Heat transfer variables affecting process determinations in conduction heating institutional size retort pouches

MISSING IMAGE

Material Information

Title:
Heat transfer variables affecting process determinations in conduction heating institutional size retort pouches
Physical Description:
xiii, 187 leaves : ill. ; 28 cm.
Language:
English
Creator:
Peterson, Wayne Roger, 1954-
Publication Date:

Subjects

Subjects / Keywords:
Food -- Packaging   ( lcsh )
Food containers   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1984.
Bibliography:
Includes bibliographical references (leaves 180-186).
Statement of Responsibility:
by Wayne Roger Peterson.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000467825
notis - ACN2241
oclc - 11630305
System ID:
AA00003424:00001

Full Text














HEAT TRANSFER VARIABLES AFFECTING PROCESS DETERMINATIONS IN
CONDUCTION HEATING INSTITUTIONAL SIZE RETORT POUCHES




By

WAYNE ROGER PETERSON


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


1984

















ACKNOWLEDGEMENTS


The author would like to give special thanks to his major

advisor, Dr. J. Peter Adams, for his guidance, friendship, and

constant moral support throughout this research program.


The author aslo wishes to thank his supervisory committee, R.H.

Schmidt, J.F. Gregory, C.D. Baird, A.A. Teixeira, and K.V. Chau, for

their help with materials, suggestions, and support.


Special thanks go to all the new friends found at the University

of Florida and to the faculty and staff of the Food Science and Human

Nutrition Department, who helped provide a pleasant working

atmosphere.


Finally the author wishes to give sincerest thanks to his

parents, who provided constant encouragement, support, and

understanding during this endeavor.
















TABLE OF CONTENTS

PAGE
ACKNOWLEDGEMENTS ................................................... ii

LIST OF TABLES ....................................................vi

LIST OF FIGURES...................................................viii

KEY TO SYMBOLS....................................................... ....... .. x

ABSTRACT ......... o .. ...... ...... ..... .......... .... .................xii

CHAPTER

I. INTRODUCTION ........................... ......................1

II. LITERATURE REVIEW ......... ......................... ...............

Heat Penetration Tests for Retort Pouches....................6

Thermocouple Errors........................................7

Introduction of the Temperature Measuring Device into
the Retort Pouch........................................ 8

Placement of the Temperature Sensing Device................10

Modeling the Retort Pouch.................................12

Thermal Processing Determinations ............................15

Critical Processing Factors for Retort Pouches..............20

Food Product Type..........................................21

Entrapped Gas............................................22

Heating Media Used for Retort Pouch Processing............26

Heat Penetration Through Retort Pouch Material............29

Racking Systems Used for Retort Pouches....................29

III. HEAT TRANSFER RESISTANCES OF RETORT POUCHES...................31









Introduction ... .................. .........................31

Materials and Methods......................................33

Polycarbonate Slab Preparation ............................33

Slab Processing.................................. ... 36

Thermal Diffusivity of Polycarbonate Slabs................38

Heat Transfer Resistances Due to Pouch Material,
Rack Type, and Flow Rate................................38

Entrapped Air............................................. 42

Results and Discussion............................. .... 43

Thermal Diffusivity of Polycarbonate Slabs................43

Heat Transfer Resistances Due to Pouch Material,
Rack Type, and Flow Rate ...............................45

Entrapped Air........................................... 50

Summary and Conclusions...................................62

IV. THERMOCOUPLE GROUNDING IN RETORT POUCHES......................64

Introduction...............................................64

Materials and Methods......................................65

Thermocouple Types.......................................66

Grounding Methods........................................68

Processing..............................................69

Experimental Design......................................70

Data Analysis............................................72

Results and Discussion.....................................73

Pressurized Water Heating................................77

Steam-air Heating........................................82

Cooling Data Analysis....................................85

Summary and Conclusions....................................90










V. TEMPERATURE SENSOR PLACEMENT IN RETORT POUCHES................95

Introduction.............................................95

Materials and Methods......................................96

Results and Discussion....................................107

Statistical Model........................................107

Worst Case Situation .....................................117

Summary and Conclusions...................................122

VI. SUMMARY AND CONCLUSIONS......................................124


APPENDIX A:


APPENDIX B:


APPENDIX

APPENDIX


APPENDIX E:


APPENDIX F:


APPENDIX G:


APPENDIX H:


PROGRAM FOR INFINITE SLAB EXACT SOLUTION,
INFINITE BIOT NUMBER..................................128

PROGRAM FOR RECTANGULAR PARALLELPIPED EXACT
SOLUTION, INFINITE BIOT NUMBER .......................130

PROGRAM TO CALCULATE HEAT PENETRATION PARAMETERS........133

PROGRAM TO SIMULATE HEAT PENETRATION INTO A
RETORT POUCH WITH ENTRAPPED AIR PRESENT..............143

PROGRAM TO FIT TEMPERATURE DISTRIBUTION DATA TO
CUBIC FUNCTION......................................150

FINITE DIFFERENCE PROGRAM TO SIMULATE A THERMAL
PROCESS DETERMINATION; FINITE BIOT NUMBER............156

FINITE DIFFERENCE PROGRAM TO SIMULATE A THERMAL
PROCESS DETERMINATION; INFINITE BIOT NUMBER..........165

PROGRAM FOR INFINITE SLAB EXACT SOLUTION, FINITE
BIOT NUMBER...........................................174


REFERENCES.........................................................180

BIOGRAPHICAL SKETCH...............................................187
















v
















LIST OF TABLES


Table 3-1.


Table 3-2.


Table 3-3.




Table 3-4.


Table 3-5.





Table 4-1.


Table

Table

Table

Table


4-2.

4-3.

4-4.

5-1.


Table 5-2.


Table 5-3.



Table 5-4.


PAGE
Vertical position of thermocouple holes in
polycarbonate slabs ...................................34

Average flow rates used to test effect of pouch
material and rack type on heat transfer................40

Division of degrees of freedom for experimental
design used to examine effects of the presence
of pouch material, confining rack design, and
flow rate on heat transfer............................41

Thermal diffusivity values obtained for the
polycarbonate slabs....................................44

Effect of entrapped air and maximum expansion
allowed by confining rack system on F received
by a 1.5 inch polycarbonate slab with a process
time of 70 minutes (includes 3.0 minute come
up time)............................................. 61

Experimental design for first replicate of
pressurized water cook................................71

Mean fh-values found for pressurized water heating.....83

Mean fh-values found for steam-air heating.............86

Mean f -values found for water cooling.................92

Verification of finite difference program
accuracy with actual data..............................99

Levels of factors used in regression model
development..................... ................... 100

Combinations of factor levels used for rotatable
central composite design for the finite Biot
number model........................................ 103

Combinations of factor levels used for rotatable
central composite design for the infinite Biot
number model............ ................. ............105









Table 5-5.


Model for process time prediction at any position
in a slab with a finite Biot number..................108


Table 5-6. Model for process time prediction at any position
in a slab with an infinite Biot number................110
















LIST OF FIGURES:


Figure 3-1.


Figure

Figure

Figure


3-2.

3-3.

3-4.


Figure 3-5.



Figure 3-6.



Figure 3-7.



Figure 4-1.


Figure 4-2.


Figure 4-3.


Figure 4-4.


Figure 4-5.


Figure 4-6.


Figure 4-7.


PAGE
Thermocouple placement in a 1.0 inch polycarbonate
slab...............................................35

Effect of Reynolds number on fh for each rack type....48

Effect of Reynolds number on fc for each rack type....49

Effect of 32 ml air on heating curve of a 1.0 inch
polycarbonate slab (center thermocouple position).....52

Heating curves obtained for all 4 positions in a
1.0 inch polycarbonate slab (slab #1) with 32 ml
air present..........................................55

Temperature profile within a 1.5 inch polycarbonte
slab at various processing times. Curves are from
cubic function predicted by least squares.............58

Predicted position of cold spot during processing
of a 1.5 inch polycarbonate slab with 98 ml of
air present..........................................59

Heating curve with residual sum of squares equal
to 0.57...............................................74

Heating curve with residual sum of squares equal
to 0.75...............................................75

Heating curve with residual sum of squares equal
to 3.00..................................... ..... 76

Residual sum of squares obtained for pressurized
water heating........................................78

Heating curve illustrating instability in temperature
measurement......................................... .. 80

Residual sum of squares obtained for heating with
90% steam, 10% air....................................84

Residual sum of squares obtained for cooling.........87


viii









Figure 4-8.


Figure 4-9.


Figure 5-1




Figure 5-2




Figure 5-3



Figure 5-4





Figure 5-5






Figure 5-6


Cooling curve with residual sum of squares equal to
10.0.................................................89

Cooling curve with residual sum of squares equal to
53.2.................................................91

Interaction of Biot number with position on relative
process time. All factors at midpoint of design
except h-value which was varied to obtain different
Biot numbers.......................................113

Interaction of half slab thickness with position on
relative process time. Curves are at equal Biot
numbers (10.8), all other factors are at the design
midpoint............................................115

Interaction of thermal diffusivity with position
on relative process time. All other factors are
at the design midpoint..............................116

Actual process time versus actual position for
0.75 and 0.25 inch half slab thicknesses. Extreme
factor levels were used to achieve maximum
sensitivity of determined process time to position
of temperature measurement........................118

Center point F-value found when process was
based on temperature measurement at different
positions in 0.75 and 0.25 inch half slab
thicknesses. Extreme factor levels were used to
achieve maximum sensitivity of determined process
time to position of temperature measurement........120

Position where 1.0 minute drop in determined
process time occurred as a function of slab half
thickness.........................................121
















KEY TO SYMBOLS

A Half slab thickness

a Half slab width

B Thermal process time. Total time at processing temperature.

b Half slab length

Bi Biot number = h*A/k

(Bi) in Incremental Biot number = h*Ax/k

c Height of air at slab center

D Time at any temperature to destroy 90% of a microorganism

F Equivalent minutes at a reference temperature, with respect
to destruction of microorganisms.

f Time for straight line portion of heating or cooling curve
plotted on semi-log graph paper to transverse 1
log cycle.

g Difference between T and temperature of the food at the
container coldspo at the start of cooling.

h Surface heat transfer coefficient

j Lag factor for heating or cooling curve

k Thermal conductivity

M (a*At)/(x)2

Pos Relative position in slab = (inches from center)/A

Re Reynolds number

t Time

T Temperature

V Volume of air present










x Position on width axis (-a
y Position on length axis (-b
z Height of air over any point (x,y) on slab surface (0
Z Number of degrees Fahrenheit required to reduce the D-value
by 90%

a Thermal diffusivity

At Time increment

Ax Distance increment

Subscripts

c Cooling

cd Come down

cu Come up

h Heating

i Initial

m Heating medium

pi Pseudo initial temperature for heating or cooling.
Intersection of the straight line of a heating or
cooling curve with zero time.

r Retort

s Surface

0 Current position in numerical equations

1 Position one increment closer to the center of the slab
than the current position.

-1 Position one increment closer to the surface of the slab
than the current position.

Superscripts

0 Time one increment previous to current time in numerical
equations.

1 Current time in numerical equations

















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


HEAT TRANSFER VARIABLES AFFECTING PROCESS DETERMINATIONS IN
CONDUCTION HEATING INSTITUTIONAL SIZE RETORT POUCHES

By

Wayne Roger Peterson

April, 1984

Chairman: J. P. Adams
Major Department: Food Science and Human Nutrition


One and 1.5 inch polycarbonate slabs were used to determine the

resistances of components in an institutional sized retort pouch

system to heat transfer. The components evaluated were 1) pouch

material, 2) rack design, 3) flow rate of processing medium, and 4)

entrapped air. Pouch material and perforated rack designs offered no

significant resistance to heat transfer. A solid rack design resulted

in variable results and reduced heat transfer rate. The solid rack

design reduced heating rates more than cooling rates. Entrapped air

in pouches greatly reduced heating rates and caused an upward

migration of the cold spot.


Four different types of thermocouples and three different

grounding methods were evaluated to determine the most consistent and

accurate temperature response in heat penetration tests of pouches in

pressurized water and steam-air retorts. Criteria for evaluating the

xii











response were the residual sum of squares from non-linear regression

analysis and fh and f values derived from the heat penetration

data. Temperature measurement errors were found using both types of

heating media for certain thermocouple types and grounding methods,

while cooling data were found to be acceptable for most combinations.

The best response was found with a thermocouple shielded with a

stainless steel tube in which the measuring junction was electrically

isolated from the tube.


The effect on process time of the positioning of the temperature

measuring device in institutional size retort pouches was examined for

low acid pouched foods Numerical techniques were used to predict

process time by the improved general method at various positions in an

infinite slab. Regression models were developed for the prediction of

process time and were used to identify the factors which significantly

interacted with position of the temperature measuring device.

Positioning was not critical within a small range near the center of

the slab. A positioning accuracy of +1/8 inch around the center of

pouches having a thickness of greater than 0.5 inch did not

significantly change the determined process time. This range

increased as slab thickness increased.


xiii
















CHAPTER I
INTRODUCTION


Thermal stabilization of food to produce a shelf stable product

was first accomplished in the late 1700's by Nicolas Appert. Through

the years a great deal of experience and research have resulted in the

canned food supply of today. Due to this experience and research, the

procedures for producing a shelf stable, commercially sterile canned

food product have been well established. Retort pouches are a

relatively new form of packaging. Initial experimental work was

completed in the 1950's. Much of the data base and concepts used in

producing and processing canned food can be applied directly to retort

pouches; however, there are many retort pouch processing parameters

which differ from those of cans. These must be identified and

controlled to insure the consistent production of a commercially

sterile, sate pouched food.


One of the major advantages of retort pouches over cans is that

the thin heating profile of the retort pouch requires a shorter

process time and thus the quality of a pouched product can be much

higher than that of a comparable canned food product.


To take advantage of the thin profile and reduced process times,

the retort pouch processor must avoid overprocessing. To avoid

overprocessing and at the same time insure a safe product for the









consumer, the variables associated with thermal processing must be

known and controlled. Unfortunately, the thermal processing variables

associated with processing pouches are greater in number and more

critical than those associated with can processing.


The purpose of this research was to examine variables associated

with the processing of institutional size retort pouches and to

identify those having the greatest effect on the process time required

to produce a commercially sterile pouched food product.


The research was conducted in three phases:

1) Heat transfer resistances encountered from the heating

medium to the surface of the food in the pouch.

2) Temperature measurement errors associated with retort

pouches.

3) Temperature sensor placement errors during heat

penetration trials in retort pouches.
















CHAPTER II
LITERATURE REVIEW


The retort pouch is a flexible food package designed to withstand

the temperatures associated with commercial sterilization of food

products. Initial investigations of the potential for flexible films

for commercial sterilization of food was conducted by Hu et al.

(1955). Results of this study indicated flexible films were available

that would withstand the usual sterilization temperature of 250 F used

for low-acid foods. Since this initial investigation, considerable

research has been conducted to obtain a viable package which would

meet the strength, integrity, and safety requirements needed for a

shelf-stable, commercially sterilized food product. The result of

this research was the current retort pouch material which usually

consists of a 3-ply laminate of polyester for strength, aluminum foil

for barrier properties, and a polyolefin for heat sealability. The

development of the retort pouch was a joint effort involving both the

government and private industry, most notably U.S. Army Natick R & D

Command, The Continental Group, Inc., and Reynolds Metal Co.

(Mermelstein, 1978).


According to Mermelstein (1978), there are a number of advantages

which retort pouches have over cans and glass jars.











1. The retort pouch has a thinner profile, therefore it

requires less processing time and there is less

peripheral overprocessing.

2. The final pouched, processed product can be reheated

rapidly by placing the pouch in boiling water.

3. The retort pouch can be easily opened using scissors or

by tearing the top off.

4. The weight of retort pouch material is less than that of

a comparable number of cans or glass jars.

5. Less storage space is required for the retort pouch

material.

6. Less energy is required to manufacture retort pouch

material.


The thin profile of the pouch compared to cans may result in

shorter process times. Numerous investigations have compared the

quality of products processed in retort pouches and cans. Sensory

properties of foods packed in retort pouches have been found to be

equal or superior to the same foods packed in cans of comparable or

smaller net weight. This quality advantage has been described for

fruits and vegetables (Thorpe and Atherton, 1972; Gould et al.,

1962); cream style corn (Tung et al., 1975; Gould et al., 1962);

chicken (Lyon and Klose, 1981); and various seafood products (Adams

et al., 1983: Chia et al., 1983). Nutrient retention in pouches has

also attracted interest. Castillo et al. (1980) used a rectangular

parallelepiped model to predict nutrient retention in a conduction










heating food. Abou-Fadel and Miller (1983) found green beans and

cherries packed in retort pouches retained a greater percent of both

thiamin and ascorbic acid than the comparable canned products. Retort

pouched sweet potato puree exhibited enhanced retention of thiamin and

riboflavin when compared to cans of equal volume (Rizvi, 1981).


Williams et al. (1982) stated that retort pouches had a

significant advantage over cans when transportation costs were

considered. This advantage was the result of the lighter weight of

the empty retort pouch and reduced syrup/brine requirements of retort

pouched fruits and vegetables. Steffe et al. (1980) found the total

energy requirements for the production of retort pouch material and

wholesale distribution to be 14% less than that required for a similar

canning operation. Although retort pouches may require less

processing time than comparable cans, this factor did not play a

significant role in the overall cost comparison between retort pouches

and cans (Williams et al., 1983). A retort pouch line compared very

favorably to a canning line when operating costs were considered;

however, the large initial capital expenditure could offset these

advantages (Williams et al., 1983).


In addition to retail size retort pouches which contain

approximately 8 ounces of product, there may be a viable market for

institutional size retort pouches (Whelan and Whitehead, 1980).

Morris (1981) predicted that the institutional size retort pouch would

show the most rapid growth in the retort pouch market. Institutional

size retort pouches usually contain approximately 5 Ibs, although the










actual size of the pouch depends on the product. In many cases

institutional size retort pouches are in direct competition with the

No. 10 can (Badenhop and Milleville, 1980). The pouches can be as

large as 12 by 18 inches and are usually 0.75 to 1.5 inches thick

(Milleville, 1980a). The institutional size retort pouch market may

be easier to break into than the retail size retort pouch market due

to a number of factors. The filling speed of No. 10 cans is much

slower than smaller cans, thus the discrepancy between the filling of

No. 10 cans and the institutional size retort pouch is not as great as

it is for the retail size pouch. Also, user education on use and

storage of the institutional size retort pouch would be less than that

required for the retail size retort pouch (Mermelstein, 1976).


Heat Penetration Tests for Retort Pouches


Conducting a heat penetration trial with retort pouches is

essentially identical to the procedure used for cans. There are a

number of excellent discussions on the subject of heat penetration

tests for cans (Ball and Olson, 1957; Stumbo, 1973; Pflug, 1975;

Alstrand and Ecklund, 1952); therefore, the details of the actual

procedure will not be discussed here. Even though the basic procedure

for conducting heat penetration tests is the same for retort pouches,

there are a few areas that present difficulties not experienced when

dealing with cans. The major difficulties are experienced in

temperature measurement, introduction of the temperature measuring

device into the retort pouch, and location of the temperature

measuring device at the cold spot of the retort pouch.










Thermocouple Errors


Davis et al. (1972) noted thermocouple measurement errors during

thermal processing of retort pouches that were not experienced during

processing of cans or glass jars. Errors as large as +20 F were found

when thermocouple data were compared to the retort mercury thermometer

temperatures. Pflug (1975) also discussed temperature measurement

errors during heat penetration tests with retort pouches. Pflug

(1975) attributed these errors to the ability of the retort pouch to

produce stray electrmotive forces (emf) in a water or steam

environment above 140 F. Both Davis et al. (1972) and Pflug (1975)

alleviated temperature measurement problems by using enamel-coated

thermocouple wire and by grounding the measuring junction to the case

of the recording potentiometer.


Peterson and Adams (1983) successfully used a different grounding

method. The stainless steel sheath of a thermocouple was grounded to

the frame of the retort during processing. This method was

accomplished inside the retort and did not require a ground wire to be

extended outside of the retort. The thermocouple used was sheathed in

a stainless steel tube, but the tube was not a true electrical shield

since the measuring junction of the thermocouple was in electrical

contact with the tube.










Introduction of Temperature Measuring Device into the Retort Pouch


Introduction of the temperature measuring device into the retort

pouch is more difficult than it is in the can due to the flexible

nature of the pouch material. The use of the stuffing box or packing

gland appears to be the most popular method of inserting the

temperature sensor into a retort pouch (Berry and Kohnhorst, 1983;

Peterson and Adams, 1983; Spinak and Wiley, 1982; Castillo et al.,

1980; Tung et al., 1975; Davis et al., 1972; Turtle and Alderson,

1971; Goldfarb, 1970; Pflug et al., 1963). Generally, stuffing

boxes obtain a hermetic seal by compression of rubber gaskets at the

opening into the pouch and around the thermocouple wire or sheath. A

hermetic seal is not guaranteed when a stuffing box is used, since the

rubber gaskets must be tight enough to insure a seal. Overtightening

can create problems by breaking small gauge thermocouple wires.

Another problem occurs when multiple wires enter the same stuffing

box: a hermetic seal is difficult to obtain in this situation.


Stuffing boxes are commonly made of metal or nylon. A metal

stuffing box can affect heat penetration due to the heat sink effect.

Care must be taken to insure the stuffing box is far enough away from

the temperature measuring device to avoid this effect. Spinak and

Wiley (1982) determined that a distance of 1.5 inches between a brass

stuffing box and the measuring junction of a thermocouple was adequate

to insure a negligible effect on heat penetration data.










The caulk method of introducing the temperature measuring device

into a pouch involves introducing thermocouple wire through a small

hole in the pouch material and sealing the hole with silicon caulk

(Roop and Nelson, 1982: Spinak and Wiley, 1982) or adhesive (Thorpe

and Atherton, 1972). There are a number of disadvantages to this

method. First, a large amount of time is required for the caulk to

cure. Second, to obtain a hermetic seal the caulk must adhere very

well to both the pouch material and the thermocouple wire entering the

pouch. If the thermocouple wire used is insulated with teflon,

adherence is very difficult to obtain.


Wornick et al. (1960') used a method similar to the caulk method.

A small slit was made in one of the pouch edges, thermocouple wire was

inserted through the slit, then the slit was sealed by means of two

rubber gaskets and a screw clamp.


Pflug et al. (1963) introduced thermocouple wires into the pouch

through the pouch seal. Insulation was stripped off the thermocouple

wires where they were to be incorporated into the seal, then the wires

were cleaned and coated with 3 to 5 layers of a lacquer that would

heat seal with the pouch material. With 30 gauge wire it was

necessary to separate the thermocouple wires by at least 0.25 inch to

avoid bridging of the pouch seal across the two wires.










Placement of the Temperature Sensing Device


To obtain valid heat penetration data for establishing a process

for a thermally stabilized low acid food product, the temperature

sensing device must be located in the slowest heating point of the

container. This is difficult when dealing with retort pouches due to

the flexible nature of the pouch (Milleville and Badenhop, 1980). A

number of methods have been developed to deal with this problem.


When a pouch contains large particulates, slices of meat, fish

fillets, etc., the measuring junction of the thermocouple is commonly

embedded in a particle or placed between slices of the food (Adams et

al., 1983: Spinak and Wiley, 1982; Berry, 1979; Pflug, 1975;

Thorpe and Atherton, 1972; Turtle and Alderson, 1971; Pflug et al.,

1963). With these products one can expect greater variation in heat

penetration data than with homogeneous products due to the natural

variation found in particle size and thickness of food slices.


The spacer block or spacer disk method uses blocks or disks

corresponding to the desired thickness of the retort pouch. The

blocks or disks have a center hole through which the thermocouple is

inserted and therefore the block or disk acts as a support to maintain

the desired thermocouple position (Adams et al., 1983; Peterson and

Adams, 1983; Milleville and Badenhop, 1980; Thorpe and Atherton,

1972). The disks or blocks may be made of any material which will

maintain its integrity at processing temperatures.










This method works well for confined pouches. Problems may be

encountered with unconfined pouches or pouches containing entrapped

air due to expansion of the pouch during processing. The thermocouple

remains where it was originally placed; therefore, any movement of

the cold spot due to pouch expansion is not adjusted for.


If wire thermocouples are used with the spacer block method, the

disks or blocks must be wide enough to minimize the possibility of

their tipping over and thus changing the position of temperature

measurement. This problem is minimized if rigid thermocouples are

used.


The blocks or disks should be situated far enough away from the

measuring junction of the thermocouple to insure that the block or

disk does not affect the heat penetration data (Milleville and

Badenhop, 1980).


The spring method for positioning the temperature measuring

device inside a retort pouch uses a wire spring of the same wound

diameter as the desired thickness of the retort pouch. The

thermocouple wire is clamped to the ends of the spring and the

measuring junction is doubled back and positioned in the center of the

spring (Tung et al., 1975; Pflug, 1975; Davis et al., 1972).

Obviously, this method was designed for wire thermocouples and would

not be applicable to rigid thermocouples. As in the spacer block

method, the thermocouple position is constant and any expansion of the

pouch during processing is not accounted for in thermocouple position.










The self-centering strip method consists of a strip of retort

pouch material which is folded into the shape of a "V" with tabs on

both ends. The tabs are heat sealed to the top and bottom of the

inside of the pouch and the thermocouple is placed through the apex of

the "V". The "V" of pouch material is somewhat fragile; therefore,

to avoid deformation this method should be used only with fairly large

gauge wire thermocouples and liquid type products (Berry and

Kohnhorst, 1983; Roop and Nelson, 1982; Spinak and Wiley, 1982;

Thorpe and Atherton, 1972; Turtle and Alderson, 1971; Pflug et al.,

1963). The advantage of this self-centering method is the ability to

maintain the thermocouple at the center of the pouch when expansion of

the contents occurs during processing.


The "H" method (Nioras, 1982) uses a polypropylene support in the

shape of an "H" which fits inside a retort pouch. The crossbar of the

"H" is used to maintain the thermocouple at the desired location. One

disadvantage of the "H" method is that a different support would be

needed for any change in pouch dimensions.


Modeling the Retort Pouch


Often it is desirable to model a process to predict process time

or retention of some quality factor during processing. An appropriate

model can also be used to aid in computer control of a thermal process

(Teixeira and Manson, 1981). Thermal processes for canned or retort

pouched food can be modeled by two methods. The heat conduction

equation for the appropriate geometry can be solved for various

conditions (Carslaw and Jaeger, 1959) and used for the prediction of










time temperature relationships. The alternate method is to use

numerical techniques. The use of numerical techniques results in more

flexibility than the use of the exact solution of the heat conduction

equation, since changes in the boundary conditions during processing,

such as those found during a process deviation, are easily handled.


A number of numerical techniques are available which can be used

to model a thermal process. Only three of these techniques will be

discussed here. The most popular numerical method used for modeling

thermal processes for canned or pouched food products is the explicit

forward direction finite difference method. This method consists of

using a representation of the partial derivative heat conduction

equation in a Taylor's series form, which results in the commonly used

explicit forward direction finite difference mathematical model. The

development of the finite difference equation is straightforward and

can be found in a number of heat transfer or diffusion textbooks

(Chapman, 1974; Crank, 1975; Holman, 1976). This explicit method

calculates a temperature at a future time by using the present

temperature of the point of interest and its surrounding points.

Thus, a future temperature is calculated directly from temperatures

that have already been calculated. One disadvantage of this explicit

method is that it is not inherently stable. The choice of a distance

increment limits the choice of time increments. As the distance

increment is decreased, the time increment must also be decreased.

This can cause a large increase in computation time as one decreases

the distance increment. However, this disadvantage has been reduced

with the availability of relatively inexpensive computer time.










Backward direction finite difference methods or implicit methods

do not have the stability problems associated with the explicit

method. The backward direction method is developed in a manner

similar to that of the forward direction method. The backward

difference method uses temperatures from a future time increment to

predict the present time temperature. Therefore, each future time

increment must be solved as a series of simultaneous equations. The

advantage of the backward direction method is that it is

unconditionally stable; thus, large time increments can be chosen for

a small distance increment. This can result in a reduction of

computation time when compared to the forward direction method.


Another numerical technique used for analysis of heat transfer

problems is the finite element method. The major advantages of the

finite element method are the relative ease with which it handles

irregular shapes and complex boundary conditions (Burden et al.,

1981). The finite element method has not been used widely for

modeling of thermal processing of cans or retort pouches, probably

because the shapes being dealt with are simple and the finite

difference methods have no problem dealing with the boundary

conditions normally found during thermal processing of food products.


Models for canned food employ either the exact solution of a

finite cylinder or numerical techniques for the prediction of the

temperature history of the food during the process. Hayakawa and Ball

(1971) developed a theoretical method based on the exact mathematical

solution of the conduction heating equation for predicting










temperatures of food in a can for various retort temperature profiles.

Teixeira et al. (1969a, 1969b) used a forward direction finite

difference technique to predict spore survival and nutrient retention

in canned conduction-heated food. Manson et al. (1974) developed a

forward direction finite difference numerical model applicable to

pear-shaped containers.


The retort pouch has been modeled as a rectangular parallelepiped

or infinite slab. Manson et al. (1970) used a forward direction

finite difference technique to model lethality and nutrient retention

in rectangular containers. Ohlsson (1980) used an infinite slab model

and a forward direction finite difference technique to optimize

quality factor retention during thermal processing. Castillo et al.

(1980) developed a model using the exact mathematical solution to the

heat conduction equation for a parallelepiped geometry. This model

was used to predict nutrient retention during processing.


The geometry of the retort pouch, especially the larger

institutional size retort pouch, can usually be modeled as an infinite

slab with only minor errors (Adams et al., 1983). However, as the

thickness of the pouch increases in relation to the length and width,

the rectangular parallelepiped geometry may become the correct model.


Thermal Processing Determinations


As defined by Title 21 of the Code of Federal Regulations, Part

113.3 (1983), a minimum thermal process is the application of heat to

a food which will ensure the destruction of all microorganisms of









public health significance. Commercial sterility accomplishes the

minimal thermal process plus the destruction of microorganisms capable

of growth at ambient temperatures encountered during storage and

distribution of the food product. Commercial sterility does not imply

that all microorganisms and spores have been destroyed. There are

usually spores present in a commercially sterile food product, but

these are spores of thermophilic microorganisms. These spores are very

heat resistant and to destroy all of them would result in severe

overprocessing of the food product. The spores remaining in a

commercially sterile food product will not reproduce under normal

storage conditions. If by temperature abuse the spores do grow, they

will spoil the product but their growth will not result in a public

health hazard.


The organism of greatest concern in thermal processing of low

acid food is Clostridium botulinum. Since microorganisms are

destroyed following first order kinetics, it is not possible to reduce

the population of a microoganism to an absolute value of zero.

Instead, the population of the microorganism is reduced to a level

where the probability of any survivors is very small. The 12D concept

establishes a process that will reduce the number of C. botulinum

spores present to a probability of less than one surviving spore in

1012 containers (Stumbo, 1973). Thermal processes usually exceed

this Botulinum cook for a safety factor and to reduce the numbers of

other more heat resistant microorganisms which are capable of spoiling

the food product.









There are a number of mathematical methods for determining a

thermal process for a food product, which have been reviewed by

Hayakawa (1978). Of the available methods, two have been used

extensively. These are the improved general method and Ball's formula

method.


The general method of process evaluation was first presented by

Bigelow et al. (1920) and improved by Ball (1928) and Schultz and

Olson (1940). The improved general method is also discussed in detail

by Ball and Olson (1957), Stumbo (1973), and Hayakawa (1978). The

improved general method is based on the fact that any temperature has

a certain amount of sterilizing value and that sterilizing value can

be converted into minutes at a reference temperature. The reference

temperature commonly used in the food industry for low acid thermally

processed foods is 250 F. The mathematical relationship used for

calculating an F-value of a process is


F = (T-Tref)/Zdt
0


where F is the equivalent number of minutes of processing at Tref t

is the total time of the thermal process, T is the temperature at any

given time, Tref is the reference temperature (usually 250 F for low

acid food products), and Z is the number of degrees Farhenheit

required to change the D-value of a microorganism by a factor of 10.

The D-value is the number of minutes at a specific temperature

required to decrease the population of a given microorganism or spore









by 90%. The F-value of a process is commonly termed F when the

reference temperature is 250 F and the Z-value is 18 F.


The Ball formula method has been used extensively in the

determination of thermal processes for food products. The method was

introduced by Ball (1923) and subsequently examined and modified

(Alstrand and Benjamin, 1949; Alstrand and Ecklund, 1952; Ball and

Olson, 1957; Stumbo, 1973; Flambert and Deltour, 1977). The

following discussion of the Ball formula method includes the

modifications of Stumbo (1973).


The Ball formula method for process time calculations requires

the measurement of a number of heat penetration parameters of the

food. These include

fh Time for the straight line portion of the heating

curve to transverse one log cycle.

fc Time for the straight line portion of the cooling

curve to transverse one log cycle.

Jh A heating lag factor. Calculated as the ratio of the
temperature differential of retort temperature and

the temperature obtained by extension of the straight

line portion of the heating curve to zero process

time to the temperature differential of retort

temperature and actual initial product temperature.

Jh=(Trh-Tpih)/(Trh-Tih)

jc A cooling lag factor. Calculated in a manner similar to jh.

Jc=(Tr -Tpic)/(Trc-Tic)









The basic formula used for process time calculations is

B = fhlog(jh(rh-Tih)/)


where g is the difference between retort temperature and temperature

of the food at the cold spot of the container at the start of cooling,

and B is the heating time required for the process starting at the

corrected zero time.


There exists a relationship between g, fh' and the

sterilization value received by the food product. Through this

relationship it was possible to develop tables to predict g knowing

fh and the desired Fo value of the process. Jen et al. (1971)

also included j in this relationship and developed tables including

3c. A listing of these tables can be found in Stumbo (1973).


A number of assumptions were made in the development of Ball's

formula method. The original method assumed a constant j of 1.41.

Stumbo (1973) included the effect of jc in his modification of the

Ball formula method. A second assumption is that fh is equal to

fc.

The calculated process time or B-value as presented earlier

represents a process time in which the retort temperature is achieved

instantaneously. In reality there is often a finite come up time for

the retort to reach processing temperature. In practice the come up

time may be disregarded and zero process time is taken as the time

when the retort reaches processing temperature (Tung and Garland,

1978). The come up time can contribute significantly to the lethality









achieved during a process, especially if come up time is long. Ball

(1923) found that 42% of the come up time could be included as time at

the retort temperature. Townsend et al. (1949) found that this

correction factor applied to processing jars and cans. Alstrand and

Benjamin (1949) determined the correction factor to be close to 42%

with a linear temperature rise; however, they found a correction

factor of approximately 70% when retort come up followed a

semi-logarithmic curve. Uno and Hayakawa (1980) used a dimensional

analysis approach to predict come up time correction factors for

conduction-heated foods. They found the shape of the retort come up

temperature profile had a significant effect on the come up correction

factor. Berry (1983) developed a method for determining the come up

time correction factor. He also found that the 42% correction factor

was adequate for linear come up time retort temperature profile but

was too small for an assymptotic retort temperature profile.


Ball's formula method was developed for use on canned food.

Spinak and Wiley (1982) evaluated the use of the Ball formula and the

improved general methods for process evaluation of retort pouches.

These authors found the Ball formula method to be a reliable method

for process evaluation of retort pouches.


Critical Processing Factors for Retort Pouches


A critical processing factor is defined as "any property,

characteristic, condition, aspect, or other parameter, variation of

which may affect the scheduled process and the attainment of

commercial sterility" (21 CFR 113, 1983). Examples of critical


1









processing factors include headspace in cans, maximum fill weight,

product consistency, etc. The critical processing factors for cans

have been established through years of practical experience and

research. Many critical processing factors for cans are directly

applicable to retort pouches; however, due to the flexible nature of

the retort pouch, there are a number of factors that are different

from or additional to those established for cans. The following

discussion will deal with some of the critical processing factors

which are unique to retort pouches.


Food Product Type


During thermal processing it is advantageous to capitalize on

convective heating in canned foods, whether natural or forced.

Compared to conduction, convection increases heating rate and

decreases process times. It would therefore be desirable to have

convective heating in retort pouches. Terajima (1975) indicated that

in retort pouches, the type of product, whether liquid or solid, may

be more important in determination of total process time than the type

of heating medium or its circulation rate. However, retort pouches do

present some interesting problems. First, the thin profile of

horizontally positioned pouches does not allow natural convection to

occur as readily as in cans. Second, retort pouches typically do not

contain an appreciable headspace which is necessary for mechanically

aided convection. There is some question as to whether retort pouches

should be rotated to enhance heat transfer rate. Some manufacturers

feel that retort pouches should not be rotated under any conditions









(Beverly, 1980: Milleville and Badenhop, 1980) due to possible damage

to the integrity of the pouch seals. However, others have found no

problems with rotation and have found a reduction in process time

required when pouches were rotated (Toska, 1982). It was not

determined whether this decrease in process time was the result of an

increase in the surface heat transfer coefficient or the result of

forced convection. Roop et al. (1983) determined the effect of

rotation on seal strength of retort pouches. They found that while

rotation during processing did not adversely affect the seal strength

after processing, the seal strength was reduced significantly during

processing. This reduced seal strength during processing may be more

significant when dealing with institutional size retort pouches due to

the large quantity of product present (Abbott, 1982). Also, a liquid

product packed in an institutional size retort pouch could experience

excessive hydraulic forces during the physical abuse of distribution

and thus not survive intact without additional case cushioning.

Entrapped Gas


It is generally recognized that it is desirable to remove excess

air from retort pouches before the final seal is made. There are a

number of methods available to accomplish this (Lopez, 1981). Retort

pouches are usually vacuum sealed either by being placed into a vacuum

chamber or by use of a snorkel type apparatus placed inside the pouch.

Pouches can also be steam flushed with either saturated or superheated

steam just prior to sealing. The steam takes the place of the

headspace gas and condenses when cooled. It has been suggested that a








maximum of 2% air by volume be allowed in a sealed retort pouch;

however, some producers feel that this is an unreasonable figure and

that more should be allowed (Milleville and Badenhop, 1980).


Entrapped air, or any noncondensible gas in a retort pouch can

come from a number of sources. Pouches that are filled and sealed

cold can contain a large amount of air dissolved in the product

itself. The order of filling a pouch may allow air to be entrapped

and not easily removed (Milleville and Badenhop, 1980). If vacuum

sealing is not adequate, air can remain in the pouch after sealing.

Air can also be entrapped in the tissue of the food being pouched.

Unblanched meats were found to have more than 20 ml of air per 100

grams, and raw vegetables were found to have more than 10 ml of air

per 100 grams at room temperature and pressure (Milleville and

Badenhop, 1980). Beverly et al. (1980) reported up to 200 ml of

occluded air in the amount of product to be put into an institutional

size retort pouch. They also reported finding 60 to 80 ml of air in

vacuum sealed institutional size retort pouches.


A number of methods are available for the measurement of

entrapped gas in a retort pouch. The simplest of these is the

destructive method in which the pouch is opened under water, the gas

escaping from the pouch is collected in a inverted graduated cylinder

and the volume is measured (Shappee and Werkowski, 1972). There are

two nondestructive methods for the measurement of entrapped gas in a

retort pouch. Shappee and Werkowski (1972) and Ghosh and Rizvi (1982)

first weighed the pouch in water, then placed it in a chamber filled








with water. Air was removed from the chamber until the pouch floated.

The amount of entrapped air in the pouch was then calculated with an

equation derived from Archimedes Principle and Boyle's Law. Gylys and

Rizvi (1983) presented a nondestructive method for entrapped air

measurement for polymeric packages that floated or did not float at

atmospheric pressure. The method was based on the measurement of

volume change of the package at two pressures. The amount of

entrapped gas could then be calculated from an equation derived from

Boyle's law.


The presence of entrapped gas in a retort pouch during processing

has two major effects. First, as the temperature increases during

processing the gas will expand, possibly to the point where the retort

pouch seals are threatened due to excessive internal pressure.

Second, the entrapped gas can significantly reduce the rate of heat

transfer during processing.


A number of researchers have expressed concern over the expansion

of entrapped gas during processing (Nelson and Steinberg, 19567 Davis

et al., 1960; Gould et al., 1962; Whitaker, 1971, Davis et al.,

1972; Milleville and Badenhop, 1980). The problem of gas expansion

during processing can be minimized by the use of overriding air

pressure in the retort vessel during processing. Whitaker (1971)

derived an equation which would calculate the amount of overriding air

pressure necessary to prevent excessive expansion and pressure

increase inside a retort pouch. The use of overriding air pressure










during processing of retort pouches is generally accepted and is

considered a critical factor (Milleville and Badenhop, 1980).


The other problem presented by the presence of entrapped air in

retort pouch processing is the possible reduction of heat transfer

rate (Nelson and Steinberg, 1956; Milleville and Badenhop, 1980).

Entrapped gas has the greatest effect on heat penetration when pouches

are processed in the horizontal position. Retort pouches are

generally processed horizontally (Milleville and Badenhop, 1980),

while institutional size retort pouches are almost exclusively

processed in the horizontal position because of the large amount of

product present. Processing vertically could put undue stress on the

pouch seals. Beverly et al. (1980) found that as the amount of

entrapped gas increased, the required process time for an equivalent

sterilization value also increased. The increase in process time was

found to be greater with pouches that were allowed to expand freely.

Huerta-Espinosa (1981) examined the effect of entrapped air on

processing requirements of pears and green beans. Process times for

both products were found to increase approximately 15% for each 100 ml

of air in 12 by 15 by 1.5 inch retort pouches. Huerta-Espinosa (1981)

also showed a movement of the cold spot upward in the presence of

entrapped air: however, this effect was not quantified. Berry and

Kohnhorst (1983) examined the effect of entrapped gas on processing of

institutional size retort pouches. They also found an upward movement

of the cold spot with increased air content. The sterilization value

of a given process was found to decrease as the amount of entrapped











gas increased; however no replicates of the tests performed on the

two products were done.


Heating Media Used for Retort Pouch Processing


The ideal heating medium for processing retort pouches would be

100% steam, as is used for cans, due to its large surface heat

transfer coefficient. Retort pouches cannot safely be processed in

100% steam for two reasons. First, the expansion of entrapped gas

during processing could put excessive stress on the pouch seals as

discussed earlier. Second, as the pouch contents approach retort

temperature, a decrease in retort pressure below the operating

pressure could cause a large internal pressure increase in the pouch

and thus threaten the integrity of the pouch (Goldfarb, 1970). For

these reasons retort pouches are processed in steam-air mixtures or in

water with overriding air pressure.


The control of a water process is generally recognized as being

less complicated than control of a steam-air process (Parcell, 1930a;

Parcell, 1930b; Pflug and Borrero, 1967; Milleville, 1980c);

however, there are a number of commercially available retorts

successfully utilizing steam-air as the processing medium (Milleville,

1981).


During steam-air or water processing the main concerns are to

maintain all parts of the retort at the processing temperature and to

maintain a constant rate of heat transfer during processing

(Milleville and Badenhop, 1980).











The maintenance of retort temperature during water processing is

usually accomplished by recirculating the heating medium (Peterson and

Adams, 1983; Davis et al., 1972; Pflug and Borrero, 1967) and/or by

bubbling air through the retort vessel to increase mixing (Berry and

Kohnhorst, 1983; Berry, 1979; Davis et al., 1972: Pflug and

Borrero, 1967). The amount of agitation or the rate of flow of the

processing water has a significant effect on the rate of heat transfer

into the retort pouch. Berry and Kohnhorst (1983) demonstrated a

large decrease in the sterilization value achieved when only natural

convection was used instead of bubbling air through the retort vessel.

This large decrease was no doubt the result of a reduced surface heat

transfer coefficient. Pflug and Borrero (1967) observed a decrease in

rate of heat transfer as the rate of processing water circulation

decreased. They attributed this decrease to a decrease in the surface

heat transfer coefficient. Peterson and Adams (1983) examined the

effect of water velocity on heat transfer rates. Apparent surface

heat transfer coefficients were calculated and ranged from 33 to 48

BTU/hr ft2 F for Reynolds numbers of 3,000 to 33,000 respectively.

Process times as calculated by the formula method (Stumbo, 1973)

increased by 9.6% from the low to high flow rate.


When retort pouches are processed in steam-air, a positive flow

of the heating media is required as it is in water processing

(Milleville, 1980b: Pflug and Borrero, 1967). This requirement is

more important than it is when processing in 100% steam due to the

possibility of air pockets forming after condensation of the steam on

the surface of the pouch.











Temperature control of a steam-air process is more difficult than

temperature control of a water process, especially during the come up

period when there is a large initial heat demand (Helmer et al., 1952;

Parcell, 1930a). Pflug and Borrero (1967) suggested using 100% steam

for the come up period, and introducing air after the processing

temperature had been reached. This method gave improved control of

temperature during the initial stages of the process.


Surface heat transfer coefficients of steam-air mixtures are

known to be smaller than those of 100% steam (Sparrow and Lin, 1964).

Pflug and Blaisdell (1961) found an increase in heat transfer rate

with an increase in the velocity of the steam-air mixture. Adams et

al. (1983) found apparent surface heat transfer coefficients near 50

BTU/hr ft2 F for steam-air mixtures. The coefficients showed a

trend to increase with an increase of flow rate of the steam-air

mixture. Ramaswamy et al. (1983) presented a method for the

determination of surface heat transfer coefficients of steam-air

mixtures based on a lumped capacity method (Kreith, 1973). They

reported coefficients much higher than those reported by Adams et al.

(1983).


Another consideration in processing with steam-air mixtures is

the percentage of steam to use during processing. Generally, little

difference has been found in observed heat transfer rates of foods or

food simulating materials with steam percentages from 75 to 100%

(Adams et al., 1983: Yamano, 1976; Terajima, 1975; Pflug et al.,

1963). Ramaswamy et al. (1983) used a more sensitive method and did











find an increase in heat transfer rate and surface heat transfer

coefficient with increasing steam percentage.


Heat Penetration Through Retort Pouch Material


The resistance to heat transfer of plastics compared to metal has

been a concern in processing retort pouches. Nelson and Steinberg

(1956) felt that the heat transfer resistance due to the plastic would

be negligible; however, no evidence was presented in support of this

statement. Wornick et al. (1960) tested the resistances of films up

to 0.003 inch thick by evaluating the time for pouches filled with

water to reach the processing temperature of 212 F. They concluded

that there was no significant effect on heat transfer due to changes

in pouch thickness; however, they experienced a large amount of

variation between replicate pouches which may have hidden any effect

of the pouch material. Chapman and McKernan (1963) calculated the

effect of high density polyethlyene of various thicknesses on thermal

process times for dog food in a 4.75 by 3.75 by 1.5 inch container.

When tinplate was used, the container required a process time of 49

minutes. When 0.008 inch polyethylene was used, 3 additional minutes

were required for an equivalent process. Eight additional minutes

were required for a container constructed of 0.020 inch polyethylene.


Racking Systems Used for Retort Pouches


The flexible nature of the retort pouch requires that it be

supported in some manner during processing. Two types of support have

been used, confined and unconfined. It is generally recommended that











the confined method be used (Milleville and Badenhop, 1980).

Confinement of a retort pouch maintains the pouch in a known geometry

during processing, thus reducing the variation in geometry that would

be experienced if the pouch was allowed to expand freely (Beverly et

al., 1980). With a confined racking system, a process can be designed

in the maximum thickness the pouch can obtain, thus insuring an

adequate process (Beverly et al., 1980; Milleville and Badenhop,

1980; Berry, 1979; Davis et al., 1972; Pflug et al., 1963).

Milleville and Badenhop (1980) also suggest that minimum pouch

thickness may also be important in racking systems designed with solid

plates. Pflug et al. (1963) calculated the resistance to heat

transfer from 0.125 inch aluminum and 0.0625 inch steel to be

negligible; however, if the top and bottom of the pouch are not in

contact with the solid plates, reduced heat transfer may result.


The other critical factor in the design of the racking system is

that there must be a channel for the heating medium on both sides of

the rack (Beverly et al., 1980; Milleville and Badenhop, 1980;

Berry, 1979: Davis et al., 1972; Pflug et al., 1963) and that the

channels must allow for a uniform flow of the heating medium between

all channels (Beverly, 1980: Milleville and Badenhop, 1980).

















CHAPTER III
HEAT TRANSFER RESISTANCES OF RETORT POUCHES

Introduction


In designing a safe thermal process for low acid food packaged in

a retort pouch, many factors, in addition to those normally considered

for processing foods in cans must be considered. This is because of

the construction of the retort pouch material and its flexible nature.


The pouch material itself, although very thin, is a poorer

conductor of heat than the metal of a can. Nelson and Steinberg

(1956) state that the pouch material should not affect heat

penetration rate, but no data were presented to support this

statement. Wornick et al. (1960) tested films up to 0.003 inch thick

and found that film thickness did not change the heating time

appreciably. However, variation in data from different pouches was

great and could have hidden any subtle differences which may have

existed. Chapman and McKernan (1963) conducted a study of the effect

of thickness of plastic material on heat penetration. Their

calculations showed that dog food processed in a 4.75 X 3.75 X 1.5

inch container constructed of 0.008 inch high density polyethylene

required a process time of 52 minutes compared to 49 minutes required

by a similar container constructed of tinplate.










The flexible nature of the retort pouch requires that it be

supported during processing in order to maintain a known geometry. It

has been established that channels for the heating medium between

pouches are needed for proper heat transfer (Beverly et al., 1980).

However, the material used to construct the racking system has not

been examined critically. Pflug et al. (1963) used a racking system

of solid 1/8 inch aluminum. They calculated the effect of the metal

rack on heating rate to be negligible. However, it has been suggested

that an underfilled retort pouch not in intimate contact with both

plates of the racking system may heat more slowly than one that is in

intimate contact with both plates (Milleville and Badenhop, 1980).


Entrapped air may present difficulties in assuring an adequate

thermal process with retort pouches. Entrapped air can cause problems

in two ways. First, it can expand during processing putting excessive

pressure on the pouch seals, possibly to the extent that the pouch

integrity is threatened. Second, it can cause a reduction in the rate

of heat penetration. The problem of expansion is controlled to a

certain extent by the presence of overriding air pressure in the

retort vessel for both water and steam-air processing (Whitaker,

1971). The other obvious precaution would be to maintain the

entrapped air at a low level (Lampi, 1977). The major concern with

entrapped air is its effect on the rate of heat penetration (Nelson

and Steinberg, 1956: Beverly et al., 1980), especially with

horizontally positioned retort pouches. Berry and Kohnhorst (1983)

and Beverly et al. (1980) found an increase in process time needed for

equivalent Fo-values as the amount of entrapped air increased.










Huerta-Espinosa (1981) also showed an increase in process time with

entrapped air along with movement of the cold spot upward from the

geometric center of the pouch.


The purpose of the present study was to examine the magnitude of

heat transfer resistances existing between the flowing heating medium

and the surface of a food product in institutional size retort

pouches. These resistances included the rack design, flow rate of the

heating medium, pouch material and entrapped air.


Materials and Methods


Polycarbonate Slab Preparation


Four polycarbonate slabs (two 10 X 9 X 1.0 inch slabs and two 10

X 9 X 1.5 inch slabs; Tuffac, Rohm and Haas, Philadelphia, PA) were

used in this investigation. Each slab was equipped with 4 holes (#50

drill, 0.07 inch diameter) to accommodate thermocouples. All

thermocouple holes were located in the infinite slab region of the

specific polycarbonate slab as determined by comparison of the exact

mathematical solutions of the heat penetration equation for the

infinite slab and parallelepiped geometries. The computer programs

used for this comparison are listed in Appendices A and B for the

infinite slab and parallelepiped geometries, respectively. The

measured thermocouple positions for the four polycarbonate slabs are

listed in Table 3-1. Fig. 3-1 illustrates the thermocouple

positioning in a 1.0 inch slab.












Table 3-1.


Vertical position of thermocouple holes in polycarbonate
slabs.


Vertical distance from midplane
(inches)


Slab Thickness

Slab number


1.0 inch


1

+0.23
+0.07
-0.01
-0.26


2

+0.29
+0.14
-0.04
-0.26


1.5 inch

3 4

+0.38 +0.28
+0.19 +0.13
0.00 -0.03
-0.38 -0.41




















































0 0 0 0
i0



10 in.


Figure 3-1.


Thermocouple placement in a 1.0 inch polycarbonate slab.
----- indicates infinite slab region of the slab
midplane.


,,,


*~










The polycarbonate slabs were processed with and without the

presence of pouch material. For slabs processed without pouch

material, 6 inch CNL needle type thermocouples (O.F. Ecklund, Cape

Coral, FL) were sealed in the thermocouple holes of the slab with a

general purpose silicon caulk (Dow Corning, Midland, MI). For slabs

processed in retort pouch material, a commercially available pouch

material (0.5 mil PET/ADH/0.5 mil Al foil/4 mil PPT American Can

Company, Greenwich, CT) was used. Six inch CNL type thermocouples

were introduced into the pouch by means of stuffing boxes (C5.2, O.F.

Ecklund, Cape Coral, FL). The pouch was formed with 1/4 inch wide

impulse seals (14H/HTV, Vertrod Corp., Brooklyn, NY). The final seal

of each pouch was accomplished with a vacuum impulse sealer (Multivac

M-3-II, Koch, Kansas City, MO). Each pouch was exposed to 29 inches

of vacuum (inches of mercury) for at least 2 minutes prior to sealing.

Following the vacuum seal, each pouch was given an additional 1/4 inch

cosmetic seal.


Slab Processing


Processing was accomplished in a retort designed and built at the

University of Florida specifically for processing institutional size

retort pouches. A detailed description of this retort is given by

Adams et al. (1983). The heating medium used was flowing pressurized

water. The retort was loaded at ambient temperature and was flooded

with 270 F water. Two to 3 minutes were required to bring the water

up to the processing temperature of 250 F after flooding. Processing

continued at 250 F with 10 psi overriding air pressure (25 psig total










system pressure) until the slowest heating thermocouple was within 5 F

of retort temperature (approximately 245 F), at which time the retort

was drained and flooded with cooling water. The cooling water was

circulated while being continuously vented and replenished. The

appropriate flow rates of the heating medium were maintained during

processing for both heating and cooling. All thermocouples were

grounded by attaching a ground wire from the stainless steel sheath of

the thermocouple to the frame of the retort inside the retort vessel.

Temperature data were collected on a recording potentiometer

(Digistrip II, Kaye Instruments, Bedford, MA).

The horizontal racking system used for the 1.0 inch polycarbonate

slab was identical to that described by Peterson and Adams (1983).

The 1.0 inch slabs were positioned in the two rear center positions of

the racking system. All positions not containing polycarbonate slabs

were filled with institutional size retort pouches containing a 10%

(w/w) bentonite suspension in water. The horizontal racking system

for the 1.5 inch polycarbonate slabs was similar to the 1.0 inch

racking system. A total of six 1.5 inch pouches could be processed at

once. The racking system contained 3 racks each able to hold two 1.5

inch pouches. A 3/4 inch channel for the heating medium surrounded

each of the racks. The 1.5 inch polycarbonate slabs were processed in

the upper and center rear positions in the racking system. All other

positions were filled with retort pouches containing a 10% bentonite

suspension.










Thermal Diffusivity of Polycarbonate Slabs


Thermal diffusivity of the 1.0 and 1.5 inch polycarbonate slabs

was determined by the transient method utilizing the fh-value of the

heating curve (Olson and Jackson, 1942). The polycarbonate slabs were

processed in flowing 100% steam (460 lbs/hr), with no pouch material,

using the large perforation rack design with no top on the rack. The

resulting fh-value was determined with a computer program using

linear regression. This program is listed in Appendix C.

Heat Transfer Resistances Due to Pouch Material, Rack Type,
and Flow Rate


A preliminary experiment was performed to determine if the

replicate slabs demonstrated significantly different heat penetration

behavior. The experimental design also allowed testing of the

significance of day to day variation in the operation of the retort.

The experiment was performed at Reynolds numbers of 6,000 and 32,000.

Reynolds numbers were calculated using the hydraulic diameter of the

channels between the racks. Slabs were processed in pouches using

confining racks constructed of 16 gauge stainless steel with 3/8 inch

holes on 9/16 inch staggered centers.


The statistical design used for the preliminary experiment was a

randomized block design with pouch thickness as main blocks and days

as sub-blocks. A total of 8 retort runs were completed; each

contained two polycarbonate slabs of the same thickness. The 8 runs

were completed over a period of 4 days.










The second experiment in this portion of the research utilized

the two 1.0 inch polycarbonate slabs. The slabs were processed with

and without pouch material for this experiment. Three different rack

designs made from 16 gauge stainless steel were used. The first rack

type was constructed of perforated stainless steel with 40% open area
and large perforations (3/8 inch holes on 9/16 inch staggered

centers). The second type had 40% open area and small perforations

(1/8 inch holes on 3/16 inch staggered centers). The third type was

constructed from a solid stainless steel sheet. The two racks in the

one inch racking system that were not being tested (the top and bottom

positions) contained the large perforation rack design for all

processing trials. Four flow rates were used; the average volumetric

flow rates along with the calculated Reynolds numbers are listed in

Table 3-2.


A completely randomized split plot design was used for this

investigation. Each processing run contained one polycarbonate slab

with pouch material and one without pouch material; therefore, pouch

material was treated as the sub-plot. Each processing run contained

one rack type and was completed at one flow rate; therefore, rack

type and flow rate were analyzed as main plots. A total of 24

processing runs were performed, each containing 2 polycarbonate slabs

as experimental units. The appropriate division of the degrees of

freedom for this experimental design are listed in Table 3-3. The

main plots, flow rate and rack design, and the interaction of the two

were tested using error A. The sub-plot, pouch material, and

interactions of the sub-plot were tested using error B. The response













Table 3-2.


Average flow rates used to test effect of pouch material
and rack type on heat transfer.


Flow rate
(gal/min)

9.5
25.1
60.1
108.6


Reynolds
number

2,800
7,500
18,000
32,000






41


Table 3-3.


Division of degrees of freedom for experimental design
used to examine effects of the presence of pouch
material, confining rack design, and flow rate on heat
transfer.


Source


Flow
Rack
Flow*Rack
Error A [Run(Flow*Rack)]
Pouch
Pouch*Flow
Pouch*Rack
Pouch*Flow*Rack
Error B [Pouch*Run(Flow*Rack)]
Thermocouple(Pouch*Run(Flow*Rack)
TOTAL


2
6
12
144
191









variables for the experimental design were fh and f which were

calculated by means of a computer program using linear regression

(Appendix C).


Entrapped Air


For this portion of the research, the sides of the 1.0 and 1.5
inch polycarbonate slabs were tapered with a general purpose silicon
caulk (Dow Corning, Midland, MI). The purpose of this was to give the

pouched polycarbonate slabs a shape more closely resembling that of an
actual food product. The appropriate amount of air was injected into

the corner of a vacuum sealed pouched slab by means of a syringe with

a 20 gauge hypodermic needle. The pouch was then resealed. The seals

of the pouch were as close to the polycarbonate slab as possible.


Pouched slabs with air were processed in high flow rate

pressurized water at 250 F with 10 psi overriding air pressure (25

psig total system pressure). The 1.0 inch slabs with air present were

processed in the small perforation confining racks, and the 1.5 inch
slabs were processed in the large perforation confining racks.


The actual amount of air present in the pouched slabs was

measured after processing by the destructive method (Shappee and

Werkowski, 1972). The pouch was opened under water, and the air

present was collected in an inverted graduated cylinder. The volume

of air was measured with the air in the inverted graduated cylinder at

the same level as the surface of the water in the reservoir.










Results and Discussion


Thermal Diffusivity of Polycarbonate Slabs


The thermal diffusivity of the 1.0 and 1.5 inch polycarbonate

slabs was determined by processing in flowing 100% steam supported on

a sheet of 16 gauge stainless steel with 3/8 inch holes on 9/16 inch

staggered centers. Thermal diffusivity was calculated from the

fh-value according to the formula given by Olson and Jackson (1942)

for an infinite slab:

a = 0.933(A2)/fh

Where a is the thermal diffusivity, A is the half thickness of the

slab, and fh is the time required for the straight line portion of

the heating curve to transverse one log cycle when plotted on semi-log

graph paper. The error introduced into the calculated thermal

diffusivity values by assuming an infinite slab geometry was 2% for

the 1.0 inch slabs and 5% for the 1.5 inch polycarbonate slabs. Table

3-4 contains the calculated thermal diffusivity values for this

experiment, the manufacturers specification, and the results of the

1.0 inch slabs processed in high flow rate water (Re=32,000, no pouch

material, large perforated racks) from the experiment examining flow

rate, pouch material, and rack design. There is a variation of less

than 4% between all of these values. Also, the thermal diffusivity

value for the high flow rate of water resulted in a value very close

to that obtained with flowing 100% steam. This indicated that high

velocity water resulted in a heat transfer rate similar to 100% steam














Table 3-4. Thermal diffusivity values obtained for the polycarbonate
slabs.


Manufacturer Specification

100% steam: 1.0 in. slabs

100% steam; 1.5 in. slabs

Flowing water (Re=32,000)

a


a
Thermal Diffusivity

2
0.012 in /min
2
0.0117 + 0.0002 in /min
2
0.0121 + 0.0001 in /min
2
0.0119 + 0.0018 in /min


Thermal diffusivity + 95% confidence interval











for materials with a thermal diffusivity equal to or greater than

polycarbonate.

Heat Transfer Resistances Due to Pouch Material, Rack Type,
and Flow Rate


This experiment was designed to determine if the presence of

pouch material resulted in a significant reduction in heat transfer

rate, and to determine if the size or presence of perforations in the

confining rack design affected heat transfer. Flow rate was known to

affect heat transfer rate (Peterson and Adams, 1983), but was included

in this experiment to identify any interactions which may have existed

between it and the other two variables. Both heating and cooling data

were analyzed using fh and fc values as response variables.


A preliminary experiment was conducted to determine if a

difference existed between the two 1.0 inch polycarbonate slabs and

between the two 1.5 inch polycarbonate slabs, and to determine if

there was a significant variation in measured heat transfer rates with

time. No significant difference was found between the two slabs of

each profile. The p-values found were 0.25 and 0.79 for the 1.0 and

1.5 inch slabs, respectively. There was also no significant

day-to-day variation in the measured heat transfer rates. These two

findings allowed the experiment examining the effects of pouch

material, rack type, and flow rate to be analyzed as a completely

randomized split plot design even though the flow rates were

experimentally performed as four randomized blocks, and pouch material

was used on one polycarbonate slab and not on the other throughout the

experiment.











Table 3-2 lists the flow rates used in this experiment along with

the calculated Reynolds numbers. The Reynolds numbers for this

experiment were chosen to concentrate on the low range of the pouch

retort's capabilities, since Peterson and Adams (1983) found the

greatest effect on heat transfer rate at the lower Reynolds numbers.

Perforations such as those in the racks are known to increase

turbulence (Schlichting, 1968), but there is no convenient method to

quantify this effect. Therefore, the turbulence levels normally

associated with calculated Reynolds numbers is an underestimation of

the actual turbulence level for the perforated racks.


Analysis of the data showed that no significant interaction

existed between pouch material and any of the other effects examined;

therefore, it was appropriate to look at the main effect of the

presence of pouch material. The presence of pouch material was not

significant during heating, which agreed with the findings of Wornick

et al. (1960) and Chapman and McKernan (1963). It should be noted

that pouch material may show a measureable effect when pouch contents

demonstrate an apparent thermal diffusivity greater than that of

polycarbonate. During cooling the effect of pouch material was not

significant at the 0.05 level. Because of the large variation of heat

penetration parameters associated with the solid racks, the effect of

pouch material was examined by excluding the solid rack data from the

analysis. Pouch material again demonstrated no significant effect on

fh-values during heating (p=0.15) but was statistically significant

during cooling (p=0.04). The magnitude of this difference was very

small. A mean fc-value of 19.90 minutes with pouch material versus











19.71 minutes with no pouch material was observed. This difference is

not large enough to be of concern when performing process time

calculations.


The effects of flow rate and rack type were highly significant

for both heating and cooling (p=O.O001). There was a significant

interaction between flow rate and rack type during heating but not

during cooling. Fig. 3-2 and 3-3 show fh and f -values,

respectively, versus Reynolds number for the different rack designs.

During heating, the solid rack resulted in fh-values 6 to 8 minutes

higher than the small and large perforated rack designs. There was a

significant difference between the two perforated rack designs but the

magnitude was small. The effect of rack design was not as dramatic on

cooling rate as it was on heating rate. The solid rack design

exhibited fc-values approximately 2 minutes higher than the

fc-values obtained for the large and small perforated rack designs.

The two perforated rack designs did not differ significantly during

cooling.


A few simple calculations can show that 1/16 inch of stainless

steel should not have had as large an effect on the fh of the

polycarbonate slabs as it did. The probable reason for this increase

of fh-values was that the solid racks had a height of approximately

1-1/16 inches and therefore the polycarbonate slab was not in intimate

contact with the stainless steel of the rack. This was coupled with

the fact that the racks were closed on the ends and did not allow the

processing water to freely enter the rack. An attempt was made to





48





30--



28--



- 26- -
0
E
O No Perforations
LL- 24- A Small Perforations
C Large Perforations


22-



20- -


0 10,000 20,000 30,000

Reynolds Number

Figure 3-2. Effect of Reynolds number on fh for each rack type.
(No direct contact of pouched slab to solid rack.)













0 No Perforations
A Small Perforations
O Large Perforations


0 10,000 20,000 30,000


Reynolds


Number


Figure 3-3. Effect of Reynolds number on fc for each rack type.
(No direct contact of pouched slab to solid rack.)


C
- 2I
0











improve the heat transfer rate of the solid racks at the highest water

flow rate (Re=32,000). A 1/2 inch channel was opened through the

center of the solid racks to allow processing water to enter the rack

freely from the ends. When the polycarbonate slabs were in the rear

of the retort, the results obtained were identical to those found

earlier. This indicated that the 1/2 inch channel was not effective

in allowing processing water to flow freely through the rear position.

When the polycarbonate slabs were placed in the forward position,

close to the entering processing water, the heating rate increased

significantly and was less variable. The mean fh-value observed was

21.69 + 0.94 minutes. These fh-values and 95% confidence intervals

were still higher than those found for either the large or small

perforated rack, but they were much less than the original fh found

without a channel and with the slabs in the rear of the racking

system. The open channel increased heat transfer rate but apparently

the water entering the rack was at least partially trapped, thereby

forming an additional barrier to heat transfer.

Entrapped Air


Initial work with included air was performed with the two 1.0

inch polycarbonate slabs. However, variable results were experienced:

entrapped air sometimes caused dramatic changes in the heat

penetration characteristics of the slab, while the same amount of air

in a replicate trial would cause no effect. This inconsistency was

determined to be caused by the air migrating to the sides and corners

of the pouch. Also, the stuffing boxes used to introduce the











thermocouples into the pouch were very close to the top and bottom of

the pouch. By tightening the stuffing boxes to insure an airtight

seal, an uneven pouch surface was occasionally produced when air was

introduced. These limitations were minimized by rounding the sides of

slabs with caulk and producing wells in the caulk to restrict

deformation of the sealed pouch by the stuffing boxes.


The first major effect of entrapped air observed with the one

inch slabs was the consistent appearance of a broken heating curve.

The first portion of the heating curve was identical to that obtained

with no air present. The second portion of the heating curve reached

an fh-value of 33 to 34 minutes. Fig. 3-4 illustrates a typical

curve obtained along with a replicate run with no air present. The

first portion of the broken heating curve indicated that entrapped air

had no effect on heat penetration in the initial phase of heating.

This phenomenon may have been produced by redistribution of air in the

pouch when a certain temperature was reached. This phenomenon may be

unique to polycarbonate slabs and may not be observed with a real food

products. The second portion of the heating curve consistently

reached an fh of 33 to 34 minutes for all air levels tested. It was

felt that this was due to the air expanding to the maximum level

allowed by the confining rack system. The racks used for this

experiment allowed a total pouch thickness of 1-1/16 inches. Thus,

the air on top of the pouch was allowed to expand to a maximum height

of 1/16 inch. To test this theory, a forward direction finite

difference computer program was developed. The program used an

infinite slab model and assumed an infinite Biot number at the surface















/

/




/


/

200--
0
a0

o

A'
P70-
70-'


0 /o
0
o0

o0
0


/


A No Air
O 32ml Air












40


Time (min)


Figure 3-4.


Effect of 32 ml air on heating curve of a 1.0 inch
polycarbonate slab (center thermocouple position).


245+


240+


L-
0




CL
E
0


225+











of the pouch material. Expansion of both air and water vapor were

accounted for through use of the steam table and the ideal gas law.

The water present was assumed to have an activity of 1.0. All air was

assumed to be on the top surface of the pouch and height of the

entrapped air was assumed to be maximal at the center of the

polycarbonate slab and to decrease in a parabolic fashion to zero at

the edges of the slab. The equation which describes this behavior is


z=c+c*x2*y2/(a2*b2)-c*x2/a2_c*y2/b2


This equation can be integrated and rearranged to give


c=9*V/(16*a*b)


Using this equation, the height of the air above the center of a slab

can be calculated from the volume of air present and the length and

width of the slab. Heat transfer through the air layer was assumed to

be conductive. The program used the height of the air at the center

of the slab for heat penetration calculations. A provision was

included in the program for maximum expansion of entrapped air allowed

by the confining rack. This program is listed in Appendix D. Using

the heat penetration characteristics of the polycarbonate slabs, the

program determined a maximum fh-value of 35 minutes when a maximum

expansion of 1/16 inch was used. This agreed very well with the

experimentally observed values of 33 to 34 minutes. However, the

program did not predict the initial portion of the heating curve

observed in Fig. 3-4. A curving line that indicated slower heating

than the observed heating curve was predicted by the computer. This











curving is what one would expect with expansion of air during the

first stages of heating.


The second major effect of entrapped air was a change in the

temperature profile within the slab. Fig. 3-5 illustrates all four

heat penetration curves obtained for a 1.0 inch polycarbonate slab

with 32 ml of air present. The bottom of the slab clearly heated

faster than the top, and the thermocouple placed directly above the

center (+0.07 inch) had essentially the same heating curve as the one

placed directly below the center (-0.01 inch). With no air present,

the top and bottom thermocouples in this slab gave nearly identical

heating curves and the thermocouple placed +0.07 inch up from the

center heated faster than the thermocouple placed -0.01 inch from the

center. The air in this pouched slab clearly resulted in a movement

of the cold spot upward.


The 1.5 inch polycarbonate slabs were used for further

investigation of this cold spot migration. The stuffing boxes used to

introduce the thermocouples into the pouched slab were not as close to

the top or bottom of the slab as they were in the case of the 1.0 inch

slabs, thus a flat surface on top of the slab was achieved. Also, 200

ml of water was added to the pouched slab to help fill the voids in

the corners and at the edge of the slab. This aided in confining the

entrapped air to the top of the pouched slab. Controls were processed

with no air and with 200 ml of water in the pouch. The added water

did not affect the heat penetration characteristics of the pouched 1.5

inch slab.
















a 09
A 0 9
A 0@


a 0
a o0
A 00@

0@



0+
0+
OA
a -


I I


30


Time (min)


Figure 3-5. Heating curves obtained for all 4 positions in a 1.0 inch
polycarbonate slab (slab #1) with 32 ml air present.


245+-


240+


0


u-
O

0)

0.
E
0)


225-


200+


0.23"
0.07"
0.01 "
.0.26"


from
from
from
from


center
center
center
center


150+


70-

0


4
40


I











A 1.5 inch slab (slab #4) was processed with 98 ml of air

present. The temperatures recorded at each of the four thermocouple

positions in the slab were fit to a cubic function at various process

times. The program used to fit the data to the cubic function

employed least squares regression and assumed the lower surface

temperature of the slab to be equal to retort temperature. This was a

valid assumption since a very large Biot number was found with high

flow rate water earlier. Also, if the calculated top surface

temperature of the slab was predicted to be greater than the retort

temperature, the program repeated the regression using a top surface

temperature equal to the retort temperature. This was necessary only

in a few instances during the first portion of heating when there was

a large temperature gradient in the slab. A listing of this program

can be found in Appendix E. The predicted temperature profiles for

various processing times can be seen in Fig. 3-6. The bottom of the

slab was found to have a higher temperature than the top of the slab

at any given process time. It was also apparent that the cold spot

moved toward the top of the slab as processing progressed.


The position of the cold spot at a given process time was found

by calculating the position of zero slope from the derivative of the

predicted cubic function. Fig. 3-7 illustrates the movement of the

cold spot from the geometric center of the slab during processing.

Initially the cold spot was at the center of the polycarbonate slab.

As processing progressed, the cold spot moved upward from the center

of the slab to approximately 0.3 inch above the center at the end of

processing.
































Figure 3-6.


Temperature profile within a 1.5 inch polycarbonate slab
at various processing times. Curves are from cubic
function predicted by least squares. = Measured
temperatures.









230-





200--




LU-
0 170-





E 140-












Top
Top


7/


I














* i


I I I


I I I
+0.4 Center -0.4

Position (in. from center)


Bottom















I C


V
4-I


r- S-
*3




I S-C
I (a
0
S-



I0
or-
\ .-



*. o



0-


\ > *r-
r -
0) Ln


qi
-m

C1L
6 d 6 o 0-
I .- Q I I I ,-


Jeiueo woj; seqoul
F + + 5 a
000dC0001











A replicate pouch with 99 ml of air present was processed to

determine if the temperature distribution in the slab and cold spot

migration could be repeated and quantified. The replicate resulted in

the same general trends experienced in the previous pouched slab (i.e.

the bottom of the slab heated faster than the top and the cold spot

migrated upward). However the results obtained with the replicate

slab were not identical to those obtained with the first slab. The

cold spot migration of the replicate slab was less than that of the

first slab, migrating a total of 0.16 inch above the center during

processing. The finite difference program listed in Appendix D was

run under matching conditions to predict cold spot migration in

between the two experimentally determined values.


It should be noted that the cold spot did not migrate with the

geometric center of the pouch containing entrapped air in any of the

three cases examined. Therefore, a self-centering thermocouple

positioning method such as that described by Pflug et al. (1963) would

not maintain a thermocouple at the cold spot of a retort pouch

containing a significant amount of air.


To illustrate the effect of entrapped air on thermal processing,

the finite difference program was used to simulate an actual process

with various amounts of air present. The results of this simulation

are presented in Table 3-5. Increasing the amount of entrapped air

from 5 ml to 50 ml with a 10 X 9 X 1.5 inch pouch dropped the center

point Fo from 6.42 to 1.69 with a maximum expansion of 1/16 inch

allowed by the confining rack system. The Fo received by the cold














Table 3-5.


Effect of entrapped air and maximum expansion allowed by
confining rack system on F received by a 1.5 inch
polycarbonate slab with a process time of 70 minutes
(includes 3.0 minute come up time).


Maximum
ml Air Expansion


1/16
1/16
1/16
1/16
1/8
1/8


Time to reach
maximum expansion


38 min.
2 min.

57 min.


Center F
0


6.42
2.97
1.69
1.31
1.56
0.63


Cold spot F


6.41
2.69
1.50
1.19
1.25
0.41


a
Calculated by improved general method.
temperature = 250 F; Z = 18 F.
b


Maximum allowed expansion not reached during processing.


Reference












spot was less than that received by the center in all cases; however,

the effect is not as dramatic as that caused by increasing the amount

of air present. Also illustrated in Table 3-5 is the effect of the

maximum expansion allowed by the confining rack system. In some

cases, an increase in the maximum expansion allowed can have an effect

as dramatic as increasing the amount of entrapped air present.

Clearly the presence of air, even with a confining rack, can

substantially reduce the achieved Fo-value to the point where

underprocessing of potential health significance may occur.


Summary and Conclusions


A number of conclusions can be drawn from the results of this

experiment. First, it was discovered that water at a sufficiently

high flow rate resulted in a surface heat transfer coefficient which

was sufficiently high to give a heat transfer rate equal to that

obtained with 100% steam.


Investigation of the rack design determined that the two

perforation sizes used showed no practical difference in rate of heat

penetration. The solid rack design, however, did show a significant

decrease in the rate of heat transfer when there was a gap between the

product and the rack. The results obtained suggest that it is

important to insure intimate contact between the pouch and the metal

plates of a solid rack as suggested by Milleville and Badenhop (1980).

Flow rate of the processing medium was also shown to have a

significant effect on heat penetration rate.











Entrapped air was found to have a very large effect on heat

penetration rate. The effect of entrapped air was also difficult to

quantify. Entrapped air resulted in two major effects. First, heat

penetration rate into the retort pouch was severely reduced. Second,

the cold spot of the pouch tended to migrate upward during processing.

The reduction in heat transfer rate as the amount of air increased had

a greater effect on the total thermal process received by the pouch

contents than did cold spot migration. When entrapped air was

present, the maximum expansion allowed by the confining rack system

was an important factor. For optimum heat transfer this maximum
expansion should be kept to a minimum. Entrapped air could present a

potential health hazard in retort pouches due to its resistance to

heat transfer and to its transitory nature.

















CHAPTER IV
THERMOCOUPLE GROUNDING IN RETORT POUCHES

Introduction



Accurate measurement of thermal processing parameters is

necessary to obtain a valid calculated process time for the

sterilization of canned and pouched foods. The validity of these

parameters is directly related to the accuracy of the temperature

measurements taken during a heat penetration test. Current

thermocouple techniques used in the food industry have been developed

for cans and glass containers and work well for those containers.

With the advent of retort pouches, some temperature measurement

problems have been observed with these standard techniques. Davis et

al. (1972) experienced temperature measurement errors in retort

pouches as large as 20 F when using copper-constantan wire

thermocouples. They solved this problem by using enamel-coated

thermocouple wire and grounding the measuring junction of the

thermocouple to the ground of the recording potentiometer. Pflug

(1975) experienced similar problems and solved them in the same

manner. Peterson and Adams (1983) used grounded thermocouples

successfully: however, they experienced problems when using

ungrounded thermocouples (unpublished data).










Temperature measurement errors in retort pouches are usually

found to be one of two types or a combination of both. The first and

most common error appears as a tailing up or down of the straight line

portion of the heating curve. This effect is similar to what is

observed when a heating curve is plotted with an erroneous retort

temperature. The other error encountered is erratic jumps up or down

in the recorded temperature. These errors can be very severe, often

resulting in recorded temperatures well above the actual retort

temperature. The occurrence of these errors is inconsistent and the

magnitude of the error will probably be different for each heat

penetration test. If errors occur, they usually become apparent as

the temperature of the pouch contents approaches within 10 to 15 F of

the retort temperature.


The purpose of this research was to examine the nature and

occurrence of these temperature measurement errors in retort pouches,

and to evaluate the effectiveness of thermocouple type and methods of

grounding in alleviating these errors.


Materials and Methods


Two 10 X 9 X 1 inch polycarbonate slabs (Tuffac, Rohm and Haas

Co., Philadelphia, PA) were used. Each slab was equipped with 4 holes

(#50 drill, 0.07 inch diameter) at various heights in the slab (slabs

1 and 2 in Table 3-1). The thermocouple holes were located in the

infinite slab region of each slab as determined by comparing results

from the exact mathematical solutions of the heat conduction equations










for the infinite slab and parallelepiped geometries, as previously

discussed in Chapter 3.


The polycarbonate slabs were processed in commercially available

retort pouch material (0.5 mil PET/ADH/0.5 mil Al foil/4 mil PP)

supplied courtesy of the American Can Company (Greenwich, CT). The

pouch material was formed into a pouch with 1/4 inch wide impulse

seals (14H/HTV, Vertrod Corp., Brooklyn, NY). The thermocouples were

introduced into the pouch by means of stuffing boxes (C5.2, O.F.

Ecklund, Cape Coral, FL). The final seal of each pouch was

accomplished with a vacuum impulse sealer (Multivac M-3-II, Koch,

Kansas City, MO). Each pouch was subjected to 29 inches of vacuum

(inches of mercury) for at least 2 minutes prior to sealing.

Following the vacuum seal each pouch was given an additional cosmetic

seal.


Thermocouple Types


Four types of copper-constantan (Type T) thermocouples were used:

1) A 6 inch CNL needle thermocouple (O.F. Ecklund, Cape Coral, FL),

consisted of 30 gauge thermocouple wire inside a 1/16 inch O.D.

stainless steel tube. The measuring junction was formed by soldering

the thermocouple wires and stainless steel tube together approximately

1/16 inch back from the end. 2) A 6 inch isolated CNL needle

thermocouple (O.F. Ecklund, Cape Coral, FL) was identical to the

standard CNL type except that the measuring junction was not in

electrical contact with the stainless steel tube; therefore, the

stainless steel tube was a true electrical shield. 3) 30 and 4) 40










gauge wire thermocouples (matched, teflon coated copper and constantan

thermocouple wire: Omega Engineering, Stamford, CT) were fused

together to form the measuring junction. An extra copper wire of the

same gauge was included in the fusion to provide a means of grounding

the measuring junction. The 40 gauge thermocouple wire was very

fragile; therefore, after approximately 6 inches of the wire was

used, the 40 gauge wire was soldered to 24 gauge thermocouple wire

(Thermo Electric, Saddle Brook, NJ). This soldered connection was

secured to the polycarbonate slab using a cement made of polycarbonate

shavings dissolved in dichloromethane. The wire thermocouples were

held in place on the slab with a general purpose sealant silicon caulk

(Dow Corning, Midland, MI). Previous experience had indicated that

when more than one wire was extended through a stuffing box, a loss of

vacuum in the pouch could result during processing. This problem was

solved by using electrical shrink tubing in conjunction with silicon

caulk at the point where the thermocouple wires entered the stuffing

box. The wire thermocouples terminated outside the pouched

polycarbonate slab at a female thermocouple connector (C-7, O.F.

Ecklund, Cape Coral, FL), and the copper ground wire terminated at a

small copper alligator clip.


Copper-constantan teflon-coated thermocouple wire (24 gauge,

Thermoelectric, Saddle Brook, NJ) was used from the measuring

potentiometer into the retort vessel where it was terminated with a

male thermocouple connector (C-6, O.F. Ecklund, Cape Coral, FL). The

back of the male connectors and female connectors (wire thermocouples










only) were filled with silicon caulk (Dow Coming, Midland, MI) to

prevent water intrusion.

Grounding Methods


Three different grounding methods were used: a retort grounding

method, a potentiometric grounding method, and no ground. The retort

grounding method was accomplished by attaching the ground wire from

the 30 and 40 gauge wire thermocouples to the frame of the retort by

means of an alligator clip. The non-isolated and isolated sheathed

thermocouples were grounded to the frame of the retort by a copper

wire having two alligator clips. One end of the ground wire was

connected to the portion of the stainless steel sheath that extended

outside the pouch; the other end was attached to the frame of the

retort. Thus, for all the retort grounded thermocouples except for
the isolated type, the measuring junction was grounded to the frame of

the retort. The measuring junction of the isolated thermocouple was
not grounded; however, the stainless steel shield surrounding it was.

For the potentiometric grounding method, a copper wire was connected

to the positive lead at the measuring potentiometer and grounded to
the frame of the retort. For the potentiometric grounding method and

no ground, the alligator clips on the 30 and 40 gauge wire

thermocouples used for the retort grounding method were wrapped in

glass cloth electrical tape.











Processing


The retort used in this study was designed and built at the

University of Florida for processing institutional size retort pouches

(Adams et al., 1983). One inch confining stainless steel racks

constructed of 16 gauge stainless steel with 40% open area (3/8 inch

holes on 9/16 inch staggered centers) were used for this study. Two

of the total eight positions in the racking system contained pouched

polycarbonate slabs; all other positions contained pouches filled

with a 10% bentonite suspension. The pouched polycarbonate slabs were

repeatedly processed until loss of vacuum occurred. At that point the

slabs were repouched as described earlier.


This study examined two heating media; flowing pressurized water

and a steam-air mixture. For pressurized water processing the retort

was loaded at ambient temperature and processing water, preheated to

270 F, was introduced into the pressurized retort. The slabs were

processed at 250 F with a come up time of approximately 2 to 3 min.

The processing water was recirculated at a rate of 110 gal/min

(Reynolds number = 33,000), and an overriding air pressure of 10 psi

was used (total system pressure of 25 psig). The polycarbonate slabs

were processed until the slowest heating thermocouple had reached

245 F at which time the processing water was drained and the retort

was flooded with cooling water. The cooling water was circulated

while being continuously vented and replenished, and required less

than 10 min. to reach a temperature within 1 F of the final

temperature.











For steam-air processing, a mixture of 90% steam and 10% air was

used. The retort was filled with the pouched polycarbonate slabs at

ambient temperature. The pouches were processed in 100% steam for the

first 1.5 minutes, at which time the air pressurization was initiated.

After this venting and come up, the slabs were processed at 250 F,

with a total system pressure of 18.4 psig, and a flow rate of 462 lbs

steam-air/hour (394 lbs steam/hour) resulting in a Reynolds number of

4,800.


Experimental Design


This experiment was designed to give comparisons between

thermocouple type and grounding methods without confounding any main

effects or interactions. Two assumptions were made for this design.

The first assumption was that the 2 one inch slabs were identical as

shown to be a valid assumption in Chapter 3. Second, it was assumed

that different positions in the same slab would result in the same

fh-value. Theoretically, this is a reasonable assumption.


Table 4-1 shows the experimental design used for the first

replicate of the pressurized water cook portion of the experiment.

Slab numbers in Table 4-1 correspond to those listed in Table 3-1.

Thermocouple holes were numbered sequentially from top to bottom for

each slab. The same design was used for the second replicate. The

major difference was that the isolated and non-isolated thermocouples

were placed in slab 2 and the 30 and 40 gauge wire thermocouples were

placed in slab 1. The thermocouple hole within a slab also contained

a different thermocouple type for the second replicate. For example,













Table 4-1.


Experimental design for first replicate of pressurized
water cook.


Slab 1

Thermocouple Type

a
I(1) NI(2) 1(3) NI(4)

b
NO NO PO RT
PO PO NO RT
RT RT PO NO
RT PO NO NO
RT NO PO PO
PO NO RT RT


Slab 2

Thermocouple Type


40(1) 30(2) 40(3) 30(4)


NO NO PO RT
PO PO NO RT
RT RT PO NO
RT PO NO NO
RT NO PO PO
PO NO RT RT


I = Isolated sheathed thermocouple type.
NI = Non-isolated sheathed thermocouple type.
30 = 30 gauge wire thermocouple.
40 = 40 gauge wire thermocouple.
( ) = Slab thermocouple hole number.

NO = No ground.
PO = Potentiometric grounding method.
RT = Retort grounding method.


Retort
Run


1
2
3
4
5
6


__ -----II--~--~--~----- ----------------- ----I--











thermocouple holes number 1 and 3 for slab 1 contained the isolated

thermocouple type during the first replicate. For the second

replicate the isolated thermocouple type would be found in holes

number 2 and 4 in slab 2. This experimental design was repeated for

steam-air processing. This design yielded eight heating curves for

each combination of thermocouple type and grounding method.


Data Analysis


Temperature histories for both pressurized water and steam-air

heating and for pressurized water cooling were recorded with a data

logger (Digistrip II, Kaye Instruments, Bedford, MA). The

potentiometer was equipped with 120 dB common mode noise rejection

circuitry. The data was analyzed using non-linear regression analysis

of the exponential function describing the straight line portion of

the heating or cooling curve (Statistical Analysis System, release

79.6, SAS Institute, Cary, NC). The model used for heating data was

(Trh-T)=C*EXP(-2.302585*t/fh)

where C is the intercept of the straight line portion of the heating

curve at time zero. The model for cooling data was the same as that

for heating data with the necessary adjustments.


In all cases, 18 temperature values, each separated by one

minute, were analyzed for both heating and cooling. For heating, the

last of the temperature values analyzed was 5 F below retort

temperature, or approximately 245 F. The last of the cooling values

analyzed was 10 F above the cooling retort temperature, or

approximately 80 F. Heating and cooling retort temperatures were











determined by averaging the temperature values of two thermocouples

located at the exit of the retort racking system.


Statistical analysis of the resulting heat penetration data was

accomplished using the Statistical Analysis System (release 79.6, SAS

Institute, Cary, NC). Response variables analyzed were both fh or

fc and the residual sum of squares generated by the non-linear

regression performed on each heating or cooling curve. The residual

sum of squares was appropriate to use since all regressions contained

an identical number of observations.


Results and Discussion


Non-linear regression was chosen for analysis of the various

heating curves as opposed to a log transformation followed by linear

regression. Non-linear regression avoided the problem of changes in

the error distribution around different temperature observations which

would be experienced with a log transformation of the temperature

data.


Fig. 4-1 through Fig. 4-3 illustrate heating curves obtained for

various residual sum of squares values. A very good straight line was

obtained with residual sum of squares values of 0.57 or less (Fig.

4-1). Deviations from the straight line became apparent when residual

sum of squares equaled 0.75 and 3.00 (Fig. 4-2 and 4-3, respectively).

No figures with a residual sum of squares greater than 3.00 are

presented because it was felt that Fig. 4-2 and 4-3 were unacceptable

heating curves for polycarbonate slabs. The typical problem observed





74







245--


240- -
u /o




'. 220-- /
S/o


200- /
0
o
/


150- /
o
100- /
0 0
50-1-

0 10 20 30
Time (min)


Figure 4-1. Heating curve with residual sum of squares equal to 0.57.











0
/oQ

/o

o/
O/
//


/


/
o


) 20 30
Time (min)


Figure 4-2. Heating curve with residual sum of squares equal to 0.75.


0
LL.


CL

I-












245- -

0

240- -
- /
L 0
0


S220-- o
E /



0
o
50-- /
o

100-- o
o
50t

0 10 20 30
Time (min)


Figure 4-3. Heating curve with residual sum of squares equal to 3.00.











as the residual sum of squares increased was a tailing off or tailing

up of the straight line portion of the heating curve as the measured

temperature approached retort temperature (Fig. 4-2 and 4-3). Since

values plotted were the differences between pouch and retort

temperatures, plotting at a higher or lower retort temperature would

straighten out these lines. The retort temperature measured by

thermocouples in the retort was known to be accurate since those

thermocouples were checked in an oil bath at 250 F and were found to

agree within 0.3 F. This accuracy in measurements was coupled with
the observation that the same thermocouple measuring a slab

temperature in the retort did not consistently give the same type or

magnitude deviation from the straight line portion of the heating

curve between replicate processing runs. Thus, this effect was

considered real and not just an artifact created by errors in

temperature measurement. Additionally, since the rate of temperature
change (slope) was the dependent parameter of interest, the effect of
error in absolute temperature value measured (+1.5 F; potentiometer

specifications) would be minimized.


Pressurized Water Heating


The residual sum of squares of the different thermocouple types

and grounding methods for heating with pressurized water are presented

in Fig. 4-4. Statistical analysis of the data revealed a significant
interaction between thermocouple type and grounding method. The

isolated thermocouple resulted in acceptable heating curves with low

residual sum of squares in all cases; thus, data obtained from this













1000+


o o

88 1


0I














0 I


0
0
o



0
0


o















0

o I
0




0

0
0
0 I


0
0







0

0


0
0















0
0
o I


0
0
0 0
0


0


0


NO RT PO NO RT PO NO RT PO NO RT PO
I NI 30go 40ga
Figure 4-4. Residual sum of squares obtained for pressurized water
heating. I = Isolated thermocouple; NI = Non-isolated
thermocouple; 30ga = 30 gauge wire thermocouple; 40 ga =
40 gauge wire thermocouple; NO = No ground; RT = Retort
ground method; and PO = Potentiometric ground method.
a indicates 2 points at the same residual sum of squares
value.


8
o


100+


en
C,



0 10




4-
E
UC

-o 1.0-
-o
<3)


0.1+


A


'''''~'~~


I I I


I


,,








type of thermocouple were not dependent on the type or presence of

ground. For the other three thermocouple types, grounding at the

positive lead of the measuring potentiometer resulted in very high
residual sum of squares with a high level of variation. This

indicated that the heating curves obtained did not fit a straight

line. The heating results using the potentiometric grounding method

often showed temperatures near the end of processing as high as 15 F

above retort temperature. Apparently, grounding at the positive lead

of the recording potentiometer led to more problems than it solved in

the processing system used. This method of grounding was included in

the experiment at the suggestion of the manufacturer of the recording

potentiometer.


The non-isolated thermocouple type gave fairly good results with

both the retort ground and no ground. Grounding in the retort

resulted in a smaller spread in the data than did no ground; however,

there was not a significant difference between the two methods.
Previous work with non-isolated thermocouples which were not grounded

resulted in more variable results. In this experiment, all but one of

the non-isolated thermocouples without a ground gave an acceptable

regression fit of the data. Previous work with a 10% bentonite

suspension and pouched seafood revealed the tailing data illustrated
in Fig. 4-2 and 4-3 and also instances of instability in temperature

measurement which were not observed in this experiment. A typical

example of this instability found with a 1.0 inch thick pouched

bentonite suspension is shown in Fig. 4-5. The previous work
indicated that the failures observed should have occurred at a rate of



















000
o


000
00
O


I0 20 30
Time (min)


Figure 4-5.


Heating curve illustrating instability in temperature
measurement.


245+


240+


LL


0


I,


220-


200+


OO0O


150-

100-

50-











approximately 3 out of 8 heating curves. Peterson and Adams (1983),

working with a 10% bentonite suspension, also revealed that the retort

grounding method with the non-isolated thermocouple type essentially

eliminated the problems.


The 30 and 40 gauge wire thermocouples gave similar results for

each grounding method. In each case, the retort grounding method

resulted in a greater degree of scatter in the residual sum of squares

values than did the no grounding method (Fig. 4-4). This was an

unexpected result and was contrary to findings of Pflug (1975). With

wire thermocouples, both the retort grounding method and the no ground

method resulted in a large percentage of heating curves which did not

have an acceptable fit of the regression equation, especially when

compared to the isolated thermocouple type.


Statistical analysis using Duncan's multiple range test of the

residual sum of squares for the pressurized water cook demonstrated no

significant difference at the 0.05 significance level between

thermocouple types and grounding methods with one exception. The

non-isolated CNL thermocouple coupled with the potentiometric

grounding method was significantly different from all other

combinations. Even though there was not a statistically significant

difference, there was a practical difference between the methods as

illustrated in Fig. 4-4. The low degree of scatter in the residual

sum of squares for the isolated and grounded non-isolated

thermocouples would suggest a greater confidence in these methods than

those demonstrating the larger scatter in data.











In addition to the consideration of how well the regression

equation fit the heat penetration data, one must be concerned with the

fh-value obtained from the regression. An acceptable regression fit

is inconsequential if the fh-value obtained is erroneous. Table 4-2

summarizes the mean fh-values observed with the pressurized water

cook. The fh-value obtained should be approximately equal to 20

minutes, which can be calculated from the thermal diffusivity of the

polycarbonate slab (Olson and Jackson, 1942). The conclusions drawn

from the fh-values obtained agreed quite well with those found in

analysis of the residual sum of squares. The isolated thermocouple

gave acceptable fh-values independent of the grounding method used

and the non-isolated thermocouple type gave acceptable results with no

ground and the retort grounding method. The 40 gauge wire

thermocouple with no ground gave an fh-value close to that expected;

however the 95% confidence interval was large in comparison to that of

the acceptable isolated and non-isolated thermocouple types. The 30

and 40 gauge wire thermocouples using the retort grounding method gave

lower fh-values than were expected. By far the best choice for

thermocouple type is the isolated type, since it gave the expected

fh-value with a very small 95% confidence interval independent of

the type of grounding method used.


Steam-air Heating


Fig. 4-6 contains the results of the residual sum of squares for

each thermocouple type and grounding method for heating with 90% steam

and 10% air as the processing media. These results are essentially













Table 4-2. Mean fh-values found for pressurized water heating.


a,b


Thermocouple
Type


Isolated


Non-isolated


30 gauge


40 gauge


No
Ground

AB
19.82 + 0.35

AB
19.12 + 0.71

A
21.46 + 3.05

AB
19.50 + 1.73


Grounding Methods

Retort
Method

AB
20.08 + 0.31

AB
19.87 + 0.57

B
18.05 + 1.06

B
18.08 + 0.85


Potentiometric
Method

AB
19.81 + 0.43

D
9.56 + 4.00

B
17.17 + 3.93

C
14.24 + 2.04


Mean fh-value.+ 95% confidence interval.


Means with same letter are not significantly different
(Duncan's Multiple Range Test, p<0.05).











1000-


0
I






























o



I


0





0
o
OI















0
o



0 I
o


SI











0 o
0 o
o


I

















0 0

0
0
81




0I
o0
0
0o


1 1 1 1 1 1 i s i l


NO RT PO NO T PO NO R PO NO RT PO
I NI 30go 40go
4-6. Residual sum of squares obtained for heating with 90%
steam, 10% air. I = Isolated thermocouple; NI = Non-
isolated thermocouple; 30ga = 30 gauge wire thermocouole;
40ga = 40 gauge wire thermocouple; NO =No ground; RT=
Retort ground method; and PO = Potentiometric ground
method. a indicates 2 points at the same residual sum
of squares value.


oo
0 @
0
0


100+


(A,

0





cC,



U,


1.0+


0.1+


0.01


Figure


' '


I


Z I











identical to those found with the pressurized water cook. Overall,

the steam-air cook residual sum of squares results demonstrated less

scatter than those obtained with the pressurized water cook. The

isolated thermocouple type with the potentiometric ground, method and

the non-isolated thermocouple type with the retort ground method,

displayed more variation than with the pressurized water cook;

however, the values were still, for the most part, acceptable.


The mean fh-values obtained for steam-air heating were less

variable for isolated and non-isolated thermocouple types and

generally more variable for the 30 and 40 gauge wire thermocouple

types when compared to fh-values obtained for pressurized water

heating. This can be seen by comparing the size of the 95% confidence

intervals in Tables 4-2 and 4-3. As in the pressurized water cook,

the isolated thermocouple with any ground type and the non-isolated

thermocouple with the retort ground method or no ground resulted in

the expected fh-value. The 30 gauge wire thermocouple with no

ground gave an fh-value close to the expected value of 20 minutes,

but had a fairly large 95% confidence interval.


Cooling Data Analysis


Cooling data obtained from the pressurized water cook was

analyzed in the same manner as the heating data. The residual sum of

squares results for cooling can be seen in Fig. 4-7. Unlike the

heating results, cooling data did not show a significant interaction

between thermocouple type and ground method. Additionally, there was

not a statistically significant difference between main effects of













Table 4-3. Mean fh-values found for steam-air heating.


a,b


Thermocouple
Type


Isolated


Non-isolated


30 gauge


40 gauge


No
Ground

CD
20.45 + 0.24

CD
20.21 + 0.59

BCD
20.90 + 1.04

BCD
21.78 + 4.02


Grounding Methods

Retort
Method

CD
20.52 + 0.31

D
20.38 + 0.50

ABC
26.00 + 4.80

A
29.46 + 7.40


Potentiometric
Method

CD
20.54 + 0.26

CD
16.54 + 2.74

AB
25.32 + 5.18

A
28.83 + 8.58


Mean fh-value + 95% confidence interval.


Means with same letter are not significantly different
(Duncan's Multiple Range Test, p<0.05).











1000+


~1
cI









01




01


0





0
0 0
I












0 0
S0 0o
S8 I



o

o 0
0




I
o I


-0

E
0
o



U,
a,


NO RT PO NO RT PO NO RT PO NO RT PO
I NI 30ga 40ga
Figure 4-7. Residual sum of squares obtained for cooling. I =
Isolated thermocouple; NI = Non-isolated thermocouple;
30ga = 30 gauge wire thermocouple; 40ga = 40 gauge
wire thermocouple; NO = No ground; RT = Retort ground
method; and PO = Potentiometric ground method.
a indicates 2 points at the same residual sum of
squares value.


I












0I
a 0 0
o 8


a o


0o I
8 I
@


100+


0


0
0
0 0
0







. I


0
8
o0




0


0

0 8
0 0


0.1+


nn I


II__~


(

(


(


I


v