real reaction rates.

The data in Table 4.2 suggests that thermal inactivation of the enzyme should be
considered. However, one must note that data in Table 4.2 are based on DE values of
0-12. For example, the half-life of Termamyl 60 L and 120 L is 1500 min at 93°C, CaH
of 70 ppm and pH 6.5. By contrast, DE values of 40 were obtained in this study after a
reaction time of only 1500 min. This is probably because the enzyme levels used in this
study are much greater than the levels used to obtain the results in Table 4.2. Therefore,
it is hypothesized that the rate of reaction was so fast that the percent of un-reacted starch

decreased rapidly before thermal inactivation could become a factor.

4.7 Conclusions

The reaction kinetics of thermostable or-amylase was determined by two

experimental methods. The ﬁrst, an initial rate experiment, was carried out at low
concentrations of pre-gelatinized starch. The second, a continuous rate experiment, was
conducted at a starch concentration of 40% in a twin screw extruder. To obtain a
complete product proﬁle, the extrudate was collected and put in a batch system to allow

the enzymatic reaction to continue to equilibrium.

62

Because the two methods were different, it was difﬁcult to compare one to the
other. At low starch concentrations, the Michaelis-Menten model is applicable. But at
high starch concentrations, an empirical ﬁrst order model was more suitable for
describing the reaction proﬁle for starch hydrolysis by a-amylase. Substrate and
product inhibition, which are mechanistically analogous to mass transfer limitation, have
been observed and reported in the technical literature. However, in this study no effects
of product inhibition were observed.

At high starch concentrations, the relationship between the rate of product
formation (rp) in terms of DE values and the rate of starch consumption (-r,) was not
linear. Instead, r, was found to be proportional to the square root of -r,. It must be noted,
however, that DE values are indices used only to indicate the extent of reaction and
cannot be used to describe the intrinsic reaction kinetics.

Initial rate studies at various high starch concentrations in the twin screw extruder
needs to be further investigated. Because the extruder provides excellent mixing and
resolves the mass transfer limitation problem, a more generalized rate expression can be
derived by this method. The initial rate studies also eliminate the effect of thermal
inactivation of the enzyme, which can occur during a continuous rate study. The reaction

kinetics should be studied at various temperatures, to account for temperature variation

during extrusion.

4.8 Nomenclature

A High MW concentration, g/l

Ai Initial activity at time 0, activity unit
A, Residual activity at time t, activity unit

B Dextrin concentration, g/l

999.00

1‘1.

k1 and k3

k, k.,, k, and k4
LS

P

m

Sc

3. [S]

S... [S].

SS

63

Low MW concentration, g/l

Normalized starch fraction, dimensionless

Actual water fraction in extrudate, dimensionless
Enzyme concentration, ml of enzyme per 100 g dry starch
Diameter of small solid particle, m

Mass diffusivity of solute 1 in solution 2, m2/s

or —amylase

Total enzyme concentration, g/l

Enzyme-substrate intermediates, g/l

gravitational acceleration, m/s2

Michaelis-Menten constant, mg/ml

Inhibitor constant, mg/ml

Mass transfer coefﬁcient for liquid suspension, m/s
Rate constants, g/l/min

Rate constants, min'1

Large oligosaccharides

Product

Mass fraction of each constituent (Table 4.2), dimensionless
Schmidt number, dimensionless

Substrate or starch concentration, g/g

Original substrate concentration, g/g

Small oligosaccharides

Temperature, K

Reaction time, min

t"2 Enzyme half life, min"z
Vm Maximum enzymatic rate, mg/g/min
VI Molar volume of liquid, m3/kg mol
Greek symbols
or Molar ratio of species i to j, dimensionless
tr, it, Viscosity of solution or continuous phase, kg/m/s
p, Density of the continuous phase, kg/m3
p, Density of solid particles, kg/m3
v Reaction rate, mg/ml/min

4.9 References

Anon. 1984. Nutritive sweetness from corn. Corn Reﬁncrs Association. Washington
DC.

Anon. 1985a. Termamyl. Novo brochure B 204, Novo Industries Inc. Denmark.

Anon. 1985b. Use of Termamyl for starch liquefaction, Novo brochure, Novo Industries
Inc. Denmark.

Anon. 1987. mmwmammmm
W Boehringer Mannheim GmbH Biochemica. W. Germany.

Bemfeld, P. 1955. Amylases 0t and B. Methods Enzymol. 1:149.

Colonna, P., Buleon, A. and Lemarie, F. 1988. Action of Bacillus subtilis or—amylase on

native wheat starch. Biotech. Bioeng. 31:895.
Dygert, S., Li, L.H., Florida, D. and Thoma, J. 1965. Determination of reducing sugar
with improved precision. Anal. Chem. 13:367.

65

GeankOPIiS. CI. 1983. Transport Exams and Unit meticns 2nd ed Allyn and

Bacon, Boston.

Greenwood, CT. and Milne, EA. 1968. III Theories of the Stepwise Degradation of
Linear Polysaccharides In: Adxanees in Carbehydrate Chemistry. Wolfrom, ML.
and Tipson, R.S. (Editors). Academic Press, New York. 23:299.

Hakkarainen, L. and Linko, P. 1985. State vector model for contherm scraped surface
heat exchanger used as an enzyme reactor in wheat starch conversion. J. Food Eng.
4:135.

Henriksnals, H. And Ldvgren, T. 1978. Chain-length distribution of starch hydrolyzed
with a- and B -amylase action. Biotech. Bioeng. 20:1303.

Marc, A. and Engasser, J.M. 1983. A kinetic model of starch hydrolysis by a- and

B —amylases during mashing. Biotech. Bioeng. 25:481.

Norman, BE. 1981. New developments in starch syrup technology. In: Enzymes and
EM Messing. Birch, G.G., Blakebrough, N. and Parker, K.J. (Editors) Applied
Sci. Pub1., London.

Pasari, A.B., Korus, RA. and Heimsch, RC. 1988. Kinetics of the amylase system of
Schwanniomyces castellii. Enzyme Micro. Technol. 10:156.

Richardson, T. and Hyslop, DB. 1985. Enzymes. In: Emd Chemistry. Fennema, O.R.
(Editor). 2nd ed., Marcel Dekker Inc., New York.

Reinikainen, P., Suortti, T., Olkku, J., Malkki, Y. and Linko, P. 1986. Extrusion cooking
in enzymatic liquefaction of wheat starch. Starch 38:20.

Rollings, J .E. and Thompson, R.W. 1984. Kinetics of enzymatic starch liquefaction:
simulation of the high-molecular weight product distribution. Biotech. Bioeng.
26:1475.

66

Saha, B.C., Shen, G.J. and Zeikus, J .G. 1987. Behavior of a novel thermostable
B-amylase on raw starch. Enzyme Microb. Technol. 9:598.

Steverson, E.M., Korus, R.A., Admassu, W. and Heimsch, RC. 1984. Kinetics of the
amylase system of Saccharomycopsis ﬁbuliger. Enzyme Micro. Technol. 6:549.

Yankov, D., Dobreva, E., Beschkov, V. and Emanuilova, E. 1986. Study of optimum
conditions and kinetics of starch hydrolysis by means of thermostable (rt—amylase.

Enzyme Microb. Technol. 8:665.

67

5 EFFECT OF SHEAR ON ENZYMATIC ACTIVITY DURING
REACTIVE EXTRUSION
5.1 Abstract

The effect of shear on the activity of Termamyl 120L or —amylase was studied in a

batch reactor and in a twin screw extruder. In the batch experiment conducted at room
temperature, strain histories above 4x104 produced a signiﬁcant reduction in enzymatic
activity. On the other hand, strain history by itself did not appear to cause signiﬁcant loss
of activity in extrusion runs conducted at ﬂow rates of 10-40 kg/hr and screw speeds of
60400 rpm. Statistically, the combination of shear rate, strain history and speciﬁc
energy consumption (SEC) was signiﬁcant in the reduction of enzymatic activity in the

extruder.

5.2 Introduction and Background

The effect of shear on enzyme activity has been studied by several researchers
(Charm and Wong, 1970; Charm and Lai, 1971; Tirrell and Middleman, 1975; and
Thomas and Dunnill, 1979). Charm and Wong (1970) examined the loss of activity of
catalase, rennet and carboxypeptidase in solutions at 4°C, using a coaxial viscometer and
a capillary rheometer, and found similar results in both systems. The effect of shear was
reported over a strain history (integral of the product of shear rate and exposure time)
range of 5x103-16x10°. Enzyme activity decreased signiﬁcantly with increasing strain
histories beyond 10‘. The shear-induced deactivation was postulated to be caused by the

breaking up of tertiary structure due to molecular reorientation in the shear ﬁeld.

68

A similar study was carried out by Tirrell and Middleman (1975). They used a
coaxial cylinder viscometer to examine the shearing effect on urease during urea
hydrolysis at a pH of 6.75 and a controlled temperature of 23°C. They found that there
was a continuous decrease in urea hydrolysis due to loss of enzyme activity with
increased strain histories beyond 10‘, in agreement with the results of Charm and Wong
(1970). Both irreversible and reversible enzyme deactivation occurred under shear.
Tirrell and Middleman (1975) postulated that the conformation of protein caused
dislocation of active sites which led to loss of enzyme activity. As the shearing force was
removed, some enzyme activity was recovered as a result of molecular relaxation of the
enzyme or the reformation of the broken tertiary structure (Charm and Lai, 1971). The
shear effect on enzyme activity depends on the size and shape of the enzyme (Tirrell and
Middleman, 1975).

The effect of shear on activity also depends on the source of the enzyme and the
equipment used. Thomas and Dunnill (1979) found that the effect of shear on the loss of
catalase activity in a closed coaxial concentric viscometer at 20-60°C was less than what
would be expected based on the results of Charm and Wong (1970). The difference was
due to the source of the enzyme and the system used. Charm and Wong (1970) used an
open system. Above 50°C, shearing reduces thermal deactivation of catalase by
interfering with protein precipitation. Thomas and Dunnill (1979) proposed that activity
can be lost by other means during processing, such as pressure and oxidation.

A search of the technical literature did not yield any information on the effect of
shear on or —amy1ase activity. The loss of enzymatic activity can be signiﬁcant, however,
especially when the enzymes are exposed to the high shear rates encountered during
extrusion. The objective of this study was to investigate the inﬂuence of high shear on

the activity of a-amylase during reactive extrusion.

69

5.3 Theory

Because material is sheared in four regions inside the extruder (Martelli, 1983), an
accurate estimation of shear rate is difﬁcult to obtain, especially with non-Newtonian
ﬂuids. However, since twin screw extruders are good mixers, the average shear rate can
be estimated using an approach analogous to that used with mixers: the average shear rate
of a non-Newtonian ﬂuid is estimated on the basis of power consumption during mixing,
using a relationship between the power number (Po) and the Reynolds number (Re)
developed in the same equipment for a Newtonian ﬂuid. This method was developed by
Metzner and Otto (1957), and used successfully by Rao and Cooley (1984), and
Mohamed (1988). The approach is valid when both ﬂuids are operated in an identical
mixing system with identical impeller speeds in the laminar region.

The power number for Newtonian ﬂuids in a twin screw extruder is (Mohamed,
1988):

E
P = —'— 5.1
0 pN3D: ( )
where
E. = Pw-APQ. (5.2)

and P, is computed from the manufacturer’s relationship:

Pw = 0.354(%Torque)N (5.3)
The Reynolds number for Newtonian ﬂuids is:

0’1)!
Re = “u p (5.4)

where the hydraulic diameter (D,) is deﬁned as:

 

7O

D,I = —— (5.5)

The hydraulic diameter is generally used to characterize complex geometries in
ﬂuid mechanics. In the extruder, it can be reliably determined only for screws in
fully-ﬁlled zones. The hydraulic diameters for different screw types have been calculated
by Mohamed (1988): 0.75 cm for single lead screws, 1.01 cm for feed screws and 1.22
cm for 30° forwarding paddles. Using the same approach, Dh for any mixed screws can

be calculated by:

42Vwi

D, = 2A”. (5.6)

 

The power number versus Reynolds number for a Newtonian ﬂuid is used to
determine the corresponding Reynolds number (Re') for the non-Newtonian ﬂuid during
extrusion at the calculated power number. The apparent viscosity (1],) for the

non-Newtonian ﬂuid can then be calculated by:

DZN
n. = ”R2. (5.7)

 

If the ﬂuid is described by a power law model, its average shear rate (4),) can be

obtained from the viscosity calculated in Eq. 5.7 by:

l

11.:

'Y. = [k,e(%)] (5.8)

71

5.4'Materials and Methods

The effect of shear forces on the activity of thermostable Termamyl 120L
or -amylase (Novo Laboratories, Connecticut) was investigated in a Haake viscometer
and in a Baker-Perkins MPF 50D twin screw extruder. Enzymatic activity, determined
by the DNS method developed by Bemfeld (1955) and modiﬁed by Saha et al. (1987),
was 30,000 units/ml.
5.4.1 Haake viscometer experiments

The enzyme solution, containing three activity units per ml, was prepared in
distilled water prior to testing. Batch experiments were carried out in a Haake RV-12
viscometer consisting of an MV paddle sensor attached to an M-500 viscometer drive
head. Fifty grams of the diluted enzyme solution (pH 5 .6) was loaded into an MV cup.
The F3-C water circulator maintained the enzyme solution at a constant temperature of
25°C. The viscometer behaves as a mixer, with a well deﬁned shear rate (Steffe and
Ford, 1985):

I: k’m (5.9)
where k’ is the paddle constant and 0) is the rotational speed of the paddle in rad/s. The

paddle constant of 4.46 was determined by Steffe and Ford (1985).

The experiment was conducted over four different reaction times at each of six
rotational speeds (Table 5.1). At the end of each reaction time, the paddle was stopped
for 15s, and 0.1 ml of the enzyme solution was pipetted to a 15 ml test tube. Duplicates
were prepared at each time step.

Residual enzyme activity was determined immediately by the following process.
One ml of 2% soluble starch was added to the tube and the mixture was immersed in a

water bath controlled at 60°C for 30 min. The reaction was stopped by adding two ml of

72

DNS reagent. The mixture was heated in an ethylene glycol bath maintained at 100°C for
10 min and then cooled to room temperature. The dark brown color developed was then
measured with a spectrophotometer at 640 run against a blank containing distilled water,
prepared in exactly the same way as the samples. The absorbance was converted to
glucose equivalence using the standard curve established from several known glucose
concentrations. One unit of enzyme activity is deﬁned as 1 micromole of glucose

liberated per ml of enzyme per minute.

 

 

 

 

Table 5.1. Experimental conditions used in the Haake viscometer.
RPM Residence time (min)
100 10, 20, 35 and 70
150 10, 20, 35 and 60
350 10, 20, 35 and 60
400 10, 20, 35 and 65
500 10, 35, 60 and 90
550 10, 20, 35 and 60

 

 

 

The depletion of the enzyme solution due to sampling over the second, third and
fourth steps was insigniﬁcant, and should not alter the shear rate in the ﬂow ﬁeld.
Therefore, the shear rate was assumed to be relatively constant at each rotational speed.
5.4.2 Extruder experiments

Soy polysaccharide (SPS) (Fibrim 2000, supplied by Ralston Purina Co., MO) was
used as a starch-free, inert material to carry on -amylase through the extruder. The SPS
was fed via a K-tron feeder and mixed with deionized water in the extruder. Termamyl
120L or -arnylase (Novo Industries, Connecticut) was diluted and fed through another
injection port. The enzyme constituted 0.0003 ml per g of the feed mixture. With a

screw proﬁle of 90° paddles, the mixture was too dry to be properly conveyed, and the

73

extrudate became cooked and discolored due to viscous heat dissipation. Although the
barrel was chilled, the rate of cooling could not keep up with the rate of heat generation.
Therefore, product temperatures in the paddle zone reached or exceeded 95°C, the
Optimum temperature of the enzyme. A higher SPS moisture content might have solved
the problem but this was not investigated. Instead, 30° forwarding paddles were used, to
reduce the rate of viscous dissipation. The Po-Re relationship for this conﬁguration had
previously been developed by Mohamed (1988).

During the experiment, the mass ﬂow rate was increased with increasing screw
rpm to maintain the die pressure within as narrow a range as possible. The extruder was
operated at ﬂow rates of 10-40 kg/hr and screw speeds of 60-400 rpm. Under these
conditions, the screws were partially ﬁlled and product temperatures remained below
55°C with chilled water circulation. Therefore, protein denaturation and thermal
deactivation of enzymes were avoided. After the extruder reached steady state, ﬁve
extrudates were collected for analysis on the same day.

RTD measurements were made for all extrusion runs. About 0.5 g of dye ball
(mixture of starch and erythrosine solution) was dropped through the feed port.
Extrudate samples were collected at the first sign of a color, and at intervals of 15s until
the color was no longer visible. The intensity of red (Ad) in the extrudates was measured
against a base extrudate containing no dye, using a CR-200 Chroma Meter (Minolta

Camera Co., Japan). The mean residence time was calculated by:

$0134.13:

I = (5.10)

 

74

where A0,- is the redness value at any discrete time t, and At is the time interval, equal to

15 s.

Use of SPS was advantageous since it does not contain any soluble starch and,
therefore, does not interfere with the determination of residual enzymatic activity. The
procedure used was as follows: One tenth of a gram of SPS extrudate was placed in a 15
ml test tube and mixed with one ml of 2% soluble starch in a 50 mM acetate buffer (pH
6.0). The mixture was incubated in a water bath maintained at 60°C for 30 min. One half
ml of 0.2 M sodium hydroxide was then added and the test tube was immediately
immersed in an ice bath for 10 min. An additional 5 ml of water was added to the test
tube which was then heated in an ethylene glycol bath at 100°C for 10 min to complete
enzyme deactivation and also to extract all soluble constituents including reducing
sugars. The mixture was brought up to 15 g by adding water and then centrifuged at
15 ,000 rpm for 10 min. The supernatant was subsequently measured for the amount of
glucose equivalence by the DN S method. The enzyme activity was then calculated.
5.4.3 Determination of the effect of shear in the TSE

An average shear rate for 30° forwarding paddles was estimated by the Power
number and Reynolds number relationship developed by Mohamed (1988):

P0 = 6.9x10‘Re'1-3‘ (5.11)

The rheological model for 30% SPS in water was obtained from Howkins (1987):

(1’2) as
1: = 2.85e " (5.12)
An alternative approach is to determine the shear effect by using the speciﬁc

energy consumption (SEC) or the ratio of mass ﬂow rate to screw rpm ('5). The SEC is

75

defined as the ratio of power input to the mass ﬂow rate, and is directly proportional to
the shear strain and shear stress during extrusion (Rauwendaal, 1981). In this study, the

SEC was expressed as:

SEC = T (5.13)
m

The ratio of mass ﬂow rate to screw speed is generally used as an indicator of the

degree of ﬁll in an extruder.

5.5 Results and Discussion
5.5.1 Shear deactivation in the batch system

The residual activity of or-amylase and the associated strain histories were

calculated (Appendix 1), and are summarized in Table 5.2 and plotted in Figure 5.1. The
effect of strain history on enzyme activity was signiﬁcant at P<.05 (Appendix 2). A
marked reduction in enzymatic activity was observed at strain histories above 4 x 10‘, in

agreement with what has been reported in the technical literature (Figure 5 .2).

76

 

Table 5.2. Residual enzymatic activity obtained from the Haake viscometer.

 

 

 

 

 

 

 

 

Screw Shear rate, s’l Residence time, Strain history x104 % Average
speed, rpm min residual activity
100 46.7 10 2.8 99.0
20 5.6 99.8
35 9.8 92.7
70 19.6 88.3
150 70.0 10 4.2 100.0
20 8.4 94.6
35 14.7 95.3
60 25.2 95.8
350 163.4 10 9.8 95.3
20 19.6 99.5
35 34.3 92.0
60 58.8 87.5
400 186.7 10 11.2 93.2
20 22.4 90.0
35 39.2 84.7
65 72.8 82.4
500 233.4 10 14.0 97.6
35 49.0 83.4
60 84.0 76.2
90 126.0 65.0
550 256.7 10 15.4 91.7
20 30.8 93.0
35 53.9 77.5
60 92.4 68.5

 

 

 

 

 

 

77

>6ng Eobm msmuo> omoSEolo E 33500 6:28.. “coocod

00+m0._.

p —-h~

—

u

.AcoyoEoom3 coxzt e E0: 8:630 303

been 585
mo+mos

— —bL~L

_

L

¢0+m0._.

0
a
10—

1

10m

j

to...
to...

100
100
a
10m

1

100

10

100

1.

 

100—

 

It. 8%:

Kit/moo ecu/(zue [onpisea 34

 

78

663m mid omoSEolo oco 3an 6:25 oco Ecocov omoozaoaxxoncoo

.3238 JEEP. .6“. @322 Eobm .m> 5.3000 6328.. “cooked N0 830E

bee; seem

 

 

00.7.: 00.1: +0.7”: 00+m:
“occcm o t r 0
$2300 0 H0—
owoozaoaxxontoo < ..
oweSEold x 10N
10m. %
1 8
to... m.
n a r W:
o a low mul
o x D
a 100 m.
. m w a a w”
. . . _. )9. ,m.
a me I In _ 4....
. u. a n .3 - om
a mo! 0 I o o I
< a 0 Di Dan! I! o o 00
a a a 2.... 9.... o o o o a
.. _. . . . to?

79

5.5.2 Shear deactivation in twin screw extruders

Extrusion data was used to calculate Po for 30% SP8 in water. Eq. 5.11 was then
used to calculate the corresponding Reynolds numbers (Table 5.3).

The Reynolds numbers and Eqs. 5.6, 5 .7 and 5 .8 were used to estimate the average
shear rates (Table 5.4). The corresponding mean residence time, strain history, SEC, and
ratio of mass ﬂow rate to screw rpm (’5) and % average residual enzyme activity are also

summarized in Table 5 .4.

Since SEC and; were found to be highly correlated (R2 = 0.97) in this study, either

parameter may be used in the analysis.

Using the enzyme activity data from extrusion runs (Appendix 3), analysis of
variance (Appendix 4) was performed and results showed that none of the three variables
(strain history, shear rate or SEC) individually signiﬁcantly affected the residual enzyme
activity, at the 0.05 probability level. Over the range of strain histories in this study, less
than a 20% loss in activity was observed. An important reason for this is that the Po-Re
relationship was developed for fully—ﬁlled screws, and may not be valid for the partially-
ﬁlled screw application in this study. If used for partially-ﬁlled zones, the method would

result in over-estimation of the shear rate, leading to strain histories higher than actual.

80

 

Table 5.3. Power and Reynolds numbers for 30% SPS in water.

 

 

 

Cond. RPM % Torque Pressure Avg. Mass Po Re
no. drop, kPa Temp ﬂow rate,
(“Q kg/hr
1 60 36.5 1758 27.1 10.0 3.0x107 0.06
2 80 36.5 1862 28.6 15.2 1.0x107 0.10
3 100 35.5 1896 31.0 14.7 1.0x107 0.13
4 150 33.5 1965 34.7 15.1 501110“ 0.22
5 200 29.5 1931 37.3 15.0 3.0x10‘ 0.36
6 300 32.5 1896 44.6 24.6 1.0x106 0.61
7 400 35.5 1965 51.1 38.7 8.011105 0.90

 

 

 

 

 

 

 

 

 

Table 5.4. Values of apparent viscosity, shear rate, mean residence time, strain
history, speciﬁc energy consumption, ratio of mass flow rate to screw rpm and %
residual enzyme activity.

 

 

 

 

 

 

 

 

 

Run 11. 7x 10'3 I 7x 10'7 SEC £- % Avg.

no. residual

activity
1 2.4 30.5 6.6 1.2 0.8 0.167 94.5
2 1.9 40.5 4.2 1.0 0.6 0.190 100.0
3 1.9 38.0 4.0 0.9 0.9 0.147 88.4
4 1.6 40.5 3.8 0.9 1.4 0.101 91.8
5 1.3 47.9 3.7 1.1 1.8 0.075 83.0
6 1.2 45.1 2.2 0.6 1.8 0.085 92.4
7 1.1 42.6 1.5 0.4 1.6 0.097 91.0

 

 

 

 

81

Strain is the product of the average shear rate and the mean residence time.
Mohamed (1988) found that the average shear rate in the extruder is a power function of
the screw rpm. As screw rpm increased, shear rate increased but residence time
decreased This indirect relationship may mask the effect of strain history on activity. In
this study, the estimated shear rates were in a narrow range of 3x10‘-5x10‘, while the
corresponding mean residence times were in a relatively broad range of 1.5-6.6 min.

Based on AN OVA results (Appendix 4), each variable was signiﬁcant at a low
probability level of 0.15. If strain history, shear rate and SEC were considered
simultaneously, however, results from analysis of variance (Appendix 5) showed that
they contributed a signiﬁcant effect to the reduction of enzyme activity at a probability
level of 0.05.

5.6 Conclusions

The effect of shear on the activity of a -amy1ase was determined in a mixer

viscometer and a twin screw extruder. Results from the mixer viscometer showed that
loss of enzymatic activity was pronounced at strain histories above 4x10‘, in agreement
with what has been reported for other enzymes in the literature. The loss of enzyme
activity in the extruder did not depend signiﬁcantly on strain history, shear rate or SEC
alone at a probability of P<0.05, although each of the three was signiﬁcant at P=.15 and
their combined effect was signiﬁcant at P<.05. Among these parameters, the SEC
seemed to be the best parameter to describe the effect of shear on enzyme activity
because it was quantitatively determined rather than estimated. In this study, only a 20%
loss of the original activity was observed. Additional studies of the three variables when

a higher loss of activity occurs will help clarify their respective effects.

82

It is worth noting that the loss of activity due to shear depends on the test system
used. In a batch system where the flow ﬁeld is well deﬁned, the shear rate can be
accurately determined. However, it is difﬁth to obtain it in a complex flow ﬁeld such
as occurs in an extruder. The rheology of the ﬂuid containing the enzyme is also critical,
since viscosity plays an important role in the degree of mixing. In a batch system,
complete mixing is obtained; however, depending on screw conﬁguration, this may not
be necessarily true in the extruder.

Enzymes in the SPS mixture formed a dispersed phase which can be protected
from exposure to high shear in the extruder. Therefore, the effect of shear may not be as
pronounced as in the batch system. This is in agreement with Thomas and Dunnill
(1979) who found that increasing viscosity enhanced the resistance of enzymes.

It is also possible that the enzyme may be undergoing reversible rather than
irreversible shear deactivation, making immediate assays on enzyme activity necessary.
This is not practical, however, when the recovery time for the enzyme is shorter than the
time required by the assay procedure. All these make the effect of shear on enzyme
activity in the extruder relatively hard to determine. Another factor that can influence
activity loss during shearing is calcium ion, which acts as a co-factor to protect the

enzyme from structural deformation.

5.7 Nomenclature

Aw, AM Wetted area for one screw type (i refers to each screw in mixed
screws), cm2

Dh Hydraulic diameter, cm

E, Viscous dissipation of mechanical energy, W

k’ Paddle constant for mixer viscometer, dimensionless

 

Po

Re, Re’

Vw, V",

St
11.

TI:

83

Consistency coefﬁcient. Pa.sll

Consistency coefﬁcient at a reference temperature, Pa.s
Mass flow rate, kg/s

Flow behavior index, dimensionless

Screw speed, rps

Power number, dimensionless
Gas constant, cal/mole/K

Reynolds number, dimensionless

Absolute temperature, K
Wetted volume for one screw type (1 refers to each screw in
mixed screws), cm3

Greek symbols

Redness value for a sample relative to a base color at discrete
time ti

Activation energy, cal/mole

Pressure drop, Pa

discrete time interval, 8

Non—Newtonian apparent viscosity, Pa. sn

Non-Newtonian viscosity index, Pa.sn

Strain history, dimensionless

Shear rate, 3‘1

Newtonian viscosity, Pa.s

84

co Rotational speed of paddle, rad/s
p Fluid density, kg/m3

1 Shear stress, Pa

5.8 References

Bemfeld, P. 1955. Amylases, a and B. Methods Enzymol. 1:149.

Charm, SE. and Wong, BL. 1970. Enzyme inactivation with shearing. Biotech. Bioeng.
12:1 103.

Charm, SE. and Lai, C]. 1971. Comparison of ultraﬁltration systems for concentration
of biologicals. Biotech. Bioeng. 13:185.

Hawkins, MD. 1987. A predictive model for pressure drop in food extruder dies. MS.
Thesis. Dept. Agr. Engr., Michigan State University.

Martelli, EC. 1983. M Sm W: A £33162 Liam. Van Nostrand
Reinhold Comparce, New York.

Metzner, AB. and Otto, RE. 1957. Agitation of non-Newtonian fluids. AIChE. J. 3(1):3.

Mohamed, 1.0. 1988. Modeling shear rate and heat transfer in a twin screw co—rotating
food extruder. Ph.D Dissertation. Dept. Agr. Engr., Michigan State University.

Rao, M.A. and Cooley, HI. 1984. Determination of effective shear rates in rotating
viscometers with complex geometry. J. Texture Studies 15:327.

Rauwendaal, CJ. 1981. Analysis and experimental evaluation of twin screw extruders.
Polym. Eng. Sci. 21(16):1092.

Saha, B.C., Shen, G.J. and Zeikus, LG. 1987. Behavior of a novel thermostable

B —amylase on raw starch. Enzyme Microb. Technol. 9:598.

85

Steffe, J .F. and Ford, E.W. 1985. Rheological techniques to evaluate the shelf-stability of
starch-thickened, strained apricots. J. Texture Studies 16:179.

Thomas, CR. and Dunnill, P. 1979. Action of shear on enzymes: studies with catalase
and urease. Biotech. Bioeng. 21:2279.

Tirrell, M. and Middleman, S. 1975. Shear modiﬁcation of enzyme kinetics. Biotech.
Bioeng. 17:299.

86

6 EFFECT OF MATERIAL RHEOLOGY ON THE DISPERSION OF
FLUIDS IN EXTRUDERS
6.1 Abstract

The effect of rheological properties on dispersion coefﬁcients in a twin screw
extruder was investigated, using corn syrup, 7% soy polysaccharide (SPS) in honey and
2.5% methocel in water. Material rheological behavior was characterized by steady shear
and dynamic experiments. It was shown that the extruder could be characterized as a
tube, with the hydraulic diameter providing a characteristic dimension. Using the
Taylor-Axis model, the effective dispersion coefﬁcient was found to be affected more by
convective diffusion than molecular diffusion, with the mean velocity exerting the
greatest effect on the dispersion of ﬂuids in the extruder. Due to equipment limitations,
no reliable relationships could be developed to link viscoelastic behavior, expressed in

terms of equilibrium creep compliance, to the degree of dispersion.

6.2 Introduction and Background

In a twin screw extruder, the extrudate is subjected to both shear and elongational
forces. Shear ﬂow occurs between the rotating screw and the stationary barrel, while
elongational ﬂow occurs in the inter-meshing region. At steady state, elongational flow
causes an exponential rate of separation of fluid elements, in comparison to the linear rate
caused by shear ﬂow (Bird et al., 1987). Both ﬂows increase the degree of mixing and

dispersion.

87

Mohamed (1988) has recently demonstrated that the average shear rate in a twin
screw extruder can be estimated by an empirical correlation between the power number
and the Reynolds number. Mohamed (1988) lumped the effects of shear and elongational
flows, describing them in terms of an average shear rate.

The degree of mixing and dispersion in the extruder is expected to be inﬂuenced by
the rheological behavior of the extruded materials. The rheology may be described in
terms of the linear viscoelastic behavior of materials, which is characterized by the stress
relaxation modulus, G(t), and the creep compliance, J (t). G(t) is the ratio of
time-dependent stress to strain at constant deformation, while I (t) is the ratio of

time-dependent strain to stress at constant stress. G(t) and J (t) are correlated by:

G(t) = (6.1)

_1_
J (t)

Instead of using transient experiments to characterize G(t) and J (t), a periodic
experiment is commonly used. In the periodic experiment, the ﬂuid undergoes small
amplitude sinusoidal oscillatory shear, and the in-phase and out-of-phase components on
the stress at steady state are measured as a function of frequency. The experiment at
frequency to is qualitatively equivalent to the transient experiment at time t = 1/co. The

oscillatory data are presented in terms of complex modulus (G') and complex viscosity

(11.) by the expressions (Ferry, 1980):

G‘ = G’+iG" (6.2)

11' = n’+in" (6.3)
where

, _ 91

n — m (6.4)

and

l.-

88

II _ _G__’
Tl - 0) (6.5)

The complex modulus is related to the complex creep compliance (1’) by:
J = — (6.6)

G’(co) is the storage modulus (stress in phase with the strain divided by the strain),

and measures the elastic energy stored. Its magnitude is related to the conformation and
rearrangement of the material during oscillatory deformation. At t = 1/0), G(t) is
approximately equivalent to G’(0)). G”((n) is the loss modulus (stress 90° out of phase
with the strain divided by the strain), and measures energy dissipated or lost.

Since the frequency is equal to the shear rate, the Cox-Merz rule (Bird et al. 1987)
is valid, and the complex viscosity (n‘) and the steady shear viscosity (11) are equivalent.
This rule produces accurate results, especially at low frequencies where the dynamic
viscositym’) is much larger than the imaginary part of the complex viscosity (11”).
Therefore, 11’ is approximately equal to n' and also equal to the zero shear viscosity (1],).
J’(co) is the storage creep compliance; its value approaches the equilibrium creep
compliance (J 3 ) at low frequency.

Viscoelastic properties depend on molecular weight, molecular weight distribution,
temperature, concentration and chemical structure, among other variables (Ferry, 1980).
Zero shear viscosity and equilibrium creep compliance are used to characterize the
terminal zone behavior of the material during extrusion (Ferry, 1980). In the terminal
zone where the material has been sheared for a long period of time, the viscoelastic

properties are pronounced.

89

The zero shear viscosity is a measure of the energy consumed in the ﬂow while the
equilibrium creep compliance is a measure of the energy stored during the ﬂow. The
zero shear viscosity is relatively temperature dependent, but the equilibrium creep
compliance is not. Both are expected to play an important role in dispersion during
extrusion. The zero shear viscosity and equilibrium creep compliance can be calculated

by (Graessley, 1984):

 

n. = 311’ = 5311' (6.7)
, 1 . G’or
J. = E3}. 0:,) (6.8)

6.2.1 Diffusion and Dispersion

Diffusion is a slow process, caused by a concentration gradient. On the other hand,
dispersion is a fast process, caused by micromolecular diffusion and convective ﬂow
(Cussler, 1984). Dispersion and diffusion are, however, described by similar
mathematical expressions. The dispersion coefﬁcient is not highly dependent on
molecular weight or structure but strongly dependent on position and ﬂow direction
(Cussler, 1984).

In laminar ﬂow, a one-dimensional dispersion model is commonly used. The
model is a modiﬁcation of the plug ﬂow model, which incorporates a dispersion
coefﬁcient (D,) to describe the overall effect of different phenomena in the system.
Taylor (1953) observed that the residence time distribution for laminar ﬂow through a
tube is inﬂuenced by the velocity proﬁle as well as molecular diffusion, as depicted in

Figure 6.1. At moderate ﬂow velocities, the dispersion of a tracer occurs by axial and

90

radial diffusion from the ﬂow front via a molecular diffusion mechanism (Taylor, 1953);
however, axial diffusion is small compared to axial convective ﬂow. Therefore,

dispersion leads to two-dimensional ﬂow, expressed by (Butt, 1980):

ac 82C 1236‘ ac
—at- — Dm[-a?+;-a—r-)-u(r)-a? (69)

where DIn is the molecular diffusion coefﬁcient for the tracer, and u(r) is the laminar ﬂow
velocity proﬁle.
For one-dimensional ﬂow, Eq. 6.9 may be simpliﬁed to (Butt, 1980):

icﬁ—Digﬂ-éﬂ 610
at — ‘32: “32 (°)

where Cm is a mean concentration, 3 is the average axial velocity and D, is an effective

dispersion coefﬁcient, deﬁned as:

 

0:18?

, 48D... (6.11)

where R is the radius of the tube.

The expression in Eq. 6.11 indicates that the dispersion coefﬁcient decreases with
increased radial molecular diffusion. If the radial molecular diffusion is large, material in
the center diffuses outward while the material near the tube walls diffuse inward,
inhibiting dispersion (Cussler, 1984).

Eq. 6.9 is valid only if (Taylor, 1954):

4L HR
_ _ O O 2
R » D... :9 6 9 (6 1 )

91

.Aommd .uuam. 3oHC em:_emH c“ co_memamwu on

co_m:amﬂu emﬁaumﬁoe ucm emamcmeu m>ﬂuum>cou Co :oﬁuanﬂeucou

 

 

 

«.6 magmaa

92

where g is a radial Peclet number. However, Ananthakrishnan et al. (1965) proposed

that the radial Peclet number should be larger than 50 for application of the
one-dimensional dispersion model. In practice, axial molecular diffusion is neglected
when the radial Peclet number is equal to or larger than 100 (Subramanian and Gill,
1976).

The dispersion coefﬁcient in Eq. 6.11 was modiﬁed by Aris (1956) to provide a
generalized expression valid for all velocity proﬁles in ﬂow vessels of any geometry,
regardless of the value of the radial Peclet number. In terms of the tube diameter (d), the
expression is:

(12172
D, - Dm+1920m

 

(6.13)

Ananthakrishnan et al. (1965) demonstrated that the Taylor-Aris dispersion model
applies to Newtonian ﬂuids at steady state only, or for values of the dimensionless time
parameter I (Dmt/Rz) ranging from 0.8 at Pe=500 to 20 at Pe=1. Later, unsteady-state
convection-diffusion in Newtonian ﬂuids at small values of t was analyzed by Gill
(1967), and Gill and Sankarasubramanian (1970, 1971). Gill and Sankarasubramanian
(1971), and Sankarasubramanian and Gill (1972, 1973) subsequently developed a
generalized dispersion model applicable to many ﬂow problems.

Booras and Krantz (1976) applied the generalized dispersion model of Gill and
co—workers to describe the effect of the velocity proﬁle on dispersion in power law ﬂuids.
They reported that at small dimensionless times, the generalized dispersion coefﬁcient for
all power law ﬂuids (0.25 < n < 2.0) were identical. As the dimensionless time increased,
the dispersion coefﬁcient increased at the same rate towards an asymptotic value. They

showed that the asymptotic dispersion coefﬁcient decreased with decreased ﬂow

93

behavior index (n), in agreement with the results of Subramanian and Gill (1976). The
result was due to the ﬂatter velocity proﬁles of pseudoplastic ﬂuids, which tended to
decrease axial dispersion.

The effect of molecular diffusion on dispersion of non-Newtonian ﬂuids is
signiﬁcant (Booras and Krantz, 1976). For large Peclet numbers, the Taylor-Aris model
can be used for non-Newtonian ﬂuids at dimensionless times above 0.60 (Booras and
Krantz, 1976).

The objective of this study was to determine the effect of rheological behavior on
the dispersion of ﬂuids during extrusion, using the dispersion coefﬁcient as a

characteristic indicator of dispersion.

6.3 Theoretical Development

An expression for the dispersion coefﬁcient was developed on the basis of the
Taylor-Axis model (Eq. 6.13). The model was modiﬁed by incorporating the following
rheological properties: apparent viscosity (1]), consistency coefﬁcient (K) and ﬂow
behavior index (n).

The molecular diffusion coefﬁcient (Dm) of a solution is expressed as a function of
temperature and viscosity via an empirical correlation, developed on the basis of the
Stokes-Einstein equation. The correlation is applied to a dilute solution, containing large
spherical solutes (A) in small molecules of solvent (B). The empirical correlation can be
expressed (Geankoplis, 1983) as:

9.961x10‘l6T
DAB = W (6.14)

94

For an isothermal system at temperature To, the molecular diffusion coefﬁcient is a

function of the inverse of the viscosity of the solution only; therefore Eq. 6.14 may be

 

rewritten as:
k
where k is a constant deﬁned by:
9.96x10"‘T,
= V?” (6.16)

This correlation may be used for a power law ﬂuid by replacing the Newtonian

viscosity (u) by the apparent viscosity (1]), given by:

 

n = K?" (6.17)
Incorporating Eqs. 6.15 and 6.17, Eq. 6.13 becomes:
k d? K?"

D = _, + __ __ 6.18

‘ [KW] (1921 k l ( ’
The mean velocity (17) of a power law ﬂuid can be calculated by (Bird et al., 1960):
_ 1,, ; nR
u - [7?] 1+3n (6°19)

Substituting Eq. 6.19 into Eq. 6.18, the dispersion coefﬁcient can be expressed as a
function of K and 11, both measurable rheological properties of the ﬂuid.

The geometry of an extruder is more complex than that of a tube. However, to
provide simplicity, the hydraulic diameter may be used to characterize the extruder as an
equivalent tube. If the conﬁguration consists of more than one screw in the fully ﬁlled

zone, the average hydraulic diameter of the screw combination may be calculated by:

95

4 f; V,“-
d, = ’ (6.20)
A...-

 

The equivalent length of the conﬁguration in terms of the above characterization
may then be calculated by:

r V -
= J 6.22
L. 32A“, < )
The mean velocity in the extruder then becomes:

L.

E

E = (6.22)

An average shear rate in the extruder can be calculated on the basis of a power

number versus Reynolds number relationship (the development is given in Chapter 7) and

 

 

is expressed as:
P0 = 9.12x103Re'1‘26 (6.23)
where
E.
P0 = pN3d: (6.24)
and
d21v
Re = kn p (6.25)

The dispersion coefﬁcient for an extrusion process can be estimated from Eq. 6.18
by appropriate substitution of db, yand 3, all of which are obtained ﬁom experiment.

Also, the molecular diffusion coefﬁcient (D...) can be calculated by Eq. 6.13. if D, is

known.

96

6.4 Materials and Methods

The effect of rheology on dispersion in an MPF-50D Baker-Perkins twin screw
extruder was investigated using corn syrup (type C, Cleveland Syrup, OH), 2.5% (w/w)
Methocel (K4M, Dow Corning, MI) and a mixture of 7% (w/w) soy polysaccharide
(SPS) (Fibrim 2000 soy ﬁber, Ralston Purina, MO) in honey. The extruder was operated
at 15 LID, using a combination of feed screws (FS), single lead (SL) screws and 90° .
forwarding paddles (90 F). The schematic of the extruder setup is shown in Figure 6.2.
Two circular full ﬂange dies, 3.2 mm in diameter and 25.4 mm in length, were used. The
barrel temperature was maintained at 10°C with chilled water. The material was fed at a
ﬁxed rate of 10 kg/hr. The extruder was operated at screw speeds of 100 and 150 rpm,
with fully-ﬁlled screws. The methocel solution could not be run at 100 rpm because it
tended to back up toward the feed port and entrap air bubbles.

RTD measurements were performed when steady state conditions were reached A
tracer ball containing 0.1-0.2 ml of 4% erythrosine and 0.2-0.3 g of the feed material was
dropped in the feed port. The extrudate was collected when a slight color was observed
and continued at intervals of 15 s until the tracer color was no longer visible. The
intensity of dye in each extrudate was measured using a CR-200 Chroma Meter (Minolta
Camera Co., Japan) against a base extrudate. The mean residence time and variance were
then calculated by using the ﬁrst and second moment formulations. The results were

used to estimate the dispersion number and dispersion coefﬁcient.

97

.o\4 m« we emuaeuxm zmeum CH3»
mcﬁxemmremxmm osmium: mcu moaaaumw mzu no owumsmcum m.m mezmﬂm

mzmeum mzoeum
ummﬁ mHmcﬁm mmﬁuumn h.om ummH mmmcmm wzoeum name
so mo.m so «m.vm so co m so m«.oa xcmnm

 

_ i

 

 

 

 

 

 

 

32. as w. “
as
_ _ _ _ _ _ _ _ _
/
n V
; \x
= :JL— __JL_ : _ ___I_L_ jJL
emuaumcmeu
meammoem

upon umom

98

6.4.1 Rheological behavior

The rheological behavior of corn syrup, 7% SPS in honey and 2.5 % methocel were
characterized by steady shear and dynamic experiments. The steady shear experiments
were carried out in a Haake concentric cylinder viscometer; the dynamic experiments
were done in a Rheometric Fluids Spectrometer. Steady shear data were collected via a
computer interface with the VI sensor of the Haake RV-12 system. The experiments
were conducted at temperatures of 20-50°C within as wide a shear rate range as possible.

The dynamic experiments were done on the Rheometric Fluids Spectrometer in
oscillatory mode, using a disk-plate system. The disk was 25 mm in diameter, and was
set at a gap of 1.0-1.4 mm. All experiments were performed at a ﬁxed strain rate of 1%,
in a frequency range of 0.1 to 100 rad/s. The viscoelastic properties of the ﬂuids were
recorded, using the RECAP2 software via a computer interface. For each ﬂuid, two
replicates were performed at or near room temperature.

All samples were withdrawn from the same batch used for extrusion runs.

6.5 Results and Discussion
6.5.1 Rheological data

Results from the steady shear experiments are presented in Table 6.1. As expected,
the corn syrup was found to be Newtonian, while 2.5% methocel in water and 7% SPS in

honey were non-Newtonian ﬂuids.

99

 

Table 6.1. Rheological models of the ﬂuids-used, derived from steady shear
experiments.

 

 

 

 

Fluid ‘ Test condition Rheological model
Comsywr Trange: 23-65°C 1 = u?
Yrange: 0—270 s'1 where
H = 2.73x10"‘exp(3%—"¥)
7% SPS inhoney Trange: 40-80°C 1 = “yum
7range: 0-180 s’1 where
11 = 5.92x10'l‘exp(3‘;3)
2.5% Methocel Trange: 23-35°C 1 = m7)”
yrange: 0-240 8'1 where

 

 

 

 

T] = 5.14exp($)

 

Results from oscillatory tests are shown in Figure 6.3 for corn syrup, Figure 6.4 for
7% SP8 in honey and Figure 6.5 for 2.5% methocel in water. The curves of G', G’, G”
and n' were different for each material. Since corn syrup is a pure liquid, 6” dominated
and its complex viscosity was frequency-independent and was constant (100 Pas) at
23.4°C. The slope of the G” curve was unity, indicating Newtonian behavior. The G’ of
corn syrup was zero at low frequencies and generally increased with increased frequency,
although this increase was inconsistent. There is no clear explanation for the shape of the
6’ curve. Theoretically, 0’ should be zero for Newtonian fluids. While this is not the

case, the contribution of G’ to G* was small.

100

.0 alum yo aPSm Eoo no”. Asv xocozvmt mzmnm> Ca

36033 $3800 vco Azov 3:69: 30. 58 22:60:. 609.9% mim 6.5mm

 

 

Gumﬁob 3
No+moé Fo+moé oo+moé Folmoé
—PL - b b b Ho ho —- n b - p h F b bk - h n n u n - POIMOoF
:0 a o o
o? 4 o o m .
. poo+moe e
o o H \1m
0 n fnru\
. . a .. 30$? 6
o o o o o o o n a a a 0 MW
¢<<<o<<<< <<< «<44 <<an<<<<< << 4 NO+MO.—. /1\
O O O D . D
o u a U
o a a Du
._ .. . .. 8+3; u.
. . . a
a a mu
.. ._ 40$? ,6
- n — - n n n b n —— n n p p p n - Pb — n p n n n -

 

- mo+moé

101

.o .mém #0 >95: 5 mam NH no“. AGV xocoacot mameo> is

 

 

 

 

36033 meEoo vco Azov 3:608 82 .68 3:69.: 600.65 To 950$
enmﬁob 3
No+moé Fo+moé oo+_moé Elmo.—
.o o No+mo e
.. :0 u ..
In C: q I
s. I
v . .. vno+moé
.. 4 4 a a a a a at
r 4 4 n n a a a j
H a D M O D D o O o .w.
w .. .. a. .. .. .. . .. . weo+moe
O 0 0 4 4 r
4 n
4 Am
-- p b p P n p —— p— - n p p n b- - u - n n r n H

mo+moé

(S'Dde PUD (0d) .9 ‘(Dd).o

102

.0 .m.NN 8.0 .8056: Rad no» Aav xocoaco: mama? 95

 

 

36083 83800 ecu Azov maznoE mac. .98 mastoE 000.55 06 830E
bumﬁob 3
No+uoé Fo+_m_oé oo+_m_o._ Elmo;
.0 o oo+mo P
I :0 D I
.l C? < .l
m . .w
.w.. . . vao+moé
- .. . . . . a
m d a a 4 o o a D D H
v . .. M. a. .. wmo+mog
4 a m u a . .T
r a a no. a 1
pho- ~ - P — L _—- - p n n p P —-h p — - n p m

 

 

no+moé

(3°od),u puo (Dd) no ‘(od),9

103

On the other hand, the SPS-honey and methocel mixtures exhibited purely
viscoelastic properties. For the SPS-honey mixture, 6’ was greater than 0” in the range
of frequencies studied, indicating elastic behavior in the transition or rubbery state.
However, the terminal state of the SPS-honey mixture could not be determined because
the ﬂuids spectrometer could not be operated at frequencies lower than 0.01. Increasing
the temperature up to 50°C did not signiﬁcantly change the viscoelastic data. No
experiments were conducted above 50°C.

In contrast to the SPS—honey mixture, methocel behaved as a viscous liquid at low
frequencies: 0” was larger than 0’. The complex viscosity decreased with increased

frequency, exhibiting shear thinning behavior.

Zero shear viscosity (7],) and equilibrium creep compliance (If) of corn syrup,

SPS-honey and methocel were determined by Eqs. 6.7 and 6.8. For the SPS-honey
mixture, the values could not be obtained because no data was available at frequencies
below 0.1 rad/s. For purposes of comparison, however, no and J," were calculated at 0.1
rad/s (reference frequency) for all three materials (Table 6.2).

J,“ is a measure of the energy stored in the molecules during ﬂow. The smaller the

value, the higher the energy stored and recovered. The more elastic the ﬂuid, the less
molecular rearrangements or molecular deformation occurs during ﬂow, and possibly the
smaller the degree of mixing or dispersion. Among the three ﬂuids, methocel produced
the greatest dispersion.

104

 

 

 

 

 

 

 

 

 

 

 

Table 6.2. Zero shear viscosity and equilibrium creep compliance of corn syrup,

7% SPS in honey and 2.5% methocel in water.

Material Temperature, C Reference 1],, Pas J," , Pa’1

frequency, rad/s

Corn syrup 23.4 .1 100 .0

7% SPS in honey 24.6 .1 5.5x10‘ .00016

2.5% methocel 22.8 .1 170 .021
6.5.2 Extrusion data

Using extrusion data (Appendix 6), Eqs. 6.23-6.25 and the rheological models of

the ﬂuids in Table 6.1, the average shear rates were calculated (Table 6.3). It is clear

from the Reynolds numbers in this table that the flow was laminar. Since the viscosity of

a Newtonian ﬂuid is shear-independent, the average shear rate in the extruder could not

be estimated for corn syrup.

Using RTD data, the dispersion number, mean velocity and dispersion coefficient

were calculated (T able 6.4). The dispersion coefﬁcient of corn syrup (n = 1) was larger

than that of SPS-honey (n = 0.416). This is in agreement with the results of Beoras and

Krantz (1976), and Subramanian and Gill (1976): the dispersion coefﬁcient of the

Newtonian ﬂuid is larger than that of the non-Newtonian ﬂuid, due to differences in the

velocity proﬁles (Figure 6.6). However, the dispersion of methocel (n = 0.3) was larger

than that of corn syrup, contradicting what was expected. It is clear, therefore, that

factors other than the ﬂow behavior index (11) may also be involved. Eqs. 6.18 shows that

the dispersion coefﬁcient depends on rheological behavior (K and n), shear rate and the

mean velocity, all of which are related in a complex fashion. Therefore, the effect of

each factor is not easy to isolate.

 

105

 

Table 6.3. Apparent viscosity and estimated average shear rate of corn syrup, 7%
SPS in honey and 2.5% methocel, at various extrusion conditions.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Material RPM Feed rate Avg. temp., c Mass Re n,Pa.s 11.. s"
(kg/hr) (s.v.) density,
g/ml
Corn syrup 100 9.4 23.5(0.8) 1.41 .0031 72.7 NA
22.1(0.9) .0029 76.5
150 9.6 23.2(1.1) 1.41 .0053 63.4 NA
.0053 63.3
7% SPS, 100 9.2 26.6(0.9) 1.24 .0027 74.0 19.9
93%honey 23.7(0.6) .0027 74.0 19.9
150 9.6 24.0(0.7) 1.24 .0049 60.0 24.7
23.8(2.0) .0049 60.0 24.7
2.5% 150 8.4 18.3(0.7) 1.10 .0081 32.2 5.1
Methocel 9.2 22.3(0.4) .0081 32.1 4.8

 

NA = Not applicable

 

Table 6.4. Dispersion number, dispersion coefﬁcient and molecular diffusion
coefﬁcient of corn syrup, 7% SPS in honey and 2.5% methocel.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Material RPM 7 (min) 0'2, min” .3;- ‘ii, cm/s 1),, Dmxlo‘,
. cm2/s cm2/s

Corn syrup 100 11.6 5.34 .0202 2.22 69.3 3.6
11.7 5.59 .0207 2.19 70.2 3.4
150 1 1.4 5.42 .0212 2.26 73.9 3.5
11.4 5.44 .0213 2.26 74.3 3.4
7% SP8, 100 13.6 6.41 .0176 1.89 51.3 5.1
93%honey 12.6 4.52 .0144 2.04 45.2 4.0
150 11.4 4.00 .0156 2.26 54.2 3.2
1 1.2 3.79 .0174 2.19 59.0 4.9
2.5% 150 9.7 4.09 .0222 2.65 91.0 3.9
Methocel 8.7 2.68 .0180 2.96 82.0 5 .3

 

 

 

106

.Aouc. :oHe maﬁa c“ ucm 4.4vc. uﬁaﬁu cmﬂcouzm21co: «Aﬁuc.
uﬁaae cmwcouzmz m we one menu emcwsma c“ mﬂwuoea >uwuoﬂm> m.m meamﬂd

 

 

 

 

 

107

Using Eq. 6.13, the effect of d,, Dm and '17 on dispersion coefﬁcient can be easily

determined. For a constant hydraulic diameter, the effect of velocity and the molecular
diffusion coefﬁcient on the dispersion coefﬁcient is plotted in Figure 6.7. It is apparent
that velocity is the dominant variable, causing a large change in the dispersion coefﬁcient
in comparison to the molecular diffusion coefﬁcient. Clearly, the dispersion of methocel
was larger than that of corn syrup because the average velocity of the methocel was large.
If the mean velocity could be maintained constant for all ﬂuids, the dispersion of
methocel would be smallest. The effect of viscosity on the molecular diffusion
coefﬁcient (Eq. 6.15) appeared to be insigniﬁcant.

Having obtained De from Eq. 6.18, the molecular diffusion coefficient (Dm) was
calculated from Eq. 6.13, using a hydraulic diameter of 0.98 cm for mixed screws (Table
6.4). Since Eq. 6.13 is quadratic, two values of Dm were obtained. The smaller value was
chosen, based on physical considerations. On the basis of these results, the Taylor model
(Eq. 6.11) which ignores axial molecular diffusion could be used with resulting errors of
less than 1%.

The axial and radial Peclet numbers, and dimensionless time were calculated and
are summarized in Table 6.5. Axial Peclet number is in a range of 45-69. The radial
Peclet numbers were larger than 1000, therefore, the one-dimensional dispersion model
can be used. The dimensionless times were in a range of 0.94-1.36; therefore, the

Taylor-Aris dispersion model is applicable.

108

.56 .5 co 883 Age E66568 863% .2822: 25
A3 b_oo_m> cooE m:m..o> Adv 2205600 56.535

1
’
’r
o. "
\ “—
‘0'
\9
‘i
o
\0
\

 

    

E

1 l L A A l l 1 l 1 l l A 1 1 1 l l l l L

8

 

E 656;

 

109

 

 

 

 

 

 

 

 

Table 6.5. Axial and radial Peclet number, and dimensionless time of corn syrup,
7% SPS in honey and 2.5% methocel.

. EL. ER. 0.:
Material RPM '5: E 7‘?
Corn syrup 100 49.5 3056 1.03

48.3 3138 1.00

150 47.2 3200 0.99

46.9 3219 0.98

7% SPS in 100 56.8 1819 1.73
honey 69.4 2518 1.25
150 64.1 3516 0.90

57.5 2208 1.36

2.5% 150 45.1 3364 0.94
methocel 55.6 2737 1.15

 

 

 

 

 

 

6.6 Conclusions

The effect of rheology on the dispersion coefﬁcient was investigated using corn
syrup, 7% SPS in honey and 2.5% methocel in water. The rheological behavior of the
ﬂuids was characterized by steady shear and dynamic experiments. Corn syrup was
conﬁrmed to be Newtonian, while SPS-honey (n = 0.416) and methocel (n = 0.3) were
non-Newtonian. The dispersion coefﬁcient of corn syrup was found to be higher than
that of SPS-honey. This is in general agreement with results from laminar tube ﬂow,
where the effect of velocity proﬁles on dispersion is more pronounced with Newtonian
ﬂuids than with non-Newtonian ﬂuids. Contrary to expectation, methocel produced more
dispersion than corn syrup.

The results showed that the extruder may be characterized as a tube if an
appropriate hydraulic diameter is used The effect of rheological behavior on the
dispersion coefﬁcient can be described by the Taylor-Aris model, as supported by the

110

experimental results of this study. Viscoelastic properties in terms of the equilibrium
creep compliance may be related to the dispersion of non-Newtonian ﬂuids: the higher
the equilibrium creep compliance, the greater the dispersion. However, additional study

is required to verify this relationship.

6.7 Nomenclature

A Solute molecule

A,i Cross-sectional area of screw type i, treated as a circular tube,
cm2

Awi Wetted area of screw i, cm2

B ' Solvent molecule

C Concentration of tracer, g/l

D, Effective dispersion coefﬁcient, cm2/s

DIn Molecular diffusion coefﬁcient, cm2/s

DAB Mass diffusion of solute A in solvent B, mZ/s

(1 Tube diameter, cm

(1, Hydraulic diameter of extruder, cm

G Transient relaxation modulus, Pa

0' Complex modulus, Pa

0’ Storage modulus, Pa

G” Loss modulus, Pa

J Transient creep compliance, Pa’1

J' Complex creep compliance, Pa’l

k Constant for a given solution under isothermal conditions

at"

n

oeapsuv

H

0"

u(r)

zl

111

Consistency coefﬁcient, Pa.sn
Equivalent length of the extruder, cm
Flow behavior index, dimensionless
Number of screw types
Tube radius, cm
Hydraulic radius, cm
Temperature, K
Isothermal temperature, K
time, s
Residence time, s
Velocity at any r, cm/s
Mean velocity in the tube or extruder, cm/s
Wetted volume of screw type i, cm3
Molar volume of solute A at boiling point, m3/kg mole
Axial length, cm
Greek symbols

Phase angle between stress and strain, radian
Apparent viscosity, Pa.s

Zero shear viscosity, Pa.s

Dynamic viscosity, Pa.s

Imaginary part of complex viscosity, Pa.s
Complex viscosity, Pa.s

Steady shear rate, s'l

112

(0 Frequency, rad/s

p Mass density, g/ml
t Dimensionless time
rw Wall shear stress, Pa
6.8 References

Ananthakrishnan, V., Gill, W.N. and Barduhn, AJ. 1965. Laminar dispersion in
capillaries: Part 1. Mathematical analysis. AIChE J. 11:1063.

Aris, R. 1956. On the dispersion of a solute in a ﬂuid ﬂowing through a tube. Proceed.

I Roy. Soc. London, Series A. 235:67.

Bird, RB, Armstrong, RC. and Hassager, O. 1987. Dynamics Qf Polymeric Liquids.
Wiley lnterscience. New York.

Bird, R.B., Stewart, W.B., and Lightfoot, EN. 1960. Transport Ehemmena. John Wiley
& Sons. New York.

Booras, GS. and Krantz, W.B. 1976. Dispersion in the laminar ﬂow of power-law ﬂuids
through straight tubes. Ind. Eng. Chem. Fundam. 15(4):249.

Butt, J .B. 1980. Ream Kinetics and Ream Design. Prentice-Hall, New Jersey.

Cussler, EL. 1984. Diffusion: mass transfer in ﬂuid ms. Cambridge University
Press. New York.

Ferry, J .D. 1980. Viscoelastic anenies of Bolymm. 3rd ed. Wiley Interscience. New
York.

Gcankoplia CJ. 1983. Imam: Processes and llnit eratiens. 2nd ed» Allyn and

Bacon, Boston.

113

Gill, W.N. 1967 . A note on the solution of transient dispersion problems. Proc. Roy. Soc.
London, Series A. 298:335.

Gill, W.N. and Sankarasubramanian, R. 1970. Exact analysis of unsteady convective
diffusion. Proc. Roy. Soc. London, Series A. 316:341.

Gill, W.N. and Sankarasubramanian, R. 1971. Dispersion of a non-uniform slug in
time-dependent ﬂow. Proc. Roy. Soc. London, Series A. 322:101.

Graessley, W.W. 1984. Viscoelasticity and ﬂow in polymer melts and concentrated
Solutions. In: Physical Empcnics cf £91m. Mark, J .E., Eisenberg, A.,
Graessley, W.W., Mandelkem, L. and Koenig, J .L. (Editors) American Chemical
Society, Washington DC.

Mohamed, 1.0. 1988. Modeling shear rate and heat transfer in a twin screw co-rotating
food extruder. Ph.D. Dissertation. Michigan State University. Lansing, MI.

Sankarasubramanian, R. and Gill, W.N. 1972. Dispersion from a prescribed
concentration distribution in time variable ﬂow. Proc. Roy. Soc. London, Series A.
329:479.

Sankarasubramanian, R. and Gill, W.N. 1973. Unsteady convective diffusion with
interphase mass transfer. Proc. Roy. Soc. London, Series A. 333:115.

Subramanian, RS. and Gill, W.N. 1976. Unsteady convection diffusion in
non-Newtonian ﬂows. Can. J. Chem. Eng. 54:121.

Taylor, G. 1953. Dispersion of soluble matter in solvent ﬂowing though a tube.
Proceed. Roy. Soc. London, Series A. 219:186.

Taylor, G. 1954. Conditions under which dispersion of a solute in a stream of solvent
can be used to measure molecular diffusion. Proceed. Roy. Soc. London, Series A.

225 :473.

114

7 A DISPERSION MODEL FOR PREDICTING THE EXTENT OF
LIQUEFACTION BY ALPHA-AMYLASE IN REACTIVE
EXTRUSION
7.1 Abstract

A Baker-Perkins co-rotating twin screw extruder was used as a bioreactor for
hydrolysis of pre-gelatinized corn starch, using thermophilic or -amylase. The extruder
was modeled as a tube, and characterized as a closed system. This characterization is not
in terms of thermodynamics, but rather relates to the nature of the tracer ﬂuid upon entry
into and exit from the reaction zone. The reaction kinetics were modeled by a modiﬁed
ﬁrst order equation, which allowed the dispersion equation to be solved analytically with
the Danckwerts boundary condition. Data from several extrusion runs were
super-imposed to obtain a proﬁle to evaluate the model. The dispersion number,
determined from the ﬁrst and second moments of the RTD curve was primarily a function
of the length of the reaction zone. Results showed good agreement between predictions
and experimental data, especially at low dispersion numbers. In general, the axial
dispersion model is suitable for analysis of enzymatic reactions up to 30% conversion.
At a ﬁxed ﬂow rate and constant temperature, the extent of starch conversion
signiﬁcantly depended on moisture content, residence time and enzyme dosage, but not

on screw speed.

7.2 Introduction and Background

The use of an HTST extruder as a biochemical reactor for enzymatic reactions is a
new application which has been extensively investigated by several researchers (Linko et

al., 1983; and Linko et al., 1984; Reinikainen et al., 1986; Komolprasert and Ofoli, 1989;

115

and Ofoli et al., 1989). Thermostable a—amylase can be used to partially hydrolyze
starch during extrusion to prepare pre-gelatinized and pre-liqueﬁed starch for subsequent
syrup manufacture (Linko and Linko, 1983). Although starch can be gelatinized and
liqueﬁed thermomechanically during extrusion, use of a -amylase provides advantages
in decreasing product viscosity and reducing or eliminating starch retrogradation.

The ﬂow behavior of the TSE is not well characterized, but generally falls between
the CSTR (complete mixing) and PFR (no mixing). Based on RTD measurements,
several models have been proposed to describe the ﬂow pattern in single screw extruders
(SSE). These include such conﬁgurations as PFR, CSTR, PFR+CSTR, CSTR in series
and PFR+CSTR in series. Few models have been proposed for the TSE because its ﬂow
patterns are more complex than those in the SSE. Bounie (1988) used Fourier functions
and multivariable optimization techniques to predict a pulse response from several ﬂow
models. He found that the model of PFR+nCSTR gave the best ﬁt to the experimental
data. Among the ﬂow models developed, the one-dimensional axial dispersion model is
used most frequently for polymer processes.

The dispersion model for laminar ﬂow in a tube was developed by Taylor and Axis
(Taylor, 1953; Aris, 1956) to describe laminar flow of a solute as a combined action of
molecular diffusion and a non-uniform velocity proﬁle (Butt, 1980):

02:2
D: “ Dr+1920,,

 

(7.1)

Bischoff (1968); and Bischoff and Levenspiel (1962) used the Taylor-Aris
dispersion model to characterize a two-dimensional system, using a ﬁrst-order reaction

equation. They found that the model produced accurate results if the reaction was slow.

116

In general, the dispersion model seems to be valid for a system where the residence time
is long enough to enable the process to be represented by the one—dimensional dispersion
model (Bischoff, 1968).

The boundary conditions used with the dispersion model depend on whether the
system is open or closed This characterization is not in the thermodynamic sense. A
closed system has a ﬂat velocity proﬁle, with no diffusion at the boundaries; an open
system involves molecular diffusion and axial dispersion at the boundaries (N auman,
1981). For a long tube where end effects can be ignored, the Peclet number is large and
the difference between open and closed systems is less critical. If an open system is
described by the dispersion model, the impulse response method can be used to
approximate the RTD function in exactly the same manner as a closed system (Gibilaro,
1978). The solution from RTD measurement depends on the injection and detection
modes (Kreft and Zuber, 1978).

The classical Taylor-Aris dispersion model was primarily used for Newtonian and
non-Newtonian ﬂuids at steady state. However, the model was modiﬁed during the
1970’s by Gill and co—workers (Gill, 1967; Gill and Sankarasubramanian, 1970; and
Sankarasubramanian and Gill, 1972) to provide a wider range of applications, including
unsteady state cases. Booras and Krantz (1976), and Subramanian and Gill (197 6)
applied the general dispersion model to non-Newtonian ﬂuids for laminar ﬂow in tubes
under unsteady state conditions. They found that the dispersion coefﬁcients increased
with time and approached a steady state value. Between pseudoplastic and dilatant

ﬂuids, the former caused more dispersion (Booras and Krantz, 1976).

117

Using RTD data and the moment method introduced by Levenspiel and Smith
(1957), Parimi and Harris (1957) found that results from a reactive tracer under unsteady
state conditions could be used to select, among model candidates, the most appropriate
model for describing a system of interest.

In general, the one-dimensional axial dispersion model is suitable for real systems
such as ﬁxed beds and ﬂow in tubes (Froment and Bischoff, 1979). The dispersion
number is normally determined from RTD data through various techniques (Froment and
Bischoff, 1979), including direct nonlinear regression, time moments (which can be
improved by incorporating a weighting factor (e "‘) or weighted moments (Michelsen,
1972)), ﬁtting RTD data by Laplace transform functions (Michelsen and Ostergaard,
1970, Pham and Keey 1977 and Van der Laan, 1958) and ﬁtting RTD data by Fourier
transform (Subramanian, 1977).

No investigation of biochemical reactions using the dispersion model was found in
the technical literature. Therefore, the goals of this study were 1) to determine the
dispersion coefﬁcients under several extrusion conditions, and 2) to use the dispersion
coefﬁcients in an axial dispersion model for predicting the extent of starch liquefaction

by a-amylase in a twin screw extruder.

7.3 Model Development
The dispersion equation derives from Fick’s law of molecular diffusion:

29 = "2222 (7.2)
at 81:

Since the extruder is modeled as a tube, the axial one-dimensional dispersion
equation may be written on the basis of the mass balance for starch. In terms of mass

ﬂow by convection, diffusion and reaction, the mass balance gives

118

d C, dC,
+D.A.

x x+Ax

 

 

C,,,'JA,-C,,,,,EA,+(—D,A, ]-(—r,)A,Ax = o (7.3)

Iqu. 7.3 is divided by A,Ax, then in the limit as Ax approaches zero, it becomes:

 

—u—-(—r) = 0 (7.4)
x

At a starch concentration of 40% at 95°C, the rate of starch conversion (- ,) may be

modeled by a modiﬁed ﬁrst order reaction (Chapter 4):

 

- dC' 0 033C ’ (7 5)
r, — dt . , .
where
C,’ = C, -0.06C, (7.6)

If the extruder is treated as a closed system, then Danckwerts boundary condition

may be used:
EC, = ECO, -D,(% L at x = 0 (7.7)
%xg = 0 at x = L (7.8)

Eq. 7.4 may be written in a non-dimensional format as:

 

1 dz? d5 .—
(157)422— dZ—ktC _ 0 (7.9)
where
'5 — C’ 710
— C0 ( ° )
Pe = “L (7.11)

p

119

and

z .—. (7.12)

i
L
The analytical solution of Eq. 7.9, in terms of C as a function of the normalized

length of the extruder (z), is (Danckwerts, 1953):

 

5(2) _ ex (Ia—,2) 2(1+§)cxp[’%‘(1-z)]-2(1-t;)exp[%‘(z-1)] (713)
p 2 (1+C)2eXP(’-;5)-(l-C)’cxp(i;5) °
where
4k:

Having obtained 5, the fractional conversion, X(z), can be calculated by:

X(z) = 1-E(z) (7.15)
Eq. 7.15 then allows the extent of starch hydrolysis to be determined along the

barrel.

7.4 Materials and Methods

A Baker-Perkins MPF-50D co-rotating twin screw food extruder (APV Baker,
Grand Rapids, MI) was used for all runs. Its barrel is 50 mm in diameter and 125 cm
long. The relevant dimensions of the extruder are given in Figure 7.1. The screw
conﬁguration in the TSE consisted of feed screws and single lead screws, followed by a
reaction zone of 90° paddles (Appendix 7) to promote mixing. The size and shape of all

screw elements used in the TSE are given in Figure 7.2.

.Lmuseuxm 2meum cﬂzu oomtddz mcﬂxemdtemxmm may do uao>m4 ﬁ.m mesmﬂd

 

 

 

 

 

 

 

so mv.mﬁ n 2 Eu 4m.n n 0
Eu mo.m n a so m4.4 n 2 “Lou umme Hmﬂamumz n dd
Eu mm.m n u Eu nm.m u m meaumemaEmu Hmeemm 1 me
new: mﬂa
m.
1. man
mm: ma EH mm; mm: 45 mm» m8. 2:.
_ _ _ _ _ _ _ _ _ +
n V a z
_ _ _ . x j
u _HL m u_ u ru_m “U_ u ru_m u _u _n—mw_ﬁu_ u _a_m _“U
and mad mad 4nd mad
Lmusumcmeu

mesmmmed

121

.Lmuaeuxm zmeum cﬂzu
meaxemdtgmxmm oomrddz mcu c“ umm: mucmsmam xenon do mcoﬂmcmeHn m.n meamwd

 

3:53 9.335. 1|" loam Tlll

 

 

 

 

 

 

 

mzogum new" mﬁmcdm mzmeum ummd
uana

 

 

L 5.1...
e In

+ J

an nmdm .- 8.3

+ L

 

 

 

 

122

The entire experiment consisted of twenty extrusion runs. The ﬁrst twelve were
run at a constant length of single lead and feed screws, but variable lengths of paddles at
two screw rpm, moisture contents and enzyme levels (Table 7 .1). The experimental
design of the last eight runs are given in Table 7 .2. The schematics of screw proﬁles and
all injection ports are shown in Figures 7.3, 7.4 and 7.5 for runs 1-12, 13-16 and 17-20,
respectively. Two circular full ﬂange dies 3.2 mm in diameter and 25.4 mm in length
were used

Pre-gelatinized corn starch (American Maize, Hammond, Indiana) was used as
substrate. Liquefaction was carried out only in the paddle zone where a high level of
mixing was obtained. Termamyl 120L and 240L or -amylase (N ovo Laboratories,
Connecticut) were used. All extrusion runs were done at a ﬁxed mass ﬂow rate of 10
kg/hr. Barrel temperatures in the paddle zone were ﬁxed at 100°C for extrusion runs 1-12
and 95°C for runs 13-20. Termamyl 120L enzyme (30,000 units/ml) was diluted in
deionized water, and Termamy1240L (80,000 units/ml) was diluted in 30 mM acetate
buffer at a pH of 6.0 before being fed into the extruder. Enzyme activity was determined
by the DNS procedure (Saha et al., 1987). Extrusion runs 13-20 were operated at 60 rpm,

60% moisture content and 45 enzyme units (activity/g starch).

123

 

Table 7.1. Experimental design for extrusion runs 1-12.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Run no. Paddle length RPM % Moisture Enzyme level

(cm) (units/g starch)

1 19.1 30 60 23

2 19.1 60 60 23

3 19.1 30 50 30

4 19.1 60 50 30

5 44.5 30 60 23

6 44.5 60 60 23

7 44.5 30 50 30

8 44.5 60 50 30

9 69.9 30 60 23

10 69.9 60 60 23

11 69.9 30 50 30

12 69.9 60 50 30

(pH = 6.0-6.5, T = 100°C and m = 10 kg/hr.)
Table 7.2. Experimental conditions for extrusion runs 13-20.
Length of screws, left to right (from feed Location of liquid
port to the die), cm streams (distance
from feed port), cm
Run L/D Feed Single 90° disks Single Buffer Enzyme
no. ratio lead lead

13 25 15.2 5.1 24.1 5.1 11.11 95.3
14 69.9
15 44.5
16 19.1
17 10 15.2 5.1 24.1 5.1 — 11.1
18 15 15.2 5.1 49.5 5.1 - 11.1
19 20 15.2 5.1 74.9 5.1 - 11.1
20 25 20.3 12.7 87.6 5.1 - 19.1

 

(40% starch, pH = 6.0, T = 95°C, 60 rpm, 45 enzyme units/g starch and ti: = 10 kg/hr)

 

 

124

.mm>u among» ucm Am. coﬂuaaom ms>~cm Lo
mucﬂoa >Lucm mcHzocm mqtu mucmsﬂgmaxm Lou aaumm emuaeuxm m.m ensued

4m “:8 .5 m... 3.58
T 1.0 L

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

26 .41: m Loam: :uemum
Twl #8 .5 m“. 6.53 '—
ummc 0.2... m>u 81.9w Loam: cULQHm
m3 / Twat ”row 4m mu. '_
/ _ _ _ _ _ _ _ _ H
/
n m
\
:rl :JL. :m1:JL_j
LmMHMMMMHM 26 .318 m Loam:

zuemum

125

.mm>u emumgu ucm Am. coﬂuaﬂom me>~cm ea

mucuon >Lucm mcﬁzocm warm“ mucmsﬁemaxm Lou aaumm amuseuxm 4.5 meamwd

 

 

 

ozoeum mzogum

new" mamcum magnuma u.om new” mamcam mzoeum come

.3 86 .3 3.8 .3 2.3 .3 ~33
can: «do
can
_ _ _ _ _ _ _ _ _
/
n L

 

 

 

 

 

_ __n_L:_r_u:_r1:r_L_:\

Luuaumcmeu m>u m>u m>u m>u
unammogd .m«.m .4«.m Am«.m Am«.m Loam: :uemum

126

.Am. coﬁusaom ms>~cm do

mucﬂon >eucm mcﬁzocm omtwﬁ mucmeﬂgmaxm no; aaumm Louzguxm m.m meamﬁd

 

 

 

 

 

 

 

ﬂ)
4-0
G
.1
U)
1.—

4m ma xcmﬂm
» De...
A5«.m am we xcmﬁm
a om
. » cuamnm
.m«.m 4m mm xcmam
h.om
+ guamnm
Am«.m
m cm ma

 

 

 

 

 

 

:r_L_
A

Louzumcmgu
mgammogd

1/
\L
:ml 7] :r1 3wa

as m 73.53

127

The extrudate was collected after the extruder reached steady state. About 10 g of
sample was collected in an aluminum pan, 5 cm in diameter and 1.5 cm in height, and
immediately immersed in liquid nitrogen to freeze the enzymatic reaction. Six samples
were collected from each run, three before and three after the RTD measurement. RTD
measurement was performed in all runs. During run 20, three additional RTD
measurements were conducted to enable determination of whether the dispersion number
changes if the tracer dye is introduced at different points of the extruder for a ﬁxed feed
port (Figure 7.6). The frozen samples were stored in a household freezer maintained at
-l7°C for subsequent analyses. The moisture content of the extrudate was determined by
the oven method at loo-105°C.

7.4.1 Determination of residence time distribution (RTD)

Residence time was determined by a tracer technique using an erythrosine solution
(4%, w/v in water). From 0.1 to 0.5 ml of the dye solution was injected via a one ml
plastic syringe through a port located near the enzyme feed port. Extrudate collection
began when a slight change in color was ﬁrst observed, and continued at intervals of 155

until discoloration was no longer visible.

128

.coﬂuzﬁom m>u Lo mycﬂoa >Lecm mcﬂzocm

m.om ucm m.om .ﬁ.om mucmsﬂLmaxm Led aapmm LmuzLuxm m.m mLsde

 

 

 

 

 

 

 

 

mzmgum mzmeum
umma mﬂmcﬂm mmﬁuoma n.om ummH mﬁmcﬂm mszum ummw
so mo.m so mm.mm eu om.m« so mm.om
umm: mﬁo
mam
_ _ __ _ _ _ _ _
/
n L
\
:_I_.u_ __.I_L_ :JL. :me __
Lmuaomcmgu m>u m>u m>u
cowuaﬂom cugmum

288.5 2 . ONE 8 . ONE 8 . omv
m2>~cm

129

The extrudates were cooled at room temperature, covered with plastic wrap and
placed in a household refrigerator overnight. All extrudates were removed from the
refrigerator and allowed to equilibrate at room temperature for 30 minutes before
analysis. Each extrudate was stirred to obtain a uniform color before measuring the
intensity of dye color against a base extrudate containing no dye with a Minolta Chroma
Meter model CR-200 (Minolta Camera Co., Japan). Because the meter had a small
window of 8 mm in diameter, tluee measurements at different spots were taken,
automatically averaged and recorded through a computer interface. The "a” values,
which represent the degree of redness, are proportional to dye concentrations. A C-curve

was obtained from a plot of the "a" values versus residence time.

The mean residence time}, was approximated by
i :,Aa,A:,

i=1

; = —..-— (7.16)
.21 4014!.-

where Ad,- is the redness value of the extrudate at discrete time values ti, and At, is the

time step (15 s).

Based on the methods used to inject and detect the dye, the extruder could be
characterized as a closed system. It is worth noting that this characterization is not in a
thermodynamic sense.

7.4.2 Estimation of dispersion coefﬁcients
Using the RTD data, the variance of the RTD curve was calculated:
5: imam,

,2 ___ ;:1____ _
$132,711,-

"‘4

(7.17)

130

For a closed reactor, the dispersion number (DJuL) could be calculated from the

implicit expression (Levenspiel, 1972):

o-2 D, D, 2 _g-r
:5 - 2(fi]‘2(ﬁ'] (1_e .) (7.18)

The dispersion coefficient (D,) can be directly determined from the dispersion
number if the average velocity and the reactor length are known. If the extrudates are
relatively constant in density, the average velocity for the fully ﬁlled screw region may
be calculated by:

_ m
— PA. (7.19)

 

where m is the mass ﬂow rate (g/min) and A, is the wetted cross sectional area of the

extruder. However, when the extruder conﬁguration consists of more than one screw
type, estimation of A, becomes difﬁcult. The mean velocity must then be computed

from:

_ L,
u = ? (7.20)

where L, is the equivalent length of a given screw section, and can be calculated as the
length equivalent to that of a circular pipe containing the same volume:

Vm‘
*th

4

L, = 2 (7.21)
D1“ in Eq. 7.21 is the hydraulic diameter of screw type i, which is 0.75 cm for single lead
screws (Mohamed, 1988) and 1.04 cm for 90° paddles. The lengths of the die head and
the die were neglected because they contribute equivalent lengths much shorter than that

of the screws.

131

7.4.3 Sample preparation and analysis for reducing sugar

Frozen extrudate was thawed rapidly in a Tappan microwave oven (650 W) for 10
seconds, or thawed slowly in a plastic bag at room temperature for 15-20 min. A mixture
containing 0.5-1.0 g of the sample, 0.5-1.0 ml of 0.1 N N aOH and 10 ml distilled water
was homogenized using a Polytron homogenizer (Brinkmann Instruments, NY) at 15,000
rpm for 10-15 seconds and then heated at 100°C for 10 minutes. The solution was cooled
and brought up to 25 g by adding water, and centrifuged at 13,000-15,000 rpm for 10—15
minutes using a Sorvall centrifuge (Dupont, IL). The supernatant was removed and
subjected to sugar analyses by the DNS and %DE methods.

The quantity of total reducing sugar in the extrudate was determined using a
colorimetric procedure developed by Bemfeld (1955) and modiﬁed by Saha et al. (1987 ).
The reducing sugars liberated from the enzymatic liquefaction of starch reacted with
3,5-Dinitrosalicyclic acid (DN S) and then formed a brown product upon heating. The
absorbance of the brown color was measured against a blank (distilled water) at 640 nm,
using a Lambda 4B Perkin-Elmer Spectrophotometer (Perkin-Elmer Corp., IL). The total
amount of reducing sugars in the sample was calculated relative to the known glucose
concentration, based on Beer’s law.

The DE value was determined, using the procedure of Dygert et al. (1965). Upon
heating, neocuproine hydrochloride (2,9-dimethyl-1,10-phenanthroline.HCl) forms a
colored complex via chelation with Cu++ ion produced by the oxidation of reducing
sugars. Glycine was used as the copper chelating agent in place of tartrate to improve the
assay’s precision. The colored complex was measured spectrophotometrically against a
blank at 450 nm. The amount of reducing sugar in dextrose equivalent was calculated
relative to a known dextrose concentration which was treated in exactly the same manner

as the sample and the blank. In the analysis, the sugar extract was diluted to obtain a

132

solution containing 50-100 micrograms of glucose equivalent per ml. The amount of

reducing sugar was estimated by using the percent reducing sugar obtained by the DNS
method

The DE value was calculated by:

amount of reducing sugar equi. to dextrose
DE = .22
wt. of sample (dry basis) x100 (7 )

7.4.4 Estimation of the apparent viscosity of starch in the extruder

The apparent viscosity of liqueﬁed starch in the extruder was calculated by the
procedure described by Mohamed (1988). A series of extrusion runs were performed
using corn syrup type C (Cleveland Syrup Corp., OH) as a Newtonian standard over a
ratio of 15 LID in a Baker-Perkins MPF 50D co-rotating twin screw food extruder. The
screw proﬁle consisted of feed screws, single lead screws and 90° forwarding paddles
(Figure 7.7). Two circular dies 3.2 mm in diameter and 25.4 mm in length were used.
With this screw proﬁle, the reaction zone (paddle region) was fully ﬁlled in all extrusion
runs operated at a ﬂow rate of 7-10 kg/hr. At low rpm, the material backed up toward the
feed port indicating that the channel was ﬁlled. As the rpm increased, the backup ﬂow
ceased but the screw region from the paddles to the dies remained fully ﬁlled

Process parameters (rpm, % torque, barrel temperature, product temperature and
die pressure) were recorded after the extruder reached steady state. The throughput was
determined by collecting the extrudate over one minute and weighing it. The product

temperature was recorded along the barrel and at the die.

133

.mucmeﬁemuxm aae>m cgoo Lee gmuaguxm
szum cﬂzu mcwxamdtgmxmm oomtdd: may do azumm uﬂumemcom 5.5 mgamﬂd

mzuaum exogum
ummﬂ oHuch mmauuma m cm umoa mamcmm mzmeum new»
.3 8.... 3 836. .3 8 m .3 2.2 6.38

 

 

 

 

 

 

 

 

32.83 I __ _ n
as
_ _ _ L _ _ _ _ _
/
n L
\
__r_L_ __rl 2D: :r1 :r1
Louaumcmgu
oeammoed

aae>m :Loo

134

Using extrusion data, the power number for corn syrup was calculated (Mohamed,
1988) by:

 

P, = p55: (7.23)
where

E, = Pw-APQ, (7.24)

Pw = 0.354(% Torgue)N (7.25)

The Reynolds number was calculated by (Mohamed, 1988):

DZN
Re = Au 9 (7.26)

 

A relationship between Po and Re was developed, using regression analysis. Since
the starch material used the same screw proﬁle as the corn syrup, the Newtonian Po and
Re relationship can be used to calculate the corresponding Re (Re’) for the starch material
for a given Po. The apparent viscosity (T1,) can then be calculated by:

 

_ prN
no _ Re’ (7.27)
7.4.5 Rheological data

The rheological properties of corn syrup were determined using a Haake RV-12
viscometer consisting of an M-SOO measuring head, SVI sensor, Haake speed programer,
Haake F3-C temperature controller, Hewlett-Packard 3495 data acquisitor and
Hewlett-Packard 85 computer for data collection and analysis. The samples used to
collect rheological data were withdrawn from the same batch used for the extrusion runs.
The rheological data were taken at ﬁve different temperatures, and over as wide a shear

rate range as possible.

135

7.5 Results and Discussion

7.5.1 Estimation of dispersion numbers and dispersion coefficients

The C-curves of extrusion runs 1-12 are plotted in Figure 7.8 and 7.9. All the
curves were similar in shape, and tended to have long tails. Using the RTD data for
extrusion runs 1-12 (Appendix 8) and Eqs. 7.16-7.18 and 7.20, the mean residence time,
dispersion number and dispersion coefﬁcient were determined. The results were grouped

by rpm and enzyme level and are summarized in Table 7.3.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 7.3. Mean residence time, dispersion number and dispersion coefﬁcient
from extrusion runs 1-12.
Run Length’, 3 0'2 5D}:— 27 De (cm2/s)
no. cm (mm) (mm’) x102 (cm/s)
1 674 4.7 0.9 2.19 2.4 35.7
5 1152 5.6 1.2 2.05 3.5 81.5
9 1630 7 .0 0.9 1.00 3.9 63.1
2 674 4.5 1.5 3.84 2.5 64.1
6 1152 5.2 1.6 2.98 3.7 126.3
10 1630 6.8 0.8 0.87 4.7 56.5
3 674 5.7 2.2 3.53 2.0 46.7
7 1152 6.4 1.1 1.39 3.0 48.1
1 l 1630 8.8 2.2 1.39 3.1 69.6
4 674 5.3 2.3 4.23 2.1 60.6
8 1152 6.3 1.7 2.19 3.0 76.4
12 1630 8.3 2.0 1.46 3.3 78.3
' Length of equivalent path of tracer dye, calculated by Eq. 7 .21.

.'—_-ﬁ__ 7,

....~_. -.

136

 

 

.ycmwcoo mLBmBE Now

a .m .m .N J mecoEtoaxo Eot moisolo mK 830E

 

 

 

#0 OF Dco
835?. J
m: .1 me or m w .v N O
p _ r _ .I .. . _ no b _ ., — L 0.0
.t! L! .9! . N ~_... _
x 279%”! mm m... __ 1
I 3.... _. h: __
1 / ,J. . i. a... m. 10.2
1 u
. , f, m r __ -
, 2,. i .5 .6
V... k u. _ wl . a
l x m; m. __ _mmw 0 ON MW.
/ r N. _ T _ w
- , m a. _ r. . w
/ k ,2 _ m m. x)
I / my '2 N _m I 10.0w MW
/ . \ ..__
,. oeaxo XIX f M 7.3% T“. 7
m axe ale \ t. 4.» z a
I. Q QXO OilAv /VJPJH / ”A"... H 7.0.0.?
.. E / e... .
r x
.. N axm Dim. (Kama x
— axe 0:0
_ _ F _ h _ h _ _ P p _ _ _ _ 0.0m.

 

 

137

.HC®#COO QLmemopt NOW

#0 NF 95 : .m K on .n epceEteaxe E0: meisolo 0K eSEa

 

 

 

meSEE J
m: [we Nu OF m o .v w O
. ’1? . L I I‘d .. h 0.0
If. . may“? a a x a he
.. / I: A a .. s x
.2 .IduWﬂ \ _.. .
/ «Sui/M ‘ A\\ ... .a .
I. / 4.)¢</\ \ \ .... a 7.0 O—
. / wt; « a h *
a. 8 \ \ u _ 7
j é .. . .u __ a
. / a _. m. __ .u.
2. a . x .u _ x w
x a ..... m .. >
.. er ‘ .... r 8.2.. m.
N— axe XIX \ .r/ a.“
.. .5 .. _ ..
I axe 616 K I. .3
K . .
1 m as is «.1... 10.0...
h axe 4 La .5
.. é axe 0.6 r
n axe ole
- _ - F n h n — p _ — — n — _ 0.0m

 

 

138

Results from ANOVA (Appendix 9) showed that the effect of the reaction zone,
moisture content and enzyme level on the mean residence time were signiﬁcant at P<.01.
The mean residence time increased with increased length of reaction zone. The effects of
reactor length and screw rpm on the dispersion number were also signiﬁcant (Appendix
10). The mean velocity and dispersion coefﬁcient ﬂuctuated, probably due to effects of
screw proﬁle and enzymatic reaction. Differences in screw profile caused the average
cross-sectional area to vary, which resulted in changing the mean velocity. At short
reactor lengths or residence times, the extent of starch hydrolysis was small. Therefore,
the extrudate was solid-like, and the flow approached plug flow, i.e. the degree of
dispersion was small. When the residence time increased, starch was more liqueﬁed and
became liquid-like; the ﬂow deviated from plug flow, with increased mixing and
dispersion.

Under the same extrusion conditions, the mean residence time during enzymatic
reaction was shorter than when no enzyme was added. This effect was due to viscosity
reduction, which resulted in increasing the mean velocity, thereby decreasing the
residence time and increasing dispersion. The results at 60% moisture were different
from those at 50% moisture. This may be due to other factors such as errors caused by
truncation of the tail of the RTD curve.

C-curves of extrusion runs 13-16 and 17-20 are plotted in Figures 7.10 and 7.11,
respectively. Mean residence time, dispersion number and dispersion coefﬁcients for

extrusion runs 13-20 were determined and are presented in Table 7.4.

139

.A>_e>$oeame._ .EE \ld— Uco mum Kay .N.N ”meEz eocev_me._ 89.5

312 meceEteaxe Eot AeE: eoceEmE m:m..e> mmeceeb eeﬁsolo o—K Semi

meSEE J.

 

 

_

m: axe
MF axe
a; axe
m; axe
_

_ L

 

0.0

10.0—

10.00

 

 

ode

(D) sseupea

140

.Abeéoeamee .EE Tow eco mtg Nd .mcw ”meE: eeceEeE cerv

owl: muceEteaxe E0: AeEz eocee_me._ mzmce> meeceeb metzolo

meSEE J

 

 

ON axe
m— axe
m_ axe

NF axe
_. . _

 

 

 

0.0+.

:K eunmE
o
0.0
10.0—
.. no
9
D:
U
roam %
S
. {an
Iodm

141

 

 

 

 

 

Table 7.4. Mean residence time, dispersion number and dispersion coefﬁcient
from extrusion runs 13-20.
.. - D, _
Run Length , t o’ 5 u D.
no.’ cm (min) (minz) x102 (cm/s) (cm /s)
13 674 2.2 0.5 5.18 5.1 179.2
14 1152 4.7 1.4 3.40 4.1 161.0
15 1630 7.6 1.4 1.21 3.6 70. 9
16 2107 10.7 1.6 0.73 3.3 50. 7
17 1 143 4.6 1.7 4.05 4.1 189.9
18 1620 5.7 0.9 1.40 4.7 107.1
19 2098 10.3 2.3 1.09 3. 4 77. 6
20 2681 10.4 2.5 1.15 4. 3 132.7

 

 

 

 

 

 

 

 

LA dye solution was used for runs 13- 16 and a dye ball for runs 17- 20.
"Length of equivalent path of tracer dye, calculated by Eq. 7. 21.

Runs 13-20 also showed that increasing the reactor length resulted in increased
mean residence time and decreased dispersion number.

In runs 13—16 the dispersion coefﬁcients fluctuated in exactly the same manner as
observed for runs 2, 6 and 10, although the magnitudes were different. For the same path
length, increased enzyme levels (runs 13-16) caused the liquefaction of starch to increase,
which resulted in greater dispersion. Dispersion was also expected to increase with the
flow behavior index (n). The higher the extent of liquefaction, the closer n is to unity,
and the closer the ﬂuid approached Newtonian behavior.

The dispersion coefﬁcient is a function of velocity, viscosity and the hydraulic
diameter of the ﬂow path in the extruder. It generally decreased with decreased velocity
(runs 13-19). In addition, longer equivalent lengths appear to cause a signiﬁcant decrease

in the dispersion number, leading to smaller dispersion coefﬁcients.

142

Experimental runs were made to determine if the extent of starch hydrolysis and
degree of mixing depended on how the enzyme was fed into the extruder. In extrusion
runs 1-16, the starch and about 80% of the water or buffer solution (pH 6.0) were
pre-mixed in the extruder before adding the enzyme solution. In contrast, the enzyme
solution was directly mixed in with starch in the extruder during extrusion runs 17-20.
Statistical analysis (Appendix 11) showed no signiﬁcant difference (P<.05) between the
dispersion numbers obtained from the two methods.

Additional RTD data (sets 20.1-20.3) were collected during extrusion run 20. The

dispersion coefﬁcients for these runs are presented in Table 7.5.

 

Table 7.5. Mean velocity and dispersion coefﬁcient at various reaction zone
lengths during extrusion run 20.

 

. - D _
Run Length , cm I (min) 0'2 (minz) :7; x102 u (cm/s) Da (cmz/s)

no.

 

 

 

 

 

 

 

 

20.1 943 3.7 0.7 2.68 4.2 106.5
20.2 1421 6.6 1.9 2.29 3.6 117.2
20.3 1898 9.3 2.2 1.26 3.4 81.5

 

 

' Length of equivalent path of tracer dye calculated by Eq. 7.21.

With the exception of the D, value of run 20.2 in Table 7.5, the dispersion
coefﬁcient decreased with increased length of the reactor. It is worth noting that the
mean velocity varied along the barrel of the reactor, an indication that the mean velocity
causes a variation in the dispersion coefﬁcient, in the same manner as observed in
Chapter 6. In fact, the change in velocity is inﬂuenced by the screw proﬁle and the
rheological properties of the ﬂuid. Therefore, the dispersion coefficient depended on

143

several complex interactions. For example, D, depends on the molecular diffusion
coefﬁcient (Du), mean velocity and hydraulic diameter of the extruder (Eq. 7.1). From
the results of this study, the dispersion number provides the best parameter for
characterizing the degree of dispersion.

7.5.2 Use of the dispersion model to predict RTD

In this study, the dispersion number was estimated by using the ﬁrst and second
moments, which were developed theoretically from the model of ﬂow behavior.
However, it is of interest to determine whether the dispersion number could be used to
predict the tracer (dye) concentration proﬁle or F function.

The extruder was treated as a closed system with the Danckwerts boundary
conditions (Eqs. 7 .7-7.8). The analytical solution in terms of E for the non-reacting

tracer can be expressed (Butt, 1980) as:

n.) = g; = ;-;.{;—1—_—] (7.2.)
where

e1f(y) = %Iexp(—Vz)dv (7.29)
and

1: = g (7.30)

Eq. 7.28 was solved numerically by using the Gauss-Legendre Quadrature method
(Carnahan etal., 1969). The results were compared to the experimental F values by using

the linear regression:

Fmdcl = bFap (7‘31)

144

Values of b and standard error for all extrusion runs are presented in Table 7.6.
The "b" value is an indication of how well the two F values are related. If the "b" value
equals to unity, the model is in perfect agreement with the experimental data. The
deviation of the model from the experimental data was measured in terms of a standard
error. The results showed that as the length of the reactor increased, b values increased
while standard errors decreased (R2 = 0.99 at P<.05). On the whole, the F values were
within 5% of each other. An example of a comparison of F-curves of the model and the
experiment is shown in Figure 7.12, and illustrates the fact that the model and the
experiment became closer as the reactor length increased This is consistent with the
technical literature: the dispersion model is valid for long reactors or long residence

times. These results also validate the characterization of the extruder as a closed reactor.

145

 

Table 7.6. Values of "b" (Eq. 7.31) and standard error for various extrusion

 

 

 

 

 

 

 

 

 

 

 

 

 

runs.
Run Length. :7‘ b Standard
no. em x103 error
1 674 21.91 0.951 .014
5 1 152 20.50 0.964 .013
9 1630 9.95 0.969 .01 1
2 674 38.42 0.951 .013
6 1152 29.80 0.958 .013
10 1630 8.74 0.968 .013
3 674 35.33 0.952 .012
7 1152 13.95 0.967 .012
10 1630 13.92 0.970 .009
4 674 42.25 0.953 .01 l
8 1152 21.90 0.962 .012
12 1630 14.60 0.967 .010
13 674 51.76 0.945 .018
14 1152 33.98 0.964 .010
15 1630 12.14 0.972 .009
16 2107 7.29 0.973 .008
17 1 143 40.48 0.959 .013
18 1620 13.99 0.971 .012
19 2098 10.93 0.973 .009
20 2681 1 1.49 0.974 .008
20. 1 942.9 26.84 0.957 .016
20.2 1420.6 22.88 0.967 .010
20.3 1898.3 12.64 0.967 .008

 

° Length of equivalent path of tracer dye, calculated by Eq. 7.21.

 

146

.55.. 5.09 .x ”SE 95 .+ 2:8 54 .o 6:: mfm .1. “meEB eocev_me._

cue: .eEz msmee> 20:05: a “632328 Use _BceEteaxm N: 9.sz

 

meSEE J
m: 3 NF 9. m m cm N o
_ . _ . _ . _ . .- v o; - 0.0
2 axe x .
.1 axe o ..
m— axe + xx + I .
m: axe x .. . . N o
veg—630.00 .II + o ..
x o 14.0
x + o . 4..
X + O I 1w 0 m
x + o .- lm.O
..... Io.—

 

147

7.5.3 Extent of starch liquefaction in the extruder

The extent of enzymatic starch hydrolysis (Appendix 12) were expressed in terms
of %reducing sugar, %DE and fractional conversion (X). Results are shown in Table 7 .7
for extrusion runs 1-12 and Table 7.8 for runs 13-20.

Since the dye could not be injected at the same port as the enzyme solution, the

actual residence time of the reaction was calculated by:

.. L.’
I = =
u

(7.32)

where L; is the equivalent length of the reaction zone, and '17 is the mean velocity

calculated from the RTD data.

Without addition of enzyme, the extrudate contained <0.7 % reducing sugar and
<0.4% DE, which were produced by thermomechanical breakdown of the starch during
extrusion. Since these values were small, they were ignored in the calculation of the
amount of starch liqueﬁed by the enzyme.

The effect of moisture content, residence time and enzyme level on % reducing
sugar and % DE were statistically signiﬁcant at P<.05 (Appendix 13). However, the
effect of screw rotation in the range of 30-60 rpm was not signiﬁcant. Starch analysis
showed that initial starch concentrations were 38.8% for runs 13-16 and 42.8% for runs
17-20. Fractional conversions for runs 13-20 were calculated. Results from AN OVA
(Appendix 14) showed that there was no signiﬁcant differences in % reducing sugar and
% DB, between runs 13-16 and runs 17-20. This is an indication that the method of
introducing the enzyme solution into the extruder does not insigniﬁcantly affect the

reaction at P<.05.

148

 

Table 7.7. Percent reducing sugar and DE (extrusion runs 1-12).

 

 

 

 

 

 

Run Length', 1', min RPM Enzyme % Reducing %DE (3 v )
no. em dosage sugar(s.v.)
1 794 5. 5 30 23 3.32 (0.13) 1.94 (0.13)
5 1271 6.1 30 23 5.28 (0.16) 2.54 (0.15)
9 1749 '7. 5 30 23 6.62 (0.35) 3.51 (0.75)
2 794 5. 4 60 23 3.61 (0.33) 1.96 (0.20)
6 1271 5. 8 60 23 5.65 (0.62) 2.66 (0.31)
10 1749 6. 2 60 23 6.71 (0.26) 3.54 (0.51)
3 794 6. 8 30 30 6.31 (0.64) 3.30 (0.45)
7 1271 7 .0 30 30 7.98 (0.52) 3.37 (0.18)
11 1749 9. 5 30 30 8.43 (0.71) 4.40 (0.35)
4 794 6. 2 60 30 6. 32 (0. 83 3.03 (0.42)
8 1271 7. 0 60 30 8.67 (0.42) 3.52 (0.17)
12 1749 8. 9 60 30 8 43 (0.34) 4.12 (0.48)

 

 

 

 

 

 

 

' Length of equivalent path of enzyme, calculated by Eq. 7.21.

 

Table 7.8. Percent of un-reacted starch, reducing sugar and DE, and fractional
conversion (from extrusion runs 13-20).

 

 

 

 

 

 

 

 

 

 

 

Run Length', 1 , min %un- % Total reducing % DE (s.v.) X
no. cm reacted sugar (s.v.)
starch
13 794 2.4 34.3 8.52(0.80) 3.28(0.26) 0.12
14 1271 5.2 34.4 9.98(0.88) 3.43(0.44) 0.11
15 1749 8.1 30.8 12.28(l.10) 4.69(0.76) 0. 21
16 2227 ' 11.3 27.6 15.12(0.90) 6.27(0.89) 0. 29
17 1 143 4.6 34.7 10.44(l.00) 2.75 (0.56) 0.18
18 1620 5 .7 33.0 17.71(1.32) 5.88(0.42) 0.23
19 2098 10.3 32.7 15.66(0.79) 4.44(0.61) 0.23
20 2681 10.4 31.4 20.36(1.28) 7.05 (0.73) 0.27

 

' Length of equivalent path of enzyme, calculated by Eq. 7.21.

 

149

7.5.4 Relationship between power number and Reynolds number for

mixed screws

The consistency coefﬁcients for corn syrup at various temperatures were

determined by regression analysis (R2 = 0.99-1.00) and are presented in Table 7 .9.

 

 

 

Table 7.9. Consistency coefﬁcients for corn syrup at various temperatures.
Temperature, °C Consistency coefﬁcient, K (Pas)
23 66.71
25 55.53
34 15.86
50 3.17
65 0.86

 

 

 

 

A plot of log K versus l/I‘ (Figure 7.13) shows an Arrhenius relationship, with an
R2 of 0.998. Since for Newtonian ﬂuids, the consistency coefﬁcient and the viscosity are
identical, the viscosity may be expressed as:

‘°"°)

p = 2.73x10'1‘e(—’— (7.33)
A hydraulic diameter of 0.98 cm was calculated for the mixed screws (combination

of single lead screws and 90° forwarding paddles) by:

4 2 V,”-
2AM

 

D, = (7.34)

150

.aPSm Eoo Lea eLBEeaEeL emce>E msmee> 35085 MFR eSEa

 

TX .._.\_.
430.0 2.53 «30.0 230.0 onmod 88.0
P L L _ b O
S
3
r O
S.
. ,m.
,T. \m
. .o
I: he
4
Eomaocaxfilmnbm n v. .I

 

oo—

151

From extrusion data, Po and Re were calculated using equations 7.23 to 7.26, and are
summarized in Table 7 .10.

 

Table 7.10. Power and Reynolds numbers for corn syrup.
RPM % Torque Pressure Avg. Temp, m, kg/hr Pox10'7 Rex103
°C

 

 

drop, kPa
60 14.0 275.8 19.7 7.0 4.0 1.3
100 14.5 275.8 22.1 9.4 1.0 3.1
150 15.5 206.8 23.2 9.6 0.7 5.3
200 15.5 137.9 ' 24.6 9.9 0.4 8.3

 

 

 

 

 

 

 

 

 

A plot of log Po against log Re is given in Figure 7.14, and yields the relationship
(R2 = 1.00):

log Po = 3.96 - 1.26log Re (7.35)

This result is similar to that reported by Mohamed (1988) who used a

homogeneous screw proﬁle. The apparent viscosity was calculated by Eq. 7.27. Since
the calculated values may not represent the true viscosities, each is referred to as a
viscosity index (11,-) instead. It is a relative value that describes how the viscosity changes

during extrusion.

152

.3530? coEotsezv

asexm See .50 LenEsc mEocxem m:m..e> LenEzc Lezoa 49K 830$

LenEzc m20c>em

00.0

 

lllll

I

Illl

 

em 8:34 team u ea 3 .l

ITTjj Tﬁ

T

II|7

 

 

00+m04

\|0+m_0.—

00+m0é

Jeqwnu JGMOd

153

7.5.5 Effect of rheology on extent of reaction

Using extrusion data (Appendix 8) with Eqs. 7.23-7.27 and Eq. 7.35, viscosity
indices of extrudates at various conditions were estimated. The viscosity indices were
normalized by the viscosity index of a base extrudate. The results are shown in Table

7.11 for extrusion runs 1-12 and Table 7.12 for runs 13-20.

 

 

 

 

 

 

Talble 7.11. Normalized viscosity indices of starch extrudates from extrusion runs
- 2.
Run Avg. temp.,°C % Reducing sugar % DE (s.v.) :—:
no. (s.v.) (s.v.)
1 96.9 (3.6) 3.32 (.13) 1.94 (. 13) 0. 85
5 96.3 (3.1) 5.28 (.16) 2.54 (.15) 0.81
9 95.6 (2.3) 6.62 (.35) 3. 51 (.75) 0.72
2 96.3 (3.3) 3.61 (.33) 1.96 (.20) 0.67
6 96.8 (3.1) 5.65 (.62) 2. 66 (.31) 0.58
10 96.3 (2.7) 6.71 (.26) 3. 54 (. 51) 0.62
3 96.7 (3.3) 6.32 (.64) 3. 30 (.45) 0.71
7 96.4 (3.1) 7.98 (.52) 3. 37 (.18) 0.75
11 95.8 (2.4) 8.43 (.71) 4. 40 (. 35) 0.68
4 96.7 (3.6) 6.32 (.83) 3. 03 (.42) 0.54
8 96.6 (3.2) 8.67 (.42) 3. 52 (.17) 0.50
12 96.7 (2.3) 8.43 (.34) 4.12 (.48) 0.54

 

 

 

 

 

 

 

154

 

 

 

 

 

 

 

 

 

 

Table 7.12. Normalized viscosity indices of starch extrudates from extrusion runs
Run Avg. temp.,°C % reducing sugar % DE (s.v.) X If
no. (s.v.) (s.v.)

13 91.6 (2.9) 8.52(0.80) 3.28(0.26) 0.12 0.95
14 91.3 (2.9) 9.98(0.8 8) 3.43(0.44) 0.11 0.84
15 91.3 (2.4) 12.28(1.10) 4.69(0.76) 0.21 0.68
16 90.9 (2.8) 15.12(0.90) 6.27(0.89) 0.29 0.56
17 89.3 (3.1) 10.44(1.00) 2.75(0.56) 0.18 0.97
18 89.5 (3.7) 17.71(l.32) 5.88(0.42) 0.23 0.72
19 90.2 (2.9) 15.66(0.79) 4.44(0.6l) 0.23 0.84
20 92.0 (2.6) 20.36(1.28) 7 .05(0.73) 0.27 0.52

 

 

The results show that the normalized viscosity index decreases with increased
extent of starch hydrolysis, measured in terms of reducing sugars and amount of
un-reacted starch (Table 7.12). However, the rate of viscosity reduction was not directly
proportional to the extent of hydrolysis because of the varying molecular weights of

products formed.

7.5.6 Dispersion model for predicting starch hydrolysis in reactive
extrusion

Since RTD measurement was performed during starch hydrolysis in the extruder, it
was assumed that the dispersion number of the dye used was the same as that of the
starch, even though the dye is a non-reacting tracer, and the starch was fed continuously
while the dye was not. The dispersion equation with a reaction term included was used,
incorporating the appropriate boundary conditions. If the starch is subjected to the same

boundary conditions as the dye, the dispersion equation could be solved in terms of starch

155

concentrations, which depended on the reaction kinetics of the enzyme.

The reaction kinetics of or —amy1ase in Chapter 4 were determined from data

obtained from runs 13-20, and the results showed that the enzyme reaction could be
characterized by a modiﬁed first order reaction with a rate constant of 0.033 min". The
kinetics obtained were used to predict the extent of reaction for extrusion runs 13-20 and
compared with the analytical solution from the axial dispersion model (Eq. 7.13). It is
worth noting that thermal inactivation of the enzyme was insigniﬁcant because the mean
product temperatures of 89-92°C (Table 7.10) were below the maximum tolerable
temperature of the enzyme (105°C); and also the residence time in the extruder was
relatively short (11 minutes or less). Shear inactivation of the enzyme was assumed to be
small and was neglected, due to the shielding effect of the enzyme by the starch
molecules, in a manner similar to that of soy polysaccharide (Chapter 5).

The fractional conversion was calculated by Eqs. 7.13 and 7.15. The predicted
proﬁles of fractional conversion for extrusion runs 13-16 and 17-20 are shown in Figures
7.15 and 7.16, respectively. It was apparent that slope of the proﬁle increased with
increased length of reactor. Therefore, the rate of fractional conversion increases with
increased length of the reactor, while the dispersion number decreases. This is true for
runs 13-16 (Figure 7.15), but not for runs 17-20, where there is no clear trend (Figure
7 .16).

156

.mwlmé maceEteaxe E0: eoeooee
”.0 £96. newsoctoc mzmue> co_m._e>coo _ocozoo: beyoﬁeea 0; e50:

Leyooee ao 50:2 03:05.52

 

 

0; 0.0 0.0 .v.0 N10 0.0
— . b p _ . h L _ L 0.0
.....
..... \\\\
rrrrrr \V\
.\.\ . _
\\\\\ I— 0 H
\xx\ 0
\\\\ m.
\
\\\\ m
\ 1N0
\\\\ m
A
9
w
w.
10.0
Am: 980 :5 ~85 n ._ l
0: 930 :5 9mm: n ._ ..I
0; 93 :5 QB: n a 3.
Am... axev E0 7450 H 4 ----

[4.0

.0Nlhw maceEteaxe E0: .6308
DB 50:2 UeN__oE:o: mamLe> :o_m._e>:oo 6:00.09: 08.2025 0_.\. 8:07..

5802. .6 50:2 veNzoEBz

No N0 0.0

0.— 0.0 020

_ e a e

 

ﬂ
1 J._
J
D
m.
m.
U
.ou
3
O
U
A
9
m
w.
1.0

8m 980 :8 seem n a I

A2 92 :8 0.38 n .. .i

a: 92 :5 new? n ._ -r

A: 980 :5 92.: u ._

 

Lao

158

The observed and predicted fractional conversion (X) of extrusion runs 13—20 are

presented in Table 7.13.

 

 

 

 

Table 7.13. Observed and predicted fractional conversion (extrusion runs 13-20).

Run %- x 103 7, min Observed X Predicted X
no.

13 51.8 2.4 0.12 0.07

14 34.0 5.2 0.11 0.14

15 12.1 8.1 0.21 0.22

16 7 .3 1 1.3 0.29 0.30

17 40.5 4.6 0.18 0.14

18 14.0 5.7 0.23 0.17

19 10.9 10.3 0.23 0.29

20 11.5 10.4 0.27 0.29

 

 

 

 

 

 

 

There is generally good agreement between the predictions and experimental data,
especially at long residence times. It is possible that the large variations at short
residence times were due to lack of complete mixing. Using regression analysis, it was
found that the predicted and the experimental data have good agreement (R2 = 0.96, at
P<.05). The experimental results from runs 13-20 were super-imposed and are plotted
against the predicted proﬁles of runs 16 and 20 in Figure 7.17.

Because the reaction was characterized as ﬁrst order, the effect of dispersion was
not signiﬁcant. For example, at a constant residence time, the outlet fractional
conversion did not change signiﬁcantly when the dispersion number increased ten-fold.
On the whole, the extent of liquefaction depended signiﬁcantly on the residence time, as

expected.

 

159

0N1? BceEteaxe E0: Loyooe: 00 50:2
0eN__0E._o: m:m..e> :o_m:e>:oo 6:058: 0e>..emno 0:0 0ee.o_0e:a CK e59“.

50:2 :e035xe 0eN__0E._oz

 

0; m0 ed 4.0 Nd 0.0
_ p _ _ _ h _ _ _ _ 0.0
313 axe o
omit axe a x
m: 80283 .i \
0N 030.025 I \\\
id 3
J
D
m.
m.
U
0..
1N0 0
O
U
A
8
m.
w
10.0

 

 

Lao

160

7.6 Conclusions

A dispersion model was used to characterize the ﬂow behavior of a twin screw
extruder during starch hydrolysis by thermostable a -amylase. The ﬂow behavior was
determined by an impulse-response experiment, using erythrosine as a tracer. The
dispersion number was calculated from RTD data, using the ﬁrst and second moments.
The extruder was characterized as a tube, and a closed system. This characterization as a
closed system is not in the thermodynamic sense. Analysis showed good agreement
between the concentration proﬁle of the tracer calculated from the model of a closed
system and the one obtained from experiment (P<.05).

The experimental results show that the dispersion number depends on the screw
rpm and length of the reactor. Therefore, it is also a function of the mean velocity. Also,
the dispersion number depends on the dispersion coefﬁcient, which is a function of
molecular diffusion, velocity and geometry of the extruder. The effects of interactions
between these parameters complicates the analysis and may make the dispersion model
difﬁcult to apply to enzymatic starch liquefaction.

The above complication occurs when the effect of each parameter is considered
separately. However, if all effects are lumped in terms of the dispersion number, several
trends are easily observed. For example, it was apparent that the dispersion number
decreases with increased length of reactor, and that at long residence times, the extent of
reaction increases while the dispersion number decreases.

The dispersion number obtained from RTD data was assumed to be the same as
that of the extrudate. Therefore, it was used to predict the extent of liquefaction. The

enzymatic reaction was described by a modiﬁed ﬁrst order reaction, allowing the

161

dispersion equation to be solved analytically. The analytical solution gave results
comparable to experimental data, especially at low dispersion numbers (long residence
times). This is in agreement with what has been reported in the literature.

As expected, increases in enzyme levels resulted in increased extents of starch
liquefaction. At an enzyme activity of 45 units/g of starch, the maximum fractional
conversion was 0.3 at a residence time of 11 minutes. Because enzymes are expensive,
increasing enzyme dosage offers limited prospects. Since under the extrusion conditions
used, thermal and shear inactivation was negligible, liquefaction should be continued in a
batch system, in a longer extruder, or in multi-passage systems. Because the twin screw
extruder is very ﬂexible, other conditions may be used to increase the extent of starch
conversion. Additional study in this area would be helpful.

Theoretically, the axial dispersion model is valid for a system with a constant
dispersion number. Results in this study indicate that the model can also be used under
conditions of variable dispersion numbers if the reaction is linear. It would be of interest
to examine the effect of varying dispersion numbers on the extent of reaction under
unsteady state conditions. The result will be useful and may explain process instabilities
when the dispersion number changes along the extruder. It is also recommended that the
dispersion equation be studied under unsteady state conditions. The solution of the
unsteady state equation is likely complex; it is an eigenvalue problem and may require

numerical methods.

7.7 Nomenclature

A, Cross-sectional area, cm2
Aw, Wetted area of screw i, cm2

a Redness value, from Chroma Meter

ﬁrrxwrntn

01"

5.

Pe
Po

Re, Re’

162

Slope (Eq. 7 .31), dimensionless
Concentration of substrate at time = 0, g/g

Concentration of substrate at time t, g/g

Modiﬁed starch concentration, g/ g

Normalized starch concentration, dimensionless

Tube diameter, cm

Effective axial dispersion coefﬁcient, cmZ/s

Hydraulic diameter, cm

Molecular diffusion coefﬁcient, cm2/s

Dextrose equivalent, dimensionless

Viscous dissipation of mechanical energy, W
F-function, dimensionless

Rate constant, min'l

Consistency coefﬁcient, Pa.sn

Axial length of reactor, cm

Equivalent length of screws used during RTD measurement, cm
Equivalent length of the screws in the reaction zone, cm
Throughput, g/min

Flow behavior index, dimensionless

Screw speed, rps

Peclet number, dimensionless

Power number, dimensionless

Power input, W

Reynolds number, dimensionless

”I

"1

al

Tl.

Tl:

163

Rate of starch consumption, g/g/min
Residence time, min

Mean residence time from RTD data, min
Corrected mean residence time of the reaction, min

Mean velocity, cm/s

Wetted volume of screw i, cm3
Axial length of the reaction vessel or extruder, cm
Fractional conversion, dimensionless
Variable, Eq. 7.24
Normalized length, dimensionless
Greek symbols

Non-Newtonian apparent viscosity, Pa. 8“
Non-Newtonian viscosity index, Pa. 5'1

Newtonian viscosity, Pa.s
Variable, Eq. 7.24

Density of extrudate , g/ml
Variance of the RTD curve, min2

Dimensionless residence time

Variable, Eq. 7.14

164

7.8 References

Aris, R. 1956. On the dispersion of a solute in a ﬂuid ﬂowing through a tube. Proc. Roy.
Soc. London. Series A. 235 :67.

Bemfeld, P. 1955. Amylase, a, [3. Methods Enzymol. 1:149.

Bischoff, KB. 1968. Accuracy of the axial dispersion model for chemical reactors.
AIChE J. l4(5):820.

Bischoff, KB. and Levenspiel, O. 1962. Fluid dispersion - generalization and comparison
of mathematical models. Chem. Eng. Sci. 17:245.

Booras, GS. and Krantz, W.B. 1976. Dispersion in the laminar ﬂow of power-law ﬂuids.
Ind. Eng. Chem. Fundam. 15(4):249.

Bonnie, D. 1988. Modeling of the ﬂow pattern in a twin-screw extruder through
residence-time distribution experiments. J. Food Eng. 7:223.

Butt, J.B. 1980. Reaction Kinetics and Ream Design. Prentice-Hall, New York.

Carnahan, B., Luther, HA. and Wilkes, 1.0. 1969. Applied Numerical Methods. John
Wiley & Sons, New York.

Danckwerts, P.V. 1953. Continuous ﬂow systems, distribution of residence times. Chem.
Eng. Sci. 2(1):1.

Dygert, S., Li, L.H., Florida, D. and Thoma, J.A. 1965. Determination of reducing sugar
with improved precision. Anal. Biochem. 13:367.

Froment, CF. and Bischoff, KB. 1979. Chemical m ADAMS and 1231811. John
Wiley & Sons, New York.

Gibilaro, LG. 1978. On the residence time distribution for systems with open
boundaries. Chem. Eng. Sci. 33:487.

Gill, W.N. 1967. A note on the solution of transient dispersion problems. Proc. Roy. Soc.
London. Series A. 298:335.

165

Gill, W.N. and Sankarasubramanian, R. 1970. Exact analysis of unsteady convective
diffusion. Proc. Roy. Soc. London. Series A. 316:341.

Komolprasert, V. and Ofoli, R.Y. 1989. A dispersion model of enzyme-catalyzed
reaction-kinetics in a twin screw extruder. Proceeding of the International
Conference on Engineering and Food (ICEF), Cologne, W. Germany, June 1989.

Kreft, A. and Zuber, A. 197 8. On the physical meaning of the dispersion equation and its
solutions for different initial and boundary conditions. Chem. Eng. Sci. 33:1471.

Levenspiel, O. 1972. Chemieal Reamer) Engimzing. 2nd edition. John Wiley & Sons,
New York.

Levenspiel, O. and Smith, WK. 1957. Notes on the diffusion - type model for the
longitudinal mixing of ﬂuids in ﬂow. Chem. Eng. Sci. 6:227.

Linko, P. and Linko, Y.Y. 1983. Extrusion cooking and bioconversions. J. Food Eng.
2:243.

Linko, P., Hakulin, S. and Linko, Y.Y. 1983. Extrusion cooking of barley starch for the
production of glucose syrup and ethanol. J. Cereal Science. 1:2745

Linko, P., Hakulin, S. and Linko, Y.Y. 1984. HTST - extrusion cooking in ethanol
production from starchy materials. Enzyme Microb. TechnoL 6:457.

Michelsen, ML. and Ostergaard, K. 1970. The use of residence time distribution data for
estimation of parameters in the axial dispersion model. Chem. Eng. Sci. 25:583.

Michelsen, ML. 1972. A least - squares method for residence time distribution analysis.
Chem. Eng. J. 4:171.

Mohamed, 1.0. 1988. Modeling shear rate and heat transfer in a twin screw co-rotating
food extruder. Ph.D. Dissertation. Dept. Agri. Engr. Michigan State University.
East Lansing, MI.

166

N auman, EB. 1981. Residence time distributions and micromixing. Chem. Eng.
Commun. 8:53.

Ofoli, R.Y., Komolprasert, V., Saha, BC. and Berglund, K.A. 1989. Production of
maltose by reactive extrusion of carbohydrates. Lebens. Wissen. und Technol. In
print.

Parimi, K. and Harris, T.R. Identiﬁcation of residence time models by reacting tracer
experiments. Can. J. Chem. Eng. 53:175.

Pham, QT. and Keey, RB. 1977. The use of graphical transfer-function methods for
residence time distribution studies. Chem. Eng. Sci. 32:786.

Reinikainen, P., Suortti, T., Olkku, J ., Malkki, Y. and Linko, P. 1986. Extrusion cooking
in enzymatic liquefaction of wheat starch. Starch. 1:20.

Saha, B.C., Shen, G.J. and Zeikus, J .G. 1987. Behavior of a novel thermostable
B -amylase on raw starch. Enzyme Microb. Technol. 9:598.

Sankarasubramanian, R. and Gill, W.N. 1972. Dispersion from a prescribed
concentration distribution in time variable ﬂow. Proc. Roy. Soc. London. Series A.
329:479.

Subramanian, RS. and Gill, W.N. 1976. Unsteady convective diffusion in
non-Newtonian ﬂows. Can. J. Chem. Eng. 54:121.

Subramanian, RS. 1977. On generalized dispersion theory. Chem. Eng. Sci. 32:788.

Taylor, G. 1953. Dispersion of soluble matter in solvent ﬂowing slowly through a tube.
Proc. Roy. Soc. London. Series A. 219:186.

Van der Laan, ET. 1958. Notes on the diffusion - type model for the longitudinal mixing
in ﬂow. Chem. Eng. Sci. 7:187.

167

8 CONCLUSIONS

An MPF 50D Baker-Perkins twin screw food extruder was used as a bioreactor to
carry out liquefaction of pre-gelatinized corn starch by thermostable a -amylase. The
reaction occurred in a fully-ﬁlled zone conﬁgured with 90° paddles. A one-dimensional
axial dispersion model was used to predict the extent of starch liquefaction. The model
was evaluated with several extrusion data obtained during studies designed to address the
following:

1. Starch hydrolysis kinetics of at —amylase;

2. Effect of shear on enzyme activity;

3. Effect of rheology on dispersion of ﬂuids in the extruder; and

4. Effect of moisture content, screw rpm, enzyme level and dispersion number on
the extent of starch liquefaction.

Based on an initial rate experiment at low starch concentrations (l-8%, w/w), the
kinetics of a -amylase was found to follow the classical Michaelis-Menten model. A
continuous rate study at a high starch concentration of 40% showed that the enzyme
kinetics may be described by a modiﬁed ﬁrst order reaction equation.

Effect of shear on enzyme activity was performed in both a concentric cylinder
viscometer and the extruder. In the viscometer, strain history had the most signiﬁcant
effect on loss of activity. However, loss of activity in the extruder depended on a
combination of strain history, shear rate and speciﬁc energy consumption (SEC). About
a 20% loss of activity was observed. This loss of activity was less than expected,
possibly due to the shielding effect of SPS during extrusion. ,

Effect of rheological behavior on the dispersion coefﬁcient (D,) was explained by a
model developed on the basis of the Taylor-Axis model. Several extrusion runs were

performed using corn syrup, 30% SP8 in honey and 2.5% methocel in water. Corn syrup

168

is a Newtonian ﬂuid (n=1), while the SPS-honey (n=.416) and the methocel-water (n=.3)
systems are pseudoplastic ﬂuids. The methocel solution had the largest dispersion
coefﬁcient, followed by corn syrup; SPS-honey had the smallest dispersion. The mean
velocity was dominant and governed the degree of dispersion. The viscoelastic
properties of the ﬂuids did not show a clear effect on the dispersion coefficient.

The extent of reaction, measured by percent DE and percent reducing sugar, was
signiﬁcantly affected by moisture content, enzyme level and residence time; the effect of
screw speeds (30-60 rpm) was not apparent. Dispersion numbers were calculated from
RTD data and increased with screw rpm but decreased with length of the reaction zone.

Several extrusion data were super-imposed to obtain a proﬁle of the extent of
conversion. Since the enzymatic reaction was ﬁrst order, an analytical solution was
obtained and compared to the experimental data; there was good agreement between the
predictions and the experimental data at P<.05. In general, the axial dispersion model is

suitable for the modeling of reactions involving starch hydrolysis.

169

9 RECOMMENDATIONS FOR FURTHER RESEARCH

Further studies are required to extend the understanding of enzymatic starch
hydrolysis in twin screw extruders. These include:

1. Initial rate studies at various high starch concentrations in the extruder.

2. Determination of the effect of temperature on the kinetics of a -amylase.

3. Effect of shear on enzyme activity under various extrusion conditions.

4. Effect of viscoelastic properties on the dispersion coefficient.

5. Evaluation of the validity of the axial dispersion model with other screw
conﬁgurations.

6. Assessment of the axial dispersion model as a tool for predicting the extent of
other reactions, such as sacchariﬁcation by B —amylase.

7. Effect of the dispersion number on the stability of the extrusion processes.

10 APPENDICES

170

 

Appendix 1. Shear deactivation data from Haake viscometer.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Strain history x 104 % Residual enzyme activity
2.80 100.0
96.6
4.20 100.0
100.0
5.60 100.0
98.4
8.40 87.6
93.6
9.80 89.7
95.7
86.7
100.0
1 1.20 93.2
93.2
14.00 92.6
100.0
14.70 97.1
93.5
15.40 97.2
86.3
19.60 99.5
99.3
96.9
79.7
22.40 93.2
86.7
25.20 96.6
95.1
30.80 100.0
85.4
34.30 93.0
91.2
39.20 82.8
86.7
49.00 83.9

 

82.8

 

 

Appendix 1. (Cont’d).

171

 

 

 

 

 

 

 

53.91 81.0
74.0

58.81 86.3
88.6

72.81 82.5
82.0

84.01 75.9
76.6

92.41 70.2
66.8

126.02 65.1
.8

 

 

 

 

172

 

Appendix 2. Analysis of variance (SAS program) of effect of strain history on %
residual enzyme activity from Haake viscometer data.

 

 

Source DF Sum of Mean F value Pr > F
Square Square
Model
strain 21 3948.2 188.0 7.31 0.0001
Error 26 668.3 25.7
Corrected 47 4616.5
Total
R square C.V. Root MSE Mean % activity

0.8552 5.702 5.070 88.92

 

 

173

 

Appendix 3. Shear deactivation data from extrusion runs.

 

RPM Residual activity x 103, units

 

60 27.6
28.1
27.6
29.5
32.3

 

100 33.3
31.1
31.8
30.3
26.4

 

150 30.3
28.4
28.3
23.7
25.6

 

200 29.1
26.2
27.8
29.4
28.9

 

250 20.7
30.6
29.5
21.5
25 .7

 

300 28.1
30.5
29.3
27.1
27.6

 

 

400 25.2
31.0
26.1
28.9
29.1

 

 

 

174

 

 

 

 

Appendix 4. Analysis of variance. (SAS program) of effect of strain history on
enzyme activity during extrusion.
Source DF Sum of Mean P value Pr > F
Square Square
Model
strain 6 640.85 106.81 1.76 0.1436
Error 28 1697.44 60.62
Corrected 34 2338.29
Total
R square C.V. Root MSE Mean % activity
0.2741 9.21 106.81 84.5

 

 

' The same result was obtained for shear rate and SEC.

175

 

Appendix 5. Analysis of variance (SAS program) of effect of strain history, shear rate
and SEC on enzyme activity during extrusion.

 

 

 

Source DF Sum of Mean F value Pr > F
Square Square

Model 18 1922.56 106.81 4.11 0.0110
strain 6 640.85 106.81 4.11 0.0110
shear rate 6 640.85 106.81 4.11 0.0110
SEC 6 640.85 106.81 4.11 0.0110
Error 16 415.74 25.98

Corrected 34 2338.29

Total

R square C.V. Root MSE Mean % activity

0.8222 6.03 5.10 84.5

 

 

176

 

Appendix 6. Extrusion data for corn syrup, 7% SPS in honey and 2.5% Methocel.

 

 

 

 

 

 

 

Material RPM % Torque Pressure drOp Flow rate, Avg. temp.,
(PSi) kg/hr C (s.v.)

Corn syrup 100 14.0 35 9.37 23.5(0.8)
15.0 22.1(0.9)
150 15.5 30 9.63 23.2(1.l)
7% SPS in 100 14.0 30 9.23 26.6(0.9)
honey 23.7(0.6)
150 15.5 30 9.62 24.0(0.7)
23.8(2.0)
2.5% 150 7.0 5 8.40 18.3(0.7)
Methocel 9.18 22.3(0.4)

 

 

 

 

 

 

 

177

Appendix 7. Paddle conﬁgurations in the twin screw extruder.

 

 

 

45'paddles

JM
AI

60°paaales 90 paddles

 

 

 

 

/’“‘\

 

 

%%

 

 

 

Cross-sections of the extruder barrel with different forwarding
paddle configurations.

178

 

Appendix 8. Extrusion data on starch liquefaction (extrusion runs 1-20).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Run RPM % Torque Pressure Flow rate Avg. temp., % MC
no. drop, psi (kg/hr) C (out)
1 30 9 50 10.98 96.9 55.4
2 60 ll 60 11.10 96.3 55.4
3 30 17 70 10.74 96.7 49.4
4 60 20 70 10.98 96.7 49.4
5 30 10 30 10.08 96.3 57.2
6 60 10 30 10.32 96.8 57.2
7 30 20 40 10.38 96.4 46.0
8 60 20 40 10.20 96.6 46.1
9 30 8 40 11.10 95.6 60.9
10 60 12 40 10.98 96.3 60.9
1 1 30 27 40 10.74 95 .8 46.5
12 60 25 50 10.32 96.7 45.3
13 60 14.5 20 10.05 91.6 56.0
14 60 12.5 20 10.11 91.3 56.0
15 60 9.5 20 10.08 91.3 56.8
16 60 7.5 20 9.99 90.9 57.3
17 60 1 1.0 30 9.90 89.3 56.0
18 60 8.5 20 10.68 89.5 55.7
19 60 12.0 30 9.60 90.2 57.1
20 60 7.5 20 10.59 92.0 53.0

 

 

 

 

 

 

 

 

179

 

Appendix 9. Analysis of variance (SAS program) of mean residence time (RTD, ?) for
extrusion runs 1-12.

 

 

 

Source DF Sum of Mean F value Pr > F
Square Square

VariablezRTD
Model 5 19.743 3.949 192.33 0.0001
Lo 2 15.095 0.021 367.63 0.0001
% MC 1 0.295 0.295 14.35 0.0091
Enz. dose 1 4.107 4.106 200.03 0.0001
RPM 1 0.247 0.247 12.01 0.0134
Error 6 0.123 0.008
Corrected Total 11 19.866

R square C.V. Root MSE Mean RTD

0.9938 2.304 0.1433 6.22

 

 

180

 

Appendix 10. Analysis of variance (SAS program) of dispersion number (D1, 724) for
extrusion runs 1-12.

 

 

 

Source DF Sum of Mean F value Pr > F
Square Square

VariablezDI
Model 5 0.0011839 0.0002368 5.93 0.0256
L, 2 0.0002397 0.0005176 12.96 0.0066
% MC 1 0.0000008 0.0000008 0.02 0.8955
Enz. dose 1 0.0000132 0.0000132 0.33 0.5859
RPM 1 0.0001347 0.0001347 3.37 0.1 160
Error 6 0.0002397 0.0000399
Corrected Total 11 0.0014236

R square C.V. Root MSE Mean DI

0.8316 27.97 0.0063 0.0226

 

 

 

181

 

Appendix 11. Analysis of variance (SAS program) of dispersion number (D1, :1) for

extrusion runs 13-20.

 

 

 

Source DF Sum of Mean P value Pr > F
Square Square
Variablele
Model
Method 1 0.00010 0.00010 0.32 0.5925
Error 6 0.00188 0.00031
Corrected Total 7 0.00198
R square C.V. Root MSE Mean DI
0.0505 77.8 0.0177 0.0228

 

 

182

 

 

Appendix 12. Percent reducing sugar and DE from extrusion runs 1—20.

 

 

 

 

 

 

 

E 31 88 57886 3 8 37 1 168 62 935126 638970
D 0.0..M.m33 n73233 m9 $6.0. w. 7.9 .8. 67.420333 36.34.33 1.33237.
% 221211 211211 333332 322223 222222 222223 333333
me
.C 82 96 05839 84 2 6 3 38 516891 51383 286579
um. AJNM33 £23933 2.3%...” .2 Mﬂjwrs 339333 M73234. 4.5.31.6.6.
%
M 0 0 0 0
W 3 w 3 w 3 w 3
LR 3 3 0 0 3 3 O
n.m 2 2 3 3 2 2 3
B
C O
0 0

M m w 5 5 m w 5
am
L. M M M M 0.3 0.3 ﬂ

6 6 6 6 H H N
we.
Rn 1 2 3 4 5 6 7

 

 

 

 

 

 

 

 

 

183

Appendix 12. (Cont’d).

 

 

<uAu25 aaoo <5 0, arnanaaqloo Aqooqlnuoznu 24000: n20;
%mﬂsmwﬁ. ﬂxmmwﬁ %m233 . 3.3M.4.m.6. 0.20.233 94.5242 329 .23
333333 434223 333433 444444 444443 233333 332343
6181 5 8 282923 53 338 9 1859
mmwwmm mﬁﬁnmm 003.3239 0&6.st . 2924.30. 29 .1.4.3 3 30.4.3
80000988 666766 666766 887978 888888 788898 99.8me

 

3O

30

 

30

23

23

30

30

45

45

 

50

60

60

50

50

60

 

1152

1630

1630

1630

1630

674

1152

 

 

8

 

9

 

10

 

ll

 

12

 

l3

 

l4

 

 

 

Appendix 12. (Cont’d).

184

 

15

1630

45

1172
1053
£384
1L96
1K359
l31¥1

 

16

2107

45

bt84
L334
15at3
htSZ
11180
iht81

 

17

1143

45

1L87
$193
1(L10

1L57
1L19

 

18

1620

45

1£L54
iﬂl34
Y131
IHi95
2H359
lti53

#UJOO womb-00H H\O\OOONN w-AQHUIOx

 

19

2098

45

13183
Iﬂi38
11183
ht45
15322
161KB

LII-hv-‘NH 'N'g'ooqun— hNOsootI-oo meumoo unouetoooto

 

20

 

2681

 

 

45

 

 

28143
18148
22J3
iZL28
1893
18£Il

 

Nov-

eesﬂss shareware 0.0.0.999. P’PPNP?‘ 9.4993»?- Names»?

 

 

185

 

Appendix 13. Analysis of variance (S AS program) of effect of moisture content,
screw rpm and enzyme level on % reducing sugar (RS) and % DE for extrusion runs

 

 

 

 

1-12.
Source DF Sum of Mean P value Pr > F
Square Square
VariablezRS
Model 3 224.539 74.846 9355.61 0.0001
MC 1 111.751 111.751 13968.62 0.0001
Enz. dose 1 111.751 111.751 13968.62 0.0001
RPM 1 1.037 1.037 129.60 0.0001
Error 68 0.544 0.008
Corrected Total 71 225.083
R square C.V. Root MSE Mean % RS
0.9976 1.3888 0.0894 6.44
Variable:DE
Model 3 31.330 10.443 42.72 0.0001
MC 1 15.652 15.652 64.03 0.0001
Enz. dose 1 15 .652 15.652 64.03 0.0001
RPM 1 0.025 0.025 0.10 0.7486
Error 68 16.624 0.224
Corrected Total 71 47.953
R square C.V. Root MSE Mean % DE
0.6533 15.66 0.4944 3.15

 

 

 

 

 

186

 

Appendix 14. Analysis of variance (SAS program) of effect of method of enzyme
addition on % reducing sugar (RS) and % DE for extrusion runs 13-20.

 

 

 

 

Source DF Sum of Mean F value Pr > F
Square Square
VariablezRS
Model
Method 1 41.68 41.68 3.21 0.1234
Error 6 77.93 1 1.07
Corrected Total 7 119.61
R square C.V. Root MSE Mean % RS
0.3485 26.19 3.604 13.76
Variable:DE
Model
Method 1 0.7 6 0.76 0.28 0.6150
Error 6 16.15 2.69
Corrected Total 7 16.90
R square C V Root MSE Mean DI

0.0448 34.74 1.640 4.72