(I.

>' h‘. ivltlg ;‘, .

g Q -’. A hay in. y
t _ U
L a w a 3.5"» ~'l i

1; T. O I o
. - - «- .flk’lmf a. i- “‘ «-
-‘---‘-—‘ def-31 Q “Le—LA;

 

 

U 3'4 :‘p’s~a.’.iq',
'-6- d ‘0- ”-51.1!

 

This is to certify that the

dissertation entitled

THE SEPARATION MECHANISM FOR THE DETERGENT
FRACTIONATION OF BEEF TALLON AND ITS
RELATION TO PROCESS VARIABLES
presentedby

David Allen Glassner

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Chemical Engineering

 

 

BL; CL‘W

Major professor

Date May 14, 1986

 

MS U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

 

MSU

LIBRARIES
m

w

 

 

RETURNING MATERIALS:
Place in book drop to
remove this checkout from
your record. FINES will
be charged if book is
returned after the date
stamped below.

 

 

 

 

 

THE SEPARATION MECHANISM FOR THE DETERGENT FRACTIONATION
0F BEEF TALLOW AND ITS RELATION TO PROCESS VARIABLES

By

David Allen Glassner

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Chemical Engineering

1986

ABSTRACT

THE SEPARATION MECHANISM AND ITS RELATION TO PROCESS

VARIABLES FOR THE DETERGENT FRACTIONATION OF BEEF TALLOW

By

David Allen Glassner

Aqueous or detergent fractionation is done by partially
crystallizing the fat and recovering the crystals by wetting them
into an aqueous solution containing a surfactant and an
electrolyte. The goal of the fractionation process is to obtain
fat fractions having more value than the original fat. The aqueous
fractionation of tallow is studied to determine the separation
mechanism and the relationship of important processing variables to

the mechanism.

Detergency phenomena was found to explain the wetting of crystals
from the oil to the aqueous phase, hence the term detergent
fractionation. Key process variables were crystallization
conditions, surfactant (sodium dodecyl sulfate) concentration,
electrolyte (sodium citrate) concentration, the weight ratio of
detergent solution to partially crystallized tallow and the
viscosity of the dispersion. These variables were studied for the

I

separation of beef tallow at 40 °C.

It was found that crystal surface area was prOportional to the
weight of surfactant required. The quantity of surfactant, not the
concentration of surfactant was important. Most of the surfactant
associated with the crystal surface. Too much surfactant caused
emulsified olein product and too little surfactant resulted in an

incomplete separation.

The electrolyte helped create an ionic solution which oriented the
surfactant and allowed the crystals to be wetted into the aqueous
phase. Low electrolyte concentrations resulted in incomplete
separations. The optimum concentration was found over a wide range
with a tendency for the formation of more emulsified olein at

higher concentrations of electrolyte.

For a given crystallization procedure an optimum weight ratio of
detergent solution to tallow existed. Ratios higher than the
optimum resulted in the same separation but required processing a
larger volume of fluid. Ratios lower than the optimum caused

incomplete separations.

The rheology of the detergent solution-partially crystallized
tallow dispersions was investigated with a paddle mixer viscometer.
Apparent viscosities of the dispersions were measured at a rotation
rate of 60 rpm. An apparent viscosity of 0.08 Pa 8 or less was
necessary for maximum olein recovery. The apparent viscosity of
the aqueous solution-olein mixture without any crystals was about

0.015 Pa 3.

ACKNOWNLEDGEMENTS

I have appreciated my association with Dr. Eric A. Grulke. My most
sincere gratitude goes to him for his guidance, patience and, most

importantly, for the professional I was able to observe.

My sincere thanks go to the members of my graduate committee, Dr.
J.I. Gray, Dr. C.A. Petty and Dr. D.K. Anderson. I appreciate the
time they spent in reviewing this dissertation. I further thank
them for comments made and questions raised about my research. I
would also like to express my special appreciation for Dr. Gray's

involvement in my research.

I thank the Amoco Foundation for their financial support during my
research studies. I greatly appreciated their support. I also

thank Dr. J. Steffe for the use of his laboratory and equipment.

Most important to me, I thank my wife, Marcie, for her patience and
support. She has always understood the time that my studies have

.4

taken and always has made them fit into our life.

ii

Table of Contents

List of Figures

List of Tables

Nomenclature

1.0 Introduction

2.0

3.0

1.1

Fractionation Processes

1.2 Detergent Fractionation

1.3 Research Plan

Literature Review

2.

2.

2.

2.

1

2

7

8

Chemistry of Tallow
Fractionation Processes
Aqueous Fractionation

Crystal Morphology

Detergency Theory

Critical Micelle Concentration
Surface Tension Measurement

Dispersion Rheology

Experimental Methods and Equipment

Tallow Sample Preparation
Crystallization Procedures
Separation Procedure
Surface Tension Measurement
Rheology Data Measurement
Crystal Size Measurements

Olein Emulsification Measurement

iii

..... v‘
. ... vii
..... ix
. ... 1
. . 3
.. 3
. 5
.. . 8
.. . 8
... 10
.. . 12
. .. 18
. . 21
. 24
. .. 25
..... 28
..... 31
.. . 31
..... 31
..... 34
... . 35
..... 39
. ... 43
...... 44

4.0 Results and Discussion

4.

.1

.2

.8

.9

Proposed Separation Mechanism
Crystallization Results

Separation Results

Electrolyte Concentration Effect

Surface Area Effect on Separation Results
Surface Coverage by SDS

CMC for the Detergent Solutions

Detergent Solution Effect on the Separation

Torque Measurements

10 Apparent Viscosity of the Dispersions

5.0 Conclusions

6.0 Recommendations

7.0 Appendices

7.

7.

8.0 List

1

2

Appendix A. Raw Data

Appendix B. Sample Calculations
B.1 Table 4.1 Calculations

8.2 Table 4.2 Calculations

of References

iv

45
..... 49
57'
61
..... 61
64
..... 67
..... 69
..... 75
81

..... 86

.... 9O

.10

.11

List of Figures

Triglyceride Chain Packing Diagram. .....

Temperature Profiles for Three
Crystallization Procedures. .....

Fractionation Process Schematic Diagram. .....
Mixer Viscometer Paddle and Cup. .....

Rolling-Up Detergency Mechanism for
Individual Tallow Crystals. .....

Rolling-Up Detergency Mechanism for Crystal
Agglomerates. .....

Photomicrograph of Procedure Two
Crystals, IOOX Magnification. .....

Photomicrograph of Procedure Three
Crystals, 100x Magnification. .....

SEM Microscopy of Fat Crystals, 1,000X
Magnification. .....

SEM Microscopy of Fat Crystals, 700x
Magnification. .....

Separation Results from Three Different
Crystallization Procedures. .....

Olein Yield as a Function of the Electrolyte
Concentration in the Detergent Solution at 40 °C.
Tallow Weight was 170 g per sample; SDS Weight

was 0.45 g per sample; and the Weight Ratio of
Detergent Solution to Tallow was 0.93. .....

Simple Models for Calculating Surface Area
Occupied by an SDS Molecule in Two Different
Orientations. .....

Surface Tension of Two Detergent Solutions
as a Function of SDS Concentration at 40 °C. .....

Correlation of the Olein Yield with the Weight

of SDS Used. Each Curve Represents a Different
Weight Ratio of Detergent Solution to Tallow

Weight. 5.0 Z Sodium Citrate was in the

Detergent Solution. .....

62

4.17

Olein Yield as a Function of the Weight Ratio

of Detergent Solution to Partially Crystallized
Tallow. 0.45 g SDS per sample and 5.0 Z Sodium
Citrate was in the Detergent Solutions. .....

Torque Measured as a Function of Time for a
Dispersion at a Rotation Rate of 120 rpm. .....

Torque Measured as a Function of the
Rotation Rate for a Dispersion. .....

Precision Measurements of the Torque as a

Function of the Rotation Rate of the Paddle

Mixer. Two sets of Data are Plotted to show the
Reproducibility of the Measurements. The Curves
Represent Detergent Solution to Tallow Weight

Ratios of 1.50, 1.14, 0.80 and 0.50 from Low to

High Torque Values. .....

Dispersion Apparent Viscosity Plotted Versus the

Olein Yield. Various Weight Ratios of Detergent
Solution to Partially Crystallized Tallow were

used to Produce the Apparent Viscosity Changes. .....

The Apparent Viscosity of the Dispersion as a

Function of the Percent Solids in the Dispersion.

Four Different Sets of Data Shown; Each

Corresponds to a Separate Partial Crystallization ......

vi

72

76

78

80

82

85

2.1

2.2

2.3

2.4

4.2

4.3

A.2

A.3

A.5

A.6

A.7

A.8

A.9

LIST OF TABLES

Tallow Fatty Acid Composition.

Aqueous Fractionation Process Variable
Recommendations.

X-ray Short Spacings.
Surface Tension Measurement Methods.

Crystallization Procedure Two and Three
Crystal Descriptions.

Calculations of Crystal Surface Area
and Surfactant Coverage.

Standard Deviations for Figure 4.12.

Separation Results from Three Different
Crystallizations.

Olein Yield as a Function of the Electrolyte
Concentration in the Detergent Solution
at 40 °C.

Preliminary Correlation of the Olein Yield
with the Ratio of Detergent Solution to
Tallow Weight.

Calibration Data for Ring Method of Surface
Tension Determination for the Cenco
Tensiometer.

Surface Tension of a Detergent Solution as
a Function of Surfactant Concentration at
40 °C.

Olein Yield as a Function of the Weight
Ratio of Detergent Solution to Partially
Crystallized Tallow.

Torque Measured as a Function of Time for
a Dispersion.

Torque Measured as a Function of the
Rotation Rate.

Precision Measurements of the Torque as a
Function of the Rotational Rate of the
Paddle Mixer.

vii

A.10 Dispersion Apparent Viscosity Plotted
Versus the Olein Yield.

A.11 The Apparent Viscosity of the Dispersion as a
Function of the Percent Solids in the Dispersion.

viii

NOTATION

A, Angstroms (Table 2.3).
B, constant depending on impeller system geometry (Eqn 2.3)-
°C, degrees Celsius (Section 2.1).
D , constant for a particular shape solid (Eqn 4.3).
d, Impeller diameter, m (Eqn 2.1).
g*, acceleration of gravity, cm/s2 (Eqn 3.1).
k', constant for a particular impeller (Eqn 2.5).
K, consistency coefficient, Pa sn (Section 2.8).
L, mean circumference of the ring, cm (Eqn 3.1).
M, torque, N m (Eqn 2.7).
n, flow behavior index (Section 2.8).
Po, Power number (Eqn 2.1).
p, power, N m s (Eqn 2.1).
Re, Reynolds number (Eqn 2.2).
ls, apparent surface tension, dynes/cm (Eqn 3.1).
S, volume fraction solids in fluid (Eqn 4.3).
W, weight placed on ring, g (Eqn 3.1).
fl , fluid density, kg/m (Eqn 2.1).
Ya , shear rate, 3'1 (Eqn 2.5).
7r", reduced viscosity of dispersion (Eqn 4.3).
If', shear stress, Pa (Eqn 2.6).
/ML , fluid viscosity, Pa 5 (Eqn 2.2).
U” , rotational speed of the impeller, rad/s (Eqn 2.1).

7 , apparent viscosity, Pa 3 (Section 2.8).
a

ix

INTRODUCTION

Fractionation of a fat by partial crystallization produces two
fractions with different properties from the original fat. The
fractions have different chemical compositions, and the melting
point range of each fraction is much more narrow than the original
fat. The fractions separated are the liquid and solid formed by a
partial crystallization of the tallow at a specific temperature.
The process is most useful on a fat which exhibits a wide melting
point range, a characteristic of a fat with a wide range of
saturated and unsaturated fatty acids. Palm oil and beef tallow
are fats commonly fractionated. Groundnut oil, partially
hydrogenated soybean oil, coconut oil, rapeseed oil, hog fat, whale
oil, menhaden oil and cod liver oil are some other oils that have

been fractionated (Rek, 1977; Stein and Hartmann, 1957).

Tallow production in the United States for 1983 totaled 6.04
billion pounds. 0f the total produced 1.22 billion pounds was
edible tallow. The edible tallow produced was used for baking or
frying. Tallow accounted for 14 percent of the baking and frying
fat market in the U.S. (USDA ERS, 1984). Fractionation of tallow
at various temperatures can result in fractions which are more
useful and valuable than the original tallow (Luddy et al., 1973;

Taylor et al., 1976).

2
Many investigators attest to the suitability of tallow fractions
for use in food preparation (DeFouw et al., 1981, Holsinger et al.,
1978, Kozempel et al., 1981, Luddy et al., 1973, Taylor et al.,
1976, Ryan, 1985). Suitable tallow fractions can be used in
shortenings, as frying media and as confectionary fats. Tallow
imparts a characteristic taste to fried potatoes and is highly

desired for this use (Ryan, 1982).

The object of a fractionation process is to completely separate the
liquid and solid components of a partially crystallized oil. A
complete liquid-solid separation maximizes the chemical composition
and physical property differences of the two fractions. In the
case of tallow, the crystals are called the stearin and the oil is
called the olein. These terms refer to the preferential
distribution of stearic and oleic acids in the triglycerides of

each fraction.

The problem in any separation of crystals and oil is that the
viscosity of the medium is generally too great to allow a good
separation between the oil and crystals. Filtration processes
require large crystals to be formed so that the crystals can be
recovered from the oil-crystal mixture. A long crystallization
period, 3-4 days, is required to produce the large crystals.
Because of the long crystallization periods and the incomplete
separation, better methods of oil-crystal separation have been

developed.

1.1 Fractionation Processes

 

Three fat fractionation processes have been commercialized: dry,
solvent and detergent (Kreulen, 1976). The dry fractionation
process involves partial crystallization of the oil followed by
vacuum filtration of the partially crystallized oil. The
filtration yields clear oil but results in significant olein
entrainment in the stearin fraction. Solvent fractionation of fats
using hexane and other solvents yields fractions containing
residual solvent. Process equipment must be explosion proof.
Solvent recovery can be a significant portion of total process
costs. In detergent fractionation, the liquid-solid separation is
accomplished by mixing a detergent solution with a partially
crystallized oil, preferentially wetting the crystals into the
aqueous phase. Detergent fractionation has the disadvantages of
emulsified olein product formation, residual electrolyte and
surfactant in the fractions obtained, and additional chemical
costs. The residual surfactant easily can be removed from the
olein and stearin fractions by hot water washings, since the
equilibrium is greatly towards the water phase. Water washing also

would remove any electrolyte present.

1.2 Detergent Fractionation

 

The detergent fractionation of a fat consists of a crystallization

and a separation step. The crystallization step is a fractional

4
crystallization at a specific temperature, resulting in a partially
crystallized oil. A detergent solution containing a surfactant and
an electrolyte is then mixed with the partially crystallized oil.
The detergent solution is at the same temperature as the partially.
crystallized oil. Centrifugation then yields a clear liquid olein
and a stearin-detergent solution layer. The stearin and detergent

solution are separated by heating and recentrifuging.

Many authors have discussed the requirements for the surfactant,
electrolyte and quantity of detergent solution to be used (Stein,
1968; Kreulen, 1976; Rek, 1977; Glassner, 1983; Bussey et al.,
1981). The chemical quantities and concentrations reported are not
consistent. The separation step of the detergent fractionation of
tallow is identified as being crucial to the design of the

fractionation process (Glassner, 1983).

Crystal size and surfactant concentration both affect the formation
of undesireable emulsion in the detergent fractionation of tallow
(Glassner, 1983). Larger crystal size causes less emulsion
formation at the same surfactant concentration. Smaller crystal
sizes resulted in poor separations. Higher surfactant
concentrations, given the same crystal size, were found to result
in more stable emulsion. Low surfactant concentration resulted in
efficient wetting of the crystals into the aqueous phase and poor
separation results. This research attempts to define the

conditions necessary for a successful separation step.

1.3 Research Plans

 

The process variables for the separation step include the
surfactant concentration, electrolyte concentration, aqueous
volume, crystal size of the partially crystallized tallow and
dispersion mixing characteristics. Since the mechanism of the
separation has not been reported, the relationships between these
process variables has not been established. Some of the difficulty
in choosing the optimum amount of surfactant and electrolyte could
be reduced if the mechanism of the separation process was better
understood. Such information may help to control the formation of

emulsified olein product.

The mechanism of the separation step will be linked to detergency
theory to explain experimentally observed relationships. Preston
(1948) and Harris (1958) described the relationship of detergency
to the critical micelle concentration. The phenomenon of
detergency was shown to increase greatly upon the surfactant

concentration reaching the critical micelle concentration.

At the critical micelle concentration, many solution properties
undergo easily detectable changes. One of the properties, the
surface tension, is quickly and easily measured (Adamson, 1982).
The surface tension of a surfactant solution exhibits a distinct
dip, or a discontinuity in a surface tension versus surfactant

concentration plot at the critical micelle concentration. The

6
change in the surface tension allows the determination of the
critical micelle concentration for a particular surfactant or
detergent solution. The surface tension measurements along with
emulsification tests will be used to determine the surfactant

distribution between crystal surface and the aqueous solution.

The crystal size dependence of the process will be investigated.
Various crystal sizes will be made by using different rates of
cooling and different intermediate temperatures during the
crystallization step. The recovery of olein from different
crystals sizes will be studied as a function of surfactant and
electrolyte concentration and detergent solution to tallow weight
ratio. The separation mechanism will then be reviewed using the

new data.

The effect of detergent solution to tallow weight ratio on the
olein yield may be related to the rheological properties of the
dispersion. Viscosity effects will be studied using a paddle mixer
viscometer. The effect of various detergent solution volumes will
be related to the rheology of the dispersion. The desired
information is the rheology of the tallow-detergent solution
dispersion as a function of the detergent solution to tallow weight
ratio. The effect of the apparent viscosity on the olein yield is

to be correlated directly.

The results of this research should help establish upper and lower

bounds on processing conditions to allow process design and control

7
to be done. The relationship between the crystals formed in the
crystallization step and the process variables include the effect
of crystal size on the required detergent strength, the type of
crystallization process required and the volume of detergent
solution required for an optimum separation. Research of the
separation mechanism is the first step in the process. The results
obtained in this research may sometimes be qualitative because of
limitations in measurement and scale-up of results from small,

batch laboratory experiments to industrial size operations.

LITERATURE REVIEW

2.1 Chemistry of Tallow and other Triglycerides

Tallow is made up of a collection of triglycerides. A triglyceride
is a glycerol molecule esterified to three fatty acids. The fatty
acid composition of the triglycerides determines the properties of
the fat or oil. The fatty acid composition of natural fats like
tallow, varies depending on geographic location and feed
characteristics. Table 2.1 shows some of the variations in tallow
fatty acid composition that have been observed. The composition
differences cause small differences in the melting point and iodine
value of the fat. The crystallization properties at a specific

temperature may vary greatly.

The melting point of fats is a function of the fatty acid
composition and position on the glycerol molecule. The melting
point of a triglyceride is directly controlled by the melting point
of the fatty acids it contains. The fatty acids may be either
saturated or unsaturated. The melting point of the unsaturated
fatty acids can be 70 °C higher than the melting point of a
saturated fatty acid. The chain length of the saturated or
unsaturated fatty acid also affects its melting point. The longer
the fatty acid chain the higher the melting point for the fatty

acid.

Table 2.1

TALLOW FATTY ACID COMPOSITION

 

 

 

Data source (1) (2) (3)
Fatty acid common name Mole Fraction
Myristic acid C14:0 .044 .036 .02-.08
Myristoleic acid C14:1 .016 .014 .004-.006
Palmitic acid C16:0 .267 .241 .24-.37
Palmitoleic acid C16:1 .050 .066 .019-.027
Stearic acid C18:0 .139 .137 .140-.290
Oleic acid C18:1 .417 .501 .400-.500
Linoleic acid C18:2 .023 ---- .010-.050

Linolenic acid C18:3 .003 .004 --------

 

1 Luddy et al., 1973
2 Bussey et al., 1981

3 Swern et al., 1979

10

Mixed triglycerides, containing a number of different fatty acids,
minimize the effect of the differing melting points of the fatty
acid components on the whole fat. Predicting the melting range of
a group of triglycerides based on the fatty acid composition is not
possible. A fat with a wide variety of fatty acids, for example
tallow, has a melting range of 10 to 45 °C. 3A fat without a great
variety of fatty acids, such as cocoa butter, has a narrow melting

point range of 6-8 °C.

The mixed nature of the triglycerides combined with the random
distribution of the fatty acids means that a fat like tallow with
eight main fatty acids has over a hundred different triglycerides.
This makes the melting and crystallization of fats complicated.
Replication of crystallization processes is difficult as exact

duplications of time-temperature conditions are necessary.

2.2 Fractionation Processes

 

There are three methods of fractionation which have industrial
importance (Braae, 1976). The methods are based on cooling the fat
to a temperature where the high-melting components crystallize

while the low-melting ones remain in liquid form.

The oldest method consists of a slow cooling of the fat in small
shallow tanks to an appropriate temperature at which the fat is
kept for 3-4 days during crystals formation (Braae, 1976). For

instance, tallow might be crystallized at 25-30 °C. It is

11
important that the crystallization is extremely slow so that large
crystals are formed. Large crystal agglomerates, larger than 100

micron diameter, are recoverable by filtration.

After the crystallization is complete, the viscous grainy mass is
put into canvas press cloths, which are stacked in presses of the.
plate type and the liquid oil is pressed out of the crystals
(Braae, 1976). This is a labor intensive process and has been
largely replaced by more efficient methods. The more efficient
methods use larger crystallization tanks and the separation is
accomplished by vacuum band filters. Careful crystallization is

still necessary to produce crystals which can be filtered.

The filtration can be made more easily if the fat is mixed with a
suitable solvent before it is crystallized (Braae, 1976). The
solvent should not prevent crystallization from occuring. Suitable
solvents include hexane, isopropyl alcohol and acetone. The
solvent will lower the liquid phase viscosity making the physical
separation of oil from crystals easy to accomplish. The
crystallization in solvent tends to be selective compared to other
crystallizations. This means the two fractions obtained have the
greatest possible chemical composition difference. The
fractionation plants are quite complicated and costly to run

because of the solvent processing requirements.

A third fractionation method is based on a method for separating

fat crystals and oil first described by Fratelli Lanza in 1905a

12
The partially crystallized fat is mixed with an aqueous solution
containing a surfactant. The surface active component of the
aqueous solution allows the crystals to be freed from the oil and
wetted into the aqueous solution. The oil can then be easily

recovered from the aqueous-crystal mixture by centrifugation.

2.3 Aqueous Fractionation

 

Aqueous fractionation has been discussed by many authors. Because
certain aspects of the process mechanism have not been defined the
results of the various authors can present a confusing array of
information. Table 2.2 has been prepared to help organize the

various findings which are discussed in the following paragraphs.

Stein and Hartmann (1957) describe a process for the separation of
various high molecular organic compound mixtures found in
vegetable, fish and animal fats. They describe several ways of
obtaining a liquid-solid fat mixture suitable for dispersion with
an aqueous solution containing a surface active material. The
liquid-solid fat mixture is dispersed with an aqueous solution
containing a surface active material. The result is an oil-in-
water emulsion or dispersion. The dispersion is then centrifuged
to yield the liquid product directly. The solid portion is
recovered as a liquid after heating the aqueous-solid mixture and

centrifuging.

13

Table 2.2

Aqueous Fractionation Process Variable Recommendations

 

 

 

 

Research Surfactant Electrolyte Aqueous Crystal
Author 8 Quantity & Quantity gguantity Size & Type'
weight weight weight
percent of percent of ratio of
aqueous aqueous aqueous soln
solution solution to fat
Stein 8 .05 to 5.0% 0.5 to 5.0
Hartmann fatty alcohol
(1957) sulfates ___
10 to 28
Carbons
Stein 0.2 to 0.5% I-ZZ 1.0-2.0
(1968) fatty magnesium aqueous to
alcohol sulfate fat by .___
sulfate volume
Haraldsson beta or
(1974) betaprime
____ ___ ___' crystals
desirable
Poot et al. 0.1-3.0
(1975) micron crystals
___ ____ ____ cause separation
problems
Rek sodium decyl 2.5-5.0Z 0.2-0.8
(1977) sulfate many mono-
O.3Z valent ___
cation
salts
Bussey sodium 0.7-4.0Z 3.5
et al. dodecyl sodium
(1981) sulfate sulfate ___
0.62
Glassner surfactant
(1983) quantity -
f(crystal

size)

14
Stein and Hartmann (1957) comment briefly on the role of the
surfactant in the separation. They state that the surface active
material reduces the interfacial tension between the organic
material and the aqueous medium. The amount of surface active
agent to be used is said to depend on the surfactant properties.
Concentrations of 0.05 to 52 based on the weight of the aqueous
medium are sufficient for good results. The last two statements
apply to a range of fatty alcohol sulfates, containing 10 to 28

carbons in the carbon chain.

Stein and Hartmann (1957) mention that properties of the aqueous
solution can be influenced by adding electrolytes which are inert
to the other components of the dispersion. Any electrolyte is
suitable as long as it does not form substantially insoluble
precipitates with the surface active material. Electrolytes help
to stabilize the hydrophilic portion of the surfactant molecules in
the water phase. The preferable concentration for the electrolytes
are not stated. The amount of the aqueous solution to be used
ranges from 0.5 to 5.0 parts by weight of aqueous solution for each

part by weight organic material.

In addition to the desirable wetting action of the surfactant
solution, the aqueous solution can exhibit emulsifying power which
can cause difficulty in the subsequent separation (Stein and
Hartmann, 1957). To combat this problem "protective colloids" are

recommended. Addition of the protective colloids reduces the

15
amount of surface active material required and thereby reduces the

emulsifying power of the aqueous solution.

The term "protective colloids" refers to any inorganic or organic
substances capable of increasing the viscosity of the aqueous phase
(Stein and Hartmann, 1957). Examples of inorganic protective
colloids are swelling clays, microcrystalline silica and other
materials which will remain suspended in colloidal form in water or
only settle slowly. Organic protective colloids include glue,
gelatin, pectins, polyglycol ethers, polyacrylates, cellulose
glycolates and methyl celluloses. The concentration of the
protective colloids recommended ranges from 0.1 to 7.0 percent by

weight of the aqueous solution.

The optimum quantity of the surface active material, the liquid
organic compounds, the electrolytes and the protective colloids and
the fineness of comminution may be determined by preliminary tests
(Stein and Hartmann, 1957). Variation of the operating conditions
during processing of the material to be separated can be used to
control liquid and solid product compositions. Some of these
operating conditions are rate of cooling, final separation

temperature and the original mixture of raw materials.

In a later publication, Stein (1968) recommends a 1-22
concentration of inorganic salt in the aqueous solution. The
preferred salt was magnesium sulfate. The Optimum surfactant

concentration was 0.2-0.5% of a fatty alcohol sulfate. The

16
recommended aqueous solution to organic phase ratio was 1:1-2:1 by
volume. These optimums were for the separation of oleic and

stearic fatty acids.

Stein (1968) found that mixing the dispersion with a small amount
of the aqueous solution containing all the surfactant and
electrolyte, and then diluting the dispersion with water to the

final aqueous to fatty acid ratio was advantageous.

Poot et al. (1975) found that extremely small crystals of fat
presented serious problems for fractionation by the aqueous method.
Crystal sizes of 0.1-3.0 microns in length made high demands on the
dispersion technique, the wetting ability of the detergent and on

the separation by centrifugation.

A necessary condition for a successful fractionation using the
aqueous dispersion technique is formation of crystals of the proper
type (Haraldsson, I974). The actual size of the crystals is of
minor importance. Haraldsson (1974) did state that 25-50 micron
crystals are large enough. If the oil is cooled too rapidly,
amorphous crystals of the alpha type are formed. Slower cooling
creates beta or betaprime crystals which are easily wetted by the
aqueous solution. The alpha crystals are not prOperly wetted and
oil recovery is not good. For fats such as palm oil, tallow and

lard the necessary crystallization time is 4-6 hours.

17
Rek (1977) states that the process, although simple, is complicated
by speed of crystallization, form of crystals, occlusion of oil in
the crystals and inexplicable and unpredictable effects. The
author does not define the effects. The problems are said to be

more severe at low temperatures and when small crystals are formed.

Rek (1977) found that use of an oil-soluble surface-active agent in
addition to the water soluble surface active agent improves the
olein yield. The preferred oil-soluble surface active agents are
unsaturated monoglycerides of twelve to twenty-eight carbon
unsaturated fatty acids. The desired water soluble surfactant is
sodium decyl sulfate. The concentrations of both surfactants
depends on the particular surface active agents, the mixture of
triglycerides and the aqueous system used. The oil soluble
surfactant is preferably added to the mixture of triglycerides but
addition to the aqueous solution is suitable. It should be noted
that some oil soluble surfactants are naturally occurring in crude

fats.

Rek (1977) found that the water soluble surface active agent,
preferably sodium decyl sulfate, should be present in a
concentration of at least 0.32 based on the triglyceride mixture
weight. The inorganic salt concentration in the aqueous solution
should be between 2.5 to 5.0% by weight of the aqueous solution.
The volume ratio of water to triglyceride mixture should be between

0.2 to 0.8:1.

18
Bussey et al. (1981) fractionated tallow and state that the optimum
surfactant, sodium dodecyl sulfate, concentration to be 0.62 by
weight of the tallow. The optimum volume of water to tallow is
stated to be 3.5:1. Bussey et al. recommend adding the surfactant
directly to the partially crystallized tallow and find this

increases olein yield.

During the fractionation of tallow, the size of the triglyceride
crystals and the concentration of the surface active agent are
important (Glassner, 1983). The olein yield increases as the
surfactant concentration increases. The olein yield reaches a
maximum and increasing the surfactant concentration above the level
needed for the maximum olein yield results in the formation of an
emulsion, which makes recovery of the olein fraction by
centrifugation difficult. The size of crystals formed effects the

amount of surfactant required for a successful fractionation.

2.4 Crystal Morphology

 

Polymorphism is the existence of two or more crystalline
modifications for the same substance. Triglycerides exhibit
multiple crystal packings and this was first demonstrated using x-

ray diffraction by Clarkson and Malkin (1934).

The monoacid saturated triglycerides or mixtures of only a few
saturated triglycerides exhibit three polymorphs described by their

x-ray short spacings (See Table 2.3).

19

Table 2.3

X-ray Short Spacings

 

Crystal Type Short Spacingg

 

alpha 4.1 X
betaprime 3.80 A, 4.20 A
beta 4.6 A

(Chapman, 1962)

 

For a natural mixture of mixed fatty acid triglycerides three
polymorphs may not be observed (Hoerr, 1960). Some exhibit only
the lower melting forms such as alpha and betaprime. Only two
polymorphs are formed because the mixed triglycerides can not pack
together closely enough to transform to the beta form, the most
closely packed triglyceride form. Fats crystallized initially in
the alpha or betaprime form will transform in the sequence, alpha
to betaprime to beta, until the most stable form for the fat is
attained. If close packing cannot be achieved, the betaprime form

may be the most stable for a mixed triglyceride.

The beta form crystals have triclinic packed chains, the betaprime
crystals have orthorhombic packed chains and the alpha form
crystals have hexagonal packed chains (Chapman, 1962). The
triglyceride packing structure is shown in Figure 2.1. The end
view of the triglycerides in Figure 2.1 would show one of the three

patterns mentioned above, thus defining the crystal type.

Figure 2.1 Triglyceride Chain Packing Diagram.

20

21
The rate of cooling determines the crystalline type initially
formed. Slow cooling will produce the more stable crystal
polymorphs, beta or betaprime (Hoerr, 1960). Fast cooling will
likely result in alpha type crystal packing. If the alpha crystals
are tempered at the correct temperature the alpha crystals will

transform to the most stable form, beta or betaprime.

Hoerr (1960) also proposed that each crystalline type has features
that can distinguish it from other crystal types when viewed using
a microscope. Microscope investigations are not too definitive
because a variety of crystal habits may occur for one crystal form
(Chapman, 1962). Alpha form crystals are fragile, transparent
platelets with a characteristic size of about five microns.
Betaprime crystals are tiny needles of one micron in length. The
beta crystals are large needles up to fifty microns in length or
longer. They form clumps up to one millimeter in diameter. The
type of fat was not stated. The sizes described by Hoerr (1960)
depend on the temperature driving force and on the type of fat

being crystallized.

2.5 Detergency Theory

 

Detergency is defined as the process of cleaning the surfaces of a
solid material by means of a liquid bath involving a
physicochemical action other than simple solution (Neiditch, 1972).
The displacement of an oil-water interface by a water-fiber

interface by an aqueous solution containing electrolyte and

22
surfactant is a detergency phenomenon (Schott, 1972). Detergency
is the sum of a variety of phenomena, including electrostatic
forces, interfacial potentials, and ion activities of both

surfactants and builder electrolytes (Goette, 1949).

Surfactants are characterized by their absorption at all types of
interfaces (Lange, 1972). The cleansing action of surfactants is
due to their absorption at the interfaces involved. The
mechanical, electrical and chemical properties of the interfaces
are all altered by the surfactant absorption. The degree of
absorption depends on the type and concentration of the surfactant,
the presence of other materials in solution, the solution

temperature and the absorbent properties.

The absorption of a surfactant strongly increases as the surfactant
concentration is increased until the critical micelle concentration
is reached (Lange, 1972). Micelles are aggregates of surfactant
molecules. They form in solutions at concentrations above the
critical micelle concentration (CMC) (Neiditch, 1972). Since the
concentration of individual surfactant molecules in the aqueous
phase is virtually constant above the CMC, the surfactant is at

essentially its optimum for many applications such as wetting.

Electrolytes have been shown to reduce the concentration of
surfactant needed to reach the CMC. The addition of electrolytes

to a surfactant solution causes the property changes characteristic

23
of reaching the CMC to occur at lower surfactant concentrations

than if the electrolytes were not in the solution (Harris, 1958).

The work of Adam (1937) showed that oily soil was removed from
fiber surfaces by wash water through a mechanism of preferential
wetting. Kling (1949) called the mechanism exchange wetting,
because the oil-fiber interface exchanged to become a water-fiber

interface.

Adam and Stevenson (1953) and Moillet et a1. (1961) gave the
following description of the preferential wetting or rolling-up
mechanism. A fiber partly covered by a film of oil was immersed in
a dilute aqueous surfactant solution, which did not completely
displace the oil. A state of equilibrium resulted. The oil and
the aqueous solution both remained in contact with the fiber. The
surfactant molecules were adsorbed at the fiber-water and oil-water
interfaces in a monolayer film. The film of surfactant molecules
being adsorbed at the fiber surface displaced the oil because the
surfactant adhered to the fiber surface more strongly than did the

oil.

As the concentration of surfactant increases, the oil rolls up
until it no longer is in contact with the fiber (Adam, 1937). The
oil droplets are now surrounded by an aqueous phase. The
surfactant is adsorbed at the oil-water interface helping to

stabilize the oil-in—water emulsion (Adam and Stevenson, 1953).

24
If a smooth film of oil completely covers a fiber surface, an
aqueous solution can not break through the oil to wet the fiber
surface (Schott, 1972). A point at which the three surfaces meet
is necessary to roll-up the oil. If a contact point for the three
phases is established, the diSplacement of the oil-fiber interface

by the aqueous-fiber interface can occur.

The mechanical force exerted upon the particles in the cleansing
process may be important in connection with the influence of the
particle size (Lange, 1972). It is well established that solid
particulate soil can not be completely removed by a surfactant
solution without mechanical action. This can be explained by the
necessity of detergent solution contact with the fiber. Contact of
the detergent solution with the fiber is necessary for wetting the

fiber.

2.6 Critical Micelle Concentration

 

In surfactant solutions, aggregations of molecules form over a
narrow concentration range called the critical micelle
concentration (Harris, 1958). The shape of micelles is first
pictured by McBain (1913). Hartley (1936) proposes another
configuration for the orientation of micelles. Many other shapes
have since been proposed and all have valid supporting evidence
(Harris, 1958). The chemical structure of the compound being

investigated may be a controlling factor.

25
In addition to shape, the size of the micelles is important
(Harris, 1958). For low molecular weight compounds as few as ten
molecules are associated while in other cases they exceed a
hundred. Studies have shown that within a homologous series of
surfactants the number of molecules associating is definite and
increases with increasing hydrophobe chain length. A micelle

molecular weight can vary from 4,000 to 40,000.

The most effective surfactants generally have the lowest CMC while
having longer, optimum hydrophobe structure. Consequently a
comparison of surfactants can be made on this basis, but of even
greater importance is the measurement of the effect of builder
electrolytes upon surfactant CMC. These generally lower the

initial CMC remarkably (Harris, 1958).

2.7 Surface Tension Measurement

 

Measurement of either surface or interfacial tension are time-
consuming but relatively effective methods for determining CMC
(Harris, 1958). A method widely used involves the determination of
the force to detach a ring or loop of wire from the surface of a

liquid (Adamson, 1984). Table 2.4 lists other possible methods.

One of the difficulties with the DuNouy (1919) method lies in
adequate temperature control, but apparatus for this purpose can be
constructed. At elevated temperatures, evaporation is also a

difficulty, but again this can be corrected (Harris, 1958).

26

Table 2.4

Methods of Surface Tension Measurement

 

 

1.

Capillary Rise

Maximum Bubble Pressure
Drop Weight

Ring

Wilhelmy Slide

Pendant Drop

Sessile Drop or Bubble

27
The method is capable of good precision. The tensiometer makes use
of a torsion wire and is commonly used for determining the
detachment force. Harkins and Jordan (1930) worked out an
empirical correction factor to use for correcting the detachment
force to the true surface tension. The detailed theory of the
method has been determined by Freud and Freud (1930). The
calculated correction factors agree with the empirical factors to

within 0.25%.

The method is quite accurate if sources of error are avoided
(Adamson, 1982). The dry weight of the ring should be used for
calibration purposes. The ring must be kept horizontal to avoid
measurement errors. An angle of 1° causes an error of 0.52 and a
deviation of 2.1° causes an error of 1.6%. The surface of the
fluid must not be disturbed as the detachment point is approached.
Flaming of the ring before measurements removes grease and other
contaminants from the ring. The fluid vessel is best if it can be

overflowed to remove surface contaminants.

A zero or near zero contact angle is required or the recorded
results will be low (Adamson, 1982). This was found to be the case
with surfactant solutions. A teflon coated or polyethylene ring
may be used to keep the contact angle at zero (Krynitsky and

Garrett, 1963).

28

2.8 Dispersion Rheology

 

Due to the wide variation in recommended detergent solution
quantities of previous investigators the dispersion rheology may be
important (See Table 2.2). Previous results of this author
revealed a separation dependence on the crystal size. Since the
solid properties affect the dispersion rheology the crystal

properties may be critical.

A paddle mixer viscometer will be used for the dispersion rheology
experiments. The paddle mixer is the choice for two reasons
(Steffe and Ford, 1985). First, wall effects due to slip are
minimized with the mixer or impeller type sensor. The second
reason is that mechanical degradation of the sample during
instrument loading is less for the paddle sensor system than the

concentric cylinder systems.

Rao (1975) and Metzner and Otto (1957) have studied the application
of mixer viscometry to food products. Power consumption in a
mixing vessel for a Newtonian fluid can be expressed in terms of a
power number,
- (p/d5w (03.,0) (2.1)

and a mixing Reynolds number,

Re - (alga/f.) (2.2)
as

Pa = B/Re (2.3)

or

29

p/a5w3,o - B’a/dzwf (2.4)
Equation (2.4) can be used for power law fluids if the fluid
viscosityyu, is replaced by an apparent viscosity, 7a' evaluated
at an average rate of shear, Ya, which is defined as

X - k'u) (2.5)

3

Equation (2.5) is valid for a particular impeller if a plot of log
(p/KuPT1d3) versus (1-n) is a straight line having slope equal to
-log k' (Rieger and Novak, 1973). K and n are the consistency
coefficient and flow behavior index used to describe the power law
fluid expressed in terms of the shear stress and shear rate as

t' «(fa )m (2.6)
When using the mixer viscometry technique, the flow behavior index
is calculated as the $10pe of the log (torque) versus log (impeller
speed) curve. The consistency coefficient is calculated as

Kx =(Mx(wk')nr Ky)/(My(wk')"¥) (2.7)
where x and y are subscripts referring to the test fluid

and a standard (fluid with known properties),

respectively.
Equation (2.7) is a ratio of equation (2.6) for two different
fluids. The torque, M, is equal to the shear stress times a
constant; the constant is equal for both fluids. The shear rate,
auk', and the consistency coefficient and consistency index are the

same as in equation 2.6.

The mixer viscometer constant, k', can be determined by using
several different fluids and evaluating K and n for them using

standard rheological techniques. Solutions of hydroxypropyl

30

methylcellulose, guar gum and corn syrup were evaluated. A plot of

log (p/kup+1d3) versus (l—n) was made. The result was a straight

line with slope equal -log k', from which the slope k' was

determined to be 4.46 (Steffe and Ford, 1985)

EXPERIMENTAL METHODS AND EQUIPMENT

3.1 Tallow Sample Preparation

 

Tallow was obtained from a commercial supplier in a 55 gallon drum.
Samples, for quantitative comparisons, were prepared by melting a
large quantity of tallow and separating it into 200 m1 aliquots.
Beakers were used as crystallization and separation vessels for the

process. The sample weight was 170 g of fat.

3.2 Crystal Preparation Methods

 

Three crystallization procedures were used to produce crystals for
the separation step. The goal of these procedures was to prepare
three different sizes of crystals. The final crystallization
temperature was 40 °C in all cases. The procedures differed during
the initial cooling period. During this period crystal nucleii
were produced. The crystal nucleii produced determine the number
and size of the final crystals. The speed of cooling determined
the initial triglyceride crystal type. Due to the small size of
the samples crystallized, agitation was not necessary for adequate

heat transfer.

Crystallization procedure one, a partial melting process, was begun
by raising the tallow temperature to 60 °C to completely melting
the tallow. The tallow was then put in a freezer which quickly

31

32
brought the tallow temperature to 0 °C in under fifteen minutes.
The tallow was rapidly nucleated and crystallized totally during
this period. The tallow was kept in the freezer for a total of two
hours. The crystallized tallow was then put into a 40 °C
temperature bath. The crystallized tallow rose to 40 °C with some
of the tallow melting as the temperature rose. The tallow remained
in the temperature bath for 16 hours which allowed equilibrium to
occur between the liquid oil and solid crystals. The total
crystallization time was 18 hours. Figure 3.1 shows the

temperature profile for the three crystallization procedures.

In procedure two the tallow was heated first to 60 °C and then
cooled at an ambient temperature of 29 °C for two hours. During
this two hour period the oil cooled to about 33 °C before the first
crystal nuclei began to form. The heat of solidification caused
the tallow temperature to slowly rise after the tallow began to
crystallize. The cooling to 33 °C took one hour. During the
second hour crystallization generated heat which caused the tallow
temperature to rise to 35 °C. The tallow sample was put into the
40 °C temperature bath for the remaining 16 hours of the

crystallization time.

Crystallization procedure three was begun by melting the tallow
sample completely by raising the tallow temperature to 60 °C. The
tallow was then cooled at an ambient temperature of 29 °C until
crystal nucleii were seen. The initial cooling took 1 hour, then

the nucleii appeared. The temperature of the tallow was 33-34 °C

Procedure
One

.o—e—e. IFVVC)
5° -— Three

TEMPERATURE , °C
10
O

10

 

 

 

O 30 60 90 120 150 180
TIME, MINUTES

Figure 3.1 Temperature Profiles for Three
Crystallization Procedures.

33

34

at this time. The sample was then put into the 40 °C temperature

bath for the remainder of the crystallization time.
3.3 Crystal-Oil Separation Procedure

After the crystallization was complete the correct amounts of
water, electrolyte and surfactant were mixed and brought to 40 °C.
Distilled water was used as the base of the detergent solution.
The electrolyte used was sodium citrate (Mallinckrodt) and the
surface active agent used was sodium dodecyl sulfate (Fisher

Scientific).

After preparation the detergent solution was then added to the
partially crystallized tallow. The detergent solution was mixed in
with a stirring rod. Mixing was carried out until a uniform
dispersion was obtained. Previous tests indicated that more
intense mixing does not increase olein yield (Glassner, 1983). The
dispersion was allowed to sit for one hour before it was
centrifuged, with a centrifugal force of 3,600 times that of
gravity. A DuPont Sorval RCZ-B centrifuge was used for the

centrifugation steps.

The dispersion separated into an olein layer on top, a solid
crystal layer in the middle and a detergent solution layer at the
bottom of the centrifuge tube. The olein was poured out of the

centrifuge tubes after the centrifugation. The yield of olein was

.I I lIIIl. III... IIIIF Ill-r I I'll. ll, i I
11‘ Jllll 1" I I 1
“'"I

35
calculated by weighing the olein collected and comparing to the

amount of tallow originally present.

The crystal layer was recovered by heating the crystal and

detergent solution until the crystals melted. The melted mixture
was centrifuged to separate the stearin, now in liquid form, and
the recyclable detergent solution. The process is schematically

diagrammed in Figure 3.2.

Fractionation data were generated using crystallization procedure
three except where otherwise noted. A detergent solution to tallow
weight ratio of 0.80 was used for all separations except those in
which other ratios were specified. The sodium citrate
concentration in the detergent solution was 5.0 percent for all
separations, unless otherwise specified. The surfactant

concentration or weight used was specified for each data point.

3.4 Surface Tension Measurement

 

A Cenco Tensiometer Model No. 70530 was used to obtain the surface
tension measurements. The detergent solution was mixed to the
desired concentration. The measurements were taken at 40 °C
solution temperature. Both a platinum ring and a teflon coated
platinum ring were used to obtain the surface tension values. The
teflon coated ring was used to insure a zero contact angle with the

material being tested (Krynitsky and Garrett, 1963). The surface

Partial

 

Crystallizer
- Dispersion
Vessel .
Olein
-
l ‘ Centrifugation
Heat Stearin
Exchange '
A

Aqueous Fractionation Process

Figure 3.2 Fractionation Process Schematic Diagram.

36

37
tension values were corrected as recommended (Zuidema and Waters,

1941; Adamson, 1982).

The surface tension measurements made required the tensiometer
previously mentioned, a four inch diameter watch glass, a teflon
coated platinum ring and a platinum ring. The platinum rings in
this case were six cm circumference rings. The watch glass was the
solution holding vessel, which was chosen for its ability to be
overflowed to insure a clean air-liquid interface. The watch glass
was cleaned for each solution by rinsing throughly with tap water

and then with distilled water.

The solutions were prepared with the desired amounts of distilled
water, sodium dodecyl sulfate and sodium citrate. Since the
capacity of the watch glass was slightly under 100ml, using 100 ml
of distilled water provided a good basis. After the solution was
heated to the correct temperature of 40 °C. the solution was poured
into the watch glass and the watch glass was slightly overfilled to
give a clean liquid-air interface. The appropriate ring was then
contacted with the liquid surface. A zero reading on the
tensiometer scale was achieved by adjusting the height of the tray
holding the watch glass and the tension control for the

tensiometer.

The tension control was then slowly increased while the height of
the holding platform was lowered so that the zero reading on the

tensiometer scale was maintained. This procedure was continued

38
until the ring was detached from the liquid surface. When the ring
detached, the surface tension was read from the graduated dial
scale. The appropriate correction factor was applied and the final
surface tension reading for the particular solution was obtained

(Zuidema and Waters, 1941).

The tensiometer calibration was an essential part of the
experimental technique. The recommended procedure was designed to
give a direct reading in dynes from the graduated dial. First the
lever arm, on which the platinum ring was hung, was clamped in
place and the ring was cleaned. The cleaning was accomplished by
heating the ring momentarily to a dull red in the oxidizing portion
of the flame. The ring was then hung on the hook at the end of the

lever arm.

A small piece of paper was put on the ring for use as a platform.
The tension control was then adjusted so the lever arm index and
its mirror image coincided with the reference line on the mirror.
The dial clamp was loosened and the graduated dial rotated until
the vernier indicated approximately zero. The tension control was

adjusted until the reading was exactly zero.

A mass of 500 to 800 mg weight was placed on the paper platform.
The weight was known accurately. The tension control was
readjusted so that the index was lined up with the reference line.
The dial reading was recorded to the nearest 0.10. Using equation

(3.1), the apparent surface tension value was determined.

39

s = Wg*/2L (3.1)

If the recorded dial reading was greater than the calculated value
then the lever arm was adjusted by shortening; if the dial reading
was less than the calculated value, the lever arm was adjusted by

lengthening.

The calibration procedure was repeated by readjustment of the zero
position with the paper on the ring after each adjustment of the
lever arm length until the dial reading agreed with the calculated
value. When the calibration was completed each unit on the dial
represented a surface tension measurement unit of one

dyne/centimeter.

After completion of the calibration procedure the paper was removed
from the ring and the zero position was reset. The tension control
knob was turned until the index was aligned with the reference line
and its mirror image on the mirror. Then the tension adjustment

was loosened until the vernier read zero. The tensiometer was now

ready for use.

3.5 Rheology Data Measurement

 

A Haake RV-12 viscometer, interfaced to a Hewlett-Packard 85
computer and 3497 data acquisition system, was chosen to measure
rheological properties of the detergent solution-partially

crystallized tallow dispersion. A MV paddle sensor was attached to

40
the viscometer drive head. Samples were placed in a viscometer cup
(See Figure 3.3), having a radius equal to 0.021 m, for testing.
Also shown in Figure 3.3 was the paddle sensor used. Its diameter
was 0.04143m and its height was 0.02692m. The blade was pitched 15
degrees from a plane through the shaft. The viscometer was
equipped with a Haake F3-C water circulator to maintain samples at

a constant temperature while collecting data.

The preparation of the dispersion to be tested was the first step.
The sample cup needed a total volume of 45 ml liquid to fill it to
a level just above the top of the paddle mixer. The partially.
crystallized tallow was already crystallized by the desired
crystallization procedure. Depending on the weight ratio of
detergent solution to tallow chosen, the appropriate amounts of
partially crystallized tallow and aqueous solution were loaded into
the sample cup. The sample cup was then loaded into the Haake
viscometer unit. The tallow and detergent solution were mixed in
the cup before the cup was loaded. The dispersion was further
mixed in the cup during the 60 minute period used to gather the

torque versus time data.

The Haake temperature bath was set to 40 °C to bring the viscometer
to the necessary 40 °C temperature about 15 minutes before the
viscometer cup was loaded. The viscometer was zeroed before and
after loading the viscometer cup. The rotational rate was set at

120 rpm and the collection of data was begun. After 60 minutes the

 

Figure 3.3 Mixer Viscometer Paddle and Cup.

41

42
data collection was ended and the data was processed. A raw data

output and a graph of torque versus time was printed.

Torque as a function of the rotation rate was then measured for the
dispersion. A rotation rate was set and the torque was recorded
at that rotational rate. The rotation rate was reset and the
torque was again recorded. After this was done for fifteen
rotation rates ranging from one rpm to 180 rpm the data were
printed. Both the raw data and a graph of torque versus rotation

rate were printed.

A program for the HP-85 for power-law fluids was used to analyze
the data collected. The program calculated the consistency
coefficient, K, and the flow behavior index, n. The torque versus
rotational rates were assumed to be those of a power law fluid near
the rotational rate in question. A plot of log (torque) versus log
(rotational rate) was therefore a straight line. The program
calculated the slope of this line from which the flow behavior
index n was obtained. The consistency coefficient, K, was then
obtained by using a ratio of the common shear stress-shear rate
equation for two fluids. Since the torque was directly
proportional to the shear stress for both the dispersion being
tested and for the known fluid, the proportionality constant
canceled out of the ratio when the torque was substituted for the
shear stress. The only unknown was the consistency coefficient, K,

of the dispersion tested, which was then calculated.

43

Both the K and n values were printed out by the program. These
parameters were necessary for calculation of an apparent viscosity.
The values were printed allowing calculation of an apparent
viscosity according to equation (3.2).

78 - K(k'w)“'1. (3.2)
Equation (3.2) was evaluated at 60 rpm giving an apparent shear

rate‘(k'w) equal to 28.02 s'l.

3.6 Crystal Size Measurement

 

Photomicroscopy was used to determine the size of crystals produced
by various crystallization procedures. The photomicroscopy was
conducted on an American Optical Phase Star microscope fitted with
a Kodak 35 mm camera. The film used was Kodak Type 2415, a high
resolution black and white film. D-19 developer was used to

develop the film. A developing time of four minutes was used.

The photomicroscopy was done in a 37 °C constant temperature
cubicle. The high temperature was necessary to keep the oil in the
oil-crystal mixture from crystallizing onto the crystals. The
partially crystallized tallow from all three procedures was

photographed.

44

3.7 Olein Emulsification Measurements

 

Tests were conducted to determine the amount of surfactant required
in the detergent solution to emulsify olein. Olein previously
obtained from completed fractionations was mixed with detergent
solutions containing various concentrations of surfactant. The
mixture was then centrifuged under the same conditions used during
normal separations. The olein was observed after centrifugation
for the presence of emulsified olein. Tests were conducted for a
wide range of surfactant concentrations with the object of
determining the highest detergent solution surfactant concentration

possible before emulsified olein was left after centrifugation.

RESULTS AND DISCUSSION

4.1 Proposed Mechanism for the Separation of Liquid and Solid

Phases

Detergency theory was used to explain the crystal size and
surfactant concentration effects on the production of emulsified
olein described by Glassner (1983). Agitation provided after the
addition of the detergent solution to the partially crystallized
oil assists in contacting the oil, crystal and aqueous phases.
When a three-phase contact point was created, the high
concentration of surfactant in the aqueous solution rolled-up the

oil from the crystal surface almost immediately.

Figure 4.1 shows the process for the small single crystals
generated by the partial crystallization procedure. Microscopic
observation of the components of the process provided the basis for
the mechanism diagrammed in Figure 4.1. Stages A and D were
observed directly. Stages B and C were proposed based on
detergency theory. The intermediate steps occur rapidly enough

that they were not observed by microscopic techniques.

The spherulitic crystals, as shown in Figure 4.2, were wetted in a
similar way with the same end result except for the size of the
olein droplets formed. The detergent wetting of the
polycrystalline spherulites into the aqueous phase was attributed

45

 

' crystal

     

Stage A

anueous

Phase
Legend

 

Stage B

 

 

 

a agitation
b roll ing- up
of the all
C complete

Stage C

wetting

 

Process
Description

 

 

 

 

Rolling-Up Detergency Mechanism for

Figure 4.1
Individual Tallow Crystals.

46

    

tfiELa

    
 

..<
«a - “find;

Stage C

Figure 4.2 Rolling—Up

 

' crystal

w oil
0 aqueous

Phase
Legend

 

 

a agitation

b rolling-up
of the oil

c complete
wetting of
one crystal
fiber

 

Process
Description

 

 

 

Stage D

Detergency Mechanism

for Crystal Agglomerates.

47

48
to the breaking up of the individual crystals from the spherulites.
The surfactant solution penetrated the boundaries between the
individual crystal fibers, and they were then separated from the
large aggregate. This was observed for triglyceride matter as well
as textile fibers (Scott, 1963; Fort, Billica and Grindstaff, 1966;
Fort, Billica and Grindstaff, 1968). As the oil rolled-off the
individual crystals it remained in contact with oil surrounding the
rest of the crystal agglomerate. When the last individual crystal
was wetted, the oil became an emulsified oil droplet that was much
larger than the droplet from a small single crystal. Stage A was
observed directly by microscope. Stages B and C were not observed
directly by microscopic methods. After stage D was formed the rest
of the crystal agglomerate was wetted quickly. Microscopic
observation revealed both single crystals and spherulitic

agglomerates after wetting was complete.

The transition from stage A to stage B depended on agitation and
the resultant contacting of the crystals and the detergent
solution. The strength of the detergent solution controlled the
transition from stage B to stage C. High surfactant concentration
is present the equilibrium lies to a stage D situation. Low
surfactant concentrations leave oil contacting the crystal surface,
as in stage B. The transition from stage C to stage D in addition
to requiring adequate detergent strength requires aqueous volume to

accomodate the crystals.

49
Increasing the surfactant concentration with no change in oil
droplet size is known to stabilize emulsions, especially with
sodium dodecyl sulfate as the surfactant. If the oil droplet size
was increased, with the surfactant concentration remaining the
same, less force was required to coalesce the larger droplet with

other oil droplets.

Poot et a1. (1975) found that when using a 52 sodium
dioctylsulfosuccinate solution, the oil droplets must be greater
than 12 microns in diameter for them to coalesce during
centrifugation. The final surfactant concentration was one to
three percent by weight of the aqueous solution, or 2.0 to 7.5% by
weight of the fat. Certainly, the allowable low range of oil

droplet size depends on the surfactant type and concentration.

The principles of detergency applied to the fractionation process
provided a mechanism to help explain fractionation results.
Further experimentation to determine operating conditions for
maximizing the olein yield were conducted based on this basic
principle. The effects of the crystal size, surfactant
concentration, aqueous solution to tallow weight ratio and

dispersion rheology were investigated.

4.2 Crystallization Results

 

Crystallization procedures two and three were similiar to

procedures that can be used with scraped surface heat exchangers to

50
produce partially crystallized fats. A typical commercial process
might use a precooler followed by a constant temperature
crystallization unit. The difference between commercial
crystallization and the procedure two and three crystallizations
was that the procedure two and three crystallizations had very
little agitation during the crystallization. In these bench scale
experiments, agitation was not required for heat transfer. The
amount of shear due to a scraped wall exchanger is small because
low rotation rates are used. The low shear rates should have
little effect on the crystal nucleation and growth. Therefore, the
crystals produced in the non-agitated vessels were assumed to be
representative of crystals that might be produced in a scraped

surface exchanger.

The crystals produced in the crystallization step are important to
the success of a fractionation (Glassner, 1983). Photomicroscopy
of the crystals from each crystallization was done to characterize
the size of crystals produced by each procedure. Figures 4.3 and
4.4 show representative crystals from the second and third
crystallization procedures. Satisfactory photographs were not
obtained for crystallization procedure one. The length of the
individual crystals and the size of the crystal agglomerates were
determined by Figures 4.3 and 4.4 and other similar
photomicrographs. The diameter of the crystals was determined by

looking at higher magnification photomicrographs of the crystals.

 

Figure 4.3 Photomicrograph of Procedure Two

Crystals, 100x Magnification.

51

 

Figure 4.4 Photomicrograph of Procedure Three
Crystals, 100x Magnification.

52

53
Scanning electron microscopy of the crystals was done to obtain
better resolution pictures of the crystals. The normal procedure
is to freeze a oil-crystal mixture and look at the cross section.
Since the crystals produced were long, thin needle-like crystals
the above procedure was not appropriate. In order to obtain the
length and diameter dimensions of the crystals it was desired to

look at the crystals alone.

Crystals were obtained after the separation step and cleansed with
ethyl alcohol to remove all traces of water (Jewell and Meara,
1970). Replicas, as desribed by Jewell and Meara (1970), could not
be produced. The crystals were mounted on an SEM stub, sputter
coated and photomicrographs were taken. Figures 4.5 (1,000X) and
4.6 (700K) show the results of SEM microscopy. The heat generated
by the electron beam in the SEM's vacuum was sufficient to melt the
crystals. The sputter coating of the crystals caused extensive
melting also. The results were no help in determining the size of

crystals produced and the SEM technique was abandoned.

Crystallization procedure one crystals were extremely small. The
light microscope resolution was poor even at 2,000 times
magnification. These crystals were initially unstable alpha type
crystals. The alpha crystals were rearranging into the next most
stable form, betaprime, at the 40 °C final crystallization

temperature used (Hoerr, 1960).

 

Figure 4.5 SEM Microscopy of Fat Crystals,
1,000X Magnification.

54

 

Figure 4.6 SEM Microscopy of Fat Crystals,
7OOX Magnification.

55

56
The crystallization procedure two and three crystals were much
easier to observe. They were more well defined than the procedure
one crystals and existed in the stable tallow crystal form,
betaprime. The procedure two crystals existed both as single
crystals and spherulitic agglomerates of individual crystals. The
procedure three crystals were primarily spherulitic agglomerates.
The average size of the procedure two crystal agglomerates was 40
microns in diameter. The single crystals making up the
agglomerates were assumed to be cylindrical in shape, 30 microns in

length and 0.3 microns in diameter.

Betaprime crystals are usually orthorhombic and, therefore, are
regular extended length polyhedrons (Chapman, 1962). The features
of the minor axes of the crystal were not discerned under the
microscope. The assumption of a cylindrical crystal simplified
surface area and volume calculations. The crystals actually grow
in layers (Jewell and Meara, 1970) and the length of the crystals
is due to the apparent ease of adding to a layer (See Figure 2.1)
as opposed to beginning a new layer. This results in the crystals

having a length/diameter of 75 or more.

The single crystals were estimated to make up about 20 percent of
the crystal mass in the procedure two crystallization. The average
size of the procedure three spherulites was 60 to 80 microns in
diameter. The individual crystals making up the spherulitic
crystals were cylindrical and had an average length of 50 microns

and an average diameter of about 0.6 microns. The single crystal

57
sizes were determined from higher magnification photographs. Table
4.1 summarizes the dimensions of the crystals from procedures two

and three. The calculations for Table 4.1 are shown in Appendix B.

4.3 Separation Results

 

Figure 4.7 shows the results of separations conducted on tallow
samples crystallized by each of the three crystallization
procedures. The crystallization procedure had a large effect on
the separation results. Procedure one had a maximum olein yield of
51 percent. This maximum occurred in a broad SDS concentration
range of 0.4 to 1.1 percent. The reason for the broad range of
surfactant concentrations giving the same olein yields was related
to the crystallization procedure and to the type of crystals
present. The yield of olein with practically no surfactant was

very close to the best yield of olein obtained.

The olein yields obtained after using crystallization procedure one
were too low for the procedure to be useful. The low yield was
attributed to the high degree of supercooling used to crystallize
the procedure one crystals. The low temperature trapped oil
components which would not have crystallized at 40 °C. except for
being subjected to the fast cooling and low intermediate

temperature of the process.

The olein yield reached a maximum of 79 percent for crystallization

procedure two. This maximum occurred at an SDS concentration of

58

Table 4.1

Crystal Sizes Produced by Procedures Two and Three

Two

Three

 

Crystal Shape
Crystal Diameter
Crystal Length
Average diameter of
spherulites.

Crystal volume
Crystal surface area

Surface area/volume

Cylindrical
0.3 microns

30 microns

50 microns

2.1 microns3

28 microns2

13 micron-1

Cylindrical
0.6 microns

50 microns

80 microns
3
14 microns

95 microns2

6.8 micron-1

% OLEIN YIELD

CRYSTALLIZATION PROCEDURES

74:1 ‘.#2 I#3

0.4 0.8 1.2 1.6
96 SDS

 

2.0

Figure 4.7 Separation Results from Three Different

Crystallization Procedures.

59

60
0.50 percent. At concentrations above 0.50 percent olein yield was
reduced by the formation of emulsion. At SDS concentrations below
0.30 percent, the olein yield was lowered due to ineffective

wetting of the crystals into the aqueous phase.

The olein yield peaked at 78 percent for the crystallization
procedure three separations. The maximum yield occurred at a SDS
concentration of 0.20 percent. At concentrations of surfactant
above 0.20 percent, olein yield was reduced by the formation of
emulsion. At SDS concentrations below 0.15 percent the olein yield

was lowered because of inefficient wetting of the crystals.

The maximum yield from crystallization procedures two and three was
essentially the same. The procedure one crystals were not
desirable for a good chemical separation. The difference in the
maximum olein yield for procedures two and three was insignificant.
The crystal mass was almost identical for the two cases based on
the maximum yield of olein. The stearin fraction left in both
cases was free of olein because it had no yellow color, which
indicated the presence of olein. At an olein yield of 5 percent
below the maximum olein yield, the yellow color was easily seen in

the stearin fraction from both procedures.

The separation data were taken as evidence of crystal size directly
affecting the amount of surfactant required for the separation.
The crystal sizes and SDS required are further analyzed in section

4.5.

61
4.4 Electrolyte Concentration Effects

The choice of sodium citrate was chosen as the electrolyte for the.
detergent solutions because of favorable status as a food additive
and as an antioxidant for fats (Dugan, 1976). The performance of
sodium citrate was no better or worse than other suitable

electrolytes such as sodium sulfate.

The role of the electrolyte, sodium citrate, concentration in the
separation step was investigated by using various electrolyte 9
concentrations for the separation step while holding other
variables constant. The weight ratio of the detergent solution to
partially crystallized tallow was 0.93. The weight of sodium
dodecyl sulfate was 0.45g per separation. The results are shown
in Figure 4.8. The olein yield increased as a function of the
electrolyte concentration until an electrolyte concentration of 5.0
percent was reached. The olein yield neither increased nor
decreased above an electrolyte concentration of 5.0 percent, but at
concentrations above 8.0 percent increased amounts of emulsified
olein were noted. The role of the electrolyte is further discussed

in section 4.6.

4.5 Analysis of Separation Results

 

The calculated total surface areas for the crystals of procedures

two and three show that procedure two produced crystals with 1.9

Percent Olein Yield

 

2 4 6 8 10 12
Percent Sodium Citrate

Figure 4.8 Olein Yield as a Function of the Electrolyte
Concentration in the Detergent Solution at 40 °C.
Tallow Weight was 170 g per sample; SDS Weight
was 0.45 g per sample; and the Weight Ratio of
Detergent Solution to Tallow was 0.93.

62

63
times as much surface area per unit volume as procedure three. The
numbers used for this calculation are shown in Table 4.1. The
optimum SDS levels for procedures two and three differed by a
factor of 2.5, based on the concentration, or weight, of surfactant
required. The ratio of optimum surfactant concentrations was
dependent on the specific concentrations used. Therefore, this
ratio could be different if a more continuous range of surfactant
concentrations was used. Since the calculated surface area for
procedures two and three differed by about the same factor as the
difference in the required weight of SDS, crystal surface area may
be the parameter determining the amount of surfactant required for

a successful fractionation.

To help determine the effect of crystal surface area on the
separation step, experiments were performed to determine where the
SDS was distributed after the dispersion of tallow and detergent
solution had formed. The method of Moore and Kolbesen (1956) for
the colorimetric determination of surfactant in the aqueous phase
was tried. The analysis was not successful on the partly refined
and commercially available tallow, because it contained either a
minor component or an emulsifier that solubilizes methylene blue

into chloroform (Rek and Appel, 1979).

Experiments were conducted to determine how much SDS in solution
was required to emulsify pure olein. It was found that a
concentration of 0.05 wt. percent SDS, with an electrolyte

concentration of 5.0 percent, would completely emulsify all the

64
olein present. The aqueous and olein mixture was centrifuged to
see whether this processing broke the emulsion and it was found

that it did not.

Wetting the solid is preferred by the SDS to remaining in solution
or emulsifying the oil. The above results supported this
conclusion. This means that very little surfactant was left in the
aqueous solution during a successful fractionation. The SDS was
almost all at the crystal surface. In cases where too much
surfactant was used, the surfactant completely coated the crystal
surface and the excess surfactant went into solution and caused

emulsified olein product.

4.6 Surface Coverage by SDS

 

The electrolyte, sodium citrate, inereased the ionic strength of
the detergent solution. This allowed the surfactant, sodium
dodecyl sulfate, to be aligned around the surface of the crystals.
The surfactant has a hydrocarbon end and an charged sulfate end.
Calculations show (See Table 4.2) that the surfactant molecules had
the necessary available area to have coated the crystal surface.
Figure 4.9 shows the SDS models used to calculate the crystal
surface area occupied by a molecule in the two extremes of possible
orientations. Standard molecular bond lengths were used in the

calculations (Streitwieser and Heathcock, 1976).

65
Table 4.2

Calculation of Crystal Surface Areagper Surfactant Molecule

 

 

 

 

Procedure Two Procedure Three
Crystal surface area,m2 28 x 10'12 95 x 10-12
Crystal weight/crystal, g 1.89 x 10.12 1.26 x 10'11
Number of crystals per
200 m1 aliquot 1.8 x 1013 2.7 x 1012
Total surface area, m2 500 260
Optimum gmoles SDS per
200ml aliquot 1.73 x 10’3 6.93 x 10'4
gmole SDS per m2 area 3.5 x 10-6 2.7 x 10"6
Crystal surface area
per molecule SDS, g2 47 61
Perpendicular area of SDS molecule 3.5 32

Parallel area of SDS molecule 40 82

 

 

 

H C H
a , e
Perpendicular

o\ ’0
o’s\o

Parallel

Figure 4.9 Simple Models for Calculating Surface Area
Occupied by an SDS Molecule in Two Different
Orientations.

66

67
The SDS molecules could be either perpendicular or partially
parallel to the crystal surface. The hydrocarbon end was in
contact with the crystal while the charged, sulfate end was
hydrated by the ionic solution. The charged layer surrounding the
crystals allowed them to wet into the aqueous phase. If the
electrolyte or surfactant concentration was too low the crystals
were not wetted into the aqueous phase completely. Low surfactant
concentration meant that there were not enough surfactant molecules
to coat the crystals. Low electrolyte concentration meant that the

surfactant molecules could not be aligned prOperly.

4.7 CMC for Detergent Solutions Used for Detergent Fractionation

Detergency theory requires the surfactant concentration to be above
the critical micelle concentration (Harris, 1958). Surface tension
was measured to determine the critical micelle concentration for
SDS solutions with and without electrolyte added. Figure 4.10
shows the surface tension measurements, for the two cases, versus
surfactant concentration. One case was with the 5.0 percent sodium
citrate in solution and one case was with no electrolyte in
solution. The critical micelle concentration for the solution with
only surfactant occurred at approximately 0.1 percent at the test
temperature of 40 °C. The critical micelle concentration for the
surfactant and electrolyte solution was at 0.01%. This shows that
the concentration of surfactant required for detergency action was
very low. This means that most of the surfactant was at the

crystal surface and the solution still had detergency properties.

Surface Tension, dyneS/cm2

‘8
BO

‘5
0

03
m

(a)
O

09
J8

0
l0

—-— SDS 81 Electrolyte
-— SDS

 

0.01 0.1
SDS Weight Percent

Figure 4.10 Surface Tension of Two Detergent Solutions
as a Function of SDS Concentration at 40 °C.

68

1.0

69
This supported the dependence of the separation solution strength

on the surface area of the crystals.

At sufficiently high initial SDS concentrations the crystal surface
became saturated or completely covered with the surfactant. The
amount of SDS required was dependent on the surface area of the
crystals, which depended on the crystallization conditions. The
added electrolyte helped orient the surfactant and provided an
ionic solution to wet the crystals into the aqueous phase.
Inefficient wetting occurred when the electrolyte or SDS
concentration was too low. If the concentration of SDS was too

high emulsified product resulted.

4.8 Olein yield as a function of the detergent solution to tallow

 

weight ratio

 

Figure 4.11 shows the results of experiments conducted using
various detergent solution to tallow weight ratios and various
surfactant weights. Detergent solution to tallow weight ratios of
0.46, 0.62, 0.78 and 0.93 were used to fractionate tallow samples
weighing about 170 g each. The results are plotted using the
weight of surfactant as the x-axis value. If the surface area was
important in determining the amount of surfactant required, all
aqueous volumes used should have given a maximum olein yield at
almost the same weight of surfactant. The expected result was
observed closely. The maximum olein yield was at 0.6 g SDS for the

0.62, 0.78 and 0.93 detergent solution to tallow ratios. The

96 Olein Yield

80

7O
60
50
4O
Detergent Solution to
30 Tallow Weight Ratio

0 0.46
O 0.62

I 0.78
V 0.93

 

.2 .4 _ .6 .8 1.0
Grams of Surfactant

Figure 4.11 Correlation of the Olein Yield with the Weight
of SDS Used. Each Curve Represents a Different Weight
Ratio of Detergent Solution to Tallow Weight. 5.0 2

Sodium Citrate was in the Detergent Solution.

70

71
maximum yield for the 0.46 ratio was at 0.45 g SDS. The olein
yield for the 0.62, 0.78 and 0.93 ratios at a surfactant weight of
0.45g was very close to the olein yield for those ratios at 0.60g
surfactant. These results supported the conclusion that the
surface area of the crystals determined the required amount of

surfactant.

Another interesting feature of the results shown in Figure 4.11 was
that the amount of aqueous solution used had a great effect on the
olein yield achieved. The olein yield was still rising at the
maximum detergent solution to tallow ratio used. The maximum olein
yields were 78.02, 75.3%, 70.62 and 70.12 at detergent solution to
tallow ratios of 0.93, 0.78, 0.63 and 0.46, respectively. The
reason for the observed effect might be related to the volume
available for the transfering crystals to occupy or to overall

dispersion viscosity effects.

Various detergent solution to tallow weight ratios were used to
show the effect of this ratio on the olein yield over a wide range
of values. The results of these tests are compiled in Figure 4.12.
It-shows that the olein yield rose until a detergent solution to
tallow weight ratio of 1.39 was exceeded. The yield of olein was
79.4 percent for the detergent solution to tallow weight ratio of
1.39. Four trials make up each of the data points used in Figure

4.12. The standard deviation of the points is shown in Table 4.3.

 

 

 

 

 

Percent Olein Yield

 

 

0.7 0.9 1.1 1.3 1.5 1.7
Detergent/ Tallow Weight Ratio

Figure 4.12 Olein Yield as a Function of the Weight Ratio
of Detergent Solution to Partially Crystallized Tallow.
0.45 g SDS per sample and 5.0 Z Sodium
Citrate was in the Detergent Solutions.

72

73
Table 4.3

Standard Deviations for Figgre 4.12

 

 

Weight Ratio Data Value Standard Deviation
0.80 75.85 0.70
0.95 77.13 1.96.
1.14 78.08 0.79
1.29 78.48 2.92
1.45 79.38 1.23

1.58 76.13 1.99

74
The drop in olein yield shown in Figure 4.12 when the detergent
solution to tallow ratio was increased to 1.54 should not be
interpreted to mean that increaSed detergent solution amounts cause
reduced olein yield. It is believed that the olein yield levels
off in this detergent solution to tallow ratio range. The lower
than expected olein yield was due to the collection methods
employed after the separation had been accomplished by
centrifugation. The olein was poured off from the stearin layer
which formed over the aqueous layer. As the aqueous layer became
larger the stearin layer in each centrifuge tube became thinner.
Eventually the layer became thin enough that pouring the olein off
the top of the stearin layer broke the stearin layer. This made
collecting all the olein difficult, which resulted in low olein

yields.

The olein yield reached a maximum because the stearin became
totally white at a detergent solution to tallow weight ratio of
about 1.40. If olein remained trapped in the crystals a yellowish
color was observed. This was observed at detergent solution to

tallow ratios below 1.00.

The relationship between the detergent solution to tallow weight
ratio and the olein yield was due to viscosity or volume available
for crystals effects. The increased aqueous phase could lower the
dispersion viscosity resulting in increased yields. The increased
aqueous phase could have provided a larger volume into which the

crystals were wetted. The rheological properties of the various

75
weight ratios of detergent solution to tallow were studied in light

of the above information.

4.9 Torque Measurements

 

The collection of rheology data was controlled by a data collection
program for the HP-85 microcomputer. Torque versus time data were
collected as shown in Figure 4.13. It shows torque versus time
data collected for a dispersion mixture with a weight ratio of
detergent solution to tallow of 0.50. These data were collected to
insure a uniform dispersion was formed and to check the time

dependent behavior of the suspension.

Although the dispersion was initially mixed, the initial mixing was
not complete and the decrease in the torque with time was caused by
the separation process itself and continued mixing. The viscosity
decreased for 20 to 30 minutes and then came to a constant value.
This was the observed pattern for all dispersions observed. The
appearance of the dispersion was less grainy after the rheology
testing. This indicated that oil-crystal groups or agglomerates

were broken apart during the mixing.

The total mixing time was 60 minutes which was equal to the time
normally allowed for the dispersion to be formed. The level torque
portion of the curve indicated that the rheological properties of
the dispersion could be evaluated independent of time. Therefore,

the rheological properties of the dispersion after the sixty minute

3.5

1.0

12 24 36 48 60
Time, Minutes

Figure 4.13 Torque Measured as a Function of Time for a
Dispersion at a Rotation Rate of 120 rpm.

76

77
torque test were representative of the dispersion as it entered the

centrifuging operation of a normal fractionation.

The rheological properties of the dispersion were recorded by
measuring the torque as a function of the rotation rate. Figure
4.14 shows the torque values collected as a function of the
rotational rate. The dispersion used was the same fluid that
generated the data for Figure 4.13. The data was then used to
determine an apparent viscosity at the chosen rotational rate of 60
rpm. Since the dispersion was non-Newtonian any viscosity was

dependent on the shear rate applied to the fluid.

The data were used to construct a plot of the ln(torque) versus the
ln(rotation rate). A computer program actually did the analysis.
The slope of the plot was determined and was the flow behavior
index n. The consistency coefficient was calculated using
previously determined standard fluid values, the torque at 60 rpm
of the dispersion and the n value of the dispersion to yield K

according to equation (4.1).

n n
x -(1~1(mk')y K mm (mm x) (4.1)
x X Y Y

With both K and n determined an apparent viscosity for the

diapersion being tested was calculated from equation (4.2).

r aura)“ (4.2)

3.0

5 10 15 20 25
Rate of Turn, S'1

Figure 4.14 Torque Measured as a Function of the
Rotation Rate for a Dispersion.

78

79

These calculations required that the dispersion was a power law
fluid in the range of rotational rates used and the relationship
between shear rate and rotational rate shown by Rieger and Novak
(1973) was valid. Both of these conditions were satisfied. The
values shown in Figure 4.14 yielded a straight line when plotted on
log-log graph paper. Steffe and Ford (1985) showed that the Rieger
and Novak (1973) relationship holds for the paddle mixer used for

these experiments.

Measurements were made for various weight ratios of detergent
solution to tallow at two different times with the same partially
crystallized tallow to determine the reproducibility of the torque
versus rotational speed measurements. It was important to use
tallow partially crystallized from the same batch with the same
time-temperature history because exactly duplicating a partial
crystallization was difficult. This is confirmed by data deveIOped

later in this paper (See Figure 4.17).

Figure 4.15 shows two sets of data for detergent solution to tallow
weight ratios of 0.50, 0.80, 1.14 and 1.50. The relative error,
based on the torque reading, between the two different data sets
for a given ratio ranges from one percent to about 28 percent at 60
rpm. However, the readings vary 125 percent from the low to high
ratios tested. It was noted that the dispersions had a large
amount of irregular solid particles present. Therefore, the

reproducibility of the measurements was thought to be good.

56

+ First Data Set
48 -— Second Data Set

&
O

*A
S!
x 32
E
Z
a? 24
a
E
.2

.5
O

 

6 12 18 24
Rotational Rate, Rad/s

Figure 4.15 Precision Measurements of the Torque as a
Function of the Rotation Rate of the Paddle Mixer. Two
sets of Data are Plotted to show the Reproducibility of

the Measurements. The Curves Represent Detergent Solution
to Tallow Weight Ratios of 1.50, 1.14, 0.80 and 0.50 from
Low to High Torque Values.

80

81
4.10 Apparent viscosity of the dispersions

The apparent viscosity for various detergent solution to tallow
weight ratio dispersions at a rotational rate of 60 rpm is shown in
Figure 4.16. The natural logarithm of the apparent viscosity was
plotted as a function of the olein yield. The olein yield
increased as the apparent viscosity was lowered. A dispersion
viscosity of about 0.08 Pa 8 or lower was needed for the maximum

olein yield of 79 percent.

There was clearly a relationship between the viscosity of the
dispersion before centrifugation and the olein yield obtained after
centrifugation. The variation in olein yield was produced by using
various weight ratios of detergent solution to tallow. This meant
that in addition to controlling the surfactant and electrolyte
concentrations, the viscosity of the dispersion formed had to be
controlled. This was accomplished by using the appropriate

detergent solution to tallow ratio in the dispersion step.

Accepted methods for correlating viscosity as a function of volume
percent solids could not be applied to this system (Jeffrey and
Acrivos, 1976). One of the methods was the one originally proposed
by Einstein (1956) for inert spherical particles of small size
compared to the fluid volume in which they were suspended. The
crystals of tallow were determined to be not suitable for this type

of correlation (eqn. 4.3).

Percent Olein Yield

 

0.1 0.3 0.9
Apparent Viscosity at 60 rpm, Pa s

Figure 4.16 Dispersion Apparent Viscosity Plotted Versus the
Olein Yield. Various Weight Ratios of Detergent Solution to
Partially Crystallized Tallow were used to Produce the
Apparent Viscosity Changes.

82

83
qr = 1+DS, (4.3)
The value of D for inert spheres in solution is 2.5. Correlating
the detergent solution-tallow dispersion in this manner yielded D
values of about 90. The aqueous phase was the continuous phase in
the dispersion. The viscosity was reduced by the viscosity of the
detergent solution-olein mixture that resulted from removing the
solids from the dispersion. This was chosen because it accurately

reflects the medium the crystals encounter.

The correlation did not hold for crystals produced by two different
partial crystallizations. Each partial crystallization produced
slightly different crystal sizes and numbers of crystals because of
the sensitivity of the nucleation and growth of the crystals to the
temperatures and processing times used. These slight processing
differences caused large variations in the rheological properties
of dispersions containing the various partially crystallized

tallows.

The tallow crystals were long thin needles with a L/D ratio of
about 75.. They existed as single crystals and in spherical
agglomerates. The spherical agglomerates were far from being inert
as the spiny protusions certainly caused them to interact with one
another. They may account for a larger effective volume fraction
than was ordinarily calculated due to voids in the spherulitic
agglomerates occupied by olein. Predicting these interactions and
effective solids volume was beyond the sc0pe of current rheological

information. Many questions about the behaviour of inert spheres

84
and cylinders in dispersions have still not been answered (Cheng,

1984; Chaffey and Porter, 1984).

The apparent viscosity has been plotted as a function of the

volume percent solids in Figure 4.17. Four different partial
crystallizations were used for the dispersion data gathered. Each
of the four different symbols represents one of the four different
partial crystallizations. Each partial crystallization gave
dispersions with their own rheological properties. The trend was
for the apparent viscosity to increase as the volume percent solids

increased.

The "protective colloids" suggested by Stein and Hartmann (1957)
were not tried in this research. They probably were used to adjust
the dispersion viscosity to avoid emulsion formation. It would
seem that increasing the aqueous viscosity was opposite to the
results obtained but the real function of the protective colloids
may have been to tie-up left over surfactant so that it could not

emulsify the olein.

1.12

    
  
  
 
 
 

in
o

b
a

Apparent Viscosity at 60 rpm, Pa 5
in '8-
no on

LL
0

8 12 16 20
Volume Percent Solids

Figure 4.17 The Apparent Viscosity of the Dispersion as a
Function of the Percent Solids in the Dispersion. Four
Different Sets of Data Shown; Each Corresponds to a
Separate Partial Crystallization.

85

CONCLUSIONS

Detergency phenomena was found to explain the separation mechanism.
At sufficiently high initial SDS concentrations, the crystal
surface became saturated or completely covered with the surfactant.
The amount of SDS required was dependent on the surface area of the
crystals, which depended on the crystallization procedure. The
added electrolyte helped orient the surfactant and provided an
ionic solution to wet the crystals into the aqueous phase. If the
concentration of SDS was too high emulsified olein product

resulted.

The crystals formed during the crystallization stage affected the
separation results. The required surfactant quantity was
determined to depend on the surface area of the crystals produced.
Increased amounts of surfactant were required for crystals with
more surface area. The relationship was directly proportional
meaning twice as much surface area required twice as much
surfactant. Use of a very rapid temperature drop (crystallization
procedure one) to crystallize the tallow resulted in more crystal
mass than more slowly cooled crystallization procedures
(crystallization procedures two and three). The larger crystal
mass affected the separation in two ways. Since more crystal mass
was present, the chemical difference between the two fractions was
less than for the fractions produced by a more slowly cooled
tallow. The surfactant required for separation was increased

86

87

because of the large surface area of the small crystals.

The electrolyte concentration affected the olein yield. Low
electrolyte concentration reduced the ability of the surfactant to
wet the crystals into the aqueous phase. A minimum level of sodium
citrate for maximum separation efficiency was 5.0 2 by weight of
the detergent solution. Using more electrolyte than necessary

showed a tendency to increase the formation of emulsified olein.

The ratio of detergent solution to tallow weight affected the olein
yield. If this ratio was too low maximum olein yield was not
obtained. The olein yield increased as the detergent solution to
tallow weight ratio was increased until the maximum olein yield was
reached. For the crystals from procedure three the optimum
detergent solution to tallow ratio was 1.4. Ratios above 1.4 gave

the same olein yield as the 1.4 ratio.

Maximum olein yield was achieved at a low apparent viscosity. An
apparent viscosity at 60 rpm of 0.8 Pa s was observed when maximum
olein yields were obtained. The apparent viscosity was dependent
on‘the crystals formed during the partial crystallization of tallow

and the weight ratio of detergent solution to tallow.

RECOMMENDATIONS

Research of the detergent fractionation has left some ideas for
exploration. These ideas relate to the development of the process
and its control as well as fundamental concepts that need to be
developed. One important factor, especially for process
development, is the world-wide slump in commodities prices. The
price of competing fats and oils have come down dramatically from
the early 1980 highs. This means that unless a superior result is
obtained by using a fractionated oil product, the economic

incentive to use the fractionated oil does not exist.

For some countries, where fractionation plants for palm oil and
beef tallow exist, economic limitations and national shortages make
continued fractionation activities rational. For other countries,
such as the United States, starting up of fractionation processes
is not economically warranted now. Because of the above reason,
recommendations concerning process piloting and process control are
not priority research topics in the opinion of this researcher.
They are included in the recommendations made because at some

future time they may be more desirable.

1. The size of the crystals, especially the diameter, needs
further investigation. The SEM could be used for the size
determination if a suitable crystal replication technique can be
developed. One suggestion would be to find a material which is

88

89
liquid at a temperature below 40 °C but could be solidified to form
a replica of a mass of crystals. A polymer of some type might be
the answer. It could be solidified and then the crystals could be

dissolved away. The replica could then be viewed by the SEM.

2. The effect of the crystallization procedure on the processing
requirements needs further investigation. A pilot scale partial
crystallization system is needed to test process variable control
as a function of the crystallization procedure. A scraped wall
vessel and a jacketed vessel are needed to develop various
crystallization results. The dispersion would then be formed in
the jacketed vessel and separation results could be estimated by
taking a small sample of the dispersion to be centrifuged, if large
scale centrifugation facilities were not available. The
crystallization requirements and the process control necessary to
ensure an efficient separation could be determined in detail from

this system.

3. The mixer viscometer might be used to control the dispersion
conditions. The olein yield dependence on the apparent viscosity
could be further tested with the pilot scale system described

above.

APPENDIX A.

Table A.1.
Separation Results from Three Different Crystallizations
Figure 4.7 Data.

Separation conditions: 1. Various surfactant concentrations, 2.
5.0% sodium citrate for electrolyte, 3. Weight ratio detergent
solution to tallow = 0.62 and, 4. Crystallization procedure one
used.

7-2-84 to 7—18—84 data

SDS Weight Olein Yield

 

(grams) (percent)
0.10 43.4
0.40 51.9
0.40 48.3
0.80 46.1
0.80 51.9
0.80 50.8
0.80 45.6

“ 1.10 50.4
1.40 42.2
1.60 46.6
2.00 0.0
2.40 30.0

90

91
Table A.1. continued
Separation conditions: 1. Various surfactant concentrations, 2.
5.0% sodium citrate, 3. Weight ratio of detergent solution to
tallow c 0.62 and, 4. Crystallization procedure two used.

SDS Weight Olein Yield

 

¥(grams) (percent)
0.10 39.5
0.30 62.2
0.40 76.2
0.50 79.5
0.60 75.0
0.70 77.2
0.80 70.4
0.90 72.9
1.30 67.6

1.70 49.2

Table A.1.

Separation conditions:

1.

92

Various surfactant concentrations,

continued

2.

5.0% sodium citrate, 3. Weight ratio of detergent solution to

tallow - 0.62 and, 4. Crystallization procedure three used.

 

SDS Weight Olein Yield
(grams) (percent)
0.05 58.1
0.10 56.9
0.15 71.7
0.20 77.8
0.20 77.6
0.30 75.6
0.35 74.1
0.40 75.2
0.45 71.4
0.50 72.6
0.50 68.6
0.60 61.5
0.80 61.3
1.20 completely emulsified

93
Table A.2.
Olein Yield as a Function of the Electrolyte Concentration in the
Detergent Solution at 40 °C.
Figure 4.8 Data.

Separation conditions: 1. Surfactant weight 0.45s. 2. Various
electrolyte concentration, 3. Weight ratio of detergent solution to
tallow - 0.93, 4. Average sample weight - 170g and, 5.
Crystallization procedure three.

1-14-85 to 1-16-85 Data.

 

Electrolyte Olein
Concentration Yield
2.0 Z 51.4 Z
4.0 Z 72.3 Z
6.0 Z 71.5 Z

8.0 Z 73.2 Z
10.0 Z 71.4 Z
12.0 Z 73.8 Z

1.0 Z 19.6 Z
3.0 Z 65.9 Z

5.0 Z 72.4%
- 7.0 Z 70.2 Z
9.0 Z 71.0 Z

11.0 Z 68.9 Z

Emulsion observed at 9 to 12Z concentrations.

94
Table A.3.
Preliminary Correlation of the Olein Yield with the Ratio of
Detergent Solution to Tallow Weight.
Figure 4.11 Data.
Separation conditions: 1. Various surfactant weights, 2. 5.0Z
sodium citrate, 3. Weight ratio of detergent solution to tallow, 4.
sample weight = 165 g and, 5. Crystallization procedure three.
9-18-84 to 10-5-84 Data

SDS Weight Aqueous Volume Olein Yield

 

(grams) (ml) (percent)
.1565 75 51.8
.1565 100 46.4
.1565 125 54.1
.1565 150 46.0
.3079 - 75 69.6
.3069 75 60.0
.3069 100 66.2
.3079 100 70.0
.3069 100 66.2
.3069 125 66.7
.3079 125 67.1
.3069 125 64.8
.3079 150 71.7
.3069 150 66.2

.4567 75 70.1

95
Table A.3 continued

SDS Weight Aqueous Volume Olein Yield

 

(grams) (ml) (percent)
.4567 100 67.7
.4567 125 74.1
.4567 150 77.0
.6262 75 58.4
.6262 100 70.6
.6102 125 75.3
.6102 150 78.0
.7717 75 59.7
.7717 100 66.3
.7691 125 65.3
.7691 150 69.8
.9174 ‘ 75 62.5
.9174 100 65.3
.9174 125 72.0

.9174 150 70.5

96
Table A.4
Calibration Data for Ring Method of Surface Tension Determination

for the Cenco Tensiometer.

 

Calibration of Tensiometer with teflon coated platinum ring.

 

Weight Tension (dynes/cm)
4-22-85 data
400 mg 30.8
500 mg 38.0
600 mg 46.0
700 mg 53.5
800 mg 61.2
900 mg 68.7
5-19-85 data
400 mg 30.8
500 mg 38.4
600 mg 46.0
700 mg 53.7
800 mg 61.5
900 mg 69.0
1000 mg 76.7

— Distilled Water measurements (dynes/cm) at 28 °C.

70.9 71.2
70.9 70.8
70.8 71.3

Actual value at 28 °C - 71.7 dynes/cm

97
Table A.5.
Surface Tension of a Detergent Solution as a Function of Surfactant
Concentration at 40 °C.
Figure 4.10 Data.
Ring method of surface tension determination.
Aqueous solution containing 5.0 Z sodium citrate and varying
amounts of sodium dodecyl sulfate.

SDS concentration Surface tension (dynes/cm)

5-19-85 data
0.1 33.4, 33.3, 33.3, 33.3, 33.4.
0.05 32.6, 32.7, 32.6, 32.7, 32.8.
0.3 32.3, 33.1, 32.9, 33.4, 33.1.
0.4 32.3. 32.5, 32.5, 32.3, 32.5.
0.5 32.5, 33.2, 32.7, 32.8, 33.0.
0.6 ' 32.8, 32.3, 32.4, 32.2, 32.6.
0.7 32.4, 32.3, 32.4, 32.2, 32.3.
0.8 32.2, 32.8, 32.8, 32.9. 32.9.
0.9 32.2, 32.3, 32.4, 32.4, 32.5.
1.2 32.7, 32.6, 32.4, 32.8, 32.7.
0.01 33.4, 33.8, 33.2, 33.3, 33.2.
0.2 32.9, 33.0, 33.1, 33.2, 33.2.

0.001 34.5, 34.2, 33.3, 33.5. 33.6.

98
Table A.5 continued.
Aqueous solution containing 5.0 Z sodium citrate and varying
amounts of sodium dodecyl sulfate.

SDS concentration Surface tension (dynes/cm)

5-23-85 data
0.002 52.5, 52.2, 50.8, 50.0, 49.5.
0.005 37.7, 36.5, 37.3, 36.7, 35.8.
0.0075 34.3, 35.6, 34.7, 34.5, 34.5.
0.009 34.2, 34.1, 34.3, 34.3, 34.4.
0.01 34.0, 33.8, 33.7, 33.8, 33.7.
0.011 34.0, 33.7, 33.7, 33.7, 33.6.
0.013 34.2, 34.0, 34.0, 33.9, 33.9.
0.016 34.1, 33.8, 33.6, 33.5, 33.6.
0.02 33.5, 33.2, 33.5, 33.4, 33.4.
0.03 33.8, 33.7, 33.7, 33.6, 33.5.
0.02 34.2, 34.0, 34.1, 33.9, 34.1.
0.04 33.7, 33.6, 33.4, 33.5, 33.6.

0.1 33.5, 33.2, 32.8, 32.6, 32.5.

Table A.5 continued.

Surfactant only solutions

SDS concentration

0.2

0.3

0.15

0.05

0.01

0.02

0.03

0.04

0.06

0.08

0.10

0.11

0.12

0.2

5-21-85 data

35

35

37.

37

36.

36.

36.

47.

39

39

36.

37

37.

35.

35.

36.

36.

35.

36.

99

Surface tension (dynes/cm)

.7,
.8,
2.
.0,
8.
3.
2.
0.
.4,
.4,
5.
.0,
2.
3.
8.
0.
5.
9.

7.

35.

36.

37.

37

36.

36.

36.

43

39.

38.

36.

37.

36.

35.

35.

35.

36.

35.

36.

3.
7.

2.

.0,

2.
0.

0.

.4,

3.

6.

35.

36.

37.

36.

36.

35.

35.

44.

39.

38.

36.

36.

36.

35.

35.

35.

36.

35.

36.

35.

36.

36

36.
36.
35.
35.
44.
38.
38.
36.
37.
36.
35.
35.
35.
36.
35.

36.

5.

2.

.9,

5.
3.
6.
9.
5.
8.
6.
3.
3.

3.

5.
5.
4.
8.

5.

35.

36.

37.

37.

36.

35.

36.

44.

38.

37.

36.

36.

36.

35.

35.

35.

36.

35.

36.

100
Table A.6.
Olein Yield as a Function of the Weight Ratio of Detergent Solution
to Partially Crystallized Tallow.
Figure 4.12 Data.
Separation conditions: 1. 0.45g sodium dodecyl sulfate, 2. 5.0%
sodium citrate, 3. various weight ratios of detergent solution to
tallow and, 4. 164 g average tallow sample and, 5. Crystallization
procedure three.

2-20-85 Data.

 

Weight ratio of detergent Olein
solution to tallow - Yield
0.85 75.1 Z

0.91 80.1 Z

1.12 79.4 Z

1.28 83.0 Z

1.41 . 80.2 Z

1.64 77.3 Z

2-6-85 Data.

0.74 75.2 Z

0.96 75.4 %

1.10 77.9 Z

1.31 76.8 Z

1.44 80.7 Z

1.54 75.7 Z

101

Table A.6 continued.

 

2-13-85 Data.
Weight ratio of detergent Olein
solution to tallow Yield
0.80 76.5 Z
0.96 75.4 Z
1.17 77.3 Z
1.33 75.2 Z
1.49 77.5 Z
1.57 73.1 Z
2-18-85 Data
0.83 76.6 Z
0.95 76.4 Z
1.16 77.7 Z
' 1.23 78.9 Z
1.44 79.1 Z
1.56 78.4 Z

102
Table A.7.
Torque Measured as a Function of Time for a Diepersion.
Figure 4.13 Data.
Detergent solution-tallow dispersion data obtained using mixer
viscometer.

Dispersion conditions: 1. Partially crystallized tallow by
procedure number three, 2. Weight ratio of detergent solution to
partially crystallized tallow 8 0.50 and, 3. The rotation rate used
was 120 rpm.

4-4-85 Data

3
Time (minutes) Torque (N-m x 10 )

 

.6512 2.772
1.918 2.579
3.181 2.412
4.444 2.287
5.705 2.178
6.971 2.102
8.234 2.042
9.497 1.986
10.76 1.948
“ 12.03 1.904
13.29 1.867
14.55 1.840
15.82 1.805
17.08 1.796
18.98 1.789

21.50 1.733

103

Table A.7 continued

24.66
27.82
30.98
34.13
37.29
40.45
43.61
46.76
49.92
53.08

56.24

4-4-85 Data

Time (minutes)- Torque (N-m x 103)

1.659
1.653
1.641
1.633
1.619
1.635
1.638
1.646
1.656

1.674

1.042 (60 rpm)

104
Table A.8.
Torque Measured as a Function of the Rotation Rate.
Figure 4.14 Data.
Dispersion data obtained using paddle mixer viscometer.
Dispersion conditions: 1. Partially crystallized tallow by

procedure number three, 2. Weight ratio of detergent solution to

partially crystallized tallow - 0.50 and, 3. Various rotation rates

 

were used. 4-4-85 Data
Rotation Rate Torque
(Rad/sec) (N-m x 103)
.5371 .2659
.2105 .2905
.3075 .2952
.4045 .3356
.6151 .3546
.8255 .4026
1.036 .4268
2.086 .5794
4.166 .7854
6.278 .9535
8.374 1.116
10.48 1.295
12.59 1.484
15.75 1.763
18.90 2.073
22.06 2.434
25.21 2.944

105
Table A.9.
Precision Measurements of the Torque as a Function of the
Rotational Rate of the Paddle Mixer.
Figure 4.16 Data.

Dispersion conditions: 1. Partially crystallized tallow by
procedure number three, 2. Various weight ratios of detergent
solution to partially crystallized tallow and, 3. Various rotation
rates were used. 3-11-85 Data

Speed of Rotation Torque(N-m)

 

Rad/sec Weight Ratio of detergent solution to tallow
0.50 0.50 0.80 ' 1.14 1.50
0.139 .000658
0.150 .000818 0.000160 0.000030
0.210 .000839 .000988 0.000205 0.000065 0.000031
0.403 .000971 .001147 0.000261 0.000083 0.000056
0.613 .001123 .001264 0.000304 0.000097 0.000051
0.824 .001150 .001321 0.000347 0.000125 0.000056
1.033 .001223 .001404 0.000397 0.000121 0.000063
2.084 .001585 .001703 0.000565 0.000171 0.000090
4.159 .001955 .002104 0.000720 0.000260 0.000149
"6.280 .002279 .002438 0.000872 0.000342 0.000203
8.362 .002546 .002729 0.001005 0.000417 0.000256
10.48 .002777 .002983 0.001130 0.000494 0.000312
12.58 .003007 .003216 0.001243 0.000564 0.000376
14.68 .003218 0.001377
15.74 .003542 0.000686 0.000481
18.88 .003598 .003848 0.001610 0.000816 0.000602

106
Table A.9 continued.
Dispersion conditions: 1. Partially crystallized tallow by
procedure number three, 2. Various weight ratios of detergent
solution to partially crystallized tallow and, 3. Various rotation
rates were used. 3-14-85 Data

Speed of Rotation Torque(N-m)

 

Rad/sec Weight Ratio of detergent solution to tallow
0.50 0.50 0.80 1.14 1.50

0.150 0.000202 0.000055 0.000047

0.166 0.000667 0.00157
0.210 0.000826 .000268 0.000047 0.000060 0.001815
0.404 0.000951 .000315 0.000064 0.000065 0.001953
0.614 0.001046 .000354 0.000074 0.000087 0.002104
0.825 0.001137 .000385 0.000089 0.000090 0.002224
1.04 0.001218 .000420 0.000098 0.000104 0.002350
2.09 0.001502 .000549 0.000144 0.000146 0.002766
3.14 0.001708 0.000182 0.003074
4.16 0.001875 .000722 0.000219 0.000215 0.003324
6.29 0.002172 .000868 0.000290 0.000284 0.003793
8.37 0.002426 .000997 0.000354 0.000341 0.004188
10.48 0.002656 .001124 0.000421 0.000408 0.004563
12.59 0.002866 .001234 0.000508 0.000476 0.004903
15.76 0.003155 .001405 0.000605 0.000588 0.005413
18.90 0.003432 .001586 0.000735 0.000713 0.005896
22.06 0.003706 .001760 0.000875 0.000851

25.22 0.003975 .001949 0.001024 0.000995 0.006852
28.4 .002153 0.001179 0.001145

107
Table A.10
Dispersion Apparent Viscosity Plotted Versus the Olein Yield
Figure 4.17 Data.
Partially crystallized tallow was fractionated and dispersion data.
was collected using various weight ratios of detergent solution to

partially crystallized tallow.

 

Weight Rad/ Torque Rad/ Torque n K Na
Ratio sec N-m sec N-m Pa 3
3-11-85 Data

0.50 1.033 1.22E—3 6.28 2.28E-3 .345 7.11 0.8013
0.80 1.035 3.97E-4 6.28 8.72E-4 .437 2.00 0.3069
1.14 1.036 1.21E-4 6.28 3.42E-4 .577 0.49 0.1201
1.50 1.037 6.27E-5 6.28 2.03E-4 .649 0.23 0.0711
0.50 1.036 1.40E-3 6.28 2.44E-3 .306 8.66 0.8571
3-14-85 Data
0.00 1.033 2.35E-3 6.28 3.79E-3 .265 15.44 1.3328
0.50 1.035 1.22E-3 6.29 2.17E-3 .320 7.37 0.7637
0.80 1.034 4.20E-4 6.29 8.68E-4 .402 2.24 0.3053
1.14 1.035 9.76E-5 6.29 2.90E-4 .603 0.38 0.1017
1.50 1.035 1.04E-4 6.29 2.84E-4 .554 0.44 0.0997
' Sheared tallow data
0.50 1.035 3.03E-4 6.28 7.29E-4 .486 1.42 0.2562
4-4-85 Data
0.50 1.036 4.27E-4 6.28 9.54E-4 .445 2.13 0.335
0.80 1.036 2.30E-4 6.28 5.15E-4 .446 1.15 0.181
1.14 1.037 1.98E-4 6.28 4.27E—4 .426 1.02 0.150

1.50 1.036 1.55E-4 6.29 2.55E-4 .276 1.00 0.090

108
Table A.10 continued.
Separation data for the various weight ratios.
Weight Olein
Ratio Yield

4-1-85 Separation Data (for 4-4-85 rheology data)

0.50 69.7%
0.80 72.4%
1.14 73.2%
1.40 77.6%

3-11-85 Separation Data (for 3-11 and 3-14-85 rheology data)

0.50 66.2Z
0.80 72.3%
1.14 78.1Z

1.50 79.4Z

109
Table A.11.
The Apparent Viscosity of the Dispersion as a Function of the
Percent Solids in the Dispersion.
Figure 4.18 Data.

Dispersion conditions: 1. Various weight ratios
of detergent solution to tallow used to generate
various solid fraction contents, 2. Various
crystallization procedures used.

Percent Apparent
Solids Viscosity

3-11-85 Data, Crystallization procedure three.

12.9 0.8013
10.6 0.3069
8.8 0.1201
7.5 0.0711
12.9 0.8571

3-14-85 Data, Crystallization procedure three.

20.0 1.3328
12.9 0.7637
10.6 0.3053
8.8 0.1017
7.5 0.0997

4-4-85 Data, Crystallization procedure three

12.9 0.335
10.6 0.181
8.8 0.150

7.5 0.0897

110
Table A.11 continued.
Percent Apparent

Solids Viscosity

 

4-15-85 Data Crystallization procedure two

12.9 0.311
10.6 0.211
8.8 0.164
7.5 0.158
12.9 0.288

5-23,24-85 Data, Crystallization procedure three.

20.0 0.922
20.0 1.014
15.2 0.237
15.2 0.241
12.9 0.272
10.6 0.172
10.6 0.177
8.8 0.124
8.8 0.105
7.5 0.0758

7.5 0.0758

APPENDIX B.

B.1 Table 4.1 Calculations.

 

1. Calculation of single crystal surface area.
The assumed crystal shape was cylindrical; The crystal diameter
was 0.3 microns and the crystal length was 30 microns.
S.A. = 3.14xDxL + 3.14xD2/4
= 3.14x0.3x30 + 3.14x0.3x0.3/4
= 28.26micron52 + 0.07 microns2

2 = 28 microns2

= 28.33 microns
2. Single crystal volume calculation
v = 3.14xD2xL/4
= 3.14x0.3x0.3x30/4

= 2.1 microns 3

3. Calculation of single crystal surface area per unit single
crystal volume.

Surface area per crystal/ Volume per crystal

8 28micron52/2.l microns3

B 13 microns"1

111

112

B.2 Table 2 Calculations.

 

1. Calculation of the weight per individual crystal
Crystal volume times crystal density equals the crystal weight.
Weight - 2.1microns3x(0.9g/cm3)x1cm3/1012microns3

- 1.89x 10'12g

2. Calculations for the number of crystals in a standard 200ml or
170g aliquot.

Number of crystals per 200 ml aliquot equals the crystal weight per
200 ml aliquot divided by the crystal weight per crystal.

No. crystals/200 ml 8 .20x170g/1.89X 10-12g

= 1.8x 1o13

3. Calculation of the total crystal surface area.
The number of crystals times the area per crystal equals the total
crystal surface area.

Total S.A. = 1.8x 1013x28 micronszx1m2/1012microns2

= 500 m2

4. Calculation of the optimum gmoles SDS per 200 ml aliquot.
Optimum gmoles SDS equals Optimum SDS weight divided by the gram
molecular weight of a SDS molecule.

gmoles SDS - .500g/288.1g/gmole

- 1.73x 10'3guoles sns

per

The

the

113
Calculation of the gmole SDS per m2 surface area.
- gmole SDS per 200 ml aliquot divided by total surface area
200 ml aliquot
= 1.73x 10'3/500 m2

= 3.5x 10-6gmole/m2

Calculation of the available surface area per molecule of SDS.
surface area per gmole SDS divided by Avogadro's number equals
crystal surface area per molecule.

. 1/(3.5x 10‘6x6.023x 1023)

= 4.8x 10'”...2 a 48 32

Calculation of the area occupied by a SDS molecule.

Case I. Perpendicular to crystal surface

Assume triangular orientation of hydrogen molecules at end of SDS

molecule. Use C—H bond lengths to estimate an area.

0
C-H bond length 1.095 A.

Using a triangle surrounding the three hydrogens by .5 times the C-

H bond length yields a triangle with area calculated as: =

O
0.5x2.86Ax2.46A - 3.5a2 (See Figure 4.9)

114
Case 11. SDS molecule parallel to the surface
Estimate by using a rectangle 11 carbon-carbon bonds long plus a
carbon-hydrogen bond by 2 carbon-hydrogen bonds wide.
A = 2.19Xx(11x1.543 + 1.093)

- 402

LIST OF REFERENCES

Adam, N.K., J. Soc. Dyers Colour., 53, 121 (1937).

 

Adam, N.K., and D.G. Stevenson, Endeavour, 12, 25 (1953).

Adamson, A.W., Physical Chemistry 2; Surfaces, 4th Ed., John
Wiley and Sons, New York, NY. (1982).

Braae, 8., Selection of Refining and Fractionation Methods
in the Manufacture of Edible Products from Animal
Fats, Paper presented at the 13th World Conference of
the International Society for Fat Research, August 30th
to September 4, (1976). '

Bussey, D.M., T.C. Ryan, J.I. Gray, and M.E. Zabik,
Fractionation and Characterization of Edible Tallow,
4; Food Sci., 46, 526 (1981).

Chaffey, C.E., and R.S. Porter, Steady Shear Flow of
Solutions of Rodlike Macromolecules, g; Rheol., 28, 249
(1984).

Chapman, D., The Polymorphism of Glycerides, Chem. Rev., 62,
433 (1962).

 

Cheng, D.C.-H., Flow properties of Suspensions of Inert
Spheres, Paper presented to the 9th International
Congress on Rheology, Acapulco, Mexico, October (1984).

Clarkson, C.E., and T. Malkin, Alternation in Long Chain
Compounds. Part II. An X-ray and Thermal Investigation
of the Triglycerides, J; Chem. Soc., 666 (1934).

DeFouw, C.L., M.E. Zabik, and J.I. Gray, Fractionated Edible
Beef Tallow as a Deep-Fat Frying Medium for French
Fries,_g; Food Sci.,_§§, 452 (1981).

Dugan, L., Lipids, Principles 2; Food Science, Part 1:,
Edited by O.R. Fennema, Marcel Dekker, New York,
pp 139-204 (1976).

DuNouy, P.D., A New Apparatus for Measuring Surface Tension,
.g, Gen. Physiol.,_1, 521 (1919).

Einstein, A., Investigations qg_the Theory q£_the
Brownian Movement, Dover, New York (1956).

 

115

116

Fort, T., H.R. Billica, and T.H. Grindstaff, Studies of
Soiling and Detergency. Part II. Detergency
Experiments with Model Fatty Soils, Text. Res. J.,
36, 99 (1966). ——' ’

Fort, T., H.R. Billica, and T.H. Grindstaff, Studies of
Soiling and Detergency, J. Amer. Oil Chem. Soc., 45,
354 (1968). -—'

 

Freud, B.B., and H.Z. Freud, A Theory of the Ring Method for
the Determination of Surface Tension, g; Amer. Chem.
Soc., 22, 1772 (1930).

 

 

Glassner, D.A., Process Analysis of the Aqueous
Fractionation of Tallow, M.S. Thesis, Michigan State
University (1983).

Glassner D.A., J.I. Gray, and E.A. Grulke, Detergent
Fractionation of Tallow, J. Amer. Oil Chem. Soc., 61,
1919 (1984). '_—

 

Glassner, D.A., and E.A. Grulke, The Effects of Surfactant
Concentration and Crystal Size on the Olein Yield from
the Detergent Fractionation of Tallow, Accepted by the
J. Amer. Oil Chem. Soc., (1986).

 

Goette, E.K., Theoretical Considerations of Detergency, J;
Coll. Sci., 4, 459 (1949).

 

Haraldsson, G., The Lipofrac System and Fractionation of
Palm Products, paper presented at the Palm Oil
Symposium, organized by the Society of Chemical
Industry, London (1974).

Harkins, W.D., and H.F. Jordan, A Method for the
Determination of Surface and Interfacial Tension
from the Maximum Pull on a Ring, J. Amer. Chem. Soc.,
52, 1751 (1930). __-

 

Harris, J.C., Detergent Application of the Measurement of
Critical Micelle Concentration, J;_Amer.Oil Chem. Soc.,
35, 670 (1958).

 

Hartley, G.S., Aqueous Solutions pf Paraffin Chain
Salts, Hermann et Cie., Paris (1936).

 

 

 

Hoerr, C.W., Morphology of Fats, Oils, and Shortenings, J;_
Amer Oil Chem. Soc., 37,539 (1960).

 

Holsinger, V.H., F.E. Luddy, C.S. Sutton, H.E. Vettel,
C. Allen, and F.E. Talley, An Oil Fraction from Edible
Beef Tallow as a Constituent of Whey-Soy Drink Mix, J;
Amer Oil Chem. Soc., 55, 473 (1978).

 

117

Jeffrey, D.J., and A. Acrivos, The Rheological Properties of
Suspensions of Rigid Particles, AIChE J;, 22, 417 '

Jewell, G.G., and M.L. Meara, A New and Rapid Method for
the Electrion Microscopic Examination of Fats, J; Amer.
Oil Chem. Soc., 41, 535-538 (1970).

 

Kling, w., Kolliod gt, _1_1_§, 37 (1949).

Kozempel, M.E., J.C. Craig, W.K. Heiland, S. Elias, and
N.C. Aceto, Development of a Continuous Process to
Obtain a Confectionary Fat from Tallow: Final Status,
g; Amer. Oil Chem. Soc., 28, 921 (1981).

 

Kreulen, H.P., Fractionation and Winterization of Edible
Fats and Oils, g; Amer. Oil Chem. Soc., 53, 393 (1976).

 

Krynitsky, J.A., and W.D. Garrett, The Use of Hydrophobic
Rings for the Measurement of Interfacial Tensions, g;
Colloid Sci., 18, 893 (1963).

 

Lange, H., Detergency, Part 1, ed. by W.G. Cutler and
R.C. Davis, Marcel Dekker, New York, NY (1972)

 

Lanza, F., DRP 191,238, (1905).

Luddy, F.E., J.W. Hampson, S.F. Herb, and H.L. Rothbart,
Development of Edible Tallow Fractions for Specialty
Fat Uses, J; Amer. Oil Chem. Soc., 59, 240 (1973).

 

Lutton, E.S., Review of the Polymorphism of Saturated Even
Glycerides, g; Amer. Oil Chem. Soc., 21, 276 (1950).

 

McBain, J.W., Mobility of Highly Charged Micelles, Trans.
Faraday Soc., 2, 99 (1913).

 

Metzner, A.B., and R.E. Otto, Agitation of Non-Newtonian
Fluids, AIChE fig, 3, 3 (1957).

Moillet, J.L., B. Collie, and W. Black, Surface Activity,
2nd Edition, Spon, London (1961).

 

 

Moore, W.A., and R.A. Kolbeson, Determination of Anionic
Detergent Concentration, Anal. Chem., 28, 161 (1956).

 

Neiditch, O. W. , Detergency, Part I, ed. by W. G. Cutler and
R. C. Davis, Marcel Dekker, Inc., New York, NY (1972)

 

Foot, 0., W. Dijkshoorn, A.J. Waighton, and C.C. Verburg,
Laboratory Separation of Crystals from Plastic Fats
Using Detergent Solution, A; Amer. Oil Chem. Soc., 52,
69 (1975).

 

118

Preston, W.G., Some Correlating Principles of Detergent
Action, 5; Phys. and Colloid Chem., 52, 84 (1948).

Rao, M.A., Measurement of Flow Properties of Food
Suspensions with a Mixer, 5; Texture Stud., 5, 533
(1975).

 

Rek, J.H.M., Process for the Separation of Triglycerides
in an Aqueous System, U.S. Pat. 4,049,687 (1977).

Rek, J.H.M., and A.C.M. Appel, Analysis of Traces of Anionic
Detergents in Oils and Fats, 5; Amer. Oil Chem. Soc., 55,
861 (1979).

 

Rieger, F., and V. Novak, Power Consumption of Agitators in
Highly Viscous Non-Newtonian Liquids, Trans. Instn.
Chem. Engrs., 51, 105 (1973).

 

 

Ryan, T.C., The Oxidation of Cholesterol in Tallow
Systems, M.S. Thesis, Michigan State University,
(1982).

Ryan, T.C., The Derivation, Stability and Functionality
of a Pourable Frying Oil from Beef Tallow, Ph.D.
Dissertation, Michigan State University, (1985).

Schott, H., Detergency, Part 1, ed. by W.G. Cutler and R.C
Davis, Marcel Dekker, Inc., New York, NY (1972).

Scott, B.A., Mechanism of Fatty Soil Removal, 5; App. Chem.,
15, 133 (1963).

 

Steffe, J.F., and E.W. Ford, Rheological Techniques to
Evaluate the Shelf-Stability of Starch-Thickened,
Strained Apricots, 5; Texture Stud., (1985).

 

Stein, W., and H. Hartmann, Separation of High Molecular
Organic Compound Mixtures, U.S. Pat. 2,800,493 (1957).

Stein, W., The Hydrophilization Process for the Separation
of Fatty Materials, 5; Amer. Oil Chem. Soc., 55, 171
(1968).

 

Streitwieser, A., and C.H. Heathcock, Introduction 52
Organic Chemistry, Macmillan Publishing Co., New
York, NY, (1976).

 

 

Swern, D., M.W. Formo, E. Jungermann, F.A. Norris, and
N.O.V. Sonntag, Bailey's Industrial Oil and Fats
Products, 4th Edition, John Wiley and Sons, New York,
NY (1979).

 

 

119

Taylor, H.H., F.E. Luddy, J.W. Hampson, and H.L. Rothbart,
Substitutability of Fractionated Beef Tallow for other
Fats and Oils in the Food and Confectionary Industries:

An Economic Evaluation, 5; Amer. Oil Chem. Soc., 55, 491
(1976).

USDA ERS, Oil Crops Outlook and Situation Report, May
(1984).

Zuidema, H.H., and G.W. Waters, Ring Method for the

Determination of Interfacial Tension, Ind. Eng. Chem.,
Anal. 52;, 15, 312 (1941).