Modelling Drying Processes A Reaction Engineering Approach

more information - www.cambridge.
org/9781107012103
Modelling Drying Processes

This comprehensive summary of the state-of-the-art and the ideas behind the reaction
engineering approach (REA) to drying processes is an ideal resource for researchers,
academics and industry practitioners.
Starting with the formulation, modelling and applications of the lumped-REA, it
goes on to detail the use of the REA to describe local evaporation and condensation,
and its coupling with equations of conservation of heat and mass transfer, called the
spatial-REA, to model non-equilibrium multiphase drying. Finally, it summarises other
established drying models, discussing their features, limitations and comparisons with
the REA.
Application examples featured throughout help fine-tune the models and implement
them for process design, and the evaluation of existing drying processes and product
quality during drying. Further uses of the principles of REA are demonstrated, including
computational fluid dynamics-based modelling, and further expanded to model other
simultaneous heat and mass transfer processes.
Xiao Dong Chen is currently the 1000-talent Chair Professor of Chemical Engineering at
Xiamen University in China, and the Head of Department of Chemical and Biochemical Engineering. He held previously Chair Professorships of Chemical Engineering at
Auckland University, New Zealand, and Monash University, Australia, respectively from
2001 to 2010. He is now a fractional Professor of Chemical Engineering and the CoDirector of the Biotechnology and Food Engineering Research Laboratory at Monash
University, Australia. He is an Elected Fellow of Royal Society of NZ, Australian
Academy of Technological Sciences and Engineering, and IChemE.
Aditya Putranto holds a BE of Chemical Engineering from Bandung Institute of Technology, Indonesia and a Master of Food Engineering from University of New South Wales,
Australia. He has a Ph.D. in Chemical Engineering from Monash University, Australia.
He has worked in Indonesia as lecturer in Parahyangan Catholic University. His research
area is heat and mass transfer. He has published a dozen journal papers in peer-reviewed
hard-core chemical engineering journals.
The Reaction Engineering Approach (REA), which captures basic drying physics, is
a simple yet effective mathematical model for practical applications of diverse drying
processes. The intrinsic fingerprint of the drying phenomena can, in principle, be
obtained through just one accurate drying experiment. The REA is easy to use with
the guidance of featured application examples given in this book. This book is highly
recommended for both academics and industry practitioners involved in any aspect of
thermal drying.
Zhanyong Li,
Tianjin University of Science and Technology,
China
An interesting book on a novel approach to mathematical modelling of an important
process. Modelling Drying Processes: A Reaction Engineering Approach is the first
attempt to summarize the REA to modelling in a single comprehensive reference source.
Sakamon Devahastin,
King Mongkuts University of Technology Thonburi,
Thailand

A Reaction Engineering Approach
XIAO DONG CHEN
Monash University, Australia
ADITY A PUTRANTO
Monash University, Australia
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town,

Singapore, Sao Paulo, Delhi, Mexico City
Cambridge University Press
The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9781107012103

C
Xiao Dong Chen and Aditya Putranto 2013
This publication is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2013
Printed and bound in the United Kingdom by the MPG Books Group
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data
Chen, Xiao Dong.
Modelling drying processes : a reaction engineering approach / Xiao Dong Chen,
Monash University, Australia, Aditya Putranto, Monash University, Australia.
pages cm
Includes bibliographical references and index.
ISBN 978-1-107-01210-3 (hardback)
1. Drying. 2. Food Drying. 3. Porous materials Drying. 4. Polymers Curing.
5. Lumber Drying. I. Putranto, Aditya. II. Title.
TP363.C528 2013
2013003983
664 .0284 dc23
ISBN 978-1-107-01210-3 Hardback
Cambridge University Press has no responsibility for the persistence or

accuracy of URLs for external or third-party internet websites referred to
in this publication, and does not guarantee that any content on such
websites is, or will remain, accurate or appropriate.
Contents
List of figures
List of tables
Preface
Historical background
1
page ix
xxvi
xxvii
xxx
Introduction
1.1
1.2
Practical background
A microstructural discussion of the phenomena of drying moist,
porous materials
1.3 The REA to modelling drying
1.3.1 The relevant classical knowledge of physical chemistry
1.3.2 General modelling approaches
1.3.3 Outline of the REA
1.4 Summary
References
6
15
15
17
18
29
30
Reaction engineering approach I: Lumped-REA (L-REA)
34
2.1
2.2
2.3
2.4
34
36
40
43
43
47
50
2.5
The REA formulation

Determination of REA model parameters
Coupling the momentum, heat and mass balances
Mass or heat transfer limiting
2.4.1 Biot number analysis
2.4.2 Lewis number analysis
2.4.3 Combination of Biot and Lewis numbers
Convective drying of particulates or thin layer products modelled
using the L-REA
2.5.1 Mathematical modelling of convective drying of droplets of
whey protein concentrate (WPC) using the L-REA
2.5.2 Mathematical modelling of convective drying of a mixture of
polymer solutions using the L-REA
2.5.3 Results of modelling convective drying of droplets of WPC
using the L-REA
50
51
53
55
vi
Contents
2.5.4
Results of modelling convective drying of a thin layer of a mixture

of polymer solutions using the L-REA
2.6 Convective drying of thick samples modelled using the L-REA
2.6.1 Formulation of the L-REA for convective drying of thick samples
2.6.2 Prediction of surface sample temperature
2.6.3 Modelling convective drying thick samples of mango tissues
using the L-REA
2.6.4 Results of convective drying thick samples of mango tissues
using the L-REA
2.7 The intermittent drying of food materials modelled using the L-REA
2.7.1 Mathematical modelling of intermittent drying of food materials
using the L-REA
2.7.2 The results of modelling of intermittent drying of food materials
using the L-REA
2.7.3 Analysis of surface temperature, surface relative humidity,
saturated and surface vapour concentration during
intermittent drying
2.8 The intermittent drying of non-food materials under time-varying
temperature and humidity modelled using the L-REA
2.8.1 Mathematical modelling using the L-REA
2.8.2 Results of intermittent drying under time-varying temperature
and humidity modelled using the L-REA
2.9 The heating of wood under linearly increased gas temperature modelled
using the L-REA
2.9.1 Mathematical modelling using the L-REA
2.9.2 Results of modelling wood heating under linearly increased gas
temperatures using the L-REA
2.10 The baking of cake modelled using the L-REA
2.10.1 Mathematical modelling of the baking of cake using
the L-REA
2.10.2 Results of modelling of the baking of cake using
the L-REA
2.11 The infrared-heat drying of a mixture of polymer solutions modelled
using the L-REA
2.11.1 Mathematical modelling of the infrared-heat drying of a mixture
of polymer solutions using the L-REA
2.11.2 The results of mathematical modelling of infrared-heat drying of a
mixture of polymer solutions using the L-REA
2.12 The intermittent drying of a mixture of polymer solutions under
time-varying infrared-heat intensity modelled using the L-REA
2.12.1 Mathematical modelling of the intermittent drying of a mixture of
polymer solutions under time-varying infrared-heat intensity using
the L-REA
57
61
61
63
64
66
69
69
69
73
80
81
82
88
89
91
95
96
97
100
101
103
104
105
Contents
2.12.2 Results of modelling the intermittent drying of a mixture of

polymer solutions under time-varying infrared heat intensity using
the L-REA
2.13 Summary
References
106
116
117
Reaction engineering approach II: Spatial-REA (S-REA)
121
3.1
3.2
3.3
121
125
127
The S-REA formulation

Determination of the S-REA parameters
The S-REA for convective drying
3.3.1 Mathematical modelling of convective drying of mango tissues
using the S-REA
3.3.2 Mathematical modelling of convective drying of potato tissues
using the S-REA
3.3.3 Results of modelling of convective drying of mango tissues using
the S-REA
3.3.4 Results of modelling of convective drying of potato tissues using
the S-REA
3.4 The S-REA for intermittent drying
3.4.1 The mathematical modelling of intermittent drying
using the S-REA
3.4.2 Results of modelling intermittent drying using the S-REA
3.5 The S-REA to wood heating under a constant heating rate
3.5.1 The mathematical modelling of wood heating using the S-REA
3.5.2 The results of modelling wood heating using the S-REA
3.6 The S-REA for the baking of bread
3.6.1 Mathematical modelling of the baking of bread using the S-REA
3.6.2 The results of modelling of the baking of bread using the S-REA
3.7 Summary
References
4
vii
Comparisons of the REA with Fickian-type drying theories, Luikovs and

Whitakers approaches
4.1
4.2
4.3
Model formulation
4.1.1 Cranks effective diffusion
4.1.2 The formulation of effective diffusivity to represent
complex drying mechanisms
4.1.3 Several diffusion-based models
Boundary conditions controversies
A diffusion-based model with local evaporation rate
4.3.1 Problems in determining the local evaporation rate
4.3.2 The equilibrium and non-equilibrium multiphase
drying models
128
130
133
138
141
141
142
148
148
151
158
158
160
164
165
169
169
171
172
173
177
179
180
182
viii
Contents
4.4
Comparison of the diffusion-based model and the L-REA on

convective drying
4.5 Comparison of the diffusion-based model and the S-REA on
convective drying
4.6 Model formulation of Luikovs approach
4.7 Model formulation of Whitakers approach
4.8 Comparison of the L-REA, Luikovs and Whitakers approaches for
modelling heat treatment of wood under constant heating rates
4.9 Comparison of the S-REA, Luikovs and Whitakers approaches for
4.10 Summary
References
Index
185
188
190
195
200
203
206
207
212
Figures
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
Some traditional dried products. (a) Broccoli-steam blanched and air

dried (kindly provided by Ms Xin Jin, Wageningen University, The
Netherlands), (b) air-dried Chinese tea leaves (taken at Xiamen
University laboratory), (c) spray dried aqueous herbal extract (particle
size is about 80 m) (taken at Xiamen University laboratory), (d) timber
stacked for kiln drying (kindly provided by Professor Shusheng Pang
(Canterbury University, New Zealand).
page 2
Chemical structures of some chemicals: (a) 1, caffeic acid; 2, gallic acid;
3, vanillic acid; (b) 1, cellulose; 2, starch; 3, pectin; (c) human insulin.
4
Air drying of a capillary assembly (a bundle) which consists of
identical capillaries (diameter and wall material) a scenario of
symmetrical hot air drying of an infinitely large slab filled with the
capillaries (modified from Chen, 2007); the air flows along both sides of
the symmetrical material.
7
Schematic showing a common scenario of air drying of a moist solid.
9
Packed particulate material.
10
Cellular structures in plant material.
10
(a) Generation of computational domains of corn geometry for the
hybrid mixture theory of corn kernels (adapted from Takhar et al.
(2011)). (b) The simulated results (isosurface plots of corn moisture
content) for a variety of drying conditions. [Reprinted from Journal of
Food Engineering, 106, P.S. Takhar, D.E. Maier, O.H. Campanella and
G. Chen, Hybrid mixture theory based moisture transport and stress
development in corn kernels during drying: Validation and simulation
results, 275282, Copyright (2012), with permission from
Elsevier.]
13
Wood cellular structures employed in pore-network modelling of drying
of wood. [Reprinted from Drying Technology, 29, P. Perre, A review of
modern computational and experimental tools relevant to the field of
drying, 15291541, Copyright (2012), with permission from Taylor &
Francis.]
15
Schematic illustration of the effect of temperature on final liquid water
content (qualitatively derived from Equation 1.3.6).
20
List of figures
(a) Drying flux versus average water content X ; (b) the CDRC
(characteristic drying rate curve). [Reprinted from Chemical
Engineering Science, 9, D.A. van Meel, Adiabatic convection batch
drying with recirculation of air, 3644, Copyright (2012), reprinted with
permission from Elsevier.]
1.11 Saturated water vapour concentration in air under 1 atm (Equation
1.3.21).
1.12 Schematic diagram showing the heat of drying as a function of water
content of a porous solid of concern (when the water content is beyond
the point where the heat of drying becomes the latent heat of pure water
evaporation, the water content may be called free water).
2.1 Equipment setup of convective drying of milk droplets (a) measuring
droplet shrinkage; (b) measuring droplet temperature; (c) measuring
mass change. [Reprinted from Chemical Engineering Science, 66, N. Fu,
M.W. Woo, S.X.Q. Lin et al., 17381747, Copyright (2012), with
permission from Elsevier.] (Adapted from Fu et al. (2011) Chemical
Engineering Science 66, 17381747).
2.2 The deflection of glass filament and a typical standard curve (a)
measuring displacement to measure weight loss; (b) correlation between
the displacement and the weight. [Reprinted from Chemical Engineering
Science, 66, N. Fu, M.W. Woo, S.X.Q. Lin et al., 17381747, Copyright
(2012), with permission from Elsevier.]
2.3 The relative activation energy of convective drying of 20%wt. skim milk
powder at a drying air temperature of 67.5 C, velocity of 0.45 m s1
and humidity of 0.0001 kg H2 O kg dry air1 . [Reprinted from AIChE
Journal, 51, X.D Chen and S.X.Q. Lin, Air drying of milk droplet under
constant and time-dependent conditions, 17901799, Copyright (2012),
with permission from John Wiley & Sons, Inc.]
2.4 Schematic diagram showing the plug-flow spray dryer.
2.5 The schematic diagram showing the parameters for the definition of the
classical Biot number. [Reprinted from Drying Technology, 23, X.D.
Chen, Air drying of food and biological materials Modified Biot and
Lewis number analysis, 22392248, Copyright (2012), with permission
from Taylor & Francis.]
2.6 The schematic diagram showing the parameters for the definition of the
modified Biot number) (ChenBiot number). [Reprinted from Drying
Technology, 23, X.D. Chen, Air drying of food and biological
materials Modified Biot and Lewis number analysis, 22392248,
Copyright (2012), with permission from Taylor & Francis.]
2.7 The relative activation energy of convective drying of WPC at different
drying air temperatures. [Reprinted from Chemical Engineering and
Processing, 46, S.X.Q. Lin and X.D. Chen, The reaction engineering
approach to modelling the cream and whey protein concentrate droplet
drying, 437443, Copyright (2012), with permission from Elsevier.]
1.10
21
26
28
37
38
39
41
44
45
52
List of figures
2.8
2.9
2.10
2.11
2.12
2.13
2.14
The droplet diameter changes during convective drying of WPC.

[Reprinted from Chemical Engineering and Processing, 46, S.X.Q. Lin
and X.D. Chen, The reaction engineering approach to modelling the
cream and whey protein concentrate droplet drying, 437443, Copyright
Heat transfer mechanisms of the convective drying of a mixture of
polymer solutions. [Reprinted from Chemical Engineering and
Processing: Process Intensification, 49, A. Putranto, X.D. Chen and P.A.
Webley, Infrared and convective drying of thin layer of polyvinyl alcohol
(PVA)/glycerol/water mixture The reaction engineering approach
(REA), 348357, Copyright (2012), with permission from Elsevier.]
Normalised activation energy and fitted curve of polyvinyl
alcohol/glycerol/water under convective drying at an air temperature of
35 C and relative humidity of 30%. [Reprinted from Chemical
Engineering and Processing: Process Intensification, 49, A. Putranto,
X.D. Chen and P.A. Webley, Infrared and convective drying of thin layer
of polyvinyl alcohol (PVA)/glycerol/water mixture The reaction
engineering approach (REA), 348357, Copyright (2012), with
The comparison between experimental and model prediction using the
L-REA of convective drying of WPC at drying air temperatures of (a)
67.5 C (b) 87.1 C (c) 106.6 C. [Reprinted from Chemical Engineering
and Processing, 46, S.X.Q. Lin and X.D. Chen, The reaction engineering
approach to modelling the cream and whey protein concentrate droplet
drying, 437443, Copyright (2012), with permission from Elsevier].
Moisture content profile of convective drying at an air temperature of
55 C, air velocity of 2.8 m s1 and air relative humidity of 12%.
[Reprinted from Chemical Engineering and Processing: Process
Intensification, 49, A. Putranto, X.D. Chen and P.A. Webley, Infrared and
convective drying of thin layer of polyvinyl alcohol
Product temperature profile of convective drying at an air temperature of
55 C, air velocity of 2.8 m s1 and air relative humidity of 12%.
Moisture content profile of convective drying at an air temperature of
35 C, air velocity of 1 m s1 and air relative humidity of 30%.
xi
53
54
55
56
57
58
xii
List of figures
2.15
2.16
2.17
2.18
2.19
2.20

The relative activation energy (Ev /Ev,b ) of convective drying of
mango tissues at an air velocity of 4 m s1 , drying air temperature of
55 C, and air humidity of 0.0134 kg H2 O kg dry air1 . [Reprinted from
Drying Technology, 29, A. Putranto, X.D. Chen and P.A. Webley,
Modelling of drying of food materials with thickness of several
centimeters by the reaction engineering approach (REA), 961973,
Copyright (2012), with permission from Taylor & Francis Ltd.]
Moisture content profile of convective mango tissues at air temperatures
of 45, 55, and 65 C (modelled using the L-REA which incorporates the
temperature distribution inside the sample). [Reprinted from Drying
Technology, 29, A. Putranto, X.D. Chen and P.A. Webley, Modelling of
drying of food materials with thickness of several centimeters by the
reaction engineering approach (REA), 961973, Copyright (2012), with
permission from Taylor & Francis Ltd.]
Temperature profile of convective mango tissues at air temperatures of
45, 55, and 65 C (modelled using the L-REA which incorporates the
59
59
60
60
65
66
67
List of figures
2.21
2.22
2.23
2.24
2.25
2.26
2.27
Moisture content profile of convective mango tissues at air temperatures

of 45, 55, and 65 C (modelled using the L-REA without approximation
of temperature distribution inside the sample). [Reprinted from Drying
Temperature profile of convective mango tissues at air temperatures of
45, 55, and 65 C (modelled using the L-REA without approximation of
Moisture content profile of mango tissues during intermittent drying at a
drying air temperature of 45 C and resting at 27 C. [Reprinted from
Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao,
X.D. Chen and P.A. Webley, Intermittent drying of mango tissues:
Implementation of the reaction engineering approach, 10891098,
Copyright (2012), with permission from the American Chemical
Society.]
Temperature profile of mango tissues during intermittent drying at a
Society.]
drying air temperature of 55 C and resting at 27 C [Reprinted from
Society.]
drying air temperature of 55 C and resting at 27 C [Reprinted from
Society.]
xiii
67
68
70
70
71
71
xiv
List of figures
2.28
2.29
2.30
2.31
2.32
2.33

Society.]
Society.]
Relative activation energy profile of mango tissues during intermittent
drying at a drying air temperature of 65 C and resting at 27 C.
[Reprinted from Industrial Engineering Chemistry Research, 50,
A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley, Intermittent drying of
mango tissues: Implementation of the reaction engineering approach,
10891098, Copyright (2012), with permission from the American
Chemical Society.]
Surface relative humidity profile of mango tissues during intermittent
drying at a drying air temperature of 65 C and resting at 27 C.
[Reprinted from Industrial Engineering Chemistry Research, 50,
A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley, Intermittent drying of
mango tissues: Implementation of the reaction engineering approach,
Chemical Society.]
Saturated vapour concentration and surface temperature profile of
mango tissues during intermittent drying at a drying air temperature of
65 C and resting at 27 C. [Reprinted from Industrial Engineering
Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A.
Webley, Intermittent drying of mango tissues: Implementation of the
reaction engineering approach, 10891098, Copyright (2012), with
permission from the American Chemical Society.]
Surface and saturated vapour concentration profile of mango tissues
during intermittent drying at a drying air temperature of 65 C and
resting at 27 C. [Reprinted from Industrial Engineering Chemistry
Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley,
Intermittent drying of mango tissues: Implementation of the reaction
engineering approach, 10891098, Copyright (2012), with permission
from the American Chemical Society.]
Surface vapour concentration and surface temperature profile of mango
tissues during intermittent drying at a drying air temperature of 65 C
and resting at 27 C. [Reprinted from Industrial Engineering Chemistry
72
72
74
75
75
76
List of figures
2.34
2.35
2.36
2.37
2.38
2.39

from the American Chemical Society].
Moisture content profile of intermittent drying of mango tissues with
heating (at a drying air temperature of 45 C) and resting periods of
4000 s each. [Reprinted from Industrial Engineering Chemistry
from the American Chemical Society.]
Saturated vapour concentration and surface temperature profile of
intermittent drying of mango tissues with heating (at a drying air
temperature of 45 C) and resting periods of 4000 s each. [Reprinted
from Industrial Engineering Chemistry Research, 50, A. Putranto,
Z. Xiao, X.D. Chen and P.A. Webley, Intermittent drying of mango
tissues: Implementation of the reaction engineering approach,
Chemical Society.]
Surface vapour concentration and surface temperature profile of
Chemical Society.]
Surface and saturated vapour concentration profile of intermittent drying
of mango tissues with heating (at a drying air temperature of 45 C) and
resting periods of 4000 s each. [Reprinted from Industrial Engineering
Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A.
Webley, Intermittent drying of mango tissues: Implementation of the
reaction engineering approach, 10891098, Copyright (2012), with
permission from the American Chemical Society.]
Surface vapour concentration and surface relative humidity profile of
Chemical Society.]
The relative activation energy (Ev /Ev,b ) of the convective drying of
kaolin. [Reprinted from Chemical Engineering Science, 66, A. Putranto,
X.D. Chen, S. Devahastin et al., Application of the reaction engineering
xv
76
77
77
78
78
79
xvi
List of figures
2.40
2.41
2.42
2.43
2.44
2.45
approach (REA) for modelling intermittent drying under time-varying

humidity and temperature, 21492156, Copyright (2012), with
Moisture content profile of intermittent drying in Case 1 (periodically
changed drying air temperatures between 6543 C). [Reprinted from
Chemical Engineering Science, 66, A. Putranto, X.D. Chen, S.
Devahastin et al., Application of the reaction engineering approach
(REA) for modelling intermittent drying under time-varying humidity
and temperature, 21492156, Copyright (2012), with permission from
Elsevier.]
Temperature profile of intermittent drying in Case 1 (periodically
Elsevier.]
Elsevier.]
Chemical Engineering Science, 66, A. Putranto, X.D. Chen,
S. Devahastin et al., Application of the reaction engineering approach
Elsevier.]
changed relative humidity between 412%). [Reprinted from Chemical
Engineering Science, 66, A. Putranto, X.D. Chen, S. Devahastin et al.,
Application of the reaction engineering approach (REA) for modelling
intermittent drying under time-varying humidity and temperature,
21492156, Copyright (2012), with permission from Elsevier.]
82
83
83
84
84
86
87
List of figures
2.46
2.47
2.48
2.49
2.50
2.51
2.52
2.53

Relative activation energy (Ev /Ev,b ) of the dehydration of wood
during heat treatment generated from the experimental data in Case 2
(refer to Table 2.10). [Reprinted from Bioresource Technology, 102,
A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley, Modelling of
high-temperature treatment of wood by using the reaction engineering
approach (REA), 62146220, Copyright (2012), with permission from
Elsevier.]
Moisture content profiles during the heat treatment of Cases 1 to 3 (refer
to Table 2.10). [Reprinted from Bioresource Technology, 102,
Elsevier.]
Temperature profiles during the heat treatment of Cases 1 to 3 (refer to
Table 2.10). [Reprinted from Bioresource Technology, 102, A. Putranto,
X.D. Chen, Z. Xiao and P.A. Webley, Modelling of high-temperature
treatment of wood by using the reaction engineering approach (REA),
Moisture content profiles during the heat treatment of Cases 4 and 5
(refer to Table 2.10). [Reprinted from Bioresource Technology, 102,
Elsevier.]
Temperature profiles during the heat treatment of Cases 4 and 5 (refer to
Table 2.10). [Reprinted from Bioresource Technology, 102, A. Putranto,
X.D. Chen, Z. Xiao and P.A. Webley, Modelling of high-temperature
treatment of wood by using the reaction engineering approach (REA),
The relative activation energy (Ev /Ev,b ) of baking of thin layer of
cake at an oven temperature of 100 C. [Reprinted from Journal of Food
Engineering, 105, A. Putranto, X.D. Chen and W. Zhou, Modelling of
xvii
87
88
90
92
93
94
95
xviii
List of figures
2.54
2.55
2.56
2.57
2.58
2.59
2.60
baking of cake using the reaction engineering approach (REA),

Moisture content profiles at baking temperatures of 100, 140 and
160 C. [Reprinted from Journal of Food Engineering, 105, A. Putranto,
X.D. Chen and W. Zhou, Modelling of baking of cake using the reaction
Moisture content profiles at baking temperatures of 50 and 80 C.
[Reprinted from Journal of Food Engineering, 105, A. Putranto, X.D.
Chen and W. Zhou, Modelling of baking of cake using the reaction
Temperature profiles at baking temperatures of 100, 140 and 160 C.
[Reprinted from Journal of Food Engineering, 105, A. Putranto, X.D.
Chen and W. Zhou, Modelling of baking of cake using the reaction
Temperature profiles at baking temperatures of 50 and 80 C. [Reprinted
from Journal of Food Engineering, 105, A. Putranto, X.D. Chen and W.
Zhou, Modelling of baking of cake using the reaction engineering
Elsevier.]
Heat transfer mechanisms of convective and infrared-heat drying.
convective drying of thin layer of polyvinyl alcohol (PVA)/glycerol/
water mixture The reaction engineering approach (REA), 348357,
Copyright (2012), with permission from Elsevier.]
Moisture content profile of convective and infrared drying at an air
temperature of 35 C, air velocity of 1 m s1 , air relative humidity of
18% and intensity of infrared drying of 3700 W m2 . [Reprinted from
Chemical Engineering and Processing: Process Intensification, 49, A.
Putranto, X.D. Chen and P.A. Webley, Infrared and convective drying of
thin layer of polyvinyl alcohol (PVA)/glycerol/water mixture The
Product temperature profile of convective and infrared drying at an air
temperature of 35 C, air velocity of 1 m s1 , air relative humidity of
18% and intensity of infrared drying of 3700 W m2 . [Reprinted from
Chemical Engineering and Processing: Process Intensification, 49, A.
Putranto, X.D. Chen and P.A. Webley, Infrared and convective drying of
thin layer of polyvinyl alcohol (PVA)/glycerol/water mixture The
97
98
98
99
100
101
103
104
List of figures
2.61
2.62
2.63
2.64
2.65
2.66
2.67
Sensitivity of the moisture content profile of cyclic drying, Case 1 (refer

to Table 2.12) towards n (on Equation 2.12.1). [Reprinted from Chemical
Engineering Science, 65, A. Putranto, X.D. Chen and P.A. Webley,
Application of the reaction engineering approach (REA) to model cyclic
drying of thin layers of polyvinyl alcohol (PVA)/glycerol/water mixture,
Sensitivity of the temperature profile of cyclic drying, Case 1 (refer to
Table 2.12) towards n (on Equation 2.12.1). [Reprinted from Chemical
Moisture content profile of cyclic drying, Case 1 (refer to Table 2.12)
using the first scheme (T* as function of infrared intensity) with n = 1.8.
[Reprinted from Chemical Engineering Science, 65, A. Putranto, X.D.
Chen and P.A. Webley, Application of the reaction engineering approach
(REA) to model cyclic drying of thin layers of polyvinyl alcohol
(PVA)/glycerol/water mixture, 51935203, Copyright (2012), with
Temperature profile of cyclic drying, Case 1 (refer to Table 2.12) using
the first scheme (T* as function of infrared intensity) with n = 1.8.
Sensitivity of the moisture content profile of cyclic drying, Case 1 (refer
to Table 2.12) towards q (on Equation 2.12.3). [Reprinted from Chemical
Sensitivity of the temperature profile of cyclic drying, Case 1 (refer to
Table 2.12) towards q (on Equation 2.12.3). [Reprinted from Chemical
using the second scheme (Ev ,b as function of infrared intensity) with
q = 1.8. [Reprinted from Chemical Engineering Science, 65, A.
Putranto, X.D. Chen and P.A. Webley, Application of the reaction
engineering approach (REA) to model cyclic drying of thin layers of
xix
106
107
108
109
110
110
xx
List of figures
2.68
2.69
2.70
2.71
2.72
2.73
polyvinyl alcohol (PVA)/glycerol/water mixture, 51935203, Copyright

the second scheme (Ev,b as function of infrared intensity) with q = 1.8.
using the second scheme (Ev ,b as function of infrared intensity) with
the second scheme (Ev ,b as function of infrared intensity) with q =
1.5. [Reprinted from Chemical Engineering Science, 65, A. Putranto,
X.D. Chen and P.A. Webley, Application of the reaction engineering
approach (REA) to model cyclic drying of thin layers of polyvinyl
alcohol (PVA)/glycerol/water mixture, 51935203, Copyright (2012),
with permission from Elsevier.]
111
111
112
112
113
113
114
List of figures

2.75 Moisture content profile of cyclic drying, Case 3 (refer to Table 2.12)
using the second scheme (Ev,b as function of infrared intensity) with
2.76 Temperature profile of cyclic drying, Case 3 (refer to Table 2.12) using
the second scheme (Ev,b as function of infrared intensity) with q = 1.6.
3.1 Schematic diagram of a cube dried in a uniform convective environment.
3.2 Moisture content profiles of the convective drying of mango tissues at a
drying air temperature of 45 C solved by the method of lines with 10
and 200 spatial increments. [Reprinted from AIChE Journal, 59,
Aditya Putranto, Xiao Dong Chen, Spatial reaction engineering
approach as an alternative for nonequilibrium multiphase mass-transfer
model for drying of food and biological materials, 5567, Copyright
(2012), with permission from John Wiley & Sons Inc.]
3.3 Average moisture content profiles of mango tissues during convective
drying at different drying air temperatures. [Reprinted from AIChE
Journal, 59, Aditya Putranto, Xiao Dong Chen, Spatial reaction
engineering approach as an alternative for nonequilibrium multiphase
mass-transfer model for drying of food and biological materials, 5567,
Copyright (2012), with permission from John Wiley & Sons Inc.]
3.4 Centre temperature profiles of mango tissues during convective drying at
different drying air temperatures. [Reprinted from AIChE Journal, 59,
3.5 Spatial moisture content profiles of mango tissues during convective
drying at drying air temperatures of 45 C. [Reprinted from AIChE
xxi
2.74
115
115
116
122
131
134
134
xxii
List of figures
3.6
3.7
3.8
3.9
3.10
3.11
3.12
3.13

Spatial water vapour concentration profiles of mango tissues during
convective drying at drying air temperatures of 45 C. [Reprinted from
AIChE Journal, 59, Aditya Putranto, Xiao Dong Chen, Spatial reaction
Spatial temperature profiles of mango tissues during convective drying
at drying air temperatures of 45 C. [Reprinted from AIChE Journal, 59,
Profiles of evaporation rates inside mango tissues during convective
drying at a drying air temperature of 55 C. [Reprinted from AIChE
Moisture content profiles in the core and cortex during convective drying
of potato tissues with a diameter of 1.4 cm. [Reprinted from AIChE
Moisture content profiles in the core and cortex during convective drying
of potato tissues with a diameter of 2.8 cm. [Reprinted from AIChE
Core temperature profiles during convective drying of potato tissues
with a diameter of 1.4 cm. [Reprinted from AIChE Journal, 59, Aditya
Putranto, Xiao Dong Chen, Spatial reaction engineering approach as an
alternative for nonequilibrium multiphase mass-transfer model for
drying of food and biological materials, 5567, Copyright (2012), with
permission from John Wiley & Sons Inc.]
Average moisture content profiles of mango tissues during intermittent
drying at different drying air temperatures.
Spatial moisture content profiles of mango tissues during intermittent
drying at a drying air temperature of 55 C.
135
136
137
138
139
140
140
143
144
List of figures
3.14
3.15
3.16
3.17
3.18
3.19
3.20
3.21
3.22
3.23
3.24
3.25
3.26
3.27
3.28
3.29
3.30
3.31
3.32
4.1
Spatial water vapour concentration profiles of mango tissues during

intermittent drying at a drying air temperature of 55 C.
Centre temperature profiles of mango tissues during intermittent drying
at different drying air temperatures.
Spatial temperature profiles of mango tissues during intermittent drying
at a drying air temperature of 55 C.
Profiles of evaporation rate inside mango tissues during intermittent
Profiles of average moisture content during heat treatment in Case 2
(refer to Table 3.5) solved by the method of lines using 10 and 100
increments.
Effect of liquid diffusivity on profiles of the moisture content during heat
treatment in Case 1 (refer to Table 3.5).
Effect of liquid diffusivity on profiles of temperature during heat
treatment in Case 1 (refer to Table 3.5).
(refer to Table 3.5).
Profiles of temperature during heat treatment in Case 1 (refer to Table
3.5).
Profiles of temperature during heat treatment in Case 2 (refer to Table
3.5).
Profiles of spatial moisture content during heat treatment in Case 2 (refer
to Table 3.5).
Profiles of spatial water vapour concentration during heat treatment in
Case 2 (refer to Table 3.5).
Profiles of spatial temperature during heat treatment in Case 2 (refer to
Table 3.5).
Profiles of average moisture content during the baking of bread at a
baking temperature of 150 C.
Spatial profiles of moisture content during the baking of bread at a
baking temperature of 150 C and air velocity of 10 m s1 .
Spatial profiles of concentration of water vapour during the baking of
bread at a baking temperature of 150 C and air velocity of 10 m s1 .
Profiles of top and bottom surface temperatures during the baking of
bread at a baking temperature of 150 C and air velocity of 1 m s1 .
Spatial profiles of temperature during the baking of bread at a baking
temperature of 150 C and air velocity of 10 m s1 .
Experimental setup for convective drying of porcine skin. [Reprinted
from Chemical Engineering Research and Design, 87, S. Kar, X.D.
Chen, B.P. Adhikari and S.X.Q. Lin, The impact of various drying
kinetics models on the prediction of sample temperaturetime and
xxiii
144
145
146
147
151
152
152
153
153
154
155
156
156
157
161
162
162
163
163
xxiv
List of figures
4.2
4.3
4.4
4.5
4.6
4.7
4.8
moisture contenttime profiles during moisture removal from stratum

corneum, 739755, Copyright (2012), with permission from Elsevier.]
(a) Overview of a sample/plate assembly for convective drying of
porcine skin. (b) Detailed of layering structure of sample support.
[Reprinted from Chemical Engineering Research and Design, 87, S. Kar,
X.D. Chen, B.P. Adhikari and S.X.Q. Lin, The impact of various drying
kinetics models on the prediction of sample temperaturetime and
moisture contenttime profiles during moisture removal from stratum
corneum, 739755, Copyright (2012), with permission from Elsevier.]
Moisture content profiles from the convective drying of mango tissues
modelled using the L-REA and diffusion-based model (Vaquiro et al.,
2009). [Reprinted from Drying Technology, 29, A. Putranto, X.D. Chen
and P.A. Webley, Modelling of Drying of Food Materials with Thickness
of Several Centimeters by the Reaction Engineering Approach (REA),
961973, Copyright (2012), with permission from Taylor & Francis Ltd.]
Temperature profiles from convective drying of mango tissues modelled
using the L-REA and diffusion-based model (Vaquiro et al., 2009).
[Reprinted from Drying Technology, 29, A. Putranto, X.D. Chen and P.A.
Webley, Modelling of drying of food materials with thickness of several
centimeters by the reaction engineering approach (REA), 961973,
Moisture content profiles from the convective drying of mango tissues
modelled using the S-REA and diffusion-based model (Vaquiro et al.,
2009). [Reprinted from AIChE Journal, A. Putranto and X.D. Chen,
Spatial reaction engineering approach as an alternative for
non-equilibrium multiphase mass-transfer model for drying of food and
biological materials, DOI 10.1002/aic.13808, Copyright (2012), with
permission from John Wiley & Sons, Inc.]
Temperature profiles from the convective drying of mango tissues
modelled using the S-REA and diffusion-based model (Vaquiro et al.,
2009). [Reprinted from AIChE Journal, A. Putranto and X.D. Chen,
Spatial reaction engineering approach as an alternative for
non-equilibrium multiphase mass-transfer model for drying of food and
biological materials, DOI 10.1002/aic.13808, Copyright (2012), with
permission from John Wiley & Sons, Inc.]
Moisture content profiles from the heat treatment of wood modelled
using the L-REA and Luikovs approach. [Reprinted from Bioresource
Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley,
Modelling of high-temperature treatment of wood by using the reaction
Temperature profiles from the heat treatment of wood modelled using
the L-REA and Luikovs approach. [Reprinted from Bioresource
Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley,
174
175
187
188
189
190
200
List of figures
4.9
4.10
4.11
4.12
4.13
4.14
Modelling of high-temperature treatment of wood by using the reaction

Moisture content profiles from the heat treatment of wood (refer to Table
4.1) modelled using the L-REA and Whitakers approach. [Reprinted
from Bioresource Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and
P.A. Webley, Modelling of high-temperature treatment of wood by using
the reaction engineering approach (REA), 62146220, Copyright
Temperature profiles from the heat treatment of wood (refer to Table 4.1)
modelled using the L-REA and Whitakers approach. [Reprinted from
Bioresource Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and P.A.
Webley, Modelling of high-temperature treatment of wood by using the
reaction engineering approach (REA), 62146220, Copyright (2012),
with permission from Elsevier.]
Moisture content profile from the heat treatment of wood modelled using
the S-REA and Luikovs approach.
Temperature profile from the heat treatment of wood modelled using the
S-REA and Luikovs approach.
Moisture content profiles from the heat treatment of wood (refer to Table
4.1) modelled using the S-REA and Whitakers approach.
Temperature profiles from the heat treatment of wood (refer to Table 4.1)
modelled using the S-REA and Whitakers approach.
xxv
201
202
203
204
204
205
205
Tables
2.1
Experimental conditions of convective drying of a mixture of

polymer solutions (Allanic et al., 2009).
page 51
2.2
R2 and RMSE of modelling of a mixture of polymer solutions using
the L-REA.
58
2.3 Experimental conditions of convective drying of mango tissues (Vaquiro
et al., 2009).
61
2.4
R2 and RMSE of modelling of convective drying of mango tissues using
the L-REA.
66
2.5 Schemes of intermittent drying of mango tissues (Vaquiro et al., 2009).
69
2.6
R2 and RMSE of modelling of intermittent drying of mango tissues using
the L-REA.
73
2.7 Settings of intermittent drying of kaolin (Kowalski and Pawlowski,
2010).
81
2
2.8
R , RMSE, average absolute deviation and maximum absolute deviation
of profiles of moisture content predicted by and Kowalski and
Pawlowskis model (2010b).
85
2.9
R2 , RMSE, average absolute deviation and maximum absolute deviation
of profiles of temperature predicted by Kowalski and Pawlowskis model
(2010b).
85
2.10 Settings of heat treatment of wood samples (Younsi et al., 2006a; 2007).
89
99
2.11 R2 of modelling using the REA.
2.12 The experimental conditions of intermittent drying of a mixture of
polymer solutions.
105
3.1 Experimental conditions of convective drying of mango tissues (Vaquiro
et al., 2009).
128
2
3.2
R and RMSE of convective drying of mango tissues using the S-REA.
135
3.3 Scheme of intermittent drying of mango tissues (Vaquiro et al., 2009).
141
142
3.4
R2 and RMSE of intermittent drying of mango tissues.
3.5 Experimental settings of wood heating under a constant heating rate
(Younsi et al., 2007).
148
3.6
R2 and RMSE of modelling of heat treatment of wood under a constant
heating rate using the S-REA.
155
4.1 Experimental settings of the heat treatment of wood (Younsi et al., 2007).
201
Preface
Drying is one of the oldest and most effective methods for preserving food and biological
materials. Low moisture content in foods prevents the growth of bacteria responsible for
their deterioration so foods can have extended shelf-lives. When foods became abundant,
trade became possible. Today, dried products are the main materials trading round the
world but this is not limited to food products. Construction materials, textiles, electronic
parts and appliances, biomass-based fuels, pharmaceutics and many other materials
important to our daily lives and the business world are all included. Essentially over 80%
of the products on Earth require drying as one of the steps in their production. Product
quality and process parameters are interactive. Industrial drying is energy hungry; a
process involving simultaneous heat, mass transfer and momentum transfer. Product
quality is determined through compositional and structural rearrangements, as well as
chemical reactions in some circumstances. For existing drying facilities, optimisation is
often needed to achieve new goals such as energy reduction, quality improvements and
development of new materials. There are also opportunities in designing dryer modifications or even brand new dryers that are superior in performance over conventional
devices. Modelling of drying processes is very useful for these purposes.
A number of drying models have been proposed, which are conveniently classified
into empirical and mechanistic models. The empirical models give advantages of being
simple in their mathematical formulation. However, these models most often cannot
explain the physics of drying and their application is limited since they are valid only
for a particular set of drying conditions. On the other hand, the mechanistic models are
derived based on fundamental phenomena that occur during drying. These phenomena
are crucial in material science (and materials processing) though material scientists
themselves may not have yet come to appreciate the process engineering aspects which
impact on the product microstructure. Some of these models can capture the physics
well. These models are, however, often mathematically complex and sometimes contain
too many parameters, which need to be determined experimentally (prior to model
predictions).
For some decades now, a comprehensive set of macroscopic equations has been developed and used to address heat and mass transfer and mechanical aspects related to drying.
The application of macroscopic descriptions of drying (temperature, moisture and sometimes pressure) has been perfected over the past two decades, and relevance has been
confirmed in many drying configurations. Some of these involve irreversible thermodynamics formulations, which are lengthy and have many model coefficients. These have
xxviii
Preface
become the classical approach. However, this classical approach has serious limitations. The concept of multi-scale and multi-physics addresses some of these limitations,
e.g. coupled meso-scale and equipment scale problems. When a local thermodynamic
equilibrium is not attained, however, the time scales usually overlap. This is a real multiscale configuration and challenging in terms of the great demand in computational power
and handling of mathematics. Several scales can be considered simultaneously, ranging
from simple exchanges between macroscopic phases to comprehensive formulations in
which time evolution of microscopic values and microscopic gradients is considered
over a representative elementary volume, according to a recent review by Patrick Perre
(for a review of modern computational and experimental tools relevant to the field of
drying, see Drying Technology, 29, 15291541, 2011).
While exploring the detailed physics involved in drying using these multi-scale and
multi-physics approaches, it is, from an engineering viewpoint, also important to develop
new ideas establishing simpler models. In general, today industrial drying applications
require mathematical models that are simple and easy to use. For practical purposes, an
effective drying model should be simple, accurate, and able to capture the major physics
of drying and its application should be robust. This model should also favour short
computation time and it should be easy to establish parameters needed (experimentally)
to help quicker decision-making in an industry environment (and with the lowest cost).
The reaction engineering approach (REA), which is a middle path approach, perhaps
between the empirical and the mechanistic models, was first thought about by the first
author of this book, Chen, in 1996. Through much of the research on its possible applications, it has been revealed that the REA is indeed simple, accurate and robust enough to
model many cases of drying, i.e., drying in a constant or variable environment. The REA
has also been implemented in industry for prediction of spray dryer performance and
shows good agreement with plant data for different scales in the dairy industry. It has also
been extended to various other challenging systems of drying, such as polymer drying,
intermittent drying, thermal-thick materials, infrared heating and microwave heating.
The model is significantly easier to implement and requires less experimentation effort
to establish the parameters needed, compared with the more fundamental models. The
REA was first taken as a lumped model which does not need us to resolve the spatial
distribution of water content, etc.; the lumped-REA (or L-REA), but in recent times, we
have also extended the approach to describe spatially distributed systems; spatial-REA
(or S-REA).
The REA approach has been initiated and exercised over the past 12 years and there is
a significant amount of successful applications already illustrated. As mentioned earlier,
it is a middle path between the rigorous theory that requires high-level mathematics and
the empirical models that do not represent much physics. We can see, through our own
practices and from other colleagues in the same area who have used the REA concept, it
is a really straightforward approach to modelling some rather complex drying processes;
hence, it is simple and cost-effective to establish accurate REA models to use in industry.
This book is the most fundamental and comprehensive description of the REA
approach to drying modelling the basic idea, rationale, mathematical description
and implementation procedures for various systems. This approach has been extended,
Preface
xxix
and experimented with, by several quality Ph.D. graduates, in particular, the second
author, Aditya Putranto. Regarding the other more established theories, this book not
only provides essential details so the readers can refer to them but also illustrates, by
comparison, the physics involved in REA concepts. The disadvantages and advantages
between theories are also briefly introduced. The book should benefit both academics in
drying research and practicing engineers in industry. Undergraduate students in process
engineering may also find it useful for quickly setting up a drying model for design
purposes. The main emphasis of this book is how to apply the REA to reality. The book
will also elaborate on potential applications of similar thinking to more complex reactive
systems that couple with drying processes, hopefully to foster their future development.
Here, the modern ideas of microstructure development and product qualities created
by drying processes, and in turn their impacts on moisture transfer, will be introduced.
This should make the book more relevant in years to come.
Xiao Dong Chen and Aditya Putranto
During my Ph.D. study in the Chemical and Process Engineering Department at

Canterbury University, Christchurch, New Zealand, (19881990), the main task was
to establish mechanistically the understanding of moisture influence on coal oxidation
and the impact of moisture transfer in a packed coal particle bed on the development
of spontaneous combustion. The experimental aspect was challenging both technically
and physically. In addition to coal oxidation and its racemic measurement, I became
very interested in the mechanisms of water evaporation and moisture transfer (liquid
and vapour) in porous material. Dr Jim Stott (Reader of Chemical Engineering) was my
main supervisor and Dr John Abrahamson (Senior Lecturer), in the same department,
was my cosupervisor. Jim published some of the pioneering literature on the subject of
spontaneous combustion of coal (1959) and built (largely by himself) ingenious experimental rigs. Dr Abrahamson was an inspirational and distinguished individual as well
who has been credited as one of the first to have made a carbon nanotube (he called it
the carbon cylinder) (1978), a theory of ball lightening (2000) and a theory of particle
collision frequency in a turbulent field (1972). John was Jims student some years back.
Working with Jim on the subject of spontaneous combustion development in a moist
coal bed has taught me that if the coal bed were completely saturated with water vapour
under near ambient pressure (the institutional voids of the bed remain saturated with
water vapour), the maximum temperature would remain at around 80 C. This was
predicted from a numerical spontaneous combustion model involving mass transfer
of moisture within the coal bed when assuming the vapour concentration in the bed
is always saturated. Jim discovered this in the late 1960s, and later, in the 1970s, a
Ph.D. student of his proved this more comprehensively. This aspect was more or less
republished in 1990s by a research group in Europe (who were perhaps unaware of
the work by Jim and his ex-students). However, if an equilibrium relationship between
moisture content in the coal particles and vapour concentration in the air surrounding
the particles can be adopted, a dry spot can be predicted and the maximum temperature
will exceed the boiling temperature of water, therefore rising to an elevated temperature
due to oxidation heat (Chen, 1992a). Of course, there are also other influences such as
porosity, oxidation rate and oxygen transfer, heat transfer and, sometimes, fluid flow due
to a pressure gradient. Nevertheless, this equilibrium relationship is what we are now so
familiar with, termed the equilibrium isotherm in drying literature. The oxidation rate
of coal itself was also found, in my own experiments, to vary with the residual water
content (Chen and Stott, 1993) and I had gone to extra lengths to try to understand this
xxxi
phenomenon. This formed the foundation of my understanding of the presence of water

affecting chemical (and biochemical) reactions. In food drying, it would mean that the
removal of water, to some extent, could significantly reduce rate of deterioration, giving
a long shelf life to products (oxidative or microbial) (Chen and Mujumdar, 2008).
As I became aware of moisture transfer, I became very aware of the existence of a
giant of drying in the same department, Professor Roger Keey, who wrote the first
book on drying principles and practice that was published in English. I had spent a lot of
time looking for information on how to model moisture transfer, coupled with chemical
reactions and heat transfer and momentum transfer. Keeys books over the years have had
an impact on my own work related to this area (especially the latest one; Keey, 1992). In
particular, I have picked up the essence of the characteristic drying rate curve (CDRC)
approach. One of my friends in the drying area, Professor Tim Langrish, a Canterbury
graduate, has worked extensively on this idea, which has extended Keeys views on the
drying of wood and some other different materials, including foods. His postdoctoral
period (after his return from Oxford University) with Professor Keey overlapped with the
final year of my Ph.D. (1990). Another distinguished individual whose work has affected
my own thinking has been Professor Shusheng Pang, another Canterbury graduate
supervised by Roger Keey, who has published some key literature in wood drying related
to the application of CDRC. CDRC captures the phenomena of drying by recognising
the existence of a constant and falling drying rate period(s). The critical or transitional
water content between any of the connecting rate schemes are recognised (Keey, 1992).
Doctors Sandeep Chu and Peter Kho, who were student colleagues at Canterbury during
the period of 19881990 and whose works were supervised by Professor Keey, also had
an impact on my later research on drying.
Some others who also influenced me positively were Professor Miles Kennedy, Dr John
Peet and Dr Maurice Allen at Canterbury. I had read many of their works during the
peaceful evenings when I had pretty much the whole department to myself and some
of the weekends during my Ph.D. study at the corner room on the top floor of Simons
Block. The surroundings of Canterbury University were beautiful and peaceful and gave
me great times (and spaces) to spend thinking about my work and, of course, my loved
ones.
I submitted my Ph.D. thesis three years after I started in late December 1987. I
started working at the New Zealand Dairy Research Institute (NZDRI) (which is now
the Fonterra Research Centre based in Palmerston North of New Zealand), first as
an engineer and then as a senior engineer, working on spray drying and milk powder agglomeration. Dr Kevin Pearce (my section manager), who was a distinguished
chemist, gave advice that I understood one has to take in order to take protein chemistry
seriously when dealing with engineering problems related to dairy products. This period
of time was very constructive for my career development. After coal research, I really
wanted to move onto biotechnology and, at the time, the food industry was the nearest
thing to biotechnology in which I could secure a good position. I was deeply involved
in milk powder technology and have become very familiar with powder technology,
dissolution properties of the powders, powder agglomeration and instantisation (Chen,
1992b), glass transition and stickiness (Lloyd et al., 1996), etc. I was lucky enough to
xxxii
Xiao Dong Chen (left) and Dr Jim Stott (right) working on the 2-m-long packed coal column
investigating spontaneous heating of coal, 1989, University of Canterbury, Christchurch, New
Zealand.
xxxiii
make a significant contribution in the area of agglomeration (hardware improvement

and macrostructural analysis) and new product development that was hampered by high
stickiness, with large financial returns for the dairy industry.
My employment as an academic at the Department of Chemical and Materials Engineering, The University of Auckland, started in late 1993, which instantly gave me
greater freedom to develop new ideas. I had great fun working at Auckland, benefitting
from being surrounded by a number of highly positive individuals at the department and
the school. One strong influence came from my colleagues who were experts in materials science. Among many other studies, in 19951996, I had developed an idea that
was initially thought to be able to unify drying kinetics to the equilibrium relationship
(Chen and Chen, 1997). The notion of unified came from, at the time, an ambitious
young man (me) but later was proven to be, well, kinetics is just kinetics and equilibrium is equilibrium, so they dont have to be 100% linked. What had emerged, however,
was that if I could find a simple relationship between the surface vapour concentration
and the water content of the porous solid material being dried, noting that this surface
vapour concentration less the vapour concentration in the gas phase is the driving force
for moisture transfer from the porous material to the drying environment, the model
could be a good alternative to the CDRC model. The obvious one for surface vapour
concentration to relate to is liquid water content. At the time, I already found some issues
with the CDRC approach, as uncertainty can be great depending on the drying processes
considered. Keey (1992) has rightly pointed out that the CDRC model was excellent for
particles or sample sizes smaller than 20 mm for constant drying conditions.
The link between the surface vapour concentration and the remaining water content in the porous material as well as the material surface temperature was eventually
constructed using an Arrhenius-type relationship, which essentially suggested the evaporation of pure water and extraction of the water from the inside the porous material
was a reaction and the condensation/adsorption was not an activation process. This is
in line with a mathematical description of evaporation and a condensation mechanism
formulated by Gray and Wake (1990). Professor Brian Gray (Professor of Applied Mathematics at Sydney University at the time) is a distinguished applied mathematician (he
is also a distinguished physical chemist) whom I came to know through the link between
me and Professor Graeme Wake (another outstanding applied mathematician from New
Zealand) and had influenced my approaches to engineering in more than just one aspect.
They were not particularly interested in drying, but they were very much interested in
the systems of reactions, both exothermic and endothermic. Evaporation is viewed as an
endothermic reaction mathematically speaking.
My father, who was a Professor of Aerodynamics at the Chinese Academy of Sciences,
visited me in New Zealand in 1996. I discussed some of my initial ideas with him and
we prepared a simple paper for Chemeca in 1997. I was also fortunate in hosting a visiting researcher from Xian Jiao-Tong University (China) during 19961997, Associate
Professor Guozhen Xie, who was a refrigeration expert but was daring enough to pick
up drying modelling as the main topic in his year of working with me. We didnt do any
experiments on drying but used data reported in the literature. However, in all cases, we
had to solve an energy balance to obtain (surface) temperature of the material tested for
xxxiv
the concept. Most of the examples used (Chen and Xie, 1997) were small-sized samples.
Once we had the temperature-time profile for the sample of concern during drying, we
could establish the activation energy in the Arrhenius equation (mentioned earlier) to
demonstrate the concept.
Then, in 2000, at Auckland, I had a masters student by research, Wayne Pirini, who
was interested in drying, so we started experimenting on thin-layer drying of various
materials measuring both weight loss and temperature as drying proceeded. The first lot
of data on activation energies obtained was reported by Chen, Pirini and Mustafa in 1996.
However, I was not aware that the Biot number defined in heat transfer literature could
not account for the conditions when evaporation occurs. This gave me an opportunity
to derive a modified Biot number later on (the so-called ChenBiot number). Then, at
Auckland in the period of 20002004, Dr Sean Lin, my Ph.D. student at that time, did a
comprehensive study on droplet drying kinetics for dairy products in particular. He had
lots of practical experience before coming to me. He designed and built an excellent
cost-effective droplet drying test rig and conducted probably the most careful, accurate
experiments on dairy droplet drying. This has allowed the comprehensive establishment
of the REA model for dairy droplets that is relevant to the spray drying industry (Chen
and Lin, 2005).
Following that, two Ph.D. students under my supervision who were from India, Drs
Saptarshi Kar and Kamlesh Patel, had made a significant contribution to the development of the REA concept. Saptarshi applied REA to a spatial distributed case for water
transport in skin relevant to transdermal drug delivery for the first time. We deliberately
ignored the liquid diffusivity to see if it really mattered. It turned out that it really did matter. Kamlesh had helped in extending the ChenBiot number concept and helped to bring
in a new concept called the composite REA approach, which describes an approach
to estimating the activation energies of sugar mixtures based on the components own
activation energies. They were both tremendous students with high aptitudes to pursue
basic research. Saptarshi in particular tended towards a more theoretical rigorousness.
They started their Ph.Ds at Auckland and finished at Monash University.
In 12 years at Auckland, I moved from (in the English system) Lecturer (1993) to
Senior Lecturer (1995) to Associate Professor (1998) to Personal Chair Professor at the
age of 36 (2001). It was the most dramatic time in my life, both in career and personal
life. I had my first child, Lisa, who was born in May 2000. Sad events had taken her
mother away from her in 2001. I must thank the Engineering Dean at the time, Professor
Peter Brothers, who, in my darkest days in 2001, promised his institutions support in
allowing me to do whatever I needed to do and go wherever I needed to go without
worrying about losing my job.
Beyond that, I enjoyed tremendous learning experiences, friendships, and support
from my colleagues at the Department of Chemical and Materials Engineering: Professor Geoff Duffy (who was most influential individual in my stay at Auckland),
Associate Professor Kevin Free, Professor John J. J. Chen, Professor Wei Gao, Professor
Mohammed Farid, Professor Neil Broom, Professor George Fergusson and Dr Necati
Ozkan. I was inspired by the genius professors such as Professor John Boys (Electrical Engineering), Professor Peter Hunter (Engineering Science) and Professor Debes
xxxv
Aditya Putranto (left) and Xiao Dong Chen (right), November 2012, International Drying
Symposium (IDS 2012) chaired by Xiao Dong Chen, Xiamen, China.
Bhattahtrayya (Mechanical Engineering) for their innovations. I benefited tremendously

from collaborating with Associate Professor Sing Kiong Nguang, who is a genius in
mathematical problems in system and dynamics engineering. In that period of time at
Auckland, I picked up the idea of combining process engineering and material science
and became familiar with microscopy and material science techniques. My colleagues
have created an incredibly creative and happy environment for me to work in. Of course,
there were giants who supported me graciously over those years; Professor John Hood
(Vice Chancellor of The University of Auckland and then, later, Vice Chancellor of
Oxford University) and Professor Diane McCarthy (Dean of Medicine at Auckland
and later President of Royal Society of New Zealand). Without their recognition of
my ability and my contribution, my rapid promotion at Auckland would not have been
possible.
Coming back to the main technical topic, can the REA model do the things that a
CDRC model cannot? For small-size particles and constant drying conditions, CDRC
seems to be very comparable with REA. With this question, and many others, I had
moved to the Department of Chemical Engineering at Monash University (Melbourne,
Australia) to take up the Chair of Biotechnology at Monash University in 2006.
In 2009, I had great fortune in that a high-calibre student from Indonesia, Dr Aditya
Putranto, a humble young man, joined my group at Monash to do a Ph.D. with me.
xxxvi
He demonstrated superior ability in testing and further developing REA for numerous
applications, which are presented in this book.
In 2010, I moved to Xiamen University on the southeast coast of China, from which my
grandparents graduated in history and English literature in 1930 and 1931, respectively
as a National Expert Professor of Chemical Engineering (also known as the 1000Elite Chair Professor). I have not stopped the excellent collaborations with Aditya and
we continue to expend the REA. Of course, my other great Ph.D. students, Nan Fu,
Winston Wu and Sam Rogers (in the period of 20082011), a postdoc fellow, Dr Yan
Jin (in 2009), and Dr Mengwai Woo (20102011), have also continued to contribute
experimentally, and theoretically, to the establishment of REA and its applications to
the real world. Notably, Nan generated significantly new data on the REA approach
to dairy droplet drying and linked drying to crystallisation and particle solubility. She
has extended the techniques of single droplet drying to a more powerful means in
order to understand drying-quality inter-relations. Dr Jin has comprehensively modelled
the three-dimensional transient flows in large-scale spray dryers and has incorporated
the REA approach. Dr Woo has independently investigated the robustness of the REA
approach for modelling droplet drying in the context of computational fluid dynamics
of spray dryers.
In no way can I claim it was only me who made REA development possible, but I can
claim the original idea and model framework to be mostly mine. I sincerely thank all the
previously mentioned individuals and others whom I have not mentioned but who have
made contributions to the development of the REA in one way or another. REA is also
the result of a belief that engineering theory should be as simple and robust as possible
in order to enable a broad range of applications.
Finally, I would like to dedicate the book as follows:
To my lovely family starting from my wife Lily and the children Lisa, Nathan and
Benjamin.
To my grandparents, my parents, my sister and brother-in-law for their neverending
love and support.
To others whom I have loved and who have loved me selflessly.
Xiao Dong Chen
Xiamen City,
Southeast Coast of China
August 2012
Aditya Putranto would like dedicate this book as follows:
To his parents and sister for their endless love and support. To the Creator and others
whom have shared, and will share the love and faithfulness of the Creator.
Aditya Putranto
Melbourne
Australia
August 2012
xxxvii
References
Chen, X.D., 1991. The Spontaneous Heating of Coal Large Scale Laboratory Assessment and
Supporting Theory, Ph.D. thesis, Chemical and Process Engineering Department, University of
Canterbury, New Zealand.
Chen, X.D., 1992a. On the mathematical modelling of transient process of spontaneous heating
in a moist coal stockpile. Combustion and Flame 90, 114120.
Chen, X.D., 1992b. Whole milk powder agglomeration Principle and practice. In Milk Powders
for the Future, X.D. Chen (ed.), Dunmore Press: Palmerston North, New Zealand.
Chen, X.D. and Chen, N.X., 1997. Preliminary introduction to a unified approach to modelling
drying and equilibrium isotherms of moist porous solids, Chemeca97, Rotorua, New Zealand,
Sept. 1997, Paper DR3b (on CD-ROM).
Chen, X.D. and Lin, S.X.Q., 2005. Air drying of milk droplet under constant and time-dependent
conditions. AIChE Journal 51(6), 17901799.
Chen, X.D. and Mujumdar, A.S. (eds.), 2008. Drying Technologies in Food Processing. Blackwell
Publishing Ltd: Oxford.
Chen, X.D. and Stott, J.B., 1993. The effect of moisture content on the oxidation rate of coal
during near equilibrium drying and wetting at 50 C. Fuel 72, 787792.
Chen, X.D. and Stott, J.B., 1997. Oxidation rates of several New Zealand coals as measured in
large scale one-dimensional spontaneous heating experiments. Combustion and Flame 109,
578586.
Chen, X.D. and Xie, G.Z., 1997. Fingerprints of the drying of particulate or thin layer food
materials established using a simple reaction engineering model. Transactions of the Institute
of Chemical Engineers Part C: Food and Bio-Product Processing 75(C), 213222.
Gray, B.F., 1990. Analysis of chemical kinetic systems over the entire parameter space 3. A wet
combustion system, Proc. Roy. Soc. A 429, 449458.
Gray, B.F. and Wake, G.C., 1990. The ignition of hygroscopic materials by water. Combustion
and Flame 79, 26.
Keey, R.B., 1992. Drying of Loose and Particular Material. Hemisphere: New York.
Lloyd, R.J., Chen, X.D. and Hargreaves, J., 1996. Glass transition and caking of spray dried
amorphous lactose. International Journal of Food Science and Technology 31, 305311.
Introduction
1.1
Practical background
Drying (removing water from wet material) has been a very important processing step
for a wide range of human endeavours in our history. The dependence of human society
on drying is highly visible. For instance, in food production and preservation, drying
is the oldest, most popular and one of the most effective ways to make solid foods
and to preserve them as long as practically required. The textile industries need drying
processes. Natural fibre-based products such as those from the wood and paper industries
also need drying as a critical step in manufacturing. In fact, anything having to do with
the particulate products, not just food particles (milk powders, vegetable soup powders
and the like) but also detergents, fertilisers, and even paints: drying is critical.
As modern material science industries have started to develop at a speed never seen
before in our history, wet chemistry is needed, which requires drying (dewetting) to
form solid products which are more usable and transportable.
Some historical and typical products are shown in Figure 1.1.
Food drying is conducted in many ways. The history of using sunlight to dry fruits
goes back thousands of years, dating back to the fourth millennium BC in Mesopotamia
( http://en.wikipedia.org/wiki/Dried fruit). Today, dried fruits have the majority of the
original water content removed either naturally, through solar drying or sometimes freeze
drying and air drying (with low-humidity air in particular), or unnaturally, through the
use of specialised dryers or dehydrators powered by electricity or combustion. These
unnatural ways include mechanical dewatering, convective air/gas drying, superheated
steam drying, electro-osmosic processes, osmotic pressure-driven processes, refractory
window drying, freeze drying, vacuum drying, and microwave-aided drying processes,
to name a few. If one includes liquid evaporation (with liquid products also), these may
be expanded to evaporation operations, such as falling film, rising film evaporation,
vacuum distillation, and the like.
Dried fruits are popular products due to their enhanced sweet taste, concentrated
nutritional value and long shelf-life because the water activity is low (Chen and
Mujumdar, 2008). As water content is removed, the material shrinks (leading to a smaller
volume), the sugar content per unit volume of the material increases, as do nutritional
components such as proteins, vitamins and so on. Today, dried fruit consumption is
widespread. Nearly half the dried fruits sold are raisins, followed in popularity by dates,
prunes (dried plums), figs, apricots, peaches, apples and pears (Hui, 2006). Many are
(a)
(c)
(b)
(d)
Figure 1.1 Some traditional dried products. (a) Broccoli-steam blanched and air dried (kindly
provided by Ms Xin Jin, Wageningen University, The Netherlands), (b) air-dried Chinese tea
leaves (taken at Xiamen University laboratory), (c) spray dried aqueous herbal extract (particle
size is about 80 m) (taken at Xiamen University laboratory), (d) timber stacked for kiln drying
(kindly provided by Professor Shusheng Pang, Canterbury University, New Zealand).
referred to as conventional or traditional dried fruits: fruits that have been dried in
the sun or in heated wind tunnel dryers. Many fruits, such as cranberries, blueberries,
cherries, strawberries and mangoes, may be infused with a common sugar (e.g. sucrose
syrup) prior to drying to enhance sweetness and microbial stability. This means these
are not necessarily healthy products, especially for diabetics. Sugar-infused and dried
papaya and pineapples are actually candied fruit. Dried fruits are usually thought to
retain most of the nutritional value of the fresh fruits. The specific nutritional content of
dried fruits reflects that of their fresh counterparts and is influenced by the processing
method or processing technology, particularly processing temperature. In general, all
dried fruits provide essential nutrients and an array of healthy protective ingredients,
making them valuable tools to both improve diet quality and help reduce the risk of
chronic disease.
Furthermore, dried fruits (and nuts) are not only important sources of vitamins, minerals and fibre in the diet but also provide a wide array of bioactive components or
Introduction
phytochemicals. These plant compounds are not designated traditional nutrients since
they are not essential to sustain life but play a role in health and longevity and have
been linked to a reduction in risk of developing major chronic diseases. Convincing
evidence suggests that the benefits of phytochemicals may be even greater than currently understood, as they seem to affect metabolic pathways and cellular reactions.
However, the precise mechanisms by which specific compounds exert their biological
effect remains largely hypothetical, which requires greater investigation. Certainly, as
is well known, dried fruits are an excellent source of polyphenols and phenolic acids
(USDA, 2007). These compounds make up the largest group of phytochemicals in the
diet and appear to be, at least partially, responsible for the potential benefits associated with the consumption of diets rich in fruits and vegetables. Different dried fruits
have unique phenolic profiles (Donovan et al., 1998); for example, the most abundant in raisins are the flavonols (quercetin and kaempferol) and the phenolic acids
(caftaric and coutaric acid) (Willamson and Carughi, 2010). Therefore, due to their
high polyphenol content, dried fruits are an important source of antioxidants in the
diet (Wu et al., 2004; Vinson et al., 2005). Antioxidants can lower oxidative stress
and so prevent oxidative damage to critical cellular components. Dried apricots and
peaches are also important sources of carotenoids. These compounds not only are precursors of vitamin A but also have antioxidant activity. Dried fruits such as dried
plums provide pectin, a soluble fibre that may lower blood cholesterol levels (Tinker
et al., 1991). Dried fruits such as raisins are a source of prebiotic compounds in
the diet. They contain fructooligosaccharides like inulin, naturally occurring fibrelike
carbohydrates that contribute to colon health (Camire and Dougherty, 2003). The chemical structures of some of these useful compounds for human health are shown in
Figure 1.2.
Since drying foods may affect the chemical structure of their components, some
of which could be undesirable, it is important to keep a balance between how fast or
efficient the drying process is in terms of energy usage or product throughput and quality
requirements. Sometimes, faster processing is not necessarily better. Like food material,
all the solid-form products, natural or processed, have interesting structures (including
microstructures) and qualities.
It is worthwhile noting that removing water in its liquid form from a solid structure
is often (perhaps better) called dewatering, which induces solid structural changes
around or near where the water used to be. In a more general situation, water removal
may also be called dehydration, especially when talking about materials of partially or
wholly biological origin.
In this book, drying is mostly referred as those processes that use gas as a drying medium so the water comes out of the material as a gas (water vapour). Later in
this book, other forms of drying, such as vacuum drying or even steam-aided drying, may be employed using the reaction engineering approach (REA) when it is
appropriate.
Furthermore, roasting (coffee, for instance), baking (biscuits and bread, etc.) and
heating of moist material (e.g., detoxification of wood) may be considered extensions of
the concept of drying, and their purposes are for improving material performance while
the moisture effect simply cannot be ignored.
(a)
OH
OH
O
HO
OH
HO
HO
OH
OH
(b)
CH2OH
CH2OH
O
OH
O
OH
CH2OH
O
OH
OH
OH
OH
OH
CH2OH
CH2OH
O
O
OH
OH
OH
CH3O
OH
O
OH
CH2OH
HO
CH3O
O
O
O
O
OH
CH3
OH
O
OH
OH
OH
OH
OH
(c)
H
Gly
lle
val
Glu
Gln
Cys
Cys
Cys
Thr
Ser
Ser
Leu
lle
Tyr
Gln
Leu
A-chain
Glu
HO
Phe
B-chain
Val
Asn
Gln
His
Leu
Asn
Cys
Tyr
Gly
Cys
Gly
Ser
Val
His
Leu
Gly
Leu
Tyr
Phe
Leu
Phe
Asn
Cys
Val
Glu
Ala
Glu
Arg
Tyr
HO
Thr
Lys
Pro
Thr
Figure 1.2 Chemical structures of some chemicals: (a) 1, caffeic acid; 2, gallic acid; 3, vanillic
acid; (b) 1, cellulose; 2, starch; 3, pectin; (c) human insulin.
Introduction
The powering mechanisms for drying can be solar radiation, electricity, steam,
microwave and ultrasound, amongst others. In all drying operations, energy consumption is a critical issue in modern times; yet, in the wider range of practical interest,
people are more concerned about the quality of the products. For high-value products
often related to nanotechnology these days, nanostructure and microstructure aspects
and functionalities are an increasing concern for both the researchers and marketers. In
food and pharmaceutical products, these aspects are even more important as they affect
the metabolism and health of our organs.
Therefore, as far as drying is concerned, it provides an excellent example of process
engineering interacting with product quality. In chemical engineering, the interactions
may be considered systematically with the ideas of chemical process engineering versus
chemical product engineering (Cussler and Mogoridge, 1997).
One may see engineering as a terminology which differs from technology. Engineering should involve developing the predictive tools (evaluating a building design using
established mathematical analysis and then building it accordingly), which need good
mathematical descriptions of the physics and chemistry that go on into the relevant processes, much like in modern times, where calculations are used to investigate the validity
of constructing an architecturally designed building before actually building it; i.e., civil
engineering. There is no question that drying technology needs to be made predictive:
this is the notion of drying engineering. In fact, for design and optimisation of the
drying processes, one needs accurate and robust mathematical models. Better still, some
of the models can actually help us explore drying mechanisms (physics, chemistry and
biochemistry), investigating the scientific aspects associated with the drying phenomena. Drying models are usually referred to as models that describe the mass and heat
transport within the material being dried, as the exterior conditions are already nicely
covered by the conventional heat and mass transfer and momentum transfer theories. The
boundary conditions are intended to connect the drying models with exterior transport
theories.
Simple, yet effective, mathematical models of drying are welcomed by practitioners
or engineers. When modelling multiphase flow in a spray dryer, for instance, one might
need to mathematically track thousands of particles of different sizes travelling inside
the drying chamber to make the simulation more realistic. If one has to solve the spatially
distributed variables (like water content) inside each particle, the computational effort
is great; hence, the whole exercise incurs a high expense. There is also a high likelihood
of computational instability. However, if one does not need to solve for the spatial
distributions of moisture, temperature and species distribution within each particle, one
only needs to integrate a model over time to obtain the average water content, temperature
and average concentrations of species for each particle; the computational effort is much
smaller.
For a large piece of material being dried, a lumped drying model that may predict
drying history accurately without the need to resolve the spatially distributed parameters
is also very handy for practical purposes. Especially when many such pieces of the same
material are placed or stacked in a large chamber to dry (wood stacks for instance),
the airflow patterns around these pieces are already complex and need significant
computational input. The lumped models can be implemented using simple software
such as an Excel spread sheet and sometimes even a simple, programmable scientific
calculator.
To be relevant to the specific material of concern, every drying model proposed needs
careful experimentation to establish the required model parameters (constants) and the
model predictive power. In other words, these constants are mostly material specific.
For a diffusion-based drying model, for example, effective mass diffusivity is the key
parameter. It may be both water content and temperature dependent, making accurate
experimental determination and data analysis difficult due to non-uniform temperature
distribution when it comes to relatively large materials.
If there was one model that incurred only minimal experimental effort and less demand
for lab facilities, that model would be welcomed by industry. Relevant experiments have
to be accurate, possessing the required resolution, and simple, robust apparatuses and
simple operating procedures are desirable.
In many ways, the idea of the REA, apart from its scientific merits and the authors
own desires to make a novel model at that time (1996), is an outcome of these rationales.
1.2
A microstructural discussion of the phenomena of drying

moist, porous materials
Drying, the process of water removal, can affect the chemical composition of where the
water molecules stayed within the domain marked by the material surfaces before drying.
The removal of water molecules leaves vacant spaces, which may be filled partially or
totally by a nearby species. These movements should affect the microstructure formation of the material being dried. These movements may include rearrangement of solute
molecules (in drying liquid in particular) and shrinkage (with shell formation or hard
surface formation as well) and solid structure breakdown (crack formation, for example).
Indeed, then a solution (droplet) is dried to form a particle, in a gaseous medium the solid
surface formation (chemical composition rearrangement and microstructure formation)
is affected significantly by water removal (Chen et al., 2011). The structural changes,
once ocurring at nanostructure/microstructural levels, can affect the bioactivity of the
biological species, i.e., cells or microbes. The removal of water can cause permanent
movement of the soft structures that support the physicochemical structure of the cell
wall or cell membranes and cell contents such as genetic material, enzymes, or proteins
within, causing irreversible damage. This damage could be minimised if there was
another structure-supporting material such as sugar (this can be done by infusing the
products, mostly food materials, with sugar molecules in osmotic treatment) which
may replace the water molecules to uphold bioactivity. It is also possible that the
dryingconcentration process alters the ionic conditions, which may be more suited for
cells survival. Drying, in high-temperature instances, with still-considerable moisture
content, can cause proteins to denature, thus affecting the solubility and heat stability
of protein products. The colour of the food material can be altered during drying, especially when heat is added. Here, Mallard reactions may be responsible where proteins
Introduction
Temperature
profile
Liquid content
Heat flux
Vapour flux
Symmetry
Mixture of water
vapour and air
Uniform
capillary
assembly
Figure 1.3 Air drying of a capillary assembly (a bundle) which consists of identical capillaries
(diameter and wall material) a scenario of symmetrical hot air drying of an infinitely large slab
filled with the capillaries (modified from Chen, 2007); the air flows along both sides of the
symmetrical material.
and carbohydrates are present, such as in the baking of bread. However, the brown
colour is welcomed by consumers due to perceptions of a traditional, wholesome and
cooked appearance.
Before we proceed with the REA model concept and theory, we first discuss the drying
itself in relation to microstructures to provide an important scientific background, which
makes drying more relevant to modern material science. Microstructure is a relatively
new term compared to the classic theories of drying. There are a number of versions
of academic descriptions on how drying proceeds into initially moist materials. Here,
an intuitive, microstructural view of the gas- (hot air) aided drying process for moist
porous media is presented, which is simplified from Chen (2007).
First, we look into a hypothetical scenario and ideal case, where the initial temperature
of the moist material is slightly lower than the wet-bulb temperature of the drying
medium for an ideal capillary system, as shown in Figure 1.3. The directions of heating
and water vapour transfer are opposite each other. The capillaries here have identical
diameters at the micro level and the walls are hydrophilic, with no interexchange of
heat and mass across the capillary walls (simply, the walls are impermeable), assuming
they were initially completely filled with water and evaporation starts to happen. The
evaporation occurs uniformly here for all tubes in the same convection condition, at their
exits, provided the convection condition is the same everywhere along the side of the
assembly. There will be an obvious receding front of the liquidgas interface moving
inwards as drying proceeds. The thickness of the moving evaporation front will reflect
the meniscus of the liquidgas interface.
A mixture of complexity arises for this ideal system as different diameters of the
capillaries and permeable walls (e.g., membranes) are involved. The interexchange of
heat is possible across the capillary wall and the evaporation rates among the tubes
(under the same drying conditions applied at the exits) now differ. There will be nonuniform receding liquidgas interfaces among the tubes, giving a distribution of the
average liquidwater content (averaged over the lateral direction) along the horizontal
direction, broader than that which is shown in Figure 1.3. Furthermore, if the walls are
made of materials that are hygroscopic, water-molecular movement or liquid spreading
along the wall surfaces is also possible.
Though the interexchange of moisture and heat between the tubes may attempt to
even the evaporation rates and liquid water content, a broader distribution of the liquid
water content is still expected. Furthermore, due to the extended liquidgas interfaces,
evaporation will not occur only at the meniscus. Evaporation will happen in a region of
finite dimension: the occurrence of an evaporation zone at the micro- as well as at the
macro-level.
In a non-ideal situation, such as in a normal porous media with pores (open and closed)
and interlinking channels, the likelihood of the occurrence of a sharply receding liquid
front is reduced. In other words, the situation shown in Figure 1.3 is not common.
When the system is mixed at the micro-level (a more realistic situation for a practical
porous media), capillaries of various dimensions are oriented in many directions and
interlinked or networked. Even locally at the micron level, the capillary diameter sizes
can be uneven. The mass transfer is multi-directional, following the laws of physics,
that is, following the directions of the driving forces. Locally and microscopically, the
receding front(s), if any, would be fuzzy, depending on the specific local microstructure
and hydrophilic (or not) nature. Liquid movement may be diffusive or driven by capillary
forces and travelling in relatively easier passages. For air and vapour transfer, certain
difficult (yet wet) patches may be bypassed by a main receding front, which may be
left to dry more gradually, thus forming a relatively wet region. Fingering phenomena
may be possible with relatively dry and relatively wet patches coexisting nearby. Here,
the sorption/desorption characteristics of the materials distributed should play key roles
as well.
The capillary walls thickness (the apparently solid structure) and the walls own
porous microstructures (yet another, smaller level of pore networks or systems),
and their unevenness in spatial distribution, add to the overall picture, making
the transition more uniform. The materials that create the walls of the microstructures
are also important as they can have quite different affinities towards water molecules
(these are reflected by their equilibrium isotherms or liquid water holding capacity at
the same relative humidity and temperature). Evaporation may occur mostly in the transitional region where the rate is dependent on the local driving force for vapour transfer.
The walls of the microstructures are also important as they can have quite different
affinities towards water molecules (these are reflected by their equilibrium isotherms or
liquid water holding capacities at the same relative humidity and temperature).
All these characteristics would make the liquid water content (averaged over these
microscopic regions) distributed over a region between the really wet core and boundary
Introduction
Mass exchange surface
Drying air
Liquid water
content
T
Ts
Temperature
Cl.s
Cv,s
Cv.
x=0
x = xs
Moist porous
material
Figure 1.4 Schematic showing a common scenario of air drying of a moist solid.
of the moist material. This leads to a spatial transition of water content distribution rather
than the sharp receding liquid waterfront. This is an important recognition of the drying
phenomena such as airgas drying. In freeze drying, on the other hand, sublimation
phenomena may induce a sharp solidgas interface. In high-temperature processes, even
for air drying, the powerful heat front may result in an apparently sharper front of water
movement than its counterpart when a much lower heat wave is encountered. However,
in this case, the previously mentione intuitive analysis is still valid, though quantitatively
it may become less influential.
Figure 1.4 shows a general depiction of air drying a moist porous material. One
may pay attention to the partitioning between the liquid phase and the vapour at the
mass exchange surface (the interface between the solid domain and the air). Into the
porous solid domain, a vapour phase can also coexist with air and moisture (liquid)
phases.
These arguments can be readily generalised to packed particulate systems where
the individual particles can have their own macrostructure and sorption characteristics
(see Figure 1.5), while the main voids (where easier vapour paths can be found) would be
the voids in between the packed particles. In fruits and vegetables, the cellular structure
plays a very important role as the cell walls present major water transfer resistance (see
Figure 1.6).
The perception of a moving (liquid) front or the (sharp) evaporation front can lead
to different approaches in drying modelling. Mass transfer from the sharp moving front
and the vapour exit surface is often modelled using a simple effective diffusion concept
10
Figure 1.5 Packed particulate material.
Intercellular
spaces
Cell content
Intact cell
walls
Figure 1.6 Cellular structures in plant material.
(with an expanding resistance layer). In general though, this is a simplification, which

is intended largely for mathematical modelling.
It is also interesting to note from the process shown in Figure 1.3 that, due to the
temperature distribution in the moist material being dried in a normal air drying situation
(where the air temperature is higher than the porous material being dried), it is not
necessary to have the highest water vapour content at the innermost boundary where
liquid water content starts departing from the initial value. The vapour concentration can
be higher than the boundary value, or else there would be little or no drying. It is possible
to intuitively reason that there is a hump that can exist somewhere in the transitional
region of the liquid water content (Chen, 2007). A condensation mechanism also may
Introduction
11
exist in the region marked in the same spot as an area of uncertainty. The process of having
a high temperature and low humidity in the same air would induce an inward transport
of vapour as well as one that goes outwards (thus, drying is evident). Furthermore, a part
of the water evaporated in the lower part of the transitional liquid water content region is
transported into the structure and condensed at the lower temperature location, as long
as there is porosity (spaces for vapour to go into). This is an interesting phenomenon, as
it is clearly a more effective heat transfer mechanism than just heat conduction. Hence,
this phenomenon helps increase the temperature of the core wet region more rapidly,
which by itself has higher heat conductivity due to the higher water content (lower or
no porosity). This mechanism has an impact on the preservation of active ingredients
such as probiotic bacteria encapsulated inside a wet porous matrix subjected to drying.
The structure and the porosity have a large impact on the thermal conductivity of this
relatively dry layer. Conversely, the porosity and structure inside the region between lines
1 and 2 are also important, affecting vapour transport in this area. One would expect
this region to have lower porosity (thus, a lower vapour transfer coefficient the vapour
diffusivity). In addition to this, the vapour transfer and condensation mechanism, as
mentioned earlier, would make this heat conductivity effectively even higher. The lowest
liquid water content near the solidgas boundary should be determined by the nature of
the material and the drying air conditions through the equilibrium water content concept
(equilibrium isotherms) (Chen, 2007).
As mentioned previously, to a great extent increasing temperature may make the trend
of the liquid water content steeper; and a waterfall-like behaviour, where the vapour
wave is apparently moving inwards and a more tidal-like liquid water content versus
distance profile emerges. Furthermore, how the material swells and shrinks locally will
have an impact on the dried products quality.
This description has been supported by the micro-scale transient observations using
magnetic resonance imaging (MRI) conducted mainly at low or moderate temperatures.
A number of studies have directly targeted moisture transfer (Guillot et al., 1991; Hills
et al., 1994; Bennett et al., 2003; Mantle et al., 2003; Reis et al., 2003; Ruiz-Cabrera
et al., 2005a,b). There is no sharp front of evaporation observed in the latest MRI studies.
The spread of the lowering liquid water content as drying proceeds relies on capillary
diffusion of liquid water (Reis et al., 2003). The moisture transfer or transport devices,
units such as capillaries, inter-cellular spaces, voids and channel networks between
packed particles (which themselves may also be porous presenting another, perhaps
finer, level of transfer devices or units), all naturally possess non-uniformity. The spread
of the evaporation zone, or a transitional, mushy zone, from the still very wet core
and the already dried surface region is, therefore, expected. The pre-treatment (soaking)
using surface-active reagent solutions may help accelerate the water transfer process.
This understanding has been particularly helpful in supporting the ideas of revising
the conventional Biot and Lewis number calculations when air drying is of interest,
so conditions for model simplification can be made more realistically, and the model
concepts may be discerned with more quantitative support (see later sections on Biot and
Lewis number analyses). The effective diffusivity functions published in the literature
can be compared and discussed based on their relationships with the scientific discoveries
12
with MRI or other insightful tools. It is now recognised that the material microstructure
and its nature (composition, pore sizes, etc.) is interactive with transfer phenomena and
formation of microstructures, which has so much to do with the speed and schedule of
the transfer processes. Drying induces shrinkage, which is the most apparent structural
change visible. As drying occurs into the pore channels in the material, the water
remaining clamps on the structural walls to create the pulling strength that drags the
structure in, so to speak.
In todays modelling exercises (also refer to Chapter 3), more and more studies are
focused on treating materials according to their structural makeup. In biological material
as an example, domains in different scales of length may be divided according to the
structural characteristics, such as structural biology (see Figure 1.6 for modelling drying
of corn kernels as an example), and solved for moisture and energy transfer with their
own specific properties (Takhar et al., 2011). This approach has been termed the hybrid
mixture theory-based moisture transport and stress development. The diffusivities for
each domain were established separately as functions of temperature (Arrhenius law)
and water content (power law) (Chen et al., 2009).
The geometry of corn kernels shown in Figure 1.7 is a complex three-dimensional one.
The computational domains representing different parts of the biological structure were
captured using X-ray computed tomography (CT) and the scans were performed using
an Xradia micro-XCT scanner (Xradia Inc., Concord, CA, USA) at voltage and power
settings of 40 kV and 4 W, respectively. For imaging, a total of 253 two-dimensional
slices were obtained with a voxel size of 2.7392 m in the x, y and z directions. Some
other techniques were also to convert this information into a digital format to be
used in simulations from the Comsol Multiphysics package. Here, the geometry was
rescaled to generate suitable computational meshes (Takhar et al., 2011), while the threedimensional distributions of the computed moisture content, temperature and the stress
are visual and useful to help understand the physics involved. The predictions of the
average moisture content during drying, especially at higher temperatures such as 85 C
or at lower temperatures such as 29 C, were not accurate. The authors had attributed
these to the extrapolations of diffusivities evaluated at a narrower temperature range, i.e.,
3565 C, to both the higher and the lower temperatures explored in the simulations.
Figure 1.8 shows compression of an early wood spruce, which shows how the material
point method (MPM) can be used to describe the movement of the shrinking material
(Frank and Perre, 2010). MPM has been defined as the domain of interest (initial moist
solid domain, for instance) treated as a collection of material points p = 1, 2 . . . np (Sulsky
et al., 1994; Sulsky et al., 1995). Each material point carries its own properties such as
position, velocity, acceleration, strain and stress (basically the Lagrangian approach).
This method allows the finite-element discretisation of rather complex material shapes
to be made, based on two- or three-dimensional images, in a robust and efficient way.
As a result, it is now possible to handle such complex material structures at the plant
cell level from images taken at various resolutions. For instance, an optical microscope
commands a spatial resolution of 0.5 m with an acquisition time of 100 frames per
second; IR microscope 10 m, a few frames per second; confocal microscope 0.2 m, a
few frames per second; Raman microscope 0.2 m, a few frames per second; scanning
A stack of 253 slices obtained using

micro-CT scanning at a voxel size of
2.7392 microns in x, y and z directions
3D surface rendering
using Avizo package
Exporting the
Avizo geometry to
Comsol using
Gmsh package
Rescaling to Lagrangian coordinates and

meshing using Comsol package
Digital cutting of 3D geometry

across the plane of symmetry
(a)
Figure 1.7 (a) Generation of computational domains of corn geometry for the hybrid mixture
theory of corn kernels (adapted from Takhar et al. (2011)). (b) The simulated results (isosurface
plots of corn moisture content) for a variety of drying conditions. [Reprinted from Journal of
Food Engineering, 106, P.S. Takhar, D.E. Maier, O.H. Campanella and G. Chen, Hybrid mixture
theory based moisture transport and stress development in corn kernels during drying: Validation
and simulation results, 275282, Copyright (2012), with permission from Elsevier.]
14
T29 C, RH 48%
T67 C, RH 15%
0.2112
0.1579
0.2061
0.1529
0.201
0.148
0.1959
0.1431
0.1908
0.1382
0.1857
0.1332
0.1806
0.1283
0.1755
0.1234
0.1704
0.1185
0.1653
0.1135
0.1602
0.1086
0.1551
0.1037
0.15
0.0988
0.1449
0.0938
0.1398
T48 C, RH 25%
0.0889
0.1181
0.094
0.1142
0.0911
0.1102
0.0881
0.1062
0.0851
0.1023
0.0822
0.0983
0.0792
0.0943
0.0762
0.0904
0.0733
0.0864
0.0703
0.0824
0.0673
0.0785
0.0644
0.0745
0.0614
0.0706
0.0584
0.0666
0.0555
T85 C, RH 14%
0.0626
(b)
Figure 1.7 (cont.)
0.0525
Introduction
15
Figure 1.8 Wood cellular structures employed in pore-network modelling of drying of wood.
[Reprinted from Drying Technology, 29, P. Perre, A review of modern computational and
experimental tools relevant to the field of drying, 15291541, Copyright (2012), with permission
from Taylor & Francis.]
electron microscope (SEM) 3 nm, several seconds for one frame; transmission electron
<1 s; atomic force microscope (AFM) <1 nm (x-y plane) (and
microscope (TEM) 1 A,
<1 A at z axis), several minutes (Perre, 2011).
1.3
The REA to modelling drying
1.3.1
The relevant classical knowledge of physical chemistry

In the forthcoming sections, the ideas and the development behind the REA for modelling
drying process will be briefly described. A short summary of the classic modelling
approaches is given first, and the newer REA approach will be subsequently introduced.
To begin the discussion on the REA concepts, we need to outline physical chemistry
principles of chemical reactions with a little originality on our part, in order to form
the basis of the REA idea. The most prominent idea in reaction engineering is the
expression of the chemical reaction rate. A chemical reaction rate of species A, involved
in the reaction of two species (A and B) yields a product commonly expressed as:
dc A
= k A cnAA cnBB ,
(1.3.1)
dt
where nA and nB are the orders of reactions associated with species A and B, respectively,
kA is a rate constant, and CA and CB are the concentrations of species A and B, respectively.
This rate constant increases with temperature T, approximately increasing by two to
16
four times with a temperature increment of 10 K. The relationship between a reaction

rate constant, k, and temperature, T, has been generically described using the famous
Arrhenius equation (Fogler, 1992):
d ln k A
EA
,
=
dT
RT 2
(1.3.2)
where EA is the activation energy of the reaction (J mol1 ). This means that the
value of ln kA change against temperature is proportional to the value of EA . The
larger activation energy EA is, the more sensitive the reaction rate towards temperature
change. EA can be variable against temperature when multiple reactions are occurring
simultaneously.
When the range of temperature is not large, EA may be considered a constant. In this
case, which is more commonly adopted in real life, the rate constant is expressed as:
k A = k Ao eE A / RT ,
(1.3.3)
where kAo is a constant.

Molecular mechanisms of evaporation and condensation at free liquid surfaces under
the vapourliquid equilibrium are investigated with molecular dynamics computer simulations for argon and methanol. Vapour molecules colliding with the surface are in the
condition of almost complete capture for both fluids but, in the case of methanol, molecular exchanges strongly affect the evaporationcondensation rate (Matsumoto et al.,
1995) (presumably water evaporationcondensation behaves similarly). Evaporation
condensation is one of the fundamental processes in many fields of science and engineering. For decades, various experiments have been done to measure the absolute values
of evaporation or condensation rate, but there still remains a lack of knowledge of the
underlying molecular mechanisms. There is still a lot of room available to explore the
fundamental aspects of evaporationcondensation. This phenomenon is further complicated by the presence of species other than water during the drying process. One could,
however, understand intuitively that removal of water (in vapour) is an energetic process
involving latent heat. One has to put in energy to activate water molecules that could
become free when the material is dried. When the water molecules are not associated
with the solid or solute molecules and they stay in the bulk liquid domain, their interactions between one another would be as if there was no solid. Evaporating them into
a gaseous form would be done mainly by providing energy sufficient to overcome the
latent heat of water evaporation. Of course, another condition is that the vapour can be
transported into the gas medium or into a vacuum. It is interesting to note that, since the
condensation process is spontaneous and does not need to overcome a kind of activation energy (an energy barrier), the activation energy for condensation is zero, and the
activation energy of the evaporation process for pure water would be equal to the heat
of reaction (the latent heat of water vaporisation); this is to say, the forward reaction
activation energy is less than the reverse reaction activation energy, which is zero for
condensation. This was the basic motive and idea in formulating the drying rate as a
competition between evaporation and condensation (to be described later).
Introduction
1.3.2
17
General modelling approaches

Before REA there were three main general approaches in the literature to formulating
drying models, summarised next (Chen and Xie, 1997):
1. The concept of a characteristic drying rate curve (CDRC model), which recognises
different drying stages; e.g. the constant rate period (which may also be called the
unhindered drying period, where the internal transport of the moisture does not
affect the surface evaporation) and falling rate (hindered drying) period(s).
2. The distributed-parameter models, mostly those based on volume averaging concepts,
employing coupled heat and mass diffusion equations involving heat conductivities
and mass diffusivities, etc.
3. The empirical models obtained entirely by simple or multivariable regression methods
(often, a series of known time-dependence functions such as the Page model etc. have
been used to simply correlate the weight loss over time).
There are many models in categories (1) and (3). Both (1) and (3) may be regarded as
the lumped drying models in that they do not need to solve for the spatially distributed
moisture content and temperature. For (2), there have been a number of continuumtype mechanisms proposed and the corresponding mathematical models established.
These include effective liquid diffusion, capillary flow, evaporationcondensation, dual
(temperature, water content gradient) and triple (temperature, water content and pressure
gradient) driving-force mechanisms by Luikov (1986), another dual driving-force mechanism by Philip and De Vries (1957) and De Vries (1958), and, finally, the dual-phase
(liquid and vapour) transfer mechanism of Krischer as summarised by Fortes and Okos
(1980).
Whitaker (1977; 1999) has proposed detailed transport equations to account for the
macro- and micro-scale structures in biological materials. Three-phase (solid, liquid and
vapour) conservations and their local volume-averaged behaviours are considered. The
mechanisms for moisture transfer are largely the same as those proposed by Luikov
(1975; 1986) and Philip and De Vries (1957), except that the small-scale phenomena
(local pores, pore channels, shells, voids, etc.) have been taken into account. This theory
is based on a known (or pre-assumed) distribution of the macro-scale and micro-scale
unit structures, which allow local volume averaging to be carried out. Pore-network
models have become popular in recent years, as the concept of multi-scale and multiphysics is expanding, e.g. coupling meso-scale problems and equipment scale problems
(Perre, 2011). When a local thermodynamic equilibrium is not attained, the time scales
usually overlap. This is a real multi-scale configuration and challenging in terms of the
great demand in computational power and mathematics. Several scales can be considered simultaneously, ranging from simple exchanges between macroscopic phases to a
comprehensive formulation, in which the time evolution of microscopic parameters and
microscopic gradients are considered over a representative elementary volume according to a recent review by Perre (2011). These models are often mathematically highly
involved.
18
For (3), the models usually have little physics explained and it is difficult to extract
any fundamental information, though some of them have attempted to show some
physical significance on somewhat weaker grounds. On the other hand, some accurate
time functions under category (3), such as the Page model, can be used to fit the data
points to generate accurate drying rate data sets for other purposes. Some models of a
comprehensive nature will be described in detailed mathematical terms in later chapters
when the authors compare the performances of REA in various practical cases against
well-known models.
In modern times, expanding what we have already in category (2), the comprehensive
modelling of drying in a spatially distributed manner (or, say, in a discrete manner)
has frequently involved more pore-level information (pore-network models) and mathematical techniques, which do not require volume-averaging procedures of some sort.
The respective pore networks can be used to systematically study the influence of the
structure of a porous medium on drying kinetics. There are potentials of this discrete
modelling for use as a virtual laboratory to improve our understanding of how structures correlate with properties and how better products may be developed (Tsotas and
Mujumdar, 2007).
In general, several scales of problem are involved in drying, modified from those
summarised by Tsotsas and Mujumdar 2007:
The molecular scale (water molecules interact with each other and with the other species in
the liquid or gas, and with the solid surfaces), the pore scale (the smallest entity for expressing
transport phenomena within the drying particles or single bodies), the particle scale (single drying
body can be identified; this larger scale can include rather large particles such as wood boards
stacked and dried in an industrial drying kiln), the particle-system scale (the equipment is designed
and properly operated at this level; the interactions between the particles, the gas flow and the
apparatus are considered at this scale), and finally the process-system scale (the drying system
interacting with other engineering systems ensuring the proper operation of the entire production
plant).
1.3.3
Outline of REA
Following the basic descriptions of physical chemistry in relation to the expression of
chemical reaction rate as described earlier, the REA was proposed by the author (XDC)
in 1996, and in the subsequent year, a couple of papers were published (Chen and Chen,
1997; Chen and Xie, 1997; Chen and Pirini, 2004). The idea was also partially inspired
by two pieces of information: a paper published by Professor Brian Gray (1990) and
a series of works on the long-established CDRC model (van Meel, 1958; Keey, 1978;
1992).
In 1990, a mathematical model for a wet-combustion system, the exothermically
reactive (porous solid) system, which is also influenced by the presence of water, was
published by Professor Gray, who was a Senior Professor of Mathematics at School of
Chemistry in Macquarie University, Sydney, Australia. However, before that landmark,
the role of water on the exothermic (solid) systems, such as spontaneous heating or spontaneous combustion, had been proposed in several models from more of an engineering
Introduction
19
perspective (Chen, 1991). However, the role of water as a direct participant in chemical
reactions (in oxidation in particular) had not been considered quantitatively. In the paper
by Gray, he added a term in the mass balance and energy balance, respectively, which
accounts for the direct participation of water in chemical reaction (exothermic), in conjunction with the water effect through evaporation (liquid to vapour) and condensation
(vapour to liquid).
In the same paper, Gray wrote the following equations to describe a wet-combustion
system where the temperature of the (combustible) solid material is assumed to be
uniform, as is the water content within the solid matrix:
dx
= c (1 x) e xe/u w xew /u ,
d
(1.3.4)
where x, u and are the dimensionless liquid water concentration, temperature of the
(combustible) material, and time, respectively. Term x is defined as the current water
content (g) divided by the initial water content (g) that is available in liquid form (in fact,
it is the total water content in the system boundary), i.e. mw /mw,o . The model constants
are represented by in Equation (1.3.4); this system is also, in the theory of thermal
ignition and combustion, known as the Semenov approach, signifying the uniformity
of the variables throughout the material of concern (Bowes, 1984). This is a useful
assumption, which paves the way for a large number of mathematical analyses that have
considerable physical meaning relevant to practical conditions.
In Equation (1.3.4), the first term on the right-hand side represents the condensation
process, denoted by subscript c, and the second term evaporation, denoted by e. The last
term on the right-hand side represents the consumption of water due to the water-induced
or -involved chemical reaction (exothermic), the wet oxidation, denoted by w. When we
remove this wet-oxidation term, an inert system, we have:
dx
= c (1 x) e xe/u .
d
(1.3.5)
Equation (1.3.5) represents the water exchange (condensation less evaporation) between
the moist material and the environment/surrounding. (1 x) signifies a conservation of
the total water content in the domain of interest. The evaporation term is considered
to be first-order as far as water reactant is concerned. The most important description
of the evaporation term is used in the Arrhenius dependence function (a well-known
function in physical chemistry), i.e. e u . The in the context of Grays analysis denotes
the dimensionless latent heat of water vaporisation. This may not have been completely
new, even at that time, but certainly was a great approach to describe the physical picture,
as we tried to understand moisture movement in and out of a porous solid matrix. Gray
(1990) and Gray and Wake (1990) took full advantage of this simple and effective
formulation to fruitfully explore the wealth of the behaviours in the wet-combustion
system. In any case, a steady state can be attained for the system described by Equation
(1.3.5) as:
c (1 x) = e xe/u .
(1.3.6)
20
x
1
Figure 1.9 Schematic illustration of the effect of temperature on final liquid water content
(qualitatively derived from Equation 1.3.6).
Rearrange to give:
1x
e /u
e
,
=
x
c
(1.3.7)
which essentially suggests that as the material gets hotter, the liquid water content
in the system reduces more. The activation energy of the evaporation reaction was
assumed to be the latent heat of water vaporisation. This has good intuitive grounding (Chen, 1998). As the dimensionless temperature u gets infinitely large, x becomes
zero (see Figure 1.9). This system, however, does not seem to capture the physics
behind when water content in an environment becomes zero; even when temperature is moderate, the liquid water content inside the material can also become zero.
In other words, this system might have neglected another dimension of the drying
system.
On the practical side, a drying kinetics approach, the CDRC model, had been
employed in order to design dryers and optimise drying operations for improving
energy efficiency. In his developing understanding of drying operations and the fundamentals of moisture transfer, van Meel (1958) postulated that, when working with
convective batch dryers, a single characteristic drying curve could be deduced for a
moist material. This model reflects the nature of the drying rate curve(s) shown in
Figure 1.10(a). It is empirical but has been successful in correlating the drying kinetics of small particles. The model assumes that, for any given sample water content,
a unique relative drying rate exists. This rate is relative to the initial unhindered drying rate and is independent of the external drying conditions (refer to Figure 1.10(b)).
These conditions include the temperature, humidity and pressure of the drying gas.
The model also implies that a region exists where, for a period of time, the rate remains
unhindered.
The physics behind what is shown in Figure 1.10(a) has been described scientifically by
Perre, Remond and Turner (2007). For non-deformable materials (negligible shrinkage)
such as building materials and natural mineral products (including fragmented rocks),
the relationship between porosity and moisture content is obvious. As drying proceeds,
the moisture content is simply replaced by drying gas. For highly deformable products
such as food, the moisture removal is related to both volume reduction and porosity. It
is necessary to know whether the loss of moisture turns into volume change or into an
increase in porosity (Perre and May, 2001). Considering these factors one can interpret
21
Introduction
Drying flux (kg.m--2.g--1)
Constant
drying flux
period
Cooling down
Falling
drying flux
period
Warming up
Critical average
water content
Equilibrium
water content
0.0
Xc
(a)
Average content (X)
Critical point (1,1)

Hindered
region
Unhindered region
0 Equilibrium
point
(b)
Figure 1.10 (a) Drying flux versus average water content X ; (b) the CDRC (characteristic drying
rate curve). [Reprinted from Chemical Engineering Science, 9, D.A. van Meel, Adiabatic
convection batch drying with recirculation of air, 3644, Copyright (2012), reprinted with
the results on constant drying rate, which should be the constant drying flux, properly.
The mass exchange surface area is an important parameter involved here.
Here, the relative drying rate is defined as:
=
N
,
Nc
(1.3.8)
where is the relative drying rate, N is the instantaneous drying rate (best defined to be
the drying flux, kg m2 s1 ) and Nc is the drying rate at the critical condition (i.e. when
the drying regime is in transition between the unhindered rate and falling rate periods;
at the critical water content, Xc ).
22
The characteristic moisture content (dimensionless water content) is defined by the

following equation:
=
X X
,
Xc X
(1.3.9)
where X is the average water content (on dry basis) at any time t, X is the equilibrium
water content (on dry basis) and Xc is the critical water content. The drying rate is
normalised to pass through the critical point and the equilibrium point, denoted by
points (1,1) and (0,0), respectively in Figure 1.10(b).
According to Keey (1992), the characteristic curve method is attractive since it leads
to a simple lumped-parameter expression for the drying rate, in the following form:
N = Nc .
(1.3.10)
This expression has been used extensively as the basis for understanding the behaviour
of industrial drying plants. Because of the simplicity of the parameters used, this has
been employed in some industrial applications.
Basically, the general form of the CDRC (Keey, 1992) is expressed as follows:

= f X , if X X c ; = 1, if X > X c .
(1.3.11)
There are a number of successful applications of this approach, especially for materials
of small dimensions such as particles or thin layers (Keey, 1992). Keey (1992) has further
specified that a unique characteristic curve can be established at Kirpichev numbers less
than 2 or, in effect, when the material is thinly spread and permeability to moisture is
high (i.e. the material has a large moisture (vapour) diffusivity).
The Kirpichev number is given as:
Ki =
Nc
,
s X o Dv,eff
(1.3.12)
where is the thickness of the sample (m), s is density of the dry solid (kg m3 ), Xo
is the initial water content (on dry basis) (kg kg1 ) and Dv,eff is the effective vapour
diffusivity (m2 s1 ). According to Keey (1992), in many cases the drying curve can be
fitted using a simple algebraic equation over a limited moisture content range of interest
by:

X X
,
(1.3.13)
=
Xc X
where j is a parameter dependent on the relative difficulty of removing moisture from a
material. For example, j was found to be about 0.5 for cellulosic fibres (Langrish, 2008).
Equation (1.3.10) is a nice, simple and user-friendly expression, if accurate enough, for
the material of concern. The author was first exposed to the idea of the CDRC during
his Ph.D. study at Canterbury University in Christchurch, New Zealand. The exposure
of the author to CDRC was significant, due to the physical presence of the well-known
advocate of the CDRC approach (and indeed one of the greatest in drying research and
development), Professor Roger Keey at Canterbury University, at the time. During the
Introduction
23
same period of time, a visiting scholar from China, Professor Yuan Wu, was working
on the application of a variation of the CDRC approach in through-drying of wool in
Keeys laboratory (which was later published in Chemical Engineering Science, Wu and
Keey, 1995). However, being young at the time, the author did not fully appreciate the
usefulness of the CDRC approach, which only became apparent later in the authors
research career in drying.
According to the understanding gained so far, there are aspects of CDRC that need to
be noted:
(1) The critical water content(s) have to be determined experimentally, which are
known to be dependent upon drying conditions (temperature, humidity and
velocity).
(2) The quality of results of data reduction is not good, as in some cases the data points
can be quite scattered (for instance, with some large relative errors from the mean
in the falling rate period(s)).
(3) In addition, the mass flux (Nc ) calculations are sometimes based on the wet-bulb
temperatures (Twb ) for the gas phase.
Aspect (3) is illustrated as follows:

Nc = h m v,sat (Ts ) v, ,
(1.3.14)
where the material surface temperature Ts (K) may not be exactly the same as the wetbulb temperature Twb (K) and hm is the mass transfer coefficient (m s ). Here, v,sat (Ts )
is the saturated water vapour concentration at the surface temperature Ts (kg m3 ) and
v, is the water vapour concentration in the environment or in the drying gas medium
(kg m3 ).
A CDRC model is usually obtained from laboratory experiments under constant
external conditions, with moist materials of similar form and size to that of interest in the real industrial dryer situation. As mentioned earlier, one also needs to
note that the CDRC model must have accurate measurement, choice or prediction
of the constant rate and the maximum rate, Nc . An accurate estimation of the critical
water content as a function of the external drying conditions must be made, as well
as the ability (or, rather, lack of it) to reduce effective data for the relative drying
rates.
For CDRC, it is clear that the actual rate of drying (or drying flux) N is written as:
N = f ()Nc

f () v,sat (Ts ) v, .
(1.3.15)
To introduce the REA concept in mathematical terms, based on the conventional transport
phenomena theory, we write the vapour flux at the boundary (solidgas) as:

N = h m (v,s v, ) = h m R Hs v,sat (Ts ) v, ,
(1.3.16)
24
where v,s is the vapour concentration at the interface of solid and gas (kg m3 ) and RHs
is the relative humidity of the air (or gas in general) at the solidgas boundary, which
may be defined as:
v,s
.
(1.3.17)
R Hs =
v,sat (Ts )
During a drying process, this surface relative humidity, RHs , is reducing but is an
unknown quantity.
Equations (1.3.15) and (1.3.16) thus show a significant contrast between them. Equation (1.3.15) defines the instantaneous drying flux as a fraction of the maximum possible;
Equation (1.3.16) follows the widely accepted mass transfer expression with the first
term in brackets on the right-hand side of the equation being the product of RHs and
saturated vapour concentration at the interface. The difficulty in using Equation (1.3.16)
is how one can express RHs as a function of some known quantities.
The concept of REA has incorporated the more conventional approaches to expressing
mass transfer at the boundary, i.e. Equation (1.3.16). It is an application of chemical reaction engineering principles to establish a workable function for RHs . In this approach,
evaporation is modelled as zero-order kinetics with activation energy while condensation
is described as a first-order wetting reaction with respect to drying air vapour concentration without an activation energy (Chen and Chen, 1997; Chen and Xie, 1997). The
REA approach employs the Arrhenius equation in the evaporation term, which has its
origins in the paper by Gray (1990) but it differs markedly overall from what he proposed
(Chen, 2008). The REA approach offers the advantage of being expressed in terms of
a simple, ordinary differential equation with respect to time. This negates the complications arising from use of the partial differential equation (Chen, 2008). REA does
need accurate experimental data to determine model parameters, accurate equilibrium
isotherm and surface area measurement. It will be shown later that REA accommodates
a natural transition because of the smooth activation energy as a function of moisture
content (Chen, 2008). When activation energy for evaporation becomes higher than the
latent heat of water evaporation, free water should have already been removed (Chen,
2008).
As a lumped approach (REA was initially proposed as a lumped parameter model),
the drying rate (flux multiplied by surface area) of materials can be expressed as:
dX
= Nc A = h m A (v,s v, ) ,
(1.3.18)
dt
where ms is the dried mass of thin layer material (kg), X is moisture content on a dry
basis (kg kg1 ), X is the mean water content on a dry basis (kg kg1 ), t is time (s),
v,s is the vapour concentration at interface (kg m3 ), v, is the vapour concentration
in the drying medium (kg m3 ), hm is the mass transfer coefficient (m s1 ) and A is
surface area of the material (m2 ). The mass transfer coefficient (hm ) is determined based
on established Sherwood number correlations or is established experimentally for the
specific drying conditions involved (Incropera and DeWitt, 1990). Equation (1.3.18) is
basically correct for all cases of water vapour transfer out of a porous solid. In other
words, no assumption of uniform water content is made in this approach, even though it
ms
25
Introduction
began with mean water content. The surface vapour concentration ( v,s ) can be correlated
in terms of saturated vapour concentration of water ( v,sat ) by the following equation
(Chen and Chen, 1997; Chen and Xie, 1997):

E v
v,sat (Ts ) ,
(1.3.19)
v,s = exp
RTs
where Ev represents the additional difficulty in removing moisture from materials on
top of the free water effect. It was thought it would be excellent (and indeed lucky) to
be able to relate Ev to average water content of the material. In other words, this Ev
is ideally moisture content (X) dependent. T is temperature of the material being dried
(K). For a small temperature range, say from 0 C to just over 100 C, v,sat (kg m3 )
can be estimated with the following equation (Chen, 1998):

Ev
,
(1.3.20)
v,sat (T ) = K v exp
RT
where Kv was found to be 1.61943 105 (kg s1 ) and Ev was found to be
38.99 kJ mol1 . Ev is similar to the latent heat of water vaporisation illustrating the
physics involved (Chen, 1998). This is in line with the idea that evaporation is an
activation process, whilst condensation is not. The activation energy of the pure water
evaporation reaction is equivalent to the value of the latent heat of water evaporation, as
suggested earlier based on classical physical chemistry.
More widely in the range of temperature, one can use the following, which correlates
the entire range of the data (0 C200 C) summarised by Keey (1992) (also see
Figure 1.11):
v,sat = 4.844 109 (T 273)4 1.4807 107 (T 273)3
+ 2.6572 105 (T 273)2 4.8613 105 (T 273) + 8.342 103 ,
(1.3.21)
where T is temperature (K) based on the given data (Putranto et al., 2010).
The mass balance is then expressed as:

dX
E v
ms
= h m A exp
v,sat (Ts ) v, .
dt
RT
(1.3.22)
For small objects, such as particles or thin layer materials, the material temperature T
is approximately the same as the surface temperature Ts . Basically this happens when
the ChenBiot number is sufficiently small (Chen and Peng, 2005; Chen, 2007; Chen
and Mujumdar, 2008). In this case, uniform temperature can be assumed throughout the
material being dried so that one only needs to couple Equation (1.3.21) with a lumped
energy balance to govern the drying process.
The activation energy (Ev ) is determined experimentally by rearranging Equation
(1.3.22):

m s ddtX h m1 A + v,
,
(1.3.23)
E v = RTs ln
v,sat
0.60
Saturated vapour density (Kg.m3)
26
0.50
0.40
0.30
0.20
0.10
0.00
0.0
20.0
40.0
60.0
Temperature (C)
80.0
100.0
Figure 1.11 Saturated water vapour concentration in air under 1 atm (Equation 1.3.21).
where d X /dt is determined from experimental data on weight loss. It has been found,
based on practical experiences of using REA, that drying experiments for generating
the REA parameters need to be conducted where the air (or gas) humidity is very low in
order to cover the widest range of E v versus X . The dependence of activation energy
on moisture content can be normalised as:
E v
= (X X ),
E v,
(1.3.24)
where is a function of moisture content difference, Ev , is the maximum when the
moisture concentration of the sample approaches relative humidity and temperature of
the drying air:
E v, = RT ln(R H ),
(1.3.25)
where X is the equilibrium moisture content corresponding to RH and T which can
be related to one another by the equilibrium isotherm (Keey, 1992). It worth noting again
that, so far, the experiments for gaining the relevant Equation (1.3.24) have been under
very dry air conditions, so the final water content attained is usually low.
For the same material, the same initial water content and the same initial sample
size (sometimes different sizes do not matter), the relationship (Equation 1.3.24) may
be viewed as unique as many experimental results obtained under these conditions, but
different drying conditions for the same material, produced more or less the same trend
quantitatively (Chen, 2008). This aspect will be shown later in various applications
described in the forthcoming chapters.
Introduction
27
The REA parameters for drying of a material can be obtained from one good drying experiment and can then be applied to other different drying conditions (different drying air temperatures and air velocities) if the normalised activation energy
collapses to the same profile in these cases. However, the REA parameters should
be generated from material with the same initial moisture content since the activation energy has been found to be dependent on initial moisture content, too (Chen,
2008).
In many scenarios tested involving different materials to be described later in this
book, Equation (1.3.23) holds a very pleasant outcome indeed. Of course, other forms
of Equation (1.3.23) are possible and one should be worried if there is a temperature
dependence function or if a material structure parameter is involved.
When the temperature of the moist material being dried does not vary much within, a
uniform temperature may be considered (more quantitative assessment of this assumption can be found in a later part of this book that describes the modified Biot number
and modified Lewis number). This leads to:
Ts T ,
where T represents the mean temperature of the material (K).
This has allowed us to present REA energy balance in a lumped capacitance (Incropera and DeWitt, 1990) for the material being dried:
mC p

dT
dX
= h A T T + Hdrying m s
,
dt
dt
(1.3.26)
where h is the heat transfer coefficient (W m2 K1 ), T is the drying air temperature

(K), Cp is the specific heat of the sample (J kg1 K1 ), ms is the dried mass of the sample
and m is the mass of the material being dried, expressed as:

m = ms 1 + X .
(1.3.27)
The heat of drying, Hdr ying (J kg1 ), is more strictly expressed as a function of water
content:
Hdrying = f (X ) ,
(1.3.28)
which increases as water content X reduces and reaches a level as X approaches zero,
which can be much higher than the latent heat of pure water evaporation (Chen, 1992;
Chen and Stott, 1992; Chen, 1994; Chong and Chen, 1999). This value is usually
unknown unless separate experiments are done to obtain it, so in the present approach
this effect has been absorbed by the water content dependence function empirically
obtained for the activation energy (Equation 1.3.21). Nevertheless, the heat of drying
may be related to the so-called isosteric heat, which may be derived from the equilibrium
desorption isotherms theoretically obtained for the moist material of interest. It may also
be measured using calorimetric methods as summarised by Shen et al. (2000). The heat
of drying may be plotted qualitatively as that shown in Figure 1.12.
Heat of drying (kJ.kg--1)
28
Latent heat
of vaporisation
0
Water content (X)
Figure 1.12 Schematic diagram showing the heat of drying as a function of water content of
a porous solid of concern (when the water content is beyond the point where the heat of
drying becomes the latent heat of pure water evaporation, the water content may be
called free water).
One more interesting point to note is that in some of the literature (Rostami et al.,
2003), the rate of evaporation is expressed as:
dcw
= ko eE / RT cwn w ,
dt
(1.3.29)
where Cw is the concentration of water, E is the activation energy (J mol1 ), T is

temperature (K) and ko is the constant.
The order of the evaporation reaction, nw , is taken to be 1 (i.e. the first-order reaction). This, though not necessarily as accurate as the current REA model, may also be
considered to be in the category of the REA. The simulation that uses this formulation,
however, does not consider the phenomena of condensation.
REA has been used to model drying of a range of food materials such as pulped
kiwifruit leather, whey protein concentrate, lactose, skim milk powder, whole milk
powder, cream and mixtures of sugars (Chen et al., 2001; Chen and Lin, 2005; Lin and
Chen, 2005; 2006; 2007). Results showed that this approach models moisture content
and temperature profile along drying time very accurately. For example, modelling
of drying of an aqueous lactose solution droplet showed that the average absolute
difference of the weight loss profile was about 1% of the initial weight while that of the
temperature profile was about 1.2 C (Lin and Chen, 2006). Moreover, application of
the REA to model the drying of cream and whey protein concentrate showed average
absolute differences in weight profiles of 1.9% and 2.1%, respectively, while that of
temperature was about 3 C and 1.9 C, respectively (Lin and Chen, 2007). Modelling of
skim milk and whole milk powder by the REA was also robust and accurate (Chen and
Lin, 2005).
Introduction
29
The REA has also been implemented in computational fluid dynamics (CFD) based
simulations using a spray dryer for coupling the dispersed phase (droplets dried) and
the continuous phase (drying air) (Woo et al., 2008; Jin and Chen, 2009a; 2009b;
2010). CFD simulations using the REA can predict outlet air temperature and outlet
particle moisture content reasonably well compared to experimental data. In addition, the REA was also implemented to predict the evaporation zone, drying rate,
trajectory of particles, and deposition of particles in a spray dryer (Woo et al., 2008).
Application of CFD in conjunction with the REA to describe the performance of an
industrial-scale spray dryer in two and three dimensions has been conducted (Jin and
Chen, 2009a; 2009b). Patel et al. (2009) have extended the single solid component
approach of the REA to a composite REA model drying kinetics for mixtures of noninteracting solutes (maltodextrin and sucrose). The activation energy of the mixture
was determined based on mass fraction of each solute and its corresponding activation
energy. It was shown that the average relative error between experimental and calculated data was below 1.5% for droplet weight and below 3% for droplet temperature
(Patel et al., 2009).
After years of exercising the REA concepts, summarised by Chen (2008), it became
clear that, for large materials being dried, both temperature and water content vary
in space, so simple usage of the REA concept needed to be extended. In the past
few years, the approach has been divided into two categories: lumped-REA (L-REA,
described previously) and spatial-REA (S-REA, to be described in full in Chapter 3).
Putranto et al. (2011) have applied the L-REA to successfully describe the situations
of time-varying boundary conditions, such as intermittent drying. Following an initial
attempt by Kar and Chen (2010), where the L-REA formulation was employed as the
source term (evaporation and condensation) in the partial differential equation set that
governs the multiphase transport phenomena in drying of porous media (porcine skin,
to be exact, in their studies), Putranto and Chen (2013) have established a comprehensive set of partial differential equations (heat and mass balances) which has been
shown to be very helpful in simulating drying, referred to as the spatial-REA; the
S-REA.
In the forthcoming chapters, the applications of both the L-REA and S-REA are
demonstrated in detail using worked examples. Comparisons between the REA model
approach and the other more established models are given. Some physical insights of
the modelling approach are discussed where appropriate.
1.4
Summary
In this chapter, some general aspects of drying related backgrounds have been described
to set the scene. A brief review of the approaches to drying modelling has been given.
A brief outline of the physical chemistry aspects of reaction engineering/kinetics has
been provided highlighting the relationship between physical chemistry and the REA.
Finally, the REA has been described. The L-REA approach has been introduced, which
30
is the basis for the later development of the more advanced S-REA, both to be described
later in this book.
References
Bennett, G., Gorce, J.P., Kedde J.L., McDonald, P.J. and Beglind, H., 2003. Magnetic resonance
profiling studies of the drying film forming aqueous dispersions and glue layers. Magnetic
Resonance Imaging 21, 293297.
Bowes, P.C., 1984. Self-Heating: Evaluating and Controlling the Hazards. Elsevier, Amsterdam.
Camire, M.E. and Dougherty, M.P., 2003. Raisins dietary fiber composition and in vitro bile acid
binding, Journal of Agricultural and Food Chemistry 51, 834837.
Chen, X.D., 1991. The Spontaneous Heating of Coal Large Scale Laboratory Assessment and
Supporting Theory, Ph.D. thesis, University of Canterbury, New Zealand.
Chen, X.D., 1992. On the mathematical modelling of the transient spontaneous heating in a moist
coal stockpile. Combustion and Flame 90, 114120.
Chen, X.D., 1994. The effect of drying heat and moisture content on the maximum temperature
rise during self-ignition of a moist coal pile. Coal Preparation 14, 223236.
Chen, X.D., 1998. The fundamentals of self-ignition of water containing combustible materials.
Chemical Engineering and Processing 37, 367378.
Chen, X.D., 2007. Simultaneous heat and mass transfer. In Handbook of Food and Bioprocess
Modelling Techniques, S.S. Sablani, M.S. Rahman, A.K. Datta and A.S. Mujumdar (eds.). CRC
Press, Boca Raton, FL; Chapter 6.
Chen, X.D., 2008. The basics of a reaction engineering approach to modelling air-drying of small
droplets or thin layer materials. Drying Technology 26(6), 627639.
Chen, X.D. and Chen, N., 1997. Preliminary introduction to a unified approach to modelling drying
and equilibrium isotherms of moist porous solids. Chemeca 1997, Rotorua, New Zealand (on
CD-ROM).
Chen, X.D. and Lin, S.X.Q., 2005. Air drying of milk droplets under constant and time-dependent
conditions. AIChE Journal 51, 17901799.
Chen, G., Maier, D.E. and Campanella, O.H. Takhar, P.S., 2009. Modelling of moisture diffusivities
for components of yellow-dent corn kernels. Journal of Cereal Science 50, 8290.
Chen, X.D. and Mujumdar, A.S., 2008. (eds.) Drying Technologies in Food Processing. Blackwell
Publishing Ltd, Oxford.
Chen, X.D. and Peng, X.F., 2005. Modified Biot number in the context of air drying of small moist
porous objects. Drying Technology 23, 83103.
Chen, X.D. and Pirini, W., 2004. The reaction engineering modelling approach to drying of thin
layer of silica gel particles. In Topics in Heat and Mass Transfer, G. Chen, S. Devahastin, and
B.N. Thorat (eds.), Vindhya Press, Mumbai. pp. 131140.
Chen, X.D., Pirini, W. and Ozilgen, M., 2001. The reaction engineering approach to modelling
drying of thin layer of pulped kiwifruit flesh under conditions of small Biot numbers. Chemical
Engineering and Processing 40, 311320.
Chen, X.D., Sidhu, H. and Nelson, M., 2011. Theoretical probing of the phenomena of formation
of the outmost surface layer of a particle including the surface chemical composition during
rapid removal of water in spray drying. Chemical Engineering Science 66, 63756384.
Chen, X.D. and Stott, J.B., 1992. Calorimetric study of the heat of drying of a sub-bituminous
coal. Journal of Fire Sciences 10, 352361.
Introduction
31
Chen, X.D. and Xie, G.Z., 1997. Fingerprints of the drying of particulate or thin layer food
materials established using a simple reaction engineering model. Transactions of the Institute
of Chemical Engineers Part C: Food and Bio-Product Processing 75(C), 213222.
Chong, L.V. and Chen, X.D., 1999. A mathematical model of the self-heating of spray-dried
food powders containing fat, protein, sugar and moisture. Chemical Engineering Science 54,
41654178.
Cussler, E.L. and Moggridge, G.D., 1997. Chemical Product Design. Cambridge University Press,
Cambridge.
De Vries, D.A., 1958. Simultaneous transfer of heat and moisture transfer in porous media.
Transactions American Geophysical Union 39, 909916.
Donovan, J.L., Meyer, A.S. and Waterhouse, A.L., 1998. Phenolic Composition and Antioxidant
Activity of Prunes and Prune Juice (Prunus domestica). Journal of Agriculture and Food
Chemistry 46, 12471252.
Fogler, H.S., 1992. Elements of Chemical Reaction Engineering, 2nd ed. Prentice-Hall International Inc., New Jersey, pp. 6172.
Fortes, M. and Okos, R., 1980. Drying theories: Their bases and limitations applied to food and
grain. Advances in Drying 1, 119154.
Frank, X. and Perre, P., 2010. The potential of meshless methods to address physical and mechanical phenomena involved during drying at pore level. Drying Technology 28, 932943.
Guillot, G., Trokiner, A., Darrasse, L., Dupas, A., Ferdossi, F., Kassab, G., Hullin, J.P., Rigord, P.
and Saint-James, H., 1991. NMR imaging applied to various studies of porous media. Magnetic
Resonance Imaging 9, 821825.
Gray, B.F., 1990. Analysis of chemical kinetic systems over the entire parameter space III: A wet
combustion system. Proceedings of Royal Society, London A429, 449458.
Gray, B.F. and Wake, G.C., 1990. The ignition of hygroscopic materials by water. Combustion
and Flame 79, 26 (see http://en.wikipedia.org/wiki/Dried fruit, accessed 9 November 2012).
Hills, B.P. Wright, K.M., Wright J.J., Carpenter, C.A. and Hall, L.D., 1994. An MRI study of
drying in granular beds of nonporous particles. Magnetic Resonance Imaging 12, 10531063.
Hui, Y.H., 2006. Handbook of Fruits and Fruit Processing. Blackwell Publishing, Ltd, Oxford,
p. 81.
Incropera, F.P. and DeWitt, D.P., 1990. Fundamentals of Heat and Mass Transfer, 4th ed. John
Wiley & Sons, Inc., New York.
Jin, Y. and Chen, X.D., 2009a. A three-dimensional numerical study of the gas(particle interactions
in an industrial-scale spray dryer for milk powder production. Drying Technology 27, 1018
1027.
Jin, Y. and Chen, X.D., 2009b. Numerical study of the drying process of different sized particles
in an industrial-scale spray dryer. Drying Technology 27, 371381.
Jin, Y. and Chen, X.D., 2010. A Fundamental model of particle deposition incorporated in CFD
simulations of an industrial milk spray dryer. Drying Technology 28, 960971.
Kar, S. and Chen, X.D., 2009. The impact of various drying kinetics models on the prediction
of sample temperature-time and moisture content-time profiles during moisture removal from
stratum corneum. Chemical Engineering Research and Design 87, 739755.
Kar, S. and Chen, X.D., 2010. Moisture transport across porcine skin: experiments and implementation of diffusion-based models. International Journal of Healthcare Technology and
Management 11, 474522.
Keey, R.B., 1978. Introduction to Industrial Drying Operations. Pergamon, London.
Keey, R.B., 1992. Drying of Particulate and Loose Materials. Hemisphere, New York.
32
Langrish, T.A.G., 2008. Characteristic drying curves for cellulosic fibres. Chemical Engineering
Journal 137, 677680.
Lin, S.X.Q. and Chen, X.D., 2005. Prediction of air drying of milk droplet under relatively high humidity using the reaction engineering approach. Drying Technology 23,
13951406.
Lin, S.X.Q. and Chen, X.D., 2006. A model for drying of an aqueous lactose droplet using the
reaction engineering approach. Drying Technology 24, 13291334.
Lin, S.X.Q. and Chen, X.D., 2007. The reaction engineering approach to modelling the cream
and whey protein concentrate droplet drying. Chemical Engineering and Processing 46,
437443.
Luikov, A.V., 1975. Heat and Mass Transfer in Capillary-Porous Bodies. Pergamon Press, Oxford.
Luikov, A.V., 1986. Drying Theory. Energia, Moscow.
Mantle, M.D., Reis, N.C., Griffith R.F. and Gladden, L.F., 2003. MRI studies of the evaporation
of single liquid droplet from porous surfaces. Magnetic Resonance Imaging 21, 293297.
Matsumoto, M., Yasuoka, K. and Kataoka, Y., 1995. Molecular simulation of evaporation and
condensation. Fluid Phase Equilibria 104, 431439.
Patel, K.C., Chen, X.D., Lin, S.X.Q. and Adhikari, B., 2009. A composite reaction engineering approach to drying of aqueous droplets containing sucrose, maltodextrin(DE6), and their
mixtures. AIChE Journal 55, 217231.
Perre, P., 2011. A review of modern computational and experimental tools relevant to the field of
drying. Drying Technology 29, 15291541.
Perre, P. and May, B., 2007. The existence of a first drying stage for potato proved by two
independent methods. Journal of Food Engineering 78, 11341140.
Perre, P., Remond, R. and Turner, I.W., 2007. Comprehensive drying models based on volume
averaging: background, application and perspective. In Modern Drying Technology, Volume 1:
Computational Tools at Different Scales, E. Tsotsas and A.S. Mujumdar (eds.), Wiley-VCH,
Weinheim. pp. 812.
Philip, J.R. and De Vries, D.A., 1957. Moisture movement in porous materials under temperature
gradients. Transactions American Geophysical Union 38 (222232), 594.
Putranto, A. and Chen, X.D., 2013. Spatial reaction engineering approach (S-REA) as an alternative for non-equilibrium multiphase mass transfer model for drying of food and biological
materials. AIChE Journal 59, 5567.
Putranto, A., Chen, X.D. and Webley, P., 2010. Infrared and convective drying of thin layer of
polyvinyl alcohol (PVA)/glycerol/water mixture the reaction engineering approach (REA).
Chemical Engineering and Processing 49, 348357.
Putranto, A., Chen, X.D., Xiao, Z.Y., Davastin, S. and Webley, P.A., 2011. Application of the REA
(reaction engineering approach) for modelling intermittent drying under time-varying humidity
and temperature. Chemical Engineering Science 66, 21492156.
Reis, N.C., Griffiths, R.F., Mantle, M.D. and Gladden, L.F., 2003. Investigation of the evaporation of embedded liquid droplets from porous surfaces using magnetic resonance imaging.
International Journal of Heat and Mass Transfer 46, 12791292.
Rostami, A., Murthy, J. and Hajaligol, M., 2003. Modelling of a smoldering cigarette. Journal of
Analytical and Applied Pyrolysis 66, 281301.
Ruiz-Cabrera, M.A., Foucat, L., Bonnym J.M., Renou, J.P. and Daudin, J.D., 2005a. Assessment
of water diffusivity in gelatine gels from moisture profiles I. Non-destructive measurement
of 1D moisture profiles during drying from 2D nuclear resonance images. Journal of Food
Engineering 68, 209219.
Introduction
33
Ruiz-Cabrera, M.A., Foucat, L., Bonnym J.M., Renou, J.P. and Daudin, J.D., 2005b. Assessment of
water diffusivity in gelatine gels from moisture profiles II. Data processing adapted to material
shrinkage. Journal of Food Engineering 68, 221231.
Shen, D., Bulow, M., Siperstein, F., Engelhard, M. and Myers, A.L., 2000. Comparison of experimental techniques for measuring isosteric heat of adsorption. Adsorption 6, 275286.
Sulsky, D., Chen, Z. and Schreyer, H.L., 1994. A particle method for history-dependent materials.
Computer Methods in Applied Mechanics and Engineering 118, 179196.
Sulsky, D., Zhou, S.J. and Schreyer, H.L., 1995. Application of a particle-in-cell method to solid
mechanics. Computer Physics Communications, 87, 236252.
Takhar, P.S., Maier, D.E., Campanella, O.H. and Chen, G., 2011. Hybrid mixture theory based
moisture transport and stress development in corn kernels during drying: validation and simulation results. Journal of Food Engineering 106, 275282.
Tinker, L.F., Schneeman, G.O., Davis, P.A., Gallaher, D.D. and Waggoner, C.R., 1991. Consumption of prunes as a source of dietary fiber in men with mild hypercholesterolemia. American
Journal of Clinical Nutrition 53, 12591265.
Tsotsas, E. and Mujumdar, A.S., 2007. Preface of Volume 1. In Modern Drying Technology Volume
1: Computational Tools at Different Scales, E. Tsotsas and A.S. Mujumdar (eds.), Wiley-VCH,
Weinheim. pp. xvxviii.
USDA, 2007. USDA database for the flavonoid content of selected foods (Release 2.1 2007).
van Meel, D.A., 1958. Adiabatic convection batch drying with recirculation of air. Chemical
Engineering Science 9, 3644.
Vinson, J.A., Zubic, L., Bose, P., Samman, N. and Proch, J., 2005. Dried fruits: excellent in vivo
and in vitro antioxidants. Journal of American College of Nutrition 24, 4450.
Wang, Z., Zhou, Y., Li, S. and Liu, J. (eds.), 2007. Physical Chemistry, Volume 2, 4th ed. Higher
Education Press, Beijing, pp. 196223 (in Chinese).
Whitaker, S., 1977. Simultaneous heat, mass and momentum transfer in porous media: A theory
of drying. Advances in Heat Transfer 13, 119203.
Whitaker, S., 1999. The Method of Volume Averaging. Kluwer Academic Publishers, Dordrecht.
Williamson, G. and Carughi, A., 2010. Polyphenol content and health benefits of raisins. Nutrition
Research 30, 511519.
Woo, M.W., Daud, W.R.W., Mujumdar, A.S., Wu, Z.H., Talib, M.Z.M. and Tasirin, S.M., 2008.
CFD evaluation of droplet drying models in a spray dryer fitted with a rotary atomizer. Drying
Technology 26, 11801198.
Wu, X., Beecher, G.R., Holden, J.M., Gebhardt, S.E. and Prior, S.L., 2004. Lipophilic and
hydrophilic antioxidant capacities of common foods in the United States. Journal of Agriculture and Food Chemistry 52, 40264037.
Wu, Y. and Keey, R.B., 1995. Design of staged through-circulation drying with air reversals.
Chemical Engineering Science 50, 99104.
Reaction engineering approach I

Lumped-REA (L-REA)
2.1
The REA formulation

As suggested in Chapter 1, the general REA is an application of chemical reaction
engineering principles to model drying kinetics, first reported in 1997 (Chen, 2008).
In this approach, evaporation is modelled as zero-order kinetics with activation energy,
while condensation is treated as a first-order wetting reaction with respect to drying air
solvent vapour concentration without activation energy (Chen, 2008). The REA offers
the advantage of being expressed in terms of simple ordinary differential equations with
respect to time. This negates the complications arising from use of partial differential
equations (Chen and Xie, 1997). This approach has been initially employed to express
the overall drying rate for the entire object being dried a lumped approach. A summary
of the developments of the lumped approach of REA is given by Chen (2008) and
described at length in Chapter 1.
Generally, with no assumptions, the drying rate of a material can be expressed as:
ms
dX
= h m A(v,s v,b ),
dt
(2.1.1)
where ms is the dried mass of thin layer material (kg), X is the average moisture content
on a dry basis (kg kg1 ), t is time (s), v,s is the water vapour concentration at interface
(kg m3 ), v,b is the water vapour concentration in the drying medium (kg m3 ), hm is
the mass transfer coefficient (m s1 ) and A is surface area of the material (m2 ).
Equation (2.1.1) is a basic mass transfer equation. The convective mass transfer
coefficient (hm ) is determined based on the established Sherwood number correlations
for the geometry and flow condition of concern or can be established experimentally
for the specific drying conditions involved (Lin and Chen, 2002; Kar and Chen, 2009).
The surface vapour concentration ( v,s ) can then be scaled against the saturated vapour
concentration ( v,sat ) using the following equation (Chen and Chen, 1997; Chen and Xie,
1997; Chen, 2008):

E v
v,s = exp
(2.1.2)
v,sat (Ts ),
RTs
where Ev represents the additional difficulty in removing moisture from the material
beyond the free water effect. This Ev is the average moisture content (X) dependent.
Ts is the surface temperature of the material being dried (K) and v,sat for water vapour
Reaction engineering approach I: L-REA
35
can still be estimated by:

v,sat = 4.844 109 (Ts 273)4 1.4807 107 (Ts 273)3 + 2.6572
105 (Ts 273)2 4.8613 105 (Ts 273) + 8.342 103 , (2.1.3)
based on the data summarised by Keey (1992).
When material is thermally thin, the surface temperature is considered to be the
same as the sample temperature (Chen and Peng, 2005; Patel and Chen, 2008), i.e.
Ts T.
The mass balance (Equation 2.1.1) is then neatly expressed as:

dX
E v
= h m A exp
v,sat (T ) v,b .
ms
(2.1.4)
dt
RT
From Equation (2.1.4), it can be observed that the REA is expressed in the firstorder ordinary differential equation with respect to time, and the model is the core
of the L-REA. It must be noted that, though the average moisture content is used, the
L-REA does not assume uniform moisture content. Of course, no spatial distribution of
moisture content can be computed using the L-REA. The L-REA may also be applied
to cases where the temperature within the material is not uniform as long as the surface
temperature can be determined or predicted accurately.
The activation energy (Ev : the characteristic of the material being dried; it is materialdependent) needs to be determined experimentally. Upon the attainment of the drying
data, notably the surface temperature of the material (or the sample temperature in the
case of a thermally thin situation) and moisture loss against time, plus the information
about the external mass transfer coefficient, one can obtain the activation energy. This
can be done by rearranging Equation (2.1.4) as follows:

m s ddtX h m1 A + v,b
E v = RTs ln
.
(2.1.5)
v,sat (Ts )
The rate of moisture loss d X/dt is experimentally determined. The surface area A
should also be recorded in drying experiments in the case of shrinkable material. The
dependence of activation energy on average moisture content on a dry basis (X ) can be
normalised as:
E v
= f (X X b ) ,
E v,b
(2.1.6)
where f is a function of water content difference and Ev,b is the equilibrium activation energy representing the maximum Ev determined by the relative humidity and
temperature of the drying air:
E v,b = RTb ln(R Hb ),
(2.1.7)
where RHb is the relative humidity of drying air, Xb is the equilibrium moisture content
under the condition of the drying air (kg kg1 ) and Tb is the drying air temperature (K).
36
2.2
Determination of REA model parameters

As mentioned in the previous section, the REA model parameters such as the relative
activation energy can be generated from one set of accurate drying runs which consists
of measuring sample mass, sample surface temperature and sample surface area during
drying. This is only true when a master curve, Equation (2.1.6), can be applied to the
range of drying conditions of interest. So far in the successful applications of the REA
concept, the drying experiments are usually performed with final moisture content being
very low. In other words, the humidity of the drying air (or gas in general) employed in
the experiments is set to be very small. This is done so the master curve can cover the
widest range of the water contents for the same material. To illustrate the procedures to
obtain and determine the REA model parameters, the experiments of convective drying
of milk droplets are described next.
The convective drying of milk droplets is conducted in a glass-filament convective
drier (Lin and Chen, 2002; 2004; Chen and Lin, 2005). The drying air conditions were
drying air temperature of 67110 C, drying air velocity of 0.451 m s1 and drying
air humidity of 0.001 kg H2 O kg dry air1 (Chen and Lin, 2005). The details and setup
of the equipment have been described previously (Chen and Lin, 2005). Generally, by
using the glass-filament method, the droplet weight is measured by deflection of the
glass filament as long as the values are corrected for the corresponding drag forces.
The single droplet is attached to the tip of a special fine glass filament. The details of
the experiments are as follows (refer to Figures 2.1 and 2.2).
The cross section of the drying tunnel is 29.97 29.97 mm. The glass filament
with a thin tip section of 30100 m (whose end has a glass knob of 100220 m)
is used to suspend the milk droplet (initial solids concentration of 2030%wt. and
initial droplet diameter about 1.42 mm) in a preconditioned (humidity, temperature and
velocity) air stream. Initial solids of greater than 30%wt. have also been successfully
tested (Fu, 2012). During drying, the deflection of the glass filament and the change
of droplet diameter are monitored using standard video camera fitted with four closeup lenses. The sample mass during drying is calculated by the deflection of the glass
filament during drying. The glass filament will deflect from the original position when
there is a mass hanging up on the tip (refer to Figure 2.1). As the drying progresses,
the deflection will reduce because of moisture evaporation. The camcorder is used to
capture the displacement history and the value of deflection is converted to that of mass
by calibration which shows the linear relationship between the deflection and mass.
The example of the linear relationship is shown in Figure 2.2. For calculating surface
area during drying, the recorded images are transferred to a digital signal using a video
capture card and analysed using image analysis software. The projected droplet area of
each figure is considered to be a perfect circle and the equivalent diameter can be then
calculated (Chen and Lin, 2005; Fu et al., 2011; Fu, 2012).
The sample temperature during drying is measured separately from the weight measurement by exactly same drying conditions, droplet size and compositions. A calibrated
thermocouple (type K, diameter of 13 or 24 m) is used to record the sample temperature
during drying. The thermocouple is connected to a Picometer (Pico Technology, UK)
(a)
37
Suspending glass filament

Video camera
Droplet
Drying chamber
Airflow from heating tower

(b)
Glass capillary tubes to

support the thermocouple
Thermocouple junction
Thermocouple wires

(c)
Mass-measuring glass
filament

Figure 2.1 Equipment setup of convective drying of milk droplets (a) measuring droplet
shrinkage; (b) measuring droplet temperature; (c) measuring mass change. [Reprinted from
Chemical Engineering Science, 66, N. Fu, M.W. Woo, S.X.Q. Lin et al., 17381747, Copyright
(2012), with permission from Elsevier.] (Adapted from Fu et al. (2011) Chemical Engineering
Science, 66, 17381747.)
and the data are obtained from the data logger (Picolog R5, 17, Pico Technology, UK).
The repeatability of the weight loss experiment is 0.01 mg and the accuracy is 0.05
mg while the accuracy of the temperature measurement is within 0.1 C. The repeatability of the weight loss experiment is 0.01 mg and the accuracy is 0.05 mg while
the accuracy of the temperature measurement is within 0.1 C (Chen and Lin, 2005; Fu
et al., 2011; Fu, 2012).
(a)
Without weight
With weight
Displacement
(b)
250
Displacement (pixel)
38
200
y = 48, 13x
R2 = 0.998
150
100
Experimental
Linear correlation
50
0
2
3
4
5
Mass of weight standards (mg)
Figure 2.2 The deflection of glass filament and a typical standard curve (a) measuring
displacement to measure weight loss; (b) correlation between the displacement and the weight.
[Reprinted from Chemical Engineering Science, 66, N. Fu, M.W. Woo, S.X.Q. Lin et al.,
From the experiments mentioned previously, the experimental data of sample mass,
surface area, volume and temperature can be obtained. The rate of moisture content
change (dX/dt) can be obtained from the weight loss curve experiment. The heat and mass
transfer coefficient is determined based on established Sherwood and Nusselt number
correlations or established experimentally for the specific drying conditions involved. For
the experiments of single-droplet drying in a convective dryer, the following correlations
can be used (Lin and Chen, 2002):
Sh = 1.63 + 0.54Re0.5 Sc0.333 ,
N u = 2.04 + 0.62Re
0.5
Pr
0.333
(2.2.1)
(2.2.2)
where Sh, Nu, Re, Sc and Pr are the Sherwood, Nusselt, Reynolds, Schmidt and Prandtl
numbers, respectively. These differ somewhat from the classical Ranz and Marshall
correlations due to the blowing effect (Lin and Chen, 2002).
The saturated water vapour concentration ( v,sat ) is evaluated based on Equation
(2.1.3) while the ambient water vapour concentration ( v,b ) is determined based on
the humidity and temperature of the drying air employed in the laboratory. By using
the experimental data, calculated heat and mass transfer coefficients, v,sat and v,b ,
39
1.0
Exp. data
The curve fitted
Ev/Ev,b
0.8
0.6
0.4
0.2
0.0
0.0
1.0
2.0
3.0
XXb (kg/kg)
4.0
5.0
Figure 2.3 The relative activation energy of convective drying of 20%wt. skim milk powder at a
drying air temperature of 67.5 C, velocity of 0.45 m s1 and humidity of 0.0001 kg H2 O kg dry
air1 . [Reprinted from AIChE Journal, 51, X.D Chen and S.X.Q. Lin, Air drying of milk droplet
under constant and time-dependent conditions, 17901799, Copyright (2012), with permission
from John Wiley & Sons, Inc.]
the activation energy (Ev ) can be calculated using Equation (2.1.5). In addition, the
equilibrium activation energy (Ev,b ) can be evaluated using Equation (2.1.7). The v,b
can be related to the equilibrium moisture content (Xb ) through the desorption isotherm
such as GAB isotherm. The coefficients of the isotherm for milk droplets are provided
in Chen and Lin (2005). The activation energy (Ev ) can be scaled (from 0 to 1) by
dividing it by the equilibrium activation energy (Ev,b ) to yield the relative activation
energy (Ev /Ev,b ) as indicated in Equation (2.1.6). The relative activation energy can
be related to the difference of moisture content during drying (XXb ). The relative
activation energy (Ev /Ev,b ) of 20%wt. of skim milk powder is shown in Figure 2.3.
It shows that, initially, the relative activation energy is zero and this increases as drying
progresses. When the equilibrium moisture content is achieved, the relative activation
energy is 1. This indicates that the difficulty in removing the moisture from the materials
being dried increases as the drying continues. It is noted that Xb for the particular test
run of concern may also be measured by prolonging the drying test under the same
condition for an extended period to achieve equilibrium.
The relationship between the relative activation energy (Ev /Ev,b ) and the difference in moisture content (XXb ) can be expressed by a simple mathematical function
R
(polynomial, exponential, logarithmic, etc.). Microsoft Excel
(Microsoft Inc., 2012)
can be used for fitting in order to obtain a simple algebraic function. For the convective
drying of 20%wt. skim milk, the relative activation energy can be represented as (Chen
and Lin, 2005):

E v
= 0.998 exp 1.405 (X X b )0.993 .
E v,b
(2.2.3)
40
When Equation (2.2.3) is generally applicable, the relative activation energy generated
from one accurate drying run can be applied to model the convective drying at other
conditions, provided the same material and similar initial moisture content is used. The
relative activation energy is the fingerprint of the REA where the physics of the drying
is captured by it. Since only one accurate drying run is required to generate the relative
activation energy, the REA is effective in terms of the number of experiments to generate
the REA parameters.
This makes the REA, at least in this lumped-form, differ markedly from some other
modelling approaches to drying. For instance, the diffusion-based approach is commonly
applied in literature in which the diffusivity function needs to be estimated from several
sets of experimental data. This means several experiments are required to obtain the
diffusivity. Vaquiro et al. (2009) mentioned that 10 experiments were needed to generate
the diffusivity function as function of temperature and moisture content. Some optimisation procedures need to be conducted subsequently to represent the dependence of
diffusivity on moisture content and/or temperature (Azzouz et al., 2002; Mariani et al.,
2008; Pakowski and Adamski, 2007; Ramos et al., 2010; Ruiz-Lopez et al., 2011; Silva
et al., 2010; Vaquiro et al., 2009). Pakowski and Adamski (2007) and Ruiz-Lopez et al.,
2011 show that various mathematical expressions of diffusivity need to be attempted to
yield the most appropriate diffusivity function. More discussion on these will be given
in Chapter 4.
2.3
Coupling the momentum, heat and mass balances

As an example, the REA has been implemented to model a plug-flow spray drying
process for lactose droplets. Basically, it is drying of mono-dispersed lactose solution
droplets with a vertical trajectory. This drying approach does not consider the dropletdroplet interactions and droplet-wall interactions for purposes of simplicity. Some innovative steps can be taken to implement this one-dimensional model to a real industrial
situation in order to obtain practical and meaningful predictions (Patel et al., 2010).
According to the methodology of Patel et al. (2010), one can devise the following
to simulate spray-drying of milk droplets. This approach follows a semi-Lagrangian
framework in which the model follows or predicts the condition of the particle. In
order to do this, the model predicts the ambient conditions corresponding to the position
of the particle within the chamber, considering the momentum, heat and energy coupling
between the two phases. The model assumes a pluglike flow of particles and uniform air
flow across the diameter of the dryer. This is illustrated in Figure 2.4.
The change in particle axial trajectory is modelled by taking the balance between the
drag force due to the relative motion and the buoyancy force by the air,
g
18 b Re
dv p
= CD
(vb v p ) +
( p b ),
dt
24 p d 2p
p
(2.3.1)
where vp is the droplet/particle velocity (m s1 ), vb is the gas velocity (m s1 ), p is the

droplet density (kg m3 ), b is the gas density (kg m3 ), g is the gravitational constant
41
Atomiser
Uniform
particle
distribution
Uniform air
velocity,
temperature
and humidity
Figure 2.4 Schematic diagram showing the plug-flow spray dryer.
(m2 s1 ) and dp is the droplet diameter (m). The drag coefficient C D can be expressed
by the following for 0.5 < Re < 1000:
CD =
24
(1 + 0.15 Re0.687 ).
Re
(2.3.2)
The Reynolds number is calculated based on relative velocity between the air and the
particle taking the diameter of the particle as the characteristic length,
Re =
b d p |v p vb |
,
b
(2.3.3)
where b is the viscosity of the gas (Pa.s). It can be seen from Equation (2.3.1) that
the momentum of the particle is coupled with the drying gas (air). Only one-way
momentum coupling is considered and the effect of the particle motion on the air is
assumed negligible. Corresponding to the position of the particle, the change of the air
temperature against the length of the chamber is given next:
dTb
dL
m s ddtX
vp
[ HV C p,v (Tb T p )]
vp
h A p (Tb T p ) U ( Ddryer ) (Tb Tamb ) V b HV

V b C p,b
dY
dL
(2.3.4)
where L is the length of the one-dimensional dryer (m) (hence, dL is the differential
control volume), X is the water content on dry basis (kg kg1 ), ms is the dry mass of
a single droplet (kg), Tb is the gas (air) temperature (K), Tp is the particle temperature
(K), Cp,b is the specific heat of the drying medium (J kg1 K1 ), Nv is the evaporation
42
flux (kg H2 O m2 s1 ), Cp,v is the specific heat of water vapour (J kg1 K1 ), Ap is

the area of droplet (m2 ),V is the volumetric flow rate of gas (air) (m3 s1 ), Ddryer is the
diameter of the spray dryer (m2 ), HV is the vaporisation heat of water (J kg1 ) and U
is the overall heat transfer coefficient (W m2 K1 ).
This equation was developed considering the enthalpy balance for a control volume
of air enveloping the particle. Equation (2.3.4) also considers heat loss from the dryer
to the ambient. The change of bulk air humidity against the length of the chamber then
takes the following form:
m s ddtX vp
dY
,
=
m b,dr y
dL
(2.3.5)
where m b,dr y is the mass flow rate of the gas (air) (kg s1 ) and is the number of droplets
going into the control volume per unit time (s1 ). Currently, for this dyer, it is basically
how many droplets are generated at the atomiser per unit time. This equation has been
developed considering the humidity balance for a control volume of air enveloping the
particles. This suggests that any change in humidity is only affected by the drying of the
particles, releasing vapour into the bulk air stream.
In Equations (2.3.4) and (2.3.5), the rate of drying of a single droplet needs to be
defined. In the general one-dimensional framework, the single droplet drying term can
be determined using the REA model already mentioned previously in Equation (2.1.4).
The parameters need to be relevant to the material being dried; for skim milk with initial
solids content of 20 %wt., the relative activation energy function is thus Equation (2.2.3).
Heat transfer across the surface of the droplet takes the following form accounting
for convective heating and evaporative cooling:
dT p
dX
= h A p (Tb T p ) + m s
HV ,
dt
dt
(2.3.6)
noting that dX/dt is negative when drying takes place.

In Equation (2.1.4) and (2.3.6), the heat and mass transfer coefficients (hm , h, respectively) can be calculated using the Ranz Marshall form of the Nusselt and Sherwood
correlation:

h m = 2 + 0.6 Re1/2 Sc1/3
,
(2.3.7)
dp
kb

,
(2.3.8)
h = 2 + 0.6 Re1/2 Pr 1/3
dp
where is the film thermal diffusivity of the vapour-air system around the droplet/particle
(m2 s1 ) and kb is the thermal conductivity of the vapour-air system around the
droplet/particle (W m1 K1 ). These correlations can be modified by taking into account
the vapour blowing effects in various well-known formats. All the fluid properties used
in the calculation of Equations (2.3.9) and (2.3.10) have to be evaluated at film temperature, which can be approximated as the average of the bulk and particle temperatures.
Therefore, the Reynolds number calculated for Equations (2.3.9) and (2.3.10) will be
different from that calculated for Equation (2.3.1). The expression for the Schmidt and
43
Prandtl number is given next:

b
,
b
C pb b
Pr =
.
kb
Sc =
(2.3.9)
(2.3.10)
The profiles of moisture content and temperature of the droplets as well as those
of humidity and temperature of the drying medium along the dryer can be predicted by simultaneously solving the momentum, mass and balances of the droplet.
The coupling of momentum, mass and heat balances of the droplets and the drying medium can be used as process simulation tool for the study of plant-wide dryer
performance.
Similarly, one can apply the approach to modelling large-scale time-dependent threedimensional spray dryers. In this case, the momentum, heat and mass balances (partial
differential equations that govern these processes) have been already solved (by Jin and
Chen, 2009a; Jin and Chen, 2010; 2011). The L-REA model for single droplet drying
has been incorporated in these simulations. Here, the advantage of using the REA is that
one does not need to solve the spatial distribution of the moisture within the particles.
For example, one does not need to use a diffusion equation for each particle. Trying to
track the moisture changes using a spatially distributed model for the particle inside,
for literally thousands if not hundreds of thousands of particles in a spray dryer to make
simulation realistic, would take far more computational time.
2.4
Mass or heat transfer limiting

The question of mass or heat transfer limitation is not a straightforward concept as far as
drying modelling is concerned. When mass transfer is the limiting process, which may
usually be the case when the temperature difference between the material to be dried
and that of the drying gas is not very large, one may not need to consider the spatial
distribution of temperature and can therefore simplify the modelling process. If heat
transfer is limiting, one has to worry about the temperature gradient within the medium
being dried (Chen, 2007).
In air drying, one expects that the spatial distribution of water content inside a material
is significant, i.e. the boundary could be relatively dry but the core could still be very
wet. The question is whether the temperature within the material being dried can be
approximated to be uniform. At first sight, all these should have much to do with two
classical numbers: the Biot number for examining the temperature uniformity and the
Lewis number for examining heat or mass transfer limiting.
2.4.1
Biot number analysis

The Biot number criterion is well known and used to investigate the temperature uniformity of a material being heated or cooled. Conventionally, the Biot number is introduced
44
Heat flux
Ts,1
Ts,2
L
T
x
x=0
x=L
Figure 2.5 The schematic diagram showing the parameters for the definition of the classical Biot
number. [Reprinted from Drying Technology, 23, X.D. Chen, Air drying of food and biological
materials Modified Biot and Lewis number analysis, 22392248, Copyright (2012), with
permission from Taylor & Francis.]
through steady-state heat conduction in a slab with one side cooled by convection (see
Figure 2.5). The conductive heat flux through the wall is set to be equal to the heat flux
due to convection (Incropera and DeWitt, 1990; 2002):
qx = k
Ts,1 Ts,2
= h (Ts,2 Tb ) ,
L
(2.4.1)
where qx is the heat flux (W m2 ), k is the thermal conductivity of the solid material
(slab) (W m1 K1 ), L is the thickness of the material (m), h is the heat transfer coefficient
(due to convection) (W m2 K1 ), Ts represents the surface temperatures, the subscripts
1 and 2 represent the high and the low temperature of the slab, respectively (K), Tb
is the temperature of the gas (air, for instance) (K) and k is the thermal conductivity
(W m1 K1 ).
The ratio of the temperature differences can then be expressed as:
hL
Ts,1 Ts,2
=
.
Ts,2 Tb
k
(2.4.2)
The classical Biot number is thus defined as the temperature ratio by:
Bi =
hL
.
k
(2.4.3)
When this ratio is less than 0.1, i.e. the internal temperature difference is smaller than
10% of the external temperature difference, the internal temperature distribution may be
neglected for simplicity in modelling. For a spherical object, the characteristic length L
may be set to be the radius of the sphere.
One can rewrite Equation (2.4.2), in the form of the temperature ratio and the resistance
ratio:
Ts,1 Ts,2
L/k
Rcond
hL
=
=
=
,
Ts,2 Tb
k
1/ h
Rconv
(2.4.4)
45
Surface
evaporation
T
Ts,2
Ts,1
x=0
x=L
Figure 2.6 The schematic diagram showing the parameters for the definition of the modified Biot
number) (ChenBiot number). [Reprinted from Drying Technology, 23, X.D. Chen, Air drying of
food and biological materials Modified Biot and Lewis number analysis, 22392248,
Copyright (2012), with permission from Taylor & Francis.]
where R represents the thermal resistances, and cond and conv represent the conduction and convection, respectively. As such, the Biot number can also be considered to
be the ratio of the internal resistance to external resistance. A small temperature ratio
and a small resistance ratio essentially suggest the same thing. More fundamentally, the
resistance ratio is a better argument. The thermal conductivity of the particle would be
affected by water content and porosity (when filled with air).
Chen introduced a new formula, which accounts for evaporation from the heat
exchange surface (Chen and Peng, 2005). Similar to the conventional analysis (Equation
2.4.1), considering the addition of evaporative loss, the following heat balance can be
obtained (here the temperature of the environment is greater than the material being
dried; see Figure 2.6):
Ts,2 Ts,1
.
(2.4.5)
h (Tb Ts,2 ) HL Sv = h (Tb Ts,2 ) = k
L
Here h* is an equivalent convection heat transfer coefficient and S v is the surface based
evaporation or drying rate (kg m2 s1 ) (taken as a positive value for evaporation):
h = h
HL Sv
.
(Tb Ts,2 )
(2.4.6)
This leads to a more appropriate Biot number for the evaporative system:
Bi = Bi
HL Sv L
.
(Tb Ts,2 ) k
(2.4.7)
When evaporation occurs from the heat exchanger surface, the modified Biot number
(Bi* ) represents the temperature uniformity more precisely than Bi. Just because a
conventional and large Bi indicates a non-uniform temperature distribution does not
mean the temperature is not reasonably uniform when evaporation occurs.
46
Bi* is a smaller value than Bi indicating a more uniform temperature distribution. The
significance of this equation can be demonstrated in the following for evaporation from
a water droplet (which is also relevant to drying of a coal particle with high moisture
content at the beginning of the drying process).
The following calculations are based on the laboratory data obtained by Lin and Chen
(2002) and Lin (2004) on evaporating of a single water droplet (spherical condition).
The laboratory conditions and the techniques employed are given:
Droplet diameter: 2L = 1.43 mm
Heat transfer coefficient measured: h = 99.1 W m2 K1
Thermal conductivity of water: k = 0.63 W m1 K1
Evaporation rate: Sv = 1.73 103 kg s1 m2
Interfacial temperature: Ts = 23.4 C
Drying air temperature: Ts = 67.5 C
Latent heat of evaporation: HV = 2445 103 J kg1
The conventional analysis, based on Equation (2.4.3), yields Bi = 0.11 for the previous
example, which is slightly greater than the critical value of 0.1 mentioned earlier. Based
on Equation (2.4.7), however, one can find that the Bi required for uniform temperature
assumption is 0.21 in order to maintain the Bi* being 0.1, i.e.:
Bi cri = Bi cri
+
2445 103 1.73 103 1.43 103

0.21,
(67.5 23.4)
2 0.63
(2.4.8)
due to the evaporation (cooling) effect, which consumes much of the temperature driving
force from the outside of the material being dried.
Equation (2.4.7) can be extended to account for the cases when internal mass transfer
resistance plays a role. Here, one can express the surface based evaporation rate using
the overall mass transfer coefficient concept instead:
Bi = Bi
HV Um (v,c v,b )

,
k
(Tb Ts,2 )
L
(2.4.9)
where L may be used as the characteristic length generically of slab, cylinder or a sphere
etc. (m), v,b is the vapour concentration in the drying medium and v,c is the vapour
concentration (based in bulk) at the location marked by the characteristic dimension
( c ) inside the material, which may be taken as the saturated vapour concentration at
the material (mean) temperature (a high bound estimate). The overall mass transfer
coefficient (Um ) may be expressed approximately as the following by Chen (2007):
Um
1
hm
1
.
+ Deffc ,v
(2.4.10)
The characteristic dimension (c ) signifies the effect of the vapour concentration profile
and the solid matrix resistance to vapour transfer. For a symmetric material being dried,
this dimension should be smaller than the corresponding characteristic dimension for
heat conduction (for example, < the 0.25 of the half-thickness for a slab being heated:
47
see Van der Sman, 2003). The effective vapour diffusivity may be estimated using the
porosity () and the tortuosity ( ) correction (McCabe et al., 2001):
(2.4.11)
Deff ,v Dv,air ,
where Dv,air is the vapour diffusivity in air (m2 s1 ). The tortuosity ( ) is a parameter
not usually known prior, so an estimate between 2 and 20 may be used in some cases. It
is estimated as an empirical function of (McCabe et al., 2001).
The mass transfer coefficient (hm ) could also be estimated using one of the heat and
mass transfer analogies, such as the one for a flat plate (Incropera and DeWitt, 1990;
2002):

Dv,air 0.3
h
kair
,
(2.4.12)
hm
Dv,air
air
which is based on the heat transfer coefficient h, depending on the geometry of the object
of concern.
One can see that the new number, the ChenBiot (ChBi) number, can be defined as
(Chen, 2007) follows for the surface evaporation case shown in Figure 2.6:
ChBi = Bi
HV Sv L
.
(Tb Ts,2 ) k
(2.4.13)
It is difficult to evaluate the ChenBiot number for a case where drying occurs inside the
material. As mentioned earlier, the following formula has been proposed (Chen, 2005a;
Chen, 2007) for cases where evaporation also takes place within:
ChBi = Bi
k
L
HV

T Ts,2
v,c v,b
c
1
+
hm
Deff ,v
.
(2.4.14)
Alternatively, the term for mass transfer resistance due to vapour diffusion in Equation
(2.4.14) is replaced by that using the effective liquid water diffusivity, Deff,l (effective
liquid water diffusivity), as it is most commonly measured or cited in drying literature.
Equation (2.4.14) reduces to the surface evaporation case when the characteristic
thickness c approaches zero.
2.4.2
Lewis number analysis

The Lewis number (Le) is defined as the ratio of the thermal diffusivity to the mass
diffusivity (for water vapour transfer). For drying moist, porous materials, this would
mean that if Le < 1, the heat penetration (through conduction) into the particle is slower
than the penetration of the water vapour front.
If Le 1, it means that the heat transfer is limiting the drying process. When Le 1,
the heat input and moisture removal are highly coupled. When Le 1, the process is
mass-transfer limited.
This becomes particularly informative and important for evaluating drying after the
initial moisture-rich condition. This process can be visualised as follows by considering
48
the heating of a semi-infinite porous media in a constant temperature environment as

summarised by Chen (2007). It is known that the thermal penetration depth ( T ) can be
approximated as:

T 12tT ,
(2.4.15)
where is the thermal diffusivity (=k/Cp ) (m2 s1 ) and tT is thermal penetration
time (s). Correspondingly, the vapour penetration (mass penetration denoted by the
subscript M) may be expressed as:
M

12Deff ,v t M ,
(2.4.16)
where tM is the mass penetration time (s). For reaching the same distance into the
material, i.e. T = M , the ratio of time required for mass to penetrate to that for heat
to penetrate is essentially the Lewis number:
Le =
tM
=
.
tT
Deff ,v
(2.4.17)
The physics is apparent that if Le 1, mass transfer occurs much faster, thus heat
transfer is limiting.
A conventional Lewis number analysis using the effective vapour diffusivity calculated
with Equation (2.4.11) would, for drying a skim milk droplet, for instance, yield a value
smaller than 0.1 (Farid, 2003), leading to the conclusion that the drying is heat-transfer
limiting.
In order to derive the new Lewis number that considers the phase change process,
Chen (2005a) has employed a fuller account of the energy and mass balances (with
source terms) for drying as follows:

Cl
Cl
=
Deff ,l
E v ;
(2.4.18)
t
x
x

C v
Cv
=
Deff ,v
+ E v ;
(2.4.19)
t
x
x

1
HV
T
T
(2.4.20)
Ev ;
=
keff
t
C p x
x
C p
where Cl and Cv are the concentrations of liquid water and water vapour, respectively
(kg m3 ), T is the sample temperature (K), is the sample density (kg m3 ), Cp is the
sample specific heat (J kg1 K1 ), keff is the thermal conductivity (W m1 K1 ), D. eff,l
and Deff,v are the effective liquid and vapour diffusivity, respectively (m2 s1 ), and E v is
the source term (kg m3 s1 ). This is for slab geometry. Similar equations can be written
for spherical and cylindrical or other three-dimensional objects.
The source term may be approximated as:
dX
.
E v s
dt
(2.4.21)
49
This E v is equivalent to the source term I in Chapter 3 where S-REA is introduced. This
represents the local rate of evaporation. Assuming that Cv is the vapour concentration
that is in equilibrium with the liquid water content inside the solid structure, which is
a conservative assumption in the sense that the process of mass transfer is perhaps the
slowest, the following relationship exists:
RH = =
Cv
= f (X, T ) or X = F(, T ),
Cv,sat (T )
(2.4.22)
where RH is the relative humidity and Cv,sat is the saturated water vapour concentration
(kg m3 ).
Equation (2.4.22) is the equilibrium isotherm function. Therefore, one may write the
following:

Cv

X
X T
X T
X
Cv,sat (T )
E v s
+
= s
+
.
t
T t
t
T t
(2.4.23)
If the temperature may be taken as an average value, especially when the equilibrium
isotherm functions are insensitive to temperature in the range considered, the following
simplified Equation (2.4.23) can be obtained:

1
C
X
v

s
.
(2.4.24)
E v s
t
Cv,sat( T t
Thus, the vapour conservation Equation (2.4.19) can be rewritten (corresponding to
Equation 2.4.14) into:

1
D
C v
Cv
2
eff ,v
r2
,
(2.4.25)
t
r r
r
X
1
1 + s C T
v,sat ( )
by taking a mean D eff ,v for simplification.
new
The equivalent mean effective diffusivity (D eff ,v ) is then:
D
eff ,v
new
D eff ,v
1 + s
X
1
C v,sat (T )
.
(2.4.26)
The effective Lewis number (i.e. the ChenLewis number) can then be written as:

eff
k eff
D
.
eff ,v
Ch Le = new =
(2.4.27)
C p
D eff
X
1
1 + s C T
v,sat ( )
Equation (2.4.27) may be used to estimate a more likely high bound of the Lewis number
(here the ChenLewis number). For the same skim milk drying case as mentioned earlier,
Equation (2.4.27) can yield an estimate of the ChenLewis number (ChLe) of the order
of 100, indicating the process is mass transfer limiting.
50
2.4.3
Combination of Biot and Lewis numbers

Though no strict scientific proof regarding why one should combine the Bi and Le
numbers together to justify isothermal, uniform temperature or otherwise. Nevertheless,
these are the two most important dimensionless parameters in literature that seem most
relevant to drying. Sun and Meunier (1987) conducted a comprehensive numerical
analysis on non-isothermal sorption in adsorbents, which showed that the following rule
exists: the isothermal model would be valid if LeBi > 100 and the uniform temperature
profile model would be a good model if Le > 10.
Note that here the Le and Bi are all based on the conventional definitions. It is expected
that this rule is conservative when drying is concerned. In the desorption process the
temperature profile inside the porous material would tend to be more gradual, thus
the criteria can be relaxed. The driving force for heat transfer, i.e. the difference between
the drying air (or the drying medium in general) and the porous material, i.e. (Tb Ts ),
would also affect the temperature uniformity when drying proceeds, the waterfall
effect.
2.5
Convective drying of particulates or thin layer products

modelled using the L-REA
The L-REA (lumped reaction engineering approach) has been used to describe the
convective drying of droplets of whey protein concentrate and thin layer of a mixture
of polymer solution (Lin and Chen, 2007; Allanic et al., 2009). For the convective
drying of droplets of whey protein concentrate, the experimental setup is similar to
that explained in Section 2.3. The droplets are suspended in a glass-filament convective
dryer, and the mass and temperature are recorded during drying. The deflection of the
glass filament is captured and converted to droplet weight. The measurement of weight
also takes into account the drag force. A video camera system is used to monitor the
droplet diameter change during drying. A calibrated thermocouple is used to record
the sample temperature during drying. The thermocouple is connected to a picometer
and the data is obtained from the data logger. The repeatability of the weight loss
and temperature measurement is 0.01 mg and 0.1 C, respectively. Drying air with
the velocity of 0.45 m s1 and a temperature of 70110 C is used. Initial droplet
diameter and solids concentration are 1.45 mm and 30% wt., respectively (Lin and Chen,
2007).
For convective drying a mixture of polymer solutions, the experimental data used in
the current work are derived from the study reported by Allanic et al. (2009) and the
experimental conditions are shown in Table 2.1.
In order to better understand the modelling presented here, the details of experiment
are briefly described here (Allanic et al., 2006; 2009). Materials used in this experiment were a mixture of equal proportions of partially hydrolysed polyvinyl alcohol
(80%wt.) and glycerol with 88%wt. of water, then 8 ml of the mixture was poured into a
90-mm-diameter Petri dish so the initial thickness of the sample was 1.3 mm. During
51
Table 2.1 Experimental conditions of convective drying of a

mixture of polymer solutions (Allanic et al., 2009).
Number
Air velocity
(m s1 )
Air temperature
(C)
Air relative
humidity (%)
1
2
3
2.8
1
1
55
35
55
12
30
12
drying, shrinkage occurred and the relationship between thickness (m) and moisture
content on a dry basis could be correlated in the linear form:
e = ed (1 + X ),
(2.5.1)
where e is the thickness of product (m), ed is the thickness of the dried product (m), X is
the moisture content on a dry basis (kg kg1 ), and is the linear shrinkage coefficient
(=1.3).
The weight measurement was accurate to about 0.2 g. During drying, regulated drying
air temperature with particular velocity and temperature was fed into the rectangular
casing so that it flowed gently above the sample. The drying air temperatures and
velocities used for each experiment are listed in Table 2.1. The stable humidity of drying
air was maintained and measured using a capacitive transmitter sensor. The temperature
of the sample surface was measured using an optical pyrometer and the temperature of
the upper and lower side of the Petri dish was measured with thermocouples with an
uncertainty of about 2.5 C. The results of this previous study showed that temperature
gradient inside the Petri dish and product can be ignored (Allanic et al., 2006).
2.5.1
Mathematical modelling of convective drying of droplets of whey

protein concentrate (WPC) using the L-REA
The L-REA explained in Section 2.1 is implemented here to model the moisture content
and temperature profiles during the convective drying of WPC. The relative activation
energy is generated from one accurate drying run. The activation energy and equilibrium
activation energy are evaluated using Equations (2.1.5) and (2.1.7), respectively. The
mass balance implementing the L-REA shown in Equation (2.1.4) is used with the
convective mass transfer coefficient (hm ), determined based on the work of Lin and Chen
(2002):
Sh = 1.54 + 0.54 Re0.5 Sc0.333 ,
(2.5.2)
where Sh is the Sherwood number, Re is the Reynolds number and Sc is the Schmidt
number.
The heat balance of the convective drying of WPC can be represented as:
d(mC p T )
dX
h A(Tb T ) + m s
Hv ,
dt
dt
(2.5.3)
1.0
67.5C
87.1C
106.6C
Curve fitted
0.8
E/Eb
52
0.6
0.4
0.2
0.0
0.0
0.5
1.0
1.5
XXb (kg/kg)
2.0
2.5
Figure 2.7 The relative activation energy of convective drying of WPC at different drying air
temperatures. [Reprinted from Chemical Engineering and Processing, 46, S.X.Q. Lin and X.D.
Chen, The reaction engineering approach to modelling the cream and whey protein concentrate
droplet drying, 437443, Copyright (2012), with permission from Elsevier.]
where m is the mass of droplets during drying (kg), Cp is the specific heat of the samples
(J kg1 K1 ), T is the sample temperature (K), Tb is the drying medium temperature
(K), Hv is the vaporisation heat of water (J kg1 ) and h is the heat transfer coefficient
(W m2 K1 ) which can be evaluated by (Lin and Chen, 2002):
N u = 2.04 + 0.62 Re0.5 Pr0.333 ,
(2.5.4)
where Nu is the Nusselt number and Pr is the Prandtl number.

The relative activation energy of the WPC is generated from one accurate drying run.
It is shown in Figure 2.7 and can be expressed as (Lin and Chen, 2007):

E v
= 1.335 0.3669 exp exp(X X b )0.3011 ,
E v,b
(2.5.5)
while the droplet diameter changes during drying can be expressed as (Lin and Chen,
2007):
d
X Xb
= 0.873 + 0.127
.
d0
X0 Xb
(2.5.6)
The good agreement between Equation (2.5.6) and experimental diameter changes during
drying is shown in Figure 2.8.
The profiles of moisture content and temperature during drying are generated by
solving the mass balance implementing the L-REA and the heat balance shown in
Equations (2.1.5) and (2.5.3), respectively, in conjunction with the equilibrium activation
energy, relative activation energy and droplet diameter changes during drying shown in
Equations (2.1.5), (2.1.7) and (2.5.6), respectively.
53
1.0
d/d0
1.0
0.9
0.9
0.0
67.5C
87.1C
106.6C
Curve fitted
0.5
1.0
1.5
XXb (kg kg1)
2.0
2.5
Figure 2.8 The droplet diameter changes during convective drying of WPC. [Reprinted from
Chemical Engineering and Processing, 46, S.X.Q. Lin and X.D. Chen, The reaction engineering
approach to modelling the cream and whey protein concentrate droplet drying, 437443,
2.5.2
Mathematical modelling of convective drying of a mixture of polymer

solutions using the L-REA
The L-REA shown in Equation (2.1.4) used here is similar to formulation of the L-REA
used in the convective drying of WPC. The activation energy and equilibrium activation
energy is also calculated using Equations (2.1.5) and (2.1.7), respectively.
During convective drying, it can be seen that the sample is heated from the upper side
due to forced convective heat transfer from the drying air, while the lower side is heated
through the Petri dish which is heated by drying air from below (refer to Figure 2.9).
The heat balance can be written as:
ms d X
d( C p eT )
= h top (Tb T ) + Ubottom (Tb T ) +
Hv (T ),
dt
A dt
(2.5.7)
where is sample density (kg m3 ), Cp is sample heat capacity (J kg1 K1 ),

e is sample thickness (m), T is sample temperature (K), Tb is drying air temperature (K), ms is mass of dried product (kg), A is surface area of product (m2 ), X is
moisture content of product (kg kg1 ), Hv is enthalpy of vaporisation (J kg1 ),
htop is heat transfer coefficients at the upper surface of the dish and Ubottom represents the overall heat transfer coefficient from the lower side including convection
(natural) along and conduction through the Petri dish. Rearranging Equation (2.5.7)
results in:
ms d X
d( C p eT )
= Utotal (Tb T ) +
Hv (T ),
dt
A dt
(2.5.8)
54
Product
Petri dish
e(t)
z
Thermocoupl
Evaporation
Convection
Diffusion
Conduction
Figure 2.9 Heat transfer mechanisms of the convective drying of a mixture of polymer solutions.
[Reprinted from Chemical Engineering and Processing: Process Intensification, 49, A. Putranto,
X.D. Chen and P.A. Webley, Infrared and convective drying of thin layer of polyvinyl alcohol
(PVA)/glycerol/water mixture The reaction engineering approach (REA), 348357, Copyright
where Utotal represents the sum of the convective heat transfer coefficient (W m2 K1 )
at the top and that at bottom. This value can be deduced from the constant rate period of
drying. The heat balance for this period can be expressed as:
Utotal (Tb T ) =
ms d X
Hv (T ).
A dt
(2.5.9)
For this experiment, the activation energy is determined based on previously published
experimental data (Allanic et al., 2009) using Equation (2.1.5). The vapour concentration
in the environment is determined from the corresponding relative humidity and drying air
temperature reported previously (shown in Table 2.1). The mass transfer coefficient was
deduced from the established Sherwood number correlation. Based on drying kinetics
data, the relative activation energy (Ev /Ev ,b ) for convective drying calculated through
this exercise is expressed as:

E v
= exp 1.0794(X X b )1.28 .
E v,b
(2.5.10)
Only one set of drying data was necessary and this was taken from experiment at a
drying air temperature of 35 C, drying air velocity of 1 m s1 and relative humidity
of 30% (Allanic et al., 2009). This is of a similar format to that proposed previously
(Chen and Xie, 1997; Chen and Lin, 2005). As Figure 2.10 shows, there is excellent
agreement between correlated and experimental activation energy (R2 = 0.9892). At
high water content, moisture removal is easy, as shown by the low activation energy, and
this increases during drying as moisture content decreases indicating greater difficulty
55
1.0
Data
Fitted curve
0.8
Ev/Ev,b
0.6
0.4
0.2
0.2
3
4
5
XXb (kg water/kg dry solid)
Figure 2.10 Normalised activation energy and fitted curve of polyvinyl alcohol/glycerol/water
under convective drying at an air temperature of 35 C and relative humidity of 30%. [Reprinted
from Chemical Engineering and Processing: Process Intensification, 49, A. Putranto, X.D. Chen
and P.A. Webley, Infrared and convective drying of thin layer of polyvinyl alcohol
in removing moisture. Also, this correlation ensures that Ev /Ev ,b is 1 when the
moisture content approaches equilibrium (i.e. X = Xb ).
In order to evaluate the moisture content and the temperature profile as a function
of drying time, the mass balance and heat balance expressed in Equations (2.1.4) and
(2.5.8), respectively were solved simultaneously in conjunction with the equilibrium and
relative activation energy shown in Equations (2.1.7) and (2.5.10), respectively.
2.5.3
Results of modelling convective drying of droplets of WPC using the L-REA

The results of modelling of the convective drying of the droplets of WPC using the
L-REA are shown in Figure 2.11. Generally, the predictions using the L-REA match
well with the experimental data. For the convective drying at a drying air temperature
of 67.5 C, the results of modelling of moisture content and temperature profiles match
the experimental data well as shown in Figure 2.11(a). In addition, the L-REA describes
well the moisture content and temperature profiles of the convective drying of WPC
at drying air temperatures of 87.1 and 106.6 C, as depicted in Figures 2.11(b) and
(c), respectively. For all experiments, the average absolute differences between the
1.6E-06
1.2E-06
60
1.0E-06
8.0E-07
40
6.0E-07
4.0E-07
(b)
50
100
150
200
Time (s)
250
300
1.6E-06
20
350
100
Droplet weight (kg)
1.4E-06
Model pred.
Exp. data
1.2E-06
1.0E-06
80
60
8.0E-07
40
6.0E-07
4.0E-07
(c)
50
100
150
200
Time (s)
250
300
1.6E-06
20
350
120
1.4E-06
100
Model pred.
Exp. data
1.2E-06
80
1.0E-06
60
8.0E-07
40
6.0E-07
4.0E-07
Droplet temperature (C)
Droplet weight (kg)
1.4E-06
80
Model pred.
Exp. data
50
100
150
Time (s)
200
250
(a)
Droplet weight (kg)
56
20
300
Figure 2.11 The comparison between experimental and model prediction using the L-REA of
convective drying of WPC at drying air temperatures of (a) 67.5 C (b) 87.1 C (c) 106.6 C.
[Reprinted from Chemical Engineering and Processing, 46, S.X.Q. Lin and X.D. Chen, The
reaction engineering approach to modelling the cream and whey protein concentrate droplet
drying, 437443, Copyright (2012), with permission from Elsevier].
57
8
Model
Data
X (kg water/kg dry solid)
6
5
4
3
2
1
0
500
1000 1500 2000 2500 3000 3500 4000 4500 5000

t(s)
Figure 2.12 Moisture content profile of convective drying at air temperature of 55 C, air velocity
of 2.8 m s1 and air relative humidity of 12%. [Reprinted from Chemical Engineering and
Processing: Process Intensification, 49, A. Putranto, X.D. Chen and P.A. Webley, Infrared and
convective drying of thin layer of polyvinyl alcohol (PVA)/glycerol/water mixture The reaction
engineering approach (REA), 348357, Copyright (2010), with permission from Elsevier.]
experiments and predictions are about 2.1% of initial droplet weight for the droplet
weight prediction and about 1.9 C for the temperature prediction (Lin and Chen, 2007).
It has been shown that the L-REA can model the convective drying of WPC accurately.
This could be due to the accuracy of the relative activation energy in capturing the
physics of the convective drying of WPC. The combination between the equilibrium
relative energy and relative activation energy shown by Equations (2.1.7) and (2.5.10),
respectively seems to be sufficient to describe the change of internal behaviour of the
WPC droplets during drying. Therefore, it can be said that the L-REA can describe the
drying kinetics of the particulates well.
2.5.4
Results of modelling convective drying of a thin layer of a mixture of

polymer solutions using the L-REA
Figures 2.12 to 2.17 present results of the simulated drying profiles and temperature
profiles using the L-REA. It can be seen that generally, all moisture content and temperature profiles agree well with the experimental data supported by R2 and RMSE of
moisture content and temperature profile presented in Table 2.2. Figures 2.12 and 2.13
show the moisture content and temperature profile of convective drying conducted at
55 C and relative humidity of 12% at air velocity of 2.8 m s1 . Figure 2.12 indicates that
Table 2.2 R 2 and RMSE of modelling of a mixture of

polymer solutions using the L-REA.
Number
R2 X
R2 T
RMSE X
RMSE T
1
2
3
0.999
0.998
0.997
0.958
0.991
0.975
0.071
0.083
0.104
2.263
0.436
1.448
330
325
320
Temperature (K)
58
315
310
305
300
295
Model
Data
0
500
1000 1500 2000 2500 3000 3500 4000 4500 5000

t(s)
Figure 2.13 Product temperature profile of convective drying at an air temperature of 55 C, air
velocity of 2.8 m s1 and air relative humidity of 12%. [Reprinted from Chemical Engineering
and Processing: Process Intensification, 49, A. Putranto, X.D. Chen and P.A. Webley, Infrared
and convective drying of thin layer of polyvinyl alcohol (PVA)/glycerol/water mixture The
reaction engineering approach (REA), 348357, Copyright (2010), with permission from
Elsevier.]
moisture content profile can be predicted very well by the L-REA. Similarly, the temperature profile shown by Figure 2.13 indicates very small differences between predicted
and experimental data. This modelling is comparable with modelling of drying kinetics
conducted by Allanic et al. (2009). Slight discrepancies with experimental temperature
data were also shown although the model employed was based on a diffusion partial
differential equation with fitted diffusivity (Allanic et al., 2009).
Figures 2.14 and 2.15 provide results of modelling of drying conducted at 35 C and
relative humidity of 30% at an air velocity of 1 m s1 using the REA. A very good
prediction of both moisture content and temperature data was observed. Compared with
the simulation using the model proposed previously, it is apparent that the L-REA gives
8
Model
Data
6
5
4
3
2
1
0
2000
4000
6000
t(s)
8000
10000
12000
Figure 2.14 Moisture content profile of convective drying at an air temperature of 35 C, air
velocity of 1 m s1 and air relative humidity of 30%. [Reprinted from Chemical Engineering and
308
306
Temperature (K)
304
302
300
298
296
294
Model
Data
0
2000
4000
6000
t(s)
8000
10000
12000
Figure 2.15 Product temperature profile of convective drying at air temperature of 35 C, air
8
Model
Data
6
5
4
3
2
1
0
1000
2000
3000
4000
t(s)
5000
6000
7000
8000
325
Temperature (K)
320
315
310
305
Model
Data
300
1000
2000
3000
4000
t(s)
5000
6000
7000
8000
61
Table 2.3 Experimental conditions of convective drying of mango tissues

(Vaquiro et al., 2009).
Number
Air velocity
(m s1 )
Air temperature
(C)
Air humidity
(kg H2 O kg dry air1 )
1
2
3
4
4
4
45
55
65
0.0134
0.0134
0.0134
better results because the diffusion model shows slight discrepancies in the moisture
content profile during drying times around 400010 000 s (Allanic et al., 2009). The
L-REA can be used to describe this well, as shown in Figure 2.14.
Similarly, Figures 2.16 and 2.17 show a good agreement between estimated and experimental moisture and temperature data. Despite the simplicity of L-REA, it compares
well with the model proposed before, which shows some discrepancies of moisture
profile during drying times around 30006000 s (Allanic et al., 2009).
Overall the L-REA can be used successfully to model the thin layer drying of a mixture
of polyvinyl alcohol, glycerol and water. The L-REA is shown to be able to model not
only the convective drying of particulate or thin layer of food materials which has been
proven before (Chen and Lin, 2005; Lin and Chen, 2005; 2006; 2007), but also that
of thin layers of non-food materials. The accuracy of the L-REA could be due to the
accuracy of the relative activation energy in describing the change of internal behaviour
during drying.
2.6
Convective drying of thick samples modelled using the L-REA

For studying simulations of convective drying of thick samples using the L-REA, the
experimental data are derived from the work of Vaquiro et al. (2009) on convective
drying of mango tissues. Mango tissues used for drying experiments were formed into
cubes of side lengths of 2.5 cm, with initial moisture content of 9.3 kg kg1 and an
initial temperature of 10.8 C. Drying was conducted in a laboratory dryer described
in detail by Sanjuan et al. (2004). The drying air temperature and air velocity were
controlled at preset values by PID control algorithms while air humidity was maintained at a constant during drying. Details of the experimental conditions are listed in
Table 2.3. The weight of the sample was measured periodically to record weight loss as
well as centre temperatures every 2 min.
2.6.1
Formulation of the L-REA for convective drying of thick samples

In order to model the convective drying of thick samples, the original formulation of the
L-REA can still be implemented. However, the temperature of concern is the surface
62
temperature (Ts ) (Putranto et al., 2011a,b). Therefore, the drying rate of the material can
be expressed as:
ms

dX
= h m A v,s (Ts ) v,b ,
dt
(2.6.1)
where ms is the dried mass of thin layer material (kg), t is time (s), X is moisture
content on a dry basis (kg kg1 ), v,s is the vapour concentration at the material-air
interface (kg m3 ), v,b is the vapour concentration in the drying medium (kg m3 ), hm
is the mass transfer coefficient (m s1 ) and A is the surface area of the material (m2 ).
The mass transfer coefficient (hm ) is determined based on the established Sherwood
number correlations for the geometry and flow condition of concern or established
experimentally for the specific drying conditions involved (Lin and Chen, 2002; Kar
and Chen, 2009). The surface vapour concentration ( v,s ) can be scaled against saturated
vapour concentration ( v,sat ) using the following equation (Chen and Xie, 1997; Chen,
2008):

v,s = exp
E v
RTs

v,sat (Ts ),
(2.6.2)
where Ev represents the additional difficulty in removing moisture from the material
beyond the free water effect. This Ev is moisture-content (X) dependent. Ts is the
surface temperature of the material being dried, and v,sat for water can be estimated at
the surface material being dried by the following equation:
v,sat = 4.844 109 (Ts 273)4 1.4807 107 (Ts 273)3 + 2.6572
105 (Ts 273)2 4.8613 105 (Ts 273) + 8.342 103 ,
(2.6.3)
based on the data summarised by Keey (1992).

The mass balance (Equation 2.6.1) is then expressed as:

dX
E v
= h m A exp
v,sat (Ts ) v,b .
ms
dt
RTx
(2.6.4)
The activation energy (Ev ) is determined experimentally by placing the parameters

required for Equation (2.6.4) in its rearranged form:
E v = RTs ln
m s ddtX
1
hm A
+ v,b
v,sat (Ts )
(2.6.5)
The equilibrium activation energy (Ev,b ) is still evaluated by Equation (2.1.7). It can
be shown that the general formulation of the L-REA shown in Equation (2.1.4) can still
be implemented but the temperature of concern is the surface temperature (Ts ).
2.6.2
63
Prediction of surface sample temperature

For large sample slabs, prediction of sample temperature may be necessary since the
temperature may be not uniform inside the sample. The sample temperature may be
approximated using a simple parabolic equation (Chen, 2008):
T = a + bx 2 .
(2.6.6)
If To is the centre sample temperature (K) and L is the half-thickness of the sample as a
characteristic slab length, Equation (2.6.6) is rewritten as:

Ts To
(2.6.7)
T = To +
x 2,
L2
Tavg is determined by:
L
T (x)d x
.
(2.6.8)
L
By combining Equations (2.6.7) and (2.6.8), Tavg is expressed as:
1
2
Tavg = Ts + To .
(2.6.9)
3
3
For a sample heated in a convective environment, the boundary condition at sample
surface (x = L) can be written as:

dT
+ |Nv |Hv .
(2.6.10)
h (Tb Ts ) = k
d x x=L
Tavg =
Also note that Equation (2.6.7) satisfies the boundary condition at centre (x = 0), which
can be expressed as:

dT
= 0.
(2.6.11)
d x x=0
By combining Equation (2.6.7) to (2.6.10), Ts and To are expressed as:
L
hL
Tavg +
Tb |Nv |Hv
3k
3k ,
(2.6.12)
Ts =
hL
1+
3k

1
hL
L
1

3

Tb |Nv |Hv
. (2.6.13)
To = Tavg
2h L
2h L
2
3k
3k
2+
2+
3k
3k
Equation (2.6.12) and (2.6.13) clearly show that Ts and To are represented as functions
of Tavg and Tb .
The temperature profile prediction described previously seems to be valid for drying
conditions suitable for the boundary conditions mentioned (i.e. there is symmetry at
centre and at the surface; heat gained by convection from drying air is balanced by
conduction heat inside the sample and heat for water evaporation). The prediction is
in agreement with Pang (1994), who conducted convective drying of softwood and
64
heartwood with a half-thickness of 2.5 cm. It was observed that the boundary conditions
indicated in Equations (2.6.10) and (2.6.11) fulfil the drying conditions of Pang (1994).
The temperatures in several positions (x = 0, 7, 13, 19 and 25 cm from centre) were
measured during the drying time and a plot of the temperature profiles against positions
during drying time revealed parabolic profiles.
For drying of mango tissue, as mentioned before, the sample was heated uniformly
from all directions (Sanjuan et al., 2004). It is reasonable to assume that the temperature
profiles would be similar in the x, y and z directions. Because of this, the approximation
of the temperature profiles can be simplified into one dimension.
It is also observed that for drying of mango tissues, there is symmetry at the centre
and at the surface; heat received by convection from drying air is balanced by conduction
heat inside the sample and heat for water evaporation is represented by the boundary
conditions shown in Equations (2.6.10) and (2.6.11) (Sanjuan et al., 2004; Incropera
and DeWitt, 2002). Therefore, similarly to Equations (2.6.7), (2.6.12) and (2.6.13),
showing the temperature distribution inside mango and apple tissues, surface and centre
temperature can be represented as:

Ts To
(2.6.14)
T = To +
r 2,
R2
R
hR
Tavg +
Tb |N V |HV
3k
3k ,
Ts =
(2.6.15)
hR
1+
3k

hR
R
1
1
3
|N
|H

T
. (2.6.16)
To = Tavg
b
V
V
2h R
2
3k
3k
2 + 2h3kR
2+
3k
The equivalent radius for cubes is the side length (Incropera and DeWitt, 2002; Radziemska and Lewandowski, 2008). Because of the symmetry principle used for Equations
(2.6.14) to (2.6.16), the equivalent radius (r) used for cubes in this study are half the
side length.
2.6.3
Modelling convective drying thick samples of mango tissues using the L-REA
For drying thick samples of mango tissues convectively, the relative activation energy
(Ev /Ev,b ) is generated from continuous convective drying runs at 55 C (Vaquiro
et al., 2009). Based on drying kinetics data, the relative activation energy (Ev /Ev,b )
of convective drying of mango tissues is expressed as:
E v
= 9.92 104 (X X b )3 + 9.74 103 (X X b )2
E v,b
0.101(X X b ) + 1.053.
(2.6.17)
A good agreement between the fitted (Equation 2.6.17) and experimental activation
energy is shown in Figure 2.18 (R 2 (0.997)). This format of correlation is similar to that
proposed by Kar (2008) to describe the activation energy of drying porcine skin. The
65
1
Data
Model
0.9
0.8
Ev/Ev,b
0.7
0.6
0.5
0.4
0.3
0.2
0.1
5
XXb
10
Figure 2.18 The relative activation energy (Ev /Ev,b ) of convective drying of mango tissues at
an air velocity of 4 m s1 , drying air temperature of 55 C and air humidity of 0.0134 kg H2 O kg
dry air1 . [Reprinted from Drying Technology, 29, A. Putranto, X.D. Chen and P.A. Webley,
Modelling of drying of food materials with thickness of several centimeters by the reaction
engineering approach (REA), 961973, Copyright (2012), with permission from Taylor &
Francis Ltd.]
format of the equation could be varied but for this study the Equation (2.6.17) seems to
represent the activation energy well. It may be observed that the decrease of moisture
content results in the increase of activation energy, which indicates greater difficulty in
removing water. This equation also yields Ev /Ev,b approaching 1 as the material is
dried.
The heat balance for convective drying of mango tissues can be written as:
dX
d(mC p Tavg )
h A (Tb Ts ) + m s
HV ,
dt
dt
(2.6.18)
where m is the sample mass (kg), Cp is the heat capacity of the sample (J kg1 K1 ),
h is the heat transfer coefficient (W m2 K1 ) and HV is the latent heat of vaporisation
of water (J kg1 ). The drying rate dX/dt is negative when drying occurs. In order to yield
both profiles of moisture content and temperature of mango tissues during drying, the
mass implementing the L-REA and heat balance shown in Equations (2.6.4) and (2.6.18)
are solved simultaneously in conjunction with the equilibrium and relative activation
energy shown in Equations (2.1.7) and (2.6.17), respectively. The surface temperature
predicted by Equation (2.6.15) is used in the mass balance implementing the L-REA
and heat balance.
66
Table 2.4 R 2 and RMSE of modelling of convective drying of mango tissues using the L-REA.
Number
Velocity
(m s1 )
Air temperature
(C)
Air humidity
R2 X
RMSE X
R2 T
RMSE T
1
2
3
4
4
4
45
55
65
0.0134
0.0134
0.0134
0.998
0.998
0.996
0.08
0.1
0.14
0.993
0.982
0.984
0.61
1.12
1.41
10
Data 45C
Data 55C
Data 65C
Model 45C
Model 55C
Model 65C
8
7
6
5
4
3
2
1
0
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 2.19 Moisture content profile of convective mango tissues at air temperatures of 45, 55
and 65 C (modelled using the L-REA which incorporates the temperature distribution inside the
sample). [Reprinted from Drying Technology, 29, A. Putranto, X.D. Chen and P.A. Webley,
engineering approach (REA), 961973, Copyright (2012), with permission from
Taylor & Francis Ltd.]
2.6.4
Results of convective drying thick samples of mango tissues using the L-REA
From Figures 2.19 and 2.20, a good agreement between the experimental and predicted
data is observed for convective drying of mango tissues at drying air temperatures of
45, 55 and 65 C. The good predictions made by using the REA are further revealed by
R 2 and RMSE presented in Table 2.4, which shows all modelling of these cases yield R 2
of moisture content and temperature profiles higher than 0.996 and 0.982, respectively,
as well as RMSE of moisture content and temperature profiles lower than 0.14 and 1.41,
respectively. On the other hand, Figures 2.21 and 2.22 show discrepancies between the
predicted and experimental data. It is clear that the L-REA with the approximation of
340
Centre temperature (K)
330
320
310
Data 45C
Data 55C
Data 65C
Model 45C
Model 55C
Model 65C
300
290
280
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 2.20 Temperature profile of convective mango tissues at air temperatures of 45, 55 and
65 C (modelled using the L-REA which incorporates the temperature distribution inside the
Francis Ltd.]
10
Data 45C
Moisture content (kg water/kg dry solid)
Data 55C
8
Data 65C
Model 45C
Model 55C
Model 65C
5
4
3
2
1
0
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 2.21 Moisture content profile of convective mango tissues at air temperatures of 45, 55
and 65 C (modelled using the L-REA without approximation of temperature distribution inside
the sample). [Reprinted from Drying Technology, 29, A. Putranto, X.D. Chen and P.A. Webley,
Francis Ltd.]
340
330
68
320
310
Data 45C
Data 55C
Data 65C
Model 45C
Model 55C
Model 65C
300
290
280
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 2.22 Temperature profile of convective mango tissues at air temperatures of 45, 55 and
65 C (modelled using the L-REA without approximation of temperature distribution inside the
Francis Ltd.]
temperature distribution inside the sample is necessary, and this model describes both
moisture content and centre sample temperature profile well during drying.
Vaquiro et al. (2009) used diffusion-based modelling to represent the data and our
REA compares well with the modelling by Vaquiro et al. (2009). Modelling by Vaquiro
et al. (2009) showed a kink in the beginning of the temperature profile that was not
observed by modelling using the REA. For drying at 65 C, both the REA and modelling
proposed by Vaquiro et al. (2009) showed a slight overestimation of the temperature
profile during drying times of 500020 000 s.
It can be said that the L-REA, with the prediction of sample temperature as explained
in Section 2.6.2, is accurate enough to describe continuous convective drying of mango
tissues well. It also compares favourably with the model proposed by Vaquiro et al.
(2009) in spite of the simplicity of the L-REA. While the results are accurate, the
modelling itself is still simple and only requires a short computational time to predict
the drying kinetics accurately. This shows that the L-REA is effective for modelling
thick samples of mango tissues.
A new and innovative application of the L-REA has been implemented in this study to
describe both the moisture content and sample temperature profile of convective drying
large samples of mango tissues. For this purpose, the activation energy and the saturation
vapour concentration are evaluated at the surface temperature. The remaining principles
69
Table 2.5 Schemes of intermittent drying of mango tissues (Vaquiro et al., 2009).
Drying air
temperature (C)
Period of first
heating (s)
Period of resting
(at 27 C 1.6) (s)
Period of second
heating (s)
45
55
65
16 200
9 480
7 800
10 800
10 800
10 800
36 360
33 720
16 200
are similar to those of the L-REA used to describe the drying kinetics of thin layers or
small objects published previously. Results indicate that the REA models both moisture
content and temperature of convective drying of large samples of mango tissues very
well. When compared to the experimental data published by Vaquiro et al. (2009), a
similar if not better agreement is observed against diffusion-based models. While the
results are accurate, the effectiveness of the L-REA is also revealed as the modelling
itself is still simple and only requires a short amount of computational time. Therefore,
this work has extended the application of the L-REA to handle drying of thick samples
substantially. The L-REA can model not only the drying of thinlayer or small objects,
but also drying of thick samples.
2.7
The intermittent drying of food materials modelled using the L-REA

In this study, the experimental data are derived from the work of Vaquiro et al. (2009)
whose experimental details are briefly reviewed in Section 2.6. The intermittency is
created by the heating and resting period listed in Table 2.5. During the resting period,
the samples stay in an environment with an ambient temperature of 27 1.6 C and a
relative humidity of 60%.
2.7.1
Mathematical modelling of intermittent drying of food materials using the L-REA

Since the sample is relatively thick, the surface temperature is incorporated in modelling.
The L-REA shown in Equation (2.6.4) is used for modelling. Equations (2.6.15) and
(2.6.16) are used to predict the surface and centre temperatures, respectively. The relative
activation energy shown in Equation (2.6.17) and the heat balance shown in Equation
(2.6.18) are also applied to the modelling here. For modelling intermittent drying, the heat
balance is employed according to the drying air temperature in each section. In addition,
the equilibrium activation energy shown in Equation (2.1.7) is evaluated according to
the corresponding drying air temperature and humidity in each drying period.
2.7.2
The results of modelling of intermittent drying of food materials using the L-REA
Figures 2.23 to 2.28 show the results of modelling of intermittent drying of mango tissues
using the REA. For intermittent drying at a drying air temperature of 45 C, the REA
10
Model
Data
8
7
6
5
4
3
2
1
0
t(s)
7
104
Figure 2.23 Moisture content profile of mango tissues during intermittent drying at a drying air
temperature of 45 C and resting at 27 C. [Reprinted from Industrial Engineering Chemistry

Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley, Intermittent drying of mango
tissues: Implementation of the reaction engineering approach, 10891098, Copyright (2012),
with permission from the American Chemical Society.]
320
315
70
310
305
300
295
290
Model
Data
285
280
4
t(s)
7
104
Figure 2.24 Temperature profile of mango tissues during intermittent drying at a drying air
71
10
Model
Data
8
7
6
5
4
3
2
1
0
3
t(s)
6
104
temperature of 55 C and resting at 27 C [Reprinted from Industrial Engineering Chemistry

330
325
320
315
310
305
300
295
290
Model
Data
285
280
3
t(s)
6
104
temperature of 55 C and resting at 27 C [Reprinted from Industrial Engineering Chemistry
10
Model
Data
8
7
6
5
4
3
2
1
0
0.5
1.5
2
t(s)
2.5
3.5
4
104

340
330
72
320
310
300
290
280
0
Model
Data
0.5
1.5
2
t(s)
2.5
3.5
104
73
Table 2.6 R 2 and RMSE of modelling of intermittent drying of

mango tissues using the L-REA.
Drying air
temperature (C)
R2 X
R2 T
RMSE X
RMSE T
45
55
65
0.998
0.998
0.998
0.996
0.997
0.997
0.083
0.087
0.082
0.483
0.554
0.686
describes both moisture content and temperature profile very well. Good agreement is
observed between experimental and predicted data. Similar results are also revealed for
intermittent drying at drying air temperatures between 55 and 65 C. The predicted
moisture content and temperature match well with experimental data. The good predictions of moisture content and temperature profile are revealed by R 2 and RMSE shown in
Table 2.6.
The benchmark of modelling proposed by Vaquiro et al. (2009) employing the diffusion model was conducted, and it revealed the REA gives comparable or even better
results. Modelling proposed by Vaquiro et al. (2009) showed a kink in the temperature
profile at the beginning of drying; which was not observed by modelling using the REA.
In addition, the underestimation of the moisture content profile at the last period of drying in drying conditions of 65 C is not revealed by the REA; as shown in the modelling
by Vaquiro et al. (2009).
It can be said that the REA is accurate enough to model intermittent drying of mango
tissues, particularly when it is represented in a lumped model. This is because the relative
activation energy (Ev /Ev,b ) implemented allows the natural transition during drying
times according to the drying scheme as revealed in Figure 2.29. The relative activation
energy keeps increasing during drying, indicating an increase of difficulty removing
water from materials. This increases significantly during the heating period but only
increases slightly during the resting period. This natural transition during drying is not
observed by empirical models and the CDRC (Baini and Langrish, 2007). It was revealed
that empirical approaches could not model the intermittent drying of banana tissues well.
The CDRC might not be able to handle this type of material since the intermittent drying
rate could not be represented simply as a linear and exponential decreasing drying rate
(Baini and Langrish, 2007).
Therefore, this has extended the application of the REA significantly to model not
only continuous drying but also intermittent drying of rather thick samples. Although
the results of modelling are accurate and robust, the simplicity of the modelling is still
proven and only a short computational time is required.
2.7.3
Analysis of surface temperature, surface relative humidity, saturated and

surface vapour concentration during intermittent drying
Analysis of surface temperature, surface relative humidity, as well as saturated and surface water vapour concentration during intermittent drying will assist the determination
1
0.9
0.8
0.7
Ev /Ev,b
74
0.6
0.5
0.4
0.3
0.2
0.1
0
0.5
1.5
2
t(s)
2.5
3.5
104
Figure 2.29 Relative activation energy profile of mango tissues during intermittent drying at a
drying air temperature of 65 C and resting at 27 C. [Reprinted from Industrial Engineering

Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley, Intermittent drying of
mango tissues: Implementation of the reaction engineering approach, 10891098, Copyright
(2012), with permission from the American Chemical Society.]
of the appropriate cycle conditions; that is the length of the drying and resting periods, in
order to minimise energy and final moisture content. The following paragraphs discuss
these parameters in the intermittent drying of mango tissues. The L-REA is applied
here and combined with several equations to yield a profile of surface temperature, surface relative humidity, saturated vapour concentration and surface vapour concentration.
The saturated vapour concentration and surface temperature are evaluated by Equations
(2.1.3) and (2.6.15) while the surface vapour concentration is calculated by Equation
(2.1.2).
The profile of surface relative humidity is shown in Figure 2.30. Humidity decreases
during the heating period while it increases during resting, representing an increase
in surface moisture content. In addition, Figure 2.31 indicates the profiles of surface
temperature and saturated vapour concentration during intermittent drying at a drying
air temperature of 65 C. The profiles of saturated vapour concentration follow the
surface temperature trend. It increases in first section, decreases in second section and
increases again in third section.
However, the profile of surface vapour concentration is different from that of saturated vapour concentration as revealed in Figures 2.32 and 2.33 because the profile of
surface vapour concentration is affected by both surface temperature and surface relative
humidity. It is apparent that the surface vapour concentration increases in the very early
part of drying because of the increase in surface temperature, followed by a decrease
0.7
Surface relative humidity
0.6
0.5
0.4
0.3
0.2
0.1
0
0.5
1.5
2.5
t(s)
3.5
104
Figure 2.30 Surface relative humidity profile of mango tissues during intermittent drying at a
drying air temperature of 65 C and resting at 27 C. [Reprinted from Industrial Engineering
Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley, Intermittent drying of
mango tissues: Implementation of the reaction engineering approach, 10891098, Copyright
(2012), with permission from the American Chemical Society.]
350
0.2
0.1
Saturated vapour
concentration
0.5
1.5
2
t(s)
2.5
300
Surface temperature (K)
Saturated water vapour concentration (kg.m3)
Surface temperature
250
3.5
104
Figure 2.31 Saturated vapour concentration and surface temperature profile of mango tissues
during intermittent drying at a drying air temperature of 65 C and resting at 27 C. [Reprinted
from Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A.
Webley, Intermittent drying of mango tissues: Implementation of the reaction engineering
approach, 10891098, Copyright (2012), with permission from the American Chemical Society.]
0.16
Water vapour concentration (kg.m3)
0.14
0.12
0.1
Saturated
Surface
0.08
0.06
0.04
0.02
0
0.5
1.5
2.5
3.5
104
t(s)
0.06
340
0.05
330
Surface temperature
0.04
320
0.03
310
Surface vapour concentration
300
0.02
0.01
Surface water vapour concentration (kg.m3)
Figure 2.32 Surface and saturated vapour concentration profile of mango tissues during
intermittent drying at a drying air temperature of 65 C and resting at 27 C. [Reprinted from
Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A.
0.5
1.5
2
t(s)
2.5
290
3.5
104
Figure 2.33 Surface vapour concentration and surface temperature profile of mango tissues during
intermittent drying at a drying air temperature of 65 C and resting at 27 C. [Reprinted from

Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A.
10
Moisture content (kg water/kg dry solid)
9
8
7
6
5
4
3
2
1
t(s)
7
104
Figure 2.34 Moisture content profile of intermittent drying of mango tissues with heating (at a
drying air temperature of 45 C) and resting periods of 4000 s each. [Reprinted from Industrial
Engineering Chemistry Research, 50, A. Putranto, Z. Xiao, X.D. Chen and P.A. Webley,
Intermittent drying of mango tissues: Implementation of the reaction engineering approach,
10891098, Copyright (2012), with permission from the American Chemical Society.]
0.08
320
0.06
310
0.04
300
0.02
290
Saturated water vapour concentration (kg.m3)
Surface temperature
Saturated vapour concentration

0.01
4
t(s)
280
104
Figure 2.35 Saturated vapour concentration and surface temperature profile of intermittent drying
of mango tissues with heating (at a drying air temperature of 45 C) and resting periods of 4000 s
each. [Reprinted from Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao,
X.D. Chen and P.A. Webley, Intermittent drying of mango tissues: Implementation of the
reaction engineering approach, 10891098, Copyright (2012), with permission from the
American Chemical Society.]
0.04
320
0.02
300
Surface temperature
Surface vapour concentration

0
7
104
t(s)
280
Figure 2.36 Surface vapour concentration and surface temperature profile of intermittent drying
of mango tissues with heating (at a drying air temperature of 45 C) and resting periods of 4000 s
each. [Reprinted from Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao,
X.D. Chen and P.A. Webley, Intermittent drying of mango tissues: Implementation of the
0.07
Water vapour concentration (kg.m3)
Saturated
Surface
0.06
0.05
0.04
0.03
0.02
0.01
4
t(s)
7
104
Figure 2.37 Surface and saturated vapour concentration profile of intermittent drying of mango
tissues with heating (at a drying air temperature of 45 C) and resting periods of 4000 s each.
[Reprinted from Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao, X.D.
Chen and P.A. Webley, Intermittent drying of mango tissues: Implementation of the reaction
engineering approach, 10891098, Copyright (2012), with permission from the American
Chemical Society.]
79
1
Surface vapour
concentration
0.02
0.5
0.04

0
4
t(s)
7
104
Figure 2.38 Surface vapour concentration and surface relative humidity profile of intermittent
drying of mango tissues with heating (at a drying air temperature of 45 C) and resting periods of
4000 s each. [Reprinted from Industrial Engineering Chemistry Research, 50, A. Putranto, Z.
Xiao, X.D. Chen and P.A. Webley, Intermittent drying of mango tissues: Implementation of the
due to decrease in surface relative humidity. During the initial part of the resting period,
the surface vapour concentration increases significantly as the surface relative humidity
increases dramatically. This is followed by a decrease in the surface vapour concentration because of the low surface temperature, leading to a decrease of saturated vapour
concentration. In the second heating period, the surface vapour concentration continues
to decrease as the surface relative humidity decreases. This analysis is in agreement
with Baini and Langrish (2007) applying a diffusion model to the intermittent drying of
banana tissues. It was revealed that, during the resting period, the surface temperature
decreased, while the surface moisture content increased initially as a result of an initial
increase in surface relative humidity.
It can be seen that, during the resting period of drying times higher than 12 000 s,
the surface vapour concentration reaches a plateau. It means there is actually no point
in extending the resting period to 18 600 s. The resting period could be shortened and
followed by a subsequent heating period. Similarly, during the second heating period,
the surface vapour concentration profile has nearly flattened after a drying time around
25 000 s. The heating time could also be shortened, followed by a subsequent resting
period to achieve higher surface vapour concentration.
80
From this analysis, it seems that a cycle with a higher frequency will give better
results. Simulation of intermittent drying at a drying air temperature of 45 C with each
heating and resting period (at 27 C) at 4000 s and a total drying time of 64 000s (total
heating time at 45 C of 32 000 s) was conducted to illustrate profiles of this scheme.
Results of the simulation, including the profiles of moisture content, surface temperature,
surface-relative humidity, and saturated surface temperature, are presented in Figures
2.342.38.
It can be seen that the trends of surface-relative humidity, saturated and surface vapour
concentration are similar to those which have been discussed in previous paragraphs.
Nevertheless, no flat profile of surface vapour concentration is shown during the resting
period which means resting is not conducted for a prolonged time. It is also observed
that the final moisture content is similar to that of intermittent drying mango tissues at
45 C using scheme listed in Table 2.5, although the total heating time of this scheme
is lower. The total heating time of this scheme is 32 000 s, while that at 45 C, listed in
Table 2.5, is 52 560 s. The total drying time of this scheme (64 000 s) is also similar to
that at 45 C, listed in Table 2.5 (62 650 s). Because the heating time is shorter, while
the total drying time and the final moisture content is not significantly altered (from a
sustainable processing perspective), it is beneficial to apply such a cycle. This is because
energy cost can be minimised while the objective of obtaining a similar target moisture
content can be achieved.
2.8
The intermittent drying of non-food materials under time-varying

temperature and humidity modelled using the L-REA
In this study, the experimental data of the intermittent drying are derived from the
work of Kowalski and Pawlowski (2010a,b) whose experimental details are shown in
Table 2.7. For better understanding of the procedures, the experimental details are briefly
reviewed here.
The materials used for drying is KOC kaolin clay supplied by Surmin-Kaolin, SA Co.,
Poland. The detailed physical and chemical properties of the samples are provided on the
companys website (Surmin-Kaolin Co., 2010). Each sample was prepared by moulding
the materials into a cylinder with a radius of 0.025 m and a height of 0.06 m, with an
initial moisture content of 0.4 kg H2 O kg dry solids1 . Each sample was placed in an
aluminium container suspended on an electronic balance with an accuracy of 0.01 g.
For measurement of the sample temperature, a parallel experiment was conducted. T-type
thermocouples were inserted inside a cylinder at different positions. The temperature
measurement indicated that the temperature inside the sample was uniform (Kowalski
et al., 2007; Kowalski and Pawlowski, 2010a,b).
Two types of intermittent drying experiments were conducted: time-varying drying
air temperature and time-varying humidity. The first type of intermittency was enabled
by supplying cool air through a special air intake. Table 2.7 shows the cases of the
intermittent drying under time-varying drying air temperature and humidity. Case 1
is the intermittent drying, which implemented a periodic change of the drying air
81
Table 2.7 Settings of intermittent drying of kaolin (Kowalski and Pawlowski, 2010).
Case number
Relative humidity
Drying air temperature (C)
7.2% (at 65 C)
4% (at 100 C)
4% (before 9000 s) periodically changed

between 4 and 12% (after 9000 s)
4% (before 9000 s) periodically changed
between 4 and 80% (after 9000 s)
65 C (before 17 000 s) periodically changed

between 65 and 43 C (after 17 000 s)
100 C (before 7000 s) periodically changed
between 100 and 50 C (after 7000 s)
100
100
temperature between 65 and 43 C (see Table 2.7), while Case 2 is similar to the
first one but with a periodic change of temperature between 100 and 50 C (see Table
2.7). For the intermittent drying under time-varying humidity and constant drying air
temperature of 100 C, the intermittency was created by a periodic change in vapour
supply to the drying chamber from a humidifier. Cases 3 and 4 (see Table 2.7) are the
intermittent drying under time-varying humidity. Case 3 applied a periodic change of
the relative humidity between 4 and 12%, while Case 4 implemented a change of the
relative humidity between 480% (Kowalski and Pawlowski, 2010a,b).
2.8.1
Mathematical modelling using the L-REA

The original formulation of the L-REA described in Section 2.1 is still implemented
here without any modification. In this study, the relative activation energy is generated
from a drying experiment at a constant drying air temperature of 50.8 C (Kowalski
et al., 2007) and can be expressed as:
E v
= 0 for X X b > 0.2,
E v,b
7.847

E v
= exp 49.391(X X b )2.103
E v,b
for X X b > 0.2.
(2.8.1)
(2.8.2)
A good fit between the experimental and predicted activation energy is shown in
Figure 2.39 and indicated by the R 2 of 0.99. As mentioned earlier, uniform temperature
profiles inside the product were observed in the work of Kowalski and Pawlowski
(2010a). It has been noted that the ChenBiot number (ChBi) (Chen and Peng, 2005)
for intermittent drying of kaolin is 0.03, which indicates the temperature inside the
sample is essentially uniform. Indeed, Kowalski and Pawlowski (2010b) implemented
modelling which did not take into account the variations of spatial temperature inside
products. Based on this observation, the assumption of uniform product temperature
profile is implemented in this study. Hence, the heat balance can be represented as:
d(mC p T )
dX
h A (Tb T ) + m s
H V ,
dt
dt
(2.8.3)
82
1
Data
Fitted curve
0.9
0.8
Ev /Ev,b
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Figure 2.39 The relative activation energy (Ev /Ev,b ) of the convective drying of kaolin.
[Reprinted from Chemical Engineering Science, 66, A. Putranto, X.D. Chen, S. Devahastin et al.,
Application of the reaction engineering approach (REA) for modelling intermittent drying under
time-varying humidity and temperature, 21492156, Copyright (2012), with permission from
Elsevier.]
where m is the sample mass (kg), Cp is the heat capacity of the sample (J kg1 K1 ), h
is the heat transfer coefficient (W m2 K1 ) and HV is the latent heat of vaporisation
of water (J kg1 ).
In order to incorporate the effects of time-varying drying air temperature or humidity,
the equilibrium activation energy (Ev,b ) shown in Equation (2.1.7) is defined according
to the corresponding drying air settings in each time period. The equilibrium activation
energy (Ev,b ) is combined with the relative activation energy (Ev /Ev,b ) represented
in Equations (2.8.1) and (2.8.2). In addition, the mass balance implementing the L-REA
and heat balance shown in Equations (2.1.4) and (2.8.3), respectively, also implement the
corresponding drying air settings in each time period. The profiles of moisture content
and temperature can be yielded by solving the mass and heat balance simultaneously
in conjunction with the equilibrium and relative activation energy shown in Equations
(2.1.7), (2.8.1) and (2.8.2).
2.8.2
Results of intermittent drying under time-varying temperature and

humidity modelled using the L-REA
Figures 2.40 to 2.43 show the results of modelling of the intermittent drying under timevarying drying air temperature using the L-REA. It can be seen that the L-REA describes
Moisture content (kg water/kg dry solids)
0.45
L-REA
Data
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.5
1.5
2
t(s)
2.5
3.5
4
104
Figure 2.40 Moisture content profile of intermittent drying in Case 1 (periodically changed drying
air temperatures between 65 and 43 C). [Reprinted from Chemical Engineering Science, 66, A.
Putranto, X.D. Chen, S. Devahastin et al., Application of the reaction engineering approach
(REA) for modelling intermittent drying under time-varying humidity and temperature,
340
Temperature (K)
330
320
310
300
290
280
L-REA
Data
0
0.5
1.5
2
t(s)
2.5
3.5
4
10
Figure 2.41 Temperature profile of intermittent drying in Case 1 (periodically changed drying air
temperatures between 65 and 43 C). [Reprinted from Chemical Engineering Science, 66, A.
0.45
L-REA
Data
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.5
1.5
t(s)
2.5
104
Figure 2.42 Moisture content profile of intermittent drying in Case 2 (periodically changed drying
air temperatures between 100 and 50 C). [Reprinted from Chemical Engineering Science, 66,
A. Putranto, X.D. Chen, S. Devahastin et al., Application of the reaction engineering approach
360
350
Temperature (K)
340
330
320
310
300
290
L-REA
Data
0
0.5
1.5
t(s)
2.5
104
Figure 2.43 Temperature profile of intermittent drying in Case 2 (periodically changed drying air
temperatures between 100 and 50 C). [Reprinted from Chemical Engineering Science, 66, A.
85
Table 2.8 R 2 , RMSE, average absolute deviation and maximum absolute deviation of profiles of moisture content
predicted by and Kowalski and Pawlowskis model (2010b).
Average
Absolute
R2
RMSE
Average Deviation
Kowalski and
Kowalski and Absolute Kowalski and
Case
R2
Pawlowskis
RMSE Pawlowskis
Deviation Pawlowskis
number REA model (2010b) REA model (2010b) REA
model (2010b)
Maximum
Absolute
Maximum Deviation
Absolute Kowalski and
REA
model (2010b)
1
2
3
4
0.012
0.012
0.023
0.009
0.997
0.997
0.991
0.998
0.992
0.983
0.994
0.967
0.006
0.007
0.011
0.005
0.009
0.016
0.009
0.019
0.005
0.006
0.009
0.084
0.007
0.014
0.006
0.015
0.024
0.031
0.026
0.037
Table 2.9 R 2 , RMSE, average absolute deviation and maximum absolute deviation of profiles of temperature predicted
by Kowalski and Pawlowskis model (2010b).
Average
Absolute
R2
RMSE
Average Deviation
Kowalski and
Kowalski and Absolute Kowalski and
Pawlowskis
RMSE Pawlowskis
Case
R2
model (2010b)
number REA model (2010b) REA model (2010b) REA
Maximum
Absolute
Maximum Deviation
Absolute Kowalski and
REA
model (2010b)
1
2
3
4
5.968
5.501
5.050
4.369
0.952
0.953
0.983
0.950
0.761
0.842
0.743
0.896
2.529
3.399
1.658
4.554
5.567
6.254
7.441
6.377
1.481
2.554
1.217
0.084
4.335
4.892
6.189
3.995
11.158
10.059
14.472
6.8326
both the moisture content and temperature profiles well. Benchmarks towards the modelling approach implemented by Kowalski and Pawlowksi (2010b) for the intermittent
drying have been conducted. For Case 1 (refer to Table 2.7, Figures 2.40 and 2.41),
L-REA can give even better agreement between the predicted moisture content, temperature profiles and the experimental data than the approach implemented by Kowalski
and Pawlowksi (2010b); as shown by the results of error analysis presented in Tables
2.8 and 2.9. For Case 2 (refer to Table 2.5, Figures 2.42 and 2.43), the L-REA also
yields closer agreement between the predicted and experimental moisture content profiles. The L-REA results in a slight deviation in the temperature profiles during drying
times between 15 000 and 20 000 s, while the modelling implemented by Kowalski and
Pawlowksi (2010b) showed a slight deviation in the temperature profiles during drying
times of 800020 000 s. The closer agreement is indeed shown by the results of error
analysis presented in Tables 2.8 and 2.9.
It can be said that the REA can model the intermittent drying of kaolin under timevarying drying air temperature well. This could be due to the flexibility of the activation
energy in allowing a change in the drying kinetics according to the drying air settings
in each time period. The relative activation energy (Ev /Ev,b ) shown in Equations
0.45
86
L-REA
Data
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.5
1.5
t(s)
2.5
104
Figure 2.44 Moisture content profile of intermittent drying in Case 3 (periodically changed
relative humidity between 4 and 12%). [Reprinted from Chemical Engineering Science, 66, A.
(2.8.1) and (2.8.2) combined with the equilibrium activation energy (Ev,b ) indicated in
Equation (2.1.7) seems to capture accurate physics during the intermittent drying under
time-varying drying air temperature. The results indicated that the application of the
REA has been extended significantly, following that of the L-REA modelling the cyclic
drying of a thin layer of polymer solution mixture of under time-varying infrared-heat
intensity (Putranto et al., 2010b).
In terms of the intermittent drying of kaolin under time-varying humidity, the results
of modelling using the L-REA are shown in Figures 2.442.47. The results of the
intermittent drying for Case 3 (refer to Table 2.7) are indicated in Figures 2.44 and 2.45.
The L-REA models both the moisture content and temperature profiles reasonably well.
Benchmarks towards modelling implemented by Kowalski and Pawlowski (2010b) have
also been conducted and the L-REA yields better agreement with the experimental data.
The observed deviations between the experimental and predicted temperature profiles
observed from the approach of Kowalski and Pawlowski (2010b) do not appear when
the L-REA is used.
Similarly to Case 3 (refer to Table 2.7, Figures 2.44 and 2.45), for Case 4 (refer to
Table 2.7, Figures 2.46 and 2.47) the REA results are in a good agreement between both
the predicted moisture content and temperature profiles and the experimental data. The
slight deviation of the temperature profiles during drying times of 13 000 and 18 000 s are
also observed in the modelling applied by Kowalski and Pawlowski (2010b). However,
370
360
Temperature (K)
350
340
330
320
310
L-REA
Data
300
290
0.5
1.5
t(s)
2.5
104
Figure 2.45 Temperature profile of intermittent drying in Case 3 (periodically changed relative
humidity between 4 and 12%). [Reprinted from Chemical Engineering Science, 66, A. Putranto,
X.D. Chen, S. Devahastin et al., Application of the reaction engineering approach (REA) for
modelling intermittent drying under time-varying humidity and temperature, 21492156,
0.4
L-REA
Data
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.5
1.5
t(s)
2.5
104
Figure 2.46 Moisture content profile of intermittent drying in Case 4 (periodically changed
relative humidity between 4 and 80%). [Reprinted from Chemical Engineering Science, 66, A.
88
370
360
Temperature (K)
350
340
330
320
310
L-REA
Data
300
290
0.5
1.5
t(s)
2.5
104
Figure 2.47 Temperature profile of intermittent drying in Case 4 (periodically changed relative
humidity between 4 and 80%). [Reprinted from Chemical Engineering Science, 66, A. Putranto,
X.D. Chen, S. Devahastin et al., Application of the reaction engineering approach (REA) for
modelling intermittent drying under time-varying humidity and temperature, 21492156,
the observed overestimation at a drying time of 800013 000 s is not evident when
modelling using the L-REA. Tables 2.8 and 2.9 indicate that the L-REA yields closer
agreement with the experimental data than the other model.
It can be shown that the L-REA is indeed accurate enough to model intermittent
drying under time-varying humidity. The relative activation energy in conjunction with
the equilibrium activation energy seems to be able to capture the periodically changed
relative humidity well. Because of this, the L-REA is shown to be able to model intermittent drying, not only under time-varying temperatures but also under time-varying
humidity, well. The simplicity of the L-REA is also proven for this challenging drying
system.
2.9
The heating of wood under linearly increased gas

temperature modelled using the L-REA
Wood may contain several harmful chemicals as a result of chemical processing, including chromated copper arsenate (CCA), creosote and pentachloro-phenol (Younsi et al.,
2006a). Several methods of disposing such harmful chemicals were reviewed in research
by Helsen and Bulck (2005). Recycling and recovery, chemical extraction, bioremediation, electrodialytic remediation and thermal destruction may be attempted to remove
89
Table 2.10 Settings of heat treatment of wood samples (Younsi et al., 2006a; 2007).
Case
Final gas
temperature (C)
Heating rate
(C h1 )
Initial moisture content

(kg H2 O kg dry solids1 )
1
2
3
4
5
220
220
220
160
200
30
20
10
20
20
0.11
0.125
0.12
0.1059
0.07
the contaminants. Reuse and recycling contaminants has several disadvantages, including contamination to people handling the process as sorting, transportation and storage
are required, while chemical extraction is not an effective process since the kinetics of
extraction is relatively slow and many steps of extraction are necessary. In addition,
the remediation may result in lower quality wood fibre. The thermal destruction process has advantages of possible energy recovery and significant reduction in volume.
However, intensive research still needs to be carried out to evaluate and optimise the
process (Helsen and Bulck, 2005). Thermal destruction can be carried out by hightemperature treatment processes in which wood samples are exposed to hot gas with
linearly increased temperatures beyond 200 C. Fundamentally, it is a drying process
under linearly increased gas temperature according to heating rate (Younsi et al., 2006a).
The accuracy and robustness of the REA for high-temperature treatment of wood is
shown by published experimental data from Younsi et al. (2006a; 2007). The experimental details were reported in Kocaefe et al. (1990; 2007) and Younsi et al. (2006b) and are
reviewed here for better understanding of the current approach. The heat treatment was
conducted in a thermogravimetric analyser (Kocaefe et al., 1990). Wood samples with
dimensions of 0.035 0.035 0.2 m were heat treated by suspending the samples on
a balance with an accuracy of 0.001 g. The heat treatment was conducted by exposing
the samples to hot gas whose temperature was linearly increased according to heating
rate. The humidity of the gas was controlled by injection of steam into a second furnace
placed under the main furnace. Initial moisture content of the samples was between
7 and 12%wt. (dry) and initial temperature of the samples was 20 C (refer to Table
2.10). The samples were first heated to 120 C and held at this temperature for half
an hour, followed by heating below the preset heating rate (refer to Table 2.10) until
the final temperature (also refer to Table 2.10) was achieved. During heat treatment,
the weights of the samples were recorded. In addition, temperatures were measured by
T-type thermocouples placed inside the samples, but the measurement indicated that the
temperatures inside the samples were essentially uniform because of their small size
(Younsi et al., 2006b).
2.9.1
Mathematical modelling using the L-REA

The original formulation of the L-REA described in Section 2.1 is used here without any
modification. As mentioned before, for modelling using the L-REA, relative activation
1
Experimentally determined data
Fitted curve
0.9
0.8
0.7
Ev /Ev,b
90
0.6
0.5
0.4
0.3
0.2
0.1
0
0.02
0.04
0.06
0.08
0.1
0.12
Figure 2.48 Relative activation energy (Ev /Ev,b ) of the dehydration of wood during heat
treatment generated from the experimental data in Case 2 (refer to Table 2.10). [Reprinted from
Bioresource Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley, Modelling of
high-temperature treatment of wood by using the reaction engineering approach (REA),
energy (Ev /Ev,b ) needs to be generated. In this study, the relative activation energy is
generated from the experimental data for a drying run of Case 2 (refer to Table 2.10).
This can be represented as:

E v
= 1 2.181(X X b )0.372
E v,b

exp 3.716(X X b )3.135
for X X b 0.05,

E v
= 1 1.462(X X b )0.207
E v,b

exp 3.716(X X b )3.137 forX X b 0.05.
(2.9.1)
(2.9.2)
The format of the relative activation energy can be varied but, in this case, Equations
(2.9.1) and (2.9.2) seem to be sufficient. The good agreement between the fitted and
experimental relative activation energy is shown in Figure 2.48 and indicated by R 2 of
0.994 and 0.999, respectively.
During heat treatment of wood, the drying gas temperature changed linearly with
time according to heating rate (refer to Table 2.10). In order to incorporate this in
the modelling using the L-REA, the equilibrium activation energy (Ev,b ) shown by
Equation (2.1.7) has been evaluated according to the gas temperature and corresponding
91
humidity during heat treatment. The equilibrium activation energy (Ev,b ) is combined
with the relative activation energy (Ev /Ev,b ) shown in Equations (2.9.1) and (2.9.2).
Since the temperature distributions inside the samples were essentially uniform during
the heat treatment (Younsi et al., 2006b), the heat balance can be written as:
d(mC p T )
dX
h A (Tb T ) + m s
H V ,
dt
dt
(2.9.3)
where m is sample mass (kg), Cp is the heat capacity of the sample (J kg1 K1 ), T is
the temperature of the sample (K), h is the heat transfer coefficient (W m2 K1 ), HV
is vaporisation heat of water (J kg1 ) and Tb is the gas temperature (K). The drying rate
dX/dt is negative when drying occurs.
In this case, the linearly increased gas temperature is applied in Equation (2.9.3).
Solving the mass balance, and implementing the L-REA and heat balance shown in
Equations (2.1.4) and (2.9.3) in conjunction with the equilibrium and relative activation energy shown in Equations (2.17), (2.9.1) and (2.9.2), results simultaneously in
the profiles of moisture content and temperature during heat treatment of wood. The
shrinkage is neglected in the modelling because Younsi et al. (2006b) indicated that the
ratio between the final and initial dimension is around 0.96. Similarly, the modelling
implemented by Younsi et al. (2006a,b, 2007) did not incorporate the shrinkage effect.
2.9.2
Results of modelling wood heating under linearly increased gas

temperatures using the L-REA
The profiles of both moisture content and temperature during heat treatment of wood for
Cases 15 (refer to Table 2.10) are presented in Figures 2.492.52. Results of modelling
for Case 1 (refer to Table 2.10) are shown in Figures 2.49 and 2.50. The L-REA-based
model system describes both the moisture content and temperature profiles well. The
predictions made using the L-REA match reasonably well with the experimental data
of moisture content and temperature. The slight discrepancies in the moisture content
profile were also found in the modelling implemented by Younsi et al. (2007) using
a far more complex model. However, as depicted in Figure 2.49, the L-REA yields
system closer agreement of the moisture content profile with the experimental data. The
L-REA system can model the heat treatment of wood which applied linearly changed
air temperature with a heating rate of 30 C h1 .
For Case 2 (refer to Table 2.10), Figures 2.49 and 2.50 indicate that the L-REA predicts both the moisture content and temperature profiles well. The results of modelling
match well with the experimental data. This is also confirmed by R 2 of moisture content and temperature profiles of 0.994 and 0.996. A benchmark against the modelling
implemented by Younsi et al. (2007) shows that the L-REA gives comparable or better
results.
For the heat treatment of wood under a heating rate of 10 C h1 (Case 3, refer to
Table 2.10), the L-REA again describes both the moisture content and temperature
reasonably well, as shown in Figures 2.49 and 2.50. Slight deviations in predicting
moisture content profiles are observed, but when benchmarking against the modelling
0.14
Case 1-experimental data
Case 1-predicted by L-REA
0.12
92
0.1
0.08
0.06
0.04
0.02
4
t(s)
7
4
10
Figure 2.49 Moisture content profiles during the heat treatment of Cases 1 to 3 (refer to Table
2.10). [Reprinted from Bioresource Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and P.A.
Webley, Modelling of high-temperature treatment of wood by using the reaction engineering
approach (REA), 62146220, Copyright (2012), with permission from Elsevier.]
applied by Younsi et al. (2007), the L-REA still yields the closer agreement towards
experimental data. The good agreement of moisture content and temperature profile is
also indicated by R 2 of moisture content and temperature profiles of 0.987 and 0.993.
It is also important to assess the accuracy of the L-REA to model the heat treatment
of wood using a different final gas temperature. Case 4 (refer to Table 2.10) applied the
heat treatment of wood samples with a final gas temperature of 160 C and a heating rate
of 20 C h1 . As depicted in Figure 2.51, the L-REA system was reasonably accurate in
modelling the moisture content profiles. Both the L-REA system and the modelling by
Younsi et al. (2006a) predict the temperature profiles well as shown in Figure 2.52. It
suggests that the L-REA system can model the heat treatment of wood using different
final gas temperatures.
In addition, Case 5 (refer to Table 2.10) implemented the heat treatment of wood with
initial moisture content of 7%wt. (dry basis), slightly lower than that of other cases.
The accuracy of the L-REA in predicting both the moisture content and temperature
profiles has been assessed. From Figure 2.51, it can be seen that the L-REA system
describes the profiles of moisture content well. In addition, the L-REA is accurate in
modelling the temperature profiles during the heat treatment of wood as shown in Figure
2.52. Results of modelling using the L-REA match well with experimental data. The
93
500
Temperature (K)
450
400
350

300
250
3
t(s)
7
10
Figure 2.50 Temperature profiles during the heat treatment of Cases 1 to 3 (refer to Table 2.10).
[Reprinted from Bioresource Technology, 102, A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley,
Modelling of high-temperature treatment of wood by using the reaction engineering approach
agreement towards the profiles of moisture content and temperature is indicated by an

accurate R 2 of moisture content and temperature of 0.993 and 0.999, respectively. When
benchmarking against the modelling implemented by Younsi et al. (2006), it is observed
that the L-REA yields better results.
From Cases 1 to 5 (refer to Table 2.10), it can be said that the L-REA system
can be implemented to describe both moisture content and temperature profiles very
successfully. The applicability of the L-REA for this purpose could be due to the
accuracy and flexibility of the equilibrium activation energy (Ev,b ) combined with the
unique relative activation energy (Ev /Ev,b ) in capturing the physics of drying during
heat treatment of wood. The effect of a linearly increased gas temperature according
to the heating rate on the drying rate seems to be captured well by the combination
of (Ev,b ) and (Ev /Ev,b ). This allows the drying kinetics to be changed flexibly
according to environment conditions.
Based on the study of Cases 1 to 5 (refer to Table 2.10), it is revealed that the L-REA
can be implemented in modelling the heat treatment of wood with various heating rates.
Therefore, the L-REA may also be applied to the similar thermal processing of biomass
that employs time-varying temperature or external conditions. Several processes apply
this principle, including ThermoWood Technology (Finnish ThermoWood Association,
2011) developed by Finnish industries. The process is essentially heating wood following
0.14
0.12
94
0.1
0.08
0.06
0.04
0.02
4
t(s)
7
4
10
Figure 2.51 Moisture content profiles during the heat treatment of Cases 13 (refer to Table 2.10).
a particular schedule, with different temperatures in each heating period aimed to reduce
cracking and burning by protecting wood with water vapour generated from the wood
samples (Younsi et al., 2010; Finnish ThermoWood Association, 2011). Another example
is thermal modification of wood to improve thermal insulation properties, maintaining
colour and enhancing water resistance conducted by heating of wood to temperature of
160260 C under various gas media and schedules (Rapp, 2001). In order to predict the
quality changes resulting from the process, extensive experiments can be carried out.
However, these may require a lot of resources and may not be suitable for large-scale
industries where quick decision-making is necessary. Simulation tools can be applied
as a means to predict the quality of wood or biomass after treatment. As the L-REA is
revealed to be simple and accurate for modelling the heat treated wood, it can be used
as a simple and valuable tool in industry for predicting quality parameters of samples
under heat treatment. This can be conducted by the L-REA by evaluating the equilibrium
activation energy and heat balance according to corresponding external conditions (i.e.
humidity and temperature) in each period of heat treatment. The equations that represent
quality change, usually represented as function of moisture content and/or temperature,
are incorporated in the REA system and solved simultaneously to yield quality profiles
during heat treatment.
95
480
460
440
Temperature (K)
420
400
380
360
340
320
300
280
0.5
1.5
2
t(s)
2.5
3.5
104
Figure 2.52 Temperature profiles during the heat treatment of Cases 4 and 5 (refer to Table 2.10).
2.10
The baking of cake modelled using the L-REA

Baking is a complex simultaneous heat and mass transfer process commonly applied
in food industries. Although the process has been practiced for long time, there is still
limited knowledge of the physical phenomena involved (Zhang and Datta, 2006). The
heat and mass transfer process is relatively complicated since it is related to physical,
chemical and structural changes during baking (Lostie et al., 2002a). The process is
also affected significantly by molecular size and structure of polymers that make up
starch and protein components. The interactions between polymer chain entanglements
and branching may determine the rheological aspects that affect the deformation during
baking (Dobraszczyk, 2004). Shrinkage could occur during baking as a result of water
loss and thermal stress, while expansion takes place because of carbon dioxide produced
by leavening agents and water vapour inside the porous medium (Lostie et al., 2002a,b;
Mayor and Sereno, 2004).
The applicability of the L-REA to modelling the baking of a thin layer of cake is
validated by the published experimental data and details of Sakin et al. (2007). For a
better understanding of the modelling using the L-REA, the experimental details are
reviewed next. The ingredients of the cake batter were 49.4%wt. of Dr Oetkers ready
96
dry cake mix (consists of wheat flour, sugar, corn starch and baking powder), 24.7%wt.
pasteurised whole liquid egg, 16.2%wt. vegetable margarine and 9.7%wt. water. The
ingredients were mixed thoroughly using a three-stage mixing method and a hand mixer.
The mixture was spread on a baking tray with diameter of 220 mm to create a batter with
initial thickness of 3 mm. This thickness was used to minimise temperature gradient
inside the samples. The initial moisture content of the samples was 53%wt. (dry basis)
(Sakin et al., 2007). The baking experiments were conducted in an electrical baking oven
with dimensions of 0.39 0.44 0.35 m (Teba High-01 Inox) at baking temperatures of
50, 80, 100, 140 and 160 C under forced convection conditions. Fresh air entered the
oven cavity and the fan on the back side was used to circulate the air at a constant speed
of 0.56 m s1 (measured by an Airflow anemometer, LCA 6000). During baking, the
weight of the batter was recorded until the equilibrium moisture content was achieved.
The product temperature was also measured by a thermocouple (J-type) inserted inside
the samples. In addition, the thickness was measured by a digital calliper (Sakin et al.,
2007).
2.10.1
Mathematical modelling of the baking of cake using the L-REA

The original formulation of the L-REA described in Section 2.1 is used here to model
baking of cake without any modification. Similar to the modelling of convective drying using the L-REA mentioned in the previous sections, for modelling baking a thin
layer of cake using the L-REA, the relative activation energy (Ev /Ev,b ) needs to be
generated from one accurate baking experiment. In this study, it was generated from
an experiment baking at a temperature of 100 C, whose experimental data of moisture
content and temperature (Sakin et al., 2007) were used to evaluate the activation energy
(Ev ) shown in Equation (2.1.5). The relative activation energy (Ev /Ev,b ) is calculated by dividing the activation energy with the equilibrium activation energy (Ev,b )
indicated in Equation (2.1.7). The relative activation energy is related with the moisture
content on dry basis (X) by simple mathematical expression obtained by a least-square
method:

E v
= 1 1.612(X X b )1.151 exp 1.28 106 (X X b )14.19 .
E v,b
(2.10.1)
Figure 2.53 shows a good agreement between the experimental and fitted relative activation energy, which is also confirmed by R 2 of 0.998. The profile of the relative activation
energy is very reasonable since it is zero near the start and keeps increasing as baking
proceeds. When the equilibrium moisture content is achieved, the relative activation
energy is 1. The format of Equation (2.10.1) can be varied but in this case, Equation
(2.10.1) seems to be sufficient to represent the relative activation energy of baking a thin
layer of cake.
For yielding the profiles of both moisture content and temperature during baking, the
mass and heat balances are solved simultaneously. The mass balance using the L-REA
is shown in Equation (2.1.4). The temperature distribution inside the cake during baking
97
1
Data
Model
0.9
0.8
Ev /Ev,b
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.05
0.1
0.15 0.2 0.25 0.3 0.35

0.4
0.45
0.5
Figure 2.53 The relative activation energy (Ev /Ev,b ) of baking of thin layer of cake at an oven
temperature of 100 C. [Reprinted from Journal of Food Engineering, 105, A. Putranto, X.D.
Chen and W. Zhou, Modelling of baking of cake using the reaction engineering approach (REA),
can be neglected as the thickness of the cake was around 3 mm (Sakin et al., 2007).
Therefore, the heat balance can be expressed as:

d m s (1 + X )C p T
dX
(2.10.2)
h A (Tb T ) + m s
HV ,
dt
dt
where Cp is the heat capacity of the sample (J kg1 K1 ), T is the temperature of the
sample (K), h is the heat transfer coefficient (W m2 K1 ), HV is vaporisation heat
of water (J kg1 ) and Tb is the baking oven temperature (K). The drying rate dX/dt is
negative when baking occurs.
2.10.2
Results of modelling of the baking of cake using the L-REA

Profiles of moisture content and temperature during baking cake at different baking
temperatures are shown in Figures 2.542.57. Figures 2.54 and 2.55 show that the profiles
of moisture content predicted using the L-REA match well with the experimental data.
This is supported by the R 2 values for moisture content profiles being higher than 0.982
(shown in Table 2.11). Benchmarking against a diffusion-based model implemented by
Sakin et al. (2007) revealed that the L-REA yielded comparable or better results, although
the diffusion-based model employed diffusivity which was split into two forms in order to
incorporate the effects of temperature and moisture content. The slight overestimation of
0.7
L-REA 100C
Data 100C
L-REA 140C
Data 140C
L-REA 160C
Data 160C
0.6
0.5
0.4
0.3
0.2
0.1
0
500
1000 1500 2000 2500 3000 3500 4000 4500 5000

t(s)
Figure 2.54 Moisture content profiles at baking temperatures of 100, 140 and 160 C.
[Reprinted from Journal of Food Engineering, 105, A. Putranto, X.D. Chen and W. Zhou,
Modelling of baking of cake using the reaction engineering approach (REA), 306311,
0.5
L-REA 50C
Data 50C
L-REA 80C
Data 80C
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
4
t(s)
7
10
Figure 2.55 Moisture content profiles at baking temperatures of 50 and 80 C. [Reprinted from
Journal of Food Engineering, 105, A. Putranto, X.D. Chen and W. Zhou, Modelling of baking of
cake using the reaction engineering approach (REA), 306311, Copyright (2012), with
99
Table 2.11 R 2 of modelling using the REA.
No.
Baking
temperature (C)
R 2 for X
R 2 for T
1.
2.
3.
4.
5.
50
80
100
120
140
0.990
0.997
0.992
0.987
0.982
0.990
0.999
0.970
0.997
0.985
460
440
Temperature (K)
420
400
380
360
340
L-REA 100C
Data 100C
L-REA 140C
Data 140C
L-REA 160C
Data 160C
320
300
280
500
1000 1500 2000 2500 3000 3500 4000 4500 5000

t(s)
Figure 2.56 Temperature profiles at baking temperatures of 100, 140 and 160 C. [Reprinted
from Journal of Food Engineering, 105, A. Putranto, X.D. Chen and W. Zhou, Modelling of
baking of cake using the reaction engineering approach (REA), 306311, Copyright (2012), with
the drying rate at initial baking period at oven temperatures of 100, 140 and 160 C was
also observed in modelling using the diffusion-based model. At baking temperatures of
50 and 80 C both the L-REA and diffusion-based model described the moisture content
profiles well. It can be inferred that the L-REA can model the moisture content during
the baking of thin layer of cake very well.
Figures 2.56 and 2.57 indicate that the L-REA modelled the profiles of temperature profiles well. A good agreement with the experimental data is observed and it is
confirmed by the R 2 values for temperature profiles being higher than 0.970 (shown
in Table 2.11). The slight discrepancies between the predicted and experimental data
during initial period of baking at baking temperature of 100 C might be due largely to
100
370
360
Temperature (K)
350
340
330
320
310
L-REA 50C
Data 50C
L-REA 80C
Data 80C
300
290
4
t(s)
7
4
10
Figure 2.57 Temperature profiles at baking temperatures of 50 and 80 C. [Reprinted from

Journal of Food Engineering, 105, A. Putranto, X.D. Chen and W. Zhou, Modelling of baking of
cake using the reaction engineering approach (REA), 306311, Copyright (2012), with
experimental error of temperature measurement, which resulted in higher temperature

profiles than those of baking temperatures at 140 and 160 C. Benchmarking against
the modelling from Sakin et al. (2007) cannot be conducted as the model in this study
did not implement the heat balance. The L-REA coupled with the heat balance (shown
in Equation 2.10.2) is indeed accurate in predicting the temperature profiles during
baking.
It has been shown that the L-REA-based heat- and mass-transfer system can be
successfully applied to model the baking of cake. The L-REA is accurate in modelling
both the profiles of moisture content and temperature during the baking of cake. It seems
that the relative activation energy (Ev /Ev,b ) shown in Equation (2.9.1) combined
with the equilibrium activation energy (Ev,b ) shown in Equation (2.1.7) can capture
the physics during baking.
2.11
The infrared-heat drying of a mixture of polymer solutions modelled

using the L-REA
The experimental data for validating the accuracy of the L-REA are derived from a study
reported by Allanic et al. (2009). The experimental setup up is similar to that described
in Section 2.5 but constant infrared-heat intensity is applied here. The velocity and
101
0
Product
e(t)
Petri dish
Thermocouple
Conduction
Long infrared
Evaporation
Convection
Diffusion
Figure 2.58 Heat transfer mechanisms of convective and infrared-heat drying. [Reprinted from
Chemical Engineering and Processing: Process Intensification, 49, A. Putranto, X.D. Chen and
P.A. Webley, Infrared and convective drying of thin layer of polyvinyl alcohol
temperature of the drying medium were set to 1 m s1 and 35 C, respectively was fed
into the canal. Constant infrared-heating with an intensity of 3.7 kW m2 was maintained
throughout the experimental run (Allanic et al., 2009).
2.11.1
Mathematical modelling of the infrared-heat drying of a mixture of polymer

solutions using the L-REA
The L-REA shown in Equation (2.1.4) is used for the mass balance here. For infrared heatdrying, the sample was heated by an infrared emitter which increased the temperature
of the sample and the Petri dish (refer to Figure 2.58).
Because of the relatively low temperature of the drying air, the sample and the Petri
dish actually release heat to the environment by convection. The heat is released from
the upper side and bottom side due to forced and natural convection, respectively (refer
to Figure 2.58). Hence, it can be written as:
d(C p eT )
ms d X
= Q IR h top (T Tb ) Ubottom (T Tb ) +
HV (T ),
dt
A dt
(2.11.1)
where is sample density (kg m3 ), Cp is sample heat capacity (J kg1 K1 ), e is sample
thickness (m), T is sample temperature (K), ms is mass of dried product (kg), A is surface
area of product (m2 ), Hv is vaporisation enthalpy of water (J kg1 ), htop is the heat
102
transfer coefficients on top of the sample (W m2 K1 ), Ubottom represents the overall

heat transfer coefficient (W m2 K1 ) from the lower side, including convection along
and conduction through the Petri dish, QIR is the intensity of radiation (W m2 ) and is
the absorptivity of the product.
Rearranging Equation (2.11.1) results in:
ms d X
d(C p eT )
= Q IR Utotal (T Tb ) +
HV (T ).
dt
A dt
(2.11.2)
It should be highlighted that the application of the L-REA requires accurate determination of activation energy as a function of moisture content. Ev /Ev ,b because
characteristics of drying kinetics are used to describe the reduction of moisture content
and temperature profiles. In convective drying, the product is heated by relatively high
drying air temperatures and the maximum activation energy Ev ,b is determined using
Equation (2.1.7). However, a different condition occurs when drying is conducted using
infrared heating because the product temperature increases to above the drying air temperature so the product releases heat to the air instead of gaining heat from the air. If
one maintains constant infrared power and lets drying continue until low water content
is reached, the product temperature would reach a constant temperature determined by
the balance between infrared power input and heat loss to the surroundings. The minor
modifications of the L-REA are taken here as explained next.
The heat transfer coefficient should be determined from the final part of drying instead
of using the initial constant rate period of drying because of the low evaporation rate at the
final part of drying, so most of the heat adsorbed by the product from the infrared emitter
is released to the air. In addition, at the final part of drying the product temperature is
essentially constant as revealed by Allanic et al. (2009) indicating thermal equilibrium
has been reached. The heat balance for final part of drying can be written as:
Q IR =
ms d X
HV (T ) + Utotal (T Tb ).
A dt
(2.11.3)
It is apparent that, for the final part of drying, the contribution of the first term on
the right-hand side is low because of the low evaporation rate, thus the second term is
dominant. Equation (2.11.3) can be simplified:
Q IR Utotal (T Tb ),
(2.11.4)
and Utotal can be determined using Equation (2.11.3) by inserting T from the recorded
final product temperature (Allanic et al., 2009). The predetermined Utotal is then used for
modelling of moisture content and temperature profile. It is emphasised that Equation
(2.11.4) only holds at this point.
In addition, a new definition of maximum activation energy (Ev ,b ) is introduced
because the product is not heated only by air so the definition of Ev ,b shown by Equation
(2.1.7) is not appropriate. The relative activation energy generated from the convective
drying run and shown in Equation (2.5.10) is used but Ev ,b has been determined from
the final product temperature and corresponding humidity of air instead of using drying
air temperature. This can be written:
E v,b = RT ln(R Hb ),
(2.11.5)
103
8
Model
Data
6
5
4
3
2
1
0
500
1000
1500
t(s)
2000
2500
3000
Figure 2.59 Moisture content profile of convective and infrared drying at an air temperature of
35 C, air velocity of 1 m s1 , air relative humidity of 18% and intensity of infrared drying of
3700 W m2 . [Reprinted from Chemical Engineering and Processing: Process Intensification,
49, A. Putranto, X.D. Chen and P.A. Webley, Infrared and convective drying of thin layer of
polyvinyl alcohol (PVA)/glycerol/water mixture The reaction engineering approach (REA),
where T is the final product temperature (K) and RHb is the relative humidity at the final
product temperature and the absolute humidity. For other drying conditions, as T is not
known prior to drying experiments, T can be determined from heat balance between the
infrared power input and heat loss to the surroundings as explained previously.
In order to yield the moisture content and temperature profiles, the mass and heat
balances shown in Equations (2.1.4) and (2.11.1), respectively are solved simultaneously.
The equations are combined with the relative and equilibrium activation energy indicated
in Equations (2.5.10) and (2.11.5), respectively. The results of modelling using the LREA are validated towards the experimental data of Allanic et al. (2009).
2.11.2
The results of mathematical modelling of infrared-heat drying of a

mixture of polymer solutions using the L-REA
The results of modelling are presented in Figures 2.59 and 2.60. It could be observed
that the discrepancies between experimental and calculated data are reasonably small.
Statistical analysis showed that the R 2 and RMSE of the moisture content profile are
0.994 and 0.181, respectively. Overestimation of the drying rate between drying times
of 6002250 s was also predicted by the previous model (Allanic et al., 2009). The
L-REA seems to model this case better and only shows slight overestimation in drying
104
360
350
Temperature (K)
340
330
320
310
300
290
Model
Data
0
500
1000
1500
t(s)
2000
2500
3000
Figure 2.60 Product temperature profile of convective and infrared drying at an air temperature of
35 C, air velocity of 1 m s1 , air relative humidity of 18% and intensity of infrared drying of
3700 W m2 . [Reprinted from Chemical Engineering and Processing: Process Intensification,
49, A. Putranto, X.D. Chen and P.A. Webley, Infrared and convective drying of thin layer of
rates between drying times of 13502250 s. In addition, it is apparent that the REA can
handle temperature profiles quite well as shown by the R 2 and RMSE of temperature
profile, which are 0.992 and 1.712, respectively. Overestimation in the temperature
profile of about 5 C during drying times of 1501200 s was indicated by the previous
model (Allanic et al., 2009). However, this overestimation is not observed by modelling
using the L-REA.
It can be observed that Ev /Ev ,b derived from convective drying combined with the
new quantification of Ev ,b shown in Equation (2.11.5) is appropriate for describing
the drying kinetics of infrared-heat drying. It may be applied to other infrared-heating
cases. In the case of drying, which exhibits product temperature higher than the drying
air temperature, application of the modification of Ev ,b shown by Equation (2.11.5) in
conjunction with the generated Ev /Ev ,b is shown to be appropriate.
2.12
The intermittent drying of a mixture of polymer solutions under

time-varying infrared-heat intensity modelled using the L-REA
The experimental data for validating the accuracy of the L-REA are derived from a
study reported by Allanic et al. (2006, 2009), whose experimental conditions are briefly
105
Table 2.12 The experimental conditions of intermittent drying of a mixture of polymer solutions.
Case number
Air velocity
(m s1 )
Air temperature
(C)
Air relative
humidity (%)
Intensity of infrared
radiation (W m2 )
1 (Allanic et al., 2009)

1
1
1
35
35
35
19
14
18
12.35.6-regulated*
13.35.6-regulated*
13.312.3-regulated*
reviewed in Section 2.11. The conditions of experiments are listed in Table 2.12, which
indicates that different intensities of the infrared heater power were applied in each stage
to induce a different condition for drying in the corresponding stage.
2.12.1
Mathematical modelling of the intermittent drying of a mixture of polymer

solutions under time-varying infrared-heat intensity using the L-REA
For modelling of this kind of cyclic drying, a new idea has to be introduced to define
the equilibrium activation energy (Ev,b ). This is necessary because each stage has
different drying conditions. If each drying condition persists for a long time, a different
terminal drying state should be reached, corresponding to a unique Ev,b . The equilibrium activation energy (Ev,b ) is defined as representing the maximum activation energy
under conditions in each stage of drying.
For this purpose, two new definitions of equilibrium activation energy are introduced.
The first employs a relationship between the infrared intensity in each stage and T* :
T = m I n + c,
(2.12.1)
where T* is the final product temperature in each stage should the infrared heating
be prolonged to equilibrium (K), I is the infrared intensity employed in each stage
(kW m2 ), m and c are the empirical constants obtained from the linear relationship, T*
and In , and n is a constant indicating sensitivity of T* towards the infrared intensity.
Using this expression, Ev,b in each stage is determined from T* and relative humidity
of air at corresponding T* and corresponding humidity. This can be written as:
E v,b = RT ln(R Hb ),
(2.12.2)
where RHb is the relative humidity of air at T* and the corresponding humidity.
Alternatively, a second scheme could be used to relate directly the equilibrium activation energy (Ev,b ) and the infrared intensity. The relationship of the infrared intensity
with the equilibrium activation energy (Ev,b ) in each stage expressed as:
E v,b = p I q + k,
(2.12.3)
where p and k are the constants obtained from linear relationship between the equilibrium
activation energy (Ev,b ) and Iq and q is the constant indicating sensitivity of the
equilibrium activation energy (Ev,b ) towards infrared intensity.
For modelling the cyclic drying here, both definitions of equilibrium activation
energy (Ev,b ) need to be combined with the relative activation energy (Ev /Ev,b )
shown in Equation (2.5.10). The relative activation energy shown in Equation (2.5.10),
106
8
Data
n = 1.4
n = 1.5
n = 1.6
n = 1.7
n = 1.8
n = 1.9
6
5
4
3
2
1
0
200
400
600
t(s)
800
1000
1200
Figure 2.61 Sensitivity of the moisture content profile of cyclic drying; Case 1 (refer to Table
2.12) towards n (on Equation 2.12.1). [Reprinted from Chemical Engineering Science, 65, A.
Putranto, X.D. Chen and P.A. Webley, Application of the reaction engineering approach (REA)
to model cyclic drying of thin layers of polyvinyl alcohol (PVA)/glycerol/water mixture,
generated from convective drying of a mixture of polymer solution, can still be used
for the modelling of cyclic drying here. The L-REA shown in Equation (2.1.4) serves
as the mass balance while the heat balance shown in Equation (2.11.1) is used by
employing the different heating intensity in each drying period. Solving Equations
(2.1.4), (2.5.10) and (2.11.1) and the appropriate equilibrium energy function described
previously results in the moisture content and temperature profiles.
2.12.2
Results of modelling the intermittent drying of a mixture of polymer solutions

under time-varying infrared heat intensity using the L-REA
For Case 1 of the cyclic drying process (refer to Table 2.12), both schemes were implemented. Using the first scheme (Equation 2.12.1) (T* as a function of the infrared
intensity), several values of n shown in Equation (2.12.1) in the range of 1.41.9 were
used to describe the drying kinetics (which we found to be more favourable than either
lower or higher values). Figures 2.61 and 2.62 show the profiles of moisture content and
temperature along the cyclic drying run with various values of n. It can be observed that
different values of n did not provide a noticeable difference in both moisture content
and temperature profiles during the constant rate period of drying. This indicates an
overriding effect of the latent heat of vaporisation. However, different values of n that
107
380
370
360
Temperature (K)
350
340
330
Data
n = 1.4
n = 1.5
n = 1.6
n = 1.7
n = 1.8
n = 1.9
320
310
300
290
200
400
600
t(s)
800
1000
1200
Figure 2.62 Sensitivity of the temperature profile of cyclic drying; Case 1 (refer to Table 2.12)
towards n (on Equation 2.12.1). [Reprinted from Chemical Engineering Science, 65, A. Putranto,
X.D. Chen and P.A. Webley, Application of the reaction engineering approach (REA) to model
cyclic drying of thin layers of polyvinyl alcohol (PVA)/glycerol/water mixture, 51935203,
gave deviations in both profiles were observed after 300 s of the drying process. The
lower values of n yielded lower moisture content and lower temperature profiles because
these lower values result in lower activation energy which projects a higher evaporation
rate and more heat is removed from materials being dried for evaporation, shifting the
temperature profile down somewhat. From Figures 2.61 and 2.62, modelling by the
value of n of 1.8 tends to agree well with the moisture content profile. Figure 2.63 shows
the prediction of moisture content profile whilst Figure 2.64 shows the temperature profile using the L-REA and the first scheme (Equation 2.12.1) in conjunction with n = 1.8.
In addition, the good fit of this model approach is shown by R 2 and RMSE of 0.996 and
0.13 for moisture content, respectively while R 2 and RMSE of temperature profile are
0.938 and 3.3, respectively. Compared with results of modelling of Allanic et al. (2009)
results of the current model seem to describe moisture content profile better (the other
model resulted in an underestimation of evaporation rate initially and the overestimation
after a drying time of 600 s).
Figures 2.65 to 2.68 indicate the results of modelling of Case 1 (refer to Table 2.12)
by the second scheme shown in Equation 2.12.3 (Ev,b as function of the intensity
of the infrared heating). Similarly, several values of q (Equation 2.12.3) were used to
8
Model
Data
7
6
108
5
4
3
2
1
0
200
400
600
t(s)
800
1000
1200
Figure 2.63 Moisture content profile of cyclic drying; Case 1 (refer to Table 2.12) using the first
scheme (T* as function of infrared intensity) with n = 1.8. [Reprinted from Chemical
Engineering Science, 65, A. Putranto, X.D. Chen and P.A. Webley, Application of the reaction
engineering approach (REA) to model cyclic drying of thin layers of polyvinyl alcohol
(PVA)/glycerol/water mixture, 51935203, Copyright (2012), with permission from Elsevier.]
describe both moisture content and temperature profiles. Figures 2.65 and 2.66 show
that no deviations were observed during constant rate period of drying but different
values of q gave different moisture content and temperature profile after drying time of
300 s. It can be examined that the value of q of 1.8 resulted in the best fit of moisture
content profile. This can be clearly observed in Figure 2.67. A reasonable agreement of
temperature prediction is also indicated by Figure 2.68. Higher value of q resulted in
higher moisture content and temperature profile because of the higher activation energy
which decreases evaporation rate and less heat is removed from materials so this can
yield a higher temperature profile. The good fit using this second scheme by application
of q of 1.8 was indicated by R 2 and RMSE of 0.995 and 0.14 for moisture content while
R2 and RMSE of temperature are 0.937 and 3.3, respectively. Moreover, modelling using
this second scheme yielded better prediction of moisture content profile than that from
Allanic et al. (2009) similar to that described previously for the first scheme.
Comparison of modelling using the first and the second scheme for Case 1 in the
current study has revealed that both the proposed schemes gave almost the same profiles
of moisture content and temperature to drying times of 300 s. After that time, although
the values are not exactly the same, the differences between predicted moisture content
using the first and the second schemes are below 0.15 kg kg1 while those of temperature
109
380
370
360
Temperature (K)
350
340
330
320
310
Model
Data
300
290
200
400
600
t(s)
800
1000
1200
Figure 2.64 Temperature profile of cyclic drying; Case 1 (refer to Table 2.12) using the first
are below 1.5 C. These indicate both schemes may have been successful in capturing
the physics of the process.
The results of modelling of cyclic drying of Case 2 (refer to Table 2.12) using the
L-REA and the two schemes mentioned previously are represented in Figures 2.69 to
2.72. Similarly to Case 1, sensitivity of n (Equation 2.12.1) and q (Equation 2.12.3) was
conducted. It was found that n = 1.5 and q = 1.5 were the most appropriate to describe
the moisture content and temperature profile. A good agreement of both moisture content
and temperature profile using the first scheme and n = 1.5 can be observed from Figures
2.69 and 2.70, respectively. Similarly to the previous case, the lower value of n produced
lower moisture content and temperature profiles. This good fit is shown by R 2 and RMSE
of moisture content of 0.992 and 0.2 while R 2 and RMSE of temperature are 0.946 and
4.4, respectively. When compared to the modelling published by Allanic et al. (2009),
this model again seems to represent moisture content better.
Results of the modelling in Case 2 (refer to Table 2.12) by implementing the second
scheme (Equation 2.12.3) (Ev,b as function of intensity of infrared intensity) and q =
1.5 are shown in Figures 2.71 and 2.72. The good agreement is shown in Figures 2.71 and
2.72 for moisture content and temperature profile, respectively. In addition, this good fit
was indicated by R 2 and RMSE of moisture content of 0.992 and 0.2 while R 2 and RMSE
8
Data
q = 1.6
q = 1.7
q = 1.8
q = 1.9
q=2
6
5
4
3
2
1
0
200
400
600
t(s)
800
100
1200
Figure 2.65 Sensitivity of the moisture content profile of cyclic drying; Case 1 (refer to Table
2.12) towards q (on Equation 2.12.3). [Reprinted from Chemical Engineering Science, 65, A.
Putranto, X.D. Chen and P.A. Webley, Application of the reaction engineering approach (REA)
to model cyclic drying of thin layers of polyvinyl alcohol (PVA)/glycerol/water mixture,
380
370
360
350
Temperature (K)
110
340
330
Data
q = 1.6
q = 1.7
q = 1.8
q = 1.9
q=2
320
310
300
290
200
400
600
t(s)
800
100
1200
Figure 2.66 Sensitivity of the temperature profile of cyclic drying; Case 1 (refer to Table 2.12)
towards q (on Equation 2.12.3). [Reprinted from Chemical Engineering Science, 65, A. Putranto,
X.D. Chen and P.A. Webley, Application of the reaction engineering approach (REA) to model
cyclic drying of thin layers of polyvinyl alcohol (PVA)/glycerol/water mixture, 51935203,
111
8
Model
Data
6
5
4
3
2
1
0
200
400
600
t(s)
800
1000
1200
Figure 2.67 Moisture content profile of cyclic drying; Case 1 (refer to Table 2.12) using the
second scheme (Ev ,b as function of infrared intensity) with q = 1.8. [Reprinted from Chemical
380
370
Temperature (K)
360
350
340
330
320
310
Model
300
290
Data
0
200
400
600
t(s)
800
1000
1200
Figure 2.68 Temperature profile of cyclic drying; Case 1 (refer to Table 2.12) using the second
scheme (Ev,b as function of infrared intensity) with q = 1.8. [Reprinted from Chemical
8
Model
Data
6
5
4
3
2
1
0
100
200
300
400
t(s)
500
600
700
800
380
370
360
Temperature (K)
112
350
340
330
320
310
Model
300
290
Data
0
100
200
300
400
t(s)
500
600
700
800
113
8
Model
Data
6
5
4
3
2
1
0
100
200
300
400
t(s)
500
600
700
800
second scheme (Ev ,b as function of infrared intensity) with q = 1.5. [Reprinted from Chemical
380
370
360
Temperature (K)
350
340
330
320
310
Model
300
290
Data
0
100
200
300
400
500
600
700
800
t(s)
Figure 2.72 Temperature profile of cyclic drying; Case 2 (refer to Table 2.12) using the second
scheme (Ev ,b as function of infrared intensity) with q = 1.5. [Reprinted from Chemical
8
Model
Data
114
5
4
3
2
1
0
100
200
300
t(s)
400
500
600
of temperature are 0.933 and 4.9, respectively. This scheme also yields better moisture
content prediction than that of Allanic et al. (2009), which shows overestimation of the
drying rate after 200 s.
Results of the modelling in Case 2 using both schemes were compared and showed
the values of moisture content and temperature were almost the same for drying times of
300 s. Both schemes generate the same profiles of moisture content and temperature during the constant rate period of drying. In addition, after that time, negligible
differences between predicted moisture content and temperature using both schemes
can be found. The differences between moisture content and temperature were around
0.03 kg kg1 and 1 C, respectively.
In addition, modelling of cyclic drying in Case 3 (refer to Table 2.12) was conducted
using both schemes and the results are shown in Figures 2.732.76. Sensitivity was also
conducted to find the most appropriate value of n (Equation 2.12.1) and q (Equation
2.12.3). For the first scheme, application of n of 1.6 matches the temperature profile very
well. Figures 2.73 and 2.74 illustrate both moisture content and temperature profile using
the value of n = 1.6. The good fit is also shown by R 2 and RMSE of moisture content
of 0.989 and 0.24, while R 2 and RMSE of temperature are 0.964 and 3.55, respectively.
The second scheme was also implemented to describe cyclic drying of Case 3 and the
results are shown in Figures 2.75 and 2.76. The good agreement of this scheme using q
of 1.6 was indicated in Figures 2.75 and 2.76. This is also supported by R 2 and RMSE of
115
380
370
360
Temperature (K)
350
340
330
320
310
Model
300
290
Data
0
100
200
300
t(s)
400
500
600
8
Model
Data
6
5
4
3
2
1
0
100
200
300
t(s)
400
500
600
second scheme (Ev,b as function of infrared intensity) with q = 1.6. [Reprinted from Chemical
116
380
370
360
Temperature (K)
350
340
330
320
310
Model
300
290
Data
0
100
200
300
t(s)
400
500
600
Figure 2.76 Temperature profile of cyclic drying; Case 3 (refer to Table 2.12), using the second
scheme (Ev,b as function of infrared intensity) with q = 1.6. [Reprinted from Chemical
moisture content of 0.991 and 0.23 as well as R 2 and RMSE of temperature of 0.942 and
4.52. The differences of moisture content and temperature profile predictions between
the two schemes are around 0.03 kg kg1 and 1 C.
Therefore, the two new formulations of Ev,b combined with Ev /Ev,b shown
in Equation (2.5.10) can predict the moisture content and temperature profiles well.
Comparison in modelling of the different cases of cyclic drying conditions under use
of the REA framework has shown that application of the first and second schemes with
the appropriate values of n or q give almost exactly the same moisture content and
temperature profiles. All this work supports the notion that L-REA is a robust modelling
framework that allows for intuitive extensions, such as the schemes proposed here, for
more complex situations. The modelling itself remains fairly simple and effective.
2.13
Summary
In this chapter, the applications of the L-REA as an accurate approach for modelling
several drying cases are described. Within the range that material properties are invariant,
the REA parameters, expressed in relative activation energy, can be applied to model
other drying runs of the same materials, provided there is similar initial moisture content.
Besides milk-droplet drying, the L-REA has been innovatively applied and shown to
accurately model the average moisture content and temperature during convective drying
117
of food and non-food materials, convective drying of several centimetre-thick materials, intermittent drying, heating of wood under linearly increased gas temperatures and
baking of cake. For modelling the discussed cases, the original formulation of the L-REA
can be implemented without major modification. The estimation of temperature distribution inside the samples needs to be combined with the L-REA formulation to describe
drying of relatively thick materials. The L-REA can be used to model intermittent drying
by evaluating the equilibrium activation energy according to the corresponding humidity and temperature in each period of drying. Similarly, by evaluating the equilibrium
activation energy according to the drying settings, the L-REA can accurately model the
heat treatment of wood under a constant heating rate. For baking, the original formulation of the L-REA can be implemented without any modifications. The accuracy of the
L-REA in modelling several cases of drying could be due to the accuracy of the relative
activation energy in capturing the physics of drying. A combination of the relative activation energy and equilibrium activation energy yields unique relationships of activation
energy which can change flexibly according to the external drying conditions and represents the change in internal behaviour of the samples during drying. While the L-REA
formulation is simple and the REA is efficient, the results are accurate. It can be used
in industrial settings for process design and maintaining product quality during drying.
References
Allanic, N., Salagnac, P. and Glouannec, P., 2006. Convective and radiant drying of a polymer
aqueous solution. Heat Mass Transfer 43, 10871095.
Allanic, N., Salagnac, P., Glouannec, P. and Guerrier, B., 2009. Estimation of an effective water
diffusion coefficient during infrared-convective drying of a polymer solution. AIChE Journal
55, 23452355.
Azzouz, S., Guizani, A., Jomma, W. and Belghith, A., 2002. Moisture diffusivity and drying
kinetic equation of convective drying of grapes. Journal of Food Engineering 55, 323330.
Baini, R. and Langrish, T.A.G., 2007. Choosing an appropriate drying model for intermittent and
continuous drying of bananas. Journal of Food Engineering 79, 330343.
Chen, X.D., 2005a. Critical Biot number for uniform temperature assumption in transient heat and
mass transfer calculations. International Journal of Food Engineering 1(6), 18.
Chen, X.D., 2005b. Lewis number in the context of drying of hygroscopic materials. Separation
and Purification Technology 48, 121132.
Chen, X.D., 2007. Simultaneous heat and mass transfer, In Handbook of Food and Bioprocess
Modelling Techniques, S. Sablani, S. Rahman, A. Datta, and A.S. Mujumdar (eds.). CRC Press,
Boca Raton, pp. 179233.
Chen, X.D., 2008. The basics of a reaction engineering approach to modelling air drying of small
droplets or thin layer materials. Drying Technology 26, 627639.
Chen, X.D. and Chen, N.X., 1997. Preliminary introduction to a unified approach to modelling
drying and equilibrium isotherms of moist porous solids, Proceeding of Chemeca97, Rotorua,
New Zealand, Paper DR3b (on CD-ROM).
Chen, X.D. and Lin, S.X.Q., 2005. Air drying of milk droplet under constant and time dependent
118
Chen, X.D. and Xie, G.Z., 1997. Fingerprints of the drying behavior of particulate or thin layer
food materials established using a reaction engineering model. Trans IChemE, Part C: Food
and Bioproducts Processing 75, 213222.
Dobraszczyk, B.J., 2004. The physics of baking: rheological and polymer molecular structure
function relationships in breadmaking. Journal of Non-Newtonian Fluid Mechanics 124, 6169.
Farid, M.M., 2003. A new approach to modelling of single droplet drying. Chemical Engineering
Science 58, 29852993.
Finnish Thermowood Association, 2011. www.thermowood.fi, 2011 (accessed 21 November,
2012).
Fu, N., 2012. Single Droplet Drying and Bioactive Particle Engineering. Ph.D. thesis, Monash
University, Australia.
Fu, N., Woo, M.W., Lin, S.X.Q., Zhou, Z. and Chen, X.D., 2011. Reaction Engineering Approach
(REA) to model the drying kinetics of droplets with different initial sizes experiments and
analyses. Chemical Engineering Science 66, 17381747.
Helsen, L. and Bulck, E.V.D., 2005. Review of disposal technologies for chromated copper arsenate
(CCA) treated wood waste, with detailed analyses of thermochemical conversion processes.
Environmental Pollution 134, 301314.
Jin, Y. and Chen, X.D., 2009a. Numerical study of the drying process of different sized particles
in an industrial-scale spray dryer. Drying Technology 27, 371381.
Jin, Y. and Chen, X.D., 2009b. A three-dimensional numerical study of the gas-particle interactions
in an industrial-scale spray dryer for milk powder production. Drying Technology 27, 1018
1027.
Jin, Y. and Chen, X.D., 2010. A numerical model of milk particle deposition in spray dryers.
Drying Technology 28, 960971.
Jin, Y. and Chen, X.D., 2011. Entropy production during the drying process of milk droplets in an
industrial spray dryer. International Journal of Thermal Sciences 50, 615625.
Kar, S., 2008. Drying of Porcine Skin-Theoretical Investigations and Experiments, Ph.D. thesis.
Monash University, Australia.
Kar, S. and Chen, X.D., 2009. The impact of various drying kinetics models on the prediction
of sample temperature-time and moisture content-time profiles during moisture removal from
stratum corneum. Chemical Engineering Research and Design 87, 739755.
Keey, R.B., 1992. Drying of Loose and Particulate Materials. Hemisphere Publishing, New York.
Kocaefe, D., Charette, A., Ferland, J., Couderc, P. and Saint-Romain, J.L., 1990. A kinetic study
of pyrolysis in pitch impregnated electrodes. The Canadian Journal of Chemical Engineering
68, 988996.
Kocaefe, D., Younsi, R., Poncsak, S. and Kocaefe, Y., 2007. Comparison of different models for
the high-temperature heat-treatment of wood. International Journal of Thermal Sciences 46,
707716.
Kowalski, S.J., Musielak, G. and Banaszak, J., 2007. Experimental validation of the heat and mass
transfer model for convective drying. Drying Technology 25. 107121.
Kowalski, S.J. and Pawlowski, A., 2010a. Drying of wet materials in intermittent conditions.
Kowalski, S.J. and Pawlowski, A., 2010b. Modelling of kinetics in stationary and intermittent
119
Lin, S.X.Q., 2004. Drying of Single Milk Droplets. Ph.D. thesis, Department of Chemical and
Materials Engineering, The University of Auckland, New Zealand.
Lin, S.X.Q. and Chen, X.D., 2002. Improving the glass-filament method for accurate measurement
of drying kinetics of liquid droplets. Trans IChemE Part A: Chemical Engineering Research
and Design 80, 401440.
Lin, S.X.Q. and Chen, X.D., 2005. Prediction of air drying of milk droplets under relatively high
humidity using the reaction engineering approach, Drying Technology 23, 13961406.
Lin, S.X.Q. and Chen, X.D., 2007. The reaction engineering approach to modelling the cream
and whey protein concentrate droplet drying. Chemical Engineering and Processing 46, 437
443.
Lostie, M., Peczalski, R., Andrieu, J. and Laurent, M., 2002a. Study of sponge cake batter baking
process. I: Experimental data. Journal of Food Engineering 51, 131137.
Lostie, M., Peczalski, R., Andrieu, J. and Laurent, M., 2002b. Study of sponge cake batter
baking process. II. Modelling and parameter estimation. Journal of Food Engineering 55, 349
357.
Mariani, V.C., de Lima, A.G.B. and Coelho, L.S., 2008. Apparent thermal diffusivity estimation
of the banana during drying using inverse method. Journal of Food Engineering 85, 569579.
Mayor, L. and Sereno, A. M., 2004. Modelling shrinkage during convective drying of food
materials: A review. Journal of Food Engineering 61, 373386.
McCabe, W.L., Smith, J.C. and Harriott, P., 2001. Unit Operations of Chemical Engineering, 6th
ed. McGraw Hill, Boston.
Microsoft Corp., 2012. http://office.microsoft.com/en-au/excel/ (accessed 21 November, 2012).
Pakowski, Z. and Adamski, A., 2007. The comparison of two models of convective drying of
shrinking materials using apple tissue as an example. Drying Technology 25, 11391147.
Pang, S., 1994. High-Temperature Drying of Pinus Radiata Boards in a Batch Kiln. Ph.D. thesis,
University of Canterbury, New Zealand.
Patel, K.C. and Chen, X.D., 2008. Surface-center temperature differences within milk droplets
during convective drying and drying-based Biot number analysis. AIChE Journal 54, 3273
3290.
Patel, K., Chen, X.D., Jeantet, R. and Schuck, P., 2010. One-dimensional simulation of co-current,
dairy spray drying systems pros and cons. Dairy Science and Technology 90, 181210.
Putranto, A., Chen, X.D. and Webley, P.A., 2010a. Infrared and Convective Drying of thin layer of
polyvinyl alcohol (PVA)/glycerol/water mixture The reaction engineering approach (REA).
Chemical Engineering and Processing: Process Intensification 49, 348357.
Putranto, A., Chen, X.D. and Webley, P.A., 2010b. Application of the reaction engineering approach
(REA) to model cyclic drying of polyvinyl alcohol (PVA)/glycerol/water mixture. Chemical
Putranto, A., Chen, X.D. and Webley, P.A., 2011a. Modelling of drying of thick samples of
mango and apple tissues using the reaction engineering approach (REA). Drying Technology
29, 961973.
Putranto, A, Xiao, Z., Chen, X.D. and Webley, P.A., 2011b. Intermittent drying of mango tissues:
implementation of the reaction engineering approach (REA). Industrial Engineering Chemistry
Research 50, 10891098.
Putranto, A., Chen, X.D., Devahastin, S., Xiao, Z. and Webley, P.A., 2011c. Application of the
reaction engineering approach (REA) to model intermittent drying under time-varying humidity
120
Putranto, A., Chen, X.D., Xiao, Z. and Webley, P.A., 2011d. Modelling of high-temperature
treatment of wood by using the reaction engineering approach (REA). Bioresource Technology
102, 62146220.
Putranto, A., Chen, X.D. and Zhou, W., 2011e. Modelling of baking of cake using the reaction
engineering approach (REA). Journal of Food Engineering 105, 306311.
Putranto, A., Xiao, Z., Chen, X.D. and Webley, P.A., 2011f. Intermittent drying of mango tissues: Implementation of the reaction engineering approach. Industrial Engineering Chemistry
Research 50, 10891098.
Radziemska, E. and Lewandowski, W.M., 2008. Experimental verification of natural convective
heat transfer phenomenon from isothermal cuboids. Experimental Thermal and Fluid Science
32, 10341038.
Ramos, I.N., Miranda, J.M.R., Brandao, T.R.S. and Silva, C.L.M. 2010. Estimation of water
diffusivity parameters on grape dynamic drying. Journal of Food Engineering 97, 519525.
Rapp, A.O., 2001. Review of heat treatment of wood. Proceedings of Antibes Seminar, France.
Ruiz-Lopez, I.I.R., Ruiz-Espinosa, H., Luna-Guevara, M.L. and Garca-Alvarado, M.A., 2011.
Modelling and simulation of heat and mass transfer during drying of solids with hemispherical
shell geometry. Computers and Chemical Engineering 35, 191199.
Sakin, M., Kaymak-Ertekin, F. and Ilicali, C., 2007. Modelling the moisture transfer during baking
of white cake. Journal of Food Engineering 80, 822831.
Sanjuan, N., Lozano, M., Garcia-Pascual, P. and Mulet, A., 2004. Dehydration kinetics of red
pepper (Capsicum annuum L var Jaranda). Journal of the Science of Food and Agriculture 83,
697701.
Silva, W.P., Precker, J.W., Silva, C.M.P.D.S. and Gomes, J.P., 2010. Determination of effective
diffusivity and convective mass transfer coefficient for cylindrical solids via analytical solution
and inverse method: Application to the drying of rough rice. Journal of Food Engineering 98,
302308.
Sun, L.M. and Meunier, F., 1987. A detailed model for non-isothermal sorption in porous adsorbents. Chemical Engineering Science 42, 15851593.
Surmin-Kaolin, 2010. Company website. (Available at http://www.quarzwerke.com/surmin,/
accessed 21 November, 2012.)
Van der Sman, R.G.M., 2003 Simple model for estimating heat and mass transfer in regular-shaped
foods. Journal of Food Engineering 60, 383390.
Vaquiro, H.A., Clemente, G., Garcia Perez, J.V., Mulet, A. and Bon, J., 2009. Enthalpy driven
optimization of intermittent drying of Mangifera indica L. Chemical Engineering Research and
Design 87, 885898.
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2006a. Transient multiphase model for the
high-temperature thermal treatment of wood. AIChE Journal 52, 23402349.
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2006b. Thermal modelling of the high
temperature treatment of wood based on Luikovs approach. International Journal of Energy
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2007. Computational modelling of heat and
mass transfer during the high-temperature heat treatment of wood. Applied Thermal Engineering
27, 14241431.
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2010. Computational and experimental
analysis of high temperature thermal treatment of wood based on ThermoWood technology.
International Communications in Heat and Mass Transfer 37, 2128.
Reaction engineering approach II

Spatial-REA (S-REA)
3.1
The S-REA formulation

In order to capture detailed information about the distributions of moisture content and
temperature throughout the material being dried, the S-REA has been developed. S-REA
is a non-equilibrium multiphase drying approach in which the REA is implemented to
represent the local phase change term. It is envisaged that, for a better understanding of
the transport phenomena, application of effective liquid diffusion alone without source
terms in both energy and mass conservation equations may not be sufficient as it cannot
represent the water vapour concentration during drying. This could be affected by gas in
the pore structure (Chen, 2007). Also, vapour generation and transfer may affect other
volatile transport in the same material. Traditionally, effective liquid diffusion has been
used to simulate the detailed profiles of temperature and moisture content. This will be
discussed further in Chapter 4.
Equilibrium and non-equilibrium approaches can be implemented in the multiphase
drying model mentioned previously (Zhang and Datta, 2004; Datta, 2007). By applying
the equilibrium approach, it is assumed that vapour pressure inside the pores of the
samples equilibrates with the liquid moisture content inside the same pores and the
relationship can be described by the relevant equilibrium isotherm (Zhang and Datta,
2004). The equilibrium model has been applied to the baking process and a good
agreement with experimental data has been shown (Ni et al., 1999; Zhang et al., 2005;
Zhang and Datta, 2006). The equilibrium model has been implemented by combining
the mass conservation of water in both liquid and vapour phases, which effectively
resulted in the elimination of the source term. The moisture content and water vapour
concentration are related by the available isotherm data (through a correlation equation)
for the materials. Reasonable agreement with experimental data was shown in the cases
investigated (Ni et al., 1999; Zhang et al., 2005; Zhang and Datta, 2006). Similarly, the
equilibrium model was applied by Aversa et al. (2010) to model the convective drying of
food materials. The model has been shown to represent the experimental data reasonably
well. Nevertheless, it has not been proven that use of equilibrium approach is valid in
the case of heating hygroscopic materials (Zhang and Datta, 2004).
For a more generic application of the multiphase drying model, it is suggested that
the non-equilibrium approach is more appropriate. The good non-equilibrium multiphase drying model is also useful in determining the appropriateness of the equilibrium
approach for the situation of concern (Zhang and Datta, 2004). In order to implement
122
x
Figure 3.1 Schematic diagram of a cube dried in a uniform convective environment.
the model, it is necessary to represent the internal evaporation rate explicitly and appropriately. The internal evaporation/condensation rate is implemented in the multiphase
drying model as a depletion term for the liquid phase and as a source term for the vapour
phase. It has been proposed that the internal evaporation/wetting rate can be related to
the difference of equilibrium vapour pressure and the vapour pressure at a particular
time inside the pore spaces (Chong and Chen, 1999; Scarpa and Milano, 2002; Zhang
and Datta, 2004). In other words, gas must be present in the pores to permit this process
to occur. The REA, in its lumped format, has been proven to model the global drying
rate of various challenging drying cases accurately (Chen and Lin, 2005; Chen, Pirini
and Ozilgen, 2001; Chen and Xie, 1997; Lin and Chen, 2005; 2006; 2007; Putranto
et al., 2010a,b, 2011ae). It was expected that formulation of the L-REA may also be
applicable in representing the source term in the S-REA approach; for instance, the same
activation energy profile can be reserved for the same material.
The S-REA consists of a mass balance of liquid water, mass balance of water vapour
and heat balance. For uniform convective drying of cubic object in a heated environment,
three-dimensional modelling can be established. The mass balance of water in the liquid
phase (liquid water) is written as (refer to Figure 3.1 and Chong and Chen, 1999; Chen,
2007; Kar and Chen, 2011; Putranto and Chen, 2013; Zhang and Datta, 2004):

(C s X )
(Cs X )
(C s X )
(C s X )
=
Dw
+
Dw
+
Dw
I,
t
x
x
y
y
z
z
(3.1.1)
where X is the concentration of liquid water (kg H2 O kg dry solids1 ), Cs is the solids
concentration (kg dry solids m3 ), which can change if the structure is shrinking, Dw is
liquid diffusivity (m2 s1 ), I is the evaporation or condensation rate (kg H2 O m3 .s1 )
and I is usually defined as positive when evaporation occurs locally. The liquid diffusivity
represents the movement of liquid water inside the pore structure of the materials due
to capillary action as a result of the water concentration gradient. In practice, the liquid
diffusivity needs to be extracted from the available effective diffusivity data (Datta,
2007).
Reaction engineering approach II: S-REA
123
The mass balance of water vapour is expressed as (Chen, 2007; Chong and Chen,
1999; Kar and Chen, 2011; Putranto and Chen, 2013):

Cv
Cv
Cv
Cv
=
Dv
+
Dv
+
Dv
+ I,
(3.1.2)
t
x
x
y
y
z
z
where Cv is the concentration of water vapour (kg m3 ) and Dv is the effective water
vapour diffusivity in pore channels (m2 s1 ).
The heat balance is represented by the following equation (Chen, 2007; Chong and
Chen, 1999; Kar and Chen, 2011; Putranto and Chen, 2013):

T
T
C p
(3.1.3)
=
k
+
k
+
k
IHv ,
t
x
x
y
y
z
z
where T is the sample temperature (K), HV is the vaporisation heat of water (J kg1 ), k
is the sample thermal conductivity (W m2 K1 ), is the sample density (kg m3 ) and
k and may be functions of temperature and moisture content.
For cubic objects being dried as an example, the initial and boundary conditions for
Equations (3.1.1) to (3.1.3) may be written as:
t = 0, X = X o , C v = Cvo , T = To (initial condition, uniform initial
concentrations and temperature), (3.1.4)
dX
dX
dX
= 0,
= 0,
= 0 (symmetrical boundary),
dx
dy
dz
dCv
dC v
dC v
= 0,
=0
dx
dy
dz
dT
dT
dT
= 0,
= 0,
dx
dy
dz

Cv,s
dX
x = L , Cs Dw
= h m w
v,b (convective boundary
dx
for liquid transfer),
x = 0, y = 0, z = 0,
dCv
Dv
= h m v
dx
Cv,s
v,b
dCv
= h m v
dy
Cv,s
v,b
(3.1.7)
(3.1.8)
(convective boundary
dT
= h(Tb T ) (convective boundary for heat transfer),
dx

Cv,s
dX
y = L , Cs Dw
= h m w
dy

Dv
(3.1.6)

for vapour transfer),
(3.1.5)
(3.1.9)
(3.1.10)
(3.1.11)

(convective boundary for vapour transfer), (3.1.12)
124
dT
(3.1.13)
dy

Cv,s
dX
= h m w
z = L , Cs Dw
dz

(3.1.14)
k
dCv
= h m v
Dv
dz
Cv,s
v,b

(convective boundary for vapour,
transfer)
dT
dz
(3.1.15)
(3.1.16)
where w and v are the fractions of surface area covered by liquid water and water
vapour, respectively.
The internal evaporation rate ( I) can be described as:
I = h m in Ain (Cv,s C v ),
(3.1.17)
where Ain is the internal surface area per unit volume available for phase change
(m2 m3 ) and hm,in is the internal surface mass transfer coefficient (m s1 ).
By implementing the REA, internal-surface water vapour concentration can be written
as (Kar and Chen, 2010; 2011; Putranto and Chen, 2013):

E v
(3.1.18)
C v,sat ,
Cv,s = exp
RT
where Cv,s is the internal-solid surface water vapour concentration (kg m3 ), Cv,sat is the
internal saturated water vapour concentration (kg m3 ) and Ev is the activation energy
(J mol1 ) similar to the one described in Equation (2.1.4).
Therefore, the internal evaporation rate can be expressed as (Kar and Chen, 2010;
2011; Putranto and Chen, 2013):

I = h m in Ain exp E v Cv,sat C v .
(3.1.19)
RT
In Equation (3.1.19), the REA is used to describe the local evaporation rate as affected
by pore structure (porosity, shrinkage, local moisture content and local temperature).
These microstructural effects can be encapsulated in the term hm,in Ain .
In particular, Ain is clearly influenced by structural or microstructural of the material of
concern. It is interesting to note that when Ain is zero (i.e. for a non-porous or voidless
material), I becomes zero. Then, the liquid transfer (may be a kind of diffusion form)
is predominant. In this case, for soft materials such as polymeric and biological entities,
the free volume concept may be used to predict the effective liquid diffusivity (Vrentas
and Duda, 1977; Van der Sman, 2007a,b; Van der Sman et al., 2012). Free Volume
FloryHuggins (FVFH) theory, an extension of the classical FloryHuggins theory, can
be used to describe the thermodynamics of food materials. The chemical potential of
moisture transfer can be assumed due to osmotic, elastic and ionic contributions.
125
The FloryHuggins theory can also be implemented to predict the mutual diffusivity
(Dm ). By using Darkens relation, the mutual diffusivity can be expressed as (Van der
Sman et al., 2012):

Dm = Q Ds,s + (1 )Ds,w ,
(3.1.20)
where is the volume fraction of polymer, is the FloryHuggins interaction parameter,
Ds,s and Ds,w are the self-diffusivity of polymer and water, respectively, and Q is described
using FloryHuggins theory as (Van der Sman et al., 2012):
Q = 1 2 (1 ).
(3.1.21)
The StokesEinstein relation can be used to predict Ds,s given by (Van der Sman et al.,
2012):
w
,
(3.1.22)
Ds,s = Ds,o
eff
where Ds,o is the polymer self-diffusivity at infinite dilution, w is the viscosity of water
and eff is the viscosity of the polymer solution.
Ds,w can be predicted by Vrentas and Dudas free volume theory (Vrentas and Duda,
1977), which can be written as:
ln
E
y w Vw + ys Vs
Ds,w
=
,
Dw,o
RT
yw K ww (K sw Tg,w + T ) + ys K ws (K ss Tg,s T )
(3.1.23)
where E is the activation energy, Kij is free volume parameter(s), Vi * is parameters

related to the volume of the molecule and is the shape factor. The mutual diffusivity
(Dm ) may then be used as effective liquid diffusivity (Van der Sman, 2012).
3.2
Determination of the S-REA parameters

In S-REA, there are several parameters involved; i.e. effective vapour diffusivity (Dv ),
tortuosity ( ), porosity (), solid concentration (Cs ), capillary diffusivity (Dw ), internal
mass transfer coefficient (hm,in ) and internal surface area per unit volume (Ain ). The
procedures used to determine these parameters are explained in this section.
The effective vapour diffusivity is deduced from (Bird et al., 2002):
(3.2.1)
Dv = Dvo ,
while Dvo is the water vapour diffusivity (m2 s1 ), which is dependent on temperature.
For food and biological materials, Dv can be expressed as (Slattery and Bird, 1958):
Dvo = 2.09 105 + 2.137 107 (T 273.15).
(3.2.2)
The tortuosity ( ) of the samples is generally related to the porosity. The relationship
can be represented as (Audu and Geffreys, 1975; Gimmi et al., 1993):
= n ,
(3.2.3)
126
where n is the value between 0 and 0.5 (Audu and Geffreys, 1975; Gimmi et al.,
1993). A more refined approach would be to somehow incorporate the information on
microstructure in mathematical terms in order to evaluate and .
Cs is the solid concentration (kg m3 ) which can be expressed by (Kar, 2008; Kar and
Chen, 2010; 2011):
Cs =
1
,
+ Xw
1
s
(3.2.4)
while is the porosity which is dependent on shrinkage and local moisture content. This
can be determined according to (Madiouli et al., 2007):
s

X +1
V0
w
= 1 (1 0 )
.
(3.2.5)
V
1 + ws X 0
Until now, there has been no method to measure effective liquid diffusivity (Chen,
2007). Many drying research papers estimated the effective liquid diffusivity based on
drying kinetics data. Several sets of drying followed by complex optimization procedures are used to generate the effective diffusivity function (Azzouz et al., 2002; Mariani
et al., 2008; Pakowski and Adamski, 2007; Thuwapanichayanan et al., 2008; Vaquiro
et al., 2009). The literature on effective diffusivity may be used as a basis to determine
the effective liquid diffusivity to be used in the S-REA. A little adjustment is required
to the effective liquid diffusivity to generate the effective liquid diffusivity function,
since the effective liquid diffusivity in these existing literatures is used to represent the
whole phenomenon in drying (liquid diffusion, vapour diffusion, Darcy flow, evaporation/condensation).
For example, Srikiatden and Roberts (2008) reported the effective liquid diffusivity
of potato tissues as:

25.77 103
.
(3.2.6)
Deff = 1.0418 105 exp
8.314T
This was uniquely generated through well-controlled isothermal drying experiments
carefully arranged so the temperature dependence is correlated against the sample temperature. The real liquid diffusivity is, however, expected to be smaller than the effective
liquid diffusivity shown in Equation (3.2.6) but the temperature dependence is expected
to remain valid. Equation (3.2.6) can still be used as the basis but is altered as:

25.77 103
Dw = Dw0 exp
.
(3.2.7)
8.314T
For convective drying of potato tissues, it was found that Dwo of 6.5 106 m2 s1 gives
the best agreement with experimental data (Putranto and Chen, 2013).
The internal mass transfer coefficient (hm,in ) shown in Equation (3.1.17) is associated
with the pore surfaces (porous media) or surfaces of the particles (packed beds), and
internal to the sample being dried. Initially, moisture is present in the void spaces of
pores and within the pores. As drying proceeds, the moisture may migrate within the
pores (on the pore surfaces) by liquid (surface) diffusion and from the surfaces of the
127
pores through evaporation (Chong and Chen, 1999; Kar and Chen, 2010; 2011). Even
at low water content, surface diffusion of liquid could occur along the pore surface
accessible to air (Chen and Mujumdar, 2008). The internal mass transfer coefficient
shown in Equation (3.1.17) incorporates the restriction factor as it may be affected by
the pore structure and pore network inside the samples. This makes the value of hm,in
grow from a small value to the value of Dv /rp (when the constriction factor = 1) (Kar
and Chen, 2010; 2011). The internal surface area per unit volume (Ain ) can be calculated
using the procedures described in Kar and Chen (2010; 2011). It is calculated based on
the area of a single cell and the number of cells per unit volume, dependent on solid
mass of the samples and mass of single cell. Of course, Ain should also be affected by
moisture content.
By using the REA, the relative activation energy (Ev /Ev,b ) generated from one
accurate drying run is used to describe the local evaporation rate shown in Equation
(3.1.19). The activation energy (Ev ) and equilibrium activation energy (Ev,b ) are
calculated using Equations (2.1.5) and (2.1.7), respectively to yield the relative activation
energy shown in Equation (2.1.6). Since in the S-REA the relative activation energy
is used to represent the local evaporation rate, the average moisture content (X ) in
the relative activation energy is replaced by the local moisture content (X). The spatial
profiles of moisture content, concentration of water vapour and temperature are generated
by solving a set of equations shown in Equations (3.1.1)(3.1.3) simultaneously in
conjunction with the initial and boundary conditions indicated in Equations (3.1.4)
(3.1.16).
3.3
The S-REA for convective drying

The validity of the S-REA in modelling convective drying is benchmarked against the
experimental data of mango tissues (Vaquiro et al., 2009) and potato tissues (Srikiatden
and Roberts, 2008). The experimental data from drying of mango tissues are derived from
the previous study (Vaquiro et al., 2009). For better understanding of the predictions,
the necessary experimental details are summarised and reviewed in this section. The
samples of mango tissues were formed as cubes with initial side lengths of 2.5 cm while
the initial moisture content and temperature were 9.3 kg kg1 and 10.8 C, respectively.
The laboratory drier was described in Sanjuan et al. (2004). During drying the weight
change of the sample and the centre temperature history were recorded. The drying air
temperature and air velocity were controlled at preset values by PID control algorithms
while air humidity was maintained constant during drying. The experimental setting for
convective drying is shown in Table 3.1.
The density, thermal conductivity, heat capacity, equilibrium moisture content and
shrinkage of the samples are presented in previous publication (Putranto et al., 2011a).
For convective drying of potato tissues, the experimental data were taken from the previous work (Srikiatden and Roberts, 2008). Their experimental details are also reviewed
here for better understanding of the modelling approach (Roberts et al., 2002; Srikiatden
and Roberts, 2006; 2008). The cylindrical samples of Russet potatoes with diameters
128
Table 3.1 Experimental conditions of convective drying of mango tissues

(Vaquiro et al., 2009).
Number
Air velocity
(m s1 )
Air temperature
(C)
Air humidity
1
2
3
4
4
4
45
55
65
0.0134
0.0134
0.0134
of 1.4 and 2.8 cm were obtained using cylindrical cutters. The samples were sealed
at their top and bottom ends with epoxy to establish approximately a one-dimensional
(radial direction) moisture transfer. The experiments were conducted in a laboratory
convective dryer with a drying air temperature of 70 C and axial velocity of 1.5 m s1 .
The experimental setup can be found in Srikiatden and Roberts (2006). The fan at the
bottom of the sample draws air downward and this reduces the turbulence effect near
the sample as the air moves downwards (Roberts et al., 2002).
The samples with the diameter of 1.4 cm were cut into two concentric parts for
measurement of moisture content distribution, i.e. core and cortex (the core is a cylinder
with the radius of 0.35 cm derived from the inner part of the potato tissues, while the
cortex is a concentric shell derived from the outer part of the potato tissues). For the
samples with the diameter of 2.8 cm, the samples were cut into four concentric parts;
i.e. core, cortex 1, cortex 2 and cortex 3. Similarly to the samples with the diameter of
1.4 cm, the core is a cylinder with a radius of 0.35 cm derived from the innermost part
of the potato tissues. The procedures were repeated for a number of intervals (Srikiatden
and Roberts, 2006).
3.3.1
Mathematical modelling of convective drying of mango tissues using the S-REA

Based on the experiments reported by Vaquiro et al. (2009) which have been used to help
establish the REA, the samples were dried from three directions (x, y and z directions)
so three-dimensional modelling of the S-REA for convective and intermittent drying of
mango tissues needs to be set up, which is presented next.
The mass balance of water in the liquid phase (liquid water), the mass balance of water
in the vapour phase (water vapour) and the heat balance are shown in Equations (3.1.1),
(3.1.2) and (3.1.3), respectively, while the initial and boundary conditions for equations
are shown in Equations (3.1.4)(3.1.16). Since the sample dried was a cube shape dried
uniformly from all directions (x, y and z directions) (Sanjuan et al., 2004; Vaquiro et al.,
2009), the mass balance of water in liquid phase can be simplified into (Incropera and
DeWitt, 2002; Van der Sman, 2003):

(Cs X )
(C s X )
=3
Dw
I,
(3.3.1)
t
x
x
while the mass balance of water in vapour phase can be expressed as:

Cv
C v
=3
Dv
+ I.
t
x
x
In addition, the heat balance can be represented as:

T
T
C p
=3
k
IHv .
t
x
x
129
(3.3.2)
(3.3.3)
The internal-surface water vapour concentration (Cv,s ) and internal evaporation rate ( I)
are evaluated using Equations (3.1.18) and (3.1.19).
The relative activation energy of convective drying of mango tissues is generated from
one accurate drying run of convective drying mango tissues under constant environment
conditions with a drying air temperature of 55 C (Vaquiro et al., 2009). The activation
energy during drying is evaluated using Equation (2.1.5) and divided with the equilibrium
activation energy represented in Equation (2.1.7) to yield the relative activation energy as
mentioned in Equation (2.1.6). The relationship between the relative activation energy
and average moisture content can be represented by a simple mathematical equation
obtained by use of the least-squares method using Microsoft Excel (Microsoft Inc.,
2012). The relative activation energy can be represented as:
E v
= 9.92 104 (X X b )3 + 9.74 103 (X X b )2
E v,b
0.101(X X b ) + 1.053.
(3.3.4)
The good agreement between the fitted and experimental relative activation energy is
shown by R 2 of 0.999. Although Equation (3.3.4) involves Xb as mentioned earlier, all
successful applications of REA so far suggest that the experiments carried out should
dry the materials to Xb s of very small values in order to allow the correlations such as
Equation (3.315) to cover the widest range of water content of practical interest. If Xb
is close to initial water content, the activation energy calculated from the laboratory data
can be misleading.
The relative activation energy correlated with Equation (3.3.4) has been implemented
to model the convective and intermittent drying of mango tissues using the L-REA and
the results of modelling already matched well with experimental data (Putranto et al.,
2011a,b). For modelling using the S-REA here, the relative activation energy shown
in Equation (3.3.4) is used but the average moisture content X in Equation (3.3.4) is
substituted by the local moisture content (X) as the REA is used to represent the local
evaporation rate instead of the overall drying rate of the whole sample. In addition, it
is emphasised that, for the S-REA, the equilibrium relative activation energy (Ev,b ) is
evaluated at corresponding humidity and temperature inside the pores of the samples
under equilibrium condition.
The effective vapour diffusivity (Dv ), tortuosity ( ), solid concentration (Cs ) and
porosity () are deduced using Equations (3.2.1)(3.2.5). Similarly, the internal mass
transfer coefficient (hm,in ) is evaluated using the procedures described in Section 3.2.
130
The effective diffusivity of mango tissues presented by Vaquiro et al. (2009) is

expressed as:

1.885102
X
38.924 103
3
. (3.3.5)
Deff = 2.933 10 exp
8.314T
X +1
Equation (3.3.5) can be used as an approximation to determine the liquid water diffusivity
of mango tissues but a little adjustment of the constant is needed in order to match the
prediction with the experimental data of moisture content and temperature. The liquid
water diffusivity used in this study can be expressed as:

1.885102
X
31.924 103
3
. (3.3.6)
Dw = 2.933 10 exp
8.314T
X +1
In order to yield the spatial profiles of moisture content, water vapour concentration
and temperature of the convective of mango tissues, the mass and heat balances shown
in Equations (3.3.1)(3.3.3) in conjunction with the initial and boundary conditions
represented in Equations (3.1.4)(3.1.16) and the relative activation energy shown by
Equation (3.3.4) are solved by the method of lines (Chapra, 2006; Constatinides, 1999).
By this method, the partial differential equations are transformed into a set of ordinary
differential equations with respect to time by firstly discretising the spatial derivatives.
The ordinary differential equations are then solved simultaneously by ode23s in Matlab
(Mathworks Inc., 2012). The spatial derivative here is discretised into 10 increments;
application of 200 increments has been conducted and there is no real difference in the
profiles observed as shown in Figure 3.2.
The shrinkage (Putranto et al., 2011a) is incorporated in the modelling by a moving mesh in which the number of intervals is kept constant but the intervals of each
increment are allowed to change according to the shrinkage relationship. The moving
mesh was found to give better agreement with experimental data than a fixed coordinate
(immobilising boundary) (Thuwapanichayanan et al., 2008).
The average moisture content of mango tissues during convective drying is evaluated
by:
L(t)

X=
X (x)d x
0
L(t)

(3.3.7)
dx
The profiles of average moisture content and centre temperature are then validated
against the experimental data of Vaquiro et al. (2009).
3.3.2
Mathematical modelling of convective drying of potato

tissues using the S-REA
In the experiments reported by Srikiatden and Roberts (2008) which are of interest, the
samples were covered at both the top and bottom end to promote the one-dimensional
131
10
S-REA-10 increments
S-REA-200 increments
Data 45C
9
8
7
6
5
4
3
2
1
0
0.5
1.5
2.5
3.5
t(s)
4.5
5
104
Figure 3.2 Moisture content profiles of the convective drying of mango tissues at a drying air
temperature of 45 C solved by the method of lines with 10 and 200 spatial increments.
[Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao Dong Chen, Spatial reaction
engineering approach as an alternative for nonequilibrium multiphase mass-transfer model for
drying of food and biological materials, 5567, Copyright (2012), with permission from John
Wiley & Sons Inc.]
drying condition with respect to radial direction (Srikiatden and Roberts, 2008) so onedimensional modelling (at radial directions) of the S-REA of the convective drying of
cylindrical potato tissues is possible and can be represented by a set of equations of
conservation next.
The mass balance of liquid water can be represented as (Chen, 2007; Chong and Chen,
1999; Kar and Chen, 2011; Zhang and Datta, 2004):

1
(Cs X )
(Cs X )
=
Dw r
I,
(3.3.8)
t
r r
r
where X is the concentration of liquid water (kg H2 O kg dry solids1 ) and Cs is the
solids concentration (kg dry solids m3 ), I is the evaporation or wetting rate (kg H2 O
m3 s1 ) and I is >0 when evaporation occurs locally.
The mass balance of water vapour can be expressed as (Chen, 2007; Chong and Chen,
1999; Kar and Chen, 2011; Zhang and Datta, 2004):

1
Cv
Cv
=
Dv r
+ I,
(3.3.9)
t
r r
r
where Cv is the concentration of water vapour (kg H2 O m3 ).
132
In addition, the heat balance can be written as (Chen, 2007; Chong and Chen, 1999;
Kar and Chen, 2011; Zhang and Datta, 2004):

T
1
T
=
kr
IHv ,
C p
(3.3.10)
t
r r
r
where T is the sample temperature (K).
The initial and boundary conditions of Equations (3.3.8)(3.3.10) are:
t = 0, X = X o , Cv = Cvo , T = To (initial condition, uniform initial
concentrations and temperature),
(3.3.11)
dCv
dT
dX
= 0,
= 0,
= 0 (symmetrical condition),
(3.3.12)
dr
dr
dr

dX
Cv,s
r = R, Cs Dw
= h m w
v,b (convective boundary for
dr
liquid water transfer), (3.3.13)

r = 0,
dCv
= h m v
Dv
dr
Cv,s
v,b

(convective boundary for water
vapor transfer),
(3.3.14)
dT
(3.3.15)
= h(Tb T ) (convective boundary for heat transfer).
dr
Similar to the convective drying of mango tissues, the internal solid-surface water
vapour concentration and the local evaporation rate (I ) are evaluated using Equations
(3.1.17) and (3.1.18). The relative activation energy of convective drying of potato tissues
is generated from one accurate drying run of the convective drying of potato tissues with
a diameter of 1.4 cm at a drying air temperature of 70 C (Srikiatden and Roberts,
2008). The activation energy during drying was evaluated using Equation (2.1.5) based
on the experimental data of moisture content and surface temperature during drying
(Srikiatden and Roberts, 2008). It is then divided with the equilibrium activation energy
represented in Equation (2.1.7) to yield the relative activation energy as mentioned in
Equation (2.1.6). The relationship between the relative activation energy and average
moisture content can be represented by a simple mathematical equation obtained by
the least-squares method using Microsoft Excel (Microsoft Inc., 2012). The relative
activation energy can be represented as:

Ev
= exp 0.364(X X b )0.876 .
(3.3.16)
E v,b
k
Similar to modelling of convective drying of mango tissues, for modelling using

the S-REA, the average moisture content X in Equation (3.3.16) is substituted by the
local moisture content (X) as the REA is then able represent the local evaporation or
condensation rate here instead of the global drying rate.
The effective liquid diffusivity (Dw ) is shown in Equation (3.2.7) while the effective
vapour diffusivity (Dv ), tortuosity ( ), solid concentration (Cs ) and porosity () are
133
deduced using Equations (3.2.1)(3.2.5). Similarly, the internal mass transfer coefficient
(hm,in ) is evaluated using the procedures explained in Section 3.2.
The average moisture content in the core of potato tissues (Xcore ) is evaluated by:
Rcore
X core =
X (r )r dr
0
Rcore
(3.3.17)
rdr
The average moisture content in cortex (Xcortex ) is evaluated by:

Rsample

X cortex3 =
Rcortex
out
X (r )r dr
.
in
Rcortex

out
Rcortex
in
(3.3.18)
r dr
The results of modelling average moisture content in core and cortex (hence the spatial distribution) are validated against the experimental data of Srikiatden and Roberts
(2008). Similarly to the convective drying of mango tissues, the mass and heat balances
are shown in Equations (3.3.8)(3.3.10) in conjunction with the initial and boundary
conditions indicated in Equations (3.3.11)(3.3.15). The application of 10 and 200
increments did not result in noticeable differences in the profiles.
3.3.3
Results of modelling of convective drying of mango tissues using the S-REA

The S-REA is used to model the convective drying of mango tissues at drying air temperatures of 45, 55 and 65 C. The original formulation of the L-REA is implemented
in the partial differential equation set for transport in porous media to represent the local
drying or condensation rate. It is thus coupled with the system of equations of conservation to describe the spatial profiles of moisture content, water vapour concentration and
temperature. It is noted that, if locally there is no vacant pore space which is connected
with other pores or channels, the internal mass transfer area should be considered to
be zero; hence the REA term is zero if the pores or channels are fully hydrated. In
this study, the internal mass transfer coefficient (hm,in : see Equation 3.3.12) is chosen to
0.01 m s1 as the sensitivity analysis indicates that hm,in is likely to be higher than
0.01 m s1 , but any higher than this and it does not give any noticeable difference in
the profiles of moisture content and temperature predicted. More importantly, the value
of 0.01 m s1 is also in the order of Dv /rp (Kar and Chen, 2010; 2011); thus it is a
fundamental value.
The good agreement between the predicted and experimental data of the average
moisture content and the centre temperature is shown in Figures 3.3 and 3.4. It is also
supported by R 2 of moisture content higher than 0.996 and R 2 of temperature higher
than 0.985 as listed in Table 3.2. The results of the S-REA modelling match well with
the experimental data. Benchmarks against the diffusion-based model (Vaquiro et al.,
10
S-REA 45C
Data 45C
S-REA 55C
Data 55C
S-REA 65C
Data 65C
9
8
7
6
5
4
3
2
1
0
0.5
1.5
2.5
t(s)
3.5
4.5
5
4
10
Figure 3.3 Average moisture content profiles of mango tissues during convective drying at
different drying air temperatures. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao
Dong Chen, Spatial reaction engineering approach as an alternative for nonequilibrium
multiphase mass-transfer model for drying of food and biological materials, 5567, Copyright
340
Temperature (K)
330
320
310
S-REA 45C
Data 45C
S-REA 55C
Data 55C
S-REA 65C
Data 65C
300
290
280
0.5
1.5
2.5
t(s)
3.5
4.5
5
4
10
Figure 3.4 Centre temperature profiles of mango tissues during convective drying at different
drying air temperatures. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao Dong Chen,
Spatial reaction engineering approach as an alternative for nonequilibrium multiphase
mass-transfer model for drying of food and biological materials, 5567, Copyright (2012), with
135
Table 3.2 R2 and RMSE of convective drying of mango tissues

using the S-REA.
Drying air
temperature
(C)
R2 for X
RMSE for X
R2 for T
RMSE for T
45
55
65
0.998
0.999
0.996
0.103
0.079
0.150
0.998
0.985
0.994
0.285
1.004
0.842
10
9
8
7
6
5
t = 1000s
t = 3000s
t = 5000s
t = 10000s
t = 20000s
t = 30000s
t = 35000s
4
3
2
1
0
0
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.5 Spatial moisture content profiles of mango tissues during convective drying at drying
air temperatures of 45 C. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao Dong
Chen, Spatial reaction engineering approach as an alternative for nonequilibrium multiphase
2009) indicate that the S-REA yields comparable, or even better, agreement towards the
experimental data.
Figure 3.5 shows the spatial profiles of the moisture content during convective drying
of mango tissues at a drying air temperature of 45 C. The moisture content at the
outer part of the samples is lower than that at the inner part, which indicates the effect
of moisture removal. Initially, the gradient of moisture content inside the samples is
relatively high but this decreases as the drying progresses. At the end of drying, no
noticeable gradient of moisture content is observed, which indicates the equilibrium
moisture content is nearly approached. If no liquid diffusion mechanism is used, the
S-REA model would not be able to project this kind of liquid water profile (Kar and
Chen, 2010; 2011).
0.012
Water vapour concentration (kg/m3)
136
t = 1000s
t = 3000s
t = 5000s
t = 10000s
t = 20000s
t = 30000s
t = 35000s
0.01
0.008
0.006
0.004
0.002
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.6 Spatial water vapour concentration profiles of mango tissues during convective drying
at drying air temperatures of 45 C. [Reprinted from AIChE Journal, 59, Aditya Putranto,
Xiao Dong Chen, Spatial reaction engineering approach as an alternative for nonequilibrium
The S-REA can generate the spatial profiles of water vapour concentration. The
spatial profiles of water vapour concentration during convective drying of mango tissues
at a drying air temperature of 45 C are shown in Figure 3.6. The profiles of water
vapour concentration are significantly affected by the local composition and structure
of the samples being dried. Along drying, the concentration of water vapour achieves
a maximum at a particular position inside the samples. This could be because, at the
core of samples, the moisture content is higher than that of the outer part which makes
the porosity of the core of samples lower. The lower porosity retards the evaporation
rate at the sample core. At the outer part of the samples, the water extraction rate may
be enhanced because of higher porosity but this seems to be balanced by high diffusive
water vapour transfer as a result of higher porosity and temperature at the outer part of
the samples. The S-REA seems to capture this physics well and can model the profiles
of water vapour concentration well qualitatively.
The spatial profiles of temperature are presented in Figure 3.7. The temperature of
the outer part of the samples is higher than that of the inner part because the samples receive heat by convection from the drying air and this is used for vaporisation;
if any is left over as such, this would penetrate further inwards by conduction. However, the gradient of temperature inside the samples is not large which may indicate
that the temperature inside the samples is essentially uniform. This is in agreement
137
330
t = 1000s
t = 3000s
t = 5000s
t = 10000s
t = 20000s
t = 30000s
t = 35000s
Temperature (K)
325
320
315
310
305
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.7 Spatial temperature profiles of mango tissues during convective drying at drying air
temperatures of 45 C. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao Dong Chen,
with the prediction of the ChenBiot number (ChBi) (Chen and Peng, 2005; Putranto
et al., 2011a) which remains low (less than 0.3) during drying reported previously
(Putranto et al., 2011a).
Figure 3.8 indicates the local evaporation rate inside mango tissues during convective
drying at a drying air temperature of 55 C. As drying proceeds, the evaporation rate
at the inner part is smaller than that of the outer part, which could be due to high
moisture content at the inner part of the sample. This means a lower porosity there
that retards the evaporation rate. The observation is also in agreement with the intuitive
explanation of profiles of water vapour concentration during drying by Chen (2007). As
drying progresses, the evaporation rate increases as the temperature increases. However,
the increase is observed up to a drying time of around 15 000 s. After this period, the
evaporation rate decreases as the moisture content inside the samples is depleted. At
the end of drying, essentially there is not much difference in evaporation rate inside the
samples because the moisture content has nearly achieved equilibrium under the drying
conditions.
Therefore, it can be said that the S-REA approach models the convective drying of
mango tissues well and the original REA is a simple alternative approach to represent
the local evaporation and condensation rates. In addition, the S-REA has been easily
138
1.2
t = 1000s
t = 5000s
t = 10 000s
t = 15 000s
t = 20 000s
t = 30 000s
t = 35 000s
Evaporation rate (kg water/(m3s))
1
0.8
0.6
0.4
0.2
0
0.2
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.8 Profiles of evaporation rates inside mango tissues during convective drying at a drying
air temperature of 55 C. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao Dong Chen,
operated to yield the profiles of water vapour concentration, local evaporation rate and
local heat evaporation rate inside the mango tissues during drying.
3.3.4
Results of modelling of convective drying of potato tissues using the S-REA

As mentioned earlier, the S-REA has also been implemented to model the convective
drying of potato tissues. The relative activation energy is generated from one accurate
drying run which is the convective drying at air temperature of 70 C. It is represented
in Equation (3.3.16). Similar to the convective drying of mango tissues, the internal
mass transfer coefficient (hm,in : on Equation 3.1.17) is chosen to be 0.01 m s1 as
the sensitivity analysis indicates that hm,in of higher than 0.01 m s1 does not give any
noticeable differences in the profiles of moisture content and temperature. Nicely, this
is also in the order of Dv /rp as suggested by Kar and Chen (2010; 2011), hence hm,in
is a fundamental value. However, Dwo (in Equation 3.2.7) is determined by sensitivity
analysis and it is found that Dwo of 6.5 106 m2 s1 gives the best agreement against
the experimental data. It is emphasised that the temperature dependence function for the
liquid diffusivity in this case was obtained in isothermal drying experiments specially
designed by Srikiatden and Roberts (2006), in contrast to many published studies that
139
6
S-REA-core
Data-core
S-REA-cortex
Data-cortex
0.5
1.5
t(s)
2.5
104
Figure 3.9 Moisture content profiles in the core and cortex during convective drying of potato
tissues with a diameter of 1.4 cm. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao
suggest the diffusivity in the material is related to drying air temperature instead (Chen,
2007).
Results of modelling convective drying of potato tissues are shown in Figures 3.9
3.11. Figure 3.9 shows the profiles of moisture content of each part of potato cylindrical
tissue with the diameter of 1.4 cm during the convective drying. It can be shown that the
results of modelling match well with the experimental data with a correlation coefficient
R2 of 0.98. Benchmarks against modelling implemented by Srikiatden and Roberts
(2008) with the liquid diffusivity concept indicate that the S-REA yields comparable
results.
In addition, the profiles of moisture content for each part of samples with the diameter
of 2.8 cm are shown in Figure 3.10. Again, a good agreement towards the experimental
data is observed (R2 of 0.992). Indeed, the S-REA describes the moisture content profiles accurately during convective drying of potato tissues with a diameter of 2.8 cm.
Benchmarks towards modelling implemented by Srikiatden and Roberts (2008) indicate
that the REA yields comparable or even better results. It can be said that the S-REA can
be used to model the profiles of moisture content very well.
Figure 3.11 indicates the core temperature during convective drying of potato tissues
with the diameter of 1.4 cm. The predictions of temperature using the S-REA match
3
S-REA-core
S-REA-cortex 1
S-REA-cortex 2
S-REA-cortex 3
Data-core
Data-cortex 1
Data-cortex 2
Data-cortex 3
0.5
1.5
2.5
t(s)
3.5
104
Figure 3.10 Moisture content profiles in the core and cortex during convective drying of potato
tissues with a diameter of 2.8 cm. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao
345
340
Core temperature (K)
335
330
325
320
315
310
305
S-REA
Data
300
295
0.5
1.5
t(s)
2.5
104
Figure 3.11 Core temperature profiles during convective drying of potato tissues with a diameter
of 1.4 cm. [Reprinted from AIChE Journal, 59, Aditya Putranto, Xiao Dong Chen, Spatial
reaction engineering approach as an alternative for nonequilibrium multiphase mass-transfer
model for drying of food and biological materials, 5567, Copyright (2012), with permission
from John Wiley & Sons Inc.]
141
Table 3.3 Scheme of intermittent drying of mango tissues (Vaquiro

et al., 2009).
Drying air
temperature
(C)
Period of first
heating (s)
Period of resting
(at 27 C 1.6)
(s)
Period of
second
heating (s)
45
55
16 200
9480
10 800
10 800
36 360
33 720
well with the experimental data with R2 of 0.99. Benchmarks against modelling implemented by Srikiatden and Roberts (2008) indicate that the S-REA yields comparable
results. Overall, S-REA seems to be a sound approach to modelling the details of spatial
distributions of temperature, liquid water and water vapour concentration in the material
being dried.
3.4
The S-REA for intermittent drying

The accuracy of the S-REA in modelling intermittent drying is validated by the experimental data of intermittent drying of mango tissues (Vaquiro et al., 2009). For better
understanding of the modelling approach, the experimental details are summarised and
reviewed in this section. The samples of mango tissues were cubes with initial side
lengths of 2.5 cm, while the initial moisture content and temperature were 9.3 kg kg1
and 10.8 C, respectively. The laboratory drier was described in Sanjuan et al. (2004).
During drying, the weight of the sample and the centre temperature were recorded. The
drying air temperature and air velocity were controlled at preset values by PID control
algorithms while air humidity was maintained constant during drying. The experimental
setting for intermittent drying is shown in Table 3.3. During the resting period, the samples stayed in an environment with an ambient temperature of 27 1.6 C and relative
humidity of 60% (Vaquiro et al., 2009). Determination of density, thermal conductivity,
heat capacity, equilibrium moisture content and shrinkage of samples being dried has
been described previously (Putranto et al., 2011a).
3.4.1
The mathematical modelling of intermittent drying using the S-REA

In experiments reported by Vaquiro et al. (2009), the subject of interest, the samples were
dried from three directions (x, y and z directions) so three-dimensional modelling of the
S-REA for intermittent drying of mango tissues needs to be set up, which is represented
next. The mass and heat balances of intermittent drying of mango tissues are similar to
those of convective drying described in Section 3.3.1.
Similarly to the convective drying of mango tissues, the mass balance of water in
liquid phase liquid water, the mass balance of water in the vapour phase (water vapour)
and the heat balance are shown in Equations (3.3.1), (3.3.2) and (3.3.3), respectively,
while the initial and boundary conditions for equations are shown in Equations (3.1.4)
(3.1.16).
142
Table 3.4 R 2 and RMSE of intermittent drying of mango tissues.
45 C
55 C
R 2 for X
RMSE for X
R 2 for T
RMSE for T
0.964
0.999
0.358
0.0774
0.992
0.994
0.705
0.874
The internal-surface water vapour concentration (Cv,s ) and internal evaporation rate
are evaluated using Equations (3.1.18) and (3.1.19), respectively. The liquid diffusivity
(Dw ) is shown in Equation (3.2.7) while the effective vapour diffusivity (Dv ), tortuosity
( ), solid concentration (Cs ) and porosity () are deduced using Equations (3.2.1)
(3.2.5). Similarly, the internal mass transfer coefficient (hm,in ) is evaluated using the
procedures explained in Section 3.2.
The relative activation energy implemented for modelling of convective drying of
mango tissues shown in Equation (3.3.4) is used here to model the intermittent drying
of mango tissues. For modelling the intermittent drying of these tissues, the equilibrium activation (Ev,b ) energy shown in Equation (2.1.7) is evaluated according to the
corresponding drying air temperature and humidity in each drying period. It is also
combined with the relative activation energy shown in Equation (3.3.4) to yield the local
drying/condensation rate. In addition, the heat balance implements the corresponding
drying air temperature in each drying period by using the corresponding drying air
temperature in the boundary conditions indicated in Equations (3.1.10), (3.1.13) and
(3.1.16). The solution procedures are similar to the one for convective drying of mango
tissues, described in Section 3.3.1.
3.4.2
Results of modelling intermittent drying using the S-REA

The S-REA is implemented here to model the intermittent drying of mango tissues
whose conditions are listed in Table 3.3. As mentioned before, for modelling of the
intermittent drying, the equilibrium activation energy needs to be evaluated according
to the corresponding drying settings in each drying period. Similarly, the heat balance
implements the corresponding drying air temperature in each drying period. Solving
Equations (3.3.9)(3.3.11) in conjunction with the initial and boundary conditions shown
in Equations (3.3.4) to (3.3.8) simultaneously yields the profiles of moisture content,
concentration of water vapour and temperature during intermittent drying. Figures 3.12
3.17 show the results of modelling of the intermittent drying.
The profiles of moisture content during intermittent drying are shown in Figure 3.12.
A good agreement between the predicted and experimental data is observed and confirmed by R 2 and RMSE listed in Table 3.4. The results of modelling match well with the
experimental data of moisture content. The S-REA models the average moisture content
during the intermittent drying at drying air temperature of 45, 55 and 65 C very well.
Benchmarks against modelling implemented by Vaquiro et al. (2009) revealed that the
REA yields better results as Vaquiro et al. (2009) showed a slight underestimation in
drying rate of intermittent drying at a drying air temperature at 45 C during drying
143
10
S-REA 45C
Data 45C
S-REA 55C
Data 55C
S-REA 65C
Data 65C
9
8
7
6
5
4
3
2
1
0
4
t(s)
7
104
Figure 3.12 Average moisture content profiles of mango tissues during intermittent drying at
different drying air temperatures.
times between 20 00050 000 s. Similarly, Vaquiro et al. (2009) also revealed a slight
overestimation in drying rate of intermittent drying at a drying air temperature of 65 C
after a drying time of 20 000 s. The slight underestimation and overestimation of the
drying rate are not shown by the modelling using the S-REA. This indicates that the
S-REA can be used to describe the moisture content of intermittent drying of mango
tissues well and the REA can also be applied to model the local evaporation and condensation rate well.
The spatial profiles of moisture content during the intermittent drying at a drying air
temperature of 55 C are indicated in Figure 3.13. The moisture content of the outer
part of the samples is lower than that in the inner part of the samples, which indicates
that moisture migrates outwards during intermittent drying. Initially, the gradient of
moisture content inside the samples is relatively high, but during a drying time between
9480 and 20 280 s the gradient is relatively low since the samples are in the resting
period. This period seems to allow the moisture to redistribute inside the samples and
low gradients of moisture content inside the samples are generated. Towards the end
of drying, although the samples are in the heating period again, relatively uniform
moisture content is observed, which indicates that the equilibrium condition is almost
approached.
Figure 3.14 presents the spatial profiles of water vapour concentration during the
intermittent drying at a drying air temperature of 55 C. The profiles of water vapour
concentration are significantly affected by the local composition and structure of the
10
9
8
7
6
5
t = 1000s
t = 3000s
t = 5000s
t = 7000s
t = 14 000s
t = 20 000s
t = 30 000s
t = 40 000s
t = 50 000s
4
3
2
1
0
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.13 Spatial moisture content profiles of mango tissues during intermittent drying at a
drying air temperature of 55 C.
0.012
144
t = 1000s
t = 3000s
t = 5000s
t = 7000s
t = 14 000s
t = 20 000s
t = 30 000s
t = 40 000s
t = 50 000s
0.01
0.008
0.006
0.004
0.002
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.14 Spatial water vapour concentration profiles of mango tissues during intermittent
145
340
330
320
310
300
S-REA 45C
Data 45C
S-REA 55C
Data 55C
S-REA 65C
Data 65C
290
280
4
t(s)
7
4
10
Figure 3.15 Centre temperature profiles of mango tissues during intermittent drying at different
drying air temperatures.
samples being dried. The water vapour concentration achieves a maximum at particular
position inside the samples. At core of the samples, the moisture content is relatively
high, which may result in lower porosity and retard the local evaporation rate. At outer
part of the samples, the local evaporation rate may be enhanced as a result of the higher
porosity, but this seems to be balanced by higher diffusive flux of water vapour due to
the higher porosity and temperature. During first heating period, the gradient of water
vapour concentration is relatively high but this decreases as drying progresses. The
relatively uniform concentration of water vapour is shown during the resting period
which could be due to a relatively low temperature. During the second heating period,
relatively uniform water vapour concentration is observed. This is in agreement with the
relatively uniform moisture content at the end of drying as explained previously. The
spatial profiles of intermittent drying are similar to those of convective drying (Putranto
and Chen, 2013). In addition, this is in agreement with a qualitative prediction by Chen
(2007) which explained that the maximum water vapour concentration is achieved at a
particular position inside the samples.
The profiles of temperature during the intermittent drying are indicated in Figures 3.15
and 3.16. Figure 3.15 shows the profiles of the centre temperature during the intermittent
drying of mango tissues at drying air temperatures of 45, 55 and 65 C. A good agreement between the predicted and experimental data is observed which is also supported
by R2 of higher than 0.992 and RMSE lower than 0.828. The results of modelling match
well with the experimental data. Benchmarks against modelling implemented by Vaquiro
330
325
320
Temperature (K)
146
315
t = 1000s
t = 3000s
t = 5000s
t = 7000s
t = 14 000s
t = 20 000s
t = 30 000s
t = 40 000s
t = 50 000s
310
305
300
295
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.16 Spatial temperature profiles of mango tissues during intermittent drying at a drying
air temperature of 55 C.
et al. (2009) show that the REA yields better results since the modelling by Vaquiro et al.
(2009) indicated kinks and an underestimation of temperature profiles in the beginning
of the first heating period. However, these are not observed by the modelling using the
S-REA. This indicates that the S-REA is indeed accurate enough to model the temperature profiles of intermittent drying of mango tissues.
Figure 3.16 represents the spatial profiles of temperature during the intermittent drying
at a drying air temperature of 55 C. During the first heating period, the temperature in
the outer part of the samples is higher than that of the inner part because the samples
receive heat by convection from the drying air used for water evaporation and this
penetrates inwards by conduction. However, the gradient of temperature inside the
samples is not high, in agreement with the prediction of ChenBiot number (ChBi)
(Chen and Peng, 2005) reported previously (Putranto et al., 2011b), which indicated
that the temperature inside the samples is essentially uniform. During the first heating
period, the temperature of samples increases but the temperature decreases between
drying times of 948020 280 s as a result of resting period. This is followed by a further
increase of temperature during the second heating period. At the end of the intermittent
drying, the temperature approaches the drying air temperature.
The S-REA has its advantages, resulting in the spatial profiles of local evaporation
rate during drying as the REA is used to predict the local evaporation rate. The profiles of
local evaporation rates during intermittent drying of Scheme 1 at a drying air temperature
of 55 C are shown in Figure 3.17. The local evaporation rate at core of samples is lower
147
1.2
Evaporation rate (kg water/(m3s))
1
0.8
t = 1000s
t = 3000s
t = 5000s
t = 9000s
t = 14 000s
t = 16 000s
t = 20 000s
t = 22 000s
t = 35 000s
t = 40 000s
t = 45 000s
t = 50 000s
t = 54 200s
0.6
0.4
0.2
0
0.2
0.002
0.004
0.006
0.008
Half thickness (m)
0.01
0.012
0.014
Figure 3.17 Profiles of evaporation rate inside mango tissues during intermittent drying at a
drying air temperature of 55 C.
than that of the outer part of samples. This could be because of lower porosity at the core
of the samples as a result of higher moisture content. During the resting period, the spatial
profiles of local evaporation rate are more uniform than those during first heating period,
which may be due to a lower temperature inside the samples. Towards the end of drying,
the gradient of local evaporation rate decreases. This could be because the moisture
content decreases as drying progresses, which increases the porosity. From the start of
drying to a drying time of around 5000 s the local evaporation rate increases, which
could be because of the increase in temperature. After this period, the local evaporation
rate tends to decrease, which could be because of the decrease in moisture content.
During the resting period, the local evaporation rate changes only slightly, which may be
because of the relatively low temperature. During the second heating period, the local
evaporation decreases. This may be because the moisture content inside the sample is
depleted. At the end of drying, essentially there is not much difference in evaporation
rate inside the samples because the moisture content has nearly achieved equilibrium
under the drying conditions.
It has been demonstrated that the S-REA is very accurate in modelling the intermittent
drying of mango tissues. The S-REA can also project the concentration of water vapour
and local evaporation rate during intermittent drying so that better understanding of
transport phenomena of drying processes can be gained. It is argued here that the
S-REA is not only robust enough to model convective drying (Putranto and Chen, 2013)
but also intermittent drying.
148
Table 3.5 Experimental settings of wood heating under a constant heating

rate (Younsi et al., 2007).
3.5
Case
Final gas
temperature (C)
Heating rate
(C h1 )

1
2
220
220
10
20
0.11
0.125
The S-REA to wood heating under a constant heating rate

The accuracy and robustness of the S-REA for heat treatment of wood under a constant
heating rate is benchmarked by the experimental data of Younsi et al. (2007). The
experimental details were reported in Kocaefe et al. (1990; 2007) and Younsi et al.
(2006b) and these are reviewed here for better understanding of the current approach.
A thermogravimetric analyser was used for the heat treatment of wood as shown in
Kocaefe et al. (1990). Wood samples with dimensions of 0.035 0.035 0.2 m were
heat treated by suspending the samples on a balance with accuracy of 0.001 g. The heat
treatment was conducted by exposing the samples to hot gas whose temperature was
linearly increased according to the heating rate. The humidity of the gas was controlled
by injection of steam into the second furnace placed under the main furnace (Younsi
et al., 2006a, 2007).
The samples were first heated to 120 C and held at this temperature for half an hour
followed by heating under the preset heating rate (refer to Table 3.5) until the final
temperature (also refer to Table 3.5) was achieved. During the heat treatment, the weight
of the samples was recorded by the balance. In addition, the temperatures were measured
by a T-type thermocouple placed inside the samples. The measurements indicated that
the temperatures inside the samples were essentially uniform, perhaps due to the small
size of the samples (Younsi et al., 2006b).
3.5.1
The mathematical modelling of wood heating using the S-REA

In the experiments reported by Younsi et al. (2006a; 2007), the subject of interest
here, two-dimensional modelling with respect to x and y directions can be set up
as the height of the sample (0.2 m) is much greater than the length (0.035 m) and
width (0.035 m) of the sample. The mass balance of water in the liquid phase is
written as (Chen, 2007; Chong and Chen, 1999; Putranto and Chen, 2013; Zhang and
Datta, 2004):

(C s X )
(Cs X )
(Cs X )
=
Dw
+
Dw
I,
(3.5.1)
t
x
x
y
y
where X is the concentration of liquid water (kg H2 O kg dry solids1 ) and Cs is the
solids concentration (kg dry solids m3 ) which can change if the structure is shrinking,
I is the evaporation or wetting rate (kg H2 O m3 s1 ), I is >0 when evaporation occurs
149
locally, while the mass balance of water in the vapour phase is expressed as (Chen, 2007;
Chong and Chen, 1999; Putranto and Chen, 2013; Zhang and Datta, 2004):

C v
C v
Cv
=
Dv
+
Dv
+ I,
(3.5.2)
t
x
x
y
y
where Cv is the water vapour concentration (kg H2 O m3 ).
The heat balance is represented as (Chen, 2007; Chong and Chen, 1999; Putranto and
Chen, 2013; Zhang and Datta, 2004):

T
T
(3.5.3)
=
k
+
k
IHv ,
C p
t
x
x
y
y
where T is the sample temperature (K), k is thermal conductivity of sample (W m1
K1 ), is the sample density (kg m3 ) and Hv is the vaporisation heat of water
(J kg1 ).
The initial and boundary conditions for Equations (3.5.1) to (3.5.3) are:
t = 0, X = X o , C v = Cvo , T = To (initial condition, uniform initial concentrations
and temperature),
(3.5.4)
dX
dC v
dT
= 0,
= 0,
= 0 (symmetry boundary),
(3.5.5)
dx
dx
dx

dX
Cv,s
= h m w
x = L , y = L , Cs Dw
dx
for liquid water transfer),

x = 0, y= 0,
dCv
= h m v
Dv
dx
Cv,s
v,b
(3.5.6)

(convective boundary for
water vapor transfer),
(3.5.7)
dT
= h(Tb T ) (convective boundary for heat transfer).
(3.5.8)
dx
Because the samples were dried uniformly from all directions and the lengths and widths
were the same, the mass balance of water in liquid phase can be simplified into (Incropera
and DeWitt, 2002; Van der Sman, 2003):

(Cs X )
(Cs X )
=2
Dw
I,
(3.5.9)
t
x
x
k
while the mass balance of water in vapour phase can be expressed as:

Cv
C v
=2
Dv
+ I.
t
x
x
In addition, the heat balance can be represented as:

T
T
C p
=2
k
IHV .
t
x
x
(3.5.10)
(3.5.11)
150
Similarly to the convective and intermittent drying described in Sections 3.3 and
3.4, the internal evaporation rate (I), effective vapour diffusivity, tortuosity, solids
concentration and porosity are evaluated using Equations 3.1.19, 3.2.1, 3.2.3, 3.2.4
and 3.2.5, respectively.
The relative activation energy of heat treatment of wood is generated from the drying
run in Case 2 (refer to Table 3.5) (Younsi et al., 2007). The activation energy during
drying is evaluated using Equation (2.1.5) and divided with the equilibrium activation
energy represented in Equation (2.1.7) to yield the relative activation energy as mentioned in Equation (2.1.6). The relationship between the relative activation energy and
average moisture content can be represented by simple mathematical equation obtained
by the least-square method using Microsoft Excel (Microsoft Corp, 2012). The relative
activation energy can be represented as:

E v
= 1 1.517(X X b )0.22 exp 3.717(X X b )3.135 .
E v,b
(3.5.12)
For modelling using the S-REA here, the relative activation energy shown in Equation
(3.5.12) is used, but the average moisture content X in Equation (3.5.12) is substituted
by the local moisture content (X) as the REA represents the local evaporation rate here,
instead of the global drying rate. In order to incorporate the effect of linearly increased
gas temperature, the equilibrium activation energy shown in Equation (2.1.7) implements
the corresponding gas temperature and humidity during heat treatment. In addition, the
linearly increased gas temperature is used in Equation (3.5.8).
and temperature in the heat treatment of wood, the mass and heat balances shown in
Equations (3.5.9) to (3.5.11), in conjunction with the initial and boundary conditions
represented in Equations (3.5.4) to (3.5.8) and the relative activation energy shown by
Equation (3.5.12), are solved by method of lines (Chapra, 2006; Constantinides, 1999).
By this method, the partial differential equations are transformed into a set of ordinary
differential equations with respect to time by firstly discretising the spatial derivatives.
The ordinary differential equations are then solved simultaneously by ode23s in Matlab
(Mathworks Inc., 2012). The spatial derivative here is discretised into 10 increments;
application of 100 increments has been conducted and there is no difference in the profiles
observed, as shown in Figure 3.18. No shrinkage is incorporated in the modelling, as
Younsi et al. (2006b) indicated that the ratio between final and initial dimension is
around 0.96. Similarly, the modelling implemented by Younsi et al. (2006a,b, 2007) did
not incorporate the shrinkage effect.
The average moisture content of wood during heat treatment is evaluated by:
L
X=
X (x)d x
L
(3.5.13)
dx
The profiles of average moisture content and temperature are then validated towards the
experimental data of Younsi et al. (2007).
151
Average moisture content (kg water/kg dry solids)
0.14
S-REA 100 increments
S-REA 10 increments
Data
0.12
0.1
0.08
0.06
0.04
0.02
0
0.5
1.5
t(s)
2.5
3
104
Figure 3.18 Profiles of average moisture content during heat treatment in Case 2 (refer to Table
3.5) solved by the method of lines using 10 and 100 increments.
3.5.2
The results of modelling wood heating using the S-REA

The S-REA is used to model the heat treatment of wood under a constant heating rate and
the results of modelling are presented in Figures 3.19 to 3.27. The REA is implemented to
model the local evaporation or condensation term and coupled with a system of equations
in order to yield the spatial profiles of moisture content, water vapour concentration and
temperature. It is noted that if locally there is no vacant pore space which is connected
to other pores or channels, the internal mass transfer area is zero, hence the REA term
is zero. In this study, the internal mass transfer coefficient (hm,in ) shown in Equation
(3.1.17) is chosen to be 0.001 m s1 . Application of hm,in higher than 0.001 m s1 does
not yield any noticeable differences in the profiles of moisture content and temperature
profiles.
As mentioned before, no method has been presented anywhere in the literature to
measure pore liquid diffusivity, and liquid diffusivity is obtained by numerical sensitivity to match the prediction with the experimental data of moisture content and
temperature. Interestingly, in this study, for all cases the profiles of moisture content and
temperature are independent of the liquid diffusivity value. Further explanation about
this phenomenon is presented next.
For Case 1 (refer to Table 3.5), the results of modelling are presented in Figures 3.19
to 3.22. The varied values of the liquid diffusivity in the range of 1 108 to 1
1030 m2 s1 have been used and there are no noticeable differences in the profiles of
moisture content and temperature as shown in Figures 3.19 and 3.20. This may indicate
0.12
Dw = 1e-8 m2/s
Dw = 1e-12 m2/s
Dw = 1e-18 m2/s
Dw = 1e-24 m2/s
Dw = 1e-30 m2/s
Data
0.1
0.08
0.06
0.04
0.02
t(s)
7
10
Figure 3.19 Effect of liquid diffusivity on profiles of the moisture content during heat treatment in
Case 1 (refer to Table 3.5).
500
152
450
400
350
Dw = 1e-8 m2/s
Dw = 1e-12 m2/s
Dw = 1e-18 m2/s
Dw = 1e-24 m2/s
Dw = 1e-30 m2/s
Data
300
250
4
t(s)
10
Figure 3.20 Effect of liquid diffusivity on profiles of temperature during heat treatment in Case 1
153
0.14
Model
Data
0.12
0.1
0.08
0.06
0.04
0.02
0
t(s)
7
104
Figure 3.21 Profiles of average moisture content during heat treatment in Case 1 (refer to
Table 3.5).
500
Temperature (K)
450
400
350
300
Model
Data
250
0
4
t(s)
7
104
Figure 3.22 Profiles of temperature during heat treatment in Case 1 (refer to Table 3.5).
0.14
154
Model
Data
0.12
0.1
0.08
0.06
0.04
0.02
0
0.5
1.5
t(s)
2.5
3
10
Figure 3.23 Profiles of average moisture content during heat treatment in Case 2 (refer to
Table 3.5).
that the modelling of heat treatment in Case 1 (refer to Table 3.5) is independent of
the liquid diffusivity so this can be neglected and the modelling will only involve the
vapour diffusion and evaporation/condensation. The phenomenon could be due to the
initial moisture content being relatively low and the heating of wood implemented at a
relatively high temperature.
By ignoring the liquid diffusion term on the mass balance of liquid water (refer
to Equation 3.5.9), a good agreement between the predicted and experimental data of
moisture content and temperature is shown in Figures 3.23 and 3.24 as well as confirmed
by R 2 and RMSE indicated in Table 3.6. Benchmarks against modelling implemented
by Younsi et al. (2007) indicate that the S-REA yields comparable or even better results.
Therefore, it can be said that the S-REA models the heat treatment of wood in Case 1
well (refer to Table 3.5).
Figures 3.23 to 3.27 show the results of modelling of heat treatment of wood in Case 2
(refer to Table 3.5). Similarly to Case 1 (refer to Table 3.5), the varied values of the
liquid diffusivity in the range of 1 108 to 1 1030 m2 s1 have been used and there
is no noticeable effect on the profiles of moisture content and temperature. By ignoring
the liquid diffusion term in the mass balance of liquid water (refer to Equation 3.5.9),
a good agreement between the predicted and experimental data of moisture content
and temperature is shown in Figures 3.23 and 3.24 and confirmed by R 2 of 0.995 and
0.997 for moisture content and temperature, respectively and RMSE of lower than 0.005
155
Table 3.6 R 2 and RMSE of modelling of heat treatment of wood under a

constant heating rate using the S-REA.
Case
R 2 for X
R 2 for T
RMSE for X
RMSE for T
1
2
0.988
0.995
0.992
0.997
0.004
0.003
4.765
3.287
500
Temperature (K)
450
400
350
300
Model
Data
250
0.5
1.5
2
t(s)
2.5
3.5
104
Figure 3.24 Profiles of temperature during heat treatment in Case 2 (refer to Table 3.5).
and 3.287 for moisture content and temperature, respectively. The S-REA describes the
profiles of moisture content and temperature very well. Benchmarks against modelling
implemented by Younsi et al. (2007) reveal that the S-REA yields better agreement with
the experimental data from moisture content.
The spatial profiles of moisture content, water vapour concentration and temperature
are presented in Figures 3.253.27. The distribution of moisture content is not determined
by liquid diffusivity. This may suggest that liquid diffusivity can be neglected, so the
drying process is not governed by liquid diffusion as the initial moisture content is
relatively low and temperature is relatively high. Figure 3.25 indicates that the moisture
content of the inner part of the samples is higher than that of the outer part, which
indicates the moisture migrates outwards during drying.
Similarly, as shown in Figure 3.26, the water vapour concentration of the inner part of
the samples is higher than that of the outer part. This could be because of the relatively
0.14
0.12
0.1
0.08
0.06
t = 1000s
t = 3000s
t = 5000s
t = 10 000s
t = 15 000s
t = 20 000s
t = 28 000s
0.04
0.02
0
0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018

Axial position (m)
Figure 3.25 Profiles of spatial moisture content during heat treatment in Case 2 (refer to
Table 3.5).
0.09
0.08
156
0.07
0.06
0.05
0.04
t = 1000s
t = 3000s
t = 5000s
t = 10 000s
t = 15 000s
t = 20 000s
t = 28 000s
0.03
0.02
0.01
0
0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018

Axial position (m)
Figure 3.26 Profiles of spatial water vapour concentration during heat treatment in Case 2 (refer
to Table 3.5).
157
460
440
Temperature (K)
420
400
380
360
t = 1000s
t = 3000s
t = 5000s
t = 10 000s
t = 15 000s
t = 20 000s
t = 28 000s
340
320
300
280
0
0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018

Axial position (m)
Figure 3.27 Profiles of spatial temperature during heat treatment in Case 2 (refer to Table 3.5).
high initial porosity of the samples, which allows evaporation at the core of the samples.
The water vapour seems to migrate outwards and at the surface it is removed by the
gas. The water vapour concentration increases until a heating time of 15 500 s, followed
by a decrease until the end of drying. The initial increase could be because the initial
moisture content is still relatively high at the beginning of drying but as the heating progresses, moisture content decreases and lower water vapour is generated. As shown in
Figure 3.27, the temperature distribution of the samples is essentially uniform, in agreement with the observations of Younsi et al. (2007).
It can be observed that the S-REA can model the heat treatment of wood under
constant heating rate very well for all cases investigated. The liquid diffusion term can
be neglected so that the model for heat treatment of wood under a constant heating rate
may be simplified as follows:
(C s X )
= I,
t

C v
C v
Cv
=
Dv
+
Dv
+ I,
t
x
x
y
dy

T
T
=
k
+
k
IHV .
C p
t
x
x
y
y
(3.5.14)
(3.5.15)
(3.5.16)
The S-REA has also the advantages of yielding profiles of water vapour concentration
during the process. This enables better understanding of the transport phenomena during
158
the process. It can be said that the S-REA is an effective multiphase approach to
modelling the heat treatment of wood.
3.6
The S-REA for the baking of bread

The S-REA is validated against the experimental data of Banooni et al. (2008a,b). For
better understanding of the modelling implemented, the experimental details of Banooni
et al. (2008a,b) are reviewed briefly here. Baking experiments were carried out in a
laboratory oven with active belt length and width of 1 m and 0.72 m, respectively. The
oven was also equipped with an electrical heater and temperature control. The dough was
mixed, divided into pieces of 250 g and kept for 15 min and then shaped and punched
to produce flat bread with an initial thickness of 0.2 cm (Banooni et al., 2008a). The
samples of bread were put in the belt of oven in which heated air was pushed at high
velocity onto their surfaces. The air was propelled by a centrifugal fan and directed onto
the samples through fingers with jet holes of 1.2 cm; its velocity was 110 m s1 .
During the baking, the online system monitored the weight of the samples and Pt-100
probes were used to measure the top and bottom surface temperatures of the samples
(Banooni et al., 2008a,b).
3.6.1
Mathematical modelling of the baking of bread using the S-REA

Here, the S-REA is set up based on the experiments reported by Banooni et al.
(2008a,b). It consists of a set of equations describing the conservation of mass
and heat transfer using the REA to describe the local evaporation/condensation
rate.
The mass balance of water in the liquid phase (liquid water) is written as (Chen, 2007;
Chong and Chen, 1999; Putranto and Chen, 2013; Zhang and Datta, 2004):

(C s X )
(Cs X )
=
Dw
I,
t
x
x
(3.6.1)
where Dw is the effective liquid water diffusivity (m2 s1 ), X is the concentration of

liquid water (kg H2 O kg dry solids1 ), Cs is the solid concentration (kg dry solids m3 ),
which can change if the structure changes, I is the evaporation or condensation rate (kg
H2 O m3 s1 ) and I is >0 when evaporation occurs locally.
The mass balance of water in the vapour phase (water vapour) is expressed as (Chen,
2007; Chong and Chen, 1999; Putranto and Chen, 2013; Zhang and Datta, 2004):
Cv
=
t
x
Cv
Dv
x

+ I,
(3.6.2)
where Dv is the effective vapour diffusivity (m2 s1 ) and Cv is the concentration of liquid
water (kg H2 O kg dry solids1 ).
159
The heat balance is represented by the following equation (Chen, 2007; Chong and
Chen, 1999; Putranto and Chen, 2013; Zhang and Datta, 2004):

T
T
C p
(3.6.3)
=
k
IHV ,
t
x
x
where T is the sample temperature (K), k is thermal conductivity of sample (W m2
K1 ), is the sample density (kg m3 ) and Hv is the vaporisation heat of water
(J kg1 ).
The initial and boundary conditions for Equations (3.6.13.6.3) are:
t = 0, X = X o , C v = Cvo , T = To ,
dX
dCv
dT
x = 0,
= 0,
= 0, k
= U (Tb T ),
dx
dx
dx

dX
Cv,s
x = L , C s Dw
= h m w
v,b ,
dx

Cv,s
dCv
= h m v
v,b ,
Dv
dx
dT
k
= h(Tb T ),
dx
(3.6.4)
(3.6.5)
(3.6.6)
(3.6.7)
(3.6.8)
where h is the top heat transfer coefficient (W m2 K1 ) and U is the overall bottom
heat transfer coefficient (W m2 K1 ).
Similarly to convective and intermittent drying, as well as wood heating under constant
heating rates (described in Sections 3.3, 3.4 and 3.5, respectively), the internal evaporation rate, effective vapour diffusivity, tortuosity, solid concentration and porosity are
evaluated using Equations (3.1.19), (3.2.1), (3.2.3), (3.2.4) and (3.2.5), respectively. In
addition, the internal mass transfer coefficient (hm,in ) described in Section 3.2 is used
here.
The relative activation energy of baking of bread is generated from one accurate
baking run; i.e. baking bread at a baking temperature of 150 C and air velocity of
10 m s1 (Banooni et al., 2008a). The activation energy during drying is evaluated
using Equation (2.1.5) and divided by the equilibrium activation energy represented in
Equation (2.1.7) to yield the relative activation energy as mentioned in Equation (2.1.6).
The relationship between relative activation energy and average moisture content can
be represented by a simple mathematical equation obtained by the least-square method
using Microsoft Excel (Microsoft Corp, 2012). The relative activation energy can be
represented as:
E v
= [1 7.424(X X b )4.471 ] exp[23.884(X X b )42.282 ].
E v,b
(3.6.9)
The good agreement between the fitted and experimental relative activation energy
is shown by R 2 of 0.995. For modelling using the S-REA here, the relative activation energy shown in Equation (3.6.9) is used but the average moisture content X in
Equation (3.6.9) is substituted for the local moisture content (X) as the REA is used
160
to represent the local evaporation rate instead of the overall drying rate of the whole
sample.
The effective liquid diffusivity (Dw ) of the baking of bread presented is expressed as
(Ni et al., 1999):
Dw = 1 106 exp(2.8 + 2X ).
(3.6.10)
and temperature of the convective of mango tissues, the mass and heat balances shown
in Equations (3.6.1)(3.6.3), in conjunction with the initial and boundary conditions
represented in Equations (3.6.4)(3.6.8) and the relative activation energy shown by
Equation (3.6.9), are solved by the method of lines (Chapra, 2006; Constantinides,
1999). In this method, the partial differential equations are transformed into a set of
ordinary differential equations with respect to time by firstly discretising the spatial
derivatives. The ordinary differential equations are then solved simultaneously by ode23s
in Matlab (Mathworks Inc., 2012). The spatial derivative here is discretised into 10
increments; application of 100 increments has been conducted and there is no noticeable
difference in the profiles observed. The shrinkage during the baking process can be
represented as:
V
2
= 162.69 X 207.61 X + 66.925 (for X 0.57),
V0
V
= 1.307 X + 1.015 (for < X 0.57),
V0
(3.6.11)
(3.6.12)
where V is the volume of sample (m3 ) and V0 is the initial volume of sample (m3 ).
The average moisture content of bread baking is evaluated by:
L(t)

X=
X (x)d x
0
L(t)

(3.6.13)
dx
The profiles of average moisture content and centre temperature are then validated
against the experimental data of Banooni et al. (2008a).
3.6.2
The results of modelling of the baking of bread using the S-REA

The S-REA is used to model the baking of bread at baking temperatures of 150
and 200 C. The original formulation of the L-REA is implemented in the partial
differential equation set for transport in porous media, to represent the local evaporation or condensation rate. It is thus coupled with the equations of conservation
to describe the spatial profiles of moisture content, water vapour concentration and
temperature.
The results of modelling the baking of bread using the S-REA are shown in Figures
3.283.32. Figure 3.28 presents the profiles of average moisture content during baking
at a baking temperature of 150 C. The S-REA models the average moisture content well
161
0.65
Data v = 1 m/s
S-REA v = 1 m/s
Data v = 5 m/s
S-REA v = 5 m/s
Data v = 10 m/s
S-REA v = 10 m/s
0.6
0.55
0.5
0.45
0.4
500
1000
1500
t(s)
Figure 3.28 Profiles of average moisture content during the baking of bread at a baking
temperature of 150 C.
at various air velocities and a baking temperature of 150 C as shown in Figure 3.28
(R 2 higher than 0.985).
The spatial profiles of moisture content during the baking of bread at a baking temperature of 150 C and velocity of 10 m s1 are shown in Figure 3.29. The moisture
content at the top and bottom surfaces of the bread are lower than at the core. This may
indicate that the moisture migrates outwards during baking since the surface temperature is higher than the core. The maximum moisture content is located at a particular
position inside the samples. It is not achieved at the centre of the sample, since the top
and bottom surface temperatures are not similar. The top surface temperature is higher
than the bottom surface temperature which may result in a higher moisture content at
the bottom part of the samples. As baking progresses, the moisture content decreases
and should approach equilibrium at the end of baking.
Figure 3.30 shows the spatial profiles of water vapour concentration during the baking
of bread at a baking temperature of 150 C and velocity of 10 m s1 . Initially, the
concentration of water vapour is relatively high and this decreases as baking progresses,
which could be because of the depletion of moisture during the baking. At the bottom
part of the samples, the water vapour concentration is higher than that of the top part of
the samples. It seems that the maximum concentration of water vapour is located at the
bottom surface of the samples. These could be because no mass transfer occurs on the
bottom surfaces since the samples are placed on top of trays, while at the top surfaces,
water vapour is transferred to the baking air.
t = 20s
t = 50s
t = 100s
t = 200s
t = 400s
t = 600s
t = 800s
t = 1000s
t = 1200s
t = 1500s
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.005
0.01
0.015
0.02
Axial position (m)
0.025
0.03
0.035
Figure 3.29 Spatial profiles of moisture content during the baking of bread at a baking
temperature of 150 C and air velocity of 10 m s1 .
0.45
t = 20s
t = 50s
t = 100s
t = 200s
t = 400s
t = 600s
t = 800s
t = 1000s
t = 1200s
t = 1500s
0.4
162
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.005
0.01
0.015
0.02
Axial position (m)
0.025
0.03
0.035
Figure 3.30 Spatial profiles of concentration of water vapour during the baking of bread at a
163
400
390
380
Temperature (K)
370
360
350
340
330
Top surface-data
Top surface-model
Bottom surface-data
Bottom surface-model
320
310
300
0
500
1000
1500
t(s)
Figure 3.31 Profiles of top and bottom surface temperatures during the baking of bread at a
420
400
Temperature (K)
380
t = 20s
t = 50s
t = 100s
t = 200s
t = 400s
t = 600s
t = 800s
t = 1000s
t = 1200s
t = 1500s
360
340
320
300
0
0.005
0.01
0.015
0.02
Axial position (m)
0.025
0.03
0.035
Figure 3.32 Spatial profiles of temperature during the baking of bread at a baking temperature of
150 C and air velocity of 10 m s1 .
164
Figure 3.31 shows the results of modelling the top and bottom surface temperatures
of bread during baking at a baking temperature of 150 C and air velocity of 1 m s1 .
A good agreement with the experimental data is shown (R 2 of 0.965 and 0.961 for
top and bottom surface temperature, respectively). The S-REA models the top surface
temperature well and shows a slight underestimation of the bottom surface temperature
during baking times of 400800 s.
Figure 3.32 shows the spatial profiles of temperature during baking at a baking
temperature of 150 C and velocity of 10 m s1 . The minimum temperature is located
at a particular position inside the samples but it is not at their centre. In addition, the
temperature of the top surface is higher than that at the bottom surface. This is in
agreement with the results of temperature measurement which indicate that the bottom
surface temperature is higher than the top surface. Since the heat received by the samples
from the upper and lower side is not equal, this may result in the minimum temperature
not being located at the centre of the samples. The profiles of temperature are also in
agreement with those of moisture content explained previously in which the maximum
moisture content is not located the centre of the samples.
The study indicates that the S-REA is excellent for modelling the baking of bread.
This provides a good basis for the S-REA modelling changes in quality during baking
of bread in the future.
3.7
Summary
In this chapter, it has been shown that the S-REA is an excellent non-equilibrium
multiphase modelling approach to simulate several challenging cases of drying. The
REA parameters, used in L-REA to describe the global drying rate, are implemented in
S-REA for modelling local evaporationcondensation rate as affected by local variables
and structures of the same material. As mentioned before, the REA parameters can be
generated from one accurate drying run on the materials of concern and in a narrow range
of relevant drying conditions. In S-REA, the REA is coupled with a set of equations of
conservation of heat and mass transfer to yield a spatial model.
The S-REA has been used to model the convective drying, intermittent drying, heat
treatment under linearly increased gas temperatures and baking. The results of modelling match the experimental data well . For modelling intermittent drying and heat
treatment under linearly increased gas temperatures, the equilibrium activation energy is
evaluated according to the corresponding humidity and temperature in each period
of treatment. Without any modification, the S-REA can model the baking process
accurately.
The S-REA can also yield the spatial profiles of concentration of water vapour and
evaporation/condensation, useful for better understanding of drying process. To the best
of our knowledge so far, the S-REA is the first model proposed and used to describe the
local evaporationcondensation rate explicitly.
165
References
Audu, T.O.K. and Jeffreys, G.V., 1975. The drying of drops of particulate slurries. Trans IChemE
Part A. 53, 165175.
Aversa, M., Curcio, S., Calabro, V. and Iorio, G., 2010. Transport phenomena modelling
during drying of shrinking materials. Computer Aided Chemical Engineering 28, 91
96.
Azzouz, S., Guizani, A., Jomma, W. and Belghith, A., 2002. Moisture diffusivity and drying kinetic equation of convective drying of grapes. Journal of Food Engineering 55, 323
330.
Banooni, S., Hosseinalipour, S.M., Mujumdar, A.S., Taheran, E., Bahiraei, M. and Taherkhani,
P., 2008a. Baking of flat bread in an impingement oven: an experimental study of heat transfer
and quality aspects. Drying Technology 26, 902909.
Banooni, S., Hosseinalipour, S.M., Mujumdar, A.S., Taheran, E. and Mashaiekhi, M., 2008b.
Impingement heat transfer effects on baking of flat bread. Drying Technology 26, 910
919.
Bird, R.B., Stewart, W.E. and Lightfoot, E.N., 2002. Transport Phenomena, 2nd international ed.
John Wiley & Sons, Inc., New York.
Chapra, S.C., 2006. Numerical Methods for Engineers. McGraw-Hill, Boston.
Chen, X.D., 2007. Moisture diffusivity in food and biological materials. Drying Technology 25,
12031213.
Chen, X.D. and Lin, S.X.Q., 2005. Air drying of milk droplet under constant and time dependent
Chen, X.D. and Mujumdar, A.S., 2008. Drying Technologies in Food Processing. Blackwell
Publishing, Oxford.
Chen, X.D., Pirini, W. and Ozilgen, M., 2001. The reaction engineering approach to modelling
drying of thin layer pulped kiwifruit flesh under conditions of small Biot numbers. Chemical
Engineering and Processing 40, 311320.
Chen, X.D. and Xie, G.Z., 1997. Fingerprints of the drying behaviour of particulate or thin layer
food materials established using a reaction engineering model. Trans. IChemE, Part C: Food
and Bioproducts Processing 75, 213222.
41654178.
Constantinides, A., 1999. Numerical Methods for Chemical Engineers with MATLAB Applications.
Prentice Hall, Upper Saddle River, NJ.
Datta, A.K., 2007. Porous media approaches to studying simultaneous heat and mass transfer in food processes. I: Problem formulations. Journal of Food Engineering 80, 80
95.
Gimmi, T, Fuhler, H., Studer, B. and Rasmuson, A., 1993. Transport of volatile chlorinated
hydrocarbons in unsaturated aggregated media. Water, Air and Soil Pollution 68, 291
305.
166
Kar, S., 2008. Drying of Porcine Skin Theoretical Investigations and Experiments. Ph.D. thesis.
Kar, S. and Chen, X.D., 2010. Moisture transport across porcine skin: Experiments and implementation of diffusion-based models. International Journal of Healthcare Technology and
Kar, S. and Chen, X.D., 2011. Modelling of moisture transport across porcine skin using reaction
engineering approach and examination of feasibility of the two phase approach. Chemical
Engineering Communication 198, 847885.
Kocaefe, D., Charette, A., Ferland, J., Couderc, P. and Saint-Romain, J.L., 1990. A kinetic study
of pyrolysis in pitch impregnated electrodes. The Canadian Journal of Chemical Engineering
68, 988996.
Kocaefe, D., Younsi, R., Poncsak, S. and Kocaefe, Y., 2007. Comparison of different models for
the high-temperature heat-treatment of wood. International Journal of Thermal Sciences 46,
707716.
Lin, S.X.Q. and Chen, X.D., 2005. Prediction of air drying of milk droplet under relatively high
humidity using the reaction engineering approach. Drying Technology 23, 13951406.
Lin, S.X.Q. and Chen, X.D., 2007. The reaction engineering approach to modelling the cream and
whey protein concentrate droplet drying. Chemical Engineering and Processing 46, 437443.
Madiouli, J., Lecomte, D., Nganya, T., Chavez, S., Sghaier, J. and Sammouda, H., 2007. A method
for determination of porosity change from shrinkage curves of deformable Materials. Drying
Technology 25, 621628.
Mariani, V.C., de Lima, A.G.B. and Coelho, L.S.. 2008. Apparent thermal diffusivity estimation
of the banana during drying using inverse method. Journal of Food Engineering 85, 569579.
Mathworks, Inc., Website, www.mathworks.com, 2012 (accessed 21 November, 2012).
Microsoft Corp., http://office.microsoft.com/en-au/excel/ 2012 (accessed 21 November, 2012).
Ni, H., Datta, A.K. and Torrance, K.E., 1999. Moisture transport in intensive microwave heating of
biomaterials multiphase porous media model. International Journal of Heat and Mass Transfer
42, 15011512.
Pakowski, Z.. and Adamski, A., 2007. The comparison of two models of convective drying of
shrinking materials using apple tissue as an example. Drying Technology 25, 11391147.
Putranto, A., Chen, X.D. and Webley, P.A., 2010a. Infrared and Convective Drying of thin layer of
Chemical Engineering and Processing: Process Intensification 49, 348357.
Putranto, A., Chen, X.D. and Webley, P.A., 2010b. Application of the reaction engineering approach
(REA) to model cyclic drying of polyvinyl alcohol (PVA)/glycerol/water mixture. Chemical
29, 961973.
Putranto, A, Xiao, Z., Chen, X.D. and Webley, P.A., 2011b. Intermittent drying of mango tissues:
implementation of the reaction engineering approach (REA). Industrial Engineering Chemistry
Research 50, 10891098.
167
Putranto, A., Chen, X.D., Devahastin, S., Xiao, Z. and Webley, P.A., 2011c. Application of the
reaction engineering approach (REA) to model intermittent drying under time-varying humidity
Putranto, A., Chen, X.D., Xiao, Z. and Webley, P.A., 2011d. Modelling of high-temperature
102, 62146220.
Putranto, A., Chen, X.D. and Zhou, W., 2011e. Modelling of baking of cake using the reaction
engineering approach (REA). Journal of Food Engineering 105, 306311.
Roberts, J.S., Tong, C.H. and Lund, D.B., 2002. Drying kinetics and time-temperature distribution
of pregelatinized bread. Journal of Food Science 67, 10801087.
Sanjuan, N., Lozano, M., Garcia-Pascual, P. and Mulet, A., 2004. Dehydration kinetics of red
pepper (Capsicum annuum L. var Jaranda). Journal of the Science of Food and Agriculture 83,
697701.
Scarpa, D. and Milano, G., 2002. The role of adsorption and phase change phenomena in the thermophysical characterization of moist porous materials. International Journal of Thermophysics
23, 10331046.
Slattery, J.C. and Bird, R.B., 1958. Calculation of the diffusion coefficient of dilute gases and of
the self-diffusion coefficient of dense gases. AIChE Journal 4, 137142.
Srikiatden, J. and Roberts, J.S., 2006. Measuring moisture diffusivity of potato and carrot (core
and cortex) during convective hot air and isothermal drying. Journal of Food Engineering 74,
143152.
Srikiatden, J. and Roberts, J.S., 2008. Predicting moisture profiles in potato and carrot during
convective hot air drying using isothermally measured effective diffusivity. Journal of Food
Engineering 84, 516525
Thuwapanichayanan, R., Prachayawarakorn, S. and Soponronnarit, S., 2008. Modelling of diffusion with shrinkage and quality investigation of banana foam mat drying. Drying Technology
26, 13261333.
Van der Sman, R.G.M., 2003. Simple model for estimating heat and mass transfer in regular-shaped
foods. Journal of Food Engineering 60, 383390.
Van der Sman, R.G.M., 2007a. Moisture transport during cooking of meat: An analysis based on
FloryRehner theory. Meat Science 76, 730738.
Van der Sman, R.G.M., 2007b. Soft condensed matter on moisture transport in cooking of meat.
AIChE Journal 53, 29862995.
Van der Sman, R.G.M., Jin, X. and Meinder, M.B.J., 2012. A paradigm shift of drying of food
materials via free volume concepts. Proceedings of the 18th International Drying Symposium
(IDS 2012), Xiamen, China (1115 September 2012).
Design 87, 885898.
Vrentas, J. and Duda, J., 1977. Diffusion in polymer-solvent systems. I. Reexamination of
the free volume theory. Journal of Polymer Science: Polymer Physics Edition 15, 403
416.
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2006a.Transient multiphase model for the
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2006b. Thermal modelling of the high
168
27, 14241431.
Zhang, J. and Datta, A.K., 2004. Some considerations in modelling of moisture transport in heating
of hygroscopic materials. Drying Technology 22, 19832008.
Zhang, J. and Datta, A.K., 2006. Mathematical modelling of bread baking process. Journal of
Food Engineering 75, 7889.
Zhang, J., Datta, A.K. and Mukherjee, S., 2005. Transport processes and large deformation during
baking of bread. AIChE Journal 51, 25692580.
Comparisons of the REA with

Fickian-type drying theories, Luikovs
and Whitakers approaches
4.1
Model formulation
Ficks law is one of the most well-known concepts of mass transfer in the literature. Its
applications in many different physical circumstances are substantial. In his book on
diffusion, Cussler vividly introduced the story of how Fick came up with the law named
after him (Cussler, 1984). Through an impressive combination of qualitative theories,
casual analogies, and quantitative experiments as presented by Cussler, Adolf Fick in
1855, described that:
[T]he diffusion of the dissolved material . . . is left completely to the influence of the molecular
forces basic to the same law . . . for the spreading of warmth in a conductor and which has already
been applied with such great success to the spreading of electricity.
Basically, diffusion can be described using the same mathematics as that of Fouriers law
of heat conduction or that of Ohms law of electrical conduction. In reality, of course,
diffusion is a mass transfer process that involves a dynamic molecular process. Parallel
to the work on heat transfer, and Fouriers law in 1822, Fick defined a one-dimensional
flux of mass J:
J = AD
c1
,
z
(4.1.1)
where J takes the units of kg m2 s1 , A is the area across which diffusion of mass
occurs, c1 is the concentration of the species of concern (kg m3 ) and z is distance (m).
The quantity D was called, the constant depending on the nature of the substances by
Fick. This quantity D is in fact the diffusion coefficient (m2 s1 ). Although there are a
few formats of Ficks law that are known in literature, their essence remains similar. The
concept has been applied to the conductive mass transfer processes in gases, liquids
and solids.
Drying of porous material has been considered to be one of the most suitable candidates for applying Ficks law of diffusion. In particular, when one is interested in knowing
what moisture content distribution is like within a material being dried, in many cases
this cannot be accurately measured. Because of the complexity of the process and the
material involved, a pure application of Ficks law is not possible. The mass diffusivity
for moisture in a porous material is taken as an effective diffusivity to encapsulate
several effects in addition to the pure diffusion process.
170
Spatial drying modelling is important since the spatial distributions of moisture content and/or temperature affect the product quality of materials being dried (Huang et al.,
2009; Li et al., 1999; Mrad et al., 2012). For food materials, drying was shown to influence the loss of ascorbic acid, volatiles, aroma and carotenoids (Di Scala and Crapiste,
2008; Mrad et al., 2012; Ramallo and Mascheroni, 2012; Timoumi et al., 2007). Degradation of ascorbic and carotenoids was shown to be more enhanced with the increase
of drying air temperature and product moisture content (Di Scala and Crapiste, 2008).
The rehydration ability and ascorbic acid retention were dependent on the drying air
temperature (Ramallo and Mascheroni, 2012). Similarly, the increase of temperature
enhanced the loss of aroma (Timoumi et al., 2007). The survival of probiotics was also
found to be very dependent on the temperature and moisture content (Huang et al.,
2009). Drying was shown to induce fissuring of rice as a result of a moisture content
high gradient inside the samples (Yang et al., 2003). By predicting the distributions of
moisture content and temperature inside the materials being dried, product quality can
be predicted. Several drying operating conditions and schemes can be made in order to
maintain the product quality during drying (Chen, 2007).
For non-food materials, drying was shown to induce cracking of kaolin samples.
Convective drying resulted in cracking at the top of cylindrical samples where the tensile stress is at maximum, while microwave drying was damaged internally as a result
of high pore pressure (Kowalski et al., 2005). Convective drying of kaolin did not
result in sample cracking but applying a combination of convection and microwaves
led to fracture of the samples (Kowalski and Pawlowski, 2010a,b). Controlled
drying can give a desirable bending strength and sintered density of ceramic samples
(Misra et al., 2002). Here, the distributions of moisture content and temperature may
affect the local stress formation and the physicochemical states of the materials (Chen,
2007).
The effective diffusion has been considered to be a fundamental mechanism of moisture transport in literature (Mariani et al., 2008; Pakowski and Adamski, 2007; Thuwapanichayanan et al., 2008; Vaquiro et al., 2009). The effective liquid diffusivity is usually
used to lump the whole phenomenon during drying including liquid diffusion, vapour
diffusion, Darcys flow, capillary flow and evaporation/condensation (Mariani et al.,
2008; Pakowski and Adamski, 2007; Thuwapanichayanan et al., 2008; Vaquiro et al.,
2009).
For one-dimensional convective drying of a slab, the simplest (and indeed typical)
mathematical model of moisture diffusion is expressed as:
C
=
t
x
C
Deff
x

,
(4.1.2)
where C can be considered to be concentration of liquid water (kg m3 ), Deff is the

effective diffusivity (m2 s1 ), t is time (s) and x is distance (m). Equation (4.1.2) does
not usually consider the shrinkage velocity effect and the most important parameter Deff
needs to be determined experimentally.
Comparisons of the REA with other theories
4.1.1
171
Cranks effective diffusion

Cranks effective diffusion has been used by several researchers and considered as a
fundamental drying model (Arslan and Ozcan, 2011; Cihan and Ece, 2001; Corzo et al.,
2008; Crank, 1975; Hassini et al., 2007). However, these studies usually neglect the Biot
number (Bi) criteria and do not fully satisfy the required boundary conditions. They
often do not report the operating conditions of their experiments in detail so that clear
justification for the requiring validity of Cranks effective diffusion approach (Crank,
1975) cannot be made (Chen, 2007). For slab geometry, the solution of Cranks effective
diffusion in Equation (4.1.1) can be written as (Crank, 1975):

2
X Xe
1
8
2
= 2
exp (2n + 1) 2 Deff ,l t ,
(4.1.3)
X0 Xe
n=0 (2n + 1)2
L
where Deff is the effective liquid diffusivity (m2 s1 ), X is the moisture content on a
dry basis (kg water kg dry solids1 ), which can be evaluated by (X = C/ s ), X0 is
the initial moisture content (kg water kg dry solids1 ), Xe is the equilibrium moisture
content (kg water kg dry solids1 ) and L is the half thickness (m) where L = 0.5b and
the materials are dried symmetrically. For long time-period of drying, only the first term
of Equation (4.1.2) is significant so that Equation (4.1.2) can be simplified into (Crank,
1975):

2

8
X Xe
ln
(4.1.4)
D eff ,l t.
ln
X0 Xe
2
4b2
Cranks effective diffusion should only be valid for the conditions of isothermal drying,
negligible shrinkage, negligible external resistance, constant diffusivity and uniform
initial moisture content (Crank, 1975). In addition, the approach should only correlate
well with the experimental data for the time towards the end of drying (Srikatden and
Roberts, 2006). However, this has been implemented largely where the assumptions may
not be fulfilled and justifications are not made (Arslan and Ozcan, 2011; Cihan and Ece,
2001; Corzo et al., 2008; Hassini et al., 2007).
For instance, Cranks effective diffusion theory is based on the surface boundary
conditions (at x = L, for slab geometry) of (Crank, 1975):
Deff ,l
dX
= (X s X e ),
dx
(4.1.5)
where Xs is the surface moisture (liquid) content (kg water kg dry solids1 ) and
is a kind of mass transfer coefficient. Equation (4.1.5) is in contrast to the boundary
conditions of vapour transfer, which can be expressed as (Chen, 2007):
Dv,eff
dCv
= h m (v,s v,b ),
dx
(4.1.6)
where Dv,eff is the effective vapour diffusivity (m2 s1 ), hm is the mass transfer coefficient
in the conventional sense (m2 s1 ), Cv is the concentration of water vapour (kg m3 ),
v,s is the surface water vapour concentration (kg m3 ) and v,b is the water vapour
concentration in the drying air.
172
Equations (4.1.5) and (4.1.6) are similar if, the moisture content is in equilibrium with
the water vapour concentration only at the interface (Chen, 2007). However, previous
publications assume this without the necessary justifications (Arslan and Ozcan, 2011;
Castell-Palou, 2011; Cihan and Ece, 2001; Corzo et al., 2008; Hassini et al., 2007).
Several researchers (Corzo et al., 2008; Doymaz, 2004; Kaya et al., 2007) also correlate
effective diffusivity with drying air temperature, which is also fundamentally incorrect
unless the material being dried has a temperature similar to that of the air (Chen, 2007).
This is in contrast with Srikiatden and Roberts (2006), who inventively set up isothermal
test rigs to conduct isothermal drying carefully, so that the effective diffusivity could be
correlated with the sample temperature.
4.1.2
The formulation of effective diffusivity to represent complex

drying mechanisms
Modelling of drying processes is relatively complex and there have been several mechanistic models proposed; including liquid diffusion (Lewis, 1921), capillary flow (Buckingham, 1907), evaporation condensation (Henry, 1939), the Luikov approach (Luikov,
1975) and the Whitaker approach (Whitaker, 1977). Luikovs approach (Luikov, 1975)
assumes the thermal and moisture potential gradient within a porous body cause the
vapour and liquid water transfer so that the flux of liquid water and water vapour is
proportional to the thermal gradient and moisture potential gradient. As mentioned
earlier, the coefficients of effective water liquid diffusivity, effective water vapour diffusivity and thermal diffusivity are implemented in linking the fluxes and gradients.
Whitakers approach (Whitaker, 1977) implements the continuity equation of the liquid
and vapour phases combined with the equation for conservation of liquid water, water
vapour and energy to describe the drying process (Whitaker and Chou, 1983). The pore
network approach has been proposed but it has not been rigorously experimentally validated (Nowicki et al., 1992; Prat, 1993). The applicability of the approach is limited
by geometries and distributions of pore structure and network, especially for drying of
food and biomaterials (Chen, 2007). The complex models mentioned here also require
a lot of constants, which need to be established from several sets of drying experiments
(Chen, 2007).
Among these, there is little question that the effective liquid diffusion model is the
simplest model proposed (Chen, 2007). The effective diffusivity should be influenced
by the composition of the materials. The multi-component diffusion model is often
implemented to couple this effect (Ferrari et al., 1989; Yoshida and Miyasita, 2002).
Therefore, the effective diffusivity is essentially a lumped parameter whose variability is
dependent on the drying condition, material structure and composition and, sometimes,
sample size. The last aspect rules out the most fundamental nature of effective liquid
diffusivity. For moderate drying conditions, it would be better to see effective diffusivity
as a liquid depletion coefficient in order to avoid confusion with the original meaning of
diffusivity (Chen, 2007).
4.1.3
173
Several diffusion-based models

Various formulations of diffusion-based models have been implemented extensively by
many researchers (Adhikari et al., 2004; Cihan and Ece, 2001; Corzo et al., 2008; Kar
et al., 2009; Vaquiro et al., 2009). As mentioned before, Cranks diffusion model has
been applied to model convective drying (Arslan and Ozcan, 2011; Cihan and Ece,
2001; Corzo et al., 2008; Crank, 1975; Hassini et al., 2007). The diffusion-based models
are usually implemented as mass balance coupled with heat balance (Adhikari et al.,
2004; Guine, 2008; Kar et al., 2009; Pakowski and Adamski, 2007). As an example,
Kar et al. (2009) used the diffusion-based model coupled with heat balance to model the
convective drying of porcine skin (see Figures 4.1 and 4.2 for experimental setup).
The skin was very thin (approximately 200 m). The diffusion-based models can be
represented as (Kar et al., 2009):
X
=
Cs
t
x
X
Deff ,l Cs
x

,
(4.1.7)
where Cs is the solids concentration (kg m3 ), X is the moisture content on a dry basis
(kg water kg dry basis1 ), x is the axial position (m), t is time (s) and Deff,l is the effective
liquid diffusivity (m2 s1 ).
The initial and boundary conditions (Kar et al., 2009):
t = 0, X 0 ,
x = 0,
x = L , Cs Deff ,l
dX
= 0,
dx
dX
= h m (v,s v,b ),
dx
(4.1.8)
(4.1.9)
(4.1.10)
where L is the sample thickness, hm is the mass transfer coefficient (m s1 ), v,s is

the surface water vapour concentration (kg m3 ) and v,b is the ambient water vapour
concentration (kg m3 ).
Since the temperature inside the sample is essentially uniform (Chen and Peng, 2005;
Kar, 2008; Kar et al., 2009), the heat balance can be written as (Kar et al., 2009):
mC p
dT
= h u A(Tb T ) + U A(Tb T ) h m A(v,s v,b )HV , (4.1.11)
dt
where m is the sample mass (kg), T is sample temperature (K), Cp is the specific heat
of the sample (J kg1 K1 ), A is the surface area (m2 ), hu is the upper heat transfer
coefficient (W m2 K1 ), U is the overall bottom heat transfer coefficient (W m2 K1 ),
Tb is the ambient temperature (K) and HV is the vaporisation heat of water (J kg1 ).
It is noted that shrinkage was also incorporated in the modelling (Kar, 2008; Kar et al.,
2009).
174
K2
K1
H2
S3
H1
P
PV1
PV2
TC2
TC1
Power
board
Computer
V-6
E-5
E-4
I-8
I-7
V-5
H1-Heater 1
H2-Heater 2
P-Sample stand and platform assembly
B-Micro balance
TC1-Temperature controller 1
TC2-Temperature controller 2
PV1-Rheostat 1
PV2-Rheostat 2
V-6-Stop valve
E-4-Drierite packed bed column
V-5-Relief valve
E-5-Air filter
I-8-Pressure regulator
I-7-Digital flowmeter
K1-Type K thermocouple 1
K2-Type K thermocouple 2
Electrical line
Data signal line
Air line
Figure 4.1 Experimental setup for convective drying of porcine skin. [Reprinted from Chemical
Engineering Research and Design, 87, S. Kar, X.D. Chen, B.P. Adhikari and S.X.Q. Lin, The
impact of various drying kinetics models on the prediction of sample temperaturetime and
moisture contenttime profiles during moisture removal from stratum corneum, 739755,
However, as other examples, several researchers (Batista et al., 2007; Garcia-Perez

et al., 2009; Loulou et al., 2006; Viollaz and Rovedo, 2002) implemented the diffusionbased model without coupling with heat balance. Garcia-Perez et al. (2009) implemented
the diffusion-based model for ultrasonic-assisted drying of cube samples, which only
(a)
175
Air flow
direction
33 mm
11.30 mm
7.16 mm
(b)
Drying channel
Cardboard
Top sample surface
Sample slot
Two plastic
layers
Drying air
Support
Aluminium plate
(bottom surface)
Electronic balance
Figure 4.2 (a) Overview of a sample/plate assembly for convective drying of porcine skin. (b)
Detailed of layering structure of sample support. [Reprinted from Chemical Engineering

Research and Design, 87, S. Kar, X.D. Chen, B.P. Adhikari and S.X.Q. Lin, The impact of
various drying kinetics models on the prediction of sample temperaturetime and moisture
contenttime profiles during moisture removal from stratum corneum, 739755, Copyright
consists of the mass balance. The model can be written as:

2

Wp
Wp
Wp
2Wp
,
= De
+
+
t
x2
y2
z 2
(4.1.12)
where Wp is the moisture content on a dry basis (kg water kg dry solids1 ), De is the
effective diffusivity (m2 s1 ), and x, y and z are the axial positions (m).
176
The initial and boundary conditions are (Garcia-Perez et al., 2009):

t = 0, W p = W p0 ,
x = 0, y = 0, z = 0,
dW p
dW p
dW p
= 0,
= 0,
= 0,
dx
dy
dz
x = L , y = L , z = L , W p = W pe ,
(4.1.13)
(4.1.14)
(4.1.15)
where L is the sample thickness (m) and Wpe is the equilibrium moisture content (kg water
kg dry solids1 ). It is noted that the external resistance and shrinkage were ignored in
the modelling (Garcia-Perez et al., 2009).
Garcia-Perez et al. (2011) implemented the diffusion-based model with and without
external resistance for ultrasonic-assisted drying of a cylindrical sample. The model can
be expressed as:
2

1 Wp
2Wp
Wp
Wp
+
= De
+
,
(4.1.16)
t
x2
r r
r 2
where r is the radial position (m), with the initial and boundary conditions (Garcia-Perez
et al., 2011):
t = 0, W p = W p0 ,
(4.1.17)
x = 0,
dW p
= 0,
dx
(4.1.18)
r = 0,
dW p
= 0.
dr
(4.1.19)
For the case where external resistance is present, the surface boundary condition can be
written as (Garcia-Perez et al., 2011):
x = L, De
dW p
= k(e air ),
dx
(4.1.20)
r = R, De
dW p
= k(e air ),
dr
(4.1.21)
where e is the activity, air is the relative humidity in the drying air and k is the mass
transfer coefficient (kg m2 s1 ).
By ignoring external resistance, the surface boundary condition can be expressed as
(Garcia-Perez et al., 2011):
x = L, W p = W pe ,
(4.1.22)
r = R, W p = W pe
(4.1.23)
It is noted that Garcia-Perez et al. (2011) did not couple the diffusion-based model with
heat balance. It is noted that the shrinkage was also not incorporated in the modelling
(Garcia-Perez et al., 2011). The results of modelling indicated that the diffusion-based
model which incorporated the external resistance resulted in better agreement with
experimental data (Garcia-Perez et al., 2011).
177
More issues with boundary conditions of mass balance also occur (Chen, 2007;
Zhang and Datta, 2004) and the controversies of the boundary conditions are elaborated
in more detail in Section 4.2. Zhang and Datta (2004) and Chen (2007) also highlighted
the importance of applying the multiphase drying approach and the use of the local
evaporationcondensation rate. The detailed explanation and several issues with the
local evaporationcondensation rate are discussed in Section 4.3.
4.2
Boundary conditions controversies

For a better understanding of transport phenomena during a drying process, a multiphase
approach of drying models should be applied. It consists of mass balance of water in
liquid and vapour phases as well as heat balance. By this approach, the spatial profiles of
moisture content, concentration of water vapour and temperature can be generated. For
the multiphase approach, which does not use the source and depletion term, the model
for symmetrical convective drying of a slab can be written as (Chen, 2007; Zhang and
Datta, 2004) follows:
The mass balance of liquid water:

(Cs X )
(Cs X )
=
Dw
;
(4.2.1)
t
x
x
The mass balance of water vapour:
Cv
=
t
x
The heat balance:

C p
C v
x
Dv
T
=
t
x

k
T
x

;
(4.2.2)
(4.2.3)
where Cs is the solid concentration (kg solids m3 ), X is the moisture content (kg water
kg dry solids1 ), Cv is the water vapour concentration (kg m3 ), T is temperature (K),
Dw is the capillary diffusivity (m2 s1 ), Dv is the effective vapour diffusivity (m2 s1 ),
t is time (s), x is the axial dimension (m), is the sample density (kg m3 ), Cp is the
sample specific heat (J kg1 K1 ) and k is thermal conductivity (W m2 K1 ).
The initial conditions of Equations (4.2.1)(4.2.3) are (Chen, 2007; Zhang and Datta,
2004):
t = 0, X = Xo , Cv = Cv o, T = To (initial condition, uniform initial
concentrations and temperature), (4.2.4)
x = 0,
X
x
C v
x
(4.2.5)
(4.2.6)
178
T
x
X
= 0 (no liquid water transfer),
x = L , Cs Dw
x
Dv
C v
= h m (v,s v,b ) (convective boundary for water
x
vapor transfer),
(4.2.7)
(4.2.8)
(4.2.9)
T
= h(Tb T ) HV h m (v,s v,b ) (convective boundary for
x
heat transfer with
vaporization heat of water), (4.2.10)
where L is the sample at half thickness, Tb is drying air temperature (K), hm is the mass
transfer coefficient (m s1 ), h is the heat transfer coefficient (W m1 K1 ), HV is
the vaporisation heat of water (J kg1 ), v,s is the surface water vapour concentration
(kg m3 ) and v,b is the water vapour concentration at the drying medium (kg m3 ).
From Equations (4.2.1)(4.2.3), it can be observed that there is no interaction among
the liquid water and water vapour, apart from the effective diffusivity of water vapour
which should be a function of porosity, dependent on the moisture content. In addition,
the boundary conditions indicate that, at the interface, the vapour diffusive transport
inside the samples is balanced by the convective water vapour. Therefore, the equilibrium
relationship between the moisture content and concentration of water vapour has to be
implemented at the boundary (Chen, 2007).
However, if Equations (4.2.1)(4.2.3) are solved simultaneously with the initial and
boundary conditions shown in Equations (4.2.4)(4.2.9), the rate of change of average
moisture content would be zero, which means no drying occurs. This is not reasonable.
In order to make this model work well, Equation (4.2.2) needs to be removed so that
only the mass balance of liquid water and heat balance are implemented. The model can
then be simplified into (Chen, 2007):

(Cs X )
(Cs X )
=
Dw
,
(4.2.11)
t
x
x

T
T
C p
=
k
.
(4.2.12)
t
x
x
The initial and boundary conditions are (Chen, 2007):
t = 0, X = X, T = To (initial condition, uniform initial
x = 0,
X
x
T
x
(4.2.13)
(4.2.14)
(4.2.15)
x = L, Cs Dw
X
= h m (v,s v,b ) (convective boundary for
x
liquid water transfer),
179
(4.2.16)
T
= h(Tb T ) HV h m (v,s v,b ) (convective boundary for heat
x
transfer with vaporization).
heat of water).
(4.2.17)
Solving Equations (4.2.11) and (4.2.12) simultaneously with the initial and boundary
conditions shown in Equations (4.2.13)(4.2.17) results in the spatial profiles of moisture
content and temperature. However, no profiles of water vapour concentration can be
generated. Equation (4.2.15) indicates that the most water evaporation occurs at the
surface. Similarly, Equation (4.2.17) indicates that the surface receives the largest thermal
impact from heat of evaporation. This may undermine the predictions of moisture content
distribution and temperature inside the samples (Chen, 2007).
4.3
A diffusion-based model with local evaporation rate

Due to the controversies of boundary conditions explained previously, it may be more
appropriate to implement the multiphase drying approach with the local evaporation
rate as pointed out by Chen (2007), Datta (2007) and Zhang and Datta (2004). The
local evaporation rate is positive when drying occurs. It needs to be coupled with mass
balance in liquid and vapour phases, as well as balance. It serves as a depletion and
source term for the mass balance of liquid water and water vapour, respectively. For the
symmetrical convective drying of a slab, the model can be written as (Model A) (Chen,
2007; Chong and Chen, 1999; Kar and Chen, 2011; Putranto and Chen, 2013; Zhang
and Datta, 2004):
The mass balance of liquid water:

(C s X )
(C s X )
=
Dw
I;
(4.3.1)
t
x
x
The mass balance of water vapour:
C v
=
t
x
The heat balance:
C p
T
=
t
x

Dv
Cv
x

+ I;

T
k
IHv ;
x
(4.3.2)
(4.3.3)
where Cs is the solid concentration (kg solids m3 ), X is the moisture content (kg water
kg dry solids1 ), Cv is the concentration of water vapour (kg m3 ), T is temperature (K),
Dw is the capillary diffusivity (m2 s1 ), Dv is the effective vapour diffusivity (m2 s1 ), t is
time (s), x is the axial dimension (m), is the sample density (kg m3 ), k is the thermal
180
conductivity (W m2 K1 ), Cp is the samples specific heat (J kg1 K1 ) and HV is the

vaporisation heat of water (J kg1 ).
The initial and boundary conditions (Chen, 2007; Chong and Chen, 1999; Kar and
Chen, 2011; Putranto and Chen, 2013; Zhang and Datta, 2004):
t = 0, X = X o , Cv = Cvo , T = To (initial condition, uniform initial
x = 0,
X
x
C v
x
(4.3.4)
(4.3.5)
(4.3.6)
T
(4.3.7)
x

Cv,s
X
(convective boundary
x = L , u C s Dw
= h m w
v,b
x
for liquid transfer), (4.3.8)

C v
Cv,s
Dv
= h m v
v,b
(convective boundary for
x
vapor transfer),
(4.3.9)
T
(4.3.10)
x
where w and v are the fraction of surface area covered by liquid water and water vapour,
respectively.
It can be observed that Equations (4.3.1)(4.3.3) require the local evaporation rate to
be expressed explicitly. However, it has not been fully understood experimentally how
to establish this until the REA approach is implemented (Chen, 2007).
k
4.3.1
Problems in determining the local evaporation rate

Several researchers (Lu et al., 1998; Sablani et al., 1998; Sahin and Dincer, 2002;
Srikiaden and Roberts, 2006) use the moisture content rate during drying as the local
evaporation rate so that for the symmetrical convective drying of a slab, the model can
be written as (Model B):
The mass balance:

M
M
=
DM
;
(4.3.11)
t
x
x
The heat balance:
C p
=
t
x

k
T
x

s
M
Hv ;
t
(4.3.12)
181
where M is the moisture content (kg water kg dry solids1 ), s is the density of dry
sample (kg m3 ), DM is the effective diffusivity (m2 s1 ), T is temperature (K), t is time
(s), x is the axial dimension (m), is the sample density (kg m3 ), k is the thermal
conductivity (J kg1 K1 ), Cp is the sample specific heat (J kg1 K1 ) and Hv is the
vaporisation heat of water (J kg1 ).
The initial and boundary conditions:
t =o , M = Mo , T = To (initial conditions, uniform initial moisture
content and temperature),
x = 0,
M
x
(4.3.13)
(4.3.14)
T
= h(Tb T ) (convective boundary for liquid transfer). (4.3.15)
x
However, Zhang and Datta (2004) and Datta (2007) mentioned that the model with
a rate of moisture content used as the local evaporation rate does not satisfy mass
conservation and it is more of an empirical model. Zhang and Datta (2004) analysed
that the combination of Equations (4.3.1) and (4.3.2) in Model B results in:

Cv
(Cs X )
Cv
(Cs X )
=
Dv
Dw
+ 2 I.
(4.3.16)
t
t
x
x
x
x
x = L, k
It can be observed that the local evaporation rate should be influenced by the moisture
content as well as water vapour concentration, not only by the rate of moisture content
as implemented in Model B. Therefore, Model B does not satisfy the conservation of
mass for water.
Moreover, Model B is found to be an empirical model as it lumps the whole
phenomenon (liquid diffusion, vapour diffusion, evaporationcondensation) into diffusion marked by effective diffusivity (DM ) shown by Equation (4.3.11). The combination
of Equations (4.3.1) and (4.3.2) of Model A yields:

Cv
Cv
(C s X )
(Cs X )
+
=
Dv
+
Dw
,
(4.3.17)
t
t
x
x
x
x
while Model B represents the mass balance as shown in Equation (4.3.11), which means
Model B assumes:
C v
(C s X )
M
=
+
,
(4.3.18)
t
t
t

DM
M
x

=

Dv
C v
x

+
(C s X )
Dw
.
x
x
(4.3.19)
Model B uses an effective diffusivity (DM ) to substitute the effective vapour diffusivity
(Dv ) and capillary diffusivity (Dw ). Similarly, Model B represents the concentration of
water vapour (Cv ) and moisture content (X) as M. Therefore, Model B is an empiric
model which looks like a fundamental model of diffusion. This is in agreement with
Chen (2007), who mentioned that effective diffusivity should be treated as a liquid
depletion coefficient in order to avoid confusion with the original Ficks diffusivity.
182
Another inaccuracy of Model B is explained by Zhang and Datta (2004). Model B

implements boundary conditions for heat transfer indicated in Equation (4.3.15). This is
in contrast with the model shown by Equations (4.2.11) and (4.2.12) with the initial and
boundary conditions indicated in Equations (4.2.13)(4.2.17). The other model (shown
by Equations 4.2.114.2.17) is reasonable since it indicates that the most evaporation
occurs at the surface, and the surface receives the largest thermal impact due to heat of
evaporation. This model indicates that the heat of evaporation is taken from the ambient
air and used for evaporation of water, while the heat left penetrates inside by conduction.
On the other hand, Model B indicates that the heat of evaporation is taken from the
inside so that a negative heat source is generated. Zhang and Datta (2004) suggested that
Model B results in a lower inner temperature of samples since the heat from inside is used
for evaporation as mentioned. The other model results in more a uniform temperature
inside the sample, as the heat for evaporation is taken from the ambient air.
4.3.2
The equilibrium and non-equilibrium multiphase drying models

Two approaches can be used in multiphase drying model; equilibrium and nonequilibrium. Referring to Model A shown in Equations (4.31)(4.33), the mass
balance of liquid water and water vapour is linked by the local evaporation rate. It is necessary to represent the local evaporation rate explicitly in order to solve Model A, which
consists of three partial differential equations with three dependent variables (i.e. X, Cv
and T ). In the equilibrium approach, the moisture content inside the samples is assumed
to be in equilibrium with water vapour concentration at any time so that the water
isotherm can be used to relate these relationships. The equilibrium approach does not
require the expression of local evaporation rate (Datta, 2007; Zhang and Datta, 2004).
By assuming water vapour is an ideal gas, water vapour concentration can be expressed
by (Zhang and Datta, 2004):
Cv =
pv M
,
RT
(4.3.20)
where pv is water vapour pressure (Pa), M is the molecular weight of water vapour
(kg kmol1 ), R is 8314 J kmol1 K1 and T is the temperature (K). By assuming
equilibrium conditions between the water vapour concentration and moisture content
inside the samples at any time, the water vapour pressure can be written as (Zhang and
Datta, 2004):
aw =
pv = pv (T, W ),
(4.3.21)
pv (T, X )
= f (T, X ),
pv,sat (T )
(4.3.22)
where pv (T,W) is equilibrium water vapour pressure (Pa) and pv,sat is the saturated water
pressure at particular temperature (Pa). The moisture sorption isotherm, such as GAB,
Henderson, Oswin and BET (Brunauer et al., 1938; Oswin, 1946; Thompson et al., 1968;
van den Berg, 1984), can be used to describe these relationships.
183
For the equilibrium approach, Model A for the symmetrical convective drying of slabs
can be rearranged into (Zhang and Datta, 2004):
The mass balance:

(Cs X ) Cv
(C s X )
Cv
+
=
Dw
+
Dv
.
(4.3.23)
t
t
x
x
x
x
It can be seen that Equation (4.3.23) is obtained by adding Equation (4.3.1) and (4.3.2).
The heat balance:

T
X
T
(Cs X )
=
k
+
Dw C s
HV . (4.3.24)
C p
t
x
x
t
x
x
Here, the local evaporation rate is obtained from water vapour conservation as mentioned
in Equation (4.3.2). It is emphasised that local evaporation rate here is not simply based
on the rate of moisture content during drying, as used by Model B (Zhang and Datta,
2004).
One more equation, i.e. the moisture sorption isotherm indicated in Equation (4.3.22),
is required to create the equilibrium multiphase drying approach. The spatial profiles of
moisture content, concentration of water vapour and temperature can be generated by
solving Equations (4.3.22)(4.3.24) simultaneously.
The equilibrium approach has been used by several researchers to describe convective
drying and baking (Aversa et al., 2010; Ni et al., 1999; Zhang et al., 2005; Zhang
and Datta, 2006). Generally, results have been in agreement with the experimental data
(Aversa et al., 2010; Ni et al., 1999; Zhang et al., 2005; Zhang and Datta, 2006). Zhang
et al. (2005) and Zhang and Datta (2006) implemented this approach to model the baking
of bread. The model described the moisture content profiles during baking reasonably
well. Similarly, the model resulted in a reasonable agreement with the experimental data
of surface temperature, but an underestimation in profiles of centre temperature was
observed (Zhang et al., 2005; Zhang and Datta, 2006). Coupling between the model
and the equations of conservation of the drying air can also be used to predict the flow
field of the drying air, as well as the spatial profiles of moisture content and temperature
inside the product (Aversa et al., 2010).
Although the equilibrium approach can model the drying process reasonably well,
the use of the non-equilibrium approach is recommended as it is more generic and can
be used to assess the validity of the equilibrium approach (Zhang and Datta, 2004). In the
non-equilibrium approach, as mentioned before, the local evaporation rate needs to be
expressed explicitly. It has been proposed that the internal evaporation rate can be related
to the difference of equilibrium vapour pressure and the vapour pressure at particular
times inside the pore spaces (Bixler, 1985; Chong and Chen, 1999; Scarpa and Milano,
2002; Zhang and Datta, 2004). Bixler (1985) proposed that the local evaporation rate
can be expressed as:
I = c(X X e )( pv,e pv ),
(4.3.25)
where Xe is the equilibrium moisture content (kg water kg dry solids1 ), pv,e is the
equilibrium water vapour pressure (Pa) and c is the coefficient. However, the actual
application of Equation (4.3.25) has not been tested so far (Bixler, 1985). The coefficient
184
c is expected to vary with the temperature and moisture content. The coefficient c should
be arbitrarily chosen in order to match the model with the experimental data. Similarly,
Ousegui et al. (2010) implemented the non-equilibrium model to describe the baking
process. The local evaporation rate was expressed as (Fang and Ward, 1999; Ousegui
et al., 2010; Zhang and Datta, 2004):
I = K (v,eq v ),
(4.3.26)
where v and v,eq is the water vapour concentration (kg m3 ) and equilibrium water
vapour concentration (kg m3 ), respectively and K is the proportionality constant dependent on ambient conditions (heat transfer coefficient, ambient fluid, etc.). K was determined by matching the model and experimental data. The increase of K was found to
increase the local evaporation rate and, thus, result in the greater overall drying rate
(Ousegui et al., 2010).
As mentioned before, the multiphase drying model should be employed for a better
understanding of transport phenomena of drying. However, as shown in Section 4.3.1,
the use of moisture loss as local evaporationcondensation rate is not appropriate;
the application of equilibrium multiphase drying model is restricted and use of the
non-equilibrium multiphase drying model is suggested (Chen, 2007; Zhang and Datta,
2004). The REA in its lumped format, which has been shown in Chapter 2 to describe
several challenging drying cases accurately, can be used to model the local evaporation
condensation rate. The internal evaporationcondensation rate can be expressed as (Kar
and Chen, 2010; 2011; Putranto and Chen, 2013):
I = h m in Ain (Cv,s C v ),
(4.3.27)
where hm,in is the internal mass transfer coefficient (m s1 ), Ain is the internal surface
area per unit volume (m2 m3 ), and Cv,s and Cv are the internal-surface water vapour
concentration (kg m3 ) and water vapour concentration (kg m3 ), respectively. The
procedures for determining the internal mass transfer coefficient (hm,in ) and internal
surface area per unit volume (Ain ) are presented in Kar and Chen (2010; 2011). The
value of hm,in should be in the order of Dv /rp and the Ain is determined based on the
area of single cells inside the samples or particles and number of cells per unit volume
inside the samples (Kar and Chen, 2010; 2011; Putranto and Chen, 2013).
Using the REA, Equation (4.3.27) can be rearranged into (Kar and Chen, 2010; 2011;
Putranto and Chen, 2013):

I = h m in Ain exp E v Cv,sat C v ,
(4.3.28)
RT
where Ev is the activation energy (J mol1 ), T is the sample temperature (K), R is ideal
gas constant (8.314 J mol1 K1 ) and Cv,sat is the saturated water vapour concentration
(kg m3 ).
Equation (4.3.28) can then be combined with a system of equations for conservation
of heat and mass transfer to yield a non-equilibrium multiphase drying model using
the REA as a source/depletion term, the S-REA, explained in Chapter 3. The following
185
section provides comparisons of the results of modelling using the diffusion-based model
and the L-REA, as well as the S-REA.
4.4
Comparison of the diffusion-based model and the L-REA on

convective drying
In this section, the results of modelling with the diffusion-based model and the L-REA
on the convective drying of mango tissues are presented. The experimental data are
derived from the work of Vaquiro et al. (2009). The review of experimental details has
been presented in Section 2.6.
Vaquiro et al. (2009) implemented the diffusion-based model which can be written
as:
The mass balance:

X
X
X
De
+
De
+
De
=
,
(4.4.1)
x
x
y
y
z
z
t
with the initial and boundary conditions:
t = 0, X = X 0 ,
x = 0, y= 0, z = 0,
x = L , De s
X
Mw
(L , y, z, t) = h m
x
R
X
Mw
y = L , De s
(x, L , z, t) = h m
x
R
X
Mw
z = L , De s
(x, y, L , t) = h m
x
R
X
X
X
=
=
= 0,
x
y
z
(4.4.2)
(4.4.3)

(L , y, z, t)Ps (L , y, z, t) P s
,
T (L , y, z, t)
T
(4.4.4)

(x, L , z, t)Ps (x, L , z, t) P s
,
T (x, L , z, t)
T
(4.4.5)

(x, y, L , t)Ps (x, y, L , t) P s
,
T (x, y, L , t)
T
(4.4.6)
where De is the effective diffusivity (m2 s1 ), s is the density of solid (kg m3 ), X

is the moisture content on dry basis (kg H2 O m3 ), hm is the mass transfer coefficient
(m s1 ), Mw is the molecular weight of water (kg kmol1 ), is the surface relative
humidity, Ps is the saturated vapour pressure (Pa), T is sample temperature (K), is
the drying air relative humidity, Ps is the vapour pressure in drying air (Pa), T is the
drying air temperature and L is the thickness of sample (m) while the heat balance can
186
be expressed as:

T
k
+
k
+
k
x
x
y
y
z
z

T
X T
X T
X T
De ds C pw
+
+
,
= ds (C pds + XC pw )
t
x x
y y
z z
(4.4.7)
with the initial and boundary conditions:
t = 0,T = T0 ,
x = 0, y = 0, z = 0,
x = L , k
y = L , k
z = L , k
T
T
T
=
=
= 0,
x
y
z
(4.4.8)
(4.4.9)
X
T
(L , y, z, t) = h[T (L , y, z, t) T ] De ds
(L , y, z, t)Q s ,
x
x
(4.4.10)
T
X
(x, L , z, t) = h[T (x, L , z, t) T ] De ds
(x, L , z, t)Q s ,
y
y
(4.4.11)
X
T
(x, y, L , t) = h[T (x, y, L , t) T ] De ds
(x, L , z, t)Q s ,
z
z
(4.4.12)
where k is the thermal conductivity of sample (W m1 K1 ), h is the heat transfer

coefficient (W m2 K1 ), ds is the density of dry solid (kg m3 ), Qs is the heat of
evaporation of water (J kg1 ), Cpds is the specific heat of the dry solid (J kg1 K1 ) and
Cpw is the specific heat of water (J kg1 K1 ).
The L-REA, as explained in Section 2.1, is implemented here. Basically, the original
formulation of the L-REA, as mentioned in Equation (2.1.4), is used and combined
with the prediction of the temperature distribution as described in Section 2.6.2 and
shown in Equation (2.6.14) since the sample is relatively thick. The relative activation
energy (Ev /Ev,b ) is generated from experiments on convective drying at a drying air
temperature of 55 C.
The comparisons of modelling using the diffusion-based model implemented by
Vaquiro et al. (2009) and the L-REA are shown in Figures 4.3 and 4.4. Figure 4.3 shows
the profiles of moisture content modelled using the diffusion-based model (Vaquiro et al.,
2009) and the L-REA. The diffusion-based model describes the moisture content profiles
of the convective drying at drying air temperatures of 55 and 65 C well, indicated by R2
higher than 0.996 (Vaquiro et al., 2009). However, a slight overestimation of the drying
rate is shown for the convective drying at a drying air temperature of 45 C (Vaquiro et al.,
2009). The L-REA models the moisture content well for convective drying at drying
air temperatures of 45, 55 and 65 C. The results of modelling using the L-REA
187
10
Data 45C
L-REA 45C
Data 55C
L-REA 55C
Data 65C
L-REA 65C
Diffusion-based by Vaquiro et al. (2009) 45C
8
7
6
5
4
3
2
1
0
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 4.3 Moisture content profiles from the convective drying of mango tissues modelled using
the L-REA and diffusion-based model (Vaquiro et al., 2009). [Reprinted from Drying
Technology, 29, A. Putranto, X.D. Chen and P.A. Webley, Modelling of Drying of Food Materials
with Thickness of Several Centimeters by the Reaction Engineering Approach (REA), 961973,
match well the experimental data indicated by an R2 higher than 0.996 (Putranto et al.,
2011a).
Figure 4.4 shows the centre temperature profiles modelled using both models. At a
drying air temperature of 45 C, the diffusion-based model (Vaquiro et al., 2009) results
in a kink at the beginning of the drying period not shown by the L-REA. Similarly,
the diffusion-based (Vaquiro et al., 2009) model shows a kink in the centre temperature
profiles at the beginning of the drying period for convective drying at drying air temperatures of 45, 55 and 65 C (Vaquiro et al., 2009). On the other hand, the L-REA models
well the centre temperature profiles of convective drying at drying air temperatures of
55 and 65 C, indicated by a R2 higher than 0.984 (Putranto et al., 2011b). Both models
show a slight overestimation in temperature profiles of convective drying at a drying air
temperature of 65 C between drying times of 10 000 and 30 000 s.
Based on the case study mentioned, the diffusion-based model shown in conjunction
with the initial and boundary conditions indicated in Equations (4.4.1)(4.4.12) seems
to not be able to model the convective drying well, particularly the temperature profiles
which show a kink at the beginning of drying. On the other hand, the L-REA combined
with the prediction of temperature distribution shown in Equation (2.6.14) can describe
both the moisture content and temperature profiles well. This indicates that the L-REA
performs better than the diffusion-based model in describing convective drying.
188
340
330
320
310
Data 45C
L-REA 45C
Data 55C
L-REA 55C
Data 65C
L-REA 65C
300
290
280
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 4.4 Temperature profiles from convective drying of mango tissues modelled using the
L-REA and diffusion-based model (Vaquiro et al., 2009). [Reprinted from Drying Technology,
29, A. Putranto, X.D. Chen and P.A. Webley, Modelling of drying of food materials with
thickness of several centimeters by the reaction engineering approach (REA), 961973,
4.5
Comparison of the diffusion-based model and the

S-REA on convective drying
In this section, the results of modelling using the diffusion-based model and the S-REA
on the convective drying of mango tissues (Vaquiro et al., 2009) are compared. As
explained in Chapter 3, the S-REA is a non-equilibrium multiphase drying model with
the REA as the local evaporation rate. The experimental details of Vaquiro et al. (2009)
are presented in Section 2.6. The diffusion-based model, together with the initial and
boundary conditions, is presented in Section 4.4 and indicated in Equations (4.4.1) to
(4.4.12). The details of S-REA modelling in conjunction with the initial and boundary
conditions for the convective drying of mango tissues (Vaquiro et al., 2009) are presented
in Section 3.3.1.
Figures 4.54.6 show the comparisons of the results of modelling convective drying
(Vaquiro et al., 2009) using the diffusion-based model and the S-REA. Figure 4.5 shows
the moisture content profiles modelled using both approaches. It can be seen that both the
diffusion-based model (Vaquiro et al., 2009) and the S-REA model the moisture content
profiles well during drying (Putranto and Chen, 2013). Both the diffusion-based model
(Vaquiro et al., 2009) and the S-REA are accurate enough to model the moisture content
profiles well (R2 higher than 0.996 for both the S-REA and diffusion-based model).
189
10
S-REA 45C
Diffusion-based model by
Vaquiro et al. (2009) 45C
Data 45C
S-REA 55C
Data 55C
S-REA 65C
Data 65C
9
8
7
6
5
4
3
2
1
0
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 4.5 Moisture content profiles from the convective drying of mango tissues modelled using
the S-REA and diffusion-based model (Vaquiro et al., 2009). [Reprinted from AIChE Journal, A.
Putranto and X.D. Chen, Spatial reaction engineering approach as an alternative for
non-equilibrium multiphase mass-transfer model for drying of food and biological materials,
DOI 10.1002/aic.13808, Copyright (2012), with permission from John Wiley & Sons, Inc.]
Figure 4.6 shows the centre temperature profiles modelled using the diffusion-based
model (Vaquiro et al., 2009) and the S-REA. As mentioned before, the diffusion-based
model shows a kink in the centre temperature profiles at the beginning of the drying
period for convective drying at drying air temperatures of 45, 55 and 65 C (Vaquiro
et al., 2009). On the other hand, the S-REA models the centre temperature accurately for
all cases of the convective drying of mango tissues. The results of modelling using the
S-REA match well with the experimental data, indicated by R2 higher than 0.985
(Putranto and Chen, 2013). In addition, the S-REA model yields advantages of generating the profiles of concentration of water vapour (as shown in Figure 3.6 in Chapter 3),
as well as local evaporation rate during drying (as shown in Figure 3.8 of Chapter 3)
which is useful for better understanding of transport phenomena during drying (Putranto
and Chen, 2013).
It can be shown here that the diffusion-based model (Vaquiro et al., 2009) does
represent convective drying of mango tissues well, as shown by a kink of the temperature
profiles. On the other hand, the S-REA is accurate at modelling both the moisture content
and temperature profiles of convective drying. The S-REA formulation in conjunction
with the initial and boundary conditions can represent the convective drying very well
(Putranto and Chen, 2013). The accuracy of the S-REA, which is a non-equilibrium
190
340
330
Temperature (K)
320
310
Data 45C
S-REA 45C
Data 55C
S-REA 55C
Data 65C
S-REA 65C
Diffusion-based model by Vaquiro et al. (2009) 45C
300
290
280
0.5
1.5
2.5
t(s)
3.5
4.5
5
104
Figure 4.6 Temperature profiles from the convective drying of mango tissues modelled using the
S-REA and diffusion-based model (Vaquiro et al., 2009). [Reprinted from AIChE Journal, A.
Putranto and X.D. Chen, Spatial reaction engineering approach as an alternative for
non-equilibrium multiphase mass-transfer model for drying of food and biological materials,
DOI 10.1002/aic.13808, Copyright (2012), with permission from John Wiley & Sons, Inc.]
multiphase model, with the use of REA for local evaporation rate, may prove the
importance of multiphase model with the source term as previously mentioned by
Zhang and Datta (2004) and Chen (2007).
4.6
Model formulation of Luikovs approach

Luikov (1975) developed a theory of simultaneous heat and mass transfer in a porous
body based on irreversible thermodynamics. A porous body can be considered to consist
of four components: a dry solid, water vapour, liquid water and air within the pore.
It was postulated that the thermal and moisture potential gradient within a porous
body cause the vapour and liquid water transfer so that the flux of liquid water and
water vapour is proportional to the thermal gradient and moisture potential gradient.
The relationships can be explained in two- and three-way coupled system of partial
differential equations (Luikov, 1975). The two-way coupled system assumes that the
pressure inside the capillary body is constant during the process. It is postulated that
the thermal and moisture potential gradient within a porous body cause vapour and
liquid water transfer so that the flux of liquid water and water vapour is proportional
to the thermal gradient and moisture potential gradient. The heat and mass transfer
191
is interdependent and several coefficients are used to explain the interdependency. For
this purpose, two sets of partial differential equations in terms of moisture content and
temperature are established (Luikov, 1975). The two-way coupled system has been used
by several researchers to model the simultaneous heat and mass transfer processes (Liu
and Cheng, 1990; Mikhailov and Shishedjiev, 1975; Younsi et al., 2006a,b; 2007).
In addition, the three-way coupled system incorporates the pressure gradient inside a
body as a result of the presence of water vapour, which results in moisture movement by
filtration. In this system, it is assumed that the thermal, moisture and pressure gradient
within a capillary body lead to moisture movement within the body. Unlike the two-way
system, another set of partial differential equations in terms of pressure is established to
represent the pressure gradient inside the body (Luikov, 1975).
For the two-way system, it is assumed that the vapour movement inside the body is due
to molecular transport and the concentration of vapour equilibrates thermodynamically
with the concentration of liquid. The mass flow of vapour can be written as (Luikov,
1975):
T
0 t,
j1 = D10 = am1 0 u am1
(4.6.1)
where j1 is the flux of vapour (kg m2 s1 ), is the dimensionless factor characterising

resistance to vapour diffusion in the moisture body, D is the diffusivity of vapour in air
(m2 s1 ), is the density (kg m3 ), 0 is the density of dry body (kg m3 ), u is the
moisture content of body (kg kg1 ), t is temperature (K) and am1 is the vapour diffusion
coefficient which can be expressed as (Luikov, 1975):

d10
am1 = D
,
(4.6.2)
0
du T
where 10 is the relative vapour concentration, while am1 T is the thermal diffusion
coefficient, which can be written as (Luikov, 1975):

d10
T
am1
= D
.
(4.6.3)
0
du u
The mass flow of liquid can be expressed in a similar way to that of vapour as (Luikov,
1975):
T
0 t,
j2 = D10 = am2 0 u am2
(4.6.4)
where j2 is the flux of liquid (kg m2 s1 ) and am2 and am2 T are the vapour and
thermal diffusion coefficients dependent on the moisture content and temperature of the
body.
Fouriers law can be used to explain the heat flux inside the body as (Luikov, 1975):
q = kt,
(4.6.5)
where q is the heat flux (W m2 ) and k is the thermal conductivity (W m2 K1 ).
192
Equations (4.6.1), (4.6.4) and (4.6.5) can be rearranged into (Luikov, 1975):
u
= K 11 2 u + K 12 2 t,
t
= K 21 2 u + K 22 2 t,
where K 11 = am = am1 + am2 ,

K 12 =
T
am1
T
am2
,
Lam
,
c
Lam
K 22 = a +
,
c
T
a T + am2
,
= m1
am1 + am2
K 21 =
(4.6.6)
(4.6.7)
(4.6.8)
(4.6.9)
(4.6.10)
(4.6.11)
(4.6.12)
where L is the specific heat of phase transition (J kg1 K1 ), is the thermal diffusivity
(m2 s1 ) and c is the specific heat of sample (J kg1 K1 ).
For a three-way coupling system, the pressure gradient may occur due to the penetration of humid air through capillary system inside the body. The transfer can be described
by Darcys law as (Luikov, 1975):
j f = k f p,
(4.6.13)
where kf is the total filtration coefficient and p is pressure.

By incorporating Equation (4.6.13), the mass and heat transfer inside the capillary
body can be expressed as (Luikov, 1975):
u
= K 11 2 u + K 12 2 t + K 13 2 p,
t
= K 21 2 u + K 22 2 t + K 23 2 p,
p
= K 31 2 u + K 32 2 t + K 33 2 p,
where K 13 = am f ,
am
K 23 = L p ,
c
am
K 33 = f
p,
Cf
am
,
K 31 =
Cf
am
K 31 =
,
Cf
kf
,
f =
C f
kf
,
f =
am 0
where Cf is the body capacity for the humid air with filtration.
(4.6.14)
(4.6.15)
(4.6.16)
(4.6.17)
(4.6.18)
(4.6.19)
(4.6.20)
(4.6.21)
(4.6.22)
(4.6.23)
193
The boundary conditions on the surface can be written as (Luikov, 1975):

am 0 (u)s + amT (t)s + js = 0,
(k +
T
)(t)s
Lam2
where
Lam2 0 (u)s + qs = 0,
(4.6.24)
(4.6.25)
js = 0 (u s u e ),
(4.6.26)
qs = h(ta ts ),
(4.6.27)
where js is the mass flux at surface (kg m2 s1 ), qs is heat flux at surface (W m2

K1 ), us is the surface moisture content, ue is the equilibrium moisture content, ta is the
ambient temperature and ts is the surface temperature.
Equation (4.6.24) implies that the moisture supplied to the surface due to thermodynamic forces is equal to the one left from the surface to the ambience, while Equation
(4.6.25) indicates that heat received from the ambience is used for evaporation and any
left penetrates inside the body.
Luikovs approach (Luikov, 1975) has been used by several researchers to model
drying of porous materials (Irudayaj and Wu, 1994; 1996; Kulasiri and Samarasinghe,
1996; Thomas et al., 1980). The two-way coupled system of Luikovs approach was
implemented to model timber drying and a reasonable agreement with the experimental
data was shown. The model was solved numerically using the finite element method.
The results of modelling confirmed that the temperature gradient inside the sample can
be neglected (Kulasiri and Samarasinghe, 1996; Thomas et al., 1980). Irudayaj and Wu
(1994) used the three-way coupled version of Luikovs approach and solved it by using
the finite element method to model the convective drying of silicon gel. The numerical
model matched the exact solution well. For low Fourier numbers, the results are not stable
but the stable results are achieved if high Fourier numbers, higher geometries and nonlinear properties are used (Irudayaj and Wu, 1994). In a similar investigation, Irudayaj
and Wu (1996) showed that the model can be used to model the convective drying of
spruce samples well. The parameters of moisture conductivity and ratio of vapour to
total diffusion were found to be very sensitive to the profiles of moisture content and
temperature. Nevertheless, these parameters need to be extracted from experimental data
and determined by error-minimization (Irudayaj and Wu, 1996).
Luikovs approach (Luikov, 1975) has been also used by (Younsi et al., 2006a) to
model the heat treatment of wood under a constant heating rate, which is essentially a
drying process under linearly increased gas temperatures. The experimental details of
Younsi et al. (2006a,b; 2007) are reviewed briefly in Section 2.9. For modelling the heat
treatment of wood using Luikovs approach (1975), it is assumed that the wood sample
is isotropic and homogeneous, shrinkage is negligible, no heat is generated inside the
wood and capillary forces are much stronger than gravity. The moisture transfer can be
represented as (Younsi et al., 2006a):

U
km
(4.6.28)
=
T + km U ,
Cm
t
Cm
while the heat transfer can be expressed as (Younsi et al., 2006a):

T
km
Cq
=
kq +
T + (km )U .
t
Cm
(4.6.29)
194
The initial and boundary conditions of Equations (4.6.28) and (4.6.29) are (Younsi et al.,
2006a):
t = 0, U = U0 ,
(4.6.30)
t = 0, T = T0 ,
dT
kq
= h q (T Tg ) + (1 )h m (U Ug ), at
dn
dU
km dT
km
=
+ h m (U Ug ), at
dn
Cm dn
(4.6.31)
(4.6.32)
(4.6.33)
where t is time (s), n is the spatial direction (x,y,z), U is the moisture potential (M1 ),
T is temperature (K), hq is the heat transfer coefficient (W m2 K1 ), hm is the mass
transfer coefficient (kg H2 O m1 s1 M1 ), Cq is the heat capacity (J kg1 K1 ), Cm
is the moisture capacity (kg H2 O, kg1 M1 ), is the thermal gradient coefficient
(kg H2 O kg1 K1 ), kq is the conductivity (W m1 K1 ), km is the moisture diffusivity
(kg H2 O m1 s1 M1 ), is the latent heat (J kg1 ), is the dry body density (kg m3 ),
Ug is the gas moisture potential (M1 ), Tg is the gas temperature (K), is the ratio of
vapour diffusion to total moisture diffusion and is the boundary surface of heat and
mass transfer.
For a more general expression, Equations (4.6.28) and (4.6.29) can be represented as
(Younsi et al., 2006a):
T
U
+ A12
= (K 11 T + K 12 U ),
t
t
T
U
+ A22
= (K 21 T + K 22 U ).
A21
t
t
A11
(4.6.34)
(4.6.35)
Aij and Kij are coefficients which can be represented as (Younsi et al., 2006a):
A11 = Cq ,
(4.6.36)
A12 = A21 = 0,
k M
K 11 = kq +
,
Cm
K 12 = km ,
km
K 21 =
,
Cm
K 22 = km .
(4.6.37)
(4.6.38)
(4.6.39)
(4.6.40)
(4.6.41)
The heat and mass transfer coefficient can be estimated from established correlations
(Incropera and DeWitt, 2002) while the properties of wood can be predicted from
Simpson and Tenwold (1999). The heat and mass balances shown in Equations (4.6.34)
and (4.6.35) in conjunction with the initial and boundary conditions shown in Equations
(4.6.36)(4.6.41) are solved simultaneously in order to give the profiles of moisture
content and temperature during drying.
4.7
195
Model formulation of Whitakers approach

Whitakers approach (Whitaker, 1977) was implemented by several researchers to
describe drying of a porous body (Baggio et al., 1997; Pavon-Melendez et al., 2002;
Whitaker and Chou, 1983; Younsi et al., 2006a,b; 2007). This approach proposed detailed
mechanisms of transport in microscale and macroscale structures based on a known distribution of the structures. Equations of conservation of momentum, mass and heat in
solid, vapour and liquid phases and their local volume behaviours followed by volume
averaging methods, are used to describe the mechanisms. Darcys law is usually used to
describe the momentum transfer in liquid and gas phases while the mass transfer considers capillary action as well as evaporationcondensation. The heat transfer incorporates
the convective transport, conduction and vaporisationcondensation heat by assuming
local thermal equilibrium within solid, gas and liquid phases (Whitaker, 1977; Whitaker
and Chou, 1983).
The detailed modelling of Whitakers approach can be expressed as (Whitaker, 1977):
The total thermal energy equation:
C p
d
T
+ [ (C p )
v +
(C p )
v ]T
dt
= (keff
T ) +
,
+ h vap
m
(4.7.1)
where t is time (s),

is the spatial average of density (kg m3 ), Cp is specific heat
(J kg1 K1 ), is density of the liquid phase (kg m3 ), is density in the gas phase
(kg m3 ), (Cp ) is specific heat in the liquid phase, (Cp ) is specific heat in the gas
phase,
v is the phase average of velocity in the liquid phase,
v is the phase
average velocity in the gas phase, T is temperature (K), hvap is the vaporisation heat of
is the evaporation rate
water (J kg1 ), keff is the thermal conductivity (W m2 K1 ),
m
per unit volume (kg m3 s1 ) and is the heat generation (W m3 ).
The liquid phase equation of motion:
v =
K
[k + k
T T ( )g],
(4.7.2)
where is the volume fraction of liquid phase, K is the liquid phase permeability (m2
s1 ), is the viscosity of liquid phase (N s m2 ), k is pc / , k
T is pc /
T ,
g is the gravitational constant (m2 s1 ), is a function topology of liquid phase and pc
is the capillary pressure (N m2 ).
The liquid phase continuity equation:
= 0;
+
v +
t
(4.7.3)
The gas phase equation of motion:
v =
K
{ [
p p0 g]};
(4.7.4)
196
where is the volume fraction of the gas phase, K is gas phase permeability (m2 s1 ),
is viscosity of gas phase (N s m2 ), p0 is the reference pressure (N m2 ) and p is
pressure of the gas phase (N m2 ).
The gas phase continuity equation:
(
) + (

v ) =
m;
t
The gas phase diffusion equation:
2
(
2 /
];
(
2 ) + (
2
v ) = [
Deff
t
(4.7.5)
(4.7.6)
where
2 is the density of water vapour (kg m3 ) and Deff 2 is the effective diffusivity
of water vapour in gas (m2 s1 ).
The volume constraint:
+ + = 1;
(4.7.7)
where is the volume fraction of solid phase.

The thermodynamics relations:
p1 =
1 R1
T ;
(4.7.8)
p2 =
2 R2
T ;
(4.7.9)
=
1 +
2 ;
(4.7.10)
p =
p1 +
p2 ;

h vap
1
1
p1 = p1 exp (2 /r R1
T ) +
;
R1
T T0
(4.7.11)
(4.7.12)
where
i is the pressure of ith species in gas phase (N m2 ), where
i is the density
of ith species in gas phase (kg m3 ), Ri is the gas constant for ith species (N m kg1
K1 ), p1 0 is the reference pressure of ith species in the gas phase (N m2 ) and T0 is the
reference temperature.
As a result, Whitakers approach provides the 12 equations shown in Equations (4.7.1)
(4.7.12) which need to be solved simultaneously in order to yield the profiles during
drying. The difficulty of this approach would be in determining the transport parameters
involved in the equations. For this purpose, a theoretical basis and simplified drying
theory are needed to estimate the parameters (Whitaker, 1977).
Whitakers approach (Whitaker, 1977) has been used to model several drying process
and the results have good agreement with the experimental data (Hager et al., 2000;
Torres et al., 2011). Torres et al. (2011) implemented the approach to model the vacuum
drying of wood and the results can yield information about transport of liquid and vapour
in wood. The model was also coupled with the equations of conservation in the drying
medium so the dynamics of water vapour concentration inside the chamber could be
predicted. It was shown that the evaporation rate depends on the levels of pressure and
temperature, while the pressure inside the sample increases as drying progresses because
the liquid water is expulsed during drying (Torres et al., 2011). In addition, Hager et al.
(2000) use Whitakers approach to model steam drying of porous Al2 O3 and ceramic
197
spheres. It was indicated that the total drying rate and outlet steam temperature matched
well with experimental data, but the internal temperature showed some deviations from
it. Although the model is fairly accurate, Whitakers approach is very computationally
demanding (Truscott and Turner, 2005).
Whitakers approach (1977) was also implemented by Younsi et al. (2006b; 2007) to
describe the heat treatment of wood under a constant heating rate whose experimental
details are presented in Section 2.6. Whitakers approach (1977), implemented by Younsi
et al. (2006b; 2007), can be expressed as:
d
M
Jx
Jy
Jz
= J =
+
+
,
t
x
y
z
(4.7.13)
where d is the dry wood density (kg m3 ) and J is the total mass flux vector (kg m2 s1 ),
which can be written as:
J = Jv + Jb + J f ,
(4.7.14)
where Jv , Jb and Jf are the water vapour flux (kg m2 s1 ), bound water flux (kg m2
s1 ) and liquid water flux (kg m2 s1 ), respectively.
The vapour flow is postulated due to the vapour pressure gradient, since the total
pressure gradient inside the sample can be neglected so that the water vapour flux can
be written as (Younsi et al., 2006b; 2007):
Jv =
m v Deff
Pv ,
RT
(4.7.15)
where mv is the molar mass of vapour (kg mol1 ), R is the ideal gas constant (J mol1 ),
Pv is the partial vapour pressure of water (Pa) and Deff is effective vapour diffusivity
(m2 s1 ).
The bound water diffusion is governed by the chemical potential difference and it
occurs when the moisture content is below the fibre saturation point (FSP). The bound
water flux can be written as (Stanish et al., 1986):

Db
Pv
T
Jb =
8.314 ln
T
187 + 35.1 ln
mv
298.15
101.325

T
(4.7.16)
+ 8.314 Pv ,
Pv
where Db is the diffusion coefficient of bound water (m2 s1 ) and T is temperature (K).
When moisture content is higher than FSP, the free water transfer occurs due to
capillary flow. The liquid water flux can be represented as (Younsi et al., 2006b;
2007):
Jf =
K l l
Pc ,
l
(4.7.17)
where Pc is the capillary pressure (Pa), Kl is the permeability, l is the density of liquid
water (kg m3 ) and l is the viscosity of liquid water (kg m1 s1 ).
198
Equation (4.7.17) can be rearranged into (Younsi et al., 2006b; 2007):

Jf =
0.61 104 K l l (Mmax MFSP )0.61

M.
l (M MFSP )
(4.7.18)
where MFSP is the moisture content at the fibre saturation point and Mmax is the moisture
content if the entire void is occupied by water.
By substituting Equations (4.7.15), (4.7.16) and (4.7.18) into Equations (4.7.13)
and (4.7.14), the conservation of mass can be written as (Younsi et al., 2006b;
2007):
d
M
= (D M M + DT T ) ,
t
(4.7.19)
where DM and DT are positive constants, which can be expressed as (Younsi et al., 2006b;
2007):
1.2146 104 m v 0.75
M
R
Db T
K l l (Mmax M F S P )
+ 8.314
,
+
mv M
l (M M F S P )(1+)
D M = Psv
DT =

Psv
1.2146 104 m v 0.75 Psv
+ Psv
R
T
T

Db
Pv
T
8.314 ln
187 + 35.1 ln
mv
298.15
101325

T Psv
+ Psv
,
8.314
Pv
T
T
(4.7.20)
(4.7.21)
where Psv is the saturation vapour pressure of water (Pa) and is the coefficient, =
104 , = 0.61.
The equation of conservation of energy can be written as (Younsi et al., 2006b;
2007):

d a u a + v u v + b u b + f u f + d u d + (Ja h a + Jv h v + Jb h b + J f h f )
= (kT ),
(4.7.22)
where , u, h and k indicate; the density, the specific internal energy, the specific enthalpy
and the thermal conductivity, respectively. The subscripts a, v, b, f and d indicate; the
dry air, the water vapour, the bound water, the free water and the dry wood, respectively.
The energy balance can be represented as (Younsi et al., 2006b; 2007):
C p

T
= k M M + keff T ,
t
(4.7.23)
199
where kM and keff are positive constants, which can be expressed as (Younsi et al., 2006b;
2007):
1.2146 104 m v 0.75
Db T
+ h b 8.314
M
R
mv M
K l l (Mmax M F S P )
+hf
,
l (M M F S P )(1+)

1.2146 104 m v 0.75 Psv
Psv
= k + hv
+ Psv
R
T
T
P !

T
v
Db
8.314
187 + 35.1
298.15
ln
ln 101325
hb
.
Psv

T
m v 8.314 Pv T + Psv T
k M = h v Psv
keff
(4.7.24)
(4.7.25)
For more general expression, Equations (4.7.19) and (4.7.23) can be represented as
(Younsi et al., 2006b; 2007):
M
T
(4.7.26)
+ A12
= (K 11 M + K 12 T ),
t
t
M
T
A21
(4.7.27)
+ A22
= (K 21 M + K 22 T ),
t
t
where Aij and Kij are coefficients, which can be represented as (Younsi et al., 2006b;
2007):
A11
A11 = d ,
(4.7.28)
A12 = A21 = 0,
(4.7.29)
A22 = (C p )eff ,
(4.7.30)
K 11 = D M ,
(4.7.31)
K 12 = DT ,
(4.7.32)
K 21 = k M ,
(4.7.33)
K 21 = keff .
(4.7.34)
The mass and heat balances shown in Equations (4.7.26) and (4.7.27) are solved with
the initial and boundary conditions, which can be written as:
t = 0, M = M0 ,
(4.7.35)
t = 0, T = T0 .
(4.7.36)
At the surface, the boundary conditions can be written as (Younsi et al., 2006b; 2007):
kM
Ts = T f ,
(4.7.37)
Cs = C f ,
(4.7.38)
dM
dT
dT f
dC f
+ keff
= kf
+ DHV
,
dn
dn
dn
dn
dM
dT
dC f
+ DT
=D
,
DM
dn
dn
dn
(4.7.39)
(4.7.40)
200
0.08
Experimental data
L-REA
Luikov approach: Younsi et al. (2006a)
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0.5
1.5
t(s)
2.5
3
104
Figure 4.7 Moisture content profiles from the heat treatment of wood modelled using the L-REA
and Luikovs approach. [Reprinted from Bioresource Technology, 102, A. Putranto, X.D. Chen,
Z. Xiao and P.A. Webley, Modelling of high-temperature treatment of wood by using the reaction
where Ts is the temperature (K), Tf is the gas temperature (K), Cs is the surface water
concentration (kg m3 ), Cf is the water vapour concentration in gas, kf is the bulk gas
thermal conductivity (W m1 K1 ), D is the diffusivity of water in bulk gas (m2 s1 )
and HV is the vaporisation heat (J kg1 ).
The spatial profiles of moisture content and temperature during the heat treatment
of wood can be generated by solving the heat and mass balances shown in Equations
(4.7.26) and (4.7.27), in conjunction with the initial and boundary conditions shown in
Equations (4.7.37)(4.7.40). The linearly increased gas temperature is implemented in
Equations (4.7.37), (4.7.39) and (4.7.40).
4.8
Comparison of the L-REA, Luikovs and Whitakers approaches for

Figure 4.7 indicates the moisture content profiles of heat treatment of wood with a final
temperature of 200 C, heating rate of 20 C h1 and initial moisture content of 0.07 kg
water kg dry solids1 . Both the L-REA and Luikovs approach (implemented by Younsi
et al., 2006a) match reasonably well with the experimental data of moisture content (R2
of 0.993 and 0.987 for the L-REA and Luikovs approach, respectively). Both approaches
slightly underestimate the drying rate during treatment times shorter than 10 000 s. The
L-REA resulted in better agreement with the experimental data during treatment times
201
Table 4.1 Experimental settings of the heat treatment of wood (Younsi

et al., 2007).
Case
Final gas
temperature (C)
Heating rate
(C h1 )

1
2
220
220
20
10
0.125
0.12
480
Experimental data
L-REA
460
440
Temperature (K)
420
400
380
360
340
320
300
280
0.5
1.5
2
t(s)
2.5
3.5
104
Figure 4.8 Temperature profiles from the heat treatment of wood modelled using the L-REA and
Luikovs approach. [Reprinted from Bioresource Technology, 102, A. Putranto, X.D. Chen,
Z. Xiao and P.A. Webley, Modelling of high-temperature treatment of wood by using the reaction
above 10 000 s. The temperature profiles of the experiment are shown in Figure 4.8.
Both the L-REA and Luikovs approach match well with experimental data (R2 of 0.999
and 0.994 for the L-REA and Luikovs approach, respectively). It can be shown here that
the L-REA gives comparable results with Luikovs approach.
The results of modelling of heat treatment of wood (refer to Table 4.1) using the
L-REA and Whitakers approach (Whitaker, 1977, implemented by Younsi et al. (2007))
are shown in Figures 4.9 and 4.10. The experimental data are derived from the work of
Younsi et al. (2007), the experimental details are presented in Section 2.8 and the experimental settings are shown in Table 4.1. Figure 4.9 shows the moisture content profiles
of heat treatment of wood modelled using both approaches. For Case 1 (heating rate of
20 C h1 , refer to Table 4.1), Whitakers approach models the moisture content profiles
well at treatment times shorter than 20 000 s, but the approach slightly overestimates
the drying rate after the treatment time of 20 000 s (R2 of 0.991). The L-REA results in
0.14
202
Case 1-Experimental data

Case 1-L-REA
0.12
Case 1-Whitaker approach

by Younsi et al. (2007)
Case 2-L-REA
0.1
0.08

0.06
0.04
0.02
0
4
t(s)
7
104
Figure 4.9 Moisture content profiles from the heat treatment of wood (refer to Table 4.1)
modelled using the L-REA and Whitakers approach. [Reprinted from Bioresource Technology,
102, A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley, Modelling of high-temperature treatment
of wood by using the reaction engineering approach (REA), 62146220, Copyright (2012), with
better agreement with experimental data, shown by R2 of 0.994. Both the L-REA and
Whitakers approach match reasonably well the experimental data for modelling of heat
treatment in Case 2 (R2 of 0.987 and 0.971 for the L-REA and Whitakers approach,
respectively). Nevertheless, the L-REA yields a better agreement with experimental
data.
The temperature profiles of heat treatment of wood (refer to Table 4.1) are indicated
in Figure 4.10. Generally, the L-REA and Whitakers approach model the temperature
profiles well (R2 higher than 0.993 and 0.991 for the L-REA and Whitakers approach,
respectively). However, Whitakers approach slightly underestimates the temperature
profiles during the stage where gas temperature is held constant at 120 C for half an
hour. During this period, the L-REA resulted in a better agreement with the experimental
data, which could be because of the flexibility of activation energy in changing according
to ambient conditions.
In conclusion, although the L-REA is much simpler in mathematical modelling, it
performs comparably well with Luikovs and Whitakers approaches (implemented by
Younsi et al., 2006a; 2007) in modelling heat treatment of wood under a constant heating
rate. The L-REA is more efficient at generating the model parameters, since these can
be generated from one accurate experiment. On the other hand, the two other approaches
require several sets of experiments and complex optimization procedures to generate
these model parameters.
203
500
Temperature (K)
450
400
Case 1-L-REA
Case 2-L-REA
350
300
250
4
t(s)
7
104
Figure 4.10 Temperature profiles from the heat treatment of wood (refer to Table 4.1) modelled
using the L-REA and Whitakers approach. [Reprinted from Bioresource Technology, 102,
A. Putranto, X.D. Chen, Z. Xiao and P.A. Webley, Modelling of high-temperature treatment of
wood by using the reaction engineering approach (REA), 62146220, Copyright (2012), with
4.9
Comparison of the S-REA, Luikovs and Whitakers approaches for

In this section, the results of modelling of the heat treatment of wood under various constant heating rates using the S-REA, Luikovs (Luikov, 1975) and Whitakers approaches
(Whitaker, 1977) are compared. The experimental details are described in Section 2.8
while the modelling using Luikovs (Luikov, 1975) and Whitakers approaches (Whitaker,
1977) are presented in Section 4.6 and 4.7, respectively.
Figure 4.11 shows the moisture content profiles of heat treatment of wood with a
final temperature of 200 C, heating rate of 20 C h1 and initial moisture content
of 0.07 kg water kg dry solids1 , modelled using the S-REA and Luikovs approach
(applied by Younsi et al., 2006a). Both the S-REA and Luikovs approach (implemented
by Younsi et al., 2006a) model the moisture content profiles reasonably well (R2 of
0.99 and 0.981 for the S-REA and Luikovs approach, respectively), but the S-REA
yields a closer agreement with the experimental data as the other model results in an
increase of moisture content at the beginning of the treatment and an overestimation
in the profiles before a treatment time of 10 000 s. The temperature profiles of the
experiment, modelled using both approaches are shown in Figure 4.12. Both the S-REA
and Luikovs approach (applied by Younsi et al., 2006a) model the temperature profiles
0.08
Experimental data
S-REA
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
0.5
1.5
t(s)
2.5
3
104
Figure 4.11 Moisture content profile from the heat treatment of wood modelled using the S-REA
and Luikovs approach.

480
Experimental data
S-REA
460
440
420
Temperature (K)
204
400
380
360
340
320
300
280
0.5
1.5
2
t(s)
2.5
3.5
104
Figure 4.12 Temperature profile from the heat treatment of wood modelled using the S-REA and
Luikovs approach.
205
0.14
Case 1-S-REA
Case 2-S-REA
0.12
0.1
0.08
0.06
0.04
0.02
0
t(s)
7
104
Figure 4.13 Moisture content profiles from the heat treatment of wood (refer to Table 4.1)
modelled using the S-REA and Whitakers approach.
500
Temperature (K)
450
400
Case 1-S-REA
Case 2-S-REA
350
300
250
t(s)
7
104
Figure 4.14 Temperature profiles from the heat treatment of wood (refer to Table 4.1) modelled
using the S-REA and Whitakers approach.
206
during the heat treatment accurately (R2 of 0.998 and 0.994 for the S-REA and Luikovs
approach, respectively).
Figures 4.13 and 4.14 show the results of modelling of heat treatment of wood under
constant heating rates (refer to Table 4.1) using the S-REA and Whitakers approach
(used by Younsi et al. (2007)). For Case 1 (refer to Table 4.1), Whitakers approach
overestimated the drying rate after a treatment time of 20 000 s. The S-REA yields a
closer agreement with the experimental data (R2 of 0.995 and 0.991 for the S-REA and
Whitakers approach, respectively). Similarly, the S-REA performs better than the other
model at representing heat treatment of wood in Case 2 (R2 of 0.988 and 0.971 for the
S-REA and Whitakers approach, respectively).
The temperature profiles of the various cases of the heat treatment are shown in
Figure 4.14. Both the S-REA and the Whitakers approach describe the temperature
profiles well along the treatment (R2 higher than 0.992 and 0.991 for the S-REA and
Whitakers approach, respectively). Compared to the other approach, the S-REA has
the advantage of generating profiles of the spatial water vapour concentration so that
better understanding of the process can be studied. In conclusion, it can be said that the
S-REA performs comparably or even better than Whitakers approach in modelling the
heat treatment of wood under a constant heating rate.
4.10
Summary
In this chapter, the diffusion-based models, Luikovs and Whitakers approaches which
have been used extensively applied to model drying processes are analysed. The formulation and limitation of Cranks diffusion-based model and several other forms of
diffusion-based model are discussed. Cranks diffusion-based model offers advantages
of simple mathematical modelling but it is only valid for conditions where there are
no shrinkage, isothermal, constant diffusivity, negligible external resistance and uniform initial moisture content. Nevertheless, it is often implemented in literature without
necessary justification. The effective diffusivity of liquid water has been interpreted
in experiments with poor documtation and control of boundary conditions, inducing a
large range in orders of magnitude in diffusivity values. Various forms of diffusion-based
models reported in literature have also been described. The crucial issues are boundary
conditions and the implementation of a source term. For a better understanding of the
drying process transport phenomena, a multiphase diffusion-based model needs to be
implemented. For this purpose, equilibrium and non-equilibrium diffusion-based models
can be implemented and are both described. The equilibrium one assumes the moisture
content inside pores of the samples equilibrates with water vapour concentration, which
essentially eliminates the source term in a way. The non-equilibrium one is more general
but it requires the explicit formulation of the source term.
Luikovs approach assumes that the mechanisms of moisture and heat transfer are
similar, which is due to thes gradient of moisture content and temperature. Therefore,
several coefficients are used to represent the interdependency of moisture content and
temperature. Whitakers approach uses detailed equations of momentum, heat and mass
207
conservation in solid, liquid and gas phases, followed by the volume averaging method
to describe drying processes.
The accuracy of the diffusion-based model has been compared to those of the LREA and S-REA to model convective drying. It has been shown that the L-REA performs comparably, or sometimes even better than the diffusion-based model. While the
results are comparable, the L-REA has the advantages of simplicity in generating the
parameters and mathematical formulation. The S-REA works very well, which offers
advantages where it not only accounts for the diffusion processes (Fickian type) but also
local evaporation or condensation.
Similarly, the accuracy of the Luikovs and Whitakers approaches are compared with
those of the L-REA and S-REA to model heat treatment of wood, which is essentially
a drying process under linearly increased gas temperatures. It has been shown that the
L-REA and S-REA perform comparably or even better than Luikovs and Whitakers
approaches. While the results are similar, the L-REA offers the advantages of efficiency
in generating the parameters and simplicity of mathematical modelling. The S-REA is
also more efficient in generating the parameters than Luikovs and Whitakers approaches.
References
Adhikari, B., Howes, T., Bhandari, B.R. and Truong, V., 2004. Effect of addition of maltodextrin on drying kinetics and stickiness of sugar and acid-rich foods during convective drying:
experiments and modelling. Journal of Food Engineering 62, 5368.
Arslan, D. and Ozcan, M.M., 2011, Dehydration of red bell-pepper (Capsicum annuum L.):
Change in drying behavior, color and antioxidant content. Food and Bioproducts Processing
89, 504513.
Aversa, M., Curcio, S., Calabro, V. and Iorio, G., 2010. Transport phenomena modelling during
drying of shrinking materials. Computer Aided Chemical Engineering 28, 9196.
Baggio, P., Bonacina, C. and Schrefler, B.A., Some considerations in modelling in heat and mass
transfer in porous media. Transport in Porous Media 28, 233251.
Batista, L.M., da Rosa, C.A. and Pinto, L.A.A., 2007. Diffusive model with variable effective
diffusivity considering shrinkage in thin layer drying of chitosan. Journal of Food Engineering
81, 127132.
Bixler, N.E., 1985. NORIA A finite element computer program for analyzing water, vapor, air,
and energy transport in porous media. Report No. SAND842057, UC-70, Sandia National
Laboratories, Albuquerque, New Mexico 87185.
Brunauer, S., Emmett, P. H. and Teller, E., 1938. Adsorption of gases in multi-molecular layers.
Journal of American Chemical Society 60, 309319.
Buckingham, E.A., 1907. Studies on the movement of soil moisture. Bulletin No. 38, Washington,
D.C., U.S. Department of Agriculture, 1907.
Castell-Palou, A., Rosello, C., Femenia, A., Bon, J. and Simal, S., 2011. Moisture profiles in
cheese drying determined by TD-NMR: Mathematical modelling of mass transfer. Journal of
Chen, X.D., 2007. Moisture diffusivity in food and biological materials, Drying Technology 25,
12031213.
208
41654178.
Cihan, A. and Ece, M.C., 2001. Liquid diffusion model for intermittent drying of rough rice,
Journal of Food Engineering 49, 327331.
Corzo, O., Bracho, N. and Alvarez, C., 2008. Water effective diffusion coefficient of mango slices
at different maturity stages during air drying, Journal of Food Engineering 87. 479484.
Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press, Oxford.
Cussler, E.L., 1984. Diffusion Mass Transfer in Fluid Systems. Cambridge University Press,
Cambridge/New York.
Datta, A.K., 2007. Porous media approaches to studying simultaneous heat and mass transfer in
food processes. I: Problem formulations. Journal of Food Engineering 80, 8095.
Di Scala, K. and Crapiste, G., 2008. Drying kinetics and quality changes during drying of red
pepper. LWT 41, 789795.
Doymaz, I., 2004. Drying kinetics of white mulberry. Journal of Food Engineering 61, 341346.
Fang, G. and Ward, C.A., 1999. Examination of the statistical rate theory expression for liquid
evaporation rates. Physical Review E 59, 441453.
Ferrari, G., Meerdink, G., Walstra, P., 1989. Drying kinetics for a single droplet of skim-milk.
Garcia-Perez, J.V., Carcel, J.A., Benedito, J. and Mulet, A., 2009. Influence of the applied acoustic
energy on the drying of carrots and lemon peel. Drying Technology 27, 281287.
Garcia-Perez, J.V., Ozuna, C., Ortuno, C., Carcel, J.A. and Mulet, A., 2011. Modelling ultrasonically assisted convective drying of eggplant. Drying Technology 29, 14991509.
Guine, R.P.F., 2008. Pear drying: Experimental validation of a mathematical prediction model.
Food and Bioproducts Processing 86, 248253.
Hager, J., Wimmerstedt, R. and Whitaker, S., 2000. Steam drying a bed of porous spheres: Theory
and experiment, Chemical Engineering Science 55, 16751698.
Hassini, L., Azzouz, S., Peczalski, R. and Belghith, A., 2007. Estimation of potato moisture
diffusivity from convective drying kinetics with correction for shrinkage, Journal of Food
Engineering 79, 4756
Henry, P.S.H., 1939. Diffusion in absorbing media. Proceedings of the Royal Society London
171A, 215241.
Huang, H., Brooks, M.S.B., Huang, H.J. and Chen, X.D., 2009. Inactivation kinetics of yeast cells
during infrared drying. Drying Technology 27, 10601068.
Irudayaj, J. and Wu, Y., 1994. Finite element analysis of coupled heat, mass and pressure transfer
in porous biomaterials. Numerical Heat Transfer Part A 26, 337350.
Irudayaj, J. and Wu, Y., 1996. Analysis and application of Luikovs heat, mass and pressure transfer
model to a capillary porous media. Drying Technology 14, 803824.
Kar, S., 2008. Drying of Porcine Skin-Theoretical Investigations and Experiments. Ph.D. thesis.
Kar, S. and Chen, X.D., 2010. Moisture transport across porcine skin: experiments and implementation of diffusion-based models. International Journal of Healthcare Technology and
209
Kar, S. and Chen, X.D., 2011. Modelling of moisture transport across porcine skin using reaction
engineering approach and examination of feasibility of the two phase approach. Chemical
Engineering Communication 198, 847885.
Kar, S., Chen, X.D., Adhikari, B.P. and Lin, S.X.Q., 2009. The impact of various drying kinetics
models on the prediction of sample temperature-time and moisture content-time profiles during
moisture removal from stratum corneum. Chemical Engineering Research and Design 87,
739755.
Kaya, A., Aydin, O., Demirtas, C. and Akgun, M., 2007. An experimental study on the drying
kinetics of quince. Desalination 212, 328343.
Kowalski, S.J. and Pawlowski, A., 2010a. Drying of wet materials in intermittent conditions.
Kowalski, S.J. and Pawlowski, A., 2010b. Modelling of kinetics in stationary and intermittent
Kowalski, S.J., Rajewska, K. and Rybicki, A., 2005. Stresses generated during convective and
microwave drying. Drying Technology 23, 18751893.
Kulasiri, D. and Samarasinghe, S., 1996. Modelling of heat and mass transfer of biological materials: a simplified approach of materials with small dimension. Ecological Modelling 86, 163
167.
Lewis, W.K., 1921. The rate of drying of solid materials. Industrial and Engineering Chemistry
13, 427432.
Li, Y.B., Cao, C.W., Yu, Q.L. and Zhong, Q.X., 1999. Study on rough rice fissuring during
intermittent drying. Drying Technology 17, 17791793.
Liu, J. and Cheng, S., 1990. A parametric study of heat and mass transfer in drying of capillaryporous media. Multiphase transport in porous media. ASME 22, 2532.
Loulou, T., Adhikari, B. and Lecomte, D., 2006. Estimation of concentration-dependent diffusion coefficient in drying process from the space-averaged concentration versus time with
experimental data. Chemical Engineering Science 61, 71857198.
Lu, L., Tang, J. and Liang, L., 1998. Moisture distribution in spherical foods in microwave drying.
Luikov, A.V., 1975. Systems of differential equations of heat and mass transfer in capillary-porous
bodies. International Journal of Heat and Mass Transfer 18, 114.
Mariani, V.C., de Lima A.G.B. and Coelho, L.S., 2008. Apparent thermal diffusivity estimation of the banana during drying using inverse method. Journal of Food Engineering 85,
569579.
Mikhailov, M.D. and Shishedjiev, B.K., 1975. Temperature and moisture distributions during
contact drying of a moist porous sheet. International Journal of Heat and Mas Transfer 18,
1524.
Misra, R., Barker, A.J. and East, J., 2002. Controlled drying to enhance properties of technical
ceramics. Chemical Engineering Journal 86, 111116.
Mrad, N.D., Boudhrioua, N., Kechaou, N., Courtoisa, F. and Bonazzi, C., 2012. Influence of air
drying temperature on kinetics, physicochemical properties, total phenolic content and ascorbic
acid of pears. Food and Bioproducts Processing 90, 433441.
Ni, H., Datta, A.K. and Torrance, K.E., 1999. Moisture transport in intensive microwave heating of
biomaterials multiphase porous media model. International Journal of Heat and Mass Transfer
42, 15011512.
Nowicki, S.C., Davis, H.T. and Scriven, L.E., 1992. Microscopic determination of transport
parameters in drying porous media. Drying Technology 10, 925946.
210
Ousegui, A., Moresoli, C., Dostie, M. and Marcos, B., 2010. Porous multiphase approach for
baking process Explicit formulation of evaporation rate. Journal of Food Engineering 100,
535544.
Pakowski, Z. and Adamski, A., 2007. The comparison of two models of convective drying of shrinking materials using apple tissue as an example. Drying Technology 25, 1139
1147.
Pavon-Melendez, G., Hernandez, J.A., Salgado, M.A. and Garcia, M.A., 2002. Dimensionless
analysis of the simultaneous heat and mass transfer in food drying. Journal of Food Engineering
51, 347353.
Prat, M., 1993. Percolation model of drying under isothermal conditions in porous media. International Journal of Multi-Phase Flow 19, 691704.
29, 961973.
Putranto, A., Chen, X.D., Xiao, Z. and Webley, P.A., 2011b. Modelling of high-temperature
102, 62146220.
Ramallo, L.A. and Mascheroni, R.H, 2012. Quality evaluation of pineapple fruit during drying
process. Food and Bioproducts Processing 90, 275283.
Oswin, C.R., 1946. The kinetics of package life. III. Isotherm. Journal of the Society of Chemical
Industry 65, 419421.
Sablani, S.S., Marcotte, M., Baik, O.D. and Castaigne, F., 1998. Modelling of simultaneous heat
and water transport in the baking process. LWT 31, 201209.
Sahin, A.Z. and Dincer, I., 2002. Graphical determination of drying process and moisture transfer
parameters for solids drying. International Journal of Heat and Mass Transfer 45, 3267
3273.
Scarpa, D. and Milano, G., 2002. The role of adsorption and phase change phenomena in the thermophysical characterization of moist porous materials. International Journal of Thermophysics
23, 10331046.
Simpson, W. and Tenwold, A., 1999. Physical Properties and Moisture Relations of Wood. Wood
Handbook. USDA Forest Service, Forest Product Laboratory, Madison, WI. pp. 123.
Srikiatden, J. and Roberts, J.S., 2006. Measuring moisture diffusivity of potato and carrot (core
and cortex) during convective hot air and isothermal drying. Journal of Food Engineering 74,
143152
Stanish, M.A., Schajer, G.S. and Kayihan F., 1986. A mathematical model of drying for hygroscopic porous media. AIChE Journal 32, 13011311.
Thomas, H.R., Morgan, K. and Lewis, R.W., 1980. A fully nonlinear analysis of heat and mass
transfer problems in porous bodies. International Journal of Numerical Methods Engineering
15, 13811393.
Thompson, T.L., Peart, R.M. and Foster, G.H., 1968. Mathematical simulation of corn drying a
new model. Transactions of the American Society of Agricultural Engineers 11, 582586.
Thuwapanichayanan, R., Prachayawarakorn, S. and Soponronnarit, S. 2008. Modelling of diffusion
with shrinkage and quality investigation of banana foam mat drying. Drying Technology 26,
13261333.
211
Timoumi, S., Mihoubi, D. and Zagrouba, F., 2007. Shrinkage, vitamin C degradation and aroma
losses during infra-red drying of apple slices. LWT 40, 16481654.
Torres, S.S., Jomaa, W., Puiggali, J.R. and Avramidis, S., 2011. Multiphysics modelling of vacuum
drying of wood. Applied Mathematical Modelling 35, 50065016
Truscott, S.L. and Turner, I.W., 2005. A heterogeneous three-dimensional computational model
for wood drying. Applied Mathematical Modelling 29, 381410.
van den Berg, C., 1984. Description of water activity of food engineering purposes by means of
the GAB model of sorption. In B.M McKenna (ed.), Engineering and Foods. Elsevier, New
York.
Design 87, 885898.
Viollaz, P.E. and Rovedo, C.O., 2002. A drying model for three-dimensional shrinking bodies.
Whitaker, S., 1977. Simultaneous heat, mass and momentum transfer in porous media: A theory
of drying. Advances Heat Transfer 13, 119203.
Whitaker, S. and Chou, W.T.H., 1983. Drying granular porous media-theory and experiment.
Yang, W., Jia, C.C., Siebenmorgen, T.J., Pan, Z. and Cnossen, A.G., 2003. Relationship of kernel
moisture content gradients and glass transition temperatures to head rice yield. Biosyst. Eng.
85, 467476.
Yoshida, M. and Miyashita, H., 2002. Drying behavior of polymer solution containing two volatile
solvents. Chemical Engineering Journal 86, 193198.
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2006a. Thermal modelling of the high
Younsi, R., Kocaefe, D., Poncsak, S. and Kocaefe, Y., 2006b. Transient multiphase model for the
27, 14241431.
Zhang, J. and Datta, A.K., 2004. Some considerations in modelling of moisture transport in heating
of hygroscopic materials. Drying Technology 22, 19832008.
Zhang, J. and Datta, A.K., 2006. Mathematical modelling of bread baking process. Journal of
Zhang, J., Datta, A.K. and Mukherjee, S., 2005. Transport processes and large deformation during
baking of bread, AIChE Journal 51, 25692580.
Index
activation energy, xxxiv, 16, 24, 27, 34, 35, 39, 40,
5154, 57, 64, 69, 73, 82, 85, 88, 90, 91, 93,
96, 100, 102, 105, 117, 127, 129, 132, 142,
150, 159
average water content, x, 5, 21, 22, 25
bacteria, xxvii, 11
baking, vi, vii, xvii, xviii, xxiii, 3, 7, 95100, 117,
119121, 158168, 183, 184, 210, 211
balance, xxxiii
biological, 17
Biot, v, x, xxxiv, 11, 25, 27, 30, 4345, 47, 50, 81,
117, 119, 146, 165, 171, 208
boundary, 5, 8, 10, 19, 23, 24, 29, 43, 63, 64, 123,
124, 127, 128, 130, 132, 133, 141, 142, 149,
150, 159, 160, 171, 173, 176179, 181, 182,
185189, 193, 194, 199, 200, 206
capillary, ix, 7, 8, 11, 17, 122, 125, 170, 172, 177,
179, 181, 190193, 195, 197, 208, 209
cell, 127
centre sample temperature, 63
Characteristic Drying Rate Curve, xxxi, 20
chemistry, v, xxxi, 1, 5, 15, 18, 19, 25, 29
combustion, xxx, xxxvii, 1, 18, 19, 31
Comparison, 185, 200
computational fluid dynamics, xxxvi, 29
concentration, 15
condensation, xxxiii, 10, 16, 17, 19, 24, 25, 28, 29,
32, 34, 122, 126, 132, 133, 137, 142, 143, 151,
154, 158, 160, 164, 170, 172, 177, 181, 184,
195, 207
conduction, 45
constant rate period, 17, 54, 102, 106, 108,
114
continuum, 17
convection, 45
convective, 40
core, xxii, 8, 11, 35, 43, 128, 133, 136, 139, 140,
145, 146, 157, 161, 167, 210
cortex, 133
coupling, 41
Crank, vii, 171, 173, 206, 208
critical water content, 2123

cycle, 74
Darcy flow, 126
deflection, 36
deformable, 20, 166
diffusion, vii, viii, xxiv, 6, 9, 11, 17, 40, 43, 47, 58,
61, 68, 69, 73, 79, 97, 117, 121, 124, 126, 133,
135, 154, 155, 157, 166, 167, 169174, 176,
181, 185191, 193195, 197, 206210
diffusivity, 48
discrete, 18
discretization, 12
distributed, xxviii, xxxiv, 5, 8, 17, 18, 43
driving force, xxxiii, 8, 17, 46, 50
droplet, 36
effective diffusivity, 130
effective liquid diffusivity, 126
empirical, xxvii, xxviii, 17, 20, 47, 73, 105, 181
energy, x, xi, xii, xiv, xv, xvii, xxvii, xxxiii, 3, 5, 12,
16, 19, 20, 2429, 3436, 39, 40, 42, 48, 5155,
57, 61, 62, 64, 65, 68, 69, 73, 74, 8082, 85,
8891, 93, 94, 96, 97, 100, 102, 103, 105, 107,
108, 116, 117, 121, 122, 124, 125, 127, 129,
130, 132, 138, 142, 150, 159, 160, 164, 172,
184, 186, 195, 198, 202, 207, 208
equilibrium, 35, 182
equilibrium model, 121
evaporation, vii, x, xxii, xxiii, xxx, xxxiii, xxxiv, 1,
79, 11, 16, 17, 19, 20, 24, 25, 27, 28, 29, 32,
34, 36, 41, 4547, 49, 63, 64, 102, 107, 108,
122, 124, 126, 127, 129, 131, 132, 136, 137,
138, 142, 143, 145148, 150, 151, 154,
157160, 164, 170, 172, 177, 179184, 186,
188190, 193, 195, 196, 207, 208, 210
fiber, 1, 2, 30, 33, 89, 197, 198
Fick, 169, 181
finite element, 12, 193, 207
first order, 34
fissuring, 170, 209
Food, 1
Index
Fourier, 169, 191, 193

free volume, 124, 125, 167
fundamental, 170
glass-filament, 36
gradient, xxx, 17, 43, 51, 96, 122, 135, 136, 143,
145147, 170, 172, 190194, 197, 206
heat balance, 52
heat transfer coefficient, 27, 42, 44, 45, 47, 52, 53,
54, 65, 82, 91, 97, 102, 159, 173, 178, 184,
186, 194
infrared-heating, 100
interface, 178
intermitent, 141
intermittent, 69
internal mass transfer coefficient, 125
internal surface area per unit volume, 125
isotherm, xxx, 24, 26, 39, 49, 121, 182,
183
isotherma, 126
kaolin, 170
Lagrangian, 12, 40
Lewis, v, x, 11, 27, 4345, 4750, 117, 172, 209,
210
linear, 88
liquid, 9
local phase change term, 121
Luikov, 172, 190
lumped, 34
Lumped-REA, 29
macro-scale, 17
Magnetic resonance imaging, 11
mass balance, 25, 52
Mass transfer, 9, 208
mass transfer coefficient, 23, 24, 34, 35, 38,
46, 47, 51, 54, 62, 124126, 129, 133, 138,
142, 151, 159, 171, 173, 176, 178, 184, 185,
194
material, ix, xxvii, xxx, xxxiii, xxxv, 1, 3, 57,
912, 16, 19, 20, 2227, 3336, 40, 4248, 50,
62, 65, 116, 121, 122, 124, 139, 141, 164, 169,
172
Material Point Method, 12
meso-scale, xxviii, 17
micro-scale, 11, 17
microwave, xxviii, 1, 5, 166, 170, 209
modelling, 15, 116
momentum, v, xxvii, xxxi, 5, 33, 40, 41, 43, 195,
206, 211
multiphase, 177
multiphase drying model, 122
213
non-equilibrium multiphase drying approach, 121

Nusselt, 38, 42, 52
Ohm, 169
optimization, 40
paper, xxx, xxxiii, 1, 18, 19, 24
parabolic, 63
physical, v, xxxiii, 6, 15, 18, 19, 22, 25, 29, 31, 80,
95, 169
physics, 18
polymer, 100
pore, 121
porosity, 124
preservation, 1, 11
quality, 170
reaction engineering approach, 15
relative humidity, xi, xii
resistance, 45, 176
Reynolds, 38, 41, 42, 51
Sherwood, 24, 34, 38, 42, 51, 54, 62
shrinkage, 124
slab, 170
solid concentration, 126
source term, 29, 48, 49, 121, 122, 179, 190, 206
spatial reaction engineering approach, 121
spatial-REA, 121
Spatial-REA, 29
stress, ix, 3, 12, 13, 33, 95, 170
structure, xxiv, 3, 6, 8, 9, 11, 12, 18, 27, 49, 95, 121,
122, 124, 127, 136, 143, 148, 158, 172, 175
surface, 16
surface sample temperature, 63
surface vapor concentration, 25, 34, 62, 73, 74, 79,
80
symmetry, 63, 64, 149
thermal conductivity, 48
thermocouple, 36
thick samples, vi, 61, 64, 66, 69, 73, 119, 166, 210
thickness, 63
thin layer, 34
timber, 193
transport, 121
transport phenomena, 121
turtuosity, 125
uniform, 35
vapor, 9
vapor diffusivity, 125
vaporization heat, 52
volume-averaged, 17
214
Index
water vapor concentration, x, xxii, xxiii, 23, 26, 34,

38, 49, 73, 121, 124, 129, 130, 132, 133, 136,
138, 141144, 149151, 154157, 160,
171173, 178, 184, 206
wetting, 34
whey protein concentrate, 51
Whitaker, 195
Whittaker, 172
wood, vi, vii, viii, ix, xvii, xxiv, xxv, xxvi, xxxi, 1, 3,
5, 12, 15, 18, 8895, 116, 118, 120, 148, 150,
151, 154, 155, 157159, 166168, 193, 194,
196198, 200207, 210, 211
zero order, 24

Modelling Drying Processes A Reaction Engineering Approach

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modelling Drying Processes A Reaction Engineering Approach

Uploaded by

Copyright:

Available Formats

more information - www.cambridge.

Modelling Drying Processes

Modelling Drying Processes

CAMBRIDGE UNIVERSITY PRESS

Cambridge, New York, Melbourne, Madrid, Cape Town,

Xiao Dong Chen and Aditya Putranto 2013

This publication is in copyright. Subject to statutory exception

Cambridge University Press has no responsibility for the persistence or

Reaction engineering approach I: Lumped-REA (L-REA)

The REA formulation

Results of modelling convective drying of a thin layer of a mixture

2.12.2 Results of modelling the intermittent drying of a mixture of

Reaction engineering approach II: Spatial-REA (S-REA)

The S-REA formulation

Comparisons of the REA with Fickian-type drying theories, Luikovs and

Comparison of the diffusion-based model and the L-REA on

Some traditional dried products. (a) Broccoli-steam blanched and air

The droplet diameter changes during convective drying of WPC.

(PVA)/glycerol/water mixture The reaction engineering approach

Moisture content profile of convective mango tissues at air temperatures

Industrial Engineering Chemistry Research, 50, A. Putranto, Z. Xiao,

Intermittent drying of mango tissues: Implementation of the reaction

approach (REA) for modelling intermittent drying under time-varying

Moisture content profile of intermittent drying in Case 4 (periodically

baking of cake using the reaction engineering approach (REA),

Sensitivity of the moisture content profile of cyclic drying, Case 1 (refer

polyvinyl alcohol (PVA)/glycerol/water mixture, 51935203, Copyright

Temperature profile of cyclic drying, Case 3 (refer to Table 2.12) using

engineering approach as an alternative for nonequilibrium multiphase

Spatial water vapour concentration profiles of mango tissues during

moisture contenttime profiles during moisture removal from stratum

Modelling of high-temperature treatment of wood by using the reaction

Experimental conditions of convective drying of a mixture of

During my Ph.D. study in the Chemical and Process Engineering Department at

phenomenon. This formed the foundation of my understanding of the presence of water

make a significant contribution in the area of agglomeration (hardware improvement

Bhattahtrayya (Mechanical Engineering) for their innovations. I benefited tremendously

Modelling Drying Processes

Modelling Drying Processes

acid; (b) 1, cellulose; 2, starch; 3, pectin; (c) human insulin.

Modelling Drying Processes

A microstructural discussion of the phenomena of drying

Modelling Drying Processes

Mass exchange surface

Modelling Drying Processes

Figure 1.5 Packed particulate material.

Figure 1.6 Cellular structures in plant material.

(with an expanding resistance layer). In general though, this is a simplification, which

Modelling Drying Processes

A stack of 253 slices obtained using

Rescaling to Lagrangian coordinates and

Digital cutting of 3D geometry

Modelling Drying Processes

<1 A at z axis), several minutes (Perre, 2011).

The REA to modelling drying

The relevant classical knowledge of physical chemistry

Modelling Drying Processes

four times with a temperature increment of 10 K. The relationship between a reaction

where kAo is a constant.

General modelling approaches

Modelling Drying Processes

Modelling Drying Processes

(qualitatively derived from Equation 1.3.6).