3 ram.

.1. f... v.‘ ..

 

 

 

THE-39r-

k.)

$52

This is to certify that the

thesis entitled

Modeling Condensing-Convective Boundary
Conditions In Moist Air Impingement Ovens

presented by

Scott Clark Millsap

has been accepted towards fulfillment
of the requirements for

M.S. degree in Biosystems Engineering

 

    

' ajor professor

Date //g‘~€ 0;

0—7639 MS U i: an Affirmative Action/Equal Opportunity Institution

 

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

13531312004

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c;/ClRC/DateDue.p65-p.15

 

MODELING CONDENSING-CONVECTIVE BOUNDARY
CONDITIONS IN MOIST AIR IMPINGEMENT OVENS

By

Scott Clark Millsap

A THESIS

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

MASTER OF SCIENCE
Department of Agricultural Engineering

2002

Abstract

Modeling Condensing-Convective Boundary Conditions in
Moist Air Impingement Ovens

by

Scott C. Millsap

Previously developed equations for impingement heat transfer coefficients
experienced by a flat plate under an array of slot nozzles were compared to values
measured for discrete model food products in a commercial moist air impingement oven.
A term was included to account for heat transfer to the edges of discrete objects under
impingement flows. The best correlation predicted the convective heat transfer
coefficients to an average absolute error of 1.9%, with a standard error of prediction
(SEP) of 1.9 W/mzK.

Condensation heat transfer was successfully modeled as a convective mass
transfer phenomenon, driven by a water vapor concentration gradient between the bulk
impinging fluid and the product surface. The analogous nature between heat and mass
transfer was used to predict the mass transfer coefficients from correlations deve10ped for
pure convective heat transfer.

The same correlation found to most accurately predict the convective heat transfer
coefficient also most accurately predicted the mass transfer coefficient for impingement
flow. When accounting for edge effects, the model predicted the mass transfer

coefficient to an average absolute error of 12.7%, with an SEP of 16.4 mm/s.

Acknowledgments

I would like to thank my family, who have always supported me in every life
endeavor I have attempted. To my mother, who taught me that you have no control over
the curve balls life throws you, but do you have control over how you deal with them. To
my father, who taught me that being educated does not make one intelligent. To all my *
grandparents, who taught me that true happiness is being comfortable with the man in the
mirror.

I thank my friends, both past and present, who have helped shape my life. To
those who have stood with me through my mistakes and personal growth, and to those
who have not. Each person we know helps shape who we are, and I thank you all for
shaping me into who I am.

My gratitude goes to Stein-DST, a business of FMC FoodTech, Sandusky, OH and
Stein-DSI employees Mr. Bob Swackhamer, Dr. Nahed Kotrola, and Mr. Todd Gerold
for providing access to, and assistance in the operation of the J SO-IV moist air
impingement oven.

Finally, I thank Dr. Bradley Marks, for his support and guidance through this
project. For his belief in my intellectual ability, and for the financial support from his

research program. Without him, this thesis would not have been possible.

iii

TABLE OF CONTENTS

LIST OF TABLES .............................................................................
LIST OF FIGURES ...........................................................................
NOMENCLATURE ...........................................................................
CHAPTER 1: INTRODUCTION
1 . 1 Background .....................................................................
1.2 Objectives ......................................................................
CHAPTER 2: LITERATURE REVIEW
2.1 Impingement to a Flat Plate ..................................................
2.1.1 Overview ...............................................................
2.1.2 Local heat transfer ...................................................
2.1.3 Average heat and mass transfer ....................................
2.1.4 Surface motion effects ...............................................
2.1.5 Numerical studies ....................................................
2.2 Impingement in the Food Industry ..........................................
2.3 Condensation from Air/Water Vapor Mixtures ...........................
2.3.1 Overview ...............................................................
2.3.2 Vapor pressure driven ................................................
2.3.3 Dew point temperature driven ......................................
2.3.4 Deficiencies ...........................................................
CHAPTER 3: METHODS AND MATERIALS
3.1 Overview ........................................................................
3.2 General Theory ................................................................
3.2.1 Assumptions ..........................................................
3.3 Model Development ..........................................................
3.4 Calculation of Gas Properties ................................................
3.4.1 Properties of dry air ..................................................
3.4.2 Properties of steam ...................................................
3.4.3 Steam—air mixtures ...................................................
3.4.4 Calculation of water vapor concentration .........................
3.5 Calculation of Aluminum Properties .......................................
3.6 Experimental Design .........................................................
3.6.1 Test conditions .......................................................
3.6.2 The oven ...............................................................
3.7 Test Protocols ..................................................................
3.7.1 Aluminum block preparation .......................................
3.7.2 Temperature measurement ..........................................

iv

Pg
vi

vii

22
23
26
27
3 l
32
33
34
37
38
4O
4O
42
46
46
47

3.8

3.7.3 Oven runs .............................................................. 48

Data Analysis .................................................................. 49
3.8.1 Estimation of havg from raw data ................................... 49
3.8.2 Estimation of hm,avg from raw data ................................. 50

CHAPTER 4: RESULTS AND DISCUSSION

4.1 Prediction of havg ............................................................... 53

4.1.1 Data Analysis ......................................................... 53

4.1.2 Evaluation of the lumped parameter assumption ................ 56

4.1.3 Gardon and Akfirat (1966) .......................................... 56

4.1.4 Saad et a1. (1980) ..................................................... 57

4.1.5 Martin (1977) ......................................................... 58

4.1.6 Summary ............................................................... 59

4.2 Effects of Surface Motion .................................................... 62

4.3 Prediction of hm,g ............................................................ 63

4.3.1 Data Analysis ......................................................... 63

4.3.2 Gardon and Akfirat (1966) .......................................... 65

4.3.3 Saad et a1. (1980) ..................................................... 66

4.3.4 Martin (1977) ......................................................... 68

4.3.5 Summary ............................................................... 70

4.4 Contribution of the Edge Effect Term ...................................... 72

4.5 Errors in Mv~90% Tests ...................................................... 73

4.6 Temperature Dependency .................................................... 75

4.7 Dew Point Temperature Prediction ......................................... 77
CHAPTER 5: CONCLUSIONS

5.1 General Conclusions .......................................................... 81

5.2 Recommendations for Future Work ......................................... 83

REFERENCES ................................................................................. 85

APPENDICES
A: Predicted and Measured Convective Heat and Mass Transfer Coefficients. . .. 89
B: Block Temperature Data ............................................................... 102

LIST OF TABLES

Table 3.1: Previously developed heat and mass transfer correlations for
ASN flow to a flat plate.

Table 3.2: Composition of aluminum alloy 2024.

Table 3.3: Heat capacities of aluminum alloy 2024 individual components.

Table 3.4: Molecular weights of selected elements (g/mol).
Table 3.5: Experimental test conditions.

Table 4.1: Measured values of the impinging gas moisture content by
volume, for MV~O% runs.

Table 4.2: Comparison of tested variables vs. range of validity of the
models used.

Table 4.3: Effects of edge consideration on % average absolute error in
predicted vs. measured havg for MV~0% runs.

Table 4.4: Effects of edge consideration on % average absolute error in
predicted vs. measured hmavg for MV>O% runs.

Table 4.5: Measured values of M" for the Mv~90% runs.

Table 4.6: Predicted dew point temperature for the measured Mv values
used.

Table A. 1: Predicted and measured convective heat and mass transfer
coefficients from the model based on Gardon and Akfirat (1966).

Table A2: Predicted and measured convective heat and mass transfer
coefficients from the model based on Saad et a]. (1980).

Table A3: Predicted and measured convective heat and mass transfer
coefficients from the model based on Martin (1977).

vi

28

38
39
39
40
53

60

73

73

74

78

90

94

98

 

LIST OF FIGURES

Figure 2.1.1: Flow from an array of slot nozzles.

Figure 2.1.2: Lateral variation of local heat transfer coefficients between a
plate and ASN (interpreted from Gardon and Akfirat, 1966).

Figure 3.1.1: Overall study design.

Figure 3.2.1: Boundary layer temperature and water vapor concentration
gradients.

Figure 3.3.1: Side view of impingement flows over a discrete object in a
JSO-IV oven.

Figure 3.6.1: Stein J SO—IV oven exit velocities.

Figure 3.6.2: Impingement gas flow inside a JSO-IV oven (provided by
Stein-DSI, Sandusky, OH).

Figure 3.6.3: Interior view of top impingement slot array (fingers) in a
JSO-IV oven.

Figure 3.6.4: Interior view of bottom impingement slot array (fingers) in a
JSO-IV oven.

Figure 3.7.1: Aluminum blocks on the carrying tray before a test.

Figure 4.1.1: Measured block temperatures for Mv~0%, F=75%, runs (rep.
1 only).

Figure 4.1.2: Predicted vs. measured havg for MV~O% runs using the model
based on Gardon and Akfirat (1966).

Figure 4.1.3: Predicted vs. measured havg for Mv~0% runs using the model
based on Saad et a1. (1980).

Figure 4.1.4: Residuals of havg vs. jet exit velocity using the model based on
Saad et al. (1980).

Figure 4.1.5: Predicted vs. measured havg for Mv~0% runs using the model
based on Martin (1977).

Figure 4.1.6: Predicted vs. measured havg values estimated from Mv~0%
runs.

vii

23

25

29

41

43

45

46

47

54

56

57

58

59

6O

Figure 4.3.1: Incremental values of hm,avg for the Too=l77°C, Mv=30%, and
F=100% test condition.

Figure 4.3.2: Predicted vs. measured hm.avg values using the model based on
Gardon and Akfirat (1966).

Figure 4.3.3: Residuals of h"mg using the model based on Gardon and
Akfirat (1966) vs. the Mv set point.

Figure 4.3.4: Predicted vs. measured hmavg values using the model based on
Saad et al. (1980).

Figure 4.3.5: Residuals of hmavg using the model based on Saad et al.
(1980) vs. the My set point.

Figure 4.3.6: Residuals of hmvg using the model based on Saad et al.
(1980) vs. jet exit velocity.

Figure 4.3.7: Predicted vs. measured hm.avg values using the model based on
Martin (1977).

Figure 4.3.8: Residuals of hmavg using the model based on Martin (1977)
vs. the Mv set point.

Figure 4.3.9: Predicted vs. measured hmalvg values for all runs (rep. 1 only).

Figure 4.6.1: Residuals of hmalvg vs. gas temperature for the model based on
Gardon and Akfirat (1966).

Figure 4.6.2: Residuals of hmavg vs. gas temperature for the model based on
Saad et a1. (1980).

Figure 4.6.3: Residuals of hmavg vs. gas temperature for the model based on
Martin (1977).

Figure 4.7.1: Predicted vs. measured block temperature for T..=232°C,
F=75%, and Mv=l3 and 30% using the model based on Martin
(1977).

Figure 4.7.2: Predicted vs. measured block temperature for T..=232°C,
F=75%, and Mv=50 and 70% using the model based on Martin
(1977).

Figure 4.7.3: Predicted vs. measured block temperature for T...=232°C,
F=75%, and Mv~90% using the model based on Martin (1977).

Figure 8.1: Data collected for the T..=121°C, Mv=0% set point runs.

viii

64

65

66

67

68

68

69

70

71

76

76

77

79

8O

80

103

 

Figure 3.2:
Figure 8.3:
Figure B.4:
Figure 8.5:
Figure B.6:
Figure B.7:
Figure 8.8:
Figure B.9:
Figure B.lO:
Figure B.1 1:
Figure B.l2:
Figure B.l3:
Figure 3.14:

Figure B.15:

Data collected for the T...=l77°C, MV=O% set point runs.

Data collected for the T..=232°C, Mv=0% set point runs.

Data collected for the T..=121°C, Mv=30% set point runs.
Data collected for the T...=177°C, Mv=30% set point runs.
Data collected for the T...=232°C, Mv=30% set point runs.
Data collected for the Tm=121°C, Mv=50% set point runs.
Data collected for the T..=177°C, Mv=50% set point runs.

Data collected for the Too=232°C, MV=50% set point runs.

Data collected for the T...=121°C, Mv=70% set point runs.
Data collected for the T..=177°C, Mv=70% set point runs.
Data collected for the T...=232°C, MV=7O% set point runs.
Data collected for the T..=12l°C, Mv=90% set point runs.
Data collected for the T...=177°C, Mv=90% set point runs.

Data collected for the Tm=232°C, Mv=90% set point runs.

104

105

106

107

108

109

110

111

112

113

114

115

116

117

92w>

ErmmeUD-QO

:3"
3 no

;;a*

c

<

:51wa

CD

mgimmgon

:J"

xg£§=~€

Nomenclature

product surface area, m2

Sutherland constant, K

Biot number, hE/2k

water vapor concentration, kg/m3

specific heat, J/kg*K

product diameter, m

mass diffusivity of water vapor in air, mz/s
disk thickness, m

fraction nozzle open area, w/ZS

fan speed, % full

nozzle-to-impingement surface spacing, m
convective heat transfer coefficient, W/(mzK)
latent heat of condensation, J/kg

mass transfer coefficient, m/s

thermal conductivity, W/(m K)

mass, kg

impingement fluid moisture content, %
surface motion parameter, pm, /pjuj

Nusselt number, hw/k
Pressure, Pa

Saturation vapor pressure, Pa
Prandtl number

heat transfer (kJ)

Reynolds number, uwp/ u

heat transfer surface half width, m
distance from jet centerline, m
Schmidt number, u/pD

Sherwood number, hmw/ D

Block temperature, K

gas velocity, m/s

heat transfer surface velocity, m/s
nozzle width, m

molecular weight, g/mol

molar fraction, -

Greek Letters

wee-o:

viscosity, kg/m*s

density, kg/m3

sub-equation in formulation for viscosity of mixtures

sub-equation in the formulation for thermal conductivity of mixtures
sub-equation in the Martin (1977) model

Subscripts

= of air

avg

eff

imp

average

condensation

dew point

over the edges

effective

of steam

impingement

at the jet exit

bulk impinging fluid

initial

at time step i, or component i
mix

at product surface, or steam
total

vapor

xi

Chapter 1
1 Introduction

1.1 Background

Impingement heat transfer has a variety of industrial applications, including:
annealing of non-ferrous sheet metals, tempering of glass, drying of paper and textiles,
the cooling of electrical components and turbine blades, and the processing of food
products. Impingement flows appeal to these industries, because impingement produces
heat transfer coefficients an order of magnitude higher than do other gaseous heat transfer
methods. Impingement flows also allow for a high degree of control over any
temperature treatment of a product with large surface areas.

This potential for process control allows processors to increase production quality
and quantity. Heat transfer rates due to impingement flows can be varied either by
adjusting flow rates and temperature differentials or by adjusting several geometric
parameters. One of the most important parameters is the choice of jet configuration.
There are four main configurations to choose from: single round nozzle (SRN), single
slot nozzle (SSN), array of round nozzles (ARN), and array of slot nozzles (ASN). The
present study will focus on slot nozzle configurations; however, much research has been
conducted in the area of round nozzle flows, which may be useful for many applications.

The geometry of the slot array, including jet width (w), jet-to-surface distance
(H), and jet center-to-center distance (28), can be adjusted as heat transfer rate

requirements change. Although the number of variables controlling impingement heat

transfer rates complicate the design of an air impingement oven, they also allow for
flexibility and creativity.

Impingement flows have been utilized in the food industry for the baking of bread
products since the late 1980’s. Currently, over 100,000 industrial dry air impingement
ovens are used in the baking industry (Ovadia & Walker, 1998). The impingement gas
temperature within impingement ovens can be reduced up to 25°C, compared with
standard convection ovens, without any loss of product quality (Wahlby et al., 2000).
Thus, thermal energy requirements can be dramatically reduced.

The benefits of condensation heat transfer, in conjunction with convective heat
transfer, is beginning to be realized in the meat processing industry. The presence of
water vapor and subsequent condensation heat transfer raises the product surface to the
impingement gas dew point temperature very quickly, thus reducing cooking times.
Processing meat products with combined condensation/convection heat transfer also
reduces mass loss from the food product for two reasons. Water condensed on the
product surface during the initial stage of cooking is evaporated off before product
moisture is removed, thus saving product mass, and the high water vapor concentration in
the impingement gas reduces water vapor transfer from the product surface after it
reaches the dew point temperature.

Although considerable research has been conducted in the field of convective
impingement heat transfer (see Chapter 2), it is insufficient to meet the needs of the meat
processing industry in modeling impingement cooking with condensation. Many
correlations have been developed for different flow configurations and experimental

conditions (Obot et al., 1979). However, the problem of selecting the correct correlation

for a particular industrial application is amplified by the limited understanding of the
fundamental effects of nozzle geometry, confinement, spent gas exits, and surface motion
effects. In addition, the effects of large temperature gradients between the impinging gas
and surface, and condensation from moist air impingement onto a cold surface, have not
been studied in sufficient detail to accurately predict the heat and mass transfer

coefficients experienced by food products in a commercial moist air impingement oven.

1.2 Objectives

Several correlations have been developed over the last four decades to predict the
convective heat transfer coefficients experienced by a flat plate under an array of slot
nozzles. Research has also been conducted to describe condensation as a mass transfer
phenomenon, driven by vapor pressure or dew point temperature gradients under flows
other than impingement. However, no studies have considered the application of
condensing-convective heat transfer to discrete objects passing through an array of slot
nozzles impinging moist air. Knowledge in this area will gain importance as the benefits
of impingement flows, combined with condensation heat transfer, are realized in the food
processing industry. Therefore, a reliable method for predicting heat transfer rates in
these systems is vital to process simulations, design, optimization, and control.

The objectives of this study were:

0 To evaluate the use of previously developed impingement convective heat transfer
correlations to the industrial application of commercial food ovens utilizing slot
nozzle arrays.

0 To describe condensation heat transfer to a cool object in a moist air impingement

oven as a mass transfer phenomenon analogous to convective heat transfer, thus

allowing condensation heat transfer to be modeled using a mass transfer

coefficient and concentration gradient.

Chapter 2

2 Literature Review

2.1 Impingement to a Flat Plate

2.1.1 Overview

Only a concise review of the literature relevant to flow from an array of slot
nozzles (ASN) to food product surfaces in air impingement ovens is presented. A slot is
hereby defined as an opening with a length to width ratio much greater than 10. More
general reviews of impingement heat transfer were previously published by Mujumdar
and Douglas (1972), Martin (1977), Obot et al. (1979), Saad (1981), and van Heiningen
(1982).

In a process as complicated as impingement flow heat and mass transfer, a
completely analytical solution is impractical. Rather, a dimensional analysis is far more
reasonable. Most researchers in this field have theorized that the equation for the Nusselt
and Sherwood numbers would be of the form (Obot et al., 1979):

Nu = f(Rej,Pr,H/w,f)
eqn.2.1

Sh : f(Rej,SC,H/W,f)
eqn.2.2

Surface motion perpendicular to the impinging flow is another factor that
influences heat and mass transfer rates under impingement. Van Heiningen (1982)

proposed the use of a dimensionless surface motion parameter (Mvs) to account for this

effect. This surface motion parameter is essentially the ratio of the surface velocity to the
jet exit velocity.

Almost all industrial applications of impingement heat transfer involve turbulent
impinging jets (i.e., Rej>2,000), rather than fully laminar flow. Some applications may
involve initially laminar flow turning turbulent as a result of jet mixing. Gardon and
Akfirat (1965) showed that the intensity of turbulence from an impinging jet appears to
be uniquely determined by the jet exit Reynolds number and the dimensionless height, or
Rej and I-I/w. In other words, no separate parameter is required to characterize the
turbulence effects. They also concluded that the effect of H/w becomes negligible for
le>8. This suggests that details of the nozzles and flow conditioning prior to
impingement may be important for short impingement distances, but are secondary for
H/w>8.

2.1.2 Local heat transfer

The majority of single slot nozzle (SSN) flow research has been aimed at
determining the local heat transfer coefficients in the stagnation zone directly under a
SSN and, subsequently, at distances away from the stagnation zone. Research has shown
that heat transfer rates are largest in the stagnation zone directly under the impinging jet,
and decrease away from this zone (Gardon and Akfirat, 1966). In applications where
H/w is less than 8, a secondary peak occurs around an s/w of 8, where s is the distance
from the jet centerline (Gardon and Akfirat, 1966). This is a result of the transition from
laminar to turbulent flow as the impinging jet impacts the surface.

Additionally, for ASN configurations (Figure 2.1.1), Gardon and Akfirat (1966)

showed that spatial variation in the local heat transfer coefficient decreases as H/w

increases (i.e., as the nozzles get smaller or farther from the flat plate). This suggests that
as the distance between the nozzle exits and the product surface increases, or as the slots
become more narrow, the details of the slot configurations become less important.

+——2$——->

allR—w ill Ill
1 I l
l l 1

UV UK? UK?

Figure 2.1.1: Flow from an array of slot nozzles.

 

 

 

 

I

 

J

3 nozzles on 5 cm

 

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0
horizontal distance (cm)

Figure 2.1.2. Lateral variation of local heat transfer coefficients between a plate and
ASN (interpreted from Gardon and Akfirat, 1966).

Gardon and Akfirat (1966) evaluated the local Nusselt number under an ASN

impinging ambient air onto an isothermal aluminum plate held 20°C above the ambient

temperature. They developed a correlation for the N usselt number at the stagnation
point. In their experimental set-up, the impingement flows were not confined (i.e., no
walls or barriers around the impingement zone to confine the impingement gas), so the
spent impingement gas was not recycled for continuous use. For those conditions, they

developed the following correlation.

Nuo =1.2Re(1).'58(I-1/w)'°'62
eqn. 2.3

for 2,000<Rej<50,000
14<H/w<60

Although this correlation is informative about the conditions experienced within the
stagnation region, the present study is better served by correlations for average Nusselt
numbers. This can be accomplished by averaging the local Nusselt numbers under an
ASN.
2.1.3 Average heat and mass transfer

Gardon and Akfirat (1966) determined the average heat transfer rates in their
study from the total electrical power input to the isothermal plate, with corrections for
lateral heat losses, and by integrating the local heat transfer rates over the surface.
However, in an attempt to simplify their correlation to two dimensionless parameters,
they defined the characteristic linear dimension for both Re and Nu as the nozzle center-
to-center distance, instead of w, and they used the “arrival velocity” (ua) in the Re

calculation, instead of the jet exit velocity (Uj). Their calculations yielded:

Nu = 0.36 Re?”
eqn. 2.4

for 6,000<Rea<600,000

H/w>8

0.0156<f<0.0625

In this form, eqn. 2.4 is of little use in an industrial application, because the arrival

velocity cannot be easily measured. The slot exit velocity is more easily measured in real

world applications. Gardon and Akfirat used the following equation to relate the arrival

velocity to the exit velocity provided H/w>7. For values of H/w<7, they stated the arrival

velocity was equal to the exit velocity:

Substituting eqn. 2.5 into eqn. 2.4 yields:

Nu = 0.66 Reg?"2 (H / w)_0‘31 f 0-38

7,000<Rej<120,000
H/w > 8

0.0156<f<0.0625

eqn. 2.5

eqn. 2.6

This correlation is more practical than eqn. 2.5, because the Reynolds number is

based on the more easily measured slot exit velocity. In addition, eqn. 2.6 is more

intuitive, because the effects of slot configuration are expressly stated. Gardon and

Akfirat (1966) found eqn. 2.4 to predict their experimental results to within a 10% error.

Martin (1977) worked with confined slot jets of air impinging onto a flat plate

composed of strips of porous stoneware. Each stoneware strip was wetted with distilled

water. This arrangement allowed the local mass transfer rates to be easily measured by

weighing the strips twice in a certain time interval during a period of constant drying.

Martin calculated the mole fractions of water vapor at the stoneware surface as a function
of the surface temperature.

In the Martin (1977) study, no exit ports for the spent impingement gas were
present in the impingement zone; therefore, spent gas was forced to exit the system
laterally between the jets (i.e., parallel to the slots). Clearly, this configuration reduces
overall heat and mass transfer rates due to exiting spent gas interfering with impingement
flows, thus producing variations from one cross-section to another across the
impingement surface. Martin (1977) developed the following equations for average

Nusselt and Sherwood numbers:

Nu Sh _3 2R6.
f//1+/1/f

131.042 _ 560'42 _ 3

eqn. 2.7
Where A = (60 + 4(H / w _ 2)2 )—1/2
1 ,5 00<Rej<40,000

0.008<f<2.5)t(H/S)

2<le<80

 

The use of the term Pix): = 55:42 was developed by Martin through the analysis of heat
. C .

transfer data collected by previous researchers compared to his mass transfer results.
Martin found eqn. 2.7 to predict his measured mass transfer rates to within 15%.

Saad et al. (1980) performed experiments with an ASN configuration comprised
of three slot nozzles, but used symmetrical exhaust ports alternating with the jet nozzles
to eliminate the cross flow effect experienced in Martin’s study. Saad et al. (1980) used

an isothermal plate (T~60°C) of electrically heated porous bronze under an array of cool

10

(ambient) air slot jets. They measured local heat transfer coefficients using a heat flux
transducer, then integrated the local heat transfer coefficient values over the area to
obtain average values. Saad et al. (1980) developed the following correlation for average

Nusselt number experienced by an isothermal flat plate under cool ASN air jets.

Nu = 0.14 Reg-775 (H/wr1286 f 0-314
eqn. 2.8

where 3,000<Rej<30,000
4<le<24
0.0156<f<0.0833
Saad et a1. (1980) found this correlation to predict the average heat transfer coefficients
measured in their study to within a 10% absolute error.

In a study of impingement drying of paper, Polat (1993) endorsed the use of the
equations by Martin (1977) for use in applications involving cross flow effect, and the
Saad et al. (1980) equations for use in applications without cross flow effect.

Kroger and Krizek (1966) studied the local coefficients of mass transfer in
impingement flow of air from a SSN using a naphthalene plate sublimation technique.
By splitting the naphthalene plate into narrow strips, they could determine the local
Sherwood number values by measuring weight and plate thickness change. As expected,
their results for mass transfer were analogues to the results of heat transfer experiments
performed by other researchers. Kroger and Krizek developed the following correlation

for the average Sherwood number under a SSN impinging air onto naphthalene plates:

Sh = K Re‘l’:77 Sc”3
eqn. 2.9

where K = f(w)

ll

6,040<Rej<37,800

0.25<H/w<40

0.05<f<0.8
However, they give no indication as to the specific nature of K = f(w) other than to state
that K is a function of slot width.

In all the previous research presented above, each study yielded similar results for
the effects of jet exit velocity (i.e., each of the correlations have a Reynolds number
power near 2/3). The correlations by Gardon and Akfirat (1966), and Saad et al. (1980)
have similar powers for the effects of H/w and f (approximately -0.3 and 0.35,
respectively). However, the effects of the fluid properties (e.g., Pr and Sc) are not
expressly stated in the Gardon and Akfirat (1966) or the Saad et al. (1980) models and
are considered part of the constant terms 0.66 and 0.14 respectively. The model
developed by Martin (1977) seems to be the most versatile, having been compared to a
wide variety of research data from both heat and mass transfer studies during
development.

2.1.4 Surface motion effects

An important parameter affecting impingement heat transfer in an industrial
application, but not considered in the literature reviewed above, is surface motion. Subba
Raju and Schlunder (1977) studied average heat transfer rates from an unconfined SSN to
a moving surface. Using an infrared thermometer to measure changes in belt
temperature, they concluded that the average Nusselt number increased sharply with belt

speed at first, then reached a maximum of 1.5 to 2.0 times those experienced by a

12

 

stationary surface. However, Subba Raju and Schlunder only looked at belt velocities an
order of magnitude higher than those utilized in the present study (e.g., O.15-5.5 m/s).

Van Heiningen (1982) introduced the concept of the surface motion parameter, or
Mvs. Van Heigingen developed and used a gold film, ultra fast response heat flux sensor
to measure average Nusselt number on a moving surface under SSN flow. Using a
surface motion parameter of 0.02<Mvs<0.86, he concluded that the average Nusselt
number slightly decreased as the surface motion increased.

Polat and Douglas (1990) studied ASN flow of air impinging onto a permeable
moving surface. Their study utilized three air jets, impinging onto a rotating isothermal
cylinder, in which various pressures could be created to cause surface throughflow. Polat
and Douglas (1990) confirmed the van Heigingen (1982) findings and developed the
following correlation for average impingement heat transfer to a moving surface under an

ASN without throughflow.

Nu = 0.094 Re‘}68 (1 + M” )*""9
eqn. 2.10

where M” = vSps/ujpj

8,100<Re,-<25,800
O<Mvs<0.38
f:0.2
H/w=5
Polat (1993) noted that the Polat and Douglas (1990) model would consistently
predict Nu and Sh values higher than those predicted by Martin (1977), due to the effects

of cross flow interference. However, because the Polat and Douglas (1990) correlation

l3

was developed for single f and H/w values, it is not applicable to other configurations,

including the present study.
Polat (1993) suggested that the factor (1+ M W )4)“ could be included as part of

any correlation for average Nusselt and Sherwood numbers, and specifically those by
Martin (1977) and Saad et al. (1980). Polat and Douglas (1990) concluded that, due to
the agreement between van Heiningen (1982) and their results, the results of Subba Raju
and Schlunder (1977) were erroneous.

In these previous studies of the effects of surface motion, the surface velocities
were an order of magnitude higher than those used in the present study. Therefore,
surface motion should have minimal effect on the overall heat and mass transfer
coefficients in the present system.

2.1.5 Numerical studies

In addition to the experimental studies conducted in the field of impingement heat
and mass transfer, several numerical investigations have been performed as well. Some
of the more notable studies are those done by Tzeng and Soong (1999), Seyedein et al.
(1995), and Behnia et al. (1998). An extensive review of numerical impingement heat
and mass transfer literature available before 1988 is summarized by Polat et a]. (1988).

Numerical studies of impingement heat transfer have shown excellent agreement
to measured data for laminar flow over a flat plate; however, the studies of turbulent flow
have yielded varying results. The various methods for accounting for eddy effects may
cause this variation in results (Polat et al., 1988). Therefore, numerical techniques were
not utilized in the present study due to this lack of consistency for turbulent flow

conditions.

14

2.2 Impingement in the Food Industry

Impingement systems are used in a range of food processes, from snack foods to
breads to meats. Henke (1984) reported that impingement to food products was initially
patented in 1975 as “Jet Sweep” technology (US. Patent # 4,154,861), because the
impinging jets would “sweep away” the stagnant boundary layer on the product. Henke
believed impingement flows held potential for the food industry, because a high degree of
process control could be achieved by changing the geometric parameters. In addition,
heat transfer coefficients 4-10 times higher than other gaseous heat transfer methods
could be achieved, thus reducing cooking times.

Henke (1984) also suggested that additional control of a process could be
achieved by designing variable geometric impingement parameters within an
impingement oven, so different heat transfer “zones” are created. This provides simple
control over cooking processes requiring various heat transfer rates, without the added
complexity of building separate ovens.

Midden (1995) reported on the use of impingement air baking for snack foods.
Fluidized bed systems, which are impingement flow systems, have been used for over 45
years in the production of ready-to-eat cereals. In these systems, grain based products
could be baked from 20% to 2% moisture content in less than one minute.

Ovadia and Walker (1998) reported that the usefulness of impingement flows in
baking was well known by 1987, and was well established in the industry by 1992.
Currently over 100,000 dry air impingement ovens are in use for baking of bread
products (Ovadia and Walker, 1998). They also stated that impingement flows create

heat transfer coefficients sufficiently high to make infrared and conduction heating

15

negligible. Many applications in the food industry also utilize porous product transfer
belts, and thus double impingement flows are possible (i.e., impingement flows are
created on both sides of the product).

Ovadia and Walker (1998) also reported on the use of impingement heat transfer
in combination with microwave or radio frequency heating for bread baking. During
impingement baking, a crust is quickly formed on the surface of the bread product. This
crust prevents moisture loss by creating a barrier to moisture migration, but also reduces
heat conduction rates from the surface to the center of the product. The use of
microwaves and radio frequency heating could allow for more uniform interior cooking
of bread products.

Also investigating bread baking, Wahlby et al. (2000) reported that because the
heat transfer coefficient is enhanced by impingement flows, the impinging gas
temperature within impingement ovens could be reduced up to 25°C, compared with
other oven flow techniques, without any loss of product quality. Thus, impingement
flows could allow for reduced energy requirements. They also studied the impact of
impingement cooking on both bread products and pork meat, reporting that both products
cooked faster and lost less weight during impingement heating than during non-
impingement heating.

Nitin and Karwe (2001) studied heat transfer from an array of round nozzles to
cookie-shaped objects. Using aluminum disks as model food products, Nitin and Karwe
found that the heat transfer correlations developed for a cookie shaped object under ARN
flow were similar to those developed for heat transfer to a flat plate, such as that

developed by Martin ( 1977) for ARN flow. They also found that the average heat

16

transfer coefficient varied little whether the product was directly under the jets or
between them. This finding suggests that correlations developed for flow over a flat plate
may be applicable to flow over discrete objects. The findings also suggest that predictive
models for commercial ovens may be able to assume even distribution of jets, even when
variations exist in the actual jet spacing. However, in the Nitin and Karwe (2001) study,
the model food products were placed on a solid surface, thus creating a flow
configuration similar to impingement over a flat plate.

No studies were found in which condensation heat transfer from impinging moist
air to food products was considered; however, Braud et a1. (2001) modeled evaporation
(i.e., the opposite mass transfer phenomenon from condensation) from corn tortilla chips
during drying under a single round impingement nozzle, using predicted mass transfer
coefficients. Braud et a1. (2001) calculated the properties of air using regressions
obtained from previously tabulated data, and predicted convective heat transfer
coefficients using the correlations of Martin (1977) for SRN flow. The convective mass
transfer coefficient was subsequently calculated from the relationship recommended by
Martin (1977):

Na Sh

Pr0.42 ' $6,042
eqn. 2.11

Martin’s correlation estimated the measured convective heat transfer coefficient in the
Brand et a1. study to within a 16% error. Braud et al. (2001) suggested that better
accuracy could have been achieved by accounting for changing diffusivity of both the

tortilla chip and the impinging gas. This suggests that the Martin (1977) correlation for

17

ASN flow may be the most applicable to the current system if accurate gas properties are

calculated as a function of temperature.

2.3 Condensation from Air/Water Vapor Mixtures

2.3.1 Overview

Extensive work has been done in the study of condensation from pure component
gases. The extensive use of steam for electric power generation has been the primary
driving force for these studies. Only a concise literature review of studies performed on
condensation from steam-air mixtures is presented here. Interested readers are directed to
Merte (1973), and Tanasawa (1991) for more extensive reviews of general condensation
heat transfer.
2.3.2 Vapor pressure driven

A study on the design of cooler condensers for mixtures of vapors with non-
condensing gases performed by Colbum and Hougen (1934) is the earliest known study
on condensation from mixed vapors. Colbum and Hougen introduced the concept of
condensation as a mass transfer operation driven by a partial water vapor pressure
gradient within the gas. They theorized that the gas near a cooled surface could have a
partial vapor pressure no greater than the saturation pressure at the surface temperature.
Water vapor would then diffuse from the higher partial water vapor pressure in the bulk
gas, dictated by the bulk gas dew point temperature, to the lower partial vapor pressure at
the surface, dictated by the surface temperature.

Sparrow et al. (1967) studied forced convection condensation with the presence of
non-condensable gases using the concept of interfacial resistance. This concept is

important in the study of steam turbines, because small quantities of air act as impurities

18

within the system. Any air in these systems collects near the condensation surface and
causes interfacial resistance to the normal pure steam heat transfer.

Sparrow et al. (1967) modeled condensation heat transfer from laminar boundary
layer flow to an isothermal plate, using mass transfer equations analogous to those for
heat transfer. They assumed that condensation mass transfer was driven by a partial
vapor pressure gradient across the non-condensable gas interface, defined by the vapor
pressure in the bulk gas and the fact that the surface temperature establishes a saturation
state for the vapor near the surface. This method accurately modeled the condensation
rate; however, this method requires the mass transfer coefficient to have inconvenient
units (s/m). The use of a concentration gradient, calculated from the vapor pressure
gradients, as a driving force allows for more standard units (m/s).

Sparrow et al. (1967) also assumed that all latent heat given up by water vapor
condensing on the plate surface was absorbed by the plate. This concept is important in
modeling forced convection/condensation heat transfer, when establishing energy
balances between the product in question and the gas.

Rose (1980) also investigated forced convection condensation in the presence of
non-condensable gases, but studied horizontal flow over a flat plate and horizontal tubes.
Rose also used the analogous relationship between heat and mass transfer to describe
condensation heat transfer as a mass transfer phenomena driven by a partial vapor
pressure gradient. Rose assumed that all heat transfer was due to condensation (i.e.,
convection heat transfer was negligible), and obtained correlations for mass transfer
coefficients by substituting the Sh and Se numbers for the Nu and Pr numbers in

correlations for heat transfer. As expected, the analogous nature between heat and mass

19

transfer allowed for this transformation; however, this method required the addition of
several terms to adequately account for dimensional continuity.
2.3.3 Dew point temperature driven

Peterson et a1. (1993) studied diffusion layer theory for turbulent vapor
condensation with non-condensable gases. While they acknowledged that condensation
from steam-air mixtures is a mass transfer phenomenon, they described the condensation
heat transfer in terms of a condensation heat transfer coefficient driven by a dew point
temperature gradient. This dew point gradient was defined as the difference between the
dew point of the bulk gas and the dew point of the gas in contact with the product
surface. This method assumes that the gas in contact with the product surface is saturated
due to water vapor being removed through condensation. Thus, the overall heat transfer

rate could be stated as:

qroral = h A(To.oo - Tm )+ hA(Tw — Ts)

C

eqn. 2.12

Using this method, a condensation heat conduction value (kc) can be defined in

terms of vapor molar fraction gradients, diffusion coefficients, and temperature data. The

 

h L
Sherwood number can then be restated as Sh = I: , where hc is the mass transfer

coefficient when using dew point temperature as a driving force. While this method
seems very straightforward, the calculation of hc and kC is actually fairly complicated. In
addition, the values of hc and kc must be recalculated for every change in vapor
concentration in the bulk gas or at the surface. This is impractical for non—isothermal

surface temperature problems.

20

2.3.4 Deficiencies

No studies were found on condensation from steam-air mixtures impinging onto a
discrete object at a temperature lower than the dew point temperature. Knowledge in this
area will gain in importance as the benefits of impingement flows, combined with steam-
air condensation heat transfer, are realized in the food processing industry. Therefore, a
reliable method for predicting heat transfer rates in these systems is vital to process

design, optimization, and control.

21

Chapter 3

3 Methods and Materials

3.1 Overview

The overall approach to this project consisted of comparing model-predicted
coefficients for convective/condensing heat transfer to measured values from tests in a
commercial-scale, moist-air impingement oven (Figure 3.1.1). Previously developed
correlations for impingement on flat plates were used to predict the heat and mass
transfer coefficients over discrete test objects (aluminum blocks) subjected to moist-air
impingement. Turbulent boundary layer theory was used to account for the heat and
mass transfer coefficients over the aluminum block edge surfaces. Average predicted
heat and mass transfer coefficients were calculated as weighted averages of the
impingement and edge coefficients. The predicted average heat and mass transfer

coefficients were then compared to values estimated from experimental data.

22

 

 

(AnalyticaD @xperimentzD

Impingement “DIV” “Wet”
Models

Tests Tests
Predict Predict Estimate stimate
himp & hm»\/imp hE & hmE havg hefl'
Predict
havg hnuvg

Estimate
Compare Compare

h
h... 11mg
Figure 3.1.1: Overall study design.

    

 

  
     

 

 

 

 

mavs

 

 

 

3.2 General Theory

Heat transfer to a food product in a moist air impingement oven can be divided
into two categories: heat transferred through the forced convection mechanism and heat
transferred through the release of latent heat as water vapor condenses on the cool
product surface. In a typical cooking process, condensation is the primary heat transfer
mechanism until the product surface temperature approaches the dew point temperature,
at which point condensation ceases and convection becomes the primary mode of heat
transfer.

In impingement flow conditions, convective heat and mass transfer coefficients

are sufficiently high to make radiant and conduction heat transfer negligible (Ovadia and

23

Walker, 1998). In this study, all measured heat transfer was attributed to either
convection or condensation.

The use of a convective heat transfer coefficient in conjunction with a temperature
differential between the bulk gas and the product surface is widely accepted to describe
convection heat transfer. The overall heat transfer rate to an object from convection can
be calculated by Newton’s law of cooling:

: havg AT (T00 _ Ts )

q (‘0'? Vt’C

eqn. 3.1

The average convective heat transfer coefficient (h) has been studied extensively for
conditions similar to those in the present study. Specifically, the correlations developed
by Gardon and Akfirat (1966), Martin (1977), and Saad et al. (1980) were tested in this
study. These correlations were selected because they are the most widely accepted
correlations in the literature, and the conditions tested in their studies most closely
resemble the flow conditions in the present study.

The condensation portion of the overall heat transfer can be described as the
diffusion of water molecules from the bulk gas to the impingement surface, driven by a
concentration differential across the boundary layer (Figure 3.2.1). If the product surface
temperature is below the dew point temperature of the bulk gas, then water vapor
condenses at the product surface, so that the concentration of water vapor molecules at
the product surface is equivalent to the saturation concentration at the product surface
temperature. Because the concentration of water vapor in the bulk gas is considered to be
constant, a concentration gradient is thereby created between the bulk gas and the gas at

the product surface. Water vapor then diffuses to the surface, where condensation occurs.

24

 

 

 

T; Cs

Figure 3.2.1: Boundary layer temperature and water vapor concentration
gradients.

The use of a concentration gradient driving force to describe the condensation rate
is analogous to the temperature gradient that drives convection heat transfer. Just as each
flow condition has a convective heat transfer coefficient to quantify the effects of flow
conditions on the heat transfer rates, a convective mass transfer coefficient can be used to
quantify the effects of flow conditions on the condensation rates. Therefore, correlations
developed for the heat transfer coefficient can be modified to yield correlations for the
mass transfer coefficient. Martin (1977) recommended the relationship;

Nu Sh

Pr0.42 " S 6042
eqn.3.2

for correlations developed for impingement flows from an array of slot nozzles
configurations.
The rate of heat transferred to the product surface through water vapor

condensation can be calculated by:

25

qcond zmhfg =h A h (C00 _Cs)

m.avg T fg
eqn. 3.3
and the total heat transferred to a cool product as it passes through a moist air
impingement oven is calculated by:
qT = qconvec + qc‘ond : havg AT (T00 _ T: )+ hm,avg AT hfg (Coo - Cs )
eqn. 3.4

In order to test this hypothesis, aluminum disks (0.127 m in diameter and 0.025 m
thick) were used as model food products in this study. A total of six (6) identical
aluminum blocks were used. Because aluminum is very thermally conductive

(kA12024~165 W/mK), a Biot number (Bi = “075-12) less than 0.1 could be achieved for

Al
surface heat transfer coefficients less than 1,320 W/mzK. Biot numbers less than 0.1
allow for a lumped parameter solution to unsteady state convective heat transfer problems
with a solid temperature error less than 5% (Ozisik, 1993). The lumped parameter
solution allows for the assumption that, at any time, the block temperature is completely
uniform; therefore, the block temperature can be measured at any location within the

block.

3.2.1 Assumptions

The primary assumptions applied to this study were:
1. Radiant and conduction heat transfer are negligible compared to convective heat
transfer in impingement ovens.
2. All latent heat lost by condensing water vapor is absorbed by the model food

product.

26

. Only latent heat is transferred from the condensed water vapor to the product
surface.

. All water vapor condensed on the product is immediately blown off the product
by the impinging gas jets.

. Slot jets are evenly spaced within the impingement oven.

. The concentration of water vapor at the product surface is dictated by the product

surface temperature.

. Correlations for heat transfer coefficients can be modified to calculate mass

Sh

transfer coefficients by the relationship: P 0 42 = S 042
r ' c '

 

. Heat transfer to the product edges can be estimated by the equations for turbulent,

boundary layer flow.

3.3 Model Development

The correlations developed by Garden and Akfirat (1966), Martin (1977), and

Saad et al. (1980) for average himp values under arrays of impinging slot nozzles (ASN)

(Table 3.1) were used in the prediction of convective heat and mass transfer coefficients.

A computer model was written (via spreadsheet) to calculate the expected havg and hm,avg

values from these correlations, and to calculate the subsequent block temperatures.

27

Table 3.1: Previously developed heat and mass transfer correlations for ASN flow

to a flat plate.

 

Reference

Correlation and Conditions

 

Martin (1977)

 

Nu _ Sh —2/t3’4 2Re, 2”,
Pr“2 _ Sow '3 f/,1+,1/ f ’ Where

,1 = (60 + 4(H / w - 2)2)’”2
1,500<Re,-<40,000; 0.008<f<2.5x(H/S); 2<le<80

 

Saad et al. (1980)

Nu = 0.14 Re?775 (H / w)”286 f 0'3'"; where
0.0 156<1<0.0833, 4<H/w<24, 3,000<Rej<30,000

 

Garden and Akfirat
( 1966)

 

 

Nu = 0.66 Re 3:62 (H / w)*’"‘ f “-38 ; where
0.0156<f<0.0625; H/w>7; 7,000<Re,-<120,000

 

Because the edge surface area (A5) of the aluminum blocks constituted 29% of

the total block surface area (AT), an edge effect term was included to account for heat

transferred across this surface. As the blocks passed through the impingement oven, the

jets impinged upon the horizontal block surfaces, but also caused parallel flow past the

block edges (Figure 3.3.1). The Nu and Sh values experienced by the edges were

calculated by the equations for turbulent, boundary layer flow (Bejan, 1991). The edge

Reynolds numbers were calculated based on the jet exit velocities and the block height.

In all cases, ReE>10,000; therefore, turbulent flow was present. The equations for

turbulent, boundary layer flow are given by:

28

 

Nu ShE
FYI/E; = Sci/3 =0.037Re:/5

 

eqn. 3.5

where the characteristic length for describing Nu, Sh, and Re is the disk height (E). The
overall heat transferred to the aluminum block over a small time increment (dt) was

computed by:

+ hmEAE 1C0, - C, )dt
eqn. 3.6

q: mc dT= [h A+hWAIT —T )dt+hfg[h

imp s m.imp A:

g 1. 4 Finger

‘I IN '1’4—SIot

 

 

>4
_ _ +L¥lt <TestProduct A} x ”I,”
l 4' impingemertllowsuface x 456,
Jr

parallelflowsuriaces

 

Figure 3.3.1: Side view of impingement flows over a discrete object in a JSO-IV
oven.

In each convection/condensing model, the geometric oven parameters, bulk gas
temperature, bulk gas moisture by volume, and flow parameters were entered for each
test condition. The initial block temperature was assumed to be 20°C, in order to
coincide with the block temperature at which data analysis was initialized.

All gas properties were calculated at the instantaneous experimental film

temperature:

29

T +T
Tfilm=(°° 5%

eqn. 3.7

Using an assumed initial block temperature, the gas properties were calculated as a
function of temperature. The Nu and Sh values were then calculated using each
impingement model in conjunction with the edge effect term. The Nu and Sh values

were then used to calculate the convective heat and mass transfer coefficients by the

 

definitions:
Nu = _h_r_v_
kmix
eqn. 3.8
and
Sh = hmw
D
eqn. 3.9

where D is the diffusion coefficient of water vapor in air (mZ/s).

The gas properties, temperature and water vapor concentration gradients, and
convective transfer coefficients were then assumed to be constant over a one (1) second
time interval. The total heat transferred to the block over that time interval, and
subsequent change in block temperature, were calculated using eqn. 3.6. The new block
temperature was then:

T

i+l

=Z+Afl
eqn. 3.10

The gas properties, temperature and water vapor concentration gradients, and convective

transfer coefficients were then recalculated, and the heat transferred to the block over the

30

next time interval was computed. This process was repeated to predict a total process
time of 120 s.

The contribution of condensation heat transfer was assumed to be zero after the
block reached the calculated dew point temperature. If this assumption were not made,
the model would predict evaporative cooling. Because the condensed water was assumed
to be immediately blown off the blocks by the impinging jets, evaporative cooling was
excluded from the predictive model. However, evaporation from a product with moisture
on the surface could theoretically be modeled as a mass transfer operation identical to
condensation, but in the reverse direction, if the water vapor concentration at the surface
could be determined by some other means (e. g., desorption isotherms).

The computed heat and mass transfer coefficients at any time step (i) are a direct
function of the gas properties, because the properties are used in the calculation of Nu,
Sh, Pr, Sc, and Re. The gas properties are computed as functions of the transient gas film
temperature, which results in small transient variations in the convective heat transfer
coefficients, even for constant oven conditions. However, even over large block
temperature changes, the variation in h and hm over all the time steps was approximately
3% for any given set of test conditions. Therefore, the predicted heat and mass transfer
coefficients using each model are reported as means of the incremental heat and mass

transfer coefficients over the 120 predicted time steps.

3.4 Calculation of Gas Properties

The calculation of therrnophysical properties of the impinging gas is important for
the prediction of heat and mass transfer coefficients using previously developed

correlations, because standard dimensionless numbers such as Nu, Sh, Pr, Sc, and Re are

31

 

(A

[ti

1.1:
m

hig

00f,

it her

lilal j

dependent on these properties. The viscosity (u), density (p), thermal conductivity (k),
mass diffusivity (D), and latent heat of condensation (hfg) were calculated as functions of

gas composition and temperature using various methods.

3.4.1 Properties of dry air

The National Bureau of Standards (NBS) Circular 564, provided equations and
tabulated data for the properties of dry air (NBS 564, 1955). All properties are a function
of dry bulb temperature (K). The viscosity of dry air (kg/mos) can be calculated by (NBS
564,1955y

1.5
_ 145.8T x104,

y“ — T+110.4
eqn. 3.11

Linear, polynomial, logarithmic, exponential, and power regressions were
performed on the tabulated data for the other gas properties. The equation with the
highest R2 for each property was selected. Tabulated data from NBS Circular 564 for the
conditions 373<T<560 K and PT=1 atm were used to develop the following equations for

k, p, and cp:

k, = (—4xlOl’°T2 + 0.0099T — 0.0308)x10'2

eqn. 3.12
pa = 368.77T"'°°7"
eqn. 3.13
cm, = 5 ><10’7T2 — 0.00037" + 1.0411
eqn. 3.14

where k is in W/m K, p in kg/m3, and cp in kJ/kg K. The R2 values from the regressions

that yielded eqn. 3.12, eqn. 3.13, and eqn. 3.14 were >0.99, >0.99, and 0.97, respectively.

32

3.4.2 Properties of steam

The National Bureau of Standards (NBS) Circular 564 also provided equations
and tabulated data for the properties of superheated steam. The viscosity of pure steam

(kg/m-s) can be calculated by (NBS 564, 1955):

11,, = (0.361T —10.2)x10'7
eqn. 3.15

Linear, polynomial, logarithmic, exponential, and power regressions were
performed on the tabulated data for the other gas properties. The equation with the
highest R2 for each property was selected. Tabulated data from NBS Circular 564 for the
conditions 373<T<560 K and PT=1 atm were used to develop the following equations for

k, p, and cp:

k” = [0.008T — 0.5863]x10’2

eqn. 3.16
pH = -0.0011T + 0.9818
eqn. 3.17
c,” = 7 x10”T2 — 0.0068T + 3.6052
eqn. 3.18

where k is in W/mK, p in kg/m3, and cp in kJ/kgK. The R2 values from the regressions

that yielded eqn. 3.16, eqn. 3.17, and eqn. 3.18 were >0.99, 0.99, and 0.91, respectively.
The latent heat released by condensing water vapor, over the temperature range

273<T<373 K, was taken from tabulated data in Irvine and Hartnett (1976). A linear

regression of the tabulated data (R2>0.99) yielded:

33

his = 3167.2 — 2.4341T
eqn. 3.19

where hfg is in kJ/kg.

3.4.3 Steam-air mixtures

Properties of pure air and steam were calculated using eqn. 3.11-eqn. 3.18. The
viscosity and thermal conductivity of steam-air mixtures were calculated using the

method recommended by Burmeister (1983). Viscosity was calculated using:

 

 

 

 

 

H... = 11. + “”
1+¢.H["%a] 1+¢H.(’%H]
eqn. 3.20
where:
0.5 0.25 2
1+ ‘1/ WV
¢ 1 ”H Wa
aH J8- 1 + Wa
WH
eqn. 3.21
and thermal conductivity was calculated using:
km = k“ + k”
x
”MW W A)
eqn. 3.22

where:

34

2

0.75 0-5 0'5
wafl=l ”iii—1&1] ”Ba/T] X(1+0.733(19,,B,,) h]

1+BH/T 1+Ba/T

 

eqn. 3.23

and B = the Sutherland constant, estimated by B =1.5TB, where TB is the boiling point of

each component at 1 atm. (79 K for air, and 373 K for water). W is the molecular weight
of each component, taken as 28.97 and 18.02 g/mol, for dry air and steam, respectively.
The values of $113 and mm were calculated by the transposition of values in eqn. 3.21 and
eqn. 3.23.

An important factor in eqn. 3.20-eqn. 3.23 is the mole fractions (x) of air and
steam in the impinging gas. For this study, an air-steam mixture was estimated to be an
ideal gas mixture. In an ideal gas mixture, the gas fraction by volume is equal to the
vapor pressure fraction, which is also equal to the mole fraction. Therefore, the molar

fractions for steam and air for all steam-air mixtures were calculated by:

P
x” :MV:F"—
T

eqn. 3.24

and:

xa = 1— x H
eqn. 3.25

The value of the diffusion coefficient of water vapor in air is important to the
calculation of condensation rates. Sparrow et a1. (1967) and Denny et al. (1971) used the
formulation for the diffusion coefficient by Mason and Monchik (1962); however, Mason

and Monchik gave predicted versus actual diffusion coefficients only in the temperature

35

range 273<T<373 K. The diffusion coefficient for this study was calculated using an

equation from Vargaflik et al. (1966), over the range of 273<T<550 K:

D = 2.16x10”(T/273)"8
eqn. 3.26

where D is in mz/s.
In accordance with ideal gas theory, the density of the steam-mixture was
estimated as the sum of the products of the molar fraction of each component and the

density of the pure component, or:

pm : paxa +pH'xH
eqn. 3.27

The heat capacity of a steam-air mixture was calculated using the method of
mixtures. The heat absorbed by the mixture must equal the sum of the heat absorbed by

the steam and air, or:

mmcpm AT = macpdAT + chpH AT
eqn. 3.28

The mass of the mixture, or each component can be calculated by:

m1 : xrpr

eqn. 3.29

Substituting eqn. 3.29 into eqn. 3.28 for the mixture, air, and steam, and assuming that
temperature change is uniform throughout the gas, eqn. 3.28 yields the following
equation for the heat capacity of a steam-air mixture as a function of the heat capacities

of the pure components and the molar fractions:

36

_ paxacpa + pHxHCpH

paxa + [3qu

cpm

 

eqn. 3.30
3.4.4 Calculation of water vapor concentration

To predict water vapor concentration gradients between the bulk impingement gas
and the gas at the product surface, the saturation vapor pressure ratio (i.e., the saturation
vapor pressure divided by atmospheric pressure), as a function of temperature, was taken
from tabulated data in the CRC Handbook of Chemistry and Physics, (1989). A third

order, polynomial regression was then performed (R2>0.99) to yield:

1%, = x, = 1.6x 10'6 (T -323.5)3 + 1.523x10‘4 (T — 323.5)2 + 0.0061221T — 1.861285
T
eqn. 3.31

in the temperature range of 273<Ts<373 K. The concentration of water vapor at the

product surface in kg/m3 was then calculated by:

Cs = xH, pH 5
eqn. 3.32

where xH and pH were evaluated using eqn. 3.31 and eqn. 3.17 at the surface temperature.
The dew point temperature was also calculated via a power regression (R2>0.99)
from the same tabulated data (temperature range 273<Ts<373 K) for the saturation vapor

pressure by:

TD = 396.5M,°~°6"‘
eqn. 3.33

The bulk impingement gas water vapor concentration was calculated by:

37

C00 = xH,oopH,oo
eqn. 3.34

where X” and pH are evaluated at the bulk gas temperature and Mv conditions. The molar

fraction of steam (xH) in the bulk gas was established as a set point in oven operation, and

the density of steam was calculated using eqn. 3.17 at the bulk gas dry bulb temperature.

3.5 Calculation of Aluminum Properties

Aluminum disks (alloy 2024) 0.127 min diameter and 0.025 m thick (Catalog No.
9034K26, McMaster-Carr, Cleveland, OH) were used as model food products. A
predictive equation for the heat capacity of aluminum alloy 2024 was required to
accurately calculate the heat transferred to the aluminum blocks as a function of
temperature change.

The composition of aluminum alloy 2024 by percent mass was taken from Perry’s
Chemical Engineering Handbook, 7’h ed. (Perry et al.,1997) (Table 3.2).

Table 3.2: Composition of aluminum alloy 2024.
Composition of Aluminum Alloy 2024(%)

Cr Cu Mg Mn Si A1
0.1 4.3 1.5 0.6 0.5 93

 

 

 

 

 

 

 

 

 

 

 

The heat capacity of aluminum alloy 2024 was calculated as the sum of the heat

capacities of the individual components weighted by the mass fractions, or:

cl’zou = 2 xicl’r

eqn. 3.35

The heat capacities of the individual components as a function of temperature were taken

from Perry et al. (1997) (Table 3.3).

38

Table 3.3: Heat capacities of aluminum alloy 2024 individual components.

 

 

 

 

 

 

 

 

 

 

 

 

Substance eqn. for cfical/dmnol) T range (K) Uncertainty
Al 4.8+0.00322T 273-931 1%
Cr 4.84+0.00295T 273-1823 5%
Cu 5.44+0.001462T 273-1357 1%
Mg 6.2+0.00133T+67800/'1‘2 273-923 1%
Mn 3.76+0.00747T 273-1 108 5%
Si 5.74+0.000617T+101000/T2 273-1174 2%

 

The molar weights of the individual elements in aluminum alloy 2024 are given in Table

3.4. The effective molecular weight of aluminum alloy 2024 was determined to be 28.71

g/mol by calculating the weighted average (mass basis) of the component molar weights.
Table 3.4: molecular weights of selected elements (g/mol).

Cr Cu Mg Mn Si Al
52.00 63.55 24.31 54.93 28.09 26.98

 

 

 

 

 

 

 

 

 

 

Substituting the equations from Table 3.3 with the component fractions from
Table 3.2 into eqn. 3.35, and converting units, yields the following equation for the heat
capacity of aluminum alloy 2024 as a function of temperature:

c 2. = 0.14583(4.847 + 3.1287 .10-3 .T + 60671-2)

P20

eqn. 3.36

where cp is in kJ/kg K and T is in K. When eqn. 3.36 was evaluated over the temperature
range 273<T<673 K, a more simple equation was developed by performing a linear
regression (R2>O.99) on the outputs produced by eqn. 3.36. The heat capacity of

aluminum alloy 2024 was calculated for use in the models by:

39

c = 0.0004(T — 273)+ 0.8401

P2024

eqn. 3.37

The thermal conductivity of the aluminum blocks was only important to the
calculation of the Biot number to verify the use of the lumped parameter solution.
Therefore, a detailed equation for k2024 as a function of temperature was not required.
Ozisik (1993) lists the thermal conductivity of Duralumin (94-96% Al, 3-5% Cu, and

trace Mg) as 164 W/mK at 20°C. The density of Duralumin is also listed to be 2,787

kg/m3 at 20°C.
3.6 Experimental Design

3.6.1 Test conditions

A JSO-IV moist air impingement oven (see section 3.6.2) was used to establish 45
test conditions in a full factorial design (Table 3.5).

Table 3.5: Experimental test conditions.

 

 

T,. uj Mv f H/w vs
(°C) (m/S) (%) (mm/S)
121 11.2 0 0.0683 10 22<v<58
177 16.5 30
232 22 50
70
90

 

 

 

 

 

 

 

 

The geometric oven parameters were held constant, given that they essentially
were fixed by the oven configuration; therefore, f and l-I/w were the same for all
conditions tested. The product horizontal surface velocity was varied between 22 and 58
mm/s by changing the product belt speed. Belt speeds were slower for lower heat

transfer rate conditions (Mv~0%) to ensure adequate data from the tests. During higher

40

heat transfer rate runs (i.e., those with condensation), the belt speed was increased,

because less time was needed for large increases in block temperature.

The impingement gas temperature (T...) was established as a set point using the

oven control system. The impingement gas slot exit velocity (Uj) was established as a

function of % fan speed. The oven manufacturer staff provided slot exit velocity data for

the oven (Figure 3.6.1), which were calculated from velocity pressure measurements

assuming incompressible flow.

 

 

 

 

 

 

 

25
- i I __ ___J -‘t E
20 X
:1
.5 a "
1? . E 1 A _ ____ _. _. _ __ ___
2.. 5, 9
510 ~ 4 — — v ~-————— -= —
R ‘. El discharge (top) i
l
R (X discharge (bottom) I
5 ., I, q ,_ __ , err—_——-—,—;-EL t4::-‘_-____
3
0 , Y . . .
0 20 40 60 80 100

Fan Setting, % of max.

Figure 3.6.1: Stein J SO-IV oven exit velocities.

Linear regressions were performed on the above data to develop separate

120

equations for the top and bottom exit velocities (um, and uBouom) as functions of fan

setting.

41

um, = 0.220F + 0.446
eqn. 3.38

= 0.210F + 0.318

u Bottom

eqn. 3.39

The exit velocity used in the model was assumed to be constant throughout the oven,
from both top and bottom plenums; therefore a predictive equation was calculated as the

average of eqn. 3.38 and eqn. 3.39. Exit velocity (m/s) was thereby calculated as:

 

u = "W +2"”""""' = 0.215F + 0.382
eqn. 3.40

Moisture by volume of the impinging gas was established as an operating set
point for each test. The oven adjusted this parameter by controlling the steam flow into
the plenum. Impingement gas moisture content was measured using an industrial
Humitrol 11 (Machine Application Corporation, Sandusky, OH, model: MAC 120-100-F,
serial No. 0989). The sensor in the Humitrol II measures capacitive response to vapor
pressure changes. The vapor pressure is then converted into moisture by volume by
dividing by the total pressure. The Humitrol 11 accuracy was reported by the

manufacturer as il% full scale from 0 to 100% My.

3.6.2 The oven

In cooperation with Stein-DSI (a business of FMC FoodTech, Sandusky, OH), a
Stein-DSI JSO-IV moist air impingement oven was used to establish the desired test
conditions. This oven allowed for the control of impingement exit velocities (via fan

speed), horizontal product velocity (via belt speed), impingement height, and moisture

42

content of the impinging gas (volumetric basis). The open slot ratio (f) was constant
through all test runs.

The JSO-IV oven continuously transported product through an impingement
cooking chamber (6.7 m long x 1 m wide). An air—steam mixture was impinged onto the
product from both top and bottom in a flow configuration known as dual-impingement.
A centrifugal fan established slightly pressurized conditions in chambers above and

below the product belt known as plenums (Figure 3.6.2).

JSO-N Airflow

Fan draws heated air
thru heat exchanger

Air is forced by
distribution damper

Heated air/steam enters
top and bottom fingers

Airflow contacts
product from
above and below
product conveyor

vvvv

   

Air returns to
heat exchanger

Figure 3.6.2: Impingement gas flow inside a J SO-IV oven (provided by Stein-DSI,
Sandusky, OH).

Stainless steel fingers (15.24 cm wide) protruded from the plenums over the
product belt. These impingement fingers were positioned perpendicular to the belt

movement direction, so product passed under the slot jets sequentially. Three (3) slot jets

43

(6.35 mm wide) were built into each impingement finger to allow gas to exit the plenums.
The slots were placed evenly across each finger on 5 cm center-to-center spacing. The
impingement fingers were positioned with 15 cm between each finger; however, the
fingers on top were positioned alternating with those from the bottom plenum (Figure
3.6.3 and Figure 3.6.4).

This finger placement results in unevenly spaced slot jets above and below the
product surface. The models developed by Gardon and Akfirat (1966), Martin (1977),
and Saad et al. (1980) assume evenly spaced ASN flow over a flat plate. Therefore, the
slot jets in the J SO-IV oven were assumed to be evenly spaced by calculating the fraction
of nozzle open area (f) as the total slot open area divided by the total belt area. The
findings of Nitin and Karwe (2001), for impingement flow from an array of round
nozzles to discrete model food products, supports the validity of this assumption.

The surface of the bottom fingers were 6.35 cm from the belt surface, while the
surface of the top fingers were 8.89 cm from the belt surface. Therefore, when the model
food product (E=2.54 cm) passed through the oven, the top and bottom impingement jets

were 6.35 cm from the product top and bottom surface, respectively.

’ ill
. ,3
l m

,_

f'

\l
(I?!
a
t.
1

I. ‘ ‘
.7 .. gun.
I!
r " .. a
w .

‘ | 11 ' r ,"

 

Figure 3.6.3: Interior view of top impingement slot array (fingers) in a J SO-IV
oven.

45

 

Figure 3.6.4: Interior viewa bottom impingement slot array (fingers) in a J SO-IV
oven.

3. 7 Test Protocols

3.7.1 Aluminum block preparation

Aluminum disks (0.127 m in diameter and 0.025 m thick) were used as model
food products. A total of six (6) identical aluminum blocks were used. The conditions
for a lumped parameter solution were met in this study; therefore, the block temperature
could be measured at any location within the block.

The block geometric center was chosen for temperature measurement. A 2.4 mm
hole was drilled from the edge to the center of each block. A K-type thermocouple was
then inserted into each block (Figure 3.7.1). Each thermocouple was coated in highly
conductive Omegatherm paste (Omega Engineering, Stamford, CT) to ensure good

thermal contact between the block and thermocouple. Once the thermocouples were

46

inserted into each block, the entrance point was sealed with epoxy to hold the

thermocouples in place, and prevent impingement gas from entering the hole.

    

Figure 3.7.1: Aluminum blocks on the carrying tray before a test.
Each block was chilled in an ice bath for more than 3 min between runs to insure
an initial block temperature near 0°C. Each block was patted dry with paper towels upon

removal from the water bath to minimize thermal interference from water adhering to the

blocks.

3.7.2 Temperature measurement

Thermocouple data were collected using a six channel Datapaq 9000 (Model:
DP9064, Datapaq, Inc., Wilmington, MA) data logger. The Datapaq 9000 can measure
K-type thermocouples in the range 0<T<1,370°C to a i 1°C accuracy.

The data logger was shielded from the high oven temperatures using a Datapaq
Thermal Barrier (Model TB0049) with heat sink (Model TB9027). This thermal barrier
was stainless steel with ceramic insulation capable of withstanding 2.8 h at 150°C or 1.8
hours at 205°C. Thermal barrier dimensions were 6.6 cm (height), 18.8 cm (width), and

38.4 cm (length). The temperature of the data logger was observed to increase minimally

47

(less than 5°C) over all test runs; thus temperature dependent errors in temperature

measurement were not considered.

3.7.3 Oven runs

Before any data were collected for each condition, the oven was allowed to reach
steady-state at the set points. Slot exit velocity was easy to change by adjusting the % fan
speed, and little time (less than 1 min) was required for the oven to reach steady-state
when fan speed was changed. The impingement gas moisture by volume could typically
be changed in less than 10 min, while the dry bulb temperature typically took more than
20 min to adjust.

The following procedures were used during data collection, once the oven
conditions had reached the desired set points:

1. Three (3) aluminum disks were placed in an ice water bath for at least three
minutes to insure an initial temperature near 0°C.

2. The aluminum blocks were removed from the ice water bath and patted dry with
paper towels to remove excess water.

3. The thermocouple inserted in each block, and an exposed thermocouple, were
connected to the Datapaq 9000 data logger.

4. The three blocks were placed in a row on a stainless steel carrying tray (Figure

3.7.1)

5. The carrying tray was placed on the center of the product belt as close to the start

of the impingement zone as possible.

48

6. Once the product belt had transported the blocks through the impingement zone
and exited the oven, the carrying tray was removed from the product transport
belt.

7. The thermocouples were disconnected from the data logger, and the aluminum
blocks were placed in the ice water bath to prepare for the next test.

8. Data were downloaded from the data logger.

9. The procedures (steps 1-8) were repeated for each test condition.

3.8 Data Analysis

The heat and mass transfer coefficients were analyzed in terms of the overall
average coefficients, which were the weighted averages of both the top and bottom
surfaces as well as the edges. The local values of h and hm experienced on the top,
bottom, and edges are different due to varying surface flows. However, because of the
lumped parameter conditions, only average heat and mass transfer coefficients could be

determined from the experimental data.

3.8.1 Estimation of havg from raw data

To model the total heat transfer experienced by a model food product as it passed
through a moist air impingement oven, the convection portion of the overall heat transfer
had to be isolated, so that the contribution of condensation could be calculated separately.
This was accomplished by including oven conditions with as close to 0% moisture by
volume as possible.

A perfectly dry environment could not be achieved within the industrial J SO-IV

oven; for the MV~0% runs, the actual impingement gas moisture range was 1%<M,,<13%.

49

This resulted in some condensation heat transfer during these runs. Therefore, havg values
were calculated using only data collected after the aluminum block temperature exceeded
the dew point of the bulk impingement gas.

The heat balance on a solid object undergoing convective heat transfer for a small
time interval is given by:

h A, (To, -T_, )dt = mcpdT

avg

eqn. 3.41

or, the heat transferred from the impinging gas to the object must equal the change in
internal energy of the object.

Assuming havg, AT, m, and cp are constant, eqn. 3.41 can be solved by separation

ln T-T, = hmA/ .t
70.]; mcp

The value of ln[(T-T...)/(To-T..)] is easily calculated from the block temperature data

of variables to give:

 

eqn. 3.42

collected. A linear regression can then be performed on these data versus time to yield

the slope, or d{ln[(T-T...)/(To-T..)] }/dt. The average convective heat transfer coefficient

T—T,
_ —d[ln[ To —T,, D mcp

“"8 dt A,

can then be calculated by:

eqn. 3.43
3.8.2 Estimation of hmvg from raw data

When water vapor is present in the impinging gas, and the model food product

temperature is below the impinging gas dew point temperature, total heat transfer is a

50

function of both convection and condensation. The heat balance on a solid object
undergoing convection and condensation heat transfer for a small time interval is given
by:

km, A, (T,, — 7, )dt + hm, 14,72, (C, - C, )d: = mcpdT
eqn. 3.44

or, the heat transferred from the impinging gas to the object by convection, plus the heat
transferred to the object through the release of latent heat as water vapor condenses on
the surface, equals the change in internal energy of the object.

Eqn. 3.44 can also be written in terms of an overall effective convective heat
transfer coefficient, or:

12,, A, (T, — T, )dt = mcpdT
eqn. 3.45

This effective convective heat transfer coefficient can be estimated from the model food
product temperature data collected by calculating finite block temperature changes over a
finite time. Time steps of two (2) seconds were chosen, due to temperature measurement
resolution. Given that the measurement resolution was 05°C, and a typical temperature
change over a 1 s interval was as low as 1°C, analysis at one (1) second intervals yielded
dramatically different he“ values from time step to time step. Therefore, the use of two
second intervals yielded more consistent hen values. The hen was calculated for runs

while condensation was occurring, by:

51

mcp (T,2 - T.)

l l

h :
efi', AT(Too _Ti'xtnz _ti)

 

eqn. 3.46

Because eqn. 3.44 equals eqn. 3.45, the convective mass transfer coefficient at time i, can

be calculated by:

_ (heflj — havg xToo — Ts )

hm r'
‘ hfg (C:m - Cs.i)

 

eqn. 3.47

where h,vg is the convective heat transfer coefficient found from purely convective (i.e.,
dry) conditions with all other parameters being equal to the condensing case (eqn. 3.43).

Analysis of the raw data involving condensation heat transfer was begun at
T=20°C, to exclude data collected before the blocks traveled on the transport belt into the
impingement zone in the oven chamber. Analysis was ended at T=(TD-8°C), because for
(TD-8°C)<T<TD, the heat transfer to the product is in a condensing-convective transition.
Over this temperature range, the condensation heat transfer is no longer the dominant
heat transfer mechanism, because as the block temperature approaches TD, convection
heat transfer becomes dominant. Over this transition region, the calculation of hm

becomes unstable.

52

Chapter 4

4 Results and Discussion

4.1 Prediction of hm

4.1.1 Data analysis

A perfectly dry environment could not be achieved within the industrial ISO-IV
oven. For the Mv~0% runs, the measured impingement gas moisture range was measured
between 1 and 13% (Table 4.1).

Table 4.1: Measured values of the impinging gas moisture content by volume, for
Mv~0% runs.

 

 

 

 

 

 

 

 

 

 

Dry Bulb Temp (°C) Fan Speed (% full) Measured Mv (%)
121 50 1
121 75 l
121 100 2
177 50 5
177 75 5
177 100 5
232 50 13
232 75 13
232 100 13

 

 

 

 

 

The increasing impingement gas moisture contents with increasing temperature
can be attributed to a water seal around the edges of the oven, which contains the
impingement gas. As the impingement gas temperature was increased, some of the water
from this seal was evaporated, thus increasing MV within the oven (Table 4.1), even when
no steam is being directly added to the system.

The effects of condensation are evident in graphs of block temperature versus

time when significant water vapor is present (Mv>l%) as compared to those when water

53

vapor is minimal (Mv=1%) (Figure 4.1.1). Naturally, the difference in bulk gas dry bulb
temperature accounts for much of the difference in the rise of the block temperatures over
the course of the runs; however, the effects of the increased condensation are evident in
the initial 30 seconds.

Data collected before the block temperature reached the calculated bulk gas dew
point temperature were excluded from analysis. In the case of the data presented in
Figure 4.1.1, the dew point temperatures were calculated using equation 3.33 as 5.5, 34.5,
and 530°C for Mv = 1, 5, 13%, respectively. For T = 232°C and Mv = 13%, the block
temperature rises more quickly to 53°C, then the rate of temperature change decreases

when only pure convection heat transfer occurred, as expected.

 

 

 

    

 

 

   

 

  

 

 

 

80

70 __1_w. "1 __ ’ —‘““ ‘‘‘‘‘ Y #7 w A_-.. "j;;:7"'wxi
A Pounts where block 232 C, MV=13°/o/x—/’
o , /
2.. 60 A emerged -lhe_d_9fli?°|fli temperatures ,- g N/
3 v 177 C, Mv=5°/o ’
g 50 I 44 ~ A k 4’ __ /
é ’ 121 c, MEm~-~ ‘-
0 40 I ifli “#us70 _ ' // * —r—::;::r—="* '-
i- _ _ , 3.»
§30 “‘ _ 7"“ f:::': T- — 4 ——4
E ,, - E

20 » —~— ~— — #—

10 . . . , ,

0 10 20 30 40 50 60 70
time (s)

Figure 4.1.1: Measured block temperatures for Mv~0%, F = 75%, runs (rep. 1
only).

Measured havg values were compared to those predicted using the models (see

Figure 3.1.1). The heat transfer rate on the product top and bottom surface was expected

54

to be higher than the heat transfer rate to the product edges, due to impingement flows on
the top and bottom surfaces versus parallel flows past the edges. However, because only
the block center temperature was measured, equation 3.41 will only estimate an average
convective heat transfer coefficient from the temperature data; therefore, the comparisons
of predicted versus experimental results are based solely on the average convective
coefficients over the total product surface.

The accuracy of the model predictions for the convective heat and mass transfer
coefficients were compared to those actually measured, in terms of the average absolute

error and the standard error of prediction (SEP), calculated by:

measured, — predicted, I)

 

1 (l
error%a,,g :2 — E
n measured,

eqn. 4.1

 

 

SEP _ J 2(measured, - predicted, )2

n - l
eqn. 4.2

The average absolute error is a measure of the % the predicted value given by a model
will be different from the actual value experienced by a food product in the oven. The
SEP is a measure of the predictive accuracy of a model for the given convective
coefficients. While the average % error has no units, the SEP has the units of the
quantity being measure, W/mzK and mm/s for h and hm, respectively.

The results for the correlations based on Garden and Akfirat (1966), Saad et al.
(1980), and Martin (1977) are presented in sections 4.1.3, 4.1.4, and 4.1.5, respectively.
A summary is presented in section 4.1.6. In addition, the effects of temperature, exit

velocity, and impingement gas moisture content on model errors were considered.

55

4.1.2 Evaluation of the lumped parameter assumption

For the test condition T..=121°C, Mv~0%, and F=50%, the aluminum block
surface temperature was measured, in addition to the center temperature, so the validity
of the lumped parameter solution could be assessed. The surface temperature was
measured by inserting a K-type thermocouple into a hole drilled 0.4 mm from the product
surface. Over the course of three replications, the difference between the surface
temperature and center temperature never exceeded a 2.7% difference. This difference

was well within the 5% allowed by the lumped parameter assumption (Ozisik, 1993).

4.1.3 Gardon and Akfirat (1966)

Measured havg values taken from the Mv~0% runs were compared to those
predicted using Gardon and Akfirat (1966) combined with edge effect consideration
(Figure 4.1.2). The predicted values had an average absolute error of 14.7% from the

measured values, with an SEP of 13.1 W/mzK.

 

 

 

 

 

 

 

140 ‘ .
E? 130 L— ~ r a—L— -W l — .‘— —
g 120 -- - i l l -1-- W1 _11 __
E 110 4-—— —-— if W, ‘ 7W *~:-———— 1 — —— -Je-—O——‘J 7F

1 1 .

51m 1 _W _ W W, , 8 l 1-,.W_ l WW
3 a) W W i 6-4W.-- 1 _W i . - ,W1.W
a m 4* W i W W l 1 1 __ .-.i._,_W , lWWW
a I | I l
g 70-.._ ‘ 1 l _ 1 _--,W

60 L ° 1 ’ L l l

60 70 H) 9) 1C!) 110 120 13) 140
Magma

Figure 4.1.2: Predicted vs. measured havg using the model based on Gardon and
Akfirat (1966).

56

Although the prediction using Garden and Akfirat (1966) does closely predict the
measured values of havg, the prediction using the model based on Gardon and Akfirat

(1966) was consistently higher than the measured values by an average of 13 W/mZK.

4.1.4 Saad et al. (1980)

Measured havg values taken from the Mv~0% runs were compared to those
predicted using the model based on Saad et al. (1980) combined with edge effect
consideration (Figure 4.1.3). The predicted values had an average absolute error of

12.2% from the measured values, with an SEP of 11.8 W/mzK.

_s

.h

o
1

 

88
l
(J
i
ll
H
i
l
*|
.11

co
0

”I I
1

Measured hm (W/m’K)
3 3
O 0
1 l
. i
l
1
. i ,
i , 1 .0 Q
l.

 

 

 

. 6*
70 4 0 4,— .— . — -— - -
60 l° . 4 4 ~ . l
60 70 80 90 100 1 10 120 130 140
Predicted h", (W/m’K)

Figure 4.1.3: Predicted vs. measured havg for Mv~0% runs using the model based on
Saad et al. (1980).

The predicted values of havg using the Saad et al. (1980) correlation were near
those measured; however, an analysis of the residuals shows a clear increase in error at
higher values of havg. As the fan speed, and subsequently the jet exit velocity, increases,
the errors between the model based on Saad et al. (1980) and the measured havg values
increase (Figure 4.1.4), which indicates that the model based on Saad et al. (1980) does

an unsatisfactory job of accounting for gas velocity effects.

57

 

 

 

 

0.0
-2.0-4444404444444444444424—4444
a; .4644 448444444 4444424444244
0
E-6.044444— 444444444 44444
§-80444§- 444444444 44444
33-1004 9446-544 E444444
$42.04 4 4 4 4 4 4 444444444 44944 4 4
0
J:5'44044 44420444 2“
46.0 444 4 44444 #4446444
-18.0
7.5 12.5 17.5 22.5

"I (mIS)

Figure 4.1.4: Residuals of havg vs. jet exit velocity using the model based on Saad et
al. (1980).

4.1.5 Martin (1977)

Measured havg values taken from the Mv~0% runs were compared to those
predicted using the model based on Martin (1977) combined with edge effect
consideration (Figure 4.1.5). The predicted values had an average absolute error of 1.9%

from the measured values, with an SEP of 1.9 W/mzK.

58

140 4, , , ,
130 4 4 -4 1 -4 5 4 ,, 4 i 4
120 WW .W W. W i -- - . W ., . _ _1WW - - ,
110 *5 m I i 1 '5’ " "r '- - ~* * - ——i*-"
100 4- 4 4 ; ; 4 4 4 ' 4 4- 4. 4 4 ‘4 4
90 + » ———— . — * ~ —‘~ — 4'« - — 4 «4 44 44 44 f -
30 W. W 1W W . - W , W 1 - . W - W - W:
70 W-WW_ W WW , _ WW 1 __ _ W : W 1.. W- -_ _.I__.._
60 ° 1 l l .

50 70 80 90 100 1 10 120 130 140
predicted hm (WlmzK)

 

measured hm (WIm’K)

 

 

 

Figure 4.1.5: Predicted vs. measured havg for Mv~0% runs using the model based on
Martin (1977).

Clearly, the model based on Martin (1977) most accurately predicted the
convection coefficients measured in the present study, as the SEP was only 1.9 W/mzK.
In addition, the average residual between the Martin (1977) prediction and the measured

values of havg was 0.3 W/mzK, indicating negligible bias between predicted and measured

values.

4.1.6 Summary

When the predicted versus measured values for the average convective heat
transfer coefficient were analyzed, the models based on Gardon and Akfirat (1966), Saad
et al. (1980), and Martin (1977) were accurate to average absolute errors of 14.7, 12.2,
and 1.9%, respectively. In addition, the SEP’s for the three models were 13.1, 11.8, and
1.9 W/mzK, respectively. Figure 4.1.6 shows the predicted havg values versus the

measured havg values from the MV~O% runs, for all three models.

59

 

140 , ,

_L
OJ
C

l

l

l

l

i

re

0

l

l

1W

1

l

l

l

l

l

,-

l

_n-dW.

"i

8
l

l
7

l

l

l

l

l

1 l

l

'.

l

l

l

l

l
_1,

l

i

<0
0
l
l
l
r
l
T
l
l
l
l
l
1
T
l
l
l
l

X , W W W W WW WW Wr-W _W W

Measured h", (W/mzK)
8
O
l
l
l
l
l
l

 

 

 

80 ,__ __7 _j __ _g * W1 _ _ +_1 oMartin(1977) l,

1 l ‘ nGardon and Akfirat(1966) l

70*” , 7 7 7 7‘1" 7, xSaad etal.(1980) ,7

60 0 1x a TMT'TT‘T"
60 70 80 90 100 1 10 120 130 140

 

Predicted it“, (W/m’K)

Figure 4.1.6: Predicted vs. measured havg values estimated from M" ~ 0% runs.

Clearly, the model based on Martin (1977) most accurately predicted the
convective heat transfer coefficient when used in conjunction with the assumption of
turbulent boundary layer flow over the edges. This result was expected, due to the
similarity in test conditions between the J SO-IV oven and Martin’s impingement
conditions with cross flow (Table 4.2). Spent impingement gas was forced to exit the
impingement zone laterally between the slot jets in the JSO-IV oven, because no exit
ports existed between the jets. This flow configuration was similar to those tested by

Martin (1977).

Table 4.2: Comparison of tested variables vs. range of validity of the models used.

 

 

 

 

 

 

 

 

 

 

Model f H/w Rej
Gardon and Akfirat (1966) 0.0156<f<0.0625 >8 >6,000
Saad et al. (1980) 0.0156<f<0.0833 8<le<24 3,000<R6j<30,000
Martin (1977) 0.008<f<0.1408 2<le<80 1,500<Rei<40,000
Present study 0.0683 10 2,500<Rej<6,000

 

60

The model based on Saad et al. (1980) contained some error in the effect of exit
velocity on havg values, as demonstrated in Figure 4.1.4. The flow condition tested in the
Saad et al. (1980) study also varied from that experienced in the present study. They
allowed spent impingement gas to exit through exhaust ports placed between the
impingement nozzles, thus eliminating the cross-flow effect present in Martin (1977), and
the present study.

The model based on Gardon and Akfirat (1966) predicted an havg value
consistently 13 W/mzK above those measured. While Gardon and Akfirat tested array of
slot nozzle (ASN) flow with cross-flow, they performed their study in an unconfined
space (i.e., no walls or barriers around the impingement zone to confine the impingement
gas). In addition, the slot open area used in the present study was outside the range of
validity tested for the Gardon and Akfirat (1966) model (Table 4.2); therefore, the
presence of a consistent error was not unexpected.

The Gardon and Akfirat (1966) and Saad et al. (1980) models for flow over a flat
plate do not explicitly state the effects of the fluid properties in terms of the Pr number.
These models were developed for air near room temperature (22°C), whereas the present
study utilized temperatures between 120°C and 235°C. Over the temperature range
22<T<235°C, the Pr number for dry air decreases by 4% (Geankoplis, 1993). When
these values are taken to the standard powers of 1/3 or 0.42 as found in the literature, the
effect of the temperature dependant Pr number is less than 1.5%. Therefore, the lack of
direct consideration of the temperature dependant Pr number in the Gardon and Akfirat
(1966) and Saad et al. (1980) models may have contributed to their consistent over-

prediction of havg, but only minimally.

61

4.2 Effects of Surface Motion

Polat (1993) incorporated an adjustment factor (eqn. 4.3) to account for the effects
of horizontal surface motion by a product under an array of impinging slot jets. They
suggested that all previously developed correlations for heat transfer to a flat plate under
an ASN flow configuration could be modified to predict heat transfer to a moving flat
plate by multiplying by eqn. 4.3.

(1+ M” )-—0.69

eqn. 4.3

The factor Me, is defined by:

MVS :psvs/pjuj

eqn. 4.4

where p is the density of the impingement gas.

During this study, three horizontal surface (i.e., belt) velocities were used: 0.058,
0.035, and 0.022 m/s. By eqn. 4.4, the effects of surface motion will be the most
prevalent when the ratio of horizontal velocity to jet exit velocity is the greatest. These
horizontal surface velocities corresponded to M,s values of 0.0064, 0.0043, and 0.0033,
respectively, at the lowest jet exit velocity tested (11.2 m/s). Substituting these values
into eqn. 4.3 predicted the approximate effect of surface motion to be a maximum of
0.5%. Therefore, the effects of surface motion, using the method recommended by Polat
(1993), were negligible for the conditions considered in this study.

Polat and Douglas (1990) performed studies with a surface motion parameter

between 0 and 0.35, and van Heiningen (1982) performed studies with a surface motion

62

parameter between 0.02 and 0.86. While the range used by Polat and Douglas
encompasses the surface motion parameters experienced in the JSO-IV oven, they did not
extensively study the effects of surface motion in the experienced range. The fact that the
surface motion parameters experienced during the present study had little effect on the
predicted heat and mass transfer coefficients was expected. Therefore, the factor
(1+M.,s)'0'69 was not included as an adjustment factor for surface motion, as
recommended by Polat (1993), in the models based on Martin (1977), Saad et al. (1980),

and Garden and Akfirat (1966).

4.3 Prediction of hm,avg

4.3.1 Data analysis

Only raw data for T>20°C were analyzed, to exclude data collected before the
blocks traveled on the transport belt into the impingement zone in the oven chamber.
Analysis was ended at T=(TD-8°C), because hm was difficult to estimate for T>(TD—8°C).

When 20°C<T<(TD-8°C), the total heat transferred to the model food product was
dominated by condensation heat transfer. The heat transferred to the product over this
temperature range by condensation was an order of magnitude higher than the heat
transferred by pure convection. However, for T>(TD-8°C), the contribution of convection
becomes more important, and eventually dominates when the block temperature was
above the bulk gas dew point temperature. Over this condensation/convection transition
region (i.e., (TD-8°C)<T<TD), hm,avg was unstable when estimated using equations 3.44
and 3.45 (Figure 4.3.1). This may have been partially a result of the difficult nature in

calculating the exact water vapor concentration at the product surface as a function of

63

product surface temperature, combined with the possibility of condensed water
evaporating and affecting the local water vapor concentration. In addition, the
assumption that all condensed water was immediately blown off the product by the
impinging jets was also observed to be somewhat invalid in the experimental trials.
Although most of the condensing water was removed from the product by the force of the
impinging jets, some water buildup on the top surface did occur, especially during the
F=50% runs, when the jet exit velocities were the lowest. This water accumulation on
the product surface may have contributed to the instability of the estimated hm,avg values

by affecting the local water vapor concentrations.

0.500
0.450 444 444 44 W W- WW
0400 MT —Lz "c -‘T T“ 8°C

A 0.350 ,____ 4 —- = 4 4 = n-

t I -444

E 0.300444 444 4 4W- \\ unsabej (To—rem}

" 0.250 444-4 4 4 44444— — - 4 +rep2,
0.200 444 - mp3

4: 0.150 4 , 44—4
0.100 44444 4 -
0.0504
0.000 7 rrrrr 1 ‘i r r r . 4

¢0e°egeeeqewebe°s\«“4e°
tlme(s)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 4.3.1: Incremental values of hm,avg for the T..=177°C, Mv=30%, and F=100%
test condition.

The results for the models based on Gardon and Akfirat (1966), Saad et al.
(1980), and Martin (1977) compared with the measured heat and mass transfer
coefficients are presented in sections 4.3.2, 4.3.3, and 4.3.4, respectively. A summary is

presented in section 4.3.5. Each model was evaluated in terms of % average absolute

 

error and SEP. In addition, the errors of each model in considering the effects of

temperature, and exit velocity were considered.

4.3.2 Gardon and Akfirat (1966)

Measured hm,lvg values estimated from the Mv>~0% runs were compared to those
predicted using the model based on Garden and Akfirat (1966), combined with the edge
consideration term (Figure 4.3.2). The predicted values had an average absolute error of

16.1% from those measured, with an SEP of 19.1 mm/s.

 

 

 

 

200 , - ,
A l ' 0
U) l 1
3 1504 4 4 4; 4 44 . 4 4 4W W- _-
5 . ° , 0 0
£100 —— — 14 4 - 44— - — 44
E 1 3° 1
c 504 44 44 44 4 W. _ W .W1WW- W__ WW
2 ° . .
0 4 .1 l
0 50 100 150 200
predicted hum, (mm/s)

Figure 4.3.2: Predicted vs. measured hm,avg values using the model based on Gardon
and Akfirat (1966).

Although the measured hmvg values are spread over a large range around the
predicted values, the average residual is —8.1 mm/s. For the Mv~0% runs, the model
based on Gardon and Akfirat (1966) was found to consistently over-predict the
convective heat transfer coefficient by 13 W/mzK; therefore, the predicted hm,avg values

were anticipated to be greater than those measured.

65

 

belo
exp]
£1166
is ex

llillUl

hm m... residual (mm/gr

Figur

4.3.3

Predit

COHSR

13,54

 

 

When the residuals were analyzed with respect to the Mv set point (Figure 4.3.3),
an error during the Mv~90% runs is apparent. The prediction of hm,avg is consistently
below the measured values for all Mv settings except for the Mv~90% runs. A further
explanation of this phenomena is included in section 4.5. However, if the Mv~90% runs
are excluded from consideration, the average residual becomes —13.7 mm/s. This result
is exactly what was expected, given the results for havg and assuming the analogous

nature between heat and mass transfer.

 

 

 

h mresldual(mmls)

'88
ll
11
ii;
i :1

:11 ,
i
ii
“it:
i
‘I
ll

 

20 30 40 50 60 70 80 90 1 00
Ill" set point (%)

Figure 4.3.3: Residuals of hm,avg using the model based on Gardon and Akfirat
(1966) vs. the Mv set point.

4.3.3 Saad et al. (1980)

Measured hm,avg values estimated from the Mv>~0% runs were compared to those
predicted using the model based on Saad et al. (1980), combined with the edge
consideration term (Figure 4.3.4). The predicted values had an average absolute error of

15.5% from those measured, with an SEP of 19.4 mm/s.

66

200

 

 

 

 

 

l

E 150 4 4 444444 L 4 4444444 144W --WW -W WW W W
-; , o . 8°§

a. . a l ,3
E100 —444444’4~4— _ I -- are
E ‘ «”1 1
3504—4—44 4444 44444444- ___,
g l l 1

O l l 1
O 50 100 150 200

predicted hm,“ (mm/s)

Figure 4.3.4: Predicted vs. measured hm,avg values using the model based on Saad et
al. (1980).

Analysis of the residuals from Figure 4.3.4 yielded an average residual of 45.9
mm/s for all hm,alvg values measured; however, that number decreases to 411.5 when the
MV~90% runs were not considered. When the residuals were plotted versus the Mv set
point (Figure 4.3.5), the predicted values of hm“,g were consistently higher than those
measured, except for the Mv~90% runs. In addition, a plot of the residuals versus exit
velocity (Figure 4.3.6) revealed an inaccurate description of effects due to jet exit
velocity, similar to the dependency of havg on jet exit velocity discussed in section 4.1.4.

As the jet exit velocity increases, the over-prediction by the model based on Saad et al.

(1980) increases.

67

O)
O

 

 

 

 

1? o

4 . WW , 4W W W WWW—W .
a 0
g 20 o 4 444 44 -,W.. .W- -WW.
“ o
.43 . 4 44:44 44
w
9-m4444 4 4444 444 44.444 .844
§~w4 44434 4444434 8 44444 4 44
'3 0

-6O

20 30 4O 50 60 70 80 90 100
Mv set point (%)

Figure 4.3.5: Residuals of hm,avg using the model based on Saad et al. (1980) vs. the

 

 

Mv set point.
60
o
40 -44- 844 4 4 4 4 444 404 4 4
m4 4 4 _i 44 444 4 43—444—

hmm residual (mm/s)
o
l

 

 

 

-20 W , W, W W W _W
40-4-4 .44 44 44 44 444444414444
4%
15 ms W5 25
mm“)

Figure 4.3.6: Residuals of hmvg using the model based on Saad et al. (1980) vs. jet
exit velocity.

4.3.4 Martin (1977)

Measured hmvg values estimated from the Mv>~0% runs were compared to those
predicted using the model based on Martin (1977), combined with the edge consideration
term (Figure 4.3.7). The predicted values had an average absolute error of 12.7% from

those measured, with an SEP of 16.4 mm/s.

68

 

 

 

 

200 ‘ 44
“ * x
E 150 4+ 4 4 44 4 4 4 44444444444 ' 44444 44444
g 0 l0 9
5 . 0‘ 13 3 ‘
35100 444 44414444 44 44444 444444-—
. 1
g 50 i i
a 4 4 4 4 4 4 4464— 44444 444444
a o . *
‘ 7.4
o 4 i
0 50 100 150 200
predicted hm”, (mm/s)

Figure 4.3.7: Predicted vs. measured hm.vg values using the model based on Martin
(1977).

Analysis of the residuals from Figure 4.3.7 yielded an average residual of 2.4
mm/s for all hm,avg values measured; thus demonstrating the even distribution similar to
that predicted for havg. Analysis of the residuals, as a function of the Mv set point, from
this model (Figure 4.3.8), again showed similar results for all the settings except the
Mv~90% runs. The predicted hm,avg values for the Mv~90% runs were consistently lower

than the measured values.

69

 

residual hmm (mm/s)
N «b
O O O
i 1 1
1
We to o ‘
g ‘ 1
. 1 r
. ‘ ,
1 i 1 ,
. 1 I .
. _ ‘ k
1 1 ‘ 1
1 i 1
M‘ 1
1 l
1 1
' 1
«not» #09 o
1 I 1
l l 1

 

 

 

20 30 40 50 60 7O 80 90 1 00
Mv set point (%)

Figure 4.3.8: Residual of hm,avg using the model based on Martin (1977) vs. the Mv
set point.

4.3.5 Summary

When the predicted versus actual measurements for the convective mass transfer
coefficient were analyzed, the models based on Gardon and Akfirat (1966), Saad et al.
(1980), and Martin (1977) were accurate to an average absolute error of 16.1, 15.5, and
12.7%, respectively. In addition, the SEP for the four methods was 19.1, 18.6, and 16.4
mm/s, respectively. Figure 4.3.9 shows the predicted versus measured hmvg values for
all Mv>~0% runs. In addition, the measured and predicted average heat and mass transfer

coefficients are tabulated in Appendix A.

70

200

 

 

 

 

 

 

:T T i
‘ F
‘70? 1 1
E 160414 4 — — — 4—[44 — — 444444434$ 444444
5 1 0 m
5 l 0 D o >P‘ '23.
212044414414 4° 51
g 1 ° e 11%?
'3 o )0 1
3 Q _L
g 801- ._ i __ I ‘0 _f-‘—“f‘_‘ 4" 4oManEY1977S" "—4— 44 44—4
5 fig ‘ iDGardon et. al. (1966) I
1 }Wsiaflfléi (1929.) ___J'
40 4:53 L T
40 80 120 160 200
Predicted hm,“I (mm/s)

Figure 4.3.9: Predicted vs. measured hm”: values for all runs (rep 1 only).

As expected, the models that most accurately predicted the measured heat transfer
coefficients also most accurately predicted the mass transfer coefficients. However, the
accuracy in predicting the mass transfer coefficients is clearly less than the accuracy in
predicting the heat transfer coefficients. The larger error in predicting h,1W8 versus havg
may be partially due to the difficulty of predicting the gas properties of air/water vapor
mixtures as a function of temperature. Errors could also have been introduced by
inaccuracies in assumptions 2, 3, and 4 (see section 3.2.1).

The second assumption states that the model food product absorbs all latent heat
lost by the condensing water vapor. This assumption may not be accurate, because the
impingement gas near the product surface is at a temperature near the product surface
temperature, thus some of the latent heat given off by the condensing water vapor may be
absorbed back into the spent gas.

The third and fourth assumptions state that only latent heat is transferred from the

condensed water vapor to the product surface, and all water vapor condensed on the

71

product is immediately blown off by the impinging gas jets. However, as water vapor
condensed onto the product surface, that condensed water was observed to collect in
small pools of water on the product until a sufficient amount had collected for an
impinging gas jet to blow the bead off the product. Between the moment of condensation
and the time the condensed water was blown off the product surface, the condensed water
would exchange sensible heat with the model food product. This phenomenon would
also introduce an error in that some of the product surface was partially covered by beads
of condensed water. Therefore, because the condensation rate is driven by a
concentration gradient, the concentration of water vapor near the surface of the
condensed water bead was different than the concentration near the exposed product
surface. This phenomenon probably introduced most of the increased error in the

prediction of the convective mass transfer coefficient.

4.4 Contribution of the Edge Effect Term

The accuracy of assuming turbulent parallel boundary layer flow to the model
product edges was not explicitly tested in the present study. Future research should
measure the edge contribution by using blocks of varying height. However, a limited
analysis was performed on the contribution of the computed edge effect on the overall
prediction of ha,vg and hmmg.

When the heat and mass transfer coefficient on the product edges was assumed to
be the same as that on the product surfaces under impingement flows (i.e., impingement
flows assumed over the entire block), the % average absolute error increased for the

prediction of havg and hmalvg for each model (Table 4.3 and Table 4.4).

72

Table 4.3: Effects of edge consideration on % average absolute error in predicted
vs. measured havg for Mv~0% runs.

 

 

 

 

model without edge consideration with edge consideration
Gardon and Akfirat 17.3 14.7
(1966)
Martin (1977) 3.7 1.9
Saad et al. (1980) 14.6 12.2

 

 

 

 

 

Table 4.4: Effects of edge consideration on % average absolute error in predicted
vs. measured hm,m for Mv>0% runs.

 

 

 

 

model without edge consideration with edge consideration
Gardon and Akfirat 16.1 16.0
(1966)
Martin (1977) 15.5 15.4
Saad et al. (1980) 13.2 12.7

 

 

 

 

 

These results suggest that the edge effect term, as included in this analysis, was a
valid means for modeling non-impingement convection on the product edges. In
particular, the increased accuracy of the model based on Martin (1977), when combined
with an edge effect term, gives credibility to the use of this model in predicting heat and

mass transfer coefficients to discrete food products in an impingement oven.

4.5 Errors in M.~90% Tests

An environment of Mv=90% was never fully achieved in the JSO-IV oven while
performing this study. For the Mv~90% runs, the impingement gas moisture content by
volume was measured to be 75%<Mv<86%. The actual measured Mv conditions for the
MV ~90% runs are listed in Table 4.5. The measured values of Mv were used in all

analyses.

73

Table 4.5: Measured values of M" for the Mv~90% runs.

 

 

 

 

 

 

 

 

 

 

 

Dry Bulb Temp (°C) Fan Speed (% full) Measured Mv (%)
121 50 82
121 75 82
121 100 75
177 50 84
177 75 85
177 100 86
232 50 83
232 75 80
232 100 80

 

 

 

 

The Humitrol 11 (Machine Applications Corporation, Sandusky, OH, model:
MAC 120-100-F) used to measure the impingement gas moisture content was mounted
near the middle of the J SO-IV oven. Therefore, because the system never reached the
volumetric moisture content set point during the Mv~90% runs, the J SO-IV oven was
constantly adding steam into the impingement plenums, and the impingement gas may
never have become homogeneous. This possible lack of sufficient mixing may have
resulted in a different Mv in the impingement gas near the oven entrance than was being
sensed by the Humitrol II at the middle of the oven.

Figure 4.3.3, Figure 4.3.5, and Figure 4.3.8 show the residuals from the prediction
of hm“,g using the four models versus My. In each case, the predicted hm,avg value for the
Mv~90% runs was consistently lower than the predicted value at the other Mv settings.
One possible explanation for this under-prediction is that the actual Mv in the
impingement gas near the entrance region of the J SO-IV oven may have been higher than
the Mv measured by the Humitrol II. A separate measurement of the moisture content by

volume was not taken, and all explanations are speculation; however, the analysis of the

74

residuals as a function of the MV does demonstrate that some error existed in either the

measurement or the prediction of the mass transfer coefficient for these runs.

4.6 Temperature Dependency

The effects of bulk gas temperature on the prediction of the heat and mass transfer
coefficients were analyzed. No effect of temperature was observed on the predictive
accuracy of the three models for the heat transfer coefficients. However, for the
prediction of the mass transfer coefficients, a clear trend was observed.

For each model, the average residual increased as the impingement gas
temperature increased from 121 to 177°C (p<0.001); however, as the bulk impingement
gas temperature was raised from 177 to 232°C, the average residual stayed relatively
constant (see Figure 4.6.1, Figure 4.6.3, and Figure 4.6.2). This may indicate an error in
the prediction of some gas property value near 121°C.

Most likely, the error is in the prediction of the mass diffusivity. The mass
diffusivity of water vapor in air was calculated using eqn. 3.26. This equation determines
the mass diffusivity as a function of temperature only, with no consideration of other
factors such as water vapor concentration, or possible phase changes of the moisture near

the boiling point.

75

 

 

 

%§ 40 . W _,W W 4 ,.,W 6., _ A W $?W W
5.20 434 44°44 4 44
'76
.43 o 4 4 4 4 4 4 [ 4
3
£420 4 4444444 404
35-40 4 44o 444444 44— 44 43-4 4
0
-60
100 1 20 1 40 160 180 200 220 240
gas temp (°C)
Figure 4.6.1: Residuals of hmIvg vs. gas temperature for the model based on Gardon
and Akfirat (1966).
60 4

 

11mm residual (mm/s)
o

 

 

 

 

Figure 4.6.2: Residuals of hm,avg vs. gas temperature for the model based on Saad et
al. (1980).

140

1 60 1 80
gas temp (°C)

76

 

 

O)
O

 

 

 

 

o
7,,‘404.-WWWW444W4WWWWWWWQ WWWWWW WWW-
E ° 0
520444434 444 4 4— 4444
2»: 3 i
s o4 -4 4 - 44 44 4 444
.C
i __ _-__ _ ~____ ______ .
:3 20 f O
in
2440-4 44 4 444 4444

-60
1 00 120 1 40 160 180 200 220 240

gas temp (°C)

Figure 4.6.3: Residuals of hm,avg vs. gas temperature for the model based on Martin
(1977).

4. 7 Dew Point Temperature Prediction

The prediction of the transition temperature between condensation heat transfer
and pure convection heat transfer was assumed, in model development, to be the dew
point temperature. Once the block reached the calculated dew point temperature of the
bulk impingement gas, condensation was assumed to stop. Table 4.6 gives the predicted
dew point temperatures at the actual Mv values measured during testing, calculated by

eqn. 3.33.

77

"L

Table 4.6: Predicted dew point temperature for the measured Mv values used.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Mv (%) Predicted dew point (°C)
1 5
5 34
13 53

30 70
50 81
7O 88
75 90
80 91
82 92
83 92
84 93
85 93
86 93

 

 

 

 

The accuracy of the above assumption was evaluated by observing curves of
predicted block temperature versus the actual measured block temperature. The
condition of T...=232°C and F=75%, for all five Mv settings, was chosen at random for
demonstration. For conditions Mv<50%, the temperature at which condensation heat
transfer ceases was estimated well by this method (Figure 4.7.1). In particular, for the
MV~O% run, in which the measured MV was 13%, the predicted temperature curve breaks
at exactly the same point as the measured block temperature curve. This break indicates

a transition from condensation heat transfer to pure convection heat transfer.

78

 

100

 

      

 

 

 

90 e
E 80 ~
a 70 4
g 60 -
1:, 50 A _ _
E 40 n A m -- 44 predicted Mv %= 13
I A actual Mv % = 30
3° * A _, @6410th Mafia) r
20 = T . , ,
0 10 20 30 40 50 60 70

time (s)

Figure 4.7.1: Predicted vs. measured block temperature for T..=232°C, F=75 %, and
Mv=13 and 30% using the model based on Martin (1977).

For impingement gas Mv>50%, the prediction of this break point becomes less
accurate. The temperature of the break is consistently under-predicted, as shown in
Figure 4.7.2 and Figure 4.7.3. The break temperature was under-predicted for the MV =
50 and 70% runs by approximately 5°C, and by approximately 10°C for the MV ~ 90%
runs. Possible explanations for the worse prediction at Mv ~ 90% were discussed in
section 4.5.

The under-prediction of the break temperature may be due to condensed water
vapor buildup on the product surface, which was discussed in section 4.3.5. The
measured temperature of the model food product may have over-shot the calculated dew
point temperature due to condensed water on the product surface giving up additional
sensible heat to the product after condensation had ceased. This also explains the re-
convergence of the predicted versus measured product temperature curves, as the

condensed water was evaporated off the product surface.

79

 

100
90 4
80 ~
70 4
60 a
50 .
40 -

Block Temp (°C)

3044

 

1

1

1

1

1
i

It

11
7!:

4+
1+
“'1"
.4.
1+
w-l-
14+
1':

_ x measuréi KAT/7% = 501‘”
— 4 44 predicted Mv °/o = 50 1

+ measured Mv % = 70

4__4 - predicted Mv % = 70 L,

 

 

 

20

 

30 40 50
time (s)

Figure 4.7.2: Predicted vs. measured block temperature for T..=232°C, F=75%, and
Mv=50 and 70% using the model based on Martin (1977).

 

".l

 

 

 

 

 

 

 

 

1:: W- WuggnEfiSW”
8 8044 4 4444 DD/ 1 ,___Ag_g
2.. DD :
a. 701A‘ _EIU7/7/F -# 4—47 44 WWWW_*
'9 60 4 4 44~E44z444444 4 4 W W W W W W WWA
g 504 --4 D 44 — W W y W- W _W
a? 40 4 — 4 44 4444 W- 1 El actual Mv %=80 {Wt
30 - WW .. 1: 44 predicted Mv % = _8fl0_1_
20 c- . . ,
O 10 20 30 40 50

time (s)

Figure 4.7.3: Predicted vs. measured block temperature for T..=232°C, F=75%, and
Mv~90 % using the model based on Martin (1977).

80

 

Chapter 5

5 Conclusions

5.1 General Conclusions

Regardless of the experimental assumptions and errors experienced during the
present study, the correlations developed by Gardon and Akfirat (1966), Saad et al.
(1980), and Martin (1977), coupled with an edge effect term, predicted the heat transfer
coefficients experienced by model food products in a commercial impingement oven to
14.7, 12.2, and 1.9% average absolute error, respectively. In addition, the standard error
of prediction for the for models was 13.1, 11.8, and 1.9 W/mzK, respectively. The
addition of the edge effect term was found to improve the accuracy of each model by
reducing the average absolute error by approximately 2%.

The analogous nature between heat and mass transfer was used to predict the
convective mass transfer coefficients experienced by a model food product from
correlations developed for pure convective heat transfer. The relationship recommended

Nu

by Martin (1977), Pr 0 42 = 555142 , was used to calculate the mass transfer coefficients
. c .

 

from correlations for the heat transfer coefficients.

Condensation heat transfer was successfully modeled as a convective mass transfer
phenomenon driven by a water vapor concentration gradient between the concentration in
the bulk impingement fluid and the concentration at the product surface. The water vapor
concentration at the product surface was calculated as the saturation vapor concentration

at the product surface temperature.

81

 

The correlations developed by Gardon and Akfirat (1966), Saad et al. (1980), and
Martin (1977) were coupled with an edge effect term, and modified to predict the
convective mass transfer coefficient. The three models were found to predict the mass
transfer coefficient to an average absolute error of 16.1, 15.5, and 12.7%, with SEP’s of
19.1, 19.4, and 16.4 mm/s, respectively

The model developed by Martin (1977) for impingement flow from an array of slot
nozzles (ASN) to a flat plate was found to most accurately predict the heat and mass
transfer coefficients experienced by discrete model food products in a moist air
impingement oven when used in conjunction with an edge effect term.

Correlations have been developed to predict the convective heat transfer
coefficients experienced by a flat plate under an array of slot nozzles. Research has also
been conducted to describe condensation as a mass transfer phenomenon, driven by vapor
pressure or dew point temperature gradients. However, no previous studies considered
the application of condensing-convective heat transfer to discrete objects passing through
an array of slot nozzles impinging moist air. The present study has shown that
correlations developed for impingement flow over a flat plate can be modified, with the
addition of an edge effect term, to predict heat transfer to discrete objects under dual
ASN flow. The present study has also shown that those correlations for heat transfer can
be modified through analogies between heat and mass transfer to predict convective mass
transfer coefficients, which subsequently can be used to predict condensation rates.

The present study utilized discrete objects with a diameter 1.4 times the average
slot spacing (i.e., d/28=1.37). The models presented here should be used with caution for

conditions of d/2S less than 1. The previously developed predictive models for flow

82

 

 

under the impingement jets were developed for flow over a flat plate. As d/ZS becomes
less than 1, the use of these models for flow over the impingement surfaces may become

invalid.

5.2 Recommendations for Future Work

The effect of varying geometric oven parameters was not evaluated in this study.

 

The model based on Martin (1977) was found to most accurately predict the heat and
mass transfer coefficients given the geometric oven parameters used.

Future work should evaluate the validity of this model under varying slot to product
surface spacing (H/w) and different fractions of nozzle open areas (f). Because the model
by Martin (1977) is most sensitive to the geometric f value, this analysis would be
important in demonstrating broad applicability of the Martin (1977) model for the
prediction of heat and mass transfer rates in different impingement ovens. The fraction
of nozzle open area could be modified by the addition or removal of slot jets to the
system.

The validity of using the correlation for turbulent, boundary layer flow as an edge
correction factor was also not explicitly tested in this study. Future work should include
an examination of the effect of this parameter. This could be accomplished by using
model food products of various edge height (E) to collect heat transfer data. The
measured overall convective heat transfer coefficients could then be compared to those
predicted with and without edge consideration.

Future work might also include the collection of condensed water vapor, so a direct
analysis of the total mass transferred could be performed. In the present study, mass

transferred to the product was indirectly calculated using heat transfer data and assuming

83

a specific release of latent heat by the condensing water vapor. A direct measurement of
the mass transfer rate would be more conclusive. An isothermal model food product
could be used for these studies, thus allowing for the establishment of a steady state mass
transfer rate.

The use of convective mass transfer coefficients to predict heat transfer due to
water vapor condensation was analyzed in the present study. These mass transfer
coefficients may also be useful in the prediction of moisture loss from a food product as
the product temperature increases above the bulk impingement fluid dew point
temperature. If water is present on the product surface from condensation or from water
migration from the interior of the product (calculated with equilibrium isotherms), the
water vapor concentration near the surface would become greater than the concentration
in the bulk gas, thus creating a concentration gradient from the surface to the bulk gas.
Evaporative cooling and moisture loss could then be modeled as a mass transfer
operation, in the reverse direction of condensation, using the convective mass transfer
coefficients predicted for condensation heat transfer.

The present study has shown that correlations developed to predict heat transfer
coefficients from ASN flow over a flat plate can be modified to predict heat and mass
transfer coefficients to discrete objects in a dual ASN flow moist air impingement oven.
However, future work is needed to evaluated the robustness of these models under a wide

variety of flow configurations, and products.

84

 

References

Behnia, M., Parneiz, S. and Durbin, RA. 1998. Prediction of heat transfer in an
axisymmetric turbulent jet impinging on a flat plate. Int. J. Heat and Mass
Transfer, Vol. 41, pp. 1845-1855.

Bejan, A. 1991. Film condensation on an upward facing plate with free edges” Int. J.
Heat and Mass Transfer, Vol 34, pp. 578-582.

Braud, L.M., Moreira, R.G., and Castell-Perez, ME. 2001. Mathematical modeling of
impingement drying of corn tortillas. Journal of Food Engineering. Vol. 50, pp.
121-128.

Burmeister, LC. 1983. Convective Heat Transfer. John Wiley & Sons Inc., New York,
NY.

Colbum, AP, and Hougen, O.A. 1934 Design of cooler condensers for mixtures of
vapors with non-condensing gases,” Ind. Eng. Chem. Vol.26. No. 11. pp. 1178-
1182.

CRC Handbook of Chemistry and Physics, 70:}; ed. 1989. Cleveland, Ohio : CRC Press.

Denny, V.E., Mills, AF, and Jusionis, VJ. 1971. Laminar film condensation from a
steam air mixture undergoing forced flow down a vertical surface. J. Heat
Transfer. Vol. 93, pp. 297-304.

Gardon, R., and Akfirat, J .C. 1965. The role of turbulence in determining the heat-

transfer characteristics of impinging jets. Int. J. Heat and Mass Transfer. Vol. 8,
pp. 1261-1272.

Gardon, R., and Akfirat, J.C. 1966. Heat transfer characteristics of impinging two-
dimensional air jets. J. Heat Transfer, Vol. 88, pp. 101-108.

Geankoplis, CJ. 1993. Transport Processes and Unit Operations, 3'd ed. Prentice-Hall,
Inc. Englewood Cliffs, New Jersey.

Henke, MC. 1984. Air impingement technology. Advances in Catering Technology.
Vol. 3, pp. 107-113.

Irvine, T.F.Jr., and Hartnett, JP. 1976. Steam and air tables in SI units. Hemisphere
Publishing Corporation. London England.

Kroger, M., and Krizek, F. 1966. Mass transfer coefficient in impingement flow from
slotted nozzles. Int. J. Heat and Mass Transfer. Vol. 9, pp. 337-344.

85

 

 

Martin, H. 1977. Heat and mass transfer between impinging gas jets and solid surfaces.
Advances in Heat Transfer, Vol. 13, pp. 1-60.

Mason, EA. and Monchik, L. 1962. Transport properties of polar gas mixtures. J.
Chem. Phys. Vol. 36, PP. 2746-2757.

Merte, Jr. 1973. Condensation heat transfer. Advances in Heat Transfer. Vol 9, pp. 181-
272.

Midden, T.M. 1995. Impingement air baking for snack foods. Cereal Foods World.
Vol. 40, pp. 532-535.

Mujumdar, AS. and W.J.M. Douglas, 1972, “Impingement heat transfer: Literature
Review,” Paper presented at TAPPI Conference, New Orleans.

National Bureau of Standards Circular No. 564, 1955.

Nitin, N., and Karwe, M.V. 2001. Heat transfer coefficient for cookie shaped objects in a
hot air jet impingement oven. Journal of Food Process Engineering. Vol. 24,
issue 1, pp. 51-69.

Obot, N.T., Mujumdar, A.S., and Douglas, W.J.M. 1979. Design correlations for heat
and mass transfer under various turbulent impinging jet configurations.
Developments in Drying. Hemisphere Publications, pp. 388-402.

Ovadia, D.Z., and Walker, CE. 1998. Impingement in food processing. Food
Technology. Vol. 52, issue 4, pp. 46—50.

Ozisik, NM. 1993. Heat Conduction, 2"" ed. John Wiley & Sons, Inc. New York. NY.

Perry et al. 1997. Perry ’s Chemical Engineers’ Handbook 7:}: ed. McGraw-Hill. New
York, NY.

Peterson, P.F., Schrock, V.E., and Kageyama, T. 1993. Diffusion layer theory for
turbulent vapor condensation with non-condensable gases. Transactions of the
ASME. Vol. 115, PP. 998-1003.

Polat et a1. 1988. Numerical flow and heat transfer under impinging jets-A Review.”
Annual Review of Numerical Fluid Mechanics and Heat Transfer, Vol. 2, pp. 157-
197.

Polat, S., and Douglas, W.J.M. 1990. Heat transfer under multiple slot jets impinging on
a permeable moving surface. AIChE J. Vol. 36. pp. 1370-1378.

Polat, S. 1993. Heat and mass transfer in impingement drying. Drying Technology. Vol.
11, pp. 1147-1176.

86

 

 

Rose, J .W. 1980. Approximate equations for forced convection condensation in the
presence of a non-condensing gas on a flat plate and horizontal tube. Int. J. Heat
and Mass Transfer. Vol. 23, pp. 539-546.

Saad, N.R., Mujumdar, AS. and Douglas, W.J.M. 1980. Heat transfer under multiple
turbulent slot jets impinging on a flat plate. In Drying ’80, Developments in
Drying, (A.S. Mujumdar, ed.) pp. 422-430, Hemisphere Publishing Corp., New
York.

Saad, N .R., 1981, “Flow and Heat Transfer for Multiple Turbulent Impinging Slot Jets”
PhD. Thesis, Chem. Eng. Dept., McGill University.

Seyedein, S.H., Hasan, M., and Mujumdar, AS. 1995. Turbulent flow and heat transfer
from confined multiple impinging slot jets. Numerical Heat Transfer, Part A.
Vol. 27. PP. 35-51.

Sparrow, E.M., Minkowycz, W.J., and Saddy, M. 1967. Forced convection condensation
in the presence of non-condensables and interfacial resistance. Int. J. Heat and
Mass Transfer. Vol. 10, pp.1243-1244.

Subba Raju, K., and Schlunder, E.U. 1977. Heat transfer between an impinging jet and a
continuously moving flat surface. Warmeund Stofi‘ubertragrung, Vol. 10, pp.
131-136.

Tanasawa. 1991. Advances in condensation heat transfer. Advances in Heat Transfer,
Vol 21, pp. 55-139.

Tzeng, P.Y., and Soong, CY. 1999. Numerical investigation of heat transfer under
confined impinging turbulent slot jets. Numerical Heat Transfer: Part A:
Applications, Vol. 35, n 8.

van Heiningen, P.R.P. 1982. Heat transfer under an impinging slot jet. PhD. Thesis,
McGill University.

Vargaflik, N.B. et al. 1966. Handbook of Physical Properties of Liquids and Gases:
Pure Substances and Mixtures, New York : Begell House.

Wahlby, U., Skjoldebrand, C., and Elmar, J. 2000. Impact of impingement on cooking
time and food quality. Journal of Food Engineering. Vol. 43, Issue 3, pp. 179-
187.

87

 

 

APPENDICES

APPENDIX A: Predicted and Measured Convective Heat and Mass Transfer
Coefficients

APPENDIX B: Block Temperature Data

88

 

 

APPENDIX A

Predicted and Measured Convective Heat and Mass Transfer Coefficients

89

 

 

Table A.l: Predicted and measured convective heat and mass transfer coefficients
from the model based on Gardon and Akfirat (1966).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L predicted measured

Condition Rgp hm hm, h; 111...; havg hm,“ hM 1h... ..,
T... M" F W/m2*K mm/s W/m2*K mm/s W/m2*K mm/s W/mzK mm/s
121 O 50 1 85 90 79 85 82 89 70

1 21 0 50 2 85 90 79 85 82 89 69

1 21 0 5O 3 85 90 79 85 82 89 71

121 0 75 1 108 116 108 117 107 116 96

121 0 75 2 108 116 108 117 107 116 93

121 0 75 3 108 116 108 117 107 116 93

121 0 100 1 128 138 134 146 129 141 118

121 O 100 2 128 138 134 146 129 141 116

121 O 100 3 128 138 134 146 129 141 117

1 77 0 50 1 83 96 75 88 80 94 71

177 0 50 2 83 96 75 88 80 94 70

1 77 O 50 3 83 96 75 88 80 94 72

177 0 75 1 105 124 102 121 103 123 91

177 O 75 2 105 124 102 121 103 123 90

177 0 75 3 105 124 102 121 103 123 90

177 0 100 1 125 148 127 152 124 149 111

177 0 100 2 125 148 127 152 124 149 110

177 0 100 3 125 148 127 152 124 149 110
232 0 50 1 81 102 72 91 77 99 63
232 0 50 2 81 1 02 72 91 77 99 63
232 0 50 3 81 102 72 91 77 99 63
232 0 75 1 1 03 1 32 97 125 1 00 1 30 86
232 0 75 2 1 03 1 32 97 1 25 1 00 1 30 86
232 0 75 3 1 03 1 32 97 1 25 1 00 1 30 87
232 0 100 1 123 158 121 157 120 158 108
232 0 100 2 123 158 121 157 120 158 105
232 O 100 3 123 158 121 157 120 158 106
1 21 30 50 1 86 92 80 87 84 90 51 40
1 21 30 50 2 86 92 80 87 84 90 50 38
1 21 30 50 3 86 92 80 87 84 90 51 50
121 30 75 1 109 118 109 119 109 118 90 114
121 30 75 2 109 118 109 119 109 118 88 118
121 30 75 3 109 118 109 119 109 118 87 139
121 30 100 1 130 141 136 150 131 143 114 116
121 30 100 2 130 141 136 150 131 143 112 124
121 30 100 3 130 141 136 150 131 143 112 154
1 77 3O 50 1 84 97 80 89 81 95 51 75
1 77 30 50 2 84 97 80 89 81 95 50 76
177 30 75 1 1 O7 125 1 04 1 23 1 05 125 82 105

 

90

 

 

 

 

 

 

 

 

 

 

 

I predicted measured
Condition Rep 11.... hm .m he hm; hm, hm h" h...
r... M., F W/r-rii'K mm/s W/m2*K mm/s wnfifik mm/s W/mzK «R731
177 30 75 2 107 125 104 123 105 125 82 110
177 30 75 3 107 125 104 123 105 125 82 101
177 30 100 1 108 150 130 154 127 151 108 130
177 30 100 2 108 150 130 154 127 151 108 132
177 30 100 3 108 150 130 154 127 151 107 129
232 30 50 1 83 1 03 73 92 79 1 00 64 89
232 30 50 2 83 1 03 73 92 79 100 62 92
232 30 50 3 83 1 03 73 92 79 1 00 64 91
232 30 75 1 105 132 100 126 102 131 87 1 18
232 30 75 2 105 132 100 126 102 131 88 1 13
232 30 75 3 105 132 100 126 102 131 88 1 14
232 30 100 1 125 158 124 159 123 159 108 125
232 30 100 2 125 158 124 159 123 159 107 118
232 30 100 3 125 158 124 159 123 159 106 134
1 21 50 50 1 83 91 79 88 82 90 57 69
121 50 50 2 83 91 79 88 82 90 51 73
121 50 50 3 83 91 79 88 82 90 53 76
121 50 75 1 106 116 108 121 106 118 96 96
121 50 75 2 106 116 108 121 106 118 94 96
121 50 75 3 106 116 108 121 106 118 93 93
121 50 100 1 126 139 135 151 128 142 120 111
121 50 100 2 126 139 135 151 128 142 119 113
121 50 100 3 126 139 135 151 128 142 118 109
177 50 50 1 82 96 76 90 80 94 39 96
177 50 50 2 82 96 76 90 80 94 43 94
177 50 50 3 82 96 76 90 80 94 44 96
177 50 75 1 105 123 103 124 103 124 88 117
177 50 75 2 105 123 103 124 103 124 86 114
177 50 75 3 105 123 103 124 103 124 87 112
177 50 100 1 125 147 129 156 125 150 104 135
177 50 100 2 125 147 129 156 125 150 101 134
177 50 100 3 125 147 129 156 125 150 109 133
232 50 50 1 81 1 01 73 93 78 99 68 93
232 50 50 2 81 1 01 73 93 78 99 67 98
232 50 50 3 81 1 01 73 93 78 99 68 90
232 50 75 1 1 03 1 30 99 127 1 01 129 97 127
232 50 75 2 1 03 1 30 99 127 1 01 129 95 124
232 50 75 3 103 130 99 127 101 129 97 1 14
232 50 100 1 122 156 124 160 121 157 110 145
232 50 100 2 122 156 124 160 121 157 109 147
232 50 100 3 122 156 124 160 121 157 108 146
121 70 50 1 78 88 76 89 77 88 41 65

 

 

 

 

 

 

 

 

 

 

 

 

 

91

 

 

 

 

 

 

 

 

predicted measured
1.. M, F W/m2*K mm/s W/m2*K mm/s W/m *K mm/s W/mzK mm/s
121 70 50 2 78 88 76 89 77 88 4O 67
121 70 50 3 78 88 76 89 77 88 46 65
121 70 75 1 100 113 105 122 101 115 64 93
121 70 75 2 100 113 105 122 101 115 67 94
121 70 75 3 100 113 105 122 101 115 63 92
121 70 100 1 119 135 131 153 122 140 104 109
121 70 100 2 119 135 131 153 122 140 107 112
121 70 100 3 119 135 131 153 122 140 104 108
177 70 50 1 78 93 74 91 76 93 45 86
177 70 50 2 78 93 74 91 76 93 44 90
177 70 50 3 78 93 74 91 76 93 42 90
177 70 75 1 99 120 101 125 99 121 60 119
177 70 75 2 99 120 101 125 99 121 92 127
177 70 75 3 99 120 101 125 99 121 77 120
177 70 100 1 118 143 126 157 119 147 120 146
177 70 100 2 118 143 126 157 119 147 121 149
177 70 100 3 118 143 126 157 119 147 116 144
232 70 50 1 77 98 71 93 74 97 68 90
232 70 50 2 77 98 71 93 74 97 68 89
232 70 50 3 77 98 71 93 74 97 70 91
232 70 75 1 98 126 97 128 97 127 93 122
232 70 75 2 98 126 97 128 97 127 95 126
232 70 75 3 98 126 97 128 97 127 93 126
232 70 100 1 116 151 121 161 116 154 108 161
232 70 100 2 116 151 121 161 116 154 108 164
232 70 100 3 116 151 121 161 116 154 90 155
121 90 50 1 74 86 74 89 74 87 40 98
121 90 50 2 74 86 74 89 74 87 33 104
121 90 50 3 74 86 74 89 74 87 . 43 92
121 90 75 1 95 110 101 122 96 114 44 97
121 90 75 2 95 110 101 122 96 114 47 94
121 90 75 3 95 110 101 122 96 114 51 98
121 90 100 1 116 133 129 153 120 139 75 132
121 90 100 2 116 133 129 153 120 139 79 134
121 90 100 3 116 133 129 153 120 139 75 128
177 90 5O 1 73 91 71 92 72 91 35 95
177 90 50 2 73 91 71 92 72 91 45 116
177 90 50 3 73 91 71 92 72 91 32 98
177 90 75 1 93 116 97 126 94 119 86 103
177 90 75 2 93 116 97 126 94 119 93 133
177 90 75 3 93 116 97 126 94 119 67 128
177 90 100 1 110 138 121 158 113 144 113 138

 

 

 

 

 

 

 

 

 

 

 

 

92

 

 

 

 

 

 

 

[ predicted measured

Condition Rep M hm hL hm; I'I._\,1 hm havg h... "91

T... M" F W/m2*K mm/s W/m2*K mm/s W/m2*K mm/s W/mzK mm/s
177 90 100 2 110 138 121 158 113 144 112 180
177 90 100 3 110 138 121 158 113 144 107 153
232 90 50 1 73 96 69 94 71 95 63 123
232 90 50 2 73 96 69 94 71 95 66 139
232 90 50 3 73 96 69 94 71 95 51 120
232 90 75 1 94 124 95 128 94 125 1 O1 1 37
232 90 75 2 94 1 24 95 1 28 94 1 25 98 146
232 90 75 3 94 124 95 128 94 125 101 135
232 90 100 1 112 148 118 161 113 152 115 167
232 90 100 2 112 148 118 161 113 152 113 175
232 90 100 3 112 148 118 161 113 152 112 161

 

 

 

 

 

 

 

 

 

 

 

 

 

93

Table A2: Predicted and measured convective heat and mass transfer coefficients
from the model based on Saad et al. (1980).

 

 

 

 

 

 

predicted measured
Condition Rep hm h... .m l1; hm; hm h... m hm,El hm avg

T... Mv F W/m2*K mm/s W/m2*K mm/s W/m2*K mm/s W/mzK mm/s
121 O 50 1 80 86 79 85 79 85 70
121 O 50 2 80 86 79 85 79 85 69
121 0 50 3 80 86 79 85 79 85 71
121 O 75 1 109 117 108 117 107 117 96
121 0 75 2 109 117 108 117 107 117 93
121 O 75 3 109 117 108 117 107 117 93
121 O 100 1 135 146 134 146 133 146 118
121 O 100 2 135 146 134 146 133 146 116
121 0 100 3 135 146 134 146 133 146 117
177 O 50 1 77 89 75 88 75 89 71
177 O 50 2 77 89 75 88 75 89 70
177 O 50 3 77 89 75 88 75 89 72
177 0 75 1 104 122 102 121 101 121 91
177 0 75 2 104 122 102 121 101 121 90
177 O 75 3 104 122 102 121 101 121 90
177 O 100 1 128 152 127 152 125 152 111
177 O 100 2 128 152 127 152 125 152 110
177 O 100 3 128 152 127 152 125 152 110
232 0 50 1 74 93 72 91 72 92 63
232 0 50 2 74 93 72 91 72 92 63
232 O 50 3 74 93 72 91 72 92 63
232 0 75 1 99 126 97 1 25 97 1 26 86
232 0 75 2 99 126 97 1 25 97 1 26 86
232 0 75 3 99 126 97 1 25 97 1 26 87
232 O 100 1 123 158 121 157 119 158 108
232 O 100 2 123 158 121 157 119 158 105
232 O 100 3 123 158 121 157 119 158 106
121 30 50 1 81 86 8O 87 8O 86 51 40
121 30 5O 2 81 86 80 87 80 86 50 38
121 30 50 3 81 86 80 87 80 86 51 50
121 30 75 1 109 117 109 119 108 118 90 114
121 30 75 2 109 117 109 119 108 118 88 118
121 30 75 3 109 117 109 119 108 118 87 139
121 30 100 1 135 146 136 150 134 148 114 116
121 30 100 2 135 146 136 150 134 148 112 124
121 30 100 3 135 146 136 150 134 148 112 154
177 30 50 1 78 89 80 89 76 89 51 75
177 30 50 2 78 89 80 89 76 89 50 76
177 30 75 1 105 122 104 123 103 122 82 105

 

 

 

 

 

 

 

 

 

 

 

 

 

94

 

 

 

 

 

predicted measured
Condition Rep hm hm... he hmg hm hm .v havg hm...“I
T... M., F W/m2*K mm/s W/m2*K mm/s W/m2*K mm/s W/mzK mm/s
177 30 75 2 105 122 104 123 103 122 32 110
177 30 75 3 105 122 104 123 103 122 32 101
177 30 100 1 130 152 130 154 123 153 103 130
177 30 100 2 130 152 130 154 123 153 103 132
177 30 100 3 130 152 130 154 123 153 107 129
232 30 so 1 75 92 73 92 73 92 64 39
232 30 so 2 75 92 73 92 73 92 62 92
232 30 so 3 75 92 73 92 73 92 64 91
232 30 75 1 101 126 100 126 96 126 37 113
232 30 75 2 101 126 100 126 93 126 33 113
232 30 75 3 101 126 100 126 93 126 33 114
232 30 100 1 125 157 124 159 122 153 103 125
232 30 100 2 125 157 124 159 122 153 107 113
232 30 100 3 125 157 124 159 122 153 106 134
121 so so 1 73 35 79 33 73 36 57 69
121 so so 2 73 35 79 33 73 36 51 73
121 so so 3 73 35 79 33 73 36 53 76
121 so 75 1 106 116 103 121 105 117 96 96
121 so 75 2 106 116 103 121 105 117 94 96
121 so 75 3 106 116 103 121 105 117 93 93
121 so 100 1 131 144 135 151 131 146 120 111
121 so 100 2 131 144 135 151 131 146 119 113
121 so 100 3 131 144 135 151 131 146 113 109
177 so so 1 7s 33 76 9o 75 39 39 96
177 so so 2 7s 33 76 9o 75 39 43 94
177 so so 3 75 33 76 9o 75 39 44 96
177 so 75 1 102 120 103 124 101 121 33 117
177 so 75 2 102 120 103 124 101 121 36 114
177 so 75 3 102 120 103 124 101 121 37 112
177 so 100 1 126 150 129 156 125 151 104 135
177 so 100 2 126 150 129 156 125 151 101 134
177 so 100 3 126 150 129 156 125 151 109 133
232 so so 1 73 91 73 93 72 91 63 93
232 so so 2 73 91 73 93 72 91 67 93
232 so so 3 73 91 73 93 72 91 63 90
232 so 75 1 93 124 99 127 97 125 97 127
232 so 75 2 93 124 99 127 97 125 95 124
232 so 75 3 93 124 99 127 97 125 97 114
232 so 100 1 122 155 124 160 120 156 110 145
232 so mo 2 122 155 124 160 120 156 109 147
232 so 100 3 122 155 124 160 120 156 103 146
121 70 so 1 73 32 76 39 74 34 41 6s

 

 

 

 

 

 

 

 

 

 

 

 

 

95

 

 

 

 

 

 

predicted measured
Condition Rep hm2E hm .m hg I'M hfiL h... "g hm 11m "9
T... M., F W/m *K mm/s W/m2*K mm/s W/m2*K mm/s W/m2K mm/s
121 70 50 2 73 82 76 89 74 84 40 67
121 70 50 3 73 82 76 89 74 84 46 65
121 70 75 1 99 112 105 122 100 115 64 93
121 70 75 2 99 112 105 122 100 115 67 94
121 70 75 3 99 112 105 122 100 115 63 92
121 70 100 1 123 140 131 153 125 143 104 109
121 70 100 2 123 140 131 153 125 143 107 112
121 70 100 3 123 140 131 153 125 143 104 108
177 70 50 1 71 85 74 91 71 87 45 86
177 70 50 2 71 85 74 91 71 87 44 90
177 70 50 3 71 85 74 91 71 87 42 90
177 70 75 1 96 116 101 125 96 119 60 119
177 70 75 2 96 116 101 125 96 119 92 127
177 70 75 3 96 116 101 125 96 119 77 120
177 70 100 1 119 145 126 157 120 149 120 146
177 70 100 2 119 145 126 157 120 149 121 149
177 70 100 3 119 145 126 157 120 149 116 144
232 70 50 1 69 88 71 93 69 90 68 90
232 70 50 2 69 88 71 93 69 90 68 89
232 70 50 3 69 88 71 93 69 90 70 91
232 70 75 1 93 120 97 128 93 122 93 122
232 70 75 2 93 120 97 128 93 122 95 126
232 70 75 3 93 120 97 128 93 122 93 126
232 70 100 1 115 150 121 161 115 153 108 161
232 70 100 2 115 150 121 161 115 153 108 164
232 70 100 3 115 150 121 161 115 153 90 155
121 90 50 1 69 80 74 89 70 83 40 98
121 90 50 2 69 80 74 89 70 83 33 104
121 90 50 3 69 80 74 89 70 83 43 92
121 90 75 1 94 109 101 122 96 113 44 97
121 90 75 2 94 109 101 122 96 113 47 94
121 90 75 3 94 109 101 122 96 113 51 98
121 90 100 1 121 138 129 153 123 142 75 132
121 90 100 2 121 138 129 153 123 142 79 134
121 90 100 3 121 138 129 153 123 142 75 128
177 90 50 1 67 83 71 92 68 85 35 95
177 90 50 2 67 83 71 92 68 85 45 116
177 90 50 3 67 83 71 92 68 85 32 98
177 90 75 1 90 113 97 126 91 116 86 103
177 90 75 2 90 113 97 126 91 116 93 133
177 90 75 3 90 113 97 126 91 116 67 128
177 90 100 1 112 140 121 158 113 145 113 138

 

 

 

 

 

 

 

 

 

 

 

 

 

96

 

 

 

 

 

 

predicted measured
condition Rep lump hmljmp h; IlmJE—_ha_vg hlflflVfl ham— hm avg
T... M., F W/m2*K mm/s W/m2*K mm/s W/m2*K mm/s W/mzK mm/s
177 90 100 2 113 140 121 158 113 145 112 180
177 90 100 3 114 140 121 158 113 145 107 153
232 90 50 1 66 86 69 94 66 88 63 123
232 90 50 2 66 86 69 94 66 88 66 139
232 90 50 3 66 86 69 94 66 88 51 120
232 90 75 1 90 118 95 128 90 121 101 137
232 90 75 2 90 118 95 128 90 121 98 146
232 90 75 3 90 118 95 128 90 121 101 135
232 90 100 1 111 147 118 161 111 151 115 167
232 90 100 2 111 147 118 161 111 151 113 175
232 90 100 3 111 147 118 161 111 151 112 161

 

 

 

 

 

 

 

 

 

 

 

 

97

 

Table A.3: Predicted and measured convective heat and mass transfer coefficients
from the model based on Martin (1977).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I predicted measured
Condition Rep hm, hm he M h... h... .39, havg hm m
T M F W/rfig‘K mm/s W/m *K mm/s W/m *K mm/s W/m K mm/s

121 0 50 1 7O 74 79 85 72 77 70
121 0 50 2 70 74 79 85 72 77 69
121 0 50 3 70 74 79 85 72 77 71
121 0 75 1 91 97 108 117 95 103 96
121 O 75 2 91 97 108 117 95 103 93
121 0 75 3 91 97 108 117 95 103 93
121 O 100 1 109 117 134 146 115 126 118
121 0 100 2 109 117 134 146 115 126 116
121 0 100 3 109 117 134 146 115 126 117
177 0 50 1 68 78 75 88 69 81 71
177 O 50 2 68 78 75 88 69 81 70
177 0 50 3 68 78 75 88 69 81 72
177 0 75 1 88 102 102 121 91 108 91
177 0 75 2 88 102 102 121 91 108 90
177 0 75 3 88 102 102 121 91 108 90
177 O 100 1 105 124 127 152 110 132 111
177 0 100 2 105 124 127 152 110 132 110
177 0 100 3 105 124 127 152 110 132 110
232 0 50 1 66 82 72 91 67 85 63
232 0 50 2 66 82 72 91 67 85 63
232 0 50 3 66 82 72 91 67 85 63
232 O 75 1 85 108 97 125 87 113 86
232 0 75 2 85 1 08 97 125 87 1 1 3 86
232 0 75 3 85 1 08 97 125 87 1 13 87
232 0 100 1 102 131 121 157 106 138 108
232 0 100 2 102 131 121 157 106 138 105
232 0 100 3 102 131 121 157 106 138 106
121 30 50 1 72 76 80 87 73 79 51 40
121 30 50 2 72 76 8O 87 73 79 50 38
121 30 50 3 72 76 8O 87 73 79 51 50
121 30 75 1 93 100 109 119 97 105 90 114
121 30 75 2 93 100 109 119 97 105 88 118
121 30 75 3 93 100 109 119 97 105 87 139
121 30 100 1 112 121 136 150 118 129 114 116
121 30 100 2 112 121 136 150 118 129 112 124
121 30 100 3 112 121 136 150 118 129 112 154
177 30 50 1 7O 80 80 89 71 83 51 75
177 30 50 2 7O 80 80 89 71 83 50 76
177 30 75 1 90 105 104 123 93 110 82 105

 

 

 

98

 

 

 

 

 

 

 

predicted measured
condition Rep hm. hm he m hay hm avg hay hm avg
T M F W/niK mm/s W/m2*K mm/s W/mg*K mm/s W/mg mm/s

177 30 75 2 90 1 05 104 123 93 1 1 0 82 1 10
1 77 30 75 3 90 1 05 104 123 93 1 1 0 82 101
177 30 100 1 109 127 130 154 113 135 108 130
177 30 100 2 109 127 130 154 113 135 108 132
177 30 100 3 109 127 130 154 113 135 107 129
232 30 50 1 68 84 73 92 69 86 64 89
232 30 50 2 68 84 73 92 69 86 62 92
232 30 50 3 68 84 73 92 69 86 64 91
232 30 75 1 88 1 1 O 100 126 90 1 14 87 1 18
232 30 75 2 88 1 10 100 126 90 1 14 88 1 13
232 30 75 3 88 1 1 0 1 00 1 26 90 1 1 4 88 1 14
232 30 1 00 1 1 06 1 33 124 1 59 1 09 140 1 08 125
232 30 1 00 2 106 1 33 124 1 59 1 09 140 107 1 18
232 30 1 00 3 1 06 1 33 124 1 59 1 09 140 1 06 1 34
1 21 50 50 1 71 77 79 88 73 80 57 69
1 21 50 50 2 71 77 79 88 73 80 51 73
1 21 50 50 3 71 77 79 88 73 80 53 76
121 50 75 1 93 101 108 121 96 107 96 96
121 50 75 2 93 101 108 121 96 107 94 96
121 50 75 3 93 101 108 121 96 107 93 93
121 50 100 1 111 122 135 151 117 131 120 111
121 50 100 2 111 122 135 151 117 131 119 113
121 50 100 3 111 122 135 151 117 131 118 109
1 77 50 50 1 70 81 76 90 71 84 39 96
1 77 50 50 2 70 81 76 90 71 84 43 94
1 77 50 50 3 70 81 76 90 71 84 44 96
1 77 50 75 1 90 1 06 103 124 93 1 1 1 88 1 17
1 77 50 75 2 90 1 06 1 03 1 24 93 1 1 1 86 1 14
1 77 50 75 3 90 1 06 1 03 124 93 1 1 1 87 1 12
177 50 100 1 109 128 129 156 113 136 104 135
177 50 100 2 109 128 129 156 113 136 101 134
177 50 100 3 109 128 129 156 113 136 109 133
232 50 50 1 68 85 73 93 69 87 68 93
232 50 50 2 68 85 73 93 69 87 67 98
232 50 50 3 68 85 73 93 69 87 68 90
232 50 75 1 88 1 1 1 99 127 90 1 15 97 127
232 50 75 2 88 1 1 1 99 127 90 1 15 95 124
232 50 75 3 88 1 1 1 99 127 90 1 15 97 1 14
232 50 100 1 106 134 124 160 109 141 110 145
232 50 100 2 106 134 124 160 109 141 109 147
232 50 1 00 3 106 1 34 124 1 60 1 09 141 1 08 146
1 21 70 50 1 70 78 76 89 71 81 41 65

 

 

 

 

 

 

 

 

 

 

 

 

 

99

 

 

 

 

 

 

 

predicted measured
Condition Rep 11...; 'hm he 11.... hgéLhm hiyag "m
T M F W/m*K mm/s W/m2*Kmm/sW/m*K mm/s W/mem/s
121 70 so 2 7o 73 76 39 71 31 4o 67
121 70 so 3 7o 73 76 39 71 31 46 65
121 70 75 1 90 102 105 122 94 103 64 93
121 70 75 2 90 102 105 122 94 103 67 94
121 70 75 3 90 102 105 122 94 103 63 92
121 70 100 1 109 123 131 153 115 132 104 109
121 70 100 2 109 123 131 153 115 132 107 112
121 70 100 3 109 123 131 153 115 132 104 103
177 70 so 1 63 32 74 91 69 34 45 36
177 70 so 2 63 32 74 91 69 34 44 90
177 70 so 3 63 32 74 91 69 34 42 90
177 70 75 1 39 107 101 125 91 112 60 119
177 70 75 2 39 107 101 125 91 112 92 127
177 70 75 3 39 107 101 125 91 112 77 120
177 70 100 1 107 129 126 157 111 137 120 146
177 70 100 2 107 129 126 157 111 137 121 149
177 70 100 3 107 129 126 157 111 137 116 144
232 70 so 1 67 35 71 93 67 37 63 90
232 70 so 2 67 3s 71 93 67 37 63 39
232 70 so 3 67 35 71 93 67 37 7o 91
232 70 75 1 37 111 97 123 33 116 93 122
232 70 75 2 37 111 97 123 33 116 95 126
232 70 75 3 37 111 97 123 33 116 93 126
232 70 100 1 104 135 121 161 107 142 103 161
232 70 100 2 104 135 121 161 107 142 103 164
232 70 100 3 104 135 121 161 107 142 90 155
121 90 so 1 63 79 74 39 69 32 4o 93
121 90 so 2 63 79 74 39 69 32 33 104
121 90 so 3 63 79 74 39 69 32 43 92
121 90 75 1 33 103 101 122 92 103 44 97
121 90 75 2 33 103 101 122 92 103 47 94
121 90 75 3 33 103 101 122 92 103 51 93
121 90 100 1 103 124 129 153 114 132 75 132
121 90 100 2 103 124 129 153 114 132 79 134
121 90 100 3 103 124 129 153 114 132 75 123
177 90 so 1 67 32 71 92 67 35 35 95
177 90 so 2 67 32 71 92 67 35 45 116
177 90 so 3 67 32 71 92 67 35 32 93
177 90 75 1 36 107 97 126 39 113 36 103
177 90 75 2 36 107 97 126 39 113 93 133
177 90 75 3 36 107 97 126 39 113 67 123
177 90 100 1 104 130 121 153 103 133 113 133

 

 

 

 

 

 

 

 

 

 

 

 

 

100

 

 

 

 

 

 

 

predicted measured
Condition Rep 11.... 11...... 11E 11...,E 11..v 11...... 11..v 11......
T M F W/mip‘iK mm/s W/m2*K mm/s W139— mm/s W/mgK mm/s
177 90 100 2 104 130 121 153 103 133 112 130
177 90 100 3 104 130 121 153 103 133 107 153
232 90 so 1 65 36 69 94 66 33 63 123
232 90 so 2 65 36 69 94 66 33 66 139
232 90 so 3 65 36 69 94 66 33 51 120
232 90 75 1 35 112 95 123 37 117 101 137
232 90 75 2 35 112 95 123 37 117 93 146
232 90 75 3 35 112 95 123 37 117 101 135
232 90 100 1 103 135 113 161 106 143 115 167
232 90 100 2 103 135 113 161 106 143 113 175
232 90 100 3 103 135 113 161 106 143 112 161

 

 

 

 

 

 

 

 

 

 

 

 

 

101

APPENDIX B

Block Temperature Data

102

 

T121 MVOFSO

 

 

 

 

 

 

 

 

 

140

1201— --_/_ 7-‘4—7 -'---*' -7‘ _fi
stood/T r~— -—--- _2- ,2__ —————-~;-—--————« '----rep1!
25309“: ~— ~—~A—--— —~--~-<§----rep21
g 60 2, ,- 2-, - —-— —*‘---- -:1---rep3]
"' 4Ofihr 77 7' 77’ "“7 ” ' ‘7"1'ILT'99§.‘

20.--, ,fii,- __f (2-, 2.774-V,7_.-V 4.,

 

 

 

lit-1173}?!
1 !—---rep2
’ 1—---rep3

’:---'ga_s_1

 

 

 

 

o ,
'\ Nooweorsoo \sewe
w‘w*$$@o°o‘o9efi&@§@@@§ 13>

time (s)

T121 MVOF1OO

 

140
120 ,1 - “:r; — "‘—":.‘ " “" ' ____ _._.,___-.
100 / ._ _ - -- ~-—— ”fl- —- —— i- — "' I'GP1

.. , , 3 -. . - 1—---rep2‘
i----mp3i

I

L' ,' 999-1

 

 

T (°C)

 

 

 

 

0') O N v v- (D In N O) (D 9 O i\ v v- a) to N

N O) O N s: ID i\ O) O N (D N 0) v- N § (D

1- v- v- v- F v- N N N N N N (O (‘0 (‘0
tlme(s)

Figure B.lz Data collected for the Too=121°C, Mv=0% set point runs.

103

T177MVOF50

 

 

 

 

 

 

 

 

 

 

 

 

 

1:765
1--—-mp2
1----mp3
*L:_"_"_9£S._g
assassssessssszggs
v-v-v-r-v-v-NNNNNNNCO (0
time(s)
T177MVOF75
200
'::__}—e'b1"‘:
:----rep2
[—---mp3
'__.'._' 'gas 1‘
0 ,.
111N030 ONSv-SION 8000K?
coo—comixoo & Inhw «gusts
1- 1- v- v- '- 1- N N N N 0') (O 0') CO
tlme(s)
T177MVOF1OO
200

 

 

 

 

 

 

 

time (s)

Figure B.2: Data collected for the T..=l77°C, M.,=0% set point runs.

104

T232 MVOFSO

 

 

 

 

 

 

 

 

 

 

_ 1

 

 

 

 

 

 

 

 

 

 

 

[\OC‘OCO Niooov-grxococoo)
NVIOCDNmov-(O mNQmO
tlme(s)

Figure B.3: Data collected for the T..=232°C, Mv=0% set point runs.

105

 

'---rei)1_;
1----rep2
J I—---rop3
‘- : t; 989 J
1 _ i311 1'5
J—---rep2
__ "----rep3
J7: 3.1.923;

T121MV30F50

 

140
120
100 ,
mfi
m”
40 .22 _ __ , W, 2, _ ._ __u.’ , 7 7
204 -2 e. a _9 99—1 ~— -~ _2 _-
o

\ bt
&@®&§$&&§@&§9$$@$

 

 

 

T (°C)

 

 

 

tlme (s)

T121Mv30 F75

 

140

 

 

T (°C)

 

 

 

mi

 

 

140
120 15* M '7 '
100 5777/.” ’ 77 fl 7 w 7 7J
60
40 1
20 *

 

 

 

T (°C)

 

 

 

 

0 Hi iii'illli'ili'il 'T T‘HJ' 1 V1111 V111 I N ' 1111' 111'111'1 ' I" WNW-111 Iii 1111111111'11111 11'11111111 1
9®$@@@¢&&§&&$&§§$
tlme (s)

Figure B.4: Data collected for the T..=121°C, Mv=30% set point runs.

106

VJEH
i-——rep1l
|----rep2

22192.31

------ gas 1
J—--rep1
1----rep2;

J
T- - 7 192%.?

 

 

T177MV3OF50

 

 

 

 

1-—-rep1
11771923

 

 

 

 

 

103
106;
112

 

 

 

 

 

 

 

 

 

 

V v to In N N 00
time(s)
T177 MV3OF75
200
150 — _.------...: """ 2' ~44 — — 4 — — _ _ — — — — 2.135;“;
amou’v 7 , ——79~# , 4 i 7 v——*——- _ 4 §---rep1
1.. l-H-repz
50 ~~ —— # * + _ -+---— —-——-——--—-~—--~ 1::"1993;
’\ ’1‘" (13‘ (13’ 03° 19’ 19 <69 65 «94‘ 15‘" e" 99.63.51
time(s)
T177 MV3OF1OO
200
150 ' ,3. """"" ~57 7‘ —— V ——* 15‘}; 1
I ___
8100 1.27 A v ___ k;_ ,_i 4‘ rep1
I. i----rep2
50 7/ ”777* ' '77“ — 1_;j_-rep 3,;
O 111TTTT<Y1THV 111 TH 17H :1’11311.11'1111’11711711 1‘111 1T11¥11111 1"1111‘11’11'11 T1 T Y

 

 

 

1P 65‘ 61‘" 5°) «‘1’ «9 e9 9‘5 .99 G" 3‘ 41> {13> (5‘0 \1‘3' \‘9

time (s)
Figure 8.5: Data collected for the T..=l77°C, Mv=30% set point runs.

107

250

T 232 M., 30 F 50

 

200 J ~ -

150
100
50 -

T (°C)

 

 

 

 

 

250

6154159 \‘L‘b <oo
ovoo'\e°\o¢\w<b°‘¢\o

time (s)

T 232 M. 30 F 75

1‘11l11'l‘ll ‘

3“ <18

 

200 -
1 50
100
50 1

T (°C)

 

O

 

 

 

 

 

3‘" 99

250

111 ‘1 .V1'1I1'11'1111'1‘11111'1'11111111111linii‘l'ifi‘iri'1-11 1’111111111111111111'1d

\ bk \
(‘0 (7’ (51' Kb?) \‘3 ~96) (‘6 ’3’; 9% '199 “139 ‘1'? ‘b '1,

time (s)

T 232 Mv 30 F 100

111‘ 1‘111111.l1 1 1'

 

T (°C)

 

 

 

 

 

 

Figure 8.6: Data collected for the T..=232°C, Mv=30% set point runs.

1"1'111‘111
O

F
|\

P

‘0

P

time (s)

204
21s

«30) N00
GOV (no:

1163
127}

105?

108

226 f

237
243 g

[_ _ -_..____

1 """ gas 1

J---rep1}
i---—rep2'

!
1: :3231'

1---?-'gas
T—--rep1
1

‘----rep2
1:;--rep§_J

“1
1
1
|

 

 

140

T (°C)

140.0
120.0
100.0
80.0
60.0
40.0 1
20.0
0.0

T (°C)

140

120 1
100 A»

T (°C)

Figure 8.7: Data collected for the T...=121°C, Mv=50% set point runs.

T121MV50F50

 

 

 

 

 

 

 

 

 

 

 

 

 

time (s)

T121Mv50 F100

143

126;
137;

1152

 

 

 

 

 

78 7

9001-

PP

time (s)

109

0301-

144:
155 f
166 g
177 i
133

133

 

[__ _ ___“

1 ------ gas

' 1

1

1 I---rep11
* J----rep2
’ L:f:_'39.1!

. C.:;:fg;§”1

1
‘—--rep1

!—---rep2

1
1:_;.—_'.'_°B_§.1

*— ______ — —'1

. ----- as 1
I— - - rep 1 '
1- - - —rep 2

L- 17199.3.

T177MV50F50

 

 

 

 

 

YT" T‘.TTTT' V1717 1*17 r77*. 1777.11 ' 1.1'1 1711-81 . 1111

"TWINW'TT ri'rfl'n‘ 1

1‘39 63° 49 e‘" 4" «9’ <5" 9‘” <39 .99 .35 .39

 

 

 

 

 

 

 

 

time (s)
T 177 M. so F 75

200

150 7 _,.--"""""m"_°m: """"""" 7.":"7 ' '7 7 7 ' _ 7
8100 1' — —#—~- ——— ——————-————
P 1—

50 .7 77 — 59' 5 - - —

o . . . , s .9. 4 1 5 w

 

 

 

 

 

time(s)
T177 M., 50F100

200

150 3"": ............. : """""""""" 7" """""""""" 7 """" 7' 7
6 "
L100 7 7 # — ——— ——
1- 1'

50 1/ _ ~ _ _ 77.

0 T’T'T'T‘YTTY I1 *1. 111 .-'T‘ 11*1 ”'1 W77 7" 77 it"r'TT' *1 Y

\" 13° q? is" '59 1? <65 69 6‘ 4‘“ 21" 23’ <3” .9" .99 (3’

time (s)

 

Figure 13.8: Data collected for the T..=l77°C, Mv=50% set point runs.

110

h 7777 "1

J ------ gas—-
'----rep1

)---—rep2‘

1: #:7197937

1 ------ gas

‘J---rep1J

:----rep2
17:973.!

-_—.—_f 7 ‘J

T 232 M., 50 F 50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11 11 11 11 11 11 1111' 11.11I1’1111 11 l '1111 I 11 -! 11111"11 11111111'1111 111111 111l11 I11111'1111'l‘11111!111 11111111 1.1111'l1111111111111‘11 1Il11llll 111l11! 1' 1 1111

time (s)

T 232 M., 50 F 100

 

 

7. .___77_.]

1' ------ gas
7“ —"' \---rep1J

__7, ,7- -7 ;---—rep2

J—-f-rep3.

__—i. . _

 

 

 

 

1 11 - 1 11 1111‘1111‘1 11‘111111 1111 1‘11 111111 11 11-1‘1 11 11-1 1111'11‘1111 1 !l 1 11 1' 1 :11} 1

N" (19’ (5" 111* 6‘” 69 4" 93’ 9" (9 411‘ .{9’ \1‘3’ (6" .31" (\Q’ (6‘

time (s)
Figure B.9: Data collected for the T..=232°C, Mv=50% set point runs.

111

T121MV7OF50

 

mo

 

 

 

T (°C)

 

 

 

 

mo
907-777:77,77777777777777_777777,
m0~ ~
30
60»
401
2o

0 111 1'11111111111111111111111111111117111'11-111111‘1117111 1111111 V17111111111111111111111111'111T111I111
‘ 94°fi’819S86’4<§4§4bé’&W§§9$5

time (s)

 

 

rem

 

 

 

 

T121 1111.70 mm

140

, 111-:--gas

 

1
1—--rep1
:---—rep2
1_

1 ------ 9881

 

mm77777777777777 4;

 

100 .7 4.7,,-r7"'"‘7 ,7 7777. 7. _7- 7-- _7 77.7 .77 _ __7_
80 :1", , ,7 7 , 7 7 7 7 _7 _ 7 7 _ 7 7_ .7 7 7-
60 5* * 7 7777 7 ~ ~ 7 7 — 7 5 — 77 *‘
40 7 3 7 7 7 7 7* 7 7 7 77* 7 ,7 9 7

20 4. 7 7 7 7, 7 7 7. 7 7 ,7 77 .7 77.. . 7 7 .7

T (°C)

 

 

 

O 1.1TT.1‘1*1_T.1111_'1111.m1‘11-711111‘.11111= .11111111'1 1711'1‘11'11-11 1‘1'1'1111111

time (s)

Figure 8.10: Data collected for the T..=121°C, Mv=70% set point runs.

112

T177MV70F50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

200
150 47 “7.. ------------- -7- -----"""7"":"': ...... 7 7 : """"""" '7' """""""""" : __. 'iéa—S ‘1
A ---re 1
8 100 7 f 77 7 - 7 7 1 p 1
,_ ----rep2
50 7 7 7 7 * 7 **-— * * — fi— fi - * 1‘__"T'7"1?.B_3l
‘6 ’3’ (13’ Q9 '5" 13’ 63° 6‘ '0‘" ’\" «9’ <6” <5” 99,9665
tlme(s)
T177 MV7OF75
200.0
150.0 7771*7":“7 7 Z 1; 7:33;}
A i 1
8100.0 —— 7 77 * -~ -7— ‘fi rep2
._ ----rep3
50.0 7 7 - 7 — 7 —— —— 7 7 .---_-_-'gas#_‘
00 W 7 W7
\%¢3‘L€bb‘bo’\ «000,00
~%\%\9179119®wwr£519w¢03’r§’q900$
tlme(s)
T177 My 70F100
200
150 1),...7"””"”""""“”3".””7“mum."""mflnfum-m 7" “7 . . :‘T'gaé
8100 ' 7 777 “um“
._ - --rep2
50 w -— * ~~~ ----- :- --reggj
o Y, 1177 ,7
\b‘ q} 03’ q," 113' 9 <06 65 0° ’0 Q1" 9" 09’No‘°.\\q’\\9(13’

tlme (s)
Figure 8.11: Data collected for the T..=l77°C, Mv=70% set point runs

113

T 232 Mv 70 F 50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

250
200 — —-- — ~-—f 777 77
'. 1 """ gas
8150 ".5 7 ‘ ""‘— ___ 7_ fl“”““ :---rep1
L l
l- 100 ’77_ ~—- + » -""f992
!—-—-re 3
50 , 7 7 7 7 - __ 7 ._ P7
0
time(s)
T 232 My 70 F 75
250
200 W _ ._ 77- 77-77777
. t """ gas 1
@150 ++~~ 7 ~ ~~ w !---rep1
2., ; 1
.— 100 _ —.—-————~ — — -777 - :----rep2
.- —-re 3
50 77 -7 ___77_ 1 _____ p.
" «‘7’ 0‘? 6.1" b‘é" <0" 6‘ «‘5 09 \0° 5‘ @1455 Nb? (6" .3940
time (s)
T 232 M" 70 F 100
250
200 ‘14 r” — — _jfiigas
A 150 7' - — 7 7 -—~ _7777 - i---re 1
8 -' f4 1 p
|_ 100 7’7 _ ___ 7 _ ___. :_ -—rep2
i— —-re 3
50 77 77 77 7 7- 777 1 p77
0 11'1 1111111 ' TT-" 111 11111 1T1-11‘ V 1. 1. 1111 11111—717'7111111111

 

 

o. \x ,3: (go (gt, ,9 he {0% go 6 «a 0" .23, (gag/$0

time (s)
Figure 8.12: Data collected for the Tw=232°C, Mv=70% set point runs.

114

T121Mv90 F50

 

 

 

 

140
120 3————— — ___—777777777 7.. _ .. _7 77
100 ;_;.;,:..~.-1~1"4~f~7-7—_-777-__ 77 7

E 80»-—~-~ 7777777 7 7- _______

,_ 6077 ~ _77 — 77 7 _ 7 7777
40555— iiiiiiiiiiiiiiii
20. 7-... 77 7 ________

Q ’\ bl ’\ Q I\ b‘ \
9x x q, 0.: 9%»?39‘Vbqé’4595 % 0,9,‘0‘?’
time(s)
T121Mv90F75
140
120 if 1‘ 7 ’ 74777777_7_7____\-:

’\100‘Pd;’7t - 7 — 7777 ___—7 77777.7.

8 307777 f — —#7—777 _ 7-7-777777

,_ 60$ 7 — —7——7-7_ ___77777777
405- — 7 - 777777
201-!“ 7.- — - ————————

o . ,, ,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
Q," ‘9 «'3‘ v?) 635 6‘ «:9 ’\’\ 91° <5” 0" .99 x .333 {5‘5 ,3" 4’9
time(s)
T121 My 90 F 100
140
1203 7 7 7‘7 77 7 7 77-777-
100 77, 7YTJ7 7 I 7’ fl _ F7; 4“ _7

6 ' 7 7 ~ 7 7 _ 77 7

L J

._ 7 7 7_

o 55"”55” 35‘. C535 ’35 333 {a "
" " °° °’ 2 :: z: E! .— 3
time(s)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

;---rep1
1----rep2

1

: T711937

i-n-repz

, 1—__-_--reps1

1""~"93$ .
'1 1
---rep1

i-----rep2’
9.7:: "69 3-!

Figure B.l3: Data collected for the T.°=121°C, Mv=90% set point runs.

115

T177Mv90F50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

200
. 7 777 ...-/"’“... . 7 77 771
150 '.‘f’—\‘\~‘ .v'N 1;" \‘1 /
8 V‘N/
L 100
.—
50 1 - 7 3 7 ********
'3’ r1,“ r5" b9 5‘? <56 125‘" «'1' 125° 6" <2? .9“ 40’ .39 {13’ (5% \bP‘
tlme(s)
T177 MV9OF75
200
150 _,....-. --------- «‘77 _ -x “-9... """" :7: .......... 7 77
a 100 -35" fig 7 *Jfl "7 777
I-
50,7777 7 7_ 777 77777
0 1 1 ‘1Y '11'T 1r1fi111fi111'1'1 111T T'TWTT T T 111'1 1.1‘111171'11'1'111111111‘11111‘1 T‘1'
’4‘ ’3’ "15° '5'” '59 b9 <55 «9 6‘ ’\°‘ Qa" <2? 9‘3 New .99 (3’
time(s)
T177 My 90F100
200
”M3“
13‘3” 3535 35‘ o L 3 35‘ 35"
1~ °° °° 0 e : z: 91. :2 :2
time(s)

---—rep2
1i'7f'rep371

 

111172571

1---—rep21
17.5551

1 .
, Lit-"gas .1

1:51:35: '1
1—--rep1:
1—3—3552.

.—'.':."9793_1

Figure 8.14: Data collected for the To.=l77°C, Mv=90% set point runs.

116

T 232 My 90 F 50

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

250
200 7 7 7.'.-_'....29"‘7"-""~"‘37-"'7':".:.-:.“.:“._””, 7 7 7 7 7 7 _ "V 7 7. 7 7
1 ------ gas 1
8150 “ f' 7 " ‘ * " ‘ " " " ""‘—" “ 7 1---rep1
L ,' . 1
.— 100 7+! 7 ~ ~— -- -- 7-7 ‘———"—~—»-v ~1 '-°--rep2'
1_._. 1
5o 77 7- 77777 777777 _ -7197'37371
9’ 4° (13’ Q9 15% $5 3,0 6‘ 0‘" ’\" «9’ <25" <5" 09\ob\\%\119
tlme(s)
T 232 My 90 F 75
250
200 7. 7. ;7_.__.7.....-..o-----_,....o---'_"""°_'_7'7"""°7""at”: ..... .7". "77' -. _- ___1 1 77 7|
------- gas 1
6 150 7 ‘7 5L7 ' i "7' 7 7 .. “—77 V—¥ ” "_— __‘— __ w" '77— .— ——rep1 1
L .' ‘
+— 100 5.3.: - -7 ~ 7_ - 77 g 7_ , 1----repZ1
50 7/ 7 _7 7 .7 77 .7 7 _. 7 7 . 1—-_—-rep‘§‘1
bt ’\ N
’\ x 0, 03’0") 191905" 65/04" 0‘9 $69.5”
tlme(s)
T 232 My 90 F 100
250
200 777..7....-""""""""""“"" ...... 7 ... . 7 . - . __ 7
1 ------ gas 1
6150 7’! ""’ ” ’ 777-77___ __77 * ” ' ---rep11
L I ~ ' 1
1. 100 7 7.77 + 77 77 1----rep21
50 7f 77 _7 77--. 1:- 19979.1

Figure B.lS: Data collected for the T55=232°C, Mv=90% set point runs.

117

   

11111111111111111111