Transport Processes in Porous Media

Advanced Structured Materials
Volume 20
Series Editors
Andreas Öchsner
Lucas F. M. da Silva
Holm Altenbach
For further volumes:

http://www.springer.com/series/8611
Frank A. Coutelieris J. M. P. Q. Delgado
•
Transport Processes
in Porous Media
123
Frank A. Coutelieris J. M. P. Q. Delgado
Department of Environmental and Natural Laboratorio de Fisica das Construccoes
Resources Management Faculdade de Engenharia Universidade do
University of Ioannina Porto
Seferi 2, 30100 Agrinio Rua Dr. Roberto Frias
Greece 4200-465 Porto
Portugal
ISSN 1869-8433 e-ISSN 1869-8441

ISBN 978-3-642-27909-6 e-ISBN 978-3-642-27910-2
DOI 10.1007/978-3-642-27910-2
Springer Heidelberg New York Dordrecht London
Library of Congress Control Number: 2011946089
Ó Springer-Verlag Berlin Heidelberg 2012

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed. Exempted from this legal reservation are brief
excerpts in connection with reviews or scholarly analysis or material supplied specifically for the
purpose of being entered and executed on a computer system, for exclusive use by the purchaser of
the work. Duplication of this publication or parts thereof is permitted only under the provisions of the
Copyright Law of the Publisher’s location, in its current version, and permission for use must always be
obtained from Springer. Permissions for use may be obtained through RightsLink at the Copyright
Clearance Center. Violations are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt
from the relevant protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)

The effort I put for this work is dedicated to
my wife, Natasa, who offered me the most
valuable gift of my life: our three children.
Frank Coutelieris
Preface
What is a ‘‘porous medium’’? Although several definitions have been proposed,

depending on the application the porous material is involved with, a common
sense of this term exists to almost everyone. This is mainly due to their abundance.
Either artificial or natural, porous materials can be found in the wide majority of
scientific applications, technological developments, and the natural environment.
Clothing, chalk, underground oil reservoirs, catalytic pellets and sandstones are
typical examples of ‘‘everyday’’ porous materials while other, more modern,
approaches consider bread, cooked rice and the human scalp as porous media.
Independent of the approach adopted, the basic concepts and properties that
characterize a porous medium (porosity, connectivity, consolidation, percolation,
tortuosity, homogeneity, isotropy and anisotropy, permeability) have to be well
defined and understood before one attempts investigating transport phenomena
occurring within them.
‘‘Transport processes in porous media’’ is a field of great industrial interest, but
it still requires more research before it can be properly understood. Fundamental
and applied research in heat and mass transfer in porous media has generated
increasing interest over the past five decades because of the importance of porous
media and transport processes in many engineering applications. A large amount
of literature has been generated on this subject and significant advances have been
made in modeling fluid flow, heat and mass transfer through porous media
including clarification of several important physical phenomena.
The main propose in writing this book is to present a new approach, in several
respects. Unlike the most recent reference books on porous media, which cata-
logue in pure and relatively abstract terms the current position of fundamental
research on porous media, this book provides a comprehensive description of the
flow and mass transport in porous materials. Authors avoid to present heat transfer
processes in porous structures because such a topic is too extensive to be covered
along with the other transport processes, better requiring an external book.
The overall approach here is based both on experimental and theoretical state-
of-the-art studies, covering the physical and engineering aspects encountered
within some of the main applications for porous materials. The text is designed to
vii
viii Preface
be intelligible to readers who have some previous knowledge of fundamentals of

transport processes as well as a considerable mathematical background. The
book’s flexible structure makes its accessible to the general engineer or scientist,
but also caters for readers with non-scientific backgrounds.
This book was jointly authored from 2010 to 2011, and authors conclude that
even if they cannot make any claim of completeness in investigating this crucial
field, they believe that some of its recent developments are significantly addressed
here Authors would also acknowledge with gratitude the support received from
Sandy Coles and Dr. Eleni Vakouftsi as well as current and previous research
teams established in the University of Ioannina—Department of Environmental
and Natural Resources Management, Greece, the National Center for Scientific
Research—Environmental Research Laboratory, Greece, the University of Wes-
tern Macedonia—Department of Mechanical Engineering, Greece, and the Uni-
versity of Porto—Faculty of Engineering, Portugal. Finally, the authors would
welcome reader comments, corrections and suggestions with the aim of improving
any future editions.
Frank Coutelieris
J. M. P. Q. Delgado
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Why Porous Materials? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Aims and Scopes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Book Contents and Structure. . . . . . . . . . . . . . . . . . . . . . . . . . 3
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 Fundamentals of Porous Structures . . . . . . . ....... ......... 5

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . ....... ......... 5
2.2 Cell Models . . . . . . . . . . . . . . . . . . . . . ....... ......... 5
2.3 Digital Reconstruction (Lattice Boltzmann Method)
of the Porous Structure. . . . . . . . . . . . . . ....... . . . . . . . . . 9
2.3.1 Porous Media Generation . . . . . . . ....... . . . . . . . . . 10
2.4 Stochastic Modeling. . . . . . . . . . . . . . . . ....... . . . . . . . . . 11
2.4.1 The Pore Space. . . . . . . . . . . . . . ....... . . . . . . . . . 13
2.4.2 Transport Properties . . . . . . . . . . ....... . . . . . . . . . 17
References . . . . . . . . . . . . . . . . . . . . . . . . . . ....... . . . . . . . . . 18
3 Flow in Porous Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.1.1 Macroscopic Description . . . . . . . . . . . . . . . . . . . . . . . 24
3.1.2 Microscopic Description . . . . . . . . . . . . . . . . . . . . . . . 25
3.2 Analytical Solutions for Single Phase Flow in Cell Models . . . . 25
3.2.1 Sphere-in-Cell Models: Kuwabara’s Approach . . . . . . . . 26
3.2.2 Sphere-in-Cell Models: Happel’s Approach . . . . . . . . . . 27
3.2.3 Cylinder-in-Cell Models. . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.4 Spheroid-in-Cell Model . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Single Phase Flow in Granular Structures. . . . . . . . . . . . . . . . . 33
3.3.1 Representation of 3-D Sphere Assemblages . . . . . . . . . . 33
3.3.2 The Flow Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 37
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
ix
x Contents
4 Transport Phenomena in Porous Structures . . . . . . . . . . . . . . . . . 39

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2.1 Constrictivity Factor . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2.2 Tortuosity Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.2.3 Porosity Factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2.4 Diffusion in Semi-Infinite Porous Media . . . . . . . . . . . . 43
4.2.5 Diffusion in a Plane Sheet . . . . . . . . . . . . . . . . . . . . . . 44
4.2.6 Diffusion in a Cylinder . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2.7 Diffusion in a Sphere . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Axial Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.3.1 Parameters Influencing Axial Dispersion:
Porous Medium . . . . . . . . . . . . . . . . . . . .......... 48
4.3.2 Parameters Influencing Axial Dispersion:
Fluid Properties . . . . . . . . . . . . . . . . . . . .......... 53
4.4 Radial Dispersion . . . . . . . . . . . . . . . . . . . . . . . .......... 58
4.4.1 Parameters Influencing Radial Dispersion:
Porous Medium . . . . . . . . . . . . . . . . . . . .......... 60
Fluid Properties . . . . . . . . . . . . . . . . . . . .......... 63
4.5 Dispersion in Packed Beds Flowing
by Non-Newtonian Fluids. . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
4.6 Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.6.1 New Correlations: Axial Dispersion . . . . . . . . . . . . . . . 74
4.6.2 New Correlations: Radial Dispersion . . . . . . . . . . . . . . . 76
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5 Modeling of Transport Processes in Porous Materials . . . . . . . . . . 87

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.2 Single Phase Transport in Unit Cells . . . . . . . . . . . . . . . . . . . . 87
5.2.1 Fundamental Quantities . . . . . . . . . . . . . . . . . . . . . . . . 88
5.2.2 Adsorption Mechanisms. . . . . . . . . . . . . . . . . . . . . . . . 89
5.2.3 Mass Transport Through Spheres . . . . . . . . . . . . . . . . . 91
5.2.4 Mass Transport Through Cylinders . . . . . . . . . . . . . . . . 98
5.2.5 Mass Transport Through Spheroids . . . . . . . . . . . . . . . . 98
5.2.6 Single Phase Mass Transport in Other-Type
Unit Cells . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
5.3 Single Phase Flow in Granular Structures. . . . . . . . . . . . . . . . . 106
5.4 Macroscopic Quantities for Single Phase Transport . . . . . . . . . . 112
5.4.1 Stochastically Constructed 3-D Sphere Assemblage . . . . 113
5.4.2 The Flow Field (Single Phase Flow) . . . . . . . . . . . . . . . 113
5.4.3 Mathematical Formulation . . . . . . . . . . . . . . . . . . . . . . 114
5.4.4 The Volume-Averaging Procedure . . . . . . . . . . . . . . . . 114
Contents xi
5.4.5 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6 Experimental and Numerical Investigation of Mass Transport

in Porous Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.1 Measurement of Molecular Diffusion Coefficients . . . . . . . . . . . 123
6.1.1 Diffusion Alone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
6.1.2 Diffusion with Convection . . . . . . . . . . . . . . . . . . . . . . 125
6.1.3 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
6.2 Measurement of Dispersion Coefficients (Axial and Radial) . . . . 128
6.2.1 Measurement of Axial Dispersion Coefficients . . . . . . . . 128
6.2.2 Measurement of Radial Dispersion Coefficients . . . . . . . 131
6.3 Measurement of Solubility at Different Temperatures . . . . . . . . 141
6.3.1 Mass Transfer Around a Buried Soluble Sphere . . . . . . . 142
6.3.2 Experimental Set-Up . . . . . . . . . . . . . . . . . . . . . . . . . . 143
6.4 Measurement of Tortuosity in Porous Media. . . . . . . . . . . . . . . 145
6.4.1 Experimental Example. . . . . . . . . . . . . . . . . . . . . . . . . 148
6.5 Mass Transfer Around Active Solids . . . . . . . . . . . . . . . . . . . . 149
6.5.1 Mass Transfer From a Soluble Flat Slab . . . . . . . . . . . . 149
6.5.2 Mass Transfer From a Soluble a Cylinder Aligned
With Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
6.5.3 Mass Transfer From a Soluble Sphere . . . . . . . . . . . . . . 157
6.5.4 Mass Transfer From a Cylinder in Cross Flow . . . . . . . . 159
6.5.5 Mass Transfer From a Prolate Spheroid . . . . . . . . . . . . . 162
6.5.6 Mass Transfer From an Oblate Spheroid . . . . . . . . . . . . 167
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7 Applications and Examples. . . . . . . . . . . . . . . . . . . . . . . . ...... 175

7.1 Contaminant Plume Sizes Associated to Different
Active Solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 175
7.1.1 Concentration Profiles From a Soluble Flat Slab. ...... 176
7.1.2 Concentration Profiles From a Cylinder Aligned
With Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 178
7.1.3 Concentration Profiles From a Soluble Sphere . . ...... 178
7.1.4 Concentration Profiles From a Cylinder
in Cross Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
7.1.5 Concentration Profiles From a Prolate Spheroid . . . . . . . 180
7.1.6 Concentration Profiles From an Oblate Spheroid. . . . . . . 181
7.2 Rising Damp in Building Walls. . . . . . . . . . . . . . . . . . . . . . . . 182
7.2.1 Rising Damp Theory . . . . . . . . . . . . . . . . . . . . . . . . . . 183
7.2.2 The Wall Base Ventilation System . . . . . . . . . . . . . . . . 184
7.2.3 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.2.4 Rising Damp Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 187
xii Contents
7.3 Bubbles and Drops in Porous Structures. . . . . . . . . . . . . . . . . . 194

7.3.1 Moments of Distribution (Sc) . . . . . . . . . . . . . . . . . . . . 195
7.3.2 Determination of Source Terms . . . . . . . . . . . . . . . . . . 197
7.3.3 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
7.3.4 Algorithm Validation and Findings . . . . . . . . . . . . . . . . 202
7.4 Fluid Flows Through Porous Media in Fuel Cells . . . . . . . . . . . 206
7.4.1 Solid Oxide Fuel Cell Configuration . . . . . . . . . . . . . . . 208
7.4.2 Electrochemical and Surface Reactions . . . . . . . . . . . . . 208
7.4.3 Transport Phenomena in Gas Channels . . . . . . . . . . . . . 210
7.4.4 Transport Phenomena in Porous Media . . . . . . . . . . . . . 211
7.4.5 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
7.5 Multi Phase Transport in Porous Media . . . . . . . . . . . . . . . . . . 217
7.5.1 Theoretical Background . . . . . . . . . . . . . . . . . . . . . . . . 218
7.5.2 Formulation of the Problem . . . . . . . . . . . . . . . . . . . . . 220
7.5.3 The Closure Problems . . . . . . . . . . . . . . . . . . . . . . . . . 221
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Chapter 1
Introduction
Fluid and flow problems in porous media have attracted the attention of industrialists,
engineers and scientists from varying disciplines, such as chemical, environmental
and mechanical engineering, geothermal physics and food science. Porous medium
domains, in which fluid phase mass, phase component mass, momentum, energy,
electric charge, and other extensive quantities are being transported while under-
going transformations, are encountered in a variety of disciplines: Civil Engineering
(concrete is a porous medium, and soil is a deformable porous medium), Hydrology
(aquifers), Environmental Engineering (groundwater pollution by toxic liquids and
hazardous wastes), Chemical Engineering (reactors, but the subsurface also behaves
like a large chemical reactor in which chemical and biological transformations of
contaminants take place), Petroleum Engineering (oil and gas production from
reservoirs), Agricultural Engineering (drainage and irrigation), Biomedical Engi-
neering (lungs, kidneys), and Geothermal Engineering (geothermal reservoirs), to
mention but a few.
It is widely accepted that most materials are to some extent porous, in the sense
that it is quite difficult to find or compose an extremely dense (non-porous) medium.
It is evident that the physical and chemical properties of these materials are depen-
dent on their pore structure. For instance, the control of porosity is of great industrial
importance in the design of catalysts, industrial adsorbents, membranes and cera-
mics, as it influences the chemical reactivity of solids and the physical interaction of
solids with gases and liquids. The complexity and variety of porous materials has
therefore led to a diverse variety of studies on the transport processes occurring there.
1.1 Why Porous Materials?
What is a porous material? The term ‘‘porous’’ is used to describe a complex

structure consisting of a compact phase (usually solid) and some void space, where
their analogy defines the most significant property of a porous material, its
F. A. Coutelieris and J. M. P. Q. Delgado, Transport Processes in Porous Media, 1

Advanced Structured Materials 20, DOI: 10.1007/978-3-642-27910-2_1,
Springer-Verlag Berlin Heidelberg 2012
2 1 Introduction
‘‘porosity’’. Generally speaking, any solid material which contains cavities,

channels or interstices may be regarded as porous, although in a particular context
a more restrictive definition may be appropriate. Thus, in describing a porous
solid, care must be exercised in the choice of terminology in order to avoid
ambiguity. More precisely, all these ‘‘pores’’, ‘‘cavities’’ etc. must be carefully
defined. Chapter 2 of this book presents a detailed discussion of these terms and
their use. For the moment, the following definition sounds accurate enough for the
reader: A porous medium is a region in space comprising of at least two homo-
geneous material constituents, presenting identifiable interfaces between them in a
resolution level, with at least one of the constituent remaining fixed or slightly
de-formable [1].
Why are porous materials interesting? Besides the wide spectrum of applica-
tions they have attained during the last centuries, they are not transparent, thus it is
necessary to describe any processes occurring within them without local obser-
vation, and based only on macroscopic data regarding the inlet and outlet of the
domain. The complexity of porous structures allows for assuming very complex
local-scale phenomena, which should be integrated to normal macroscopic ones.
For instance, even very low volumetric flow-rate would lead to extremely high
local velocity and Reynolds number, when the porous structure contains very thin
(capillary) throats locally.
The term ‘‘porous’’ is often used instead of the term ‘‘dense’’, which is the most
suitable fundamental consideration for the continuum in physics and applied
sciences in general. What is important is that dense materials do not actually exist.
Even the most compact artificially developed material is of small porosity and
must be treated as porous. As a consequence, the study of flow and transport in
porous media is applicable to a very wide range of fields, with practical applica-
tions in modern industry and environmental areas. In all these disciplines, transport
phenomena and transformation occur in porous media domains at a wide range of
scales, from the microscopic to the field scale. Understanding these transport
phenomena, modeling them at different scales, coping with the uncertainties
inherent in such models, and solving the models in order to provide data for
management and decision making is the subject of this book.
1.2 Aims and Scopes
This book examines the transport processes that take place in the pore structure of
porous materials. Models of pore structure are presented with a discussion of how
such models can be used to predict macroscopic quantities that are useful for
practical applications. The book is in principal devoted to interpretations of
experimental results and simulation predictions in this area, giving directions for
future research. Practical applications are given where applicable.
1.2 Aims and Scopes 3
As a first step towards the study of transport processes in porous media, several
techniques for artificial representation (modeling) of porous media are discussed.
Single phase flows in simplistic and complex porous structures are described in
terms of macroscopic and microscopic equations as well as their analytical and
numerical solutions. Furthermore, macroscopic quantities such as permeability are
introduced and discussed. Mass transport processes in porous media are also
discussed and further strengthened by experimental validation and specific tech-
nological applications.
The special features of this book are (a) the use of state-of-the-art techniques for
modeling transport processes in porous structures, (b) the consideration of realistic
sorption mechanisms, (c) the application of advanced mathematical techniques for
up-scaling major quantities, and (d) experimental investigation and application,
namely, experimental methods for the measurement of relevant transport
properties.
1.3 Book Contents and Structure
This book contains seven chapters about transport processes in porous media,
organized as follows:
• Initially, Chap. 2 introduces the reader to the Fundamentals of Porous Struc-
tures by presenting some methods for the description of porous structures. These
methods correspond to analytical descriptions (cell models) and numerical ones,
either deterministic (digital reconstruction) or probabilistic (stochastic
modeling).
• Chapter 3 presents the Flow in Porous Media, which is necessary for the ade-
quate estimation of any other transport process. More precisely, this chapter
presents the analytical expressions derived in cell models when single phase
flow is considered. This is then further exploited to more realistic granular
structures.
• Transport Phenomena in Porous Structures are presented in Chap. 4, where
details of basic phenomena (diffusion and dispersion) are discussed, while a
selected application (Dispersion in Packed Beds Flowing by Non-Newtonian
Fluids) has been used as an abundant exemplar.
• Chapter 5 deals with the Modeling of Transport Processes in Porous Materials,
presenting single phase transport in the selected geometries specified in Chap. 3
(cells, granular structures) as well as a novel scale-up technique to obtain
macroscopic quantities from microscopic measures/calculations (‘‘Macroscopic
quantities for single phase transport’’).
• The theoretical results of Chap. 5 are further strengthened by the Experimental
Investigation of Transport in Porous Media, presented in Chap. 6. These
experiments contain measurements of molecular diffusion and axial and radial
dispersion coefficients, calculation of the solubility of organic compounds at
4 1 Introduction
different temperatures, as well as estimation of tortuosity in porous media.

Finally, an application of mass transfer around active solids is presented.
• Five applications of all the above theoretical concepts and experimental results
are discussed in Chap. 7 (Applications and Examples). These particular appli-
cations are: (a) Contaminant Plume Sizes Associated to Different Active Solids,
(b) Rising Damp in Building Walls, (c) Bubbles and Drops in Porous Structures,
(d) Transport Processes in Solid Oxide Fuel Cells (SOFCs), and (e) Multi Phase
Transport in Porous Media.
Each chapter is systematically detailed to be easily understood by a reader with
basic knowledge of fluid mechanics, heat and mass transfer, and computational and
experimental methods. Although much of the subject matter contained here is
available in research literature, journals and conference proceedings, this book
presents an overview of the most interesting developments over a wide variety of
the most important topics in transport phenomena in porous media. The funda-
mental concept for this book was to present the complete story of transport pro-
cesses in porous structures, and for this reason the chapters are self-contained,
citing separate references, and presenting their own mathematics, explanations and
terminology.
Reference
1. Lage, J.L., Narasimhan, A.: Porous media enhanced forced convection fundamentals and
applications. In: Vafai, K. (ed.) Handbook of Porous Media. Marcel Dekker, New York (2000)
Chapter 2
Fundamentals of Porous Structures
2.1 Introduction
The understanding of multiphase flow in porous media is of great importance in many

fields such as hydrology, contaminant clean-up and petroleum engineering. Macro-
scopic properties—principally capillary pressure and relative permeability—are often
needed when modelling flow and transport at the continuum scale, whether it is
transport of non-aqueous phase liquids (NAPL) in contaminant clean-up or the pro-
duction of oil during reservoir water flooding. These macroscopic properties are,
however, difficult to obtain. It is possible to conduct physical experiments on samples
of the reservoir, but this will only reflect one set of conditions. Furthermore, the scale
of the reservoir itself is so much larger that a few experiments are unlikely to describe
the variation likely to be present. Hence, there is a need to develop physically-based
models that can predict multiphase flow and transport properties and their likely
variation in a reservoir setting, based on readily available experimental data.
This chapter emphasizes the mathematical modelling of transport processes in
porous media by the cell models, the digital reconstruction of the porous structure
using the Lattice Boltzmann method and the stochastic modelling to obtain a
realistic description of the porous structure.
2.2 Cell Models
Mathematical modeling of transport processes in porous media is a powerful tool

employed whenever experimentation is either expensive or difficult due to the nature
of the process. On the other hand, the realistic description of the porous structure
significantly increases the complexity of the mathematics involved, due to the
coupling between the physicochemical mechanisms and the geometrical complexity
of the porous medium, thus requiring massive computational power. Given the poor

6 2 Fundamentals of Porous Structures
continuous
phase
spherical
grain
b a
GRANULAR POROUS MEDIUM

SPHERE-IN-CELL
Fig. 2.1 Schematic representation of sphere-in-cell model idealization (a and b denote the radii
of the inner and outer spheres, respectively)
computational power available sixty years ago, along with the industry-driven
increased necessity for research on porous media structure, in the late 1940s
modeling aimed towards the achievement of analytical solutions. The conceptual idea
behind this was the cell approach, where the medium is considered as an assemblage
of unit cells gathered in a regular manner. Accordingly, it was widely accepted that the
unit cell is the adequate representative of the whole medium, therefore processes
occurring through the porous structure are described sufficiently by those occurring in
the unit cell. As many industrial and technological applications are related with
swarms of cylindrical or spherical grains (filtration, separation, fibers, etc.) and the
analytical solution for the flow field around an isolated sphere was obtained by Stokes
one hundred years ago, what was introduced in those days were the sphere-in-cell
models, whose fundamental idea is schematically represented in Fig. 2.1.
Sphere-in-cell models are based on the representation of the overall solid mass
of the swarm by a spherical or cylindrical solid body, which is embedded in a
spherical or cylindrical liquid envelope, respectively. The boundary conditions
imposed on the outer surface of the envelope is supposed to adequately represent
the interactions with the other grains of the swarm. Obviously, the thickness of the
surrounding fluid layer is adjusted so the ratio of the solid volume to the volume of
the liquid envelope to represents exactly the solid volume fraction of the porous
medium. The main advantage of these models is that an analytical expression for
the stream function can be obtained demanding significantly less effort than that
needed for numerical investigations. The spherical shape corresponds to a for-
mulation which leads to axially symmetric flow that has a simple analytical
solution of closed form, and can thus be used readily for heat and mass transport
calculations. Although this analytical solution is actually an approximation of the
real flow field in a complex porous structure, it was sufficient for the engineering
applications raised until the early 1980s. On the other hand, the model has one
disadvantage in that the outer envelope is not space filling, a difficulty which must
be dealt with when a scale-up from the single unit cell to an assemblage of
particles is necessary.
2.2 Cell Models 7
The first introduction of sphere-in-cell model was by Cunningham [18]. When

considering particle sedimentation, Cunningham postulated that the movement of
each spherical particle was allowed only within a concentric mass of fluid boundary.
This model had problems in relating the size of the outer fluid sphere to the bed
voidage (or particle concentration), therefore Uschida proposed at 1959 an
improvement. He assumed that the porous medium is divided into numerous cubes
and a particle is placed into the center of each cube. Although space filling, this
approach failed to predict accurate flow field solutions due to discontinuities imposed
internally by the cubic schema. Richardson and Zaki [59] developed a hexagonal cell
model to predict the sedimentation velocity of uniform spheres under viscous flow
conditions. In this model, two configurations were considered depending on the
vertical particle arrangement. For the first, the distance between neighboring parti-
cles in the vertical direction was assumed to be equal to that in the horizontal
direction, while the second configuration assumed that vertical neighboring particles
are essentially in contact. Due, at least in part, to the oversimplification of the
pressure gradient and questionable assumptions regarding the angles between the
streamlines and the vertical axis, predictions of the sedimentation velocity from the
two configurations were not applicable to a dilute concentration of particles. Happel
[26] and Kuwabara [33] presented two independently devised sphere-in-cell models
to address the flow problem in sphere packing. The major differences between these
two models are in the boundary conditions imposed in the outer surface (A detailed
description can be found in Chap. 3). Despite their fundamental differences, both
models have been proved to yield very similar flow fields over a wide range of
porosity values [67]. At the same time, using formulations analogous to those of
spherical geometry, both Happel [26] and Kuwabara [33] also proposed cylinder-in-
cell models that consider particles of cylindrical instead of spherical shape. These
models were based on the cylindrical-cell model of Kawaguti [32] and found
applications in modeling flow through fibrous mats and arrays of parallel fibers.
Neale and Nader [45] proposed an important improvement over the aforementioned
models. They considered that the basic spherical cell is embedded in an unbounded,
continuous, homogeneous and isotropic permeable medium of the same porosity and
permeability as those of particle swarm. To describe the flow through the porous
surrounding, they used Brinkmans’s equation [8]. A decade later, a new spherical cell
model for predicting the bed expansion of a liquid–solid fluidized bed was presented
by Moritomi et al. [43]. They assumed that Stokes’ flow is valid for the fluid within
the outer cell boundary while outside the cell the flow is approximated by a potential
flow. However, prediction of the bed voidage did not agree well with their experi-
mental data. Another use of Brinkman’s equation in cell models is encountered
Prasad et al. [54], which replaces the solid sphere of the sphere-in-cell model with a
solid sphere surrounded by a concentric spherical shell of homogeneous and isotropic
porous material. This model combines features of those developed by Happel and
Kuwabara, and uses Brinkman’s equation to describe the flow through the porous
shell. The work of Prasad et al. is based on the work of Masliyah et al. [42] whose
‘‘solid-sphere-with-porous-shell’’ is embedded in an unbounded fluid instead of the
fluid envelope presented by Prasad et al. [54].
continuous
phase
spherical
grain
GRANULAR POROUS MEDIUM

SINUSOIDAL CELL
Fig. 2.2 Schematic representation of sinusoidal cell concept
The models described above all considered either spherical or cylindrical

grains, which corresponds to two-dimensional spherical geometry after some
physical assumptions and mathematical manipulations. However, grain shape is
closer to spheroidal than spherical for the majority of practical applications and
this is a crucial limitation for model accuracy. Epstein and Masliyah [22] solved
numerically the flow-field through clusters of spheroids under creeping flow
conditions, whereas the analytical solution of Ammar and Hsieh [4] for Stokes
flow inside an oblate hemispheroidal cap was the precursor for the work of Dassios
et al. [19, 20]. The latter presented a complete spheroid-in-cell model, quite
analogous to those of Happel and Kuwabara, and based on the introduced concept
of ‘‘semi-separation’’, an analytical formulation for the stream-function under
creeping flow conditions was obtained in terms of series expansions.
Depending on the application considered, several different non grain-type cell
models have been proposed and used to simulate the usually complex porous
structures. To model the transient behavior of deep-bed filtration systems, Payatakes
et al. [50] and Tien et al. [68] proposed the representation of the granular medium by
an ensemble of unit-bed elements (UBE), which was found to successfully describe
the initial stages of the filtration process. Each UBE is constructed of a number of
tube-type cells (collectors), sometimes surrounded by liquid envelopes [58].
Although the trend was to use straight capillaries for the sake of simplicity in the
calculations, representation of the pore space by sinusoidal cells, as suggested by
Payatakes et al. [51], seemed to produce results in good agreement with experimental
observations. The fundamental idea of sinusoidal cells is schematically represented
in Fig. 2.2.
The theoretical results of using UBEs for filtration simulations (Payatakes et al.
[52]) were found to be in good agreement with experimental data, thus indicating
that this simulator type could be used to provide an order-of-magnitude estimate
on a truly predictive basis for filtration processes. Some years later, Chiang and
Tien [12] used constricted tubes as collectors and employed the concept of UBE in
their analysis. Unfortunately, this model cannot describe a complete filter cycle, as
it does not deal with the problem of pore clogging and the simultaneous decrease
2.2 Cell Models 9
σ-phase σ-phase
σ-phase β-phase
γ-phase
γ-phase β-phase
β-phase
β-phase
β-phase
σ-phase σ-phase
POROUS MEDIUM 2-D UNIT CELL
γ-phase
β-phase
σ-phase
1-D UNIT CELL 3-D UNIT CELL
Fig. 2.3 Various unit cells
in permeability. Based on the UBE concept, Choo and Tien [14] presented another
approach and assumed that each element contained a number of cylindrical tubes
of various sizes. However, it was already known that the assumption of straight
capillary tubes oversimplifies pore structure, leading to a substantial underesti-
mation of the particle deposition rate in each cell [49].
Further to representation purposes, unit cells of various dimensions have been
introduced and used for the study of diffusion, convection, dispersion and inter-
facial transport in homogeneous porous media [3, 17, 55–57]. Some of those unit
cells are presented in Fig. 2.3.
2.3 Digital Reconstruction (Lattice Boltzmann Method)

of the Porous Structure
Transport in porous media is a topic related to many technological and environ-

mental applications. The accurate prediction of the transport coefficients in porous
media is a challenging problem due to the complexity of transport mechanisms in
fluid–solid systems and the difficulty in representing accurately the complicated

and tortuous nature of a porous medium.
The digital reconstruction of porous materials is a relatively new, powerful
methodology that enables the reliable representation of the complex biphasic
structure induced by porosity. The significant advances in both experimental and
numerical practices have allowed the development of reconstruction techniques,
the most advantageous being (a) the statistical methods leading to the stochastic
reconstruction of porous media in three dimensions based on statistical informa-
tion obtained from one or several two dimensional images of thin sections of the
actual material, and (b) the process-based methods, where the computational
procedure tries to imitate the physical processes that commence during the for-
mation of the medium. Although the latter seem to be closer to the physical
system, they frequently exhibit severe computational requirements and hence are
limited to the specific material considered in each case (Kainourgiakis et al. 31).
The Lattice Boltzmann method (LBM) constitutes a very powerful tool for the
study of the hydrodynamical problem of fluid flow inside porous structures, mainly
due to the simplified handling of the complicate boundary conditions, as well as,
due to the efficiency of the method with regard to parallelization [63, 65]. Lattice
Boltzmann method (LBM) is a mesoscopic approach for simulating computational
fluid dynamics by solving a discretized Boltzmann equation [11, 60, 71]. An
attractive feature of LBM is the ease of addressing complex boundary conditions
by implementing very simple schemes. Numerous works have successfully applied
LBM in modeling fluid flow in porous media and quantification of porous media
permeability [10, 27, 48, 60]. LBM models fluids as particle distributions residing
on a discrete lattice, propagating to their adjacent lattice nodes, and colliding with
other particles to redistribute momentum.
Aharonov and Rothman [2] first used the Lattice Boltzmann method to simulate
the flow of non-Newtonian fluids. Their pioneering work addresses two dimen-
sional pipes and random media. They found that the flux is related to the driving
force by a simple scaling law. Similar results were reported by Boek et al. [7].
Recently, Gabbanelli et al. [24] studied the flow of truncated power-law fluids in
reentrant flow geometries and found a very good agreement between the results
obtained by the Lattice Boltzmann method and those obtained by standard finite
element methods, while Sullivan et al. [66] explored the relationship between
lattice resolution and simulation accuracy as a function of the power-law index.
2.3.1 Porous Media Generation
Accurate numerical simulation of fluid flow in porous media requires detailed

descriptions of porous media morphology, which should include geometric
properties such as particle or pore shape and volume, and topological properties
such as pore interconnectivity. In many cases, however, the type of model that can
be employed is dependent on the modeling method, and more importantly, limited
2.3 Digital Reconstruction (Lattice Boltzmann Method) of the Porous Structure 11
computational resources. It is thus important to construct models that are able to

closely mimic the heterogeneity of actual porous media, and at the same time are
sufficiently efficient to allow simulation of flow and transport phenomena with
reasonable computational effort. In this study, porous media are envisioned as a
statistical distribution of non-overlapping circular disks representing soil particles
distributed in a rectangular two-dimensional uniform continuum representing the
pore space through which a fluid flows.
As first proposed by Gardner, particle size distributions in soil are often
assumed to be lognormal in nature (Lerman [36]). Buchan noted that approxi-
mately one-half of the US Department of Agriculture textual classification triangle
could be adequately modeled by a lognormal distribution. Since a standard log-
normal distribution implies zero and infinity for the smallest and largest particle
sizes, respectively, modified lognormal distributions were developed to constrain
the upper and lower extremes of the particle size. Fredlund et al. [23] proposed a
new model based on a unimodal mathematical function, which is believed to
provide improved representations of particle size distributions relative to lognor-
mal distributions. This model’s ease of use, however, is limited by its employment
of five fitting parameters; our study thus employs a modified lognormal distribu-
tion to describe particle size distribution, assuming that all particle sizes reside in a
95% confidence interval to eliminate extremely large or small particles.
Li et al. [37] modified the algorithm proposed by Yang et al. [72] for a three
dimensional sphere packing, a two-step collective rearrangement technique to
generate random porous media. In this work, the particles with size distributions
following a modified lognormal distribution are generated until the required
porosity is satisfied. The particles are then assigned to a two dimensional domain
by assuming a uniform distribution of particle locations. Based on this initial,
possibly overlapped configuration (i.e., one particle may overlap another particle),
an iterative arrangement process is applied to achieve an overlap free condition.
During each iterations the largest particle is selected for relocation if there is any
overlap with another particle; if overlap occurs, its spatial location is adjusted until
the overlap is removed, and then registered in the final non-overlap location. The
procedure continues with the next largest particle, etc. until all particles are
registered in their final non-overlap location. Periodic boundary conditions are
maintained at all boundaries throughout the iteration process. Figure 2.4 provides an
illustration of several of the generated random porous media employed in this study.
2.4 Stochastic Modeling
Most natural and biological phenomena, such as solute transport in porous media,
exhibit variability which can not be modeled using deterministic approaches. There
is evidence in natural phenomena to suggest that some observations cannot be
explained using models which give deterministic solutions. Stochastic processes
Fig. 2.4 Example of generated porous media
have a rich repository of objects which can be used to express the randomness
inherent in the system and the evolution of the system over time.
Transport processes are generally modeled by initial/boundary value problems
for a partial differential equation (PDE) describing the local balance of the con-
centration fields. However, during probabilistic analysis of flow and transport
properties in porous media, the uncertainties due to spatial heterogeneity of
governing parameters will be often taken into account. The definition of the
properties of porous media in space and time using the concept of random func-
tions provides means for (1) studying the inherent heterogeneity, (2) evaluating the
spatiotemporal variability of the properties, and (3) assessing the uncertainty
associated with their estimated values.
For example, if a small control volume is considered, a mass balance on the
solute, in one-dimension, without chemical reaction, leads to

@ @C @C
De uC ¼ ð2:1Þ
@x @x @t
where De is the diffusion coefficient, C is the mean solute concentration, u is the

interstitial velocity of fluid, and t the time. The velocity field u(x,t) and diffusion
coefficients De(x,t) instead account for the spatial heterogeneity of the porous
media. Within the frame of Scheidegger’s theory of dispersion in porous media the
2.4 Stochastic Modeling 13
components of the diffusion tensor are proportional with those of the velocity field
[61]. The velocity field is modelled by the solution of flow equation. For instance,
the stationary (in time) and divergence free flow through a saturated porous
medium, is governed by Darcy and continuity equations
K @H X n
@ui
ui ¼ ; ð2:2Þ
e @xi i¼1 @xi
where K is the isotropic hydraulic conductivity, e is the porosity supposed to be

constant and H is the hydraulic head. The solution of Eq. 2.2 describes flow
through isotropic porous media.
For example, the hydraulic conductivity, K, as a function of x, it is convenient
to use its logarithm Y(x) = lnK(x). This quantity can be inferred from field
experiments consisting of pumping tests performed in observation wells. In most
experimental settings the locations of the wells are chosen so that an optimum
sampling of concentration during tracer tests is obtained. These results in a sparse
and non-regular distribution of wells and, therefore interpolation techniques are
used to estimate spatial correlations of Y(x). Based on experimentally inferred
correlations, stochastic models are proposed. For example, it is assumed that Y is a
statistically homogeneous in space random function, normal distributed with
constant mean and exponential covariance function [13].
Consequently, Eq. 2.1 has random coefficients. The stochastic approach con-
siders ensembles of solutions of (2.1), corresponding to the ensemble of K field
realizations. Expectations and standard deviations of the concentration are com-
puted as ensemble averages. In many applications an upscaled transport model
with simpler structure (in general with constant coefficients) is used to describe the
process at the desired observation scale. An observable transport process corre-
sponds to a solution of (2.1) for a fixed realization of hydraulic conductivity.
Therefore, to obtain realistic predictions, ergodicity assessments have to complete
stochastic modelling. Ergodicity can be quantified, for instance, by root mean-
square distances between single realization solutions and the output of the upscale
model [64]. Technical issues related to the construction of the numerical solutions
for PDEs of form (2.1) also provide arguments for a stochastic approach.
2.4.1 The Pore Space
The pore space description can be generated directly using X-ray microtomogra-
phy (see [16]), where the rock is imaged at resolutions of around a few microns.
This can be done due to the differences in X-ray absorption of rock matrix and
void space. An example of a 3D image, of Fontainebleau sandstone, is shown in
Fig. 2.5a along with a 2D cross section in Fig. 2.6a. This method has the
advantage of directly reconstructing the pore space, but due to the specialized
scanners required, it is not readily available and also very costly. It is, however,
Fig. 2.5 Comparison between different 3D voxel representations of a Fontainebleau sandstone [6],
generated using different reconstruction techniques. The side length of each sample is 2.25 mm.
a X-ray microtomography. b Object-based modelling. c Gaussian field technique. d Simulated
annealing
very useful for validating numerical reconstruction techniques of the microstruc-

ture, described below.
Most stochastic reconstruction algorithms are based on threshold Gaussian field
techniques [1]. These algorithms are based on porosity and two-point correlation
functions, both of which can be readily obtained by image analysis of 2D thin
sections. The technique is similar to that used in geostatistics. A continuous cor-
related field is generated using Fourier transform methods and threshold to retrieve
the binary phases (pore space and matrix) with the correct porosity and correlation
function, as sketched in Figs. 2.5c and 2.6c. This method can also be extended to
include more phases, such as clays.
Realising that the earlier Gaussian based methods were not very good at
reproducing the underlying particulate structures of the porous media, as evident
from Fig. 2.6c, Yeong et al. [73] developed a stochastic method based on simu-
lated annealing, later extended by Manwart et al. [40]. Rather than being restricted
to one- and two-point correlation functions, the objective function used can be
made to match additional quantities such as multi-point correlation functions,
Fig. 2.6 Comparison between different 2D cross-sections of a Fontainebleau sandstone [6],

generated using different reconstruction techniques. The side length of each sample is 2.25 mm.
a X-ray microtomography. b Object-based modelling. c Gaussian field technique. d Simulated
annealing
lineal-path function or pore size distribution function to name a few. Since the
method is based on moving pore-space voxels around to minimize the objective
function, the correct porosity is always retained. They subsequently reconstructed
a 3D Fontainebleau and Berea sandstones from 2D sections using an objective
function based on both two-point correlation function and either lineal-path or pore
size distribution function, Figs. 2.5d and 2.6d. The estimated percolation proba-
bility was closer to the reference than when only using the two-point correlation
function. Incorporating more higher-order information into the objective function,
such as the local percolation probability, would most likely improve the recon-
struction further, but that would also increase the computational cost of the method
significantly.
Another technique is the use of object-based modelling whereby the actual

rock-forming process is simulated. Packing algorithms for spheres are well
established (see [1]) and algorithms for arbitrary shapes have also been presented
by several authors [15, 34, 35, 53]. An algorithm that takes into account the whole
rock-forming process where primary grain sedimentation is followed by com-
paction and diagenesis has been presented by Bakke and Øren [5] and later by Jin
et al. [30]. A grain size distribution is determined from image analysis of thin
sections. Grains are then randomly picked from this distribution during the sedi-
mentation process. Compaction is modelled by moving the centre of each grain
down in proportion to its original vertical position. Finally, diagenetic processes
are modelled. Quartz overgrowth is modelled by increasing the radii of the grains,
and clays are precipitated on quartz surfaces. The amount of diagenesis to include
is also determined from thin section analysis. Though this reconstruction method
does not guarantee that statistical geometrical parameters such as the two-point
correlation functions are honoured, it is assumed that these will be adequately
reproduced since the actual rock-forming process is simulated, Figs. 2.5b and 2.6b.
Biswal et al. [6] performed a quantitative comparison of three reconstruction
methods (Gaussian field, simulated annealing and object-based) on a Fontaine-
bleau sandstone, where the reference, containing 3,003 voxels, was obtained using
microtomography. They found that the object-based technique reproduced the two-
point correlation function reasonably well. When comparing connectivity using
local percolation probability it was clear that the object-based technique was far
superior to the statistical techniques. This result agrees with analysis performed by
other authors as Hilfer [28].
Recently several statistical reconstruction techniques have been suggested that
try to improve on the deficiencies of the Gaussian field techniques, while not being
as computationally demanding as simulated annealing. Thovert et al. [69] intro-
duced a method that is a hybrid between the statistical and object-based methods.
Their technique is statistical, using only porosity and two-point correlation func-
tion, but conditioned to an underlying granular structure, defined using a pore size
distribution derived from the two-point correlation function. This is based on a
Poissonian penetrable sphere model. They verified their method using a 3D
Fontainebleau sample and reported good agreement with the reference, even for
the percolation probability. Hilfer and Manswart [29] also developed a hybrid
between statistical and object-based method. Initially a close packing of spheres is
laid down. Matrix voxels are then randomly distributed in the pore space until a
prescribed porosity is attained. The configuration of the added matrix voxels are
then updated using simulated annealing, matching the two-point correlation
function. They also verified their method using a 3D Fontainebleau sample and the
local percolation probability was found to be significantly better than traditional
simulated annealing though not quite as good as a traditional object-based method.
For many types of porous media it might, however, be difficult to define the
structures that make up the matrix. In carbonate rocks post-depositional diagenesis
often completely dominates the matrix structure (see [39]), making object-based
techniques difficult to use. In field-scale reservoir characterization the same
problems are often experienced where there are multiple faces with distinctive
connected geometries; for example, in fluvial reservoirs. Gaussian field techniques
will not reproduce the channel connectivity whereas object-based methods rely on
the faces geometries being easily parameterized. Strebelle [62] suggested a
statistical algorithm in which the multiple-point statistics were inferred from
exhaustive 2D training images of equivalent reservoir structures and then used to
reconstruct the reservoir, adhering to any conditioning data. This method was
applied successfully to both fluvial and more complex patterned reservoirs. The
ability to reproduce any pattern makes this method highly attractive for recon-
structing complex porous media like carbonates. Okabe et al. [46] have used this
algorithm to reconstruct a 3D Fontainebleau sandstone from a 2D training image.
Although the granular structure is not as well reproduced as in object-based
methods, the local percolation probability is significantly better reproduced than
that achieved by other statistical methods such as Gaussian field techniques.
2.4.2 Transport Properties
When estimating single phase transport properties like absolute permeability it is

possible to conduct flow experiments directly on the 3D reconstructed sample.
Many authors have reported good agreement with experimentally obtained values
when solving the Stokes equation for single phase flow using finite difference
methods (see [1, 29]) or the Lattice-Boltzmann method [41]. The importance of
reproducing long-range connectivity in any reconstruction technique is clearly
evident from calculations of absolute permeability. Whereas object-based methods
reproduce permeability values within a few percentage points, traditional Gaussian
field techniques and simulated annealing typically under-predict permeability by
about a factor of ten for low porosity systems that are close to the percolation
threshold [29, 41]. For higher porosity systems, the statistical methods generally
perform somewhat better at reproducing the local percolation probability [40],
with permeability predicted within a factor of about five (see [47]).
Even single phase simulations conducted directly on the 3D reconstructed
sample are computationally very expensive. A common way to model both single
and multiphase flow quickly on large systems is to represent complex porous media
by an equivalent network of pores and connecting throats. Absolute permeability
has been successfully predicted, as already mentioned, by Bryant et al. [9] using this
method. Another approach is to skip the 3D reconstruction process and estimate
absolute permeability directly from the 2D thin sections. Lock et al. [38] computed
a distribution of fluid conductance’s by recoding pore areas and perimeter lengths.
From this a single effective conductance was found using effective medium theory,
resulting in a direct estimate of absolute permeability. This approach was applied to
a number of sandstone thin sections with predicted absolute permeability’s
generally within a factor of 2 of the measured values.
For multiphase flow it is possible to use Lattice-Boltzmann techniques

(see [11, 25]) to solve the transport equations on the 3D reconstructed sample.
Being computationally very demanding, their use in multiphase flow problems is,
however, limited to relatively small systems. As a consequence their applications
are more tailored towards understanding the fundamental physics of flow in
complex systems rather than for quantitative predictions.
A whole range of empirical or semi-empirical methods for predicting relative
permeability have suggested over the last five decades (see [21]). One of the most
widely used is found by combining the model by van Genuchten [70], relating
effective wetting saturation to capillary pressure, with the model for relative
permeability by Mualem [44]. Empirical fitting parameters are found by matching
the van Genuchten model to experimental capillary pressure. These parameters are
subsequently used for the relative permeability predictions. However, what most
of these models have in common is that they are only applicable for strongly
water-wet data, and many of them also rely on difficult to define parameters such
as tortuosity exponents.
References
1. Adler, P.M., Thovert, J.F.: Real porous media: local geometry and macroscopic properties.
Appl. Mech. Rev. 51, 537–585 (1998)
2. Aharonov, E., Rothman, D.H.: Non-newtonian flow (through porous media): a lattice
Boltzmann method. Geophys. Res. Lett. 20, 679–682 (1993)
3. Ahmadi, A., Quintard, M., Whitaker, S.: Transport in chemically and mechanically
heterogeneous porous media. Part V: two-equation model for solute transport with
adsorption. Adv. Water Resour. 22, 59–86 (1998)
4. Ammar, L., Hsieh, C.K.: Solution of the velocity and pressure fields for stokes flow inside an
oblate hemispheroidal cap. Eur. J. Mech. B/Fluids 10, 171–178 (1991)
5. Bakke, S., Øren, P.E.: 3-D pore-scale modeling of sandstones and flow simulations in the
pore networks. SPE. J. 2, 136–149 (1997)
6. Biswal, B., Manwart, C., Hilfer, R., Bakke, S., Øren, P.E.: Quantitative analysis of
experimental and synthetic microstructures for sedimentary rock. Phys. A: Stat. Mech. Appl.
273, 452–475 (1999)
7. Boek, E.S., Chin, J., Coveney, P.V.: Lattice Boltzmann simulation of the flow of non-
newtonian fluids in porous media. Int. J. Mod. Phys. B 17, 99–102 (2003)
8. Brinkman, H.C.: A calculation of the viscous force exerted by a bowing fluid on a dense
swarm of particles. Appl. Sci. Res. Al, 27–34 (1947)
9. Bryant, S.L., Mellor, D.W., Cade, C.A.: Physically representative network models of
transport in porous-media. AIChE J. 39, 387–396 (1993)
10. Cancelliere, A., Chang, C., Foti, E., Rothman, D.H., Succi, S.: The permeability of a random
medium: comparison of simulation with theory. Phys. Fluids A: Fluid 12, 2085–2088 (1990)
11. Chen, S., Doolen, G.D.: Lattice Boltzmann method for fluid flow. Ann. Rev. Fluid. Mech. 30,
329–364 (1998)
12. Chiang, H.W., Tien, C.: Dynamics of deep bed filtration: analysis of two limiting situations.
AIChE J. 31, 1349–1355 (1985)
13. Chiles, J.P., Delfiner, P.: Geostatisctics: Modeling Spatial Uncertainty. John Wiley and Sons,
New York (1999)
References 19
14. Choo, C.U., Tien, C.: Simulation of hydrosol deposition in granular media. AIChE J. 41,
1426–1431 (1995)
15. Coelho, D., Thovert, J.F., Adler, P.M.: Geometrical and transport properties of random
packings of spheres and aspherical particles. Phys. Rev. E 55, 1959–1978 (1997)
16. Coker, D.A., Torquato, S., Dunsmuir, J.H.: Morphology and physical properties of
fontainebleau sandstone via a tomographic analysis. J. Geophys. Res-Sol. Ea. 101,
17497–17506 (1996)
17. Coutelieris, F.A., Kainourgiakis, M.E., Stubos, A.K., Kikkinides, E.S., Yortsos, Y.C.:
Multiphase mass transport with partitioning and inter-phase transport in porous media. Chem.
Eng. Sci. 61, 4650–4661 (2006)
18. Cunningham, M.A.: On the velocity of steady fall of spherical particles through fluid
medium. In: Proceedings of the Royal Society, London, A83, pp 357–365 (1910)
19. Dassios, G., Hadjinicolaou, M., Payatakes, A.C.: Generalized eigenfunctions and complete
semi-separable solutions for stokes flow in spheroidal coordinates. Quart. Appl. Math. 52,
157–191 (1994)
20. Dassios, G., Hadjinicolaou, M., Coutelieris, F.A., Payatakes, A.C.: Stokes flow in spheroidal
particle-in-cell models with Happel and Kuwabara boundary conditions. Int. J. Eng. Sci. 33,
465–1490 (1995)
21. Dullien, F.A.L.: Porous Media: Fluid Transport and Pore Structure, 2nd edn. Academic, San
Diego (1992)
22. Epstein, N., Masliyah, J.H.: Creeping flow through clusters of spheroids and elliptical
cylinders. Chem. Eng. J. 3, 169–175 (1972)
23. Fredlund, M.D., Fredlund, D.G., Wilson, G.W.: An equation to represent grain-size
distribution. Can. Geotech. J. 37, 817–827 (2000)
24. Gabbanelli, S., Drazer, G., Koplik, J.: Lattice Boltzmann method for non-newtonian (power-
law) fluids. Phys. Rev. E 72, 046312–046319 (2005)
25. Gunstensen, A.K., Rothman, D.H.: Lattice-Boltzmann studies of immiscible 2-phase flow
through porous-media. J. Geophys. Res-Sol. Ea. 98, 6431–6441 (1993)
26. Happel, J.: Viscous flow in multiparticle systems: slow motion of fluids relative to beds of
spherical particles. AIChE J. 4, 197–201 (1958)
27. Heijs, A.W.J., Lowe, C.P.: Numerical evaluation of the permeability and Kozeny constant for
two types of porous media. Phys. Rev. E 51, 4346–4352 (1995)
28. Hilfer, R.: Local porosity theory and stochastic reconstruction for porous media. In: Mecke,
K.R., Stoyan, D. (eds.) Lecture Notes in Physics, 554th edn. Springer, Berlin (2000)
29. Hilfer, R., Manwart, C.: Permeability and conductivity for reconstruction models of porous
media. Phys. Rev. E 64, 021304–021307 (2001)
30. Jin, G., Patzek, T.W., Silin, D.B.: Physics-based reconstruction of sedimentary rocks. In:
Proceedings of the SPE Western Regional/AAPG Pacific Section Joint Meeting, SPE 83587,
Society of Petroleum Engineers, Long Beach, USA (2003)
31. Kainourgiakis, M.E., Kikkinides, E.S., Steriotis, T.A., Stubos, A.K., Tzevelekos, K.P.,
Kanellopoulos, N.K.: Structural and transport properties of alumina porous membranes from
process-based and statistical reconstruction techniques. J. Colloid Interf. Sci. 23, 158–167
(2000)
32. Kawaguti, M.: A hydrodynamical model for the sedimentation. J. Phys. Soc. Jpn. 3, 209–215
(1958)
33. Kuwabara, S.: The forces experienced by randomly distributed parallel circular cylinder or
spheres in a viscous flow at small reynolds numbers. J. Phys. Soc. Jpn. 14, 527–531 (1959)
34. Latham, J.P., Lu, Y., Munjiza, A.: A random method for simulating loose packs of angular
particles using tetrahedral. Geotechnique 51, 871–879 (2001)
35. Latham, J.P., Munjiza, A., Lu, Y.: On the prediction of void porosity and packing of rock
particulates. Powder Techn. 125, 10–27 (2002)
36. Lerman, A.: Geochemical Processes: Water and Sediment Environments. Wiley, New York
(1976)
37. Li, Y., LeBoeuf, E.J., Basu, P.K., Mahadevan, S.: Stochastic modeling of the permeability of
randomly generated porous media. Adv. Water Resour. 28, 835–844 (2005)
38. Lock, P.A., Jing, X.D., Zimmerman, R.W., Schlueter, E.M.: Predicting the permeability of
sandstone from image analysis of pore structure. J. Appl. Phys. 92, 6311–6319 (2002)
39. Lucia, F.J.: Carbonate Reservoir Characterization. Springer, Berlin (1999)
40. Manwart, C., Torquato, S., Hilfer, R.: Stochastic reconstruction of sandstones. Phys. Rev. E
62, 893–899 (2000)
41. Manwart, C., Aaltosalmi, U., Koponen, A., Hilfer, R., Timonen, J.: Lattice-Boltzmann and
finite-difference simulations for the permeability for three dimensional porous media. Phys.
Rev. E 66, 016702–016713 (2002)
42. Masliyah, J.H., Neale, G., Malysa, K., Van De Ven, G.M.: Creeping flow over a composite
sphere: solid core with porous shell. Chem. Eng. Sci. 42, 245–253 (1987)
43. Moritomi, H., Yamagishi, T., Chiba, T.: Prediction of complete mixing of liquid-fluidized
binary solid particles. Chem. Eng. Sci. 41, 297–305 (1986)
44. Mualem, Y.: A new model for predicting the hydraulic conductivity of unsaturated porous
media. Water Resour. Res. 12, 513–522 (1976)
45. Neale, G.H., Nader, W.: Prediction of transport processes within porous media: creeping flow
relative to a fixed swarm of spherical particles. AIChE J. 20, 530–538 (1974)
46. Okabe, H., Blunt, M.J.: Pore space reconstruction using multiple-point statistics. J. Petrol.
Sci. Eng. 46, 121–137 (2005)
47. Øren, P.E., Bakke, S.: Reconstruction of berea sandstone and pore-scale modeling of
wettability effects. J. Petrol. Sci. Eng. 39, 177–199 (2003)
48. Pan, C., Hilpert, M., Miller, C.T.: Pore-scale modeling of saturated permeabilities in random
sphere packings. Phys. Rev. E 64, 0066702-1–0066702-9 (2001)
49. Paraskeva, C.A., Burganos, V.N., Payatakes, A.C.: Three-dimensional trajectory analysis of
particle deposition in constricted tubes. Chem. Eng. Commun. 108, 23–48 (1991)
50. Payatakes, A.C., Tien, C., Turian, R.M.: A new model for granular porous media. Part I:
model formulation. AIChE J. 19, 58–67 (1973)
51. Payatakes, A.C., Tien, C., Turian, R.M.: Trajectory calculation of particle deposition in deep
bed filtration. Part I: model formulation. AIChE J. 20, 889–900 (1974)
52. Payatakes, A.C., Brown, D.H., Tien, C.: On the transient behavior of deep bed filtration. In:
Proceedings of the 83rd National AIChE Meeting, Houston, USA (1977)
53. Pilotti, M.: Reconstruction of clastic porous media. Transport. Porous Med. 41, 359–364
(2000)
54. Prasad, D., Narayan, K.A., Chhabra, R.P.: Creeping fluid flow relative to an assemblage of
composite spheres. Int. J. Eng. Sci. 28, 215–230 (1990)
55. Quintard, M., Whitaker, S.: One- and two-equation models for transient diffusion processes in
two-phase systems. Adv. Heat Transf. 23, 369–464 (1993)
56. Quintard, M., Whitaker, S.: Convection, dispersion, and interfacial transport of contaminants:
homogeneous porous media. Adv. Water Resour. 17, 221–239 (1994)
57. Quintard, M., Whitaker, S.: Transport in chemically and mechanically heterogeneous porous
media I: theoretical development of region-averaged equations for slightly compressible
single-phase flow. Adv. Water Resour. 19, 29–47 (1996)
58. Rajagopalan, R., Tien, C.: Trajectory analysis of deep bed filtration using the sphere-in-cell
model. AIChE J. 22, 523–528 (1976)
59. Richardson, J.F., Zaki, W.N.: The sedimentation of a suspension of uniform spheres under
conditions of viscous flow. Chem. Eng. Sci. 3, 65–73 (1954)
60. Rothman, D.H.: Cellular-automaton fluids: a model for flow in porous media. Geophysics 53,
509–518 (1988)
61. Scheidegger, A.E.: General theory of dispersion in porous media. J. Geophys. Res. 66,
3273–3278 (1961)
62. Strebelle, S.: Conditional simulation of complex geological structures using multiple-point
statistics. Math. Geol. 34, 1–21 (2002)
References 21
63. Succi, S.: The Lattice Boltzmann Equation for Fluid Dynamics and Beyond. Oxford
University Press, Oxford (2001)
64. Suciu, N., Vamos, C., Vanderborght, J., Hardelauf, H., Vereecken, H.: Numerical
investigations on ergodicity of solute transport in heterogeneous aquifers. Water Resour.
Res. 42, W04409–W04419 (2006)
65. Sukop, M.C., Thorne Jr, D.T.: Lattice Boltzmann Modeling an Introduction for Geoscientists
and Engineers. Springer, Berlin (2006)
66. Sullivan, S.P., Gladden, L.F., Johns, M.L.: Simulation of power-law fluid flow through
porous media using lattice Boltzmann techniques. J. Non-Newton. Fluid Mech. 133, 91–98
(2006)
67. Tien, C.: Granular Filtration of Aerosols and Hydrosols. Butterworths, USA (1989)
68. Tien, C., Turian, R.M., Pendse, H.P.: Simulation of the dynamic behavior of deep bed filters.
AIChE J. 25, 385–395 (1979)
69. Thovert, J.F., Yousefian, F., Spanne, P., Jacquin, C.G., Adler, P.M.: Grain reconstruction of
porous media: application to a low-porosity fontainebleau sandstone. Phys. Rev. E 63,
061307–061323 (2001)
70. Van Genuchten, M.T.: A closed-form equation for predicting the hydraulic conductivity of
unsaturated soils. Soil Sci. Soc. Am. J. 44, 892–898 (1980)
71. Wolf-Gladrow, D.A.: Lattice-Gas Cellular Automata and Lattice Boltzmann Models: An
Introduction. Lecture Notes in Mathematics. Springer, New York (2000)
72. Yang, A., Miller, C.T., Turcoliver, L.D.: Simulation of correlated and uncorrelated packing
of random size spheres. Phys. Rev. E 53, 1516–1524 (1996)
73. Yeong, C.L.Y., Torquato, S.: Reconstructing random media. Phys. Rev. E 57, 495–506
(1998)
Chapter 3
Flow in Porous Media
3.1 Introduction
As mentioned in Chap. 1, a porous medium is a particulate phase (usually solid) that

contains void spaces (microscopic pores). These pores may either be connected to
each other or unconnected, and are distributed in the medium in either a regular or a
random manner. Porous media can be distinguished as granular or fractured in
form with either consolidated or unconsolidated mechanical properties. In a con-
solidated porous medium, the particles (grains) are connected by an intermediate
cementing material, while in an unconsolidated porous medium the grains are loose.
The spatial distribution of matter in a porous medium can be typically repre-
sented by the phase function Z(x), defined as:
8
<1 x belongs to the pore space
Z ð xÞ ¼ ð3:1Þ
:
0 otherwise
where x is the position vector from an arbitrary origin.

The motion of a continuum is generally described by a usually linear relation
between some fluxes and the relative driving forces. Following the trend of deter-
minism, a large number of constitutive (phenomenological) equations describing
relationships between fluxes and driving forces exist in several fields of physics.
These include Newton’s law (which correlates developing forces with acceleration,
i.e. velocity gradient), Fourier’s law (which correlates heat flow with temperature
gradient), Fick’s law (which correlates mass flow with concentration gradient),
Ohm’s law (which correlates current with potential gradient), etc. The description of
flow in porous media is especially difficult due to the media’s complex geometry.
Furthermore, a general law of the continuum theory is the law of conservation of
extensive properties such as mass, momentum and energy. The resulting equations of
continuity are commonly referred to as field equations and must be considered along
with flow equations to adequately describe flow in porous structures.

24 3 Flow in Porous Media
Experimental and theoretical investigations are normally associated with the

scale of the application considered. In general, these scales are microscopic (or
‘‘pore’’ scale), mesoscopic (or ‘‘local’’ scale), and macroscopic (or ‘‘field’’ scale). As
far as the phenomena considered (and the governing equations as well as the relative
solutions) are strongly dependent on the scale, the transition between different scales
is of great importance. This transition is usually considered from the lowest ‘‘pore’’
level, where experimental and simulation results could be easily obtained, to the
higher ‘‘field’’ scale, where data are hard to be measured. This up-scaling process is
the subject of many studies (see Sects. 5.5 and 7.5).
3.1.1 Macroscopic Description
During his experiments on flow in pipes, Darcy found and proved that the pressure
drop caused by the flow is proportional to the velocity. This result is expressed by
the phenomenological Darcy law equation [7]:
k Dp
q¼A ð3:2Þ
l L
which generally describes the fluid flow through a porous medium. In the above
formula, q is the volumetric flowrate through a cross-section, A of the porous med-
ium, which is perpendicular to the flow direction, L is the length of the porous media
in direction of flow, Dp is the pressure difference along the porous medium, l is the
viscosity of the flowing fluid, and k is the permeability as a material property of the
porous medium. This equation can be written in differential form as:
l
rp ¼ u ð3:3Þ
k
where u denotes the velocity vector of the fluid. It is important to note that the
Darcy law is only valid for a laminar and steady state one-phase flow through a
porous medium. In addition, the fluid must be largely incompressible.
An extension to the Darcy law is the isotropic law introduced by Brinkman law [3]:
l
rp ¼ u þ leff r2 u ð3:4Þ
k
where leff is an effective viscosity which may be different from l: The Brinkman
law is a slight modification of the Stokes equation and one of its practical
advantages is that applications and analytical tools devoted to the Stokes approach
can be used with small adaptations.
One of the most significant applications of Darcy’s law (3.2) is the calculation
of permeability of a porous material. After solving the equation for k:
q L
k¼ l ð3:5Þ
A Dp
3.1 Introduction 25
the permeability of a porous media is defined as 1 Darcy, if a fluid of 1 cP viscosity,

flowing through a cross-section of 1 cm2 at a rate of 1 cm3/s, causes a pressure drop
of 1 atm/cm. Many years later, Klinkenberg observed that gas permeability is not
the same as liquid permeability for the same porous structure, since gas perme-
ability depends on pressure [16]. The correlation between gas permeability kgas ;
liquid permeability kliq and mean pressure inside the core pmean ; is given as:

b
kgas ¼ kliq 1 þ ð3:6Þ
pmean
where parameter b depends on the gas studied.

All the above calculations are valid under the major restriction that perme-
ability should be a number, which means that the porous medium is homogeneous
and isotropic. For non-homogeneous and anisotropic media, permeability becomes
a tensor, thus the above equations should be transformed accordingly.
3.1.2 Microscopic Description
Modeling approaches for microscopic conditions can be described by the appropriate

flow conditions equations (creeping, laminar, turbulent, etc.). It is important to note
that these equations are defined in the pore space, therefore the boundary conditions
in the solid–fluid interface are essential to solve the flow problem. Obviously,
equations of motion and the accompanying boundary conditions, are strongly
dependent on the application considered, thus many different approaches have been
proposed in the literature. The problem here is the derivation of representative
macroscopic quantities from the microscopic results, a procedure which hardly
allows for a generalized unique solution. As the extensive discussion of these issues
is not within the scope of this book, our presentation is limited to granular porous
media, where analytical solutions are also available through the cell models.
3.2 Analytical Solutions for Single Phase Flow in Cell Models
As mentioned in Chap. 2, cell models have been widely used for mathematical
simulation of the flow conditions through porous structures, especially granular
materials. According to these models, the grains are distributed throughout the
predefined space, and each grain is enclosed in a cell formed by the liquid phase.
The cell model can be used to reduce the solution of the boundary-value problem
for the flow around a system of particles to the problem for a single particle, where
the flow field solution can be obtained by the analytical solution of the Stokes
problem. The cell models differ in their boundary conditions, as discussed below.
A wide variety of assumptions exists for the shape of the grains, and the cells in
general: spherical, cylindrical, spheroidal, etc. Different approaches for the physical
conditions and mathematical formulations have also been presented for each geom-
etry considered, therefore producing an extensive range of combinations between
geometrical and physical considerations. These combinations are described below.
3.2.1 Sphere-in-Cell Models: Kuwabara’s Approach
Consider a solid sphere of radius a, surrounded by another concentric spherical

liquid envelope of radius b, whose thickness is adjusted so the porosity of the
medium is equal to that of the model. The internal sphere is solid and stationary
while a Newtonian fluid flows around the solid core under constant approaching
velocity. The governing equation for quasi-steady creeping flow of incompressible
(i.e. constant density) viscous (i.e. constant dynamic viscosity) fluids is the well-
known Stokes equation, given as:
rp ¼ lr2 v ð3:7Þ
along with the continuity equation:

rv¼0 ð3:8Þ
where p is the pressure field, v is the velocity vector, and l is the dynamic
viscosity. As the system is axially symmetrical, the problem is two-dimensional.
Accordingly, by using the spherical coordinates system ðr; hÞ; the above equation
can be written in terms of the steam-function w; as:

E2 E2 w ¼ 0 ð3:9Þ
where:
#2 sin h #2
E2 ¼ 2 ð3:10Þ
#r 2 r #h2
Consequently, the velocity components are given as:
1 #w 1 #w
ur ¼ ; uh ¼ ð3:11Þ
r2 sin h #h r sin h #r
To solve the above problem, it is necessary to consider the appropriate
boundary conditions. According to Kuwabara’s approach [17], zero vorticity is
assumed on the outer surface, as depicted in Fig. 3.1.
The velocity components are expressed as:

F1 F2
ur ¼ 2 3 þ þ F3 þ F4 r 2 cos h ð3:12Þ
r r

F1 F2 2
uh ¼ 3 2F3 4F4 r sin h ð3:13Þ
r r
3.2 Analytical Solutions for Single Phase Flow in Cell Models 27
Fig. 3.1 Kuwabara’s sphere- constant velocity

in-cell model
non-slip condition
zero vorticity
where:

U1 a3
F1 ¼ 12 3 ð3:14Þ
4F5 b
3U1
F2 ¼ ð3:15Þ
4F5

U1 a3
F3 ¼ 1þ 3 ð3:16Þ
2F5 2b

3U1 b3
F4 ¼ ð3:17Þ
20F5 a3

a 3 6b 3b2 b3
F5 ¼ 1 1þ þ 2 þ 3 ð3:18Þ
b 5a 5a 5a
and where U1 is the magnitude of the uniform approaching velocity.

Typical results for the stream function are presented in Fig. 3.2 which also
depicts a comparison with Happel’s approach (for details on Happel’s model, see
Sect. 3.2.2).
3.2.2 Sphere-in-Cell Models: Happel’s Approach
The model proposed by Happel [10, 11] is similar to that of Kuwabara but differs
in the fluid motion. Kuwabara’s stationary grain is located in a flowing fluid,
however, Happel proposed a grain moving under constant velocity in an otherwise
Fig. 3.2 Sample stream lines

in two sphere-in-cell models Kuwabara Happel
quiescent surrounding fluid. This dissimilarity imposes different boundary condi-

tions, as depicted in Fig. 3.3.
Compared to Kuwabara’s approach, Happel’s model has the additional advantage
of being autonomous from an energy point of view. This benefit is the main reason for
the wider acceptance of Happel’s approach rather than Kuwabara’s model.
In Happel’s approach, the governing equations are the same as (3.7–3.11),
where it is assumed that the solid spherical core is moving under a constant
velocity gradient q. After all, the velocity components are expressed as [10]:

6 3
ur ¼ 6r 3 F1 þ 2rF2 þ 2 F3 4 F4 þ rq sin h cos h ð3:19Þ
r r

3 1 rq 2
uh ¼ 5r F1 þ rF2 þ 4 F4 þ cos h sin2 h ð3:20Þ
r 2
where:
0 7 1
a
5q BB b C
C
F1 ¼ 2 B 7 CF5 ð3:21Þ
4a @ a A
10 þ 4
b
0 7 1
a
4 þ 10
5q B
B b C
C q
F2 ¼ B 7 CF5 ð3:22Þ
4 @ a A 2
10 þ 4
b
5qa3
F3 ¼ F5 ð3:23Þ
12
Fig. 3.3 Happel’s sphere-in-

cell model
constant velocity
zero shear stress
5a5
F4 ¼ 7 F5 ð3:24Þ
a
10 þ 4
b
7
a
10 þ 4
b
F5 ¼ " 10 # 3 " 4 # ð3:25Þ
a a a
10 1 25 1
b b b
The visual interpretation of sample flow in Happel’s sphere-in-cell is depicted

in Fig. 3.2.
3.2.3 Cylinder-in-Cell Models
Further to ‘‘sphere-in-cell’’ models, both Happel and Kuwabara also proposed

similar ‘‘cylinder-in-cell’’ ones. The similarity in the formulations and the con-
sequent analytical solutions is due to the two-dimensionality of the mathematical
description that allows us to ignore the shape imposed by the third dimension.
3.2.4 Spheroid-in-Cell Model
As described above, grain are often closer in shape to spheroids rather than spheres.
This observation leads to the development of ‘‘spheroidal-in-cell’’ models, similar to
those of Happel and Kuwabara. One difficulty of such a mathematical interpretation
arises from the orientation of the spheroidal geometry, which can be either prolate or
oblate, as illustrated in Fig. 3.4.
prolate
oblate
Fig. 3.4 Prolate and oblate spheroidal cells
In cases of prolate geometry, the inner solid spheroid has long semiaxis a3
and short semiaxis a1 ; where a3 [ a1 : The semifocal distance a is defined as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a ¼ a23 a21 : The outer confocal prolate spheroid is of long and short semiaxes
b3 and b1 ; respectively, while its dimensions are determined so that the volume
fraction of the spheroid-in-cell is equal to that of the original swarm of spheroidal
particles. Thus, given the dimensions of the inner spheroid, porosity is related to
the outer semiaxes by the following equations:
ð1 eÞb21 b3 ¼ a3 ð3:26Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
b1 ¼ b23 a2 ð3:27Þ
The axisymmetric, steady state, incompressible, viscous creeping flow is

again described in terms of stream function by Eq. 3.9, where the operator E2 is
given as:

2 1 2 o2 2 o o2 o
E ¼ 2 coth g 2 coth g þ 2 cot h ð3:28Þ
a sinh2 g þ sin2 h og og oh oh
Under this respect, the velocity components are given in the prolate spheroidal
coordinates system ðg; hÞ as [20]:
1 ow
ug ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:29Þ
a2 sinh g þ sin h sinh g sin h oh
2 2
1 ow
uh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:30Þ
a2 sinh g þ sin h sinh g sin h og
2 2
By using the semi-separation concept introduced by Dassios et al. [8], the

stream function is analytically given by an infinite series expansion of the form:
X
1
wðg; hÞ ¼ gn ðcosh gÞGn ðcos hÞ ð3:31Þ
n¼2;4;...
The leading term of the above infinite series has proved to be sufficient enough
to adequately represent the whole stream function [4], thus the stream function is
given as:

wðg; hÞ ¼ A1 G1 cosh g þ A2 G2 cosh gÞ þ A3 G4 cosh g
ð3:32Þ
þ A4 H2 cosh gÞgG2 ðcos hÞ
where A1 ; A2 ; A3 and A4 are constants coefficients, the values of which depend on the
geometrical parameters ga and gb (the values of the coordinate g on the inner and
outer surface, respectively). Obviously, their expressions are also dependent on the
model considered (Kuwabara or Happel), i.e. on the boundary conditions applied.
More specifically, Kuwabara’s approach corresponds to the expression:
( " #
a 5G4 cosh gb
wðg; hÞ ¼ K2 G2 ðcosh gÞ þ K3 G1 ðcosh gÞ þ G4 ðcosh gÞ
D G1 cosh gb
)
þ K4 H2 ðcosh gÞ G2 ðcos hÞ ð3:33Þ
where D, K2 ; K3 and K4 are g and h dependent coefficients, defined by Dassios

et al. [9], and GN ð xÞ and HN ð xÞ are the Gegenbauer polynomials of the first and
1
second kind, respectively, of degree and order N.
2
The Happel approach results in the algebraic linear system [9]:
2 3
G1 ðcosh ga Þ G2 ðcosh ga Þ G4 ðcosh ga Þ H2 ðcosh ga Þ
6 G0 ðcosh g Þ G02 ðcosh ga Þ G04 ðcosh ga Þ H20 ðcosh ga Þ 7
6 1 a 7
6 7
6 G1 cosh gb G2 cosh gb G4 cosh gb H2 cosh gb 7
6 7
6 7
6 KG001 cosh gb KG002 cosh gb KG004 cosh gb KH200 cosh gb 7
4 5

þKG01 cosh gb þKG02 cosh gb þKG04 cosh gb þKH20 cosh gb
2 3 2 3
A1 MG2 ðcosh ga Þ
6 A 7 6 MG ðcosh g Þ 7
6 27 6 1 a 7
6 7¼6 7
4 A3 5 4 0 5
A4 0
ð3:34Þ

in two identical prolate Kuwabara Happel
spheroids-in-cell
axis ratio = 2
porosity = 0.95
where:
1
K ¼ cosh2 gb ð3:35Þ
5
K ¼ 2 cosh gb ð3:36Þ
2
M¼ 2
ð3:37Þ
cosh ga 1
and where the primes and double primes denote the first and second derivatives of
the corresponding functions, respectively.
Some typical results for the flow field in a spheroidal-in-cell are shown in
Fig. 3.5.
The spheroid is oblate when a3 \a1 ; where the semifocal distance is given by
a ¼ a21 a23 : As shown by various researchers (see, for example, Happel and
Brenner [12]) the stream function for the oblate case can be obtained from the
same formulation as above, by using the following transformation from the prolate
coordinates system ðg; hÞ to the oblate one ð
g; hÞ:
cosh g ¼ i cosh
g ð3:38Þ
a ¼ i
a ð3:39Þ
Following this, the solution for stream function can be easily obtained; some
typical results of which are shown in Fig. 3.6.
3.3 Single Phase Flow in Granular Structures 33

in two identical oblate Kuwabara Happel
spheroids-in-cell
axis ratio = 0.5

porosity = 0.95
3.3 Single Phase Flow in Granular Structures
In the presence of porous media, the need for a realistic description of the structure of a
porous medium significantly increases the mathematical complexity of a model. As
porous media are generally characterized by highly complex internal geometry,
appropriate modeling is necessary for the derivation of meaningful conclusions.
Numerous industrial and technological applications involving fluid flow and mass
transport processes within multi-particle assemblages have attracted scientific interest
in the last decades, mainly focusing on industrial physicochemical processes (sedi-
mentation, catalysis, etc.), alternative energy sources (fuel cells, etc.), and separation
techniques (chromatography, filters, etc.). Although arrays of regularly spatially dis-
tributed spheres represent an idealization of real granular media, they have been widely
studied from both the fluid dynamics and mass transport points of view [19, 21]. On the
other hand, due to their complex geometry, random particle distributions were the
subject of rather limited investigations until the 1990s (see [2, 18]). Since then, fast
advances in computational capabilities have contributed to reviving interest on this
topic with emphasis placed on hydrodynamic aspects [5, 13, 22].
3.3.1 Representation of 3-D Sphere Assemblages
To define a realistic domain to solve flow and transport problems, a granular porous
medium was constructed in the form of a spherical particle assemblage. The structure
was digitized and the phase function (equal to zero for solid and unity for the pore
space) was determined to obtain the specified porosity. More specifically, repre-
sentation of the domains under consideration was achieved as follows:
Step (1) Using a random number generator, the position of the sphere’s center was
selected, being in a box of specified dimensions (3 9 2 9 3 mm)
Step (2) Using a random number generator, a radius assumed to follow the log-
normal distribution was selected
Fig. 3.7 A three-

dimensional representation
of a porous medium
Step (3) The void space around the sphere was checked. If it was free the radius
value was accepted, otherwise Step 2 was repeated
Step (4) The sphere was posed
Step (5) Steps 1–4 were repeated until the volume of the positioned spheres
satisfied the pre-defined porosity value.
Obviously, many three-dimensional representations can be generated by the
above algorithm for a specific porosity value. Therefore, it is necessary to validate
the results against these configurations to ensure that the macroscopic results are
independent of each specific realization. A graphic representation of a repre-
sentative medium for a typical porosity of e = 0.43 is presented in Fig. 3.7.
A randomly selected two-dimensional cut of this domain in is shown in
Fig. 3.8, which also depicts the grid of the numerical solution.
3.3.2 The Flow Field
For the numerical simulations, the velocity field was computed numerically by
solving the Stokes equations
rp ¼ lr2 v ð3:40Þ
rv¼0 ð3:41Þ
v ¼ 0 at liquidsolid interface ð3:42Þ

Fig. 3.8 A selected two-

dimensional cut of the
simulated geometry
discretized by an unstructured
grid
where v, p, and l are the velocity vector, pressure field, and fluid viscosity,
respectively.
The procedure for solving the Stokes flow problem involves discretization in
terms of cubic elements as follows [1, 6, 14, 15]: At the pore level, a staggered
marker-and-cell (MAC) mesh is used, with the pressure defined at the center of the
cell, and the velocity components defined along the corresponding face bound-
aries. The resulting linear system of equations is solved by a successive over-
relaxation (SOR) method. An initial estimate of p is determined by solving the
Laplace equation. Next, the velocity vector v is calculated from the corresponding
momentum balance and the continuity equation r v ¼ 0: The pressure is cor-
rected through an artificial compressibility equation of the form:
dp
¼rv ð3:43Þ
dt
Essentially, the method adds an artificial density time derivative related to the
pressure by an artificial equation of state p = bq, where b is an artificial
compressibility factor. Similar to the compressible momentum equation, c = b1/2
is an artificial speed of sound and for stability during the iterative procedure,
its magnitude should be such that the respective artificial Mach number,

P 2 1=2
M ¼ Rc max ui is low (M 1), where R is the relevant Reynolds number.
D i
In the limiting case of R ? 0, which is the present case, any finite value of b
should meet this criterion. Thus, b = 1 was selected although it is evident that the
Fig. 3.9 Snapshot of the

a pressure field, b velocity,
and c stream line simulations
through the representative
porous medium. The flow
direction is from left to right
exact value cannot have any effect on the final (steady state) results, since at steady
state the artificial density time derivative is equal to zero.
The above steps are repeated until convergence is reached. This numerical
scheme for the determination of the velocity field has been widely validated in
terms of both the velocity field and the corresponding permeability [1, 6, 14, 15].
3.3.3 Results and Discussion
Figure 3.9 show the results of the (a) pressure field, (b) velocity, and (c) stream
function for a typical porous medium of e = 0.72. A randomly selected two-
dimensional cut of the medium is considered so the results can be visualized
clearly. The boundary condition at the closed walls is non-slip, at the left
boundary an inflow was imposed, and at the right boundary an outflow condition.
Small vortices and recirculating flow are produced in the medium depending on
the pore size, while smoother profiles are obtained at the inlet and outlet. The
velocity gradient observed from top to bottom at the inlet surface is generated
because the inflow condition ensures constant molar flux instead of a plug-type
velocity vector. Finally, it is interesting to observe that the faster flow paths are
generated by the porous structure and they appear wherever pore diameters are
fairly small.
References
1. Adler, P.M., Jacquin, C.J., Quiblier, J.A.: Flow in simulated porous media. Int J Multiphas
Flow 16, 691–712 (1990)
2. Berryman, J.G.: Effective conductivity by fluid analogy for a porous insulator filled with a
conductor. Phys. Rev. B 27, 7789–7792 (1983)
3. Brinkman, H.C.: A calculation of the viscous force exerted by a flowing fluid on a dense
swarm of particles. Appl. Sci. Res. A 1, 27–34 (1949)
4. Burganos, V.N., Coutelieris, F.A., Dassios, G., Payatakes, A.C.: On the rapid convergence of
the analytical solution of stokes flow around spheroids-in-cells. Chem. Eng. Sci. 50,
3313–3317 (1995)
6. Coutelieris, F.A., Kainourgiakis, M.E., Stubos, A.K.: Low peclet mass transport in
assemblages of spherical particles for two different adsorption mechanisms. J. Colloid
Interface Sci. 264, 20–29 (2003)
7. Darcy H (1857) Recherches Experimentales Relatives au Mouvement de l’Eau dans les
Tuyaux, Mallet-Bachelier, Paris
semiseparable solutions for stokes flow in spheroidal coordinates. Quart. Appl. Math. 52,
157–191 (1994)
9. Dassios, G., Hadjinicolaou, M., Coutelieris, F.A., Payatakes, A.C.: Stokes flow in spheroidal
particle-in-cell models with Happel and Kuwabara boundary conditions. Int. J. Eng. Sci. 33,
1465–1490 (1995)
10. Happel, J.: Viscosity of suspensions of uniform spheres. J. Appl. Phys. 28, 1288–1292 (1957)
11. Happel, J.: Viscous flow in multiparticle systems: Slow motion of fluids relative to beds of
spherical particles. AIChE J. 4, 197–201 (1958)
12. Happel, J., Brenner, H.: Low Reynolds Number Hydrodynamics. Prentice-Hall, Englewood
Cliffs (1965)
13. Jullien, R., Meakin, P.: Simple three-dimensional models for ballistic deposition with
restructuring. Europhys. Lett. 4, 1385–1390 (1987)
14. Kainourgiakis, M.E., Kikkinides, E.S., Stubos, A.K.: Diffusion and flow in porous domains
constructed using process-based and stochastic techniques. J. Porous. Mat. 9, 141–154 (2002)
15. Kikkinides, E.S., Burganos, V.N.: Permeation properties of three-dimensional self-affine
reconstructions of porous materials. Phys. Rev. E 62, 6906–6915 (2000)
16. Klinkenberg, L.J.: The permeability of porous media to liquids and gases, drilling and
production practice. Am. Petr. Inst. V, 200–213 (1941)
17. Kuwabara, S.: The forces experienced by randomly distributed parallel circular cylinder or
spheres in a viscous flow at small Reynolds numbers. J. Phys. Soc. Jpn. 14, 527–532 (1959)
18. Meakin, P., Skjeltorp, A.T.: Application of experimental and numerical models to the physics
of multiparticle systems. Adv. Phys. 42, 1–127 (1993)
19. Meredith, R.E., Tobias, C.W.: Resistance to potential flow through a cubical array of spheres.
J. Appl. Phys. 31, 1270–1273 (1960)
20. Moon, P., Spencer, D.E.: Field Theory Handbook: Including Coordinate Systems,
Differential Equations, and Their Solutions, 2nd edn. Springer, Berlin (1971)
21. Rayleigh, R.S.: On the influence of obstacles arranged in rectangular order upon the
properties of a medium. Philos. Mag. 34, 481–502 (1892)
22. Visscher, W.M., Bolsterli, M.: Random packing of equal and unequal spheres in two and
three dimensions. Nature 239, 504–507 (1972)
Chapter 4
Transport Phenomena in Porous
Structures
4.1 Introduction
The problem of solute dispersion during underground water movement has

attracted interest from the early days of the last century [127], but it was only since
the 1950s that the general topic of hydrodynamic dispersion, or miscible dis-
placement, became the subject of more systematic study. This topic has interested
hydrologists, geophysicists, petroleum and chemical engineers, among others, and
for some time now it is treated at length in books on flow through porous media
(e.g. [9, 116]). Some books on chemical reaction engineering (ex: [24, 55, 141])
treat the topic of dispersion (axial and lateral) in detail and it is generally observed
that data for liquids and gases do not overlap, even in the ‘‘appropriate’’ dimen-
sionless representation.
Since the early experiments of Slichter [127] and particularly since the analysis
of dispersion during solute transport in capillary tubes, developed by Taylor [133]
and Aris [5, 6], much work has been done on the description of the principles of
solute transport in porous media of inert particles (ex: soils) and in packed bed
reactors (see [9, 44]).
Gray [59], Bear [9] and Whitaker [143] derived the proper form of the transport
equation for the average concentration of solute in a porous medium, by using the
method of volume or spatial averaging, developed by Slattery [126].
Brenner [20] developed a general theory for determining the transport prop-
erties in spatially periodic porous media in the presence of convection, and showed
that dispersion models are valid asymptotically in time for the case of dispersion in
spatially periodic porous media, while Carbonell and Whitaker [26] demonstrated
that this should be the case for any porous medium. These authors presented a
volume–average approach for calculating the dispersion coefficient and carried out
specific calculations for a two-dimensional spatially periodic porous medium.
Eidsath et al. [48] have computed axial and lateral dispersion coefficients in
packed beds based on these spatially periodic models, and have compared the

40 4 Transport Phenomena in Porous Structures
results to available experimental data. The axial dispersion coefficient calculated

by Eidsath et al. [48] shows a Peclet number dependence that is too strong, while
their radial dispersion. However, in soils or underground reservoirs, large scale
non-uniformities lead to values of dispersion coefficients that differ much from
those measured in packed beds, and for these cases spatially periodic models
cannot be expected to provide excellent results without modifications.
There have been other attempts at correlating and predicting dispersion coef-
ficients based on a probabilistic approach [38, 60, 70, 114] where the network of
pores in the porous medium is regarded as an array of cylindrical capillaries with
parameters governed by probability distribution functions.
Dispersion in porous media has been studied by a significant number of
investigators; using various experimental techniques. However, measurements of
axial and lateral dispersion are normally carried out separately, and it is generally
recognised that ‘experiments on lateral dispersion are much more difficult to
perform than those on axial dispersion’ [116].
When a fluid is flowing through a bed of inert particles, one observes the
dispersion of the fluid in consequence of the combined effects of molecular dif-
fusion and convection in the spaces between particles. Generally, the dispersion
coefficient in axial direction is superior to the dispersion coefficient in radial
direction by a factor of 5, for values of Reynolds number larger than 10. For low
values of the Reynolds number (say, Re\1Þ; the two dispersion coefficients are
approximately the same and e qual to molecular diffusion coefficient.
The detailed structure of a porous medium is greatly irregular and just some
statistical properties are known. An exact solution to characterize flowing fluid
through one of these structures is basically impossible. However, by the method of
volume or spatial averaging it is possible to obtain the transport equation for the
average concentration of solute in a porous medium [9, 143].
At a ‘‘macroscopic’’ level, the quantitative treatment of dispersion is currently
based on Fick’s law, with the appropriate dispersion coefficients; cross stream
dispersion is related to the radial dispersion coefficient, DT ; whereas stream-wise
dispersion is related to the axial dispersion coefficient, DL :
If a small control volume is considered, a mass balance on the solute, without
chemical reaction, leads to

@2C 1 @ @C @C @C
DL 2 þ DT r u ¼ ð4:1Þ
@z r @r @r @z @t
where C is the mean solute concentration, u ð¼ U=e; where U is the superficial

velocity and e the porosity of the porous media of inert particles with diameter d)
the mean interstitial velocity of fluid and t the time.
A large number of theories, namely the theories based on a probabilistic
approach, have been proposed to explain dispersion in porous media; however, the
theory of Saffman [114], who modelled the microstructure of a porous media as a
network of capillary tubes of random orientation, and Koch and Brady [83] were
the most referred.
4.2 Diffusion 41
4.2 Diffusion
Diffusion in porous media is a general subject that involves many fields of

research, such as chemistry (e.g. porous catalytic pallets), biology (e.g. porous
cellular organelles), and materials science (e.g. porous polymer matrixes for
controlled-release and gas-storage materials).
Diffusivity or diffusion coefficient is a proportionality constant between the
molar flux due to molecular diffusion and the gradient in the concentration of the
species (or the driving force for diffusion). Diffusivity is encountered in Fick’s law
and numerous other equations of physical chemistry.
The effective diffusion coefficient, De, describes diffusion through the pore space
of porous media. It is macroscopic in nature, because it is not individual pores but the
entire pore space that needs to be considered. Under steady state conditions the mass
flux, F, depends on the concentration gradient and is expressed by Fick’s first law:
dC
F ¼ De ð4:2Þ
dx
The effective diffusion coefficient for transport through the pores is estimated as
follows:
Dm ed
De ¼ ð4:3Þ
s
where s and d are dimensionless factors accounting for tortuosity ([ 1) and
constrictivity (B1) of the pores, respectively, and e is the effective porosity which
accounts for the reduced cross-sectional area available for diffusion when diffusion
occurs only in the pore space. The constrictivity describes the slowing down of
diffusion by increasing the viscosity in narrow pores as a result of greater prox-
imity to the average pore wall. It is a function of pore diameter and the size of the
diffusing particles.
4.2.1 Constrictivity Factor
The constrictivity factor d depends on the ratio of the solute diameter to the pore
diameter (kp):
molecule diameter
kp ¼ ð4:4Þ
pore diameter
may be quantified by empirical equations, as developed by Beck and Schultz [11]
and Chantong and Massoth [28], respectively:
4
d ¼ 1 kp ð4:5Þ

d ¼ 1:03 exp 4:05kp ð4:6Þ
4.2.2 Tortuosity Factor
The tortuosity factor, s, in Eq. 4.3 accounts for the pore geometry and is defined as
the square of the ratio of the effective path length, le, in the pore to the shortest
distance, l, in a porous medium:
2
le
s¼ ð4:7Þ
l
Tortuosity is not a physical constant and depends first of all on other porous
media characteristics, like porosity, pore diameter, channel shape, etc. In general,
in granular packings or beds the value of tortuosity lies in the region 1.1–1.7 (see
[9] and Dullien [43]), and our experimental results obtained with packed beds of
narrow size particles are in this range.
Several empirical correlations, suggesting a relationship between tortuosity and
porosity, have been found in the literature since the end of the nineteen century:
s ¼ 1:5 0:5e Ref: ½113 ð4:8Þ
s ¼ en Ref: ½4 ð4:9Þ
s ¼ 1 0:5 lnðeÞ Ref: ½168 ð4:10Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s¼ 1 lnðe2 Þ Ref: ½22 ð4:11Þ
s ¼ 1 0:41 lnðeÞ Ref: ½42 ð4:12Þ
pffiffi
s ¼ 1= e Refs: ½27; ½113 ð4:13Þ
Recently, Yun et al. [148] presented a theoretical geometry model for tortuosity
of tortuous streamtubes in a porous media with spherical particles. The authors
suggested an average value of tortuosity given by
s1 þ s2
s¼ ð4:14Þ
2
4.2 Diffusion 43
with
h pffiffi i pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffi 1

3ðp 2Þ 2 þ 3P2 =4 þ 3Pþarcsinð 3ð2 þ PÞ=2Þ
1þ þ pffiffi
6 þ 3P 3ð2 þ PÞ=2
s1 ¼
4
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð4:15Þ
i2ffi
1 1
h i2 þ h
1 pffi 1
1 pffi 1
3ð2 þ PÞ=2 3ð2 þ PÞ=21
þ
4

P1 p2 P1
s2 ¼ 1 1þ þ ð4:16Þ
P1 þ 1 P1 þ 2 P1 þ 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi ffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where P ¼ 2p 3ð1 eÞ 2 and P1 ¼ p=ð1 eÞ 2.
Equations 4.8–4.14 all satisfy the condition s ¼ 1 for e ¼ 1; and this is con-
sistent with the physical situation observed. Yun et al. [148] and Sen et al. [120]
showed that for an isotropic medium with spherical particles the tortuosity of
porous and granular media decreases with increasing bed voidage and increase for
non-spherical particles.
4.2.3 Porosity Factor
The overall porosity, e, of porous media can be determined (pore size distribution
and tortuosities are unknown). Therefore the relative diffusivity ðD ¼ De =Dm Þ is
often defined as an empirical function of e alone (d = 1):
De ¼ Dm em ð4:17Þ
where m is an empirical exponent.
Archie [4] found that the exponent m varied between 1.8 and 2.0 in consolidated
materials. In unconsolidated sand he found a value of 1.3. For an isotropic packing of
spherical particles a theoretical value of m = 3/2 was derived [21]. Adler et al. [1]
reported a value of m = 1.64 for Fontainebleau sandstone. Probst and Wohlfahrt
[107] found that m equals 1.43 for loose packings of catalyst particles and Millington
and Quirk [96] reported a value of 4/3 for diffusive flow of gases at normal pressures
or diffusion of ions in solution in soils. In experiments on the diffusion of gases
through compacted sands a value of m = 1.5 was determined [122].
4.2.4 Diffusion in Semi-Infinite Porous Media
In the limit of very low fluid velocity, where u ! 0; dispersion is determined

solely by molecular diffusion, with DT ¼ DL ¼ Dm =s ¼ De : Diffusion may be the
dominant mass transfer mechanism (compared to advection) in zones of low
hydraulic conductivity. The geometry of these layers may be considered as plane

sheets and analytical solutions of Fick’s second law are available for a variety of
initial and boundary conditions [36].
Thick confining layers and bedrock formations of low permeability may rep-
resent a sink for contaminants which are spread out in an aquifer or a landfill. On
the other hand, once such low-permeability domains are contaminated, they may
become a long-term source during remediation of the aquifer (e.g. pump-and-
treat). These formations can be considered as semi-infinite media for diffusion. If
the low conductivity zone is free of the contaminant initially and then exposed to a
constant concentration at the surface for a given period of time, the initial and
boundary conditions are:
t¼0 x[0 C ¼ C1 ð4:18aÞ
t[0 x¼0 C ¼ C0 ð4:18bÞ
t[0 x!1 C ¼ C1 ð4:18cÞ

The concentration profile at a given time is:

C C1 x
¼ 1 erf pffiffiffiffiffiffiffi ð4:19Þ
C0 C1 2 De t
pffiffiffiffiffiffiffi
The quantity 2 De t ¼ dp can be considered as ‘‘penetration depth’’ which rep-
resents the time dependent distance within which 87% of the mass of the diffusing
substance occurs. Accordingly, the ‘‘penetration time’’ then represents the time after
which 87% of the diffusing molecules have not yet moved beyond a given distance.
The special case of zero surface concentration is obvious. The rate of loss
diffusing substance from the semi-infinite medium when the surface concentration
is zero, is

@C De C0
De ¼ pffiffiffiffiffiffiffi ð4:20Þ
@x x¼0 De t
and the total amount Mt of diffusing substance which has left the medium at time
t is given by integrating Eq. 4.19 and is
rffiffiffiffiffiffiffi
De t
Mt ¼ 2C0 ð4:21Þ
p
4.2.5 Diffusion in a Plane Sheet
In this case we consider one-dimensional diffusion through a plane sheet of

thickness 2d, initially at a uniform concentration C0 and with the surface at con-
stant concentration C1.
4.2 Diffusion 45
The boundary conditions are:

t¼0 d\x\d C ¼ C1 ð4:22aÞ
t[0 x¼0 C ¼ C0 ð4:22bÞ
t[0 x!1 C ¼ C1 ð4:22cÞ

The mass of solute per unit area which has diffused after a certain time is:
!
C C1 4X 1
ð1Þn ð2n þ 1Þpx ð2n þ 1Þ2 p2 De t
¼1 cos exp ð4:23Þ
C 0 C1 p n¼0 2n þ 1 2d 4d 2
and the total amount Mt of diffusing substance which has entered the plane sheet at
time t is given by
!
Mt X1
8 ð2n þ 1Þ2 p2 De t
¼1 2 2
exp ð4:24Þ
M1 n¼0 ð2n þ 1Þ p
4d 2
where M? is the corresponding quantity after infinite time. The corresponding

solutions for small times are:
C C1 X
1
ð2n þ 1Þd x X 1
ð2n þ 1Þd þ x
¼ ð1Þn erfc pffiffiffiffiffiffiffi þ ð1Þn erfc pffiffiffiffiffiffiffi ð4:25Þ
C0 C1 n¼0 2 De t n¼0
2 De t
and
" #
Mt 2 pffiffiffiffiffiffiffi 1=2 X1
n nd
¼ De t p þ2 ð1Þ ierfc pffiffiffiffiffiffiffi ð4:26Þ
M1 d n¼1
De t
4.2.6 Diffusion in a Cylinder
If we consider a circular cylinder in which the diffusion is radial, concentration is

then a function of radius and time, and the diffusion equation is obtained:

@C 1 @ @C
¼ rDe ð4:27Þ
@t r @r @r
Considering the following boundary conditions:
t¼0 0\r\a C ¼ C1 ð4:28aÞ
t[0 r¼a C ¼ C0 ð4:28bÞ

t[0 r!1 C ¼ C1 ð4:28cÞ
the solution obtained is:
C C1 2X 1
1
¼1 expðDe a2n tÞJ0 ðran Þ ð4:29Þ
C 0 C1 a n¼1 an J1 ðaan Þ
where an is the positive roots of J0 ðaan Þ ¼ 0; J0 ðxÞ is the Bessel function of the
first kind of order zero, and J1 ðxÞ is the Bessel function of the first order.
The total amount Mt of diffusing substance which has entered or left the cyl-
inder at time t is given by
Mt X1
4
¼1 2 a2
expðDe a2n tÞ ð4:30Þ
M1 n¼1
a n

rffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffi
C C1 a ar ða rÞ De at ar
¼ erfc pffiffiffiffiffiffiffi þ 3=2
ierfc pffiffiffiffiffiffiffi
C0 C 1 r 2 De t 4ar 2 De t
ð4:31Þ
ð9a2 7r 2 2arÞDe t 2 ar
þ i erfc pffiffiffiffiffiffiffi þ . . .
32a3=2 r5=2 2 De t
and
rffiffiffiffiffiffiffiffi
Mt 4 De De 1 De 3=2
¼ pffiffiffi t 2 t pffiffiffi 2 t þ. . . ð4:32Þ
M1 p a2 a 3 p a
4.2.7 Diffusion in a Sphere
If we consider a sphere in which the diffusion is radial, concentration is then a

function of radius and time, and the diffusion equation is obtained:
@C @ 2 C 2 @C
¼ D e 2 þ De ð4:33Þ
@t @r r @r
Considering the following boundary conditions:
t¼0 0\r\a C ¼ C1 ð4:34aÞ
t[0 r¼a C ¼ C0 ð4:34bÞ
t[0 r!1 C ¼ C1 ð4:34cÞ
the solution obtained is:

4.2 Diffusion 47

C C1 2a X 1
ð1Þn np r D e n 2 p2
¼1þ sin exp t ð4:35Þ
C0 C1 p r n¼1 n a a2
And the concentration at the centre of the sphere, r ! 0; is given by

X1
C C1 D e n 2 p2
¼1þ2 ð1Þn exp t ð4:36Þ
C0 C 1 n¼1
a2
The total amount Mt of diffusing substance which has entered or left the sphere
at time t is given by

Mt 6X 1
1 D e n 2 p2
¼1 2 exp t ð4:37Þ
M1 p n¼1 n2 a2
1
C C1 aX ð2n þ 1Þa r ð2n þ 1Þa þ r
¼ erfc pffiffiffiffiffiffiffi erfc pffiffiffiffiffiffiffi ð4:38Þ
C0 C1 r n¼0 2 De t 2 De t
and
rffiffiffiffiffiffiffiffi" #
Mt De X1
na De
1=2
¼6 2
t p þ2 ierfc pffiffiffiffiffiffiffi 3 2 t ð4:39Þ
M1 a n¼0
De t a
4.3 Axial Dispersion
Over the past 5 decades, axial dispersion in porous media has been measured and
correlated extensively for gaseous and liquid systems. Many publications are
available for a variety of applications, including: packed bed reactors [29, 47, 65,
89, 135] and soil column systems [9, 99, 102, 103].
One of the first results published about axial dispersion in packed beds of
inertial particles was in the 1950s by Danckwerts [37], who published his cele-
brated paper on residence time distribution in continuous contacting vessels,
including chemical reactors, and thus provided methods for measuring axial dis-
persion rates. The author studied dispersion along the direction of flow for a step
input in solute concentration (CS) in a bed of Raschig rings (with length L), crossed
by water (C0) with a value of Reð¼ qUd=lÞ approximately equal to 25 and
obtained a PeL ð¼ ud=DL Þ value of 0.52.
Kramers and Alberda [85] followed Danckwerts’s study with a theoretical and
experimental investigation by the response to a sinusoidal input signal. These
authors proposed that packed beds could be represented as consecutive regions of
well-mixing rather than a sequence of stirred tanks (mixing-cell model) and
suggested a PeL ffi 1; for Re ! 1: McHenry and Wilhelm [94] assumed the axial
distance between the mixing-cells in a packing to be equal to particle diameter and

showed that PeL must be about two for high Reynolds number. The difference in
the two results may be explained on the basis of experimental results of Kramers
and Alberda [85] while are obtained with L=D 4:6; a value significantly less
than L=D [ 20 [66]. Klinkenberg et al. [81] and Bruinzeel et al. [22] show that
radial dispersion can be neglected for a small ratio of column diameter to length
and large fluid velocity.
Brenner [19] presented the solution of a mathematical model of dispersion for a
bed with finite length, L, and the most relevant conclusion of his work was that for
Pea ð¼ uL=DL Þ 10; the equations obtained by Danckwerts [37] for an input step
in solute concentration and Levenspiel and Smith [89] for a pulse in solute
concentration, that assumed an infinite bed, are corrected.
Hiby [75] proposed a better empirical correlation to cover the range of Rey-
nolds numbers to 100. The author reported experimental results with the aid of
photographs to compare the two dispersion mechanisms presented above: diffu-
sional model in turbulent flow and the mixing-cell model.
Sinclair and Potter [124] used a frequency response technique applied to the
flow of air through beds of glass ballotini in a Reynolds number range between 0.1
and 20. A further investigation in the intermediate Reynolds number region has
been carried out by Evans and Kenney [50] who used a pulse response technique in
beds of glass spheres and Raschig rings.
Experiments reported by Gunn and Pryce [68] showed that axial dispersion
coefficients given by the theoretical equation for the diffusional model and the
theoretical equation for the mixing-cell model are very similar. The authors also
showed that neither the mixing-cell model nor the axially dispersed plug flow
model could describe axial dispersion phenomena.
The description of solute transport in packed beds by dispersion models has
been studied since the 1950s and has long attracted the attention of engineers and
scientists.
Typically, the boundary conditions adopted, by the vast majority of the
investigators reported above, have corresponded to the semi-infinite bed, i.e., L is
sufficiently large ðL=D [ 20Þ: Dispersion of the given tracer was measured at two
points in the outlet and the distortion of a tracer forced by a pulse input (ex: [13,
25, 128]), frequency response (ex: [39, 45, 85, 94, 130]) and step input (ex: [37,
75, 95, 100]). Figure 4.1 illustrates some experimental data points for axial dis-
persion in liquid and gaseous systems. The experimental dispersion data are most
frequently presented in logarithmic plots of PeL (or DL =Dm Þ vs. Pem ð¼ ud=Dm Þ;
spanning six or more orders of magnitude.
4.3.1 Parameters Influencing Axial Dispersion: Porous Medium
Perkins and Johnston [102] in their article review showed some of the variables
that influence axial and radial dispersion. However, before attempting in the
4.3 Axial Dispersion 49
10
Pe L
Edwards and Richardson (1968), Sc=0.72

-1 Gunn and Pryce (1969), Sc=0.88
10 Hiby (1962), Sc=545
Chung and Wen (1968), Sc=675
Harleman and Rumer (1963), Sc=710
Miller and King (1966), Sc=730
Jacques and Vermeulen (1957), Sc=820
Cairns and Prausnitz (1960), Sc=770
Ebach and White (1958), Sc=1858
-2
Carberry and Bretton (1958), Sc=1858
10 -2 -1 2 3 4 5 6 7
10 10 1 10 10 10 10 10 10 10
Pem
Fig. 4.1 Some experimental data points for axial dispersion in liquid systems and gaseous
systems
parameters influencing dispersion, it is important to consider the effect of the

packing of the bed on dispersion coefficients. Gunn and Pryce [68] and Roemer
et al. [111] showed that when particles in packed beds are not well packed the
dispersion coefficient is increased. Experimental results of Gunn and Pryce [68]
showed that different re-packing of the bed gave deviations of 15% in radial Peclet
values. These experiments confirm that fluid mechanical characteristics are not
only defined by the values of the porosity and tortuosity (easy to reproduce), but
depend of the quality of packing in the bed.
The effect of radial variations of porosity and velocity on axial and radial
transport of mass in packed beds was analytically quantified by Choudhary
et al. [30], Lerou and Froment [88], Vortmeyer and Winter [138] and Delmas
and Froment [41].
A rigorous measurement of the porosity in a packed bed is fundamental to
minimize the errors in the experimental measurements, because the porosity
between the inert particles of the bed helps the diffusion of a tracer and gradually
increases dispersion.
A more coherent interpretation of the experimental data may be obtained
through the use of dimensional analysis. As a starting point it is reasonable to
accept the functional dependence
DL ¼ /ðL; D; u; d; q; l; Dm Þ ð4:40Þ
for randomly packed beds of mono sized particles with diameter d, where q and l
are the density and viscosity of the liquid, respectively, and Dm is the coefficient of
molecular diffusion of the solute. Making use of Buckingham’s p theorem,

Eq. 4.40 may be rearranged to give

DL L D ud l
or PeL ¼ U ; ; ; ð4:41Þ
Dm D d Dm qDm
and it is useful to define Pem ¼ ud=Dm and Sc ¼ l=qDm : This result suggests that
experimental data be plotted as ðDL =Dm Þ vs. Pem :
4.3.1.1 Effect of Column Length
One first aspect to be considered, as a check on the experimental method (infinite

medium), is the influence of the length of the bed (L) on the measured value of
axial dispersion. In reality, if an experimental method is valid, values of the
dispersion coefficient measured with different column lengths, under otherwise
similar conditions, should be equal, within the reproducibility limits.
The dependence of the axial dispersion coefficient on the position in packed
beds was first examined by Taylor [133]. The author showed that, in laminar flow,
dispersion approximation would be valid if the following equation is satisfied,
Dm t
h¼ 0:14 ð4:42Þ
R2
where R is the tube radius. Carbonell and Whitaker [26] concluded that the axial
dispersion coefficient becomes constant if the following expression is satisfied

1 e 2 Dm t
h¼ 1 ð4:43Þ
e d2
Han et al. [69], see Fig. 4.2, showed that values of the axial dispersion coef-
ficient, for uniform size packed beds, measured at different positions in the bed are
function of bed location unless the approximate criterion

L 1 1e 2 Dm t
0:3 or h ¼ 2 0:15 ð4:44Þ
d Pem e d
is satisfied. The authors showed that for Pem \700; axial dispersion coefficients
were nearly identical for all values of x ¼ L; and for Pem [ 700 observed an
increase in the value of dispersion coefficients with increasing distance down the
column.
4.3.1.2 Ratio of Column Diameter to Particle Diameter
It is well known (e.g. [137]) that the voidage of a packed bed (and therefore the
fluid velocity) is higher near a containing flat wall. The effects of radial variations
5
10
Approximative criterium of
Han et al . (1987):
2
L 1 ⎛1− ε ⎞
⎜ ⎟ ≥ 0 .3
d Pe m ⎝ ε ⎠
4
10
D L/D m
3
10 L=3.0 m (L/d=6490)
L=2.025 m (L/d=4380)
L=0.75 m (L/d=1620)
L=1.476 m (L/d=420), Han et al. (1987)
L=1.121 m (L/d=320), Han et al. (1987)
L=0.74 m (L/d=210), Han et al. (1987)
L=0.257 m (L/d=73.4), Han et al. (1987)
2
10 2 3 4 5
10 10 10 10
Pe m
Fig. 4.2 Effect of bed length on axial dispersion
of porosity and velocity on axial and radial transport of mass in packed beds were
analytically quantified by several investigators like Choudhary et al. [30], Lerou
and Froment [88], Vortmeyer and Winter [138] and Delmas and Froment [41].
Schwartz and Smith [118] were the first to present experimental data showing
zones of high porosity extending two or three particle diameters from the
containing flat wall. The results indicated that unless D=d [ 30 important velocity
variations exist across the packed bed. Other studies showed that packed bed
velocity profiles significantly differ from flows with large diameter particles in
small diameter tubes ([23, 34]).
Hiby [75] showed that the effect of D=d is not significant in the measured of
axial dispersion coefficient when the ratio is greater than 12.
Stephenson and Stewart [129] showed that the area of high fluid velocities
limits to the area of high porosities, and this area does not extend more than a
particle diameter of the wall and the assumption of a flat velocity profile is rea-
sonable. This work confirms the earlier experiments reported by Roblee et al.
[110], Schuster and Vortmeyer [117] and Vortmeyer and Schuster [137].
A similar effect was observed in measuring pressure drops across packings, so an
empirical rule can be considered that the variations, in radial position, of the fluid
velocity, porosity and dispersion coefficient can be negligible, if D=d [ 15 [3, 66].
4.3.1.3 Ratio of Column Length to Particle Diameter
Strang and Geankoplis [130] and Liles and Geankoplis [90] make much of the
effect of L=d but the evidence from fluid mechanical studies [67] was that the
effect is confined to a dozen layers of particles and is not very important.
Experimental results of Guedes de Carvalho and Delgado [61], presented in

Fig. 4.3, with two different spherical particles diameter and the same length of the
packed bed showed that axial dispersion coefficient does not increased with par-
ticle diameter, as long as the condition D=d [ 15 is satisfied (see Vortmeyer and
Schuster [137] and Ahn et al. [2], for wall effects).
4.3.1.4 Particle Size Distribution
Another aspect of dispersion in packed beds that needs to receive attention is the
effect of porous medium structure. In a packed bed of different particle sizes, the
small particles accumulate in the interstices between large particles, and porosity
tends to decrease.
Raimondi et al. [108] and Niemann [98] studied the effect of particle size
distribution on axial dispersion and concluded that DL increases with a wide
particle size distribution. Eidsath et al. [48] indicated a strong effect of particle size
distribution on dispersion. As the ratio of particle diameters went from a value of 2
to 5, the axial dispersion increased by a factor of 1.5, and radial dispersion
decreased by about the same factor.
Han et al. [69] showed that for a size distribution with a ratio of maximum to
minimum particle diameter equal to 7.3, axial dispersion coefficient are 2–3 times
larger than the uniform size particles (see Fig. 4.3).
Wronski and Molga [146] studied the effect of particle size non-uniformities on
axial dispersion coefficients during laminar liquid flow through packed beds (with
a ratio of maximum to minimum particle diameter equal to 2.13) and proposed a
generalized function to determine the increase of the axial dispersion coefficients
in non-uniform beds relative to those obtained in uniform beds.
Guedes de Carvalho and Delgado [61] obtained the same conclusion in their
experiments, with ballotini and a ratio of maximum to minimum particle diameter
equal to 3.5 in comparison with glass ballotini that have the same size.
4.3.1.5 Particle Shape
The effect of particle shape on axial dispersion has been studied by several
investigators, such as Bernard and Wilhelm [14], Ebach and White [45], Car-
berry and Bretton [25], Strang and Geankopolis [130], Hiby [75], Klotz [82] and
more recently Guedes de Carvalho and Delgado [61]. The authors have used
beds of spheres, cubes, Raschig rings, sand, saddles and other granular material,
and have concluded that generally axial dispersion coefficient tend to be greater
with packs of nonspherical particles than with packs of spherical particles, with
the same size.
Figure 4.4 shows that particle shape is a significant parameter, with higher
values of DL (i.e., lower PeL Þ being observed in packed beds of sand and
Raschig rings comparatively with the results obtained with spherical beds.
5
10
4
10
D L/D m
3
10
2
10 d=3.0 mm (uniform)
d=0.462 mm (uniform)
d= 3.5 mm (uniform), Han et al. (1987)
d=3.5 mm (dmax/dmin=2.2), Han et al. (1987)
10
2 3 4 5
10 10 10 10 10
Pe m
Fig. 4.3 Effect of particle size distribution on axial dispersion
Therefore, increased particle sphericity correlates with decreased dispersion, with

a sphericity defined as the surface area of a particle divided by the surface area
of a sphere of volume equal to the particle.
4.3.2 Parameters Influencing Axial Dispersion: Fluid Properties
4.3.2.1 Viscosity and Density of the Fluid
Some investigators, like Hennico et al. [74], used glycerol and obtained significant
effect of viscosity, at large Reynolds number, on axial dispersion coefficient. In
vertical miscible displacements, if a less viscous fluid displaced another fluid
viscous fingers will be formed [102]. However, if a more viscous fluid displaced a
different fluid the dispersion mechanisms are unaffected, but the situation will tend
to reduce convective dispersion. This leads to increased dispersion relative to the
more viscous fluid displacing a less viscous one.
The importance of density gradients was recently investigated by Benneker
et al. [12] and their experiments showed that axial dispersion coefficient is con-
siderably affected by fluids with different densities due the action of gravity forces.
Fluid density creates similar effects to fluid viscosity. In a displacement with a
denser fluid above the less-dense fluid, gravity forces cause redistribution of the
fluids. However, if a denser fluid is on the bottom, usually, a stable displacement
occurs.
5
10
4
10
D L/D m
3
10
Glass spheres
Sand
Glass spheres, Ebach and White (1958)
Raschig rings, Ebach and White (1958)
Glass spheres, Strang and Geankoplis (1958)
Raschig rings, Strang and Geankoplis (1958)
2
10
2 3 4 5
10 10 10 10
Pe m
Fig. 4.4 Effect of particle shape on axial dispersion
4.3.2.2 Fluid Velocity
The first two groups of Eq. 4.8 have importance only when D=d is less than 15 and
L=D is so small that the characteristics of dispersion are affected by changing velocity
distributions. So, for packed beds we will usually have DL =Dm ¼ UðPem ; ScÞ:
In order to understand the influence of fluid velocity on the dispersion coefficient,
it is important to consider the limiting case where u ! 0: If DL was defined based on
the area open to diffusion (see Eq. 2), in the limit u ! 0; solute dispersion is
determined by molecular diffusion, with DL ¼ D0m ¼ Dm =s (s being the tortuosity
pffiffiffi
factor for diffusion and it is equal to 2 as suggested by Sherwood et al. [121]).
As the velocity of the fluid is increased, the contribution of convective
dispersion becomes dominant over that of molecular diffusion (see [144]) and
DL ¼ ud=PeL ð1Þ; where u is the interstitial fluid velocity and PeL ð1Þ ffi 2 for
gas or liquid flow through beds of (approximately) isometric particles, with
diameter d [23, 79].
Assuming that the diffusive and convective components of dispersion are
0
additive, the same authors suggest that DL ¼ Dm þ ud=PeL ð1Þ; which may be
written in dimensionless form [66] as
DL 1 1 ud 1 1 e 1
¼ þ or ¼ þ ð4:45Þ
Dm s 2 Dm PeL s ReSc 2
This equation is expected to give the correct asymptotic behaviour in gas and
liquid flow through packed beds, at high and low values of Pem ð¼ ud=Dm Þ: For
gases this is confirmed in Fig. 4.5, but for liquids (Fig. 4.6) the data do not cover
the extreme conditions.
10
Hiby (1962) Langer et al . (1978)
Evans&Kenney (1966)
Edwards&Richardson (1968)
1 Scott et al . (1974)
Johnson&Kapner (1990)
Pe L
1 1 1 1
= +
Pe L Pe m 2
Jonhson and Kapner (Sc=0.22)
-1
10 Carberry and Bretton (Sc=0.22)
Jonhson and Kapner (Sc=1.12)
Edwards and Richardson (Sc=1.12)
Gunn and Pryce (Sc=1.12)
Sinclair and Potter (Sc=1.2)
Blackwell et al. (Sc=1.90)
-2 data compiled by Perkins and Johnson (1963)
10
-2 -1 2 3 4
10 10 1 10 10 10 10
Pe m
Fig. 4.5 Axial dispersion in gases
But these figures show that Eq. 4.45 is inaccurate over part of the intermediate
range of Pem : In the case of gas flow, shown Fig. 4.5, significant deviations are
observed only in the range 0:6\Pem \60; as pointed out by several of the authors
[47, 66, 75, 136]. The experimental values of PeL ð¼ ud=DL Þ are generally higher
than predicted by Eq. 4.12. Several equations have been proposed to represent the
data in this intermediate range and the equations of Hiby [75], Edwards and
Richardson [47], Evans and Kenney [50], Scott et al. [119], Langer et al. [80] and
Johnson and Kapner [80] are shown to fit the data points reasonably well (see
Fig. 4.5).
With liquids, deviations from Eq. 4.45 occur over the much wider range
2\Pem \106 ; the experimental values of PeL being significantly lower than
predicted by that equation. The difference in behaviour between gases and liquids
has to be ascribed to the dependence of PeL on Sc ð¼ l=qDm Þ:
4.3.2.3 Fluid Temperature (or Schmidt Number)
The coefficient of axial dispersion for gas flow ðSc ffi 1Þ is predicted with good
accuracy by Eq. 4.45, except in the approximate range 0:5\Pem \100; where
experimental values may be more than twice those given by the equation, as
confirmed by Fig. 4.5.
For liquid flow, a large number of data are available, that were obtained with
different solutes in water at near ambient temperature, corresponding to values of
Sc in the range 500\Sc\2000: Most of the data reported in the literature, for this
range of Sc, are shown in Fig. 4.6, and they form a ‘‘thick cloud’’ running parallel
10
1 1 1 1
= +
Pe L Pe m 2
PeL
10 -1
Rifai et al. (Sc=560) Chung and Wen (Sc=675)

Pfannkuch (Sc=680) Bruinzeel et al (Sc=700)
Miller and King (Sc=730) Cairns and Prausnitz (Sc=770)
Jacques and Vermeulen (Sc=820) Liles and Geankopolis (Sc=883)
Strang and Geankopolis (Sc=1080) Ebach and White (Sc=1858)
Carberry and Bretton (Sc=1858) data compiled by Perkins and Jonhston
10 -2
-2 -1 2 3 4 5 6 7
10 10 1 10 10 10 10 10 10 10
Pe m
Fig. 4.6 Axial dispersion in liquids
to the line defined by Eq. 4.45, though somewhat below it (at approximately,
0:3\PeL \2Þ:
In recent years, data on axial dispersion have been made available for values of
Sc between the two extremes of near ideal gas ðSc ffi 1Þ and cold water
ðSc [ 550Þ: Such data were obtained either supercritical carbon dioxide
ð1:5\Sc\20Þ or heated water ð55\Sc\550Þ and are presented in Fig. 4.7.
Figure 4.7 show a consistent increase in PeL with a decrease in Sc and it may be
seen that the dependence is slight for the higher values of Sc (say for Sc of order
750 and above). At the lower end of the range of Pem investigated there seems to
be a tendency for PeL to become independent of Sc, even if the values of DL are
still significantly above Dm . In the intermediate range, 100\Pem \5000; values of
PeL are very nearly constant, for each value of Sc. The convergence of the different
series of points at about Pem ffi 20 seems to suggest that PeL is insensitive to Sc
below this value of Pem ; for the range of Sc presented.
A good additional test of the consistency of the data of Guedes de Carvalho and
Delgado [61] is supplied by the plot in Fig. 4.8, where it may be seen that all the
series of points converge at high Re, as would be expected for turbulent flow. The
agreement with the data of Jacques and Vermeulen [79] and Miller and King [95],
for cold water, is worth stressing.
Recently, some workers have measured axial dispersion for the flow of
supercritical carbon dioxide through fixed beds and this provides important new
data in the range 1:5\Sc\20: However, the various authors fail to recognize the
direct dependence of PeL on Sc. Catchpole et al. [27] represent their data and those
of Tan and Liou [132] in a single plot (their Fig. 4.3) of PeL vs. Re: The majority
of points are in the range of 1\Re\30 and the data of both groups, together,
1.4
Sc=1930 Sc=424 Sc=102
Sc=1358 Sc=283 Sc=71
1.2 Sc=993 Sc=200 Sc=57
Sc=754 Sc=143
1.0
Pe L
0.8
0.6
0.4
0.2
0.0
2 3 4 5
10 10 10 10 10
Pe m
Fig. 4.7 Dependence of PeL on Pem for different values of Sc
2.0
Sc=1930 Sc=424 Sc=102
Sc=1358 Sc=283 Sc=71 G.Carvalho and
Sc=993 Sc=200 Sc=57 Delgado (2003)
Sc=754 Sc=143
1.6 Jacques and Vermeulen (1958), Sc=820
Miller and King (1966), Sc=730
1.2
PeL
0.8
0.4
Re = 25
0.0
10 -2 10 -1 1 10 10 2 10 3
Re
Fig. 4.8 Dependence of PeL on Re for different values of Sc
define a horizontal cloud with mid line at about PeL ffi 0:8; spreading over the
approximate range 0:3\PeL \1:1:
The data of Yu et al. [147] are for 0:01\Re\2 and 2\Sc\9: It is worth
referring here that the modelling work of Coelho et al. [32] gives theoretical
support to experimental findings for low Re, both for spherical and non-spherical
particles. No influence of Sc on PeL is detected, but unfortunately the results are
not very consistent, particularly in the range 1\Pem \20; where the scatter is high
and the values of PeL are much too low.
10
Hiby (1962)
Molecular diffusion
PeL
1 1 1 1 Edwards and Richardson (Sc=1.12)

= + Gunn and Pryce (Sc=1.12)
10
-1 Pe L Pe m 2 Blackwell et al. (Sc=1.90)
Sc=57
Sc=102
Our data Sc=424
Sc=754
Rifai et al. (Sc=560)
Miyauchi and Kikuchi (Sc=665)
Miller and King (Sc=730)
-2 Supercritical Fluid Flow Yu et al. (Sc=1.9 - 8.5)
10
-2 -1 2 3
10 10 1 10 10 10
Pem
Fig. 4.9 Dependence of PeL on Pem for Stokes flow regime
Figure 4.9 shows that for low values of Pem (Stokes flow regime) there seems to
be a tendency for PeL to become independent of Sc. The values of PeL reported by
Miller and King [95], for 6\Pem \100; are much too low; this may be because the
particles used in most experiments are too small (particle sizes of 55 and 99 lm)
and this is known to yield enhanced dispersion coefficients, possibly due to particle
agglomeration (see [64, 75]). The data reported by Miyauchi and Kikuchi [97] and
plotted in Fig. 4.9, for 6\Pem \300; are higher than our experimental data.
There are considerable experimental difficulties in the measurement of axial
dispersion in the liquid phase at small Reynolds number, because the usual method
of obtaining low Reynolds number is to reduce particle size and this is known to
yield enhanced dispersion coefficients.
4.4 Radial Dispersion
Generally, radial dispersion coefficients are measured in non-reactive conditions,

because the rate of mass transfer, observed experimentally, is directly related to
the coefficient of radial dispersion in the bed.
The most popular technique for the measurement of radial dispersion consists in
feeding a continuous stream of tracer from a ‘‘point’’ source somewhere in the bed
(usually along the axis, if there is one) and measuring the radial variation of tracer
concentration at one or more downstream locations.
The first study of mass transfer by radial dispersion in gaseous systems was
carried out by Towle and Sherwood [134]. The results presented were very
important for packed bed dispersion because they showed that dispersion was not
influenced by the tracer molecular weight.
4.4 Radial Dispersion 59
Bernard and Wilhelm [14] reported the first measurements, in liquid systems, of
experimental values of radial dispersion coefficients in packed beds of inerts by a
Fickian model. The authors took into account the wall effect condition and their
experiments suggested that for high values of Reynolds number the value of PeT is
constant and between 11 and 13.
Baron [7] proposed a new model of radial dispersion in which a particle
of tracer executes a simple random-walk displacement of ± particle diameter to
give a transversal Peclet number between 5 and 13, when Re ! 1: The basis for
this prediction is the random-walk theory, in which a statistical approach is
employed. This method does not take into account effects of radial variations in
velocity and void space. Latinen [87] extended the random-walk concept to three
dimensions and predicted a value of 11.3, for PeT ð1Þ:
Klinkenberg et al. [81] solved Eq. 4.1 for anisotropic dispersion, but considered
that dispersion occurs in an infinite medium. In the same work were considered the
particulate cases of isotropic dispersion and axial dispersion neglected.
Plautz and Johnstone [105] used the equation derived by Wilson [145], for heat
transfer, and suggested a PeT between 11 and 13, for Re ! 1: Fahien and Smith
[51] assumed that for Reynolds numbers in the range between 40 and 100, the
Peclet number is independent of fluid velocity and equal to 8. The authors were the
first to consider that the tracer pipe can be of significant diameter compared to the
diameter of the bed.
Dorweiler and Fahien [42] used the equation derived by Fahien and Smith [51]
to study mass transfer in laminar and transient flows. The results showed that for
Re \200; the Peclet number based on the radial dispersion coefficient is a linear
function of the fluid velocity and for Re [ 200; at room temperature, the Peclet
number is constant as also shown by Bernard and Wilhelm [14], Plautz and
Johnstone [105] and Fahien and Smith [51]. The authors have demonstrated a
difference in the Peclet number with radial position. The transversal Peclet number
is constant from the axis to 0.8 times the radius and then rises near the wall.
Hiby and Schummer [76], and later Roemer et al. [111], presented the solution
of the mass balance equation (Eq. 4.1), considering the tracer pipe to be of sig-
nificant diameter compared to the diameter of the packed bed.
Saffman [114] considered the packed bed as a network of capillary tubes ran-
domly orientated with respect to the main flow. At high Peclet number and at very
long time, Saffman found that the dispersion never becomes truly mechanical, with
zero velocity of the fluid at the capillary walls, the time required for a tracer
particle to leave a capillary would become infinite as its distance from the walls
goes to zero. The author proposed that DT ¼ ð3=16Þud when Re ! 1; but this
prevision of radial dispersion coefficient is higher than observed experimentally.
Hiby [75] and Blackwell [16] presented an experimental technique in which they
divided the sampling region into two annular regions and calculated the transversal
dispersion coefficient from the averaged concentrations of each of the two samples.
The experimental data points of Wilhelm [144] suggested that PeT ð1Þ ¼ 12;
for beds of closely sized particles, and this value is accepted for the majority of the
investigators (ex: [15, 33, 66, 75, 144]).
Roemer et al. [111] studied radial mass transfer in packed beds at low flow
rates, Re \100: The authors considered the tracer pipe to be of significant
diameter compared to the diameter of the bed (‘‘finite source’’ model) and axial
and radial dispersion are equal. In this work the authors compared the values of
PeT obtained with two methods (‘‘instantaneous finite source’’ and ‘‘point source’’)
and concluded that the values of PeT obtained with the ‘‘point source’’ method
were 10% less that the values obtained with the ‘‘instantaneous finite source’’
method. The authors estimated that the neglecting the axial dispersion in calcu-
lations of DT ; for low values of Reynolds numbers, can cause errors of 10%.
Coelho and Guedes de Carvalho [33] developed a new experimental technique,
based on the measurement of the rate of dissolution of planar or cylindrical sur-
faces, buried in the bed of inert particles and aligned with the flow direction. This
alternative technique is simple to use, allows the determination of the coefficient of
radial dispersion in packed beds over a wide range of flow rates, and it is easily
adaptable to work over a range of temperatures above ambient, as shown by
Guedes de Carvalho and Delgado [63] and Delgado and Guedes de Carvalho [40].
In recent years, nuclear magnetic resonance has been used to determine both
diffusion and dispersion coefficients (e.g. [8, 57]), with significant advantages, but
this technique were limited to low fluid velocities.
It is important to remember that, at high Reynolds numbers, the main mecha-
nism of radial dispersion is the fluid deflection caused by deviations in the flow
path caused by the particles in the bed (axial dispersion is caused by differences in
fluid velocity in the flow), i.e., dispersion is caused by hydrodynamic mechanisms
(macroscopic scale) and not by molecular diffusion (Brownian motion).
The result is a poor mixture at the ‘‘microscopic scale’’. In fact, there are
detected different values of solute concentration over a distance of the order of a
particle diameter or less, what explains the convenience of use of an efficient
averaging procedure [66]. This is probably one of the reasons that explain the
difference observed in some experimental results of dispersion (see Fig. 4.10).
Gunn and Pryce [68] showed that the standard deviation without repacking in the
measurement of PeT was 5%, while when the bed was repacked each time of
measurement, the standard deviation found was 15%.
4.4.1 Parameters Influencing Radial Dispersion: Porous Medium
4.4.1.1 Length of the Packed Column
Han et al. [69] showed that values of the radial dispersion coefficient, for uniform
size packed beds, measured at different positions in the bed are not a function of
bed location, i.e., they observed no time dependent behaviour for radial dispersion,
because radial dispersion is caused by mechanical mechanism alone.
An important aspect to be considered, as a check on the experimental method of
Coelho and Guedes de Carvalho [33], is the influence of the length of the test
2
10
10
PeT
1
Plautz and Johnstone (1955), Sc=0.94
Coelho and G.Carvalho (1988), Sc=1.86
Blackwell (1962), Sc=500
Hiby (1962), Sc=540
-1 Harleman and Rumer (1963), Sc=710
10
Hartman et al. (1958), Sc=750
Bernard and Wilhelm (1950), Sc=754
Coelho and G.Carvalho (1988), Sc=986
Simpson (1962), Sc=1270
Grane and Gardner (1961), Sc=1476
-2
10
-2 -1 2 3 4 5 6 7
10 10 1 10 10 10 10 10 10 10
Pem
Fig. 4.10 Some experimental data points for radial dispersion in liquid systems and gaseous
systems
cylinder on the measured value of DT : In reality, the two variables are indepen-
dent, provided that the criterion given by Eq. 4.46 is satisfied (see Fig. 4.11 )

L ud
0:62 ð4:46Þ
d Dm
4.4.1.2 Ratio of Column Diameter to Particle Diameter
Several investigators, like Fahien and Smith [51], Latinen [87] and Singer and
Wilhelm [125], have studied the wall effect on radial dispersion coefficient. The
experiments suggested that in a packing structure characterized by significant
variations of void fraction in radial direction, up to distance of about two particle
diameters from the wall, a non-uniform radial velocity profile is induced, with a
maximum just near the wall. As result, wall effects occur due large voidage
fluctuations near the wall. The above investigators also showed that the increase in
radial dispersion in the laminar region would be the same order of magnitude as in
the turbulent region.
4.4.1.3 Particle Size Distribution
Eidsath et al. [48] studied the effect of particle size distribution on dispersion. As
the ratio of particle diameter went from a value of 2 to 5, the radial dispersion
decreased by a factor of 3, but perhaps the results were a cause of the simple
0.04
d =0.496 mm
0.03
D T (mm 2/s)
0.02
0.01
L = 155 mm
L = 215 mm
L = 250 mm
0.00
0.0 0.5 1.0 1.5 2.0
u (mm/s)
Fig. 4.11 Effect of length of soluble cylinder on the measurement of radial dispersion
geometry employed in these computations (packed bed of cylinders). Steady-state

measurements of radial dispersion reported by Han et al. [69], with the same void
fraction and mean particle diameter, but different particle size range (ratio of
maximum to minimum particle diameter equal to 2.2 and 7.3), showed that there
was no evidence to indicate a change in radial dispersion with particle size dis-
tribution (see Fig. 4.12a).
The effect of a distribution of particle sizes within the bed, on the radial dis-
persion coefficient, may be assessed from Guedes de Carvalho and Delgado [63].
In particular, lot D was prepared by carefully blending lots B and E in a proportion
of 1:1 (by weight). In Fig. 4.12b, dispersion data obtained with the mixed lot are
seen to fall in between the data for the original separate lots, as might be expected.
Figure 4.12a shows that in a plot of DT =Dm vs. Pem ; the data for the three lots fall
along the same line, when d (in Pem Þ is taken to represent the average particle size
in the bed.
4.4.1.4 Particle Shape
The effect of particle shape on the radial dispersion coefficient has been given
attention by several investigators both for gaseous and liquid systems. England and
Gunn [49] measured the dispersion of argon in beds of solid cylinders and beds of
hollow cylinders and have concluded that DT tend to be greater with packs of
hollow cylinders than with packs of solid cylinders, and these results were greater
than obtained with packs of spherical particles (see Fig. 4.13).
The same conclusion, in liquid systems, was been obtained by Hiby [75], who
used packed beds of glass spheres and Rachig rings, and Bernard and Wilhem [14],
Fig. 4.12 Effect of particle (a) 10 3

size distribution on radial
dispersion. a DT =Dm vs. Pem ;
Eq. (4.32)
b DT =Dm vs. u:
10 2
D T/D m
10
d=0.496 mm (uniform, lot E - narrow caut)
d=0.297 mm (uniform, lot B - narrow cut)
d=0.409 mm (lot D - mixed)
d= 3.5 mm (uniform), Han et al. (1987)
1
2 3 4 5
10 10 10 10 10
Pe m
(b) 25
20
D T/D m
15
10
5 d=0.409mm (lot D - mixed)

d=0.297mm (lot B - narrow cut)
d=0.496mm (lot E - narrow cut)
0
0.0 0.5 1.0 1.5 2.0
u (mm/s)
who used packed beds of cubes, cylinders and glass spheres. Figure 4.13 shows
that the radial dispersion coefficient tends to be greater in packed beds of non-
spherical particles.
However, Blackwell [16], List [91], Guedes de Carvalho and Delgado [63] and
others reported experiments with packed beds of sand and showed that DT
obtained with glass ballotini are very close to those for sand (not pebble or gravel)
and the conclusion seems to be that particle shape has only a small influence on
lateral dispersion, for random packings of ‘‘isometric’’ particles.

Fluid Properties
4.4.2.1 Viscosity and Density of the Fluid
The effect of fluid densities and viscous forces on radial dispersion has been
studied by Grane and Garner [58] and Pozzi and Blackwell [106]. They concluded
that when a fluid is displaced from a packed bed by a less viscous fluid, the viscous
forces create an unstable pressure distribution and the less viscous fluid will
penetrate the medium in the form of fingers, unless the density has an opposing
effect.
2
10
1 1 1 1
= +
Pe T Pe m 12
PeT
10
Sand, G.Carvalho and Delgado (2001) Sand, List (1965)

Spheres, Bernard and Wilhelm (1950) Spheres, Hiby (1962)
Spheres, Coelho and G.Carvalho (1988) Spheres, Grane and Gardner (1961)
Cubes, Bernard and Wilhelm (1950) Cylinders, Bernard and Wilhelm (1950)
Pebble, Grane and Gardner (1961) Cylinders, England and Gunn (1970)
Hollow cylind, England and Gunn (1970) Raschig Rings, Hiby (1962)
1
2 3 4 5 6 7
1 10 10 10 10 10 10 10
Pem
Fig. 4.13 Effect of particle shape on radial dispersion
4.4.2.2 Fluid Velocity
For very low fluid velocities, u, dispersion is the direct result of molecular
diffusion, with DT ¼ D0m : As the velocity of the fluid is increased, the contri-
bution of convective dispersion becomes dominant over that of molecular dif-
fusion and DT becomes less sensitive to temperature. According to several authors
(see [15, 33, 66, 75, 144]) DT ! ud=PeT ð1Þ; for high enough values of u, where
d is particle size and PeT ð1Þ ffi 12 for beds of closely sized particles. Assuming
that the diffusive and convective components of dispersion are additive, the same
authors suggest that DT ¼ D0m þ ud=K; which may be written in dimensionless
form as
DT 1 1 ud 1 1 e 1
¼ þ or ¼ þ ð4:47Þ
Dm s 12 Dm PeT s ReSc 12
This equation has been shown (see [33]) to give a fairly accurate description of
radial dispersion in gas flow through packed beds, but it is not appropriate for the
description of dispersion in liquids, over an intermediate range of values of
ud=Dm ; as pointed out by several of the authors mentioned above.
Figure 4.14a–b shows that the value of the radial dispersion coefficient is seen
to increase with fluid velocity and comparison between the two plots shows that
DT also increases with particle size.
Data on dispersion in randomly packed beds of closely sized, near spherical
particles, lend themselves to simple correlation by means of dimensional analysis.
Making use of Buckingham’s theorem it may therefore be concluded that

DT ud l
¼U ; or PeL ¼ UðRe, ScÞ ð4:48Þ
Dm Dm qDm
and it is useful to make Pem ¼ ud=Dm and Sc ¼ l=qDm :
4.4.2.3 Fluid Temperature (or Schmidt Number)
The dependence of DT on liquid properties and velocity is best given in plots of

PeT vs. Pem ; for different values of Sc. Not surprisingly, Fig. 4.15 shows that the
variation of PeT with Pem gets closer to that for gas flow as the value of Sc is
decreased. For the lowest Sc tested (Sc = 54; T = 373 K), PeT does not differ by
more than 30% from the value given by Eq. 4.47, with PeT ð1Þ ¼ 12; over the
entire range of Pem : However, for the higher values of Sc, the experimental values
of PeT may be up to four times the values given by Eq. 4.47.
Delgado and Guedes de Carvalho [40] had studied the dependence of DT =Dm
on Sc, up to Pem ffi 1350; and they reported a smooth increase in DT =Dm with Pem ;
for all values of Sc. But the data in Fig. 4.15 show that there is a sudden change in
the trend of variation of PeT with Pem ; somewhere above Pem ffi 1; 350; a maxi-
mum being reached in the approximate range 1; 400\Pem \1; 800 (depending on
Sc). The fact that the change in trend corresponds to a much enhanced increase in
DT (i.e. a decrease in PeT Þ; in response to a small increase in u (i.e. in Pem Þ;
strongly suggests a connection with the transition from laminar to turbulent flow in
the interstices of the packing. The plot of PeT vs. Re, shown in Fig. 4.16, seems to
support this view, since the maxima in PeT are reached for 0:3\Re\10
(depending on Sc) and this is the approximate range of values of Re for the
transition from laminar to turbulent flow. The range 1\Re\10 is often indicated
for that transition (see for example [9]), but Scheidegger [116] as giving Re ¼ 0:1
for the lower limit of that transition.
The plot in Fig. 4.16 also suggests that ‘‘purely mechanical’’ fluid dispersion
will be observed above about Re ¼ 100; this value is estimated as the convergence
of the data points for liquids with the line representing Eq. 4.47. Figure 4.17 shows
the data reported by most other authors (all for Sc C 540) in a plot of PeT vs. Pem :
With the exception of the data of Hoopes and Harleman [78] and some of the points
of Grane and Gardner [58] and Bernard and Wilhelm [14], general agreement is
observed with Guedes de Carvalho and Delgado [62] data for high Sc.
4.5 Dispersion in Packed Beds Flowing by Non-Newtonian

Fluids
Hilal et al. [77], Edwards and Helail [46], Payne and Paker [101] and Wen and Yin
[142] reported results of axial dispersion coefficients for the flow of two polymer
solutions through a packed bed and their results were similar to the corresponding
(a) 3.0
T=373 K T=363 K
T=353 K T=343 K
2.5 T=333 K T=323 K
T=313 K T=303 K
T=293 K
D T (m 2/s).10 11
2.0
1.5
1.0
0.5
0.0
0 2 4 6 8 10 12 14 16
u (mm/s)
(b) 3.0
T=373 K T=363 K
T=353 K T=343 K
2.5
T=333 K T=323 K
T=303 K T=293 K
D T (m 2/s).10 11
2.0
1.5
1.0
0.5
0.0
0 1 2 3 4 5 6 7 8
u (mm/s)
Fig. 4.14 Variation of radial dispersion coefficient with fluid velocity. a sand size
d = 0.297 mm; b sand size d = 0.496 mm
Newtonian results (see Table 4.1). Wen and Fan [140] correlate the previous
results for packed beds with the following expression:

Ud qd n U 2n
Pe ¼ ¼ 0:2 þ 0:011Re0:48
n with Ren ¼ ð4:49Þ
DL m
where m is the power law consistency coefficient. Note that Eq. 4.49 for n = 1
(Newtonian fluids) reduces to the correlation obtained by Chung and Wen [31], for
Newtonian fluid through packed beds.
The only study on the influence of Non-Newtonian fluid in radial dispersion
coefficients is reported by Hassell and Bondi [73]. who showed that the quality of
mixing deteriorate with increasing viscosity.
4.6 Correlations 67
2
10
PeT
10
1 1 1 1
Pe T Pe m 12
Sc=54 Sc=71 Sc=101

Sc=142 Sc=200 Sc=288
Sc=420 Sc=754 Sc=993
Sc=1358 Sc=1930
1
1 10 10 2 10 3 10 4 10 5
Pem
Fig. 4.15 Dependence of PeT on Pem for different values of Sc
60
Sc=54 Sc=71
Sc=101 Sc=142 Re = 25
Sc=200 Sc=288
50 Sc=420 Sc=754
Sc=993 Sc=1358
Sc=1930
40
PeT
30
20 Eq. (4.32), Sc=750
10
0
0.001 0.01 0.1 1 10 100
Re
Fig. 4.16 Dependence of PeT on Re for different values of Sc
4.6 Correlations
Many equations have been proposed to correlate dispersion experimental data in

porous media, and the most used are the equations proposed by Bear and Verruijt
[10] in groundwater works and the equations proposed by Koch and Brady [83].
2
10
PeT 1 1 1 1
Pe T Pe m 12
10
Sc=754 (sand), G.Carvalho and Delgado (2001)

Sc=540 (glass spheres), Hiby (1962)
Sc=550 (sand), Blackwell (1962)
Sc=710 (plastic spheres), Harleman and Rumer (1963)
Sc=710 (sand), List (1965)
Sc=750 (ion-exchangeresin), Hartman et al. (1958)
Sc=754 (glass spheres), Bernard and Wilhelm (1950)
Sc=986 (glass spheres), Coelho and G.Carvalho (1988)
Sc=1070 (sand), Hoopes and Harleman (1965)
Sc=1270 (sand bonded by resin), Simpsom (1962)
Sc=1476 (glass spheres), Grane and Gardner (1961)
1
2 3 4 5 6
1 10 10 10 10 10 10
Pem
Fig. 4.17 Comparison between our data points and the results of other authors for Sc 540
The coefficients of radial and longitudinal dispersion are not linear function of
velocity and expressions with the form DL ¼ D0m þ aL un or DT ¼ D0m þ aT un are
suggested. The coefficients aT and aL are the radial and longitudinal dispersivities,
respectively, of the porous medium in the direction of transport and n is an
empirically constant, n ¼ 1 2 [53]. In most applications, the exponent, n, is
assumed to be unity, i.e., dispersion coefficient is assumed to be a linear function
of fluid velocity. However, n may be greater than unity in many situations [10].
Also, the dispersivity, aL ; is probably scale dependent with larger values for aL
being associated with greater transport distances [104]. For example, values of aL
reported from the results of field studies may be as much as four or six orders of
magnitude greater than the corresponding laboratory measured values which
commonly are found to range between 0.1 and 10 mm [53]. Ratios of aL =aT of
5:1–100:1 have been reported in the literature [10].
Some of most referred works were developed by Fried and Combarnous [54]
and Bear and Verruijt [10, p. 166]; the authors showed the existence of five
dispersion regimes, in unconsolidated porous media. Sahimi [115] and Marsily
[92] analyze the data compiled by Fried and Combarnous [54] to characterize
longitudinal dispersion in five dispersion regimes and radial dispersion in four
dispersion regimes and a hold-up dispersion. The (1) pure molecular diffusion
regime, for very low fluid velocity, is represented by Koplik et al. [84]
DL DT 1
¼ ¼ ð4:50Þ
Dm Dm s
The (2) superposition regime ð0:3\Pem \5Þ not quantify by an equation. In
this zone the effect of molecular diffusion and dispersion are of the same order of
magnitude. The (3) predominant mechanical dispersion ð5\Pem \300Þ:
4.6 Correlations
Table 4.1 Summary of the previous work with experimental data on axial and radial dispersion of non-newtonian fluids in packed beds
Reference Experimental method Packed bed e d (mm) L (mm) D (mm) Rec Sc
Wen and Yin [142] Pulse response Glass spheres 0.40 4.76 1066.8 50.8 7–800 %106
(H2O+Polyox 0.50 14.3
301)—Na2C20H10O5
(n = 0.90 and 0.81)
Payne and Parker [101] Step function Glass spheres 0.365 0.374 595 38.1 9 9 10-6– 8 9 106
(H2O+Polyox 1.5 9 10-3 –109
301)—blue dye
(n = 0.95 and 0.52)
Edwards and Helail Glass spheres
[46]
Hilal et al. [77] Frequency response Glass spheres 0.39 2–51.05 9
(Potassium parallelepiped 5.0
ferricyanide and particles (d = 0.42)
ferrocyanid +CMC
sodium
salt+Na2CO3)
Hassel and Bondi [73] Glass spheres 5.0
69
DL 1
¼ þ 0:5Pe1:2
m ð4:51aÞ
Dm s
DT 1
¼ þ 0:025Pe1:1
m ð4:51bÞ
Dm s

The (4) pure mechanical dispersion 300\Pem \105 represented by
DL 1
¼ þ ð1:8
0:4ÞPem ð4:52aÞ
Dm s
DT 1
¼ þ 0:025Pem ð4:52bÞ
Dm s
In this zone (yet in the range of validity of Darcy’s law) the effect
of molecular

diffusion is negligible. The (5) dispersion out of Darcy domain Pem [ 105 not
quantify by an equation. This is the region in which the effects of inertia and
turbulence cannot be neglected.
The (6) holdup dispersion [83]:
DL DT
¼ Pe2m ð4:53Þ
Dm Dm
Probably one of the most important studies in dispersion topic was presented by
Saffman [113, 114]. These works are the most detailed analysis of dispersion in
porous media, but the correlation proposed for axial and radial dispersion diverge
slightly from the existing experimental data, in the literature (see Fig. 4.18). The
author proposed the following equation, for longitudinal dispersion:

DL Pe0m 3 0 1
¼ ln sPem valid for Pe0m 1 ð4:54Þ
D0m 6 2 4
Another important study was presented by Koch and Brady [83] who showed
that the ratio of the dispersion coefficient to the molecular diffusivity is only a
function of the product ReSc: However, experimental measurements show that the
dispersion Peclet group is a function of both Reynolds and Schmidt groups as
Pe ¼ f ðRe; Sc) and not Pe ¼ f ðReSc): For Stokes flow through a random packed
bed of spheres, Koch and Brady [83] derived analytical expressions for the lon-
gitudinal and radial dispersion coefficients. Since the solid phase is not permeable
to the tracer, for Pe0m [ 1; DL =D0m and DT =D0m are given by:
0
DL 3 Pe0m p2 Pe0m Pem
¼1þ þ ð1 eÞ ln ð4:55Þ
D0m 4 2 6 2 2
pffiffiffi
DT 63 2 pffiffiffiffiffiffiffiffiffiffiffi Pe0m
¼1þ 1e ð4:56Þ
D0m 320 2
4.6 Correlations 71
The works of Saffman [113, 114] and Koch and Brady [83] have shown that
DL =D0m Pe0m ln(Pe0m Þ and DT =D0m Pe0m : The logarithmic dependence results
from the fact that velocity of fluid at the walls is zero. Different Peclet number
contributions
0 to dispersion were identified:
hold-up
in a closed circulation region
Pem 2 ; particle boundary layer Pe0m ln(Pe0m Þ and random velocity field contri-

bution Pe0m : The agreement with the experimental results of Fried and Com-
barnous [54] was very good for both longitudinal and radial dispersivities. Saffman
[113] modelled random porous media with randomly oriented capillary tubes and
found Fickian dispersion at long times with dispersivity approaching a Pe0m ln(Pe0m Þ
behaviour.
It’s important to enhance the fact that in literature, it was possible to find a large
number of empirical correlations. For example, Fetter [52] used the experimental
data obtained by Carberry and Bretton [25], Raimondi et al. [108], Blackwell et al.
[17] and Rosenberg [112] to conclude that longitudinal dispersion coefficient is
well represented by the fitted curve:
DL
¼ 1 þ 1:75Pe0m valid for Pe0m \50s ð4:57Þ
D0m
and the data of Blackwell [16] and Grane and Gardner [58], to suggest the fol-
lowing equation for radial dispersion:
DT
¼ 1 þ 0:055Pe0m ð4:58Þ
D0m
Gunn [65] admitted the existence of two regions in the packing, one of fast
flowing and the other of nearly stagnant fluid, to deduce the following expression
for the axial dispersion coefficient in terms of probability theory
2
1 ePem ePem
¼ 2 ð1 pÞ2 þ pð1 pÞ3
PeL 4a1 ð1 eÞ 4a21 ð1 eÞ

ð4:59Þ
4ð1 eÞa21 1
exp 1 þ
pð1 pÞe Pem s Pem
where a1 is the first zero of equation J0 ðUÞ ¼ 0 and p is defined, for a packing of
spherical particles, by

24 pffiffiffi
p ¼ 0:17 þ 0:20 exp for spheres; s ¼ 2 ð4:60aÞ
Re

24
p ¼ 0:17 þ 0:20 exp for solid cylinders; s ¼ 1:93 ð4:60bÞ
Re

24
p ¼ 0:17 þ 0:20 exp for hollow cylinders; s ¼ 1:8 ð4:60cÞ
Re
(a) 10 Gunn and Pryce (Sc=0.88) Edwards and Richardson (Sc=0.72)

Sc=1930 Sc=754
Sc=102 Sc=424
Sc=57 Sc=200
PeL
Sc=57
Sc=57
1
Sc=1930
Sahimi (1995)
Saffman (1960)
Sc=1930
Eq. Gunn (1969) Koch and Brady

Eq. Tsotsas and Schlunder (1988) (1985)
-1
10
-1 2 3 4 5 6 7
10 1 10 10 10 10 10 10 10
Pem
(b) 50
Wen and Fan
Wen and Fan
(1975), Sc=1930
(1975), Sc=54
40 Sahimi (1995)
Gunn (1969),
30 Sc=54
PeT
Gunn (1969),
20 Sc=1930
10
Koch and Brady Sc=54 Sc=101 Sc=200

Sc=288 Sc=420 Sc=754
(1985) Sc=1930
0
2 3 4 5 6
1 10 10 10 10 10 10
Pem
Fig. 4.18 Comparison between experimental data and correlations presented in the literature
Tsotsas and Schlunder [135] deduced an alternative correlation for the pre-
diction of PeL : The authors defining two zones in a simple flow model consisting
of a fast stream (central zone in the model capillary) and a stagnant fluid, but the
mathematical expressions associated with it are a little cumbersome,

1 1 1 1 1 Dc 2
¼ þ 2
1 nc þ Per;1 n2c f1 ðnc Þ þ Pe0m f2 ðnc Þ
PeL s Pez;1 Pe0m 32 d
ð4:61Þ
4.6 Correlations 73
where the axial and radial Peclet number of the fast stream is
1 1 1
¼ þ ð4:62aÞ
Pez;1 Pe01 1:14ð1 þ 10=Pe01 Þ
1 1 1
¼ 0 þ ð4:62bÞ
Per;1 Pe1 8
u1 d
Pe01 ¼ ð4:62cÞ
D0m
and u1 ¼ u=n2c is the interstitial velocity of the fast stream, with nc (the dimen-
sionless position of the velocity jump, i.e. the ratio between the radius of the zone
of high velocities and the radius of packed bed) equal to
Re 0:1 ! nc ¼ 0:2 þ 0:21exp(2:81yÞ ð4:62dÞ
Re 0:1 ! nc ¼ 1 0:59 exp½fðyÞ ð4:62eÞ

with
y ¼ logðReÞ þ 1 ð4:62fÞ

f(yÞ ¼ y 10:274y þ 0:086y2 ð4:62gÞ
Finally, the distributions functions f1 ðnc Þ and f2 ðnc Þ are defined by:
2
f1 ðnc Þ ¼ 1n2c ð4:62hÞ
f2 ðnc Þ ¼ 4n2c 3 4ln(nc Þ n4c ð4:62iÞ

In Fig. 4.18a, the lines corresponding to the correlations of Gunn [65] and of
Tsotsas and Schlunder [135] are represented, for the higher and lower values of Sc
in our experiments (Sc = 57 and 1,930), as well as for gas flow (Sc = 1). It may
be seen that the correlation of Gunn [65] is not sensitive to changes in Sc, for
Pem \103 ; and the correlation of Tsotsas and Schlunder [135] is much too sen-
sitive to variations in Sc; however, this correlation describes dispersion in gas flow
with good accuracy.
In this context it is interesting to consider, for radial dispersion, the predicting
accuracy of some alternative empirical correlations that have been proposed to
represent the experimental data in liquid flow, as the equation of Gunn [65]:
1 1 1 e
¼ þ ð4:63Þ
PeT Pef s ReSc
where the fluid-mechanical Peclet number, Pef ; is defined by,
pffiffiffi
Pef ¼ 40 29e7=Re for spheres; s ¼ 2 ð4:64aÞ
Pef ¼ 11 4e7=Re for solid cylinders; s ¼ 1:93 ð4:64bÞ
Pef ¼ 9 3:3e7=Re for hollow cylinders; s ¼ 1:8 ð4:64cÞ

And the empirical equation proposed by Wen and Fan [141],
17:5
PeT ¼ þ 11:4 ðfor high values of Pem Þ ð4:65Þ
Re0:75
In Fig. 4.18b, the lines corresponding Gunn [65] and Wen and Fan [141]
correlations are represented for the two extreme values of Schmidt observed in our
experiments (Sc = 54 and 1,930); comparison with the experimental points shows
that the correlations are very inadequate over significant ranges of Pem :
In conclusion, we can say that it was possible to find a large number of
equations in the literature that have been proposed to correlate dispersion exper-
imental data in porous media. However, the vast amount of data available is
obtained mostly for air and water at room temperature (the influence of Schmidt
number is not taking into account).
4.6.1 New Correlations: Axial Dispersion
In the limit of very low fluid velocity, dispersion is determined solely by molecular
diffusion, with DL ¼ Dm =s (s being the tortuosity factor for diffusion). At high
fluid velocities, dispersion is purely ‘‘fluid mechanical’’ (see [144]), with DL ¼
ud=PeL ð1Þ; where u is the interstitial fluid velocity and PeL ð1Þ ffi 2 for gas or
liquid flow through beds of (approximately) isometric particles, with diameter d. A
common approximation for the intermediate range of fluid velocities is to assume
that the effects of molecular diffusion and fluid mechanical dispersion are additive
and the resulting expression is given by
DL ¼ D0m þ ud=PeL ð1Þ ð4:66Þ
where D0m ¼ Dm =s: This equation is expected to give the correct asymptotic
behaviour at high and low values of Pem : In the case of gas flow, see Fig. 4.19,
significant deviations are observed in the range 0:6\Pem \60; the experimental
values of PeL are generally higher than predicted by Eq. 4.66, with PeL ð1Þ ffi 2:
Several equations have been proposed to represent the data in this intermediate
range and the equation presented by Hiby [75],
DL 0:65Pe0m
¼ 1 þ pffiffiffiffiffiffiffiffiffiffiffiffiffi ðvalid for Re\100Þ ð4:67Þ
D0m 1 þ 7 s=Pe0m
is shown to fit the data points reasonably well.

For most gaseous mixtures, far from the critical point, the value of Sc is close to
unity, whereas for most solutes in cold water, it is 550\Sc\2; 000: In recent
4.6 Correlations 75
years, data on longitudinal dispersion have been made available for values of Sc
between these ranges, such data were obtained with either supercritical carbon
dioxide ð1:5\Sc\20Þ or heated water ð55\Sc\550Þ:
Some workers have measured axial dispersion for the flow of supercritical
carbon dioxide through fixed beds and this provides important new data in the
range 1:5\Sc\20 (see Fig. 4.19). The experiments of Catchpole et al. [27]
were performed in the range 8\Sc\20; but the individual values of Sc, for the
data points represented, are not given by the authors. One would expect that the
data of Catchpole et al. [27] would lie somewhere between the series of points for
Sc = 57 and the line corresponding to Eq. 4.67. It turns out that values of PeL are
a little lower than expected, possibly as a result of low accuracy; the scatter is
certainly very pronounced. The data of Tan and Liou [132] represents some 90
data points by those authors, for 0:3\Re\135 and 1:5\Sc\3; the points are
more or less evenly distributed, meaning that scatter is very significant.
The data of Yu et al. [147] are for 0:01\Re\2 and 2:0\Sc\9:1: Unfortu-
nately they are not very consistent, particularly in the range 1\Pem \20; where
the scatter is high and the values of PeL are much too low. The experiments of
Ghoreishi and Akgermanb [56], again for the flow of supercritical carbon dioxide,
are for 0:1\Re\0:3 and 3:3\Sc\5:8: No influence of Sc on PeL is detected, but
this is what might be expected, considering that the values of Re are generally very
low, with the consequence that the points are partly in the range where dispersion
is totally determined by molecular diffusion.
For the case of liquid flow in a porous media, our group used the division in five
dispersion regimes to obtain the expressions presented below.
(1) Diffusion regime (valid for Pem \0:1Þ:
DL
¼1 ð4:68Þ
D0m
(2) Predominant diffusional regime (valid for 0:1\Pem \4Þ :
DL Pe0m
¼ ð4:69Þ
Dm 0:8=Pe0m þ 0:4
0
with a an average relative deviation lesser than 14%.

(3) Predominant mechanical dispersion (valid for 4\Pem and Re\10Þ :
DL Pe0m
¼ q ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi ð4:70Þ
D0m 18 Pe1:2 þ 2:35 Sc0:38
m
with a deviation lesser than 11%, over the entire range of Pe0m and Sc.
10
⎛1 1 ⎞
=⎜ ⎟
1 0.65
+
Pe L ⎜ Pe m 1 + 7 Pe −1 ⎟
⎝ m ⎠
1 1 1 1
= +
Pe L Pe m 2
PeL
-1
10
Sc=1.12, Johnson and Kapner (1990)
Sc=0.22, Johnson and Kapner (1990)
Sc=1.20, Suzuki and Smith (1972)
Sc=1.12, Gunn and Pryce (1969)
Sc=1.12, Edwards and Richardson (1968)
Sc=1.12, Sinclair and Potter (1965)
Sc=1.90, Blackwell et al. (1959)
Sc=0.22, Carberry and Breton (1955)
-2
10
-2 -1 2 3 4
10 10 1 10 10 10 10
Pem
Fig. 4.19 Axial dispersion in gas flow
(4) Pure mechanical dispersion (valid for 10\Re and Pem \106 Þ :
DL Pe0m
¼ ð4:71Þ
D0m 25Sc1:14 =Pe0m þ 0:5
with a an average relative deviation lesser than 16%, over the entire range of
Pe0m and Sc.
(5) Dispersion out of Darcy domain (valid for Pem [ 106 Þ :
DL Pe0m
¼ ð4:72Þ
D0m 2
The correlations proposed are shown (see Fig. 4.20) to be significantly more
accurate than previous correlations (see Fig. 4.18) and they cover the entire
spectrum of values of Pem and Sc expected to be useful. It is important to have in
mind that Eqs. 4.68–4.72 are recommended only for random packings of
approximately ‘‘isometric’’ particles.
4.6.2 New Correlations: Radial Dispersion
For gas flow, by simply adding the contributions of molecular diffusion and tur-
bulent dispersion, we suggested the well-known equation (in dimensionless form)
4.6 Correlations 77
DT Pe0m
¼ 1 þ ð4:73Þ
D0m 12
with a an average relative deviation lesser than 12%. Equation 4.73 give the
correct asymptotic behaviour (both for very high and very low Pem Þ for both gases
and liquids, as reported by several workers (see [65, 144]).
In the intermediate range of Pem they are still a reasonable approximation for
gases, the wider deviation being observed in the intervals 3\Pem \300; as shown
in Fig. 4.21.
For the case of liquid flow in a porous media, our group used the division in
four dispersion regimes to obtain the expressions presented below.
(1) Diffusion regime (valid for Pem \1Þ :
DT
¼1 ð4:74Þ
D0m
(2) Predominant mechanical dispersion (valid for 1\Pem \1600Þ :

DT 1
¼1þ for Sc\550 ð4:75aÞ
D0m 2:7 105 Sc þ 12=Pe0m
DT 1
¼1þ for Sc 550 ð4:75bÞ
0
Dm 0:017 þ 14=Pe0m
with a deviation lesser than 8 and 5%, respectively, over the entire range of
Pe0m and Sc.
(3) Pure mechanical dispersion (valid for 1; 600\Pem \106 Þ:
DT Pe0m
¼ for Sc\550 ð4:76aÞ
D0m 500Sc0:5
ð0:058Sc þ 14Þ ð0:058Sc þ 2Þ exp
Pe0m
DT Pe0m
¼ for Sc 550 ð4:76bÞ
D0m 21Sc
45:9 33:9 exp 0
Pem
and the experimental data do not deviate by more than 6 and 4% from the
values given by Eqs. 4.76a and 4.76b, respectively.
(4) Dispersion out of Darcy domain (valid for Pem [ 106 Þ:
DT Pe0m
¼ ð4:77Þ
D0m 12
10
Eqs. (4.68) - (4.72): Sc=57, Sc=102, Sc=200, Sc=424, Sc=754, Sc=1930
1 1 1 1
= +
PeL Pe L Pe m 2
Catchpole et al.
(1996), 8<Sc<20
Sc=1930 Sc=754
Sc=424 Sc=200
Ghoreishi and Akgermanb (2004), 3.3<Sc<5.8 Sc=102 Sc=57
Yu et al. (1999), 2.0<Sc<9.1 Jacques and Vermeulen (1958), Sc=820
-1
10
10-1 1 10 102 103 104 105 106 107
Pem
Fig. 4.20 Comparison between experimental data and correlations given by Eqs. 4.68–4.72
2
10
10 1 1 1 1
= +
Pe T Pe m 12
PeT
1
Sc=1.87, Coelho and Carvalho (1988)
Sc=0.77, Gunn and Pryce (1969)
Sc=1.20, Sinclair and Potter (1965)
Sc=1.10, Roemer et al. (1962)
Sc=1.00, Plautz and Johnstone (1955)
Sc=0.88, Plautz and Johnstone (1955)
Sc=0.94, Bernard and Wilhelm (1950)
-1
10
-1 2 3 4
10 1 10 10 10 10
Pem
Fig. 4.21 Radial dispersion in gas flow

4.6 Correlations 79
50
Sc=54 Sc=101 Sc=200
Sc=288 Sc=420 Sc=754
Sc=1930
40 Eqs. (4.74) - (4.77)
30
PeT
20
10
1 1 1 1
= +
Pe T Pe m 12
0
-1 2 3 4 5 6 7
10 1 10 10 10 10 10 10 10
Pem
Fig. 4.22 Comparison between experimental data and our correlations, for radial dispersion
The experimental data are shown in Fig. 4.22, alongside the solid lines corre-
sponding to Eqs. 4.74–4.77, for the values of Sc indicated in the figure. The
agreement is seen to be generally very good, even when the values of PeT are
represented on a linear scale. For Sc [ 550 (see experiments with values of Sc of
754 and 1,930), the above equations representing the data must take into account
that PeT is only dependent on Pem ; in the ascending part of the curve PeT vs. Pem
and that PeT only depends on Reð¼ e Pem =ScÞ; in the descending part of the same
curve.
In conclusion, the present chapter increases our knowledge about diffusion and
dispersion in packed beds by providing a critical analysis on the effect of fluid
properties and porous medium on the values of axial and radial dispersion
coefficients.
Different experimental techniques are presented in full detail and the data
obtained from these techniques are very similar. An improved technique for the
determination of the coefficient of radial dispersion in fluid flow through packed
beds is described more detailed, which is based on the measurement of the rate of
dissolution of buried flat or cylindrical surfaces.
A large number of experimental data on dispersion available in literature for
packed beds were examined to pave the way for the formulation of new correla-
tions for the prediction of PeT and PeL : The correlations proposed are shown to be
more accurate than previous correlations and they cover the entire range of values
of Pem and Sc. The axial dispersion coefficient can be calculated by Eqs. 4.68–
4.72 and the radial dispersion coefficient by Eqs. 4.73–4.77.
References
1. Adler, P.M., Jacquin, C.G., Thovert, J.F.: The formation factor of reconstructed porous
media. Water Resour. Res. 28, 1571–1576 (1992)
2. Ahn, B.J., Zoulalian, A., Smith, J.M.: Axial dispersion in packed beds with large wall effect.
AIChE J. 32, 170–174 (1986)
3. Akehata, T., Sato, K.: Flow distribution in packed beds. Chem. Eng. Jpn. 22, 430–436
(1958)
4. Archie, G.E.: The electrical resistivity log as an aid in determining some reservoir
characteristics. Trans. AIME 146, 54–61 (1942)
5. Aris, R.: On the dispersion of a solute in a fluid flowing through a tube. Proc. Roy. Soc. A,
Math. Phy. 235, 67–77 (1956)
6. Aris, R.: On the dispersion of a solute by diffusion, convection and exchange between
phases. Proc. Roy. Soc. A, Math. Phy. 252, 538–550 (1959)
7. Baron, T.: Generalized graphical method for the design of fixed bed catalytic reactors.
Chem. Eng. Prog. 48, 118–124 (1952)
8. Baumeister, E., Klose, U., Albert, K., Bayer, E.: Determination of the apparent transverse
and axial dispersion coefficients in a chromatographic column by pulsed field gradient
nuclear magnetic resonance. J. Chromatogr. A 694, 321–331 (1995)
9. Bear, J.: Dynamics of Fluids in Porous Media. Dover Publications, New York (1972)
10. Bear, J., Verruijt, A.: Modelling Groundwater Flow and Pollution, 1st edn. Springer, Berlin
(1987)
11. Beck, R.E., Schultz, J.S.: Hindered diffusion in microporous membranes with known pore
geometry. Sci 170, 1302–1305 (1970)
12. Benneker, A.H., Kronberg, A.E., Post, J.W., Van der Ham, A.G.J., Westerterp, K.R.: Axial
dispersion in gases flowing though a packed bed at elevated pressures. Chem. Eng. Sci. 51,
2099–2108 (1996)
13. Beran, M.J.: Dispersion of soluble matter in slowly moving fluids. Dissertation, Harvard
University, Cambridge (1955)
14. Bernard, R.A., Wilhelm, R.H.: Turbulent diffusion in fixed beds of packed solids. Chem.
Eng. Prog. 46, 233–244 (1950)
15. Bischoff, K.B.: A note on gas dispersion in packed beds. Chem. Eng. Sci. 24, 607–608
(1969)
16. Blackwell, R.J.: Laboratory studies of microscopic dispersion phenomena. Soc. Petrol. Eng.
J. 225, 1–8 (1962)
17. Blackwell, R.J., Rayne, J.R., Terry, W.M.: Factors influencing the efficiency of miscible
displacement. Petrol. Trans. AIME 216, 1–8 (1959)
18. Boudreau, B.P.: The diffusive tortuosity of fine-grained unlithified sediments. Geochim.
Cosmochim. Acta 60, 3139–3142 (1996)
19. Brenner, H.: The diffusion model of longitudinal mixing in beds of finite length. Numerical
values. Chem. Eng. Sci. 17, 229–243 (1962)
20. Brenner, H.: Dispersion resulting from flow through spatially periodic porous media. Philos.
Trans. R. Soc. A 297, 81–133 (1980)
21. Bruggeman, D.A.G.: Calculation of different physical constants of heterogeneous
substances i, dielectric constant and conductivity of media of isotropic substances. Ann.
Phys. Ser.-5 24, 636–664 (1935)
22. Bruinzeel, C., Reman, G.H., van der Laan, E.T.H.: Axial dispersion. Proceedings of the 3rd
Congress of the European Federation of Chemical Engineering, Olympia, London, UK
(1962)
23. Cairns, E.J., Prausnitz, J.M.: Longitudinal mixing in packed beds. Chem. Eng. Sci. 12,
20–34 (1960)
24. Carberry, J.J.: Chemical and Catalytic Reaction Engineering. McGraw-Hill Chemical
Engineering Series, USA (1976)
References 81
25. Carberry, J.J., Bretton, R.H.: Axial dispersion of mass in flow through fixed beds. AIChE J.
4, 367–375 (1958)
26. Carbonell, R.G., Whitaker, S.: Dispersion in pulsed systems. Part II: Theoretical
developments for passive dispersion in porous media. Chem. Eng. Sci. 38, 1795–1801
(1983)
27. Catchpole, O.J., Berning, R., King, M.B.: Measurement and correlations of packed bed axial
dispersion coefficients in supercritical carbon dioxide. Ind. Eng. Chem. Res. 35, 824–828
(1996)
28. Chantong, A., Massoth, F.E.: Restrictive diffusion in aluminas. AIChE J. 29, 725–731
(1983)
29. Chao, R., Hoelscher, H.E.: Simultaneous axial dispersion and adsorption in packed beds.
AIChE J. 12, 271–278 (1966)
30. Choudhary, M., Szekely, J., Weller, S.W.: The effect of flow maldistribution on conversion
in a catalytic packed bed reactor. AIChE J. 22, 1021–1032 (1976)
31. Chung, S.F., Wen, C.Y.: Longitudinal dispersion of liquid flowing through fixed and
fluidized beds. AIChE J. 14, 857–866 (1968)
33. Coelho, M.A.N., Guedes de Carvalho, J.R.F.: Transverse dispersion in granular beds. Part I:
Mass transfer from a wall and the dispersion coefficient in packed beds. Chem. Eng. Res.
Des. 66, 165–177 (1988)
34. Collins, M.: Velocity distribution in packed beds. Dissertation, University of Delaware
(1958)
35. Comiti, J., Renaud, M.: A new model for determining mean structure parameters of fixed
beds from pressure drops measurements, application to beds packed with parallelepipedal
particles. Chem. Eng. Sci. 44, 1539–1545 (1989)
36. Crank, J.: The Mathematics of Diffusion, 2nd edn. Oxford University Press, Oxford (1975)
37. Danckwerts, P.V.: Continuous flow systems. Chem. Eng. Sci. 2, 1–13 (1953)
38. De Jong, G.J.: Longitudinal and transverse diffusion in granular deposits. Trans. AGU 39,
67–75 (1958)
39. Deisler, P.F., Wilhelm, R.H.: Diffusion in beds of porous solids - measurement by frequency
response techniques. Ind. Eng. Chem. 45, 1219–1227 (1953)
40. Delgado, J.M.P.Q., Guedes de Carvalho, J.R.F.: Measurement of the coefficient of
transverse dispersion in packed beds over a range of values of Schmidt number 50–1000.
Transp. Porous Med. 44, 165–180 (2001)
41. Delmas, H., Froment, G.F.: Simulation model accounting for structural radial
nonuniformities in fixed bed reactors. Chem. Eng. Sci. 43, 2281–2287 (1988)
42. Dorweiler, V.P., Fahien, R.W.: Mass transfer at low flow rates in a packed column. AIChE
J. 5, 139–144 (1959)
43. Dullien, F.A.L.: Single phase flow through porous media and pore structure. Chem. Eng. J.
10, 1–34 (1975)
44. Dullien, F.A.L.: Porous media, fluid transport and pore structure. Academic, San Diego
(1979)
45. Ebach, E.A., White, R.R.: Mixing of fluids flowing through beds of oacked solids. AIChE J.
4, 161–169 (1958)
46. Edwards, M.F., Helail, T.R.: Axial dispersion in porous media. Proceedings of the 2nd
European Conference of Mixing, BHRA Fluids Engineering, Cranfield, UK (1977)
47. Edwards, M.F., Richardson, J.F.: Gas dispersion in packed beds. Chem. Eng. Sci. 23,
109–123 (1968)
48. Eidsath, A., Carbonell, R.G., Whitaker, S., Herrmann, L.R.: Dispersion in pulsed systems.
Part III: Comparison between theory and experiments for packed beds. Chem. Eng. Sci. 38,
1803–1816 (1983)
49. England, R., Gunn, D.J.: Dispersion, pressure drop, and chemical reaction in packed beds of
cylindrical particles. Trans. IChemE. 48, T265–T275 (1970)
50. Evans, E.V., Kenney, C.N.: Gaseous dispersion in packed beds at low Reynolds numbers.
Trans. IChemE. 44, T189–T197 (1966)
51. Fahien, R.W., Smith, J.M.: Mass transfer in packed beds. AIChE J. 1, 28–37 (1955)
52. Fetter, C.W.: Contaminant hydrogeology, 2nd edn. Prentice-Hall, New Jersey (1999)
53. Freeze, R.A., Cherry, J.A.: Groundwater. Prentice-Hall Inc, Englewood Cliffs (1979)
54. Fried, J.J., Combarnous, M.A.: Dispersion in porous media. Advances in Hydroscience n8 7.
Academic, New York (1971)
55. Froment, G.F., Bischoff, K.B.: Chemical reactor analysis and design, 2nd edn. Wiley, New
York (1990)
56. Ghoreishi S.M., Akgermanb, A.: Dispersion coefficients of supercritical fluid in fixed beds.
Sep. Purif. Technol. 39, 39–50 (2004)
57. Gibbs, S.J., Lightfoot, E.N., Root, T.W.: Protein diffusion in porous gel filtration
chromatography media studied by pulsed field gradient NMR spectroscopy. J. Phys.
Chem. 96, 7458–7462 (1992)
58. Grane, F.E., Gardner, G.H.F.: Measurements of transverse dispersion in granular media.
J. Chem. Eng. Data 6, 283–287 (1961)
59. Gray, W.G.: A derivation of the equations for multi-phase transport. Chem. Eng. Sci. 30,
229–233 (1975)
60. Greenkorn, R.A., Kessler, D.P.: Dispersion in heterogeneous nonuniform anisotropic porous
media. Ind. Eng. Chem. 61, 8–15 (1969)
61. Guedes de Carvalho, J.R.F., Delgado, J.M.P.Q.: The effect of fluid properties on dispersion
in flow through packed. AIChE J. 49, 1980–1985 (2003)
62. Guedes de Carvalho, J.R.F., Delgado, J.M.P.Q.: Radial dispersion in liquid flow through
packed beds for 50 \ Sc \ 750 and 103 \ Pem \ 105. Proceedings of the 5th World
Conference on Experimental Heat Transfer, Fluid Mechanics and Thermodynamics,
Thessaloniki, Greece (2001)
63. Guedes de Carvalho, J.R.F., Delgado, J.M.P.Q.: Lateral dispersion in liquid flow through
packed beds at Pem \ 1,400. AIChE J. 46, 1089–1095 (2000)
64. Gunn, D.J.: Axial and radial dispersion in fixed beds. Chem. Eng. Sci. 42, 363–373 (1987)
65. Gunn, D.J.: Theory of axial and radial dispersion in packed beds. Trans. IChemE. 47, T351–
T359 (1969)
66. Gunn, D.J.: Mixing in packed and fluidised beds. Chem. Eng. J. 5, E153–CE172 (1968)
67. Gunn, D.J., Malik, A.A.: Flow through expanded beds of solids. Trans. IChemE. 44, T371–
T379 (1966)
68. Gunn, D.J., Pryce, C.: Dispersion in packed beds. Trans. IChemE. 47, T341–T350 (1969)
69. Han, N.W., Bhakta, J., Carbonell, R.G.: Longitudinal and lateral dispersion in packed beds,
effect of column length and particle size distribution. AIChE J. 31, 277–288 (1985)
70. Haring, R.E., Greenkorn, R.A.: Statistical model of a porous medium with nonuniform
pores. AIChE J. 16, 477–483 (1970)
71. Harleman, D.R.F., Rumer, R.: Longitudinal and lateral dispersion in an isotropic porous
medium. J. Fluid Mech. 16, 1–12 (1963)
72. Hartman, M.E., Wevers, C.J.H., Kramers, H.: Lateral diffusion with liquid flow through a
packed bed of ion-exchange particles. Chem. Eng. Sci. 9, 80–82 (1958)
73. Hassel, H.L., Bondi, A.: Mixing of viscous non-newtonian fluids in packed beds. AIChE J.
11, 217–221 (1965)
74. Hennico, A., Jacques, G., Vermeulen, T.: Longitudinal dispersion in single-phase liquid
flow through ordered and random packings. Lawrence Rad. Laboratory Report UCRL,
10696 (1963)
75. Hiby, J.W.: Longitudinal and transverse mixing during single-phase flow through granular
beds. In: Interaction between fluid and particles, London IChemE, pp. 312–325, (1962)
76. Hiby, J.W., Schummer, P.: Zur messung der transversalen effektiven diffusion in
durchstromten fullkorpersaulen. Chem. Eng. Sci. 13, 69–74 (1960)
77. Hilal, M., Brunjail, D., Combi, J.: Electrodiffusion characterization of non-newtonian flow
through packed-beds. J. Appl. Electrochem. 21, 1087–1090 (1991)
References 83
78. Hoopes, J.A., Harleman, D.R.F.: Waste water recharge and dispersion in porous media. MIT
Hydrodynamics Laboratory Report 75, 55–60 (1965)
79. Jacques, G.L., Vermeulen, T.: Longitudinal dispersion in solvent-extraction columns, peclet
numbers for random and ordered packings. University of California Rad. Laboratory Report
N8 8029, US Atomic Energy Commission, Washington DC, USA (1958)
80. Johnson, G.W., Kapner, R.S.: The dependence of axial-dispersion on non-uniform flows in
beds of uniform packing. Chem. Eng. Sci. 45, 329–3339 (1990)
81. Klinkenberg, A., Krajenbrink, H.J., Lauwerier, H.A.: Diffusion in a fluid moving at uniform
velocity in a tube. Ind. Eng. Chem. 45, 1202–1208 (1953)
82. Klotz, D.: Untersuchungen zur dispersion in porosen medien. Z. Deutsch. Geol. Ges. 124,
523–533 (1973)
83. Koch, D.C., Brady, J.F.: Dispersion in fixed beds. J. Fluid Mech. 154, 399–427 (1985)
84. Koplik, J., Redner, S., Wilkinson, D.: Transport and dispersion in random networks with
percolation disorder. Phys. Rev. A 37, 2619–2636 (1988)
85. Kramers, H., Alberda, G.: Frequency response analysis of continuous flow systems. Chem.
Eng. Sci. 2, 173–181 (1953)
86. Langer, G., Roethe, A., Roethe, K.P., Gelbin, D.: Heat and mass-transfer in packed beds.
Part III: Axial mass dispersion. Int. J. Heat Mass Tran. 21, 751–759 (1978)
87. Latinen, G.A.: Mechanism of fluid-phase mixing in fixed and fluidised beds of uniformly
sized spherical particles. Dissertation, Princeton University (1951)
88. Lerou, J.J., Froment, G.F.: Velocity, temperature and conversion profiles in fixed bed
catalytic reactors. Chem. Eng. Sci. 32, 853–861 (1977)
89. Levenspiel, O., Smith, W.K.: Notes on the diffusion-type model for the longitudinal mixing
of fluids in flow. Chem. Eng. Sci. 6, 227–233 (1957)
90. Liles, A.W., Geankopolis, C.J.: Axial diffusion of liquids in packed beds and end effects.
AIChE J. 6, 591–595 (1960)
91. List, E.J.: The stability and mixing of a density-stratified horizontal flow in a saturated
porous medium. WM Keck Lab Hydraulics Water Resources Reporter KH-R-11, USA
(1965)
92. Marsily, G.: Quantitative Hydrogeology, 1st edn. Academic, Orlando (1986)
93. Maxwell, J.C.: Treatise on Electricity and Magnetism. Oxford University Press, London
(1873)
94. McHenry, J.R., Wilhelm, R.H.: Axial mixing of binary gas mixtures flowing in a random
bed of spheres. AIChE J. 3, 83–91 (1957)
95. Miller, S.T., King, C.J.: Axial dispersion in liquid flow through packed beds. AIChE J. 12,
767–773 (1966)
96. Millington, R.J., Quirk, J.P.: Transport in porous media. Proceedings of the 7th International
Congress of Soil Science, Madison, Wisc., USA (1960)
97. Miyauchi, T., Kikuchi, T.: Axial dispersion in packed beds. Chem. Eng. Sci. 30, 343–348
(1975)
98. Niemann, E.H.; Dispersion during flow nonuniform heterogeneous porous media.
Dissertation, Purdue University (1969)
99. Ogata, A.: The spread of a dye stream in an isotropic granular medium. US Geol. Surv. Prof.
Pap. 411-G, pp. 1–11. USA (1964)
100. Otake, T., Kunugita, E.: Axial dispersion of the gas phase in countercurrent packed bed
columns. Chem. Eng. Jpn. 22, 144–150 (1958)
101. Payne, L.W., Parker, H.W.: Axial dispersion of non-newtonian fluids in porous media.
AIChE J. 19, 202–204 (1973)
102. Perkins, T.K., Johnston, O.C.: A review of diffusion and dispersion in porous media. Soc.
Petrol. Eng. J. 2, 70–84 (1963)
103. Pfannkuch, H.O.: Contribution a l’etude des déplacements de fluids miscibles dans un
milieu poreux. Rev. Inst. Franc. Pétrol. 18, 215–219 (1963)
104. Pickens, J.F., Grisak, G.E.: Scale-dependent dispersion in a stratified granular aquifer.
Water Resour. Res. 17, 1191–1211 (1981)
105. Plautz, D.A., Johnstone, H.F.: Heat and mass transfer in packed beds. AIChE J. 1, 193–199
(1955)
106. Pozzi, A.L., Blackwell, R.J.: Design of laboratory models for study of miscible
displacement. Soc. Petrol. Eng. J. 3, 28–40 (1963)
107. Probst, K., Wohlfahrt, K.: Empirische abschätzung effektiver diffusionskoeffizienten in
porösen systemen. Chem. Eng. Tech. 1, 737–739 (1979)
108. Raimondi, P., Gardner, G.H.F., Petrick, C.B.: Effect of pore structure and molecular
diffusion on the mixing of miscible liquids flowing in porous media. In: AIChE-SPE joint
symposium, San Francisco (1959)
109. Rifai, M.N.E., Kaufman, W.J., Todd, D.K.: Dispersion phenomena in laminar flow through
porous media. University of California, Sanitary Engineering Report 3. Inst. Eng. Res. Ser.
90, 1–157 (1956)
110. Roblee, L.H.S., Baird, R.M., Tierney, J.W.: Radial porosity variations in packed beds.
AIChE J. 4, 460–468 (1958)
111. Roemer, G., Dranoff, J.S., Smith, J.M.: Diffusion in packed beds at low flow rates. Ind. Eng.
Chem. Fund. 1, 284–287 (1962)
112. Rosenberg, D.V.: Axial dispersion in porous media. J. Am. Inst. Chem. Engrs. 2, 55–64
(1956)
113. Saffman, P.G.: A theory of dispersion in a porous medium. J. Fluid Mech. 6, 321–325
(1959)
114. Saffman, P.C.: Dispersion in flow through a network of capillaries. J. Fluid Mech. 7,
194–207 (1960)
115. Sahimi, M.: Flow and transport in porous media and fractured rock. VCH
Velagsgesellschaft mbH, Weinheim (1995)
116. Scheidegger, A.E.: The physics of flow through porous media, 3rd edn. University of
Toronto Press, Toronto (1974)
117. Schuster, J., Vortmeyer, D.: Ein einfaches verfahren zur näherungsweisen bestimmung der
porosität in schttungen als funktion des wasdabstandes. Chem. Eng. Tech. 52, 848–855
(1980)
118. Schwartz, C.E., Smith, J.M.: Flow distribution in packed beds. Ind. Eng. Chem. 45,
1209–1218 (1953)
119. Scott, D.S., Lee, W., Papa, J.: The measurement of transport coefficients in gas-solid
heterogeneous reactions. Chem. Eng. Sci. 29, 2155–2167 (1974)
120. Sen, P.N., Scala, C., Cohen, M.H.: A self-similar model for sedimentary-rocks with
application to the dielectric-constant of fused glass-beads. Geophysics 46, 781–795 (1981)
121. Sherwood, T.K., Pigford, R.L., Wilke, C.R.: Mass transfer. McGraw-Hill, Tokyo (1975)
122. Shimamura, K.: Gas diffusion through compacted sands. Soil Sci. 153, 274–279 (1992)
123. Simpson, E.S.: Transverse dispersion in liquid flow through porous media. US Geol. Surv.
Prof. Pap. 411-C, pp. 1–30. USA (1962)
124. Sinclair, R.J., Potter, O.E.: The dispersion of gas in flow through a bed of packed solids.
Trans. IChemE. 43, T3–T9 (1965)
125. Singer, E., Wilhelm, R.H.: Heat transfer in packed beds, analytical solution and design
method. Fluid flow, solids flow and chemical reaction. Chem. Eng. Prog. 46, 343–352
(1950)
126. Slattery, J.C.: Momentum, energy and non transfer in continua. McGraw-Hill, New York
(1972)
127. Slichter, C.S.: Field measurement of the rate of movement of underground waters. US Geol.
Surv. Water Supply Pap. 140, USA (1905)
128. Smith, W.D., Bretton, R.H.: Paper presented at AIChE Houston meeting, Texas, USA
(1967)
129. Stephenson, J.L., Stewart, W.E.: Optical measurements of porosity and fluid motion in
packed beds. Chem. Eng. Sci. 41, 2161–2170 (1986)
130. Strang, D.A., Geankopolis, C.J.: Longitudinal diffusivity of liquids in packed beds. Ind.
Eng. Chem. 50, 1305–1308 (1958)
References 85
131. Suzuki, M., Smith, J.M.: Dynamics of diffusion and adsorption in a single catalyst pellet.
AIChE J. 18, 326–333 (1972)
132. Tan, C.S., Liou, D.C.: Axial dispersion of supercritical carbon dioxide in packed beds. Ind.
Eng. Chem. Res. 28, 1246–1250 (1989)
133. Taylor, G.: Dispersion of soluble matter in solvent flowing slowly through a tube. Proc.
Roy. Soc. A, Math. Phy. 219, 186–203 (1953)
134. Towle, W.L., Sherwood, T.K.: Studies in eddy diffusion. Ind. Eng. Chem. 31, 457–467
(1939)
135. Tsotsas, E., Schlunder, E.U.: On axial dispersion in packed beds with fluid flow. Chem. Eng.
Proc. 24, 15–31 (1988)
136. Urban, J.C., Gomezplata, A.: Axial dispersion coefficients in packed beds at low Reynolds
numbers. Can. J. Chem. Eng. 47, 353–363 (1969)
137. Vortmeyer, D., Schuster, J.: Evaluation of steady flow profiles in rectangular and circular
packed beds by a variational method. Chem. Eng. Sci. 38, 1691–1699 (1983)
138. Vortmeyer, D., Winter, R.P.: Impact of porosity and velocity distribution of the theoretical
prediction of fixed-bed chemical reactor performance, comparison with experimental data.
ACS symposium series, pp. 49–61. USA (1982)
139. Weissberg, H.L.: Effective diffusion coefficient in porous media. J. Appl. Phys. 34,
2636–2639 (1963)
140. Wen, C.Y., Fan, L.S.: Axial dispersion of non-Newtonian liquids in fluidized beds. Chem.
Eng. Sci. 28, 1768–1773 (1973)
141. Wen, C.Y., Fan, L.T.: Models for systems and chemical reactors. Marcel Dekker, New York
(1975)
142. Wen, C.Y., Yin, J.: Axial dispersion of a non-newtonian liquid in a packed bed. AIChE J.
17, 1503–1504 (1971)
143. Whitaker, S.: Diffusion and dispersion in porous media. AIChE J. 13, 420–432 (1967)
144. Wilhelm, R.H.: Progress towards the a priori design of chemical reactors. Pure Appl. Chem.
5, 403–421 (1962)
145. Wilson, H.A.: On convection of heat. P. Camb. Philos. Soc. 12, 406–423 (1904)
146. Wronski, S., Molga, E.: Axial dispersion in packed beds: the effect of particle size non-
uniformities. Chem. Eng. Prog. 22, 123–135 (1987)
147. Yu, D., Jackson, K., Harmon, T.C.: Dispersion and diffusion in porous media under
supercritical conditions. Chem. Eng. Sci. 54, 357–367 (1999)
148. Yun, M.J., Yu, B.M., Zhang, B., Huang, M.Y.: A geometry model for tortuosity of
streamtubes in porous media with spherical particles. Chin. Phys. Lett. 22, 1464–1467
(2005)
Chapter 5
Modeling of Transport Processes
in Porous Materials
5.1 Introduction
As far as porous materials are characterized by high complexity in their invisible

internal structure, modeling seems to be a grateful tool for understanding the
transport processes that take place in porous media. Several modeling approaches,
applicable in porous domain, have been introduced and used in a variety of
scientific and industrial applications. Numerous industrial and technological
applications involving fluid flow and mass transport processes within multi-par-
ticle assemblages have attracted scientific interest in recent decades. These focus
mainly on industrial physicochemical processes (e.g. sedimentation, catalysis),
alternative energy sources (e.g. fuel cells) and separation techniques (e.g. chro-
matography, filters). This chapter focuses on the modeling of mass transport in
granular structures as well as on the estimation of macroscopic quantities (such as
mass transport coefficient, adsorption efficiency, etc.).
5.2 Single Phase Transport in Unit Cells
Although arrays of regularly spatially distributed grains represent an idealization of

real granular media, they have been widely studied from both the fluid dynamics and
mass transport points of view. Due to their complex geometry, random particle
distributions were the subject of few investigations until about twenty years ago.
Since then, fast advances in computational capabilities have contributed to reviving
research on this topic with emphasis on hydrodynamic aspects. Very often the par-
ticles are assumed to be sufficiently small and the physical properties of the fluid
(i.e. viscosity and density) independent of the concentration of the transferred species,
thus making it possible to uncouple the fluid problem from the mass-transport
problem. Furthermore, the case of low Reynolds number flow is very important from
an application point of view, where analytical solutions can be obtained for the flow
field and the mass diffusion and/or adsorption process.

88 5 Modeling of Transport Processes in Porous Materials
5.2.1 Fundamental Quantities
Generally speaking, the transient mass transport of a passive solute in the void
space of a porous medium is described by the convection–diffusion equation [4]:
@cA
þ v rcA ¼ Dr2 cA þ RðcA Þ ð5:1Þ
@t
where cA is the concentration of the passive solute, v is its velocity in the medium,
D is its diffusivity in the intermediate occupying the medium’s void space, and
RðcA Þ describes the reactions taking place in the bulk phase.
For the sake of simplicity, the most common assumptions made in the relative
literature and applied in the forthcoming analysis are:
@cA
• steady state condition (i.e. ¼ 0)
@t
• inert bulk phase (i.e. RðcA Þ ¼ 0).
Consequently, the governing differential equation for mass transport in a porous
material can often be written as:
v rcA ¼ Dr2 cA ð5:2Þ

The LHS of the above equation describes the mass transport due to convection
while RHS stands for the mass transport due to diffusion. The relative dominance
between these simultaneously occurring phenomena is expressed by the dimen-
sionless Peclet number, defined as:
characteristic velocity characteristic length U L
Pe ¼ ¼ ð5:3Þ
diffusivity D
Obviously, the higher the Pe, the more convective the mass transport regime.
The microscopic quantity cA is related to useful macroscopic quantities, such as
the mass transport coefficient, the Sherwood (Nusselt) number, and the adsorption
efficiency. A brief description of these quantities is given below.
The mass transport coefficient, k0 ; describes the ability of the mass to be
transferred from the bulk phase to solid absorbers. It is obviously dependant on the
physical properties of the materials involved in the procedure, the porous structure,
and some physical parameters such as pressure, temperature and volumetric flow-
rate, which could be considered critical for the process. Mathematically, the mass
transport coefficient is defined by the expression:
ZZ
k0 ðDcA ÞSabs ¼ ½NA surf dSabs ð5:4Þ
Sabs
where DcA is the absolute difference of the concentration upon the absorbing
surface Sabs from the incoming concentration, and ½NA surf is the component of
molar flux which is normal on the collectors surface. The dimensionless overall
5.2 Single Phase Transport in Unit Cells 89
Sherwood number is a direct consequence of the mass transport coefficient and is

defined as:
mass transport coefficient characteristic length ko L
Sho ¼ ¼ ð5:5Þ
diffusivity D
For practical applications, the most useful macroscopic quantity is the
adsorption efficiency, k0 ; defined by the ratio of the solute adsorption rate divided
by the rate of the upstream influx:
RR
cA v ndS
k0 ¼ 1 RRSoutlet ð5:6Þ
Sinlet cA v ndS
where n is the vector normal to the surface. For cases regarding the adsorption
efficiency of a grain-in-cell, the above expression is modified to:
RR
½NAr surf dSabs
Sabs
k0 ¼ ð5:7Þ
u1 cA;1 Sinlet
where u1 is the approaching velocity, cA;1 is the constant bulk concentration

sufficiently far away from the collector surface, and Sinlet is the inlet surface that
the solute faces to pass through the porous medium.
5.2.2 Adsorption Mechanisms
The majority of scientific and technological applications that deal with mass
transport in porous materials, involve a sorption mechanism. Generally speaking,
sorption includes adsorption and ion exchange, processes that involve the transfer
and resulting equilibrium distribution of one or more solutes between a fluid phase
and a solid surface. More precisely, adsorption generally involves the accumula-
tion (or depletion) of solute molecules at an interface (gas–solid and liquid–solid
interfaces), with solute distributed selectively between the fluid and solid phases.
On the other hand, ion exchange occurs when positively (cations) or negatively
(anions) charged ions from the fluid replace dissimilar ions of the same charge,
being initially in the solid surface. Furthermore, the partitioning of a single solute
between fluid and sorbed phases can be also included in the sorption mechanism.
Phase equilibrium between liquid and sorbed phases for one or many compo-
nents in adsorption, or two or more species in ion exchange, is considered the
single most important factor affecting process performance. This equilibrium is
usually expressed by a boundary condition upon the liquid–solid interface, while it
is of great importance to obtain a simple expression for the detailed description of
the sorption mechanism. In relevant literature, several different cases for the
adsorption of the solute upon the solid surface have been considered and used.
The first and simplest case corresponds to instantaneous adsorption upon the
solid surface [17] and can be expressed as:
cA ¼ 0; on the solidliquid interface ð5:8Þ
Despite its mathematical simplicity, it is clear that the instantaneous adsorption
approach pertains to a very limited range of applications, being a fairly rare
physicochemical phenomenon. Thus, detailed models referring to adsorption fol-
lowing isotherms have also been adopted. The most common is that of Langmuir’s
isotherm, which can be formulated as [12]:
k
Dn rcA ¼ cs ; on the solidliquid interface ð5:9Þ
K
where cs is the concentration of the sorpted material, K is a constant defined by the
Langmuir isotherm:
Kcb
Heq ¼ ð5:10Þ
1 þ Kcb
and k is a reaction rate defined from the relation
Rðcs Þ ¼ kcb ðcmx cs Þ ð5:11Þ
where R(cs) is the overall adsorption rate given as a function of the surface con-
centration cs, cb is the concentration of the diluted mass in the neighborhood of the
solid surface, cmx is the maximum concentration attained when the surface is
completely covered by substance A, and Heq is the ratio of the covered to the total
surface, defined as:
cs
Heq ¼ ð5:12Þ
cmx
Another more detailed approach is the more complicated but realistic sorption
mechanism, which involves adsorption-reaction-desorption [13]. More specifi-
cally, it can be considered that the solute diluted in the bulk phase is initially
adsorbed by the solid surface where a heterogeneous reaction takes place, and its
products, assumed to be inactive and of very low concentrations, are again des-
orbed in the bulk phase. The adsorption is assumed to occur due to vacant sites [2]
that are normally distributed over the surface area while the whole process is
determined by an overall rate according to thermodynamics. The above mecha-
nism is often described by an expression of the form:
Dn rcA ¼ Rn ; on the solidliquid interface ð5:13Þ
With Rn denoting the overall adsorption rate, given as:
Rn ¼ ks cnAS ð5:14Þ
where ks is the rate constant of the heterogeneous reaction upon the surface, and
the concentration of solute on the solid surface, cAS ; is calculated by solving the
non linear equation:
h i
ks cnAS þ kAd þ kAa cA Þsurf N cAS kAa cA Þsurf nm ¼ 0 ð5:15Þ
Equation 5.15 correlates the hard-to-measure surface concentration of A, cAS ;

with its concentration in the bulk phase very close to the solid surface cA Þsurf by
considering the balance for the active sites on the adsorbing surface. In the above
equation, the terms kAa and kAd denote the adsorption and desorption rate constants of
solute, respectively, nm is the concentration of the vacant sites of the solid surface,
and N is Avogadro’s number, while n denotes the order of the heterogeneous
reaction upon the surface. In general, only the cases of n ¼ 0; n ¼ 1 and n ¼ 2 are
of practical importance, but zero order reactions are also used in limited appli-
cations [2].
Finally, it should be noted that some models consider neutral solid surface i.e.
absolute absence of adsorption [11]. This case can be described by the following
boundary condition upon the solid surface:
n rcA ¼ 0; on the solidliquid interface ð5:16Þ
5.2.3 Mass Transport Through Spheres
As discussed in Sect. 2.2, sphere-in-cell geometry is a simple model that represents

the actual complex geometry of the pore space in spherical particle assemblages
and the approximation of the flow-field therein. Consider a typical sphere-in-cell
model where the approaching fluid is a dilute solution of a substance A, which is
moving towards the solid adsorbing surface. After taking into account the plane
symmetry of the problem, the governing equation for the steady state mass
transport in the fluid phase within the model can be written in spherical coordi-
nates ðr; hÞ as:
2
@cA uh @cA @ cA 2 @cA 1 @ 2 cA cot h @cA
ur þ ¼D þ þ þ 2 ð5:17Þ
@r r @h @r 2 r @r r 2 @h2 r @h
where ur and uh are the r and hvelocity components, given either by Eqs. 3.12
and 3.13 or by Eqs. 3.19 and 3.20, where the set chosen depends on the model
considered.
The axial symmetry of the mass transfer problem is expressed by the following
boundary conditions:

@cA
¼ 0; a\r b ð5:18Þ
@h h¼p
Fig. 5.1 Mass transport

taking place in a sphere-in-
cell. Bold arrows indicate ∂c
=0
flow direction ∂θ
sorption
mechanism
∂c
=0
∂r
∂c
=0
c=1 ∂θ

@cA
¼ 0; a\r b ð5:19Þ
@h h¼0
To ensure the continuity of concentration for any Peclet number, Coutelieris et al.
[9] proposed the following boundary condition at the outer boundary of the cell:
cA ðr ¼ b; h ¼ pÞ ¼ 1 ð5:20Þ

@cA
¼ 0; 0 h\p ð5:21Þ
@r r¼b
For high Peclet values, the boundary condition (5.20) and (5.21) is completely
equivalent to the well-known Levich approach for unbounded fluids, given else-
where as [8, 17]:
cA ¼ 1; for r ! 1 or r ¼ b ð5:22Þ
Figure 5.1 shows the extent of the concentration boundary layer surrounding
the solid surface.
There are several approaches for describing the sorption mechanism (see Sect. 5.2.2).
By using the geometrical specification of the sphere-in-cell model, sorption in specific
spherical geometries can be written as follows:

@cA
½ð5:16Þ neutral surface ) ¼ 0; 0 h\p ð5:23Þ
@r r¼a
½ð5:8Þ instantaneous adsorption ) cA ðr ¼ a; hÞ ¼ 0; 0 h p ð5:24Þ


@cA k
½ð5:9Þ Langmuir isotherm ) D ¼ cs ; 0hp ð5:25Þ
@r r¼a K

@cA
½ð5:13Þ adsorpionndash;reactionndash;desorption ) D ¼ Rn ; 0 h p
@r r¼a
ð5:26Þ
At high Pe numbers, the above described mass transport problem (considering
each possible adsorption case) is amenable to analytical treatment with sufficient
accuracy, as shown by various research groups [8, 17]. Mathematically, it is
notable here that the RHS of Eq. 5.2 is simplified enough to accept a variables’
separated function as an analytical solution. This is because the concentration
boundary layer is very thin compared to the local radius of curvature of the
particle, therefore the curvature term and the tangential diffusion terms can be
neglected. In particular, Eq. 5.2 becomes:
@cA uh @cA @ 2 cA
ur þ ¼D 2 ð5:27Þ
@r r @h @r
which is parabolic on h and can be solved analytically in a manner quite similar to
that of Levich [17]. By using dimesionless quantities, the concentration profile in
the fluid phase is of the form:
Zz
4 3
cA ðzÞ ¼ c2 e9t dt þ c3 ð5:28Þ
0
where:
rffiffiffiffiffiffiffiffiffi
3 3 Pe sin h
z ¼ r ð5:29Þ
4a3 h 0:5 sin 2h
and c2 and c3 are coefficients that can impose the appropriate boundary conditions
for mass transport.
For the case of low and moderate Peclet numbers, all the terms of the transport
equation (5.17) survive, therefore it is necessary to apply numerical approaches to
obtain the solution. What is important for cases of low Peclet number is the
boundary condition on the outer boundary, as described by Eqs. 5.20 and 5.21, is
equivalent to that of Levich for high Peclet numbers, while it allows the boundary
layer to violate the concept of a very thin boundary layer imposed by the Levich
approach [17]. The difference between high and low Peclet mass transport in
sphere-in-cell geometry, is graphically summarized in Fig. 5.2.
Obviously, mass transport problems assuming neutral solid grains (no adsorption)
result in a uniform concentration profile without scientific or industrial interest. For
instantaneous adsorption and using Kuwabara’s approach for the velocity field, the
concentration profiles for a moderate Peclet value (Pe = 70) are presented in Fig. 5.3.
Fig. 5.2 Schematic

visualization of boundary low Peclet high Peclet
layers for low and high Peclet Diffusive regime Convective regime
wide boundary layer thin boundary layer
numbers
For Langmuir-type adsorption with the same geometry and flow conditions, the
concentration profiles are shown in Fig. 5.4.
It should be stressed however, that the boundary condition (5.26) in combi-
nation with (5.14)–(5.15), imply non-linearity in the whole approach and, there-
fore, an iterative numerical technique is necessary to obtain the solution. Thus, the
integration of the transport equation must be numerical, although the flow field can
be obtained analytically through Happel or Kuwabara type boundary conditions.
Figure 5.5 visualizes the concentration profiles for the realistic sorption mecha-
nism with the same geometry and flow conditions as those for the other adsorption
types, presented above.
The above Figs. 5.3, 5.4 and 5.5 show a gradual decrease of the concentration
for constant r as the angular position approaches the outlet because of the shape
and extent of the concentration boundary layer. Therefore, the possibility for the
solute to be captured by the solid surface decreases with angular position because
of the convective regime. In general, higher concentration gradients at any radial
and angular positions are found in cases of instantaneous adsorption compared to
those for other sorption mechanisms, as the surface concentration, cA(a,h), is much
higher in the cases of Langmuir type or realistic adsorption mechanisms, taking its
highest value at the impact point and decreasing monotonically as h tends to p.
For all sorption mechanisms considered, the overall Sherwood number, Sho, is
calculated as follows [6]:

@cA
Z0
1 @r r¼a
Sho ¼ dh ð5:30Þ
4pa2 cA ðb; hÞ cA ða; hÞ
p
Fig. 5.3 Spatial 1.0

concentration profiles for θ=0
e = 0.9 at three different 0.8 θ = π/2
angular positions of the
θ=π
sphere-in-cell for
0.6
instantaneous adsorption
cA
0.4
0.2
0.0
0 0.2 0.4 0.6 0.8 1
α β α

θ=0
concentration profiles for
e = 0.9 at three different θ = π/2
0.8
angular positions of the θ=π
sphere-in-cell forLangmuir
0.6
k
type adsorption ¼1
cA
K
0.4
0.2
0.0
0 0.2 0.4 0.6 0.8 1
α β α

concentration profiles for θ=0
e = 0.9 at three different 0.8 θ = π/2
angular positions of the
sphere-in-cell for realistic θ=π
0.6
sorption mechanism
cA
0.4
0.2
0.0
0 0.2 0.4 0.6 0.8 1
α β α
Finally, the adsorption efficiency of a sphere-in-cell model can be written as:

RR
½NAr r¼a dSsphere
Ssphere
k0 ¼ ð5:31Þ
u1 cA;1 4pb2
where it can be easily shown for the specified spherical geometry that:
0 1
cA
Zp @
a2 B cA;1 C
ko ¼ sin h B C
@ r A dh ð5:32Þ
4pb2 @
0 a r¼a
To initially investigate the macroscopic behavior of the sphere-in-cell model,

Fig. 5.6 presents the overall Sherwood number as a function of the Peclet number
for a relatively high porosity value (e = 0.9) in the case of instantaneous
adsorption. An almost monotonic increment of Sho is observed as the Peclet values
increase. This is due to the higher concentration gradients close to the adsorbing
surfaces in highly convective regimes. It has been also shown that such analytical
results are in very good agreement with experimental measurements and numerical
simulations [11]. This agreement can be attributed to the high porosity value used,
taking into account that the analytical approach has been shown to be highly
accurate in that range [5, 9, 26]. High porosity values favor the accuracy of
transport results because mass transport under high Peclet values develops very
thin boundary layers, a consideration which corresponds to an outer sphere of
radius significantly higher than the inner one, i.e. porosity takes quite high values.
To further clarify the situation, the dependence of the overall Sherwood number on
porosity for the case of instantaneous adsorption is shown in Fig. 5.7. Evidently,
increasing porosities do not favor mass exchange efficiency and the Sho drops.
High porosity values correspond to low available surface for mass transfer leading
to low adsorption rates for the porous material. Again experimental data and
numerical predictions are generally in agreement with the analytical results [11]
although an erratic behavior is observed. This behavior is due to the fact that high
porosities, where analytical models are in principle applicable, correspond to a
very low number of spheres for the numerical representation of the porous med-
ium. It should be noted that at porosities lower than 0.8, the available analytical
approaches become gradually less accurate and therefore less dependable.
To include the more detailed and realistic sorption mechanism of Eqs. 5.26,
5.14 and 5.15 into the whole approach, it is necessary to numerically solve the
boundary value problem described above either using the Kuwabara or Happel
expression for the velocity components. Frequently, a non-uniform finite-
difference scheme is employed where the value of k0 can be calculated once the
radial component of the concentration gradient upon the surface is known by using
a numerical integration method (the most suitable choice could be the modified
Newton–Cotes with an adjustable step). The effect of the sorption mechanism
Fig. 5.6 Dependence of the 6

overall Sherwood number on
the Peclet for a high porosity
value (instantaneous 5
Sh 0
adsorption)
4
3
50 100 150 200 250 300 350 400
Pe
Fig. 5.7 Influence of the 6

porosity on the overall
Sherwood number Pe = 70
(instantaneous adsorption) 5
Sh0
3
0.7 0.8 0.9 1
ε
-3
k s = 8 x 10 (n=1) sec-1
Fig. 5.8 Adsorption 1.0 k s = 8 x 10 -5 (n=2) kg m-2 sec-
k d = 8 x 10-3 sec-1
efficiency of the sphere- k a = 1 x 10-30 m3 sec-1
0.8
in-cell model as a func-
tion of Pe for several 0.6
instantaneous
adsorption mechanisms, Langmuir
λ
compared to experi- 0.4 ads/reac/des, n=1

mental data presented ads/reac/des, n=2
by discrete points [29] 0.2 ε=0.7243
κ/Κ=1 experimental
0.0
0 20 40 60 80
Pe
considered, is clearly depicted in Fig. 5.8, which presents the adsorption efficiency
as a function of Pe for various adsorption mechanisms.
In general, higher efficiency is found for instantaneous adsorption than for the
Langmuir type, as the concentration gradients are lower in the latter case. Indeed,
in Langmuir adsorption concentration on the solid surface attains non zero values
thus rendering the overall driving force cA ðb; hÞ cA ða; hÞ smaller. A general
decrement of the efficiency with Peclet is observed, a trend expected as the more
convective flows (increasing Peclet) tend to prevent solute from being captured by
the solid surface. The discrete points represent the experimental data of Wilson
and Geankoplis [29] for the rather high porosity of 0.7 (the highest porosity used in
their experiments). As these authors measured the overall Sherwood number, it is

necessary to transform it into adsorption efficiency to allow direct comparison with
the predictions. Starting from the definition of adsorption efficiency in Eq. 5.31 the
numerator (r-component of the molar flux on the adsorbing surface) can be
expressed by the use of the mass transfer coefficient as k0 DC Ssphere : After some
algebraic manipulations, this leads to the following linear relation between the
overall Sherwood number and k0:
Pe a cA:1
Sh ¼ k0 ð5:33Þ
ð1 eÞ L hcA i
where hcA i is the spatially averaged concentration of solute, and L is the charac-
teristic length, used when defining the Peclet number (see Eq. 5.3) As the
experimental adsorption mechanism is not clearly known, the agreement between
predictions and experimental data is considered sufficient.
5.2.4 Mass Transport Through Cylinders
The 2-D sphere-in-cell models are completely equivalent to 2-D cylinder-in-cell

models. For this reason, the solutions of the previous Sect. 5.2.3 are valid for both
cases.
5.2.5 Mass Transport Through Spheroids
Spheroidal geometry was more difficultly approached until the 1990s than
spherical one because the flow field had to be estimated numerically, even for
creeping flow conditions. This is because the governing Eq. 3.9 in terms of stream
function is not separable in spheroidal coordinates whereas it is separable in
Cartesian, cylindrical and spherical ones. In 1994, Dassios et al. [14] overcome
this barrier by introducing the method of variable semi-separation, thus obtaining
analytical expressions for the velocity components when either Happel or
Kuwabara type boundary conditions were considered. (For a detailed description
of the spheroid-in-cell model, see Sect. 3.2.4.)
By considering this cell approach, the governing equation for steady state mass
transport in the fluid phase within the model can be written in prolate spheroidal
coordinates ðg; hÞ as:
2
@cA @cA D @ cA @cA @ 2 cA @cA
ug þ uh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi þ coth g þ þ cot h
@g @h a sinh2 g þ sin2 h @g2 @g @h2 @h
ð5:34Þ
where ug and uh are the g and hvelocity components, depend on the model
considered (Happel or Kuwabara type boundary conditions).
The axial symmetry of the mass transfer problem is assured by the boundary
conditions (5.18) and (5.19), where the continuity of concentration for any Peclet
number is described by the boundary condition at the outer boundary of the cell [10]:

cA g ¼ g b ; h ¼ p ¼ 1 ð5:35Þ

@cA
¼ 0; 0 h\p ð5:36Þ
@g g¼gb
Again several approaches for describing the sorption mechanism can be

adopted (see Sect. 5.2.2), as follows:

@cA
½ð5:16Þ neutral surface ) ¼ 0; 0 h\p ð5:37Þ
@g g¼ga
½ð5:8Þ instantaneous adsorption ) cA ðg ¼ ga ; hÞ ¼ 0; 0hp ð5:38Þ

D @cA k
½ð5:9Þ Langmuir isotherm ) pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ cs ; 0 h p
a sinh ga þ sin h @g
2 2
g¼ga K
ð5:39Þ
#
D @cA
½ð5:13Þ adsorpionndash;reactionndash;desorption ) pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a sinh2 ga þ sin2 h @r r¼a
¼ Rn ; 0 h p
ð5:40Þ
As shown above, the consideration of high Pe numbers corresponds to the
analytical solution to the mass transport problem [8, 10]. The assumption of a very
thin boundary layer leads to a transformed formation of Eq. 5.34 as follows:
@cA @cA D @ 2 cA
ug þ uh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2
ð5:41Þ
@r @h a sinh2 ga þ sin2 h @g
which can be solved analytically in a manner similar to that of Levich [17],
providing concentration profiles in the fluid phase of the form given by (5.28)
while the transformed variable z is given as:

13
aK Pe sin h
z ¼ g
13 ð5:42Þ
4K sinh2 ga sin 2h h

4 2
where K and K are given by Coutelieris et al. [8]. Whenever Peclet attains low or
moderate values, a numerical approach must be involved, as discussed previously.
Fig. 5.9 Concentration (a)

profiles in the vicinity of
1.0
adsorbing prolate (a) and
oblate (b) spheroids-in-cell at
0.8
three different angular
positions for instantaneous
0.6 a 3=2
adsorption and high Pe
cA
θ= 0 Y=0.1
0.4 Pe=10000
θ=π/2
0.2 θ=π
0.0
0.00 0.05 0.10 0.15
(η-ηα)/(ηβ-ηα)
(b)
1.0
0.8
0.6 θ= 0
cA
0.4 θ=π/ 2 a 3=0.5

Y= 0.1
θ=π Pe=10000
0.2
0.0
0.00 0.10 0.20 0.30
By employing the transformation for the oblate spheroidal coordinates, given

by (3.38) and (3.39), the concentration profile is again described by (5.28), where:
" #13

aK Pe sin h
z ¼ g
13 ð5:43Þ

4K sinh2 ga þ1 sin 2h h

4 2
K
and K; are given by Coutelieris et al. [8].
Again the assumption of neutral solid grains clearly results in a uniform con-
centration profile that lacks scientific interest. For instantaneous adsorption and
using Kuwabara’s approach for the velocity field, the concentration profiles for a
high Peclet value (Pe = 10,000) are presented in Fig. 5.9.
The assumption of instantaneous adsorption for low Peclet values corresponds
to the concentration profiles in Fig. 5.10
For Langmuir type adsorption with the same geometry and flow conditions, the
concentration profiles are shown in Fig. 5.11.
Figure 5.12 depicts the concentration profiles for the realistic sorption mech-
anism with the same geometry and flow conditions as above.

profiles in the vicinity of
1.0
adsorbing prolate (a) and θ= 0
oblate (b) spheroids-in-cell at θ=π
three different angular 0.8
positions for instantaneous a 3=2
adsorption and low Pe 0.6 Y= 0.1
Pe=10
cA
0.4
0.2
0.0
0.00 0.25 0.50 0.75 1.00
(b)
1.0
θ= 0
0.8 θ=π
0.6
cA
0.4
a3=0.5
Y= 0.1
0.2
Pe=10
0.0
0.00 0.25 0.50 0.75 1.00
Clearly, the results are qualitatively similar to those for spheres with the same
transport and adsorption conditions.
For all the sorption mechanisms considered, and for the case of prolate sphe-
roidal coordinates, the overall Sherwood number, Sho, is calculated as follows [9]:
Z0
1 @cA
Sho ¼ sin h dh ð5:44Þ
a3 a @g
1 þ sin1 p
g¼ga
a a3
and, for oblate spheroidal coordinates [9]:

Z0
1 @cA
Sho ¼ sin h dh ð5:45Þ
a23 1 þ a
@g g¼
ga
1þ ln p
a 1
2 a
Fig. 5.11 Spatial (a)

concentration profiles at three 1.0
θ= 0
different angular positions of
θ=π/ 2
the prolate (a) and oblate 0.8
θ=π
(b) spheroid-in-cell for
Langmuir
type adsorption 0.6 a 3=2
cA
k Y=0.1
¼1 0.4 Pe=15
K
0.2
0.0
0.00 0.25 0.50 0.75 1.00
(b)
1.0
θ= 0
0.8 θ=π/ 2
θ=π
0.6
cA
0.4 a 3=0.5
Y= 0.1
0.2 Pe= 15
0.0
0.00 0.25 0.50 0.75 1.00
Finally, the adsorption efficiency of prolate cells can be written as:

Z0
2 @cA
k0 ¼ sin h dh ð5:46Þ
Pe b21 @g g¼ga
p
and
Z0
2 @cA
k0 ¼ sin h dh ð5:47Þ
Pe
b21 g
@ g¼ga
p
The effect of the Peclet number on the Sherwood number for instantaneous
adsorption is presented in Fig. 5.13.
As shown above, the boundary condition on the outer boundary is crucial for
the estimation of macroscopic quantities such as the overall Sherwood number and
adsorption efficiency. The dashed lines in Fig. 5.13 correspond to a ‘‘Levich-
equivalent’’ approach, where a constant uniform concentration profile is consid-
ered on the surface of the outer spheroid. This condition is compatible with the
assumption of a thin boundary layer (i.e. high Peclet regime), unless it breaks
down when diffusion becomes significant and the boundary layer thickens. Indeed,

profiles for prolate (a) and
1.0
oblate (b) spheroidal shapes
for high and low Peclet θ= 0
0.8 Pe=1000
numbers at three different
angular positions Pe=15
0.6 a3=2
cA
θ=π / 2 Y= 0.1
0.4
0.2 θ=π
0.0
0.0 0.2 0.4 0.6 0.8 1.0
(b)
1.0
0.8 Pe=1000
θ= 0 Pe=15
0.6
cA
θ=π/ 2
0.4 a 3 =0.5
Y= 0.1
0.2 θ=π
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Fig. 5.13 Dependence of the 10.0

overall Sherwood number on
Axis Ratio = 2.0
the Peclet for prolate and
oblate spheroids-in-cell Axis Ratio = 0.5
(instantaneous adsorption)
Sho
1.0
Y= 0.1
0.1
1 10 100
Pe
for Pe 20 the Levich type approach predicts the correct Sho value with sufficient
accuracy while, for lower Pe values, it yields Sho values weakly dependent on Pe,
and converges to a limiting value as Pe ! 0: This limiting value is a function of
porosity and the axis ratio, given by [9]:
1 1
a2 a a þ b3
Sh0 ¼ 2a 1 þ 3 sin1 ln ð5:50Þ
a a3 b1 ða þ a3 Þ
for prolate spheroids-in-cell, and by:

1 b3 1
a2 1 þ
a a3
Sh0 ¼ 2
a 1 þ 3 ln tan1 tan1 ð5:51Þ
a 1
2 a a a
for oblate spheroids-in-cell. The above equations result from solving the Laplace
equation, which is valid for the pure diffusive regime, i.e. as Pe ! 0 [5, 9].
However, conditions on the outer boundary (5.35) and (5.36) are obviously more
consistent, resulting in more realistic Sho . For the realistic adsorption mechanism,
typical analogous results are shown in Fig. 5.14.
In Fig. 5.14, a monotonic decrease of Sho , with increasing aspect ratio, is
observed for all cases because prolate spheroids present higher diffusional resis-
tance than oblate ones [6, 8]. Oblate spheroids present larger impact surface than
prolate ones, and thus their capacity for adsorption is higher. This advantage of
oblate shapes becomes very weak in the case of small Pe values because almost all
the parts of the adsorbing solid surface become active as diffusion becomes
dominant. For small Peclet values, unlike large ones, the overall Sherwood number
depends appreciably on the order of the heterogeneous reaction because the
concentration gradients become less significant. In this case, a decrease of Sho is
observed when the reaction is of second-order (rather than first-order) as a higher
reaction order causes lower concentration gradients and larger amounts of
adsorbed mass. The overall Sherwood number can be 15–50% higher, depending
on Pe, the shape, porosity, and order of the reaction, compared to the values
obtained assuming instantaneous adsorption for either high or low Peclet values
[8, 9]. This occurs because the concentration
on the solid surface attains nonzero
values and, thus, the difference cA gb ; h cA ðga ; hÞ becomes almost nil for some
h-values. The decrease of this driving force is significantly larger than the decrease
of the concentration gradients observed when a realistic adsorption process is
assumed rather than an instantaneous one.
Figure 5.15 presents adsorption efficiency results for prolate and oblate
spheroids-in-cell, as functions of the Peclet number. The Levich approach on the
outer boundary overestimates the adsorption efficiency for low Peclet values,
leading to values larger than unity as Pe ! 0 for both geometries. On the contrary,
the modification (5.35) and (5.36) predict ko values, which remain lower than unity
even in the very low Pe range. What is observed in general is that convention is a
favorable parameter allowing the escape of diluted material from the adsorbing
solid surface.
Fig. 5.14 Dependence of the (a)

overall Sherwood number on 15.0
the axis ratio of the spheroid
for a high (a) and a low Pe=1000
(b) Peclet number (realistic 12.5 Y= 0.1
adsorption)
Sho
10.0
7.5
5.0
0.5 1.0 1.5 2.0
Axis Ratio, a 3
(b)
4.5 n=1
Pe=15
n=2 Y=0.1
3.0
Sho
1.5
0.0
0.5 1.0 1.5 2.0
Axis Ratio, a 3
1.00
Fig. 5.15 Dependence of the
adsorption efficiency on the
Peclet number for prolate and
oblate spheroids-in-cell
(instantaneous adsorption)
0.10
λo
Axis Ratio = 2.0
Axis Ratio = 0.5
Y= 0.1
0.01
1 10 100
Pe
The dependence of the overall adsorption efficiency on the Peclet number for
both prolate (a) and oblate (b) geometry is given in Fig. 5.16, where the hetero-
geneous reaction is assumed to be either of first (n = 1) or second (n = 2) order.
A significant decrease of the adsorption efficiency occurs as Pe increases. Values
of overall adsorption efficiency obtained using the instantaneous adsorption model
are 10–35% higher than those obtained here, because of the higher concentration
gradients that instantaneous adsorption generates.
Fig. 5.16 Influence of the (a)

Peclet number on the
adsorption efficiency for 0.12 n=1
prolate (a) and oblate n=2
(b) spheroids-in-cell (realistic
0.08 a 3=2
adsorption mechanism)
Y= 0.1
λo
0.04
0.00
10 100 1000 10000
Pe
(b)
0.12
n=1
n=2
0.08 a3=0.5
λo
Y= 0.1
0.04
0.00
10 100 1000 10000
Pe
5.2.6 Single Phase Mass Transport in Other-Type Unit Cells
Further to the models described above, several unit cells have also been proposed in
recent decades. These assemblies are all characterized by: (a) complex geometry
which enforces numerical solutions of flow and transport problems, (b) the iteration
of their structure to occupy the volume of interest (periodic boundary conditions),
and (c) a high degree of freedom in the geometrical shapes considered to represent the
real porous material. Increased computational power nowadays allows for the con-
sideration of huge complex structures, therefore current research trends digress from
the unit cell formulation, as analytical solutions no longer essential.
5.3 Single Phase Flow in Granular Structures
As described in Sect. 3.3, transport processes in granular media are of high

scientific and technological interest, and mathematical modeling plays an impor-
tant role in the in-depth study and understanding of these processes. In the majority
of models on transport in granular media, the particles of the assemblage are
assumed to adsorb mass instantaneously, which is obviously a fairly rare physi-
cochemical phenomenon that pertains to a very limited range of applications. Only
recently have several detailed models of the sorption mechanism been presented
for porous media processes [10, 13].
The aim of this Section is to present an adequate simulation for the adsorption
heterogeneous reaction–desorption mechanism, which accurately describes the
sorption upon a solid surface of a solute diluted in a flowing fluid [2, 20]. More
specifically, it can be considered that the solute diluted in the bulk phase is initially
adsorbed by the solid surface where a heterogeneous reaction takes place and its
products, assumed to be inactive and of very low concentrations, are again des-
orbed in the bulk phase. The adsorption is assumed to occur due to vacant sites that
are normally distributed over the surface area, while the whole process is deter-
mined by an overall rate according to the flow regime and thermodynamics [25].
As described in Sect. 3.3, the procedure of random deposition under specific
restrictions must be followed in order to represent the stochastically constructed
three-dimensional assemblages of spheres which are considered a typical dem-
onstration of a granular medium (see Fig. 3.7). The flow-field within such an
assemblage is typically obtained by the numerical solution of the Stokes equations
coupled with the continuity equation, as described in Sect. 3.3.
Assuming no reactions occur in the bulk phase, the time-dependent mass
transport of a passive solute (namely, component A) in the stochastically con-
structed medium is described by the convection–diffusion equation:
@cA
þr ðucA Þ ¼ Dr2 cA ð5:52Þ
@t
To investigate the importance of the adsorption mechanism in such complex
structures, all the previously discussed different adsorption mechanisms are adopted
here. Neutral solid surfaces again lack scientific and engineering interest. The other
sorption mechanisms considered are instantaneous adsorption (5.8), adsorption
under the Langmuir isotherm (5.9)–(5.11), and the realistic sorption mechanism
(5.13)–(5.15). Typical conditions are also considered at the other boundaries.
The validity of the solution scheme for the mass transport problem can be
checked in the limiting case of a periodic array of spheres with radius R, which is
x ¼ 0 by a concentration
fed at pulse. The concentration profile with time at the
51
outlet x ¼ L ¼ R is then expressed as [23]:
8
" #
1 ðL huitÞ2
cðtÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffi exp ð5:53Þ
4pD t 4D t
where hui is the average fluid velocity, and D the dispersion coefficient. There-
L2
fore, the dimensionless time (defined by using as characteristic time) at which
D
the maximum outlet concentration is recorded can be calculated as:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 þ Pe2 1
tmax ¼ ð5:54Þ
Pe2
Fig. 5.17 Outlet

concentration profile for
Pe = 20 in a periodic array of
spheres
huiL
where Pe ¼ is the Peclet number.
D
The dispersion coefficient can be determined based on the approach of Salles
et al. [24] and in these simulations is found to be D ¼ 1:34D; thus giving tmax ¼
0:047 for Pe ¼ 20, when the numerically calculated value is 0.048 (see Fig. 5.17).
Note that the dimensionless time step dt must be small enough (\10-5) to attain
acceptable accuracy in the calculations.
The spatial distribution of the concentration, obtained numerically for Pe = 20
and e = 0.7243 along the flow direction, is presented in Fig. 5.18 where a ran-
domly selected two-dimensional cut of the stochastically constructed medium is
visualized for both instantaneous and Langmuir-type adsorption. The concentra-
tion is higher on the left side of the images (inlet) and gradually decreases across
the medium in a manner significantly steeper for the instantaneous than the
Langmuir-type adsorption. The adsorption efficiency for several porosities
(e = 0.9883, e = 0.8136 and e = 0.7243) and Peclet numbers was calculated by
the full-numerical scheme in the sphere assembly for both instantaneous and
Langmuir-type adsorption and presented in Fig. 5.19. The effects of porosity and
Peclet on the adsorption efficiency are the same as those above (higher efficiencies
correspond to lower porosity, and increasing Peclet numbers lead to lower k0
values). Instantaneous adsorption leads to higher adsorption efficiency values than
those calculated under other adsorption mechanisms with the same Peclet number
and porosity.
The effect of the k/K ratio used in the boundary condition for Langmuir-type
adsorption, on the computed adsorption efficiency is presented in Fig. 5.20 for
Pe = 20 and e = 0.8136. As this ratio increases, a significant increase of k0 is
observed because higher k/K corresponds to lower ratios of the covered surface,
Heq. This, in turn, corresponds to lower surface concentrations and higher con-
centration gradients, i.e. to higher k0. On the other hand, an increased k/K ratio can
be viewed as a higher adsorption rate for a given K value and therefore, less mass
of substance A can escape from the solid surfaces resulting in higher k0 values.
Fig. 5.18 Spatial

(a)
distribution of the
concentration of substance A
within a two-dimensional cut
of the three-dimensional
sphere assemblage for
a instantaneous and
b Langmuir-type adsorption.
(The flow inlet is on the left
side. Darker areas
correspond to higher
concentrations.)
(b)
It is interesting to consider the relative agreement between the results produced

by considering the sphere-in-cell approximation and those obtained from the
numerical solution of the three-dimensional sphere assemblages. Figure 5.21a
compares the respective adsorption efficiencies and instantaneous adsorption for
porosity approaching unity (e = 0.9883). The agreement is perfect as the semi-
analytical sphere-in-cell model can adequately predict reality for such high
porosity values. When lower porosities are considered, the fundamental assump-
tions of the sphere-in-cell approximation are less satisfactory. Indeed, in the low
Pe regime characterized by the gradual dominance of the diffusive over the con-
vective terms, the semi-analytical approach of the sphere-in-cell model cannot
adequately describe the mass transport process, as the diffusion layer is very thick
Fig. 5.19 Adsorption (a)

efficiency of the three- 1.0
dimensional sphere
assemblage as a function of 0.8
Pe for various porosities in
0.6 ε = 0.7243
a instantaneous and
λο
b Langmuir-type adsorption 0.4 ε = 0.8136
ε = 0.9883
0.2
0.0
0 20 40 60 80
Pe
(b)
1.0
0.8 ε 0.7243
ε = 0.8136
0.6
ε = 0.9883
λο
0.4
0.2
0.0
0 20 40 60 80
Pe
Fig. 5.20 The effect of k/K 1.0

ratio on adsorption efficiency
0.8
0.6
λo
ε = 0.8136
0.4 Pe = 20
0.2
0.0
0 2 4 6 8 10
k/K
(larger than the cell itself) and tends to infinity as Pe ! 0: On the other hand, as
Pe increases, flow becomes more convective and the analytical flow-field of the
sphere-in-cell is no longer a sufficient approximation of the actual flow-field in real
granular media, thus leading to model discrepancies. However, for e = 0.8136 and
e = 0.7243 the agreement between the two approaches ranges from very good to
satisfactory depending on the Pe range considered. This result is expected since
both the sphere-in-cell and sphere assemblage geometries used in the present study
were constructed to be characterized by the same porosity and internal (adsorbing)
surface area. This was accomplished by selecting the correct radius and sphere
population number in the assemblage. Of course, it is possible to construct several
sphere assemblages of the same porosity but with varying sphere radius and
Fig. 5.21 Comparison (a)

between adsorption efficiency 1.0
in sphere-in-cell and sphere sphere -in-cell, instant. ads.
assemblies for different 0.8 sphere assemblage, instant. ads.
porosities
0.6
λο
0.4
0.2
0.0
0 20 40 60 80
Pe
(b)
1.0
0.8
0.6
λο
0.4 ε = 0.8136
0.2
0.0
0 20 40 60 80
Pe
(c)
1.0
0.8
0.6
ε = 0.7243
λο
0.4
0.2
0.0
0 20 40 60 80
Pe
population number and therefore with different internal surfaces. Figure 5.22
shows how these different arrangements of spheres affect the calculated adsorption
efficiency for instantaneous and Langmuir adsorption. The horizontal axes repre-
sent the internal surface ratio which changes when different media are considered.
The case of adsorption surface ratio equal to unity corresponds to the results
discussed so far. Evidently under instantaneous adsorption conditions k0 is not
Fig. 5.22 The effect of using (a)

sphere assemblages with 1.0
varying internal surface area
Instantaneous
on adsorption efficiency and 0.8 ε = 0.7243
for two porosity values: Langmuir
0.7243 (a) and 0.8136 (b) 0.6
λo
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0
adsorption surface ratio
(b)
1.0
0.8
ε = 0.8136
0.6
λo
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0
adsorption surface ratio
influenced significantly by changes in the medium. This is again due to the high
concentration gradients prevailing that mask the effect of internal surface varia-
tion. However, in the more realistic Langmuir-type of adsorption a very pro-
nounced influence on k0 is observed implying that caution should be exercised
when using the sphere-in-cell model to determine adsorption efficiency in granular
media. This strongly indicates that matching porosity alone is not sufficient for a
reliable result.
5.4 Macroscopic Quantities for Single Phase Transport
There are many industrial and technological applications of mass transport within
porous media in a variety of scientific fields, such as environment, energy, biology,
etc. [15, 18, 27]. Mathematical modeling of transport processes in porous med,ia is a
powerful tool, especially whenever experimental observations are difficult, time
consuming and expensive. Due to the coupling between the physicochemical
mechanisms and the local geometry of the porous medium mathematical descrip-
tions of mass transport in realistic porous media are highly complex. Modeling
becomes more difficult when moving from the pore level to the field level, because
different length scales result in complicated descriptions of the problem’s physics
and therefore increased computational power is usually required.
5.4 Macroscopic Quantities for Single Phase Transport 113
From the late 1950s, special effort has been given to mathematically describe
and solve flow and mass transport problems in porous media. Initially quite
simplified geometries were considered, where analytical solutions can be obtained
for the flow field and the mass diffusion and/or adsorption process (see previous
chapters on cell models). Numerical solutions in realistic reconstructions of porous
media for the Stokes equations and related transport problems have been obtained
for several specific applications during the last decades [1, 3, 16, 19, 26]. In the
majority of these works, the particles were assumed to adsorb mass instanta-
neously however, this is a rare physicochemical phenomenon that pertains to a
very limited range of applications. Recently, further detailed models of the sorp-
tion mechanism have been presented for porous media processes.
This Section aims to simulate an adsorption—heterogeneous reaction—
desorption mechanism for a complex granular porous geometry that can accurately
describe the sorption upon a solid surface of a solute diluted in the flowing fluid
[2, 20]. Details about this sorption mechanism can be found in Sect. 5.2.
The major issue of typical macroscopic modeling for such, or simpler, cases can
be identified at the a priori definition of the macroscopic quantities necessary to solve
these equations, although they are normally derived from the solution of these
equations. So far, mainly empirical or semi-empirical correlations for these
parameters have been proposed based on experimental measurements of specific
systems [5, 22]. The generalized treatment of such a problem corresponds to theo-
retical estimations of these quantities where the volume averaging concept is a
frequently employed tool for large-scale modeling of processes taking place in
porous media, thus eliminating the influence of porous geometry on the transport
results [28, 30, 31]. Starting with transport equations at the micro-scale (pore) level,
the spatial averaging theorem is applied with the correct assumptions, leading to the
estimation of macroscopic quantities such as mass transfer coefficient and dispersion
tensor [21]. To further simplify the modeling and eliminate the simulation effort, the
majority of the above-mentioned models have been applied to simplistic domains,
such as unit cells, since the focus was on the interfacial mass exchange rather than the
representation of the medium in a realistic manner.
5.4.1 Stochastically Constructed 3-D Sphere Assemblage
Representation of the biphasic domains under consideration is achieved by the

random deposition of spheres of a given radius in a box of specified length. (For
specific details on the efficient algorithm (ballistic deposition) used for the
domain’s stochastic construction, see Sect. 3.3).
5.4.2 The Flow Field (Single Phase Flow)
The velocity field is computed numerically by solving the Stokes equations as

described previously (see Sect. 3.3).
5.4.3 Mathematical Formulation
The area of interest is a porous domain consisting of a flowing aqueous phase

(L-phase) and a solid phase (S-phase). A tracer is advected by the flowing liquid phase,
being sorpted in the liquid–solid interface. The governing processes are diffusion and
advection in the liquid phase, and the liquid–solid interface is characterized by the
realistic adsorption/reaction/desorption mechanism of the tracer (see below).
Assuming that the bulk phase is chemically neutral, the pore-level transport of
the tracer in the b-phase is described by the convection–diffusion equation:
@CL
þ r ðvCL Þ ¼ DL r2 CL ð5:55Þ
@t
where CL is concentration, t is time, v is the fluid velocity, and DL is the diffusivity
in the liquid phase.
To ensure the continuity of the mass fluxes on the solid–liquid interfaces, the
following boundary condition (similar to Eq. 5.13) is applied:
nLS rCL ¼ Rn ; at the solidliquid interface ALS ð5:56Þ
where the overall sorption rate Rn is dependent on the type of sorption process
considered. In accordance to Sect. 5.2, the rate Rn is given as:
n
Rn ¼ ks CLS ð5:57Þ
where the concentration of the tracer upon the solid surface, CLS ; is described by
the relation:
n
ks CLS þ ½kd þ ka CL N CLS ka CL nm ¼ 0 ð5:58Þ
5.4.4 The Volume-Averaging Procedure
Following the volume-averaging procedure [22], local concentrations and veloc-

ities are next decomposed into interstitial averages and fluctuations
CL ¼ hCL iL þCL0 ð5:59Þ
v ¼ hviL þv0 ð5:60Þ
which are subsequently substituted in the governing differential equations.

Invoking separation of scales to discard small terms, linearizing and following
Quintard and Whitaker [22], assuming an isotropic medium of uniform porosity
and constant volume fractions, the following representation is obtained:
CL0 ¼ b rhCL iL sL hCL iL ð5:61Þ

where b and sL are closure variables satisfying specific boundary value problems.
It can be readily shown that variable s of the b-phase satisfies the boundary
value problem:
v rsL ¼ Dr2 sL e1 a ð5:62Þ
with
nLS rsL ¼ 0; at the solidliquid interface ALS ð5:63Þ
and
hsL i ¼ 0 ð5:64Þ
The mass-transfer coefficient, a, on Eq. 5.62, is given by:
Z
D
a¼ nLS rsL dA ð5:65Þ
V
ALS
The above can be simplified by introducing the transformation sL ¼ 1 þ awLr

and in dimensionless form:
sL ¼ 1 þ a fL ð5:66Þ
where the dimensionless mass-transfer coefficient is defined as:
al2L
a ¼ ð5:67Þ
D
and lL denotes the characteristic length of the liquid phase. In dimensionless
notation, the boundary value problems then read as follows:
Pe u rfL ¼ r2 fL e1
L in the liquid phase ð5:68Þ
nLS rfL ¼ rn at the solidliquid interface ALS ð5:69Þ

L
where Pe ¼ huDi lL is the Peclet number defined in the liquid phase, u is the
dimensionless velocity vector, fL is the scalar variable used for the decomposition,
and rn is the dimensionless sorption rate.
Following the above, the dimensionless mass transfer coefficient simply
becomes:
e
a ¼ ð5:70Þ
hfL i
Brackets denote averages over the total volume V or the volume of the aqueous
phase, VL , where the superficial volume average is defined as:
Z
1
hyL i ¼ yL dV ð5:71Þ
V
VL
Fig. 5.23 A selected two-

dimensional cut of the
simulated geometry
discretized by an unstructured
grid
and the interstitial volume average as:

Z
1
hyL iL ¼ yL dV ð5:72Þ
VL
VL
5.4.5 Simulations
To adequately simulate the above-described problem, an algorithmic procedure

has been developed as follows:
• Solve the flow problem at the pore level and calculate interstitial and superficial
velocity fields
• Formulate the mass transport problem at the pore level
• Decompose the local velocity and concentration in terms of an interstitial
average and a fluctuation
• Describe the concentration fluctuations in terms of linear combinations of
interstitial averaged concentration and its gradient
• Solve the closure problem
• Integrate the resulting quantities to calculate macroscopic coefficients.
The numerical scheme used for all the simulations is described in Sect. 3.3 and
a typical grid representation is depicted in Fig. 5.23 (for the same two-dimensional
cut as the previous Fig. 5.18).
Table 5.1 Independence on Different random assemblages Pe n a*

random deposition
ðe ¼ 0:43Þ: Case 1 1 1 20.93
Case 2 1 1 20.83
Case 3 1 1 21.01
Case 4 1 1 20.77
Case 5 1 1 20.96
Case 1 100 1 46.55
Case 2 100 1 45.87
Case 3 100 1 46.11
Case 4 100 1 47.02
Case 5 100 1 46.39
Case 1 1 2 19.36
Case 2 1 2 19.91
Case 3 1 2 19.07
Case 4 1 2 18.66
Case 5 1 2 19.12
Case 1 100 2 45.88
Case 2 100 2 45.01
Case 3 100 2 46.99
Case 4 100 2 45.85
Case 5 100 2 46.02
The results were initially validated against the randomness of the structure, as
previously discussed. More precisely, the mass transfer coefficient a was calcu-
lated for several different depositions of spheres while the porosity value was kept
constant (=0.43). It was found that the random deposition does not significantly
affect mass transfer to the solid phase, as shown in Table 5.1, where each ‘‘case’’
corresponds to a different deposition of a variable number of spheres of different
radii, and porosity is kept constant in all cases. (Five different cases were chosen
for the same porosity to verify whether the results were independent of the medium
construction). Solution dependence on the grid was also examined in terms of the
mass transfer coefficient. It was found that the discretization used is more than
sufficient for adequate calculations. It should be noted that the parametric analysis
of the grid influence is limited to only reaction order, as this parameter introduces
non-linearity, thus the solution it is depended on this reaction order.
The relative agreement between the results produced by considering the approx-
imation presented in Sect. 5.3 and those obtained using the above technique is pre-
sented in Fig. 5.24. By assuming a typical value of nm (1 active site per Å2), the values
of the sorption constants were (before the non-dimensionalization of the problems)
ka = 1 9 10-30 m3sec-1, kd = 8 9 10-3 sec-1 and ks = 8 9 10-3 9 100(1-n)
(kg m-2)1-nsec-1. These values can be considered as typical [2] and are used in the
Fig. 5.24 Comparison 75

between the results of the
mass transfer coefficient, α*

current model and those pore scale
current model (n=1)
found in the literature
current model (n=2)
50
25
0
1 10 100
Peclet
Fig. 5.25 The influence of

porosity on mass transport
simulations presented here unless otherwise stated. Figure 5.24 compares the
respective mass transfer coefficient for the standard porosity e = 0.43 while the
realistic adsorption/reaction/desorption mechanism includes the heterogeneous
reaction of first (n = 1) and second (n = 2) order. Regarding the pore-level simu-
lations, a discrepancy from the results of the current model is always observed, thus
indicating the underestimation of macroscopic mass transport quantities when cal-
culated using pore-level approaches in small-scale domains [7].
In terms of physical interpretation, Fig. 5.24 depicts the effect of convection on
mass transport. It can be seen that the stronger the convection, the more efficient
the tracer transport from the fluid to the solid phase, at least for low and inter-
mediate porosity values, which correspond to relatively high amounts of active
solid absorbers in the medium and absorb the tracer. Finally, the order of the
reaction does not seem to significantly affect the results, or the agreement between
the two approaches. Only the first order reaction is considered hereafter.
The influence of the medium’s volume porosity on mass transport is presented
in Fig. 5.25. It is clear that porosity is an unfavorable parameter for adsorption,
mass transfer coefficient,, α*

α
1,0E+02
1,0E+00
1,0E-02
1,0E-04
1,0E-06
1,0E-08
1,0E-10
1,0E-08 1,0E-04 1,0E+00 1,0E+04 1,0E+08
Reaction Rate / Adsorption rate
Fig. 5.26 Dependence of mass transport on the ratio of the reaction to the adsorption rates
mass transfer coefficient,, α*
1,0E+02
1,0E+00
1,0E-02
1,0E-04
1,0E-06
1,0E-08
1,0E-10
1,0E-08 1,0E-04 1,0E+00 1,0E+04 1,0E+08
Desorption Rate / Adsorption rate
Fig. 5.27 Dependence of mass transport on the ratio of desorption to adsorption rates
since void space increases as porosity increases (although the active solid surface
area does not necessarily decrease), thus corresponding to high possibility for the
tracer to escape from the porous material through the void space. For high porosity
values, as Peclet values increase, the transport process becomes more and more
convective, thus mass transport becomes less effective, i.e. high amounts of the
tracer can escape from the medium. On the other hand, low porosity values cor-
respond to large amounts of solid phase in the medium, thus convection favors
mass transport since the tracer is forced to approach the absorbing surfaces. These
two competitive phenomena are shown by the cross of the curves in Fig. 5.25,
which correspond to a porosity value where both mechanisms are of equal
strength. This value obviously depends on the specific flow and transport
characteristics.
Figures 5.26 and 5.27 depict the relative influence of the sorption mechanism,
i.e. the reaction, adsorption and desorption rates, on mass transport. More pre-
cisely, Fig. 5.26 presents the mass transfer coefficient as a function of the ratio of
tracer destruction rate due to the reaction, divided by the tracer destruction rate due
to adsorption. In any case, the values of the rate constants not involved in these
ratios were kept standard. It is observed that the decrement of the reaction rate
(for constant adsorption rate) corresponds to a consequent decrement of mass
transport because the tracer has been adsorbed but not destroyed at the same rate
and, therefore remains on the surface filling the vacant sites, i.e. setting barriers in
the tracer sorption process. The increased reaction rate forces adsorption to tend
asymptotically to a constant value which depends on the geometrical character-
istics of the medium. Figure 5.27 shows the influence of the ratio desorption/
adsorption rate on mass transport for a heterogeneous reaction of the first order. It
is observed that increasing the desorption rate beyond a critical value corresponds
to a decrease in mass transport. It is important to note that the value attained by the
mass transfer coefficient before the critical desorption rate is the same as the
asymptotic value of Figure 5.26, thus further underlying its independence on
the reaction characteristics.
References
1. Ahmadi, A., Aigueperse, A., Quintard, M.: Calculation of the effective properties describing
active dispersion in porous media: from simple to complex porous media. Adv. Water
Resour. 24, 423–438 (2001)
2. Atkins, P., de Paula, J.: Physical Chemistry, 7th edn. Oxford University Press, Oxford (2002)
3. Bekri, S., Thovert, J.F., Adler, P.M.: Dissolution and deposition in fractures. Eng. Geo. 48,
283–308 (1997)
4. Bird, R.B., Stewart, W.E., Lightfoot, E.N.: Transport Phenomena. Wiley, New York (1960)
5. Burganos, V.N., Coutelieris, F.A., Payatakes, A.C.: Sherwood number for mass transfer to a
swarm of adsorbing spheroidal particles at any Peclet number. AIChE J. 43, 844–848 (1997)
6. Coutelieris, F.A.: The influence of axial orientation of spheroidal particles on the adsorption
rate in a granular porous medium. Stud. Surf. Sci. Catal. 144, 745–751 (2002)
7. Coutelieris, F.A.: Modeling of flow and mass transport in granular porous media. Cent. Eur.
J. Phys. 9, 962–968 (2011)
8. Coutelieris, F.A., Burganos, V.N., Payatakes, A.C.: On mass transfer from a newtonian fluid
to a swarm of adsorbing spheroidal particles for high Peclet numbers. J. Colloid Interface Sci.
161, 43–52 (1993)
9. Coutelieris, F.A., Burganos, V.N., Payatakes, A.C.: Convective diffusion and adsorption in a
swarm of spheroidal particles. AIChE J. 41, 1122–1134 (1995)
10. Coutelieris, F.A., Burganos, V.N., Payatakes, A.C.: Model of adsorption–reaction–desorption
in a swarm of spheroidal particles. AIChE J. 50, 779–785 (2004)
11. Coutelieris, F.A., Kainourgiakis, M.E., Stubos, A.K.: The effect of the porosity on the
adsorption rate in granular porous media. Stud. Surf. Sci. Catal. 144, 753–760 (2002)
12. Coutelieris, F.A., Kainourgiakis, M.E., Stubos, A.K.: Low Peclet mass transport in
Interface Sci. 264, 20–29 (2003)
13. Coutelieris, F.A., Kainourgiakis, M.E., Stubos, A.K.: Low to moderate Peclet mass transport
in assemblages of spherical particles for a realistic adsorption–reaction–desorption
mechanism. Powder Tech. 159, 173–179 (2005)
semiseparable solutions for stokes flow in spheroidal coordinates. Quart. Appl. Math. 52,
157–191 (1994)
15. Ellis, T.G., Eliosov, E., Schmit, C.G., Jahan, K., Park, K.Y.: Activated sludge and other
aerobic suspended culture processes. Water Environ. Res. 74, 385–410 (2002)
16. Kapellos, G.E., Alexiou, T.S., Payatakes, A.C.: A multiscale theoretical model for diffusive
mass transfer in cellular biological media. Math. Biosci. 210, 177–237 (2007)
17. Levich, V.G.: Physicochemical Hydrodynamics. Pentice-Hall, Englewood Cliffs (1962)
18. Meakin, P., Skjeltorp, A.T.: Application of experimental and numerical models to the physics
of multiparticle systems. Adv. Phys. 42, 1–127 (1993)
References 121
19. Mourzenko, V.V., Bekri, S., Thovert, J.F., Adler, P.M.: Deposition in fractures. Chem. Eng.
Commun. 150, 431–464 (1996)
20. Peters, M.H., Jalan, R.K., Gupta, D.: A dynamic simulation of particle deposition on
spherical collectors. Chem. Eng. Sci. 40, 723–731 (1985)
21. Quintard, M., Whitaker, S.: Transport in ordered and disordered porous media: volume
averaged equations, closure problems and comparison with experiments. Chem. Eng. Sci. 48,
2537–2564 (1993)
22. Quintard, M., Whitaker, S.: Convection, dispersion and interfacial transport of contaminants:
Homogeneous porous media. Adv. Water Resour. 17, 221–239 (1994)
23. Sahimi, M.: Flow phenomena in rocks: from continuum models to fractals, percolation,
cellular automata, and simulated annealing. Rev. Mod. Phys. 65, 1393–1534 (1993)
24. Salles, J., Thovert, J.F., Delannay, R., Prevors, L., Auriault, J.L., Adler, P.M.: Taylor
dispersion in porous media. Determination of the dispersion tensor. Phys. Fluids A - Fluid 5,
2348–2367 (1993)
25. Smith, J.M.: Chemical Engineering Kinetics. McGraw-Hill, Tokyo (1981)
26. Tardos, G.I., Gutfinger, C., Abuaf, N.: High Peclet mass transfer to a sphere in a fixed or
fluidized bed. AIChE J. 22, 1147–1150 (1976)
27. Valdes-Parada, F.J., Ochoa-Tapia, J.A., Alvarez-Ramirez, J.: Effective medium equations for
fractional fick’s law in porous media. Phys. A 373, 339–353 (2007)
29. Wilson, E.J., Geankoplis, C.J.: Liquid mass transfer at very low reynolds numbers in packed
beds. Ind. Eng. Chem. Fund 5, 9–14 (1966)
30. Zanotti, F., Carbonell, R.G.: Development of transport equations for multiphase systems I:
general development for two-phase systems. Chem. Eng. Sci. 39, 263–278 (1984)
31. Zanotti, F., Carbonell, R.G.: Development of transport equations for multiphase systems II:
application to one-dimensional axi-symmetric flows of two-phases. Chem. Eng. Sci. 39,
279–297 (1984)
Chapter 6
Experimental and Numerical
Investigation of Mass Transport
in Porous Media
6.1 Measurement of Molecular Diffusion Coefficients
Mass transfer is a topic of central importance in chemical engineering, and

molecular diffusion is the basic physical mechanism underlying mass transfer
processes, even when convection comes into play. As a result, values of the
molecular diffusion coefficient ðDm Þ are required for mass transfer calculations
and extensive tabulations of this parameter have been prepared in the past
(see [38, 41]). Equations for the prediction of Dm are also available and, in general,
they have reasonable accuracy [44]. Nevertheless, it is important to determine
values Dm experimentally, in many instances.
In the simple experiment described here, consideration is given to the process of
mass transfer from a volatile solid sphere (a moth ball), buried in a packed bed of
inert particles (sand or glass ballotini), through which air is forced to flow con-
tinuously. And also, to the similar process of dissolution of slightly soluble spheres
buried in a packed beds of inerts through which water flows. These are important
‘‘model situations’’ for the understanding of such processes as char combustion in
fluidised beds [42] and leaching of ore (or contaminant) from buried rocks
(or buried waste).
When these processes are performed with very low fluid velocities, the rates of
mass transfer are strongly determined by molecular diffusion and the experiment
may be used to provide an accurate method for the measurement of the diffusion
coefficient. It is an entirely novel method that has the added interest of easily
providing data at temperatures and pressures that differ significantly from ambient
values. This is not always easy to achieve with other methods [35].
As an introduction to the present analysis, it is useful to consider the simple
situation depicted in Fig. 6.1a, where the vapour liberated by a buried sphere
travels through the interstices of a packed bed of inerts, as a result of molecular
diffusion.

124 6 Experimental and Numerical Investigation
(a) (b)
Fig. 6.1 Sketch of iso-concentration surfaces around active sphere of radius b for a pure
diffusion, and b diffusion with convection
6.1.1 Diffusion Alone
The sphere of slightly volatile solid is assumed to be buried in a packed bed of

sand, of ‘‘infinite extent’’, the interstices of the bed being filled with a stagnant gas
(e.g. air) that is assumed to be free of solute at a large distance from the sphere (i.e.
C ! 0 as r ! 1). At r ¼ b; where the active sphere contacts the gas phase, the
molar concentration of solute in the gas is C ¼ PV =RT; where PV is the equi-
librium vapour pressure of the solute at the temperature ðTÞ of the experiment.
The concentration gradient will give rise to a process of outwards diffusion of
solute, but if diffusion is slow, the size of the sphere is taken to be constant, during one
experiment; in other words, the assumption of a quasi-steady state is legitimate.
Under that assumption, the rate of diffusion across a spherical surface of radius
r, concentric with the solid sphere, will be independent of r. From Fick’s law,
dC
n ¼ D0m 4pr2 e ð6:1Þ
dr
where D0m ¼ Dm =s is the effective diffusion coefficient (s being the tortuosity factor,
pffiffiffi
accepted to be 2 for packed beds of granular materials [48] and e is the bed voidage
(assumed constant throughout). Since n is independent of r, integration of Eq. 6.1,
between the limits ðr ¼ b; C ¼ C Þ and ðr ! 1; C ! 0Þ, gives
D0m
n ¼ 4D0m e p b ðC 0Þ ¼ 2 e p d12 ðC 0Þ ð6:2Þ
d1
where d1 ð¼ 2bÞ is the diameter of the active sphere. Identifying ðp d12 Þ as the area
of the active sphere, helps recognize the expression for the mass transfer
6.1 Measurement of Molecular Diffusion Coefficients 125
coefficient as k ¼ 2e D0m =d1 : Introduction of the Sherwood number, Sh0 ¼ kd1 =D0m ;
shows that for mass transfer by pure diffusion, around a buried sphere,
Sh0
¼2 ð6:3Þ
e
an expression similar to the well known result, Sh ¼ 2; for pure diffusion around a
sphere in an unbounded fluid [48].
6.1.2 Diffusion with Convection
Figure 6.1b represents again the buried sphere, but now exposed to a constant flow
of gas, with average interstitial velocity u0 : The solute still diffuses away from the
surface of the sphere, but the rate of mass transfer is now enhanced, since the
solute is continuously swept away by the moving fluid. The surfaces of equal
concentration are no longer spheres. The analysis of the physical situation is
complicated, but it has been worked out in detail [23]. For the conditions of
interest in the present work, the expression
" 2=3 #1=2
Sh0 4 0 d1 4 0 d1
¼ 4þ Pep þ Pep ð6:4Þ
e 5 d p d
has been shown to give accurate values of Sh0 =e; where Pe0p ¼ u0 d=D0m is the Peclet
number based on the diameter of the inert particles (d) making up the bed. It may be
easily seen that Eq. 6.4 reduces to Eq. 6.3, in the limit of low Pe0 ð¼ Pe0p d1 =dÞ:
As Pe0p is increased, it is known that Eq. 6.4 becomes inaccurate [23]. At Pe0p ffi 1;
Eq. 6.4 is found to be accurate to within 5%, and therefore this value of Pe0p should
not be exceeded. This is because convective dispersion (rather than molecular
diffusion) would then become the relevant mechanism of mass transfer in the
packed bed. However, in the case of the experiments described below, conditions
are strictly restricted to the range of applicability of Eq. 6.4.
The essence of the experimental method proposed relies on burying a weighed
sphere of naphthalene (with initial mass m0 and diameter d1 ) in a packed bed of
sand of known porosity (e) and continuously forcing a metered stream of air
(volumetric flowrate, m) through the packed bed (cross sectional area, A) for a
given time interval, Dt. Weighing the sphere at the end of the time interval
ðmass mt Þ gives the rate of sublimation as ðm0 mt Þ=Dt and the mass transfer
coefficient may be calculated from
m 0 mt
k¼ ð6:5Þ
Dt M pd12 ðpV =RT Þ
where M is the molecular weight of naphthalene.

With the interstitial velocity given by u0 ¼ m=ðAeÞ; the only unknown in Eq. 6.4
is D0m : A simple way of solving this equation relies on re-arranging it, through
pffiffiffi
multiplication by D0m ð¼ Dm = 2Þ; to obtain
" 2=3 #1=2
kd1 4 u 0 d 1 4 u 0 d 1
¼ D0m 4 þ þ ð6:6Þ
e 5 D0m p D0m
Several models can be found in the literature for the prediction of diffusion
coefficients in binary systems [48]. Fuller et al. [21] suggest the use of the fol-
lowing equation

MA þ MB 1=2 1:75
T
M M
Dm ¼ h A B i 2 10
7
ð6:7Þ
1=3 1=3
P ðRvÞA þðRvÞB
where P is the absolute pressure (in bar), MA and MB are, respectively, the
molecular weights of components A and B (in g/mol), and ðRvÞA and ðRvÞB are
the values of the atomic diffusion volumes of components A and B, which are
tabulated in Reid et al. [44].
Equation 6.7 predicts a proportionality between Dm and T 1:75 =P; and this is an
important result that may be demonstrated using the experimental technique
described in the following section.
6.1.3 Experiments
An example of the experimental setup is sketched in Fig. 6.2. The air supply may
determine the maximum pressure at which the experiment can be performed. The
tubing and valves connecting the air supply to the low pressure end of the rig will
be standard material, for the maximum working pressure intended. The test col-
umn, containing the packed bed of sieved sand (or glass ballotini) may be made
from a short piece of stainless steel tube (typically 80 mm internal diameter and
110 mm long) flanged at both ends. The test column is best kept vertical, to avoid
settling of the packing to one side, and a downward gas flow will prevent
unwanted fluidization of the bed material. A piece of some sort of gauze should be
placed over the bottom plate of the test column to prevent the granular material
from going into the tubing.
If the rotameter gives the flowrate mR ; at pressure PR and temperature TR ;
the actual flowrate in the test column is given as m ¼ mR ðPR =PÞðT=TR Þ; where P is
the absolute pressure indicated in the manometer connected to the test column. The
vapour pressure is calculated using Uno’s correlation [53].
6.1 Measurement of Molecular Diffusion Coefficients 127
PR Manometer
Compressed Flow Manometer Atmosphere

air meter P
V1 V2
Pressure
regulator
Test column in water bath
Fig. 6.2 Experimental set-up
In experiments at temperatures differing from ambient, the test column and a

significant length of the tubing feeding it have to be kept in a constant temperature
bath.
The naphthalene spheres used in the experiments were solidified from the melt,
in simple moulds made of silicone rubber for the purpose. The diameter of each
sphere was measured with callipers (along three perpendicular directions, to ensure
near sphericity) and it was typically about 20 or 25 mm.
Each sphere was weighed accurately ðinitial mass; m0 Þ; in an analytical bal-
ance, before burying it (with some care, to avoid unwanted erosion) near the
middle of the packed bed, which had been previously immersed in the constant
temperature bath for a long enough period of time. The top plate of the test column
was then bolted in place and the column was tapped gently, a few times, to ensure
close packing of the granular material. The test column and associated tubing were
then immersed again in the constant temperature bath and the air supply was then
connected, to give the intended test pressure and gas flowrate. The air flowrate was
kept constant for a time Dt, following which it was interrupted and the test column
was opened, to remove the sphere for weighing ðfinal mass; mt Þ:
Each experiment lasted between 50 min and 60 h, the time being chosen to give
a measurable loss of weight (typically, 0.05 g), that nevertheless would not cor-
respond to a significant variation in the diameter of the sphere. The time intervals
were however sufficiently long to reduce inaccuracies due to evaporation in the
stages of sphere introduction and removal from the test column.
The experimental measurements of mass transfer in water were performed on
the dissolution of individual spheres of for example, 2-napthol or benzoic acid,
buried in beds of sand through which a metered stream of distilled water (which
had been previously deaerated under vacuum) was forced to flow steadily.
Instead of weighing the sphere, the concentration of solute in the outlet stream,
Cout ; was continuously measured to give the rate of dissolution. Cout was measured
by means of a UV/VIS spectrophotometer. When steady state was reached, the rate
of dissolution of the solid could be found directly from n ¼ QCout ; where Q is the
measured volumetric flow rate of water.
6.2 Measurement of Dispersion Coefficients

(Axial and Radial)
6.2.1 Measurement of Axial Dispersion Coefficients
Since the development of the dispersion approximation for the study of solute
transport in capillary tubes by Taylor [52], the flow of the tracer is described by
dispersion due to molecular diffusion and radial velocity variations. In packed
beds, with D=d [ 15; the assumption of flat velocity profiles and porosity is
reasonable as point out by Akehata and Sato [1] and Gunn [25] and later showed
by the experimental studies of Stephenson and Stewart [50] and Gunn and Pryce
[26], that suggested D=d [ 10:
Imagine a packed bed of uniform porosity ðeÞ; contained in a long column of
length L along which liquid flows at a superficial velocity U (the interstitial
velocity is then u ¼ U=e) and initial concentration of solute C0 ; in which a tracer
with continuously injection and concentration of soluteCS ; is dispersed in radial
and axial direction. Taking a small control volume inside this boundary layer, a
material balance on the solute, with length dz and width dr, leads to (see [25])

@2C 1 @ @C @C @C
DL 2 þ DT r u ¼ ð6:8Þ
@z r @r @r @z @t
Klinkenberg et al. [33] and Bruinzeel et al. [9] show that radial dispersion can
be neglected in comparison with axial dispersion for a small ratio of column
diameter to length ðD=LÞ and large fluid velocity. The partial differential equation
describing tracer transport in the bed reduces then to
@ 2C @C @C
DL 2
u ¼ ð6:9Þ
@z @z @t
where z measures length along the bed, and if L is sufficiently large (semi-infinite
bed) the appropriate boundary conditions are
C ¼ C0 0zL t¼ 0 ð6:10aÞ
@C
uCS ¼ uC DL z¼0 t[0 ð6:10bÞ
@z
@C
¼0 z¼L t[0 ð6:10cÞ
@z
For a step input, the concentration at the outlet of the bed ðz ¼ LÞ can be
obtained by Carslaw and Jaeger [10], who give the exact solution of the equivalent
heat transfer problem. However, a study developed by Harrison et al. [29] showed
that the boundary conditions developed by Danckwerts [15], for an infinite system,
hold adequately for a finite system provided uL=DL 10; So, for a step input
6.2 Measurement of Dispersion Coefficients (Axial and Radial) 129
(from C0 to CS), the concentration at the outlet of the bed ðz ¼ LÞ is known [15] to
be given, if L is sufficiently large, by
" rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!#
1 LPeL ð1 hÞ
F(hÞ ¼ 1 erf ð6:11Þ
2 hd 2
or for a pulse response by

1=2 " #
1 LPeL LPeL ð1 hÞ2
EðhÞ ¼ exp ð6:12Þ
2 phd 4hd
Rifai et al. [45] and Ogata and Banks [39] showed that the solution of Eq. 6.9
with the boundary conditions and initial condition given by Eqs. 6.10a–c is
" rffiffiffiffiffiffiffiffiffiffi !#
1 LPeL ð1 hÞ
F(hÞ ¼ 1 erf
2 hd 2
" rffiffiffiffiffiffiffiffiffiffi !# ð6:13Þ
1 LPeL ð1 þ hÞ L
þ 1 erf exp PeL
2 hd 2 d
However, Ogata and Banks [39] showed that for large molecular Peclet num-
bers ðsay; uL=DL [ 100Þ; the advection dominates and the second term in the
right-hand side can be neglected, with an error lesser than 5%, and Eq. 6.13
reduces to Eq. 6.11.
6.2.1.1 Experiments
Typically, dispersion along the direction of flow is studied by following the dis-
tortion of some concentration wave (of a tracer), as it progresses along the packing.
In our experiments, a step in tracer concentration was introduced at the top of a
long packed bed, of constant cross section, and the variation of tracer concen-
tration, continuously in the stream leaving the bed, was recorded.
The partial differential equation describing tracer transport in the bed is given
by Eq. 6.9 and for a step input (from C0 to CS), the concentration at the outlet of
the bed ðz ¼ LÞ is known [15] to be given by

C C0 1 L ut
¼ 1 erf pffiffiffiffiffiffiffiffi ð6:14Þ
CS C0 2 2 DL t
if L is sufficiently large. This result may be written as Eq. 6.11, where
F ¼ ðC C0 Þ=ðCS C0 Þ is the dimensionless concentration rise and h ¼ t=t is the
dimensionless time (t being the mean residence time of fluid in the bed).
For each set of C vs. t values, obtained in one experiment, the values of F vs.
h were calculated and the value of PeL determined to give the least deviation
1.0
PeL=0.90
PeL=0.50
∑ (deviation) =1.57×10
2 –2 2
0.8 i
2 –2
0.6
F
0.4
PeL=0.71
Experiment
2 –6
0.2 PeL = 0.71
i
PeL = 0.50
PeL = 0.90
0.0
0.85 0.90 0.95 1.00 1.05 1.10 1.15
θ
Fig. 6.3 Comparison of experiment with Eq. 6.4 for three values of PeL
between the experimental points and the line representing Eq. 6.11. Figure 6.3
helps illustrate the method and it also gives some idea about its sensitivity to
variations inPeL .
The rig used in the majority of our experiments (including all those at tem-
perature above ambient) is sketched in Fig. 6.4, but the various items are not to
scale. Two large stainless steel reservoirs (about 0.3 m in diameter and 1.0 m tall)
were used to keep the distilled water and the dilute solution of sodium chloride
(up to 1.5 kg/m3 in salt) immersed in a silicone oil thermostatic bath. The use of
very dilute salt solutions and the care taken in equalizing the temperature of both
liquids is needed to avoid dispersion by natural convection. Both the distilled
water and the salt solution were degassed ‘‘in situ’’, by bubbling under vacuum, to
avoid liberation of small air bubbles inside the test column, at the higher
temperatures. Both reservoirs were connected (at the top) to the compressed air
line (approx.: 4 bar) to have the rig permanently pressurised and help discharge the
liquid through the flow regulating valve V, open to the atmosphere.
At the top of the test column (3.0 m long and 0.047 m in diameter, made of
brass), a three-way valve (T) could be manipulated to select the feed. At the
beginning of each experiment it was turned to let distilled water flow through the
bed, until no salt was detected in the conductivity meter monitoring the exit
stream. After that, it was turned to allow the salt solution to go through the column,
at a constant flowrate, measured by an orifice meter connected to a differential
pressure transducer. The dead space between the three-way valve and the top of
the bed was less than 0.1% of the void space in the bed proper and this is important
to ensure that a sharp step input of tracer is obtained in the bed. The conductivity
Pressure
transducers
Compressed
air Termocouple
Three-way valve (T)
Distilled Solution
water of NaCl
to drain
Valve (V)
Conductivity Termocouple
Pre-heater in silicone oil bath cell
Test column immersed in
silicone oil bath
Fig. 6.4 Diagram of experimental set-up
cell at the exit of the test column was connected to a micro-computer and values of
the salt concentration were recorded at a frequency of 10 Hz. Care was taken to
have a negligible volume of liquid between the bottom of the bed and the con-
ductivity cell in order to avoid distortion of the signal read by the cell. A typical
record of outlet tracer concentration is shown in Fig. 6.3 and from the sensitivity
of the curves representing Eq. 6.11, to the value of PeL ; it may be guessed that the
values of DL obtained are accurate to within 15%.
6.2.2 Measurement of Radial Dispersion Coefficients
The radial dispersion coefficient can be determined by plotting ð% composition :

C10 and C90 Þ vs (distance from 50% composition) on arithmetic-probability paper
[40]. The dispersion coefficient can be calculated by

u C90 C10 2
DT ¼ ð6:15Þ
L 3:625
The most widely used techniques for the measurement of lateral dispersion
are the continuous point source and the instantaneous finite source methods
(see e.g. [46]), which rely on the injection of tracer in a flowing liquid, followed by
tracer detection at several points, downstream of the injection point. If at time
t ¼ 0 a tracer is injected into the porous medium from an injector, for the con-
tinuous point source method the tip of the injector is taken as the tracer origin. For
the instantaneous finite source method the origin lays just down-gradient of the
tracer injector.
Several authors like Roemer et al. [47] and Gunn and Pryce [26] used the
solution of Eq. 6.15 when the axial dispersion coefficient is taken equal to radial
dispersion coefficient. However, in this work we only consider experimental
techniques where axial dispersion is neglected.
6.2.2.1 ‘‘Instantaneous Finite Source’’ Method
The method adopted by some authors like Dorweiler and Fahien [16] and Fahien
and Smith [19] is based that the tracer is fed into the main stream at a point on the
axis on the column.
The analytical model for an instantaneous finite source in one dimension is first
presented by Crank [13]. Baetsle [6] extended the model to three-dimensional
dispersion. Hunt [32] and Sun [51] provided the three-dimensional solution to the
advection–dispersion equation [6] using different mathematical analysis. Van
Genuchten and Alves [54] presented a number of analytical solutions of the one-
dimensional convective–dispersive solute transport equation.
Tracer concentration should be low enough to avoid density-induced flow
effects. The tracer should be conserved (i.e. not destroyed) in the experiment and
the distribution of flow rates at the outlet must be the same as in the feed so as not
to induce complications in the flow field.
Radial dispersion may be evaluated by injecting a steady flow of a tracer at a
point of a test section column. For a boundary layer, which is thin in comparison
with the length of the axial distance (L), axial dispersion will be negligible. Taking
a radial co-ordinate, r, to measure distance to the axis of the bed and a co-ordinate
z, to measure distance along the average flow direction, the differential mass
balance on the solute reads

DT @ @C @C
r ¼u ð6:16Þ
r @r @r @Z
where DT is the radial dispersion coefficient. Fahien and Smith [19] solved the
differential dispersion Eq. 6.16 with
z ¼0 0\r\Ri C ¼ C0 ð6:17aÞ
z¼0 Ri \r\R C¼0 ð6:17bÞ
@C
all z r ¼ R; r ¼ Ri ¼0 ð6:17cÞ
@r
and the solution of Eq. 6.16 with the boundary conditions (6.17a–c) is

C 2R X1
J1 ðbn Ri =RÞJ0 ðbn r=RÞ z b2n z
¼1þ exp ð6:18Þ
C0 Ri n¼1 bn J20 ðbn Þ R PeT R
z = X0
–∞ +∞
tracer z = XM
Fig. 6.5 A schematical diagram of test section for radial dispersion
where J0 and J1 are the Bessel functions of the first kind, of order 0 and 1,
respectively, and the bn are the positive roots of the Bessel function of the first
kind, of order 1.
6.2.2.2 ‘‘Continuous Point Source’’ Method
This method is based on the measurement of radial mass exchange between two
coaxial portions of a packed bed, along which liquid flows, parallel to the axis; the
feed to the central portion is water containing a small amount of sodium chloride
and that to the outer portion is pure water.
Klinkenberg et al. [33] derived an analytical solution for Eq. 6.16, neglecting
the effect of injector radius (see Fig. 6.5), with the boundary conditions given by
z ¼ þ1 all R C ¼ C0 ð6:19aÞ
z ¼ 1 all R C¼0 ð6:19bÞ
@C
all z r ¼ R; r ¼ 0 ¼0 ð6:19cÞ
@r
and the solution of Eq. 6.16 with the boundary conditions (6.19a–6.19c) is
X1 2
C J0 ðbn r=RÞ bn D T z
¼1þ 2 exp 2 ð6:20Þ
n¼1 J0 ðbn Þ
C0 Ru
where J0 is the Bessel function of the first kind, of order 0, and the bn are the
positive roots of the Bessel function of the first kind, of order 1.
Plautz and Johnstone [43] and Sinclair and Potter [49] used Eq. 6.9 for an
infinite case, where no boundary is present, of mass diffusion from a point source.
The result with axial dispersion neglected was given by Carslaw and Jaeger [10]

C R2 u r2 u
¼ exp ð6:21Þ
C0 4DT z 4DT z
This solution includes a simplification possible when z=r [ 5 (axial dispersion
neglected).
z= 0 z z=L
u u
r
2R
u 2Ri 2Ri u
u u
C=0 C = Cω
C = C0 C=C
C=0 C = Cω
Fig. 6.6 Sketch of boundary conditions proposed by [30]
Blackwell [8] assumes the effect of radius injector and gives the analytical
solution to the differential equation describing radial dispersion in the absence of
axial dispersion. Hiby and Schummer [30] presented a solution of Eq. 6.16 that
considered the tracer pipe to be of significant diameter compared to the diameter of
the bed (see Fig. 6.6), and the boundary conditions adopted were
@C
z[0 r¼R ¼0 ð6:22aÞ
@r
z ¼ 0 r\Ri C ¼ C0 ð6:22bÞ
z¼0 Ri \ r\R C ¼ 0 ð6:22cÞ
On the assumption that DT and u are independent of position, the solution of

Eq. 6.9 following Hiby and Schummer [30] gives, for the resulting outlet average
concentration in the inner stream of liquid,
" 2 #
C X 1 2
J1 ðbn Ri =RÞ Ld bn
¼4 2 2
exp ð6:23Þ
C0 n¼0 bn J0 ðbn Þ
PeT R
where J0 and J1 are the Bessel function of the first kind, of orders 0 and 1, respectively,
and the bn are the positive roots of the Bessel functions of the first kind, of order 1.
The measurement of C and C0 provides a method for the determination of PeT
(and therefore of DT), since all other parameters in the equation are known.
Harleman and Rumer [28] and Han et al. [27] consider a steady-state experi-
ment in a rectangular column. The authors solved the differential equation with the
boundary conditions,
C ¼ C0 x¼0 0\y\ þ 1 ð6:24aÞ
C¼0 x¼0 1\y\ 0 ð6:24bÞ
@C
¼ 0 all x y!1 ð6:24cÞ
@y
and the solution obtained for a step input in concentration, is
" rffiffiffiffiffiffiffiffi !#
C 1 PeT y
¼ 1 erf ð6:25Þ
C0 2 Ld 2
6.2.2.3 Mass Transfer From a Flat Surface Aligned With the Flow
Coelho and Guedes de Carvalho [12] developed a new experimental technique,

based on the measurement of the rate of dissolution of planar or cylindrical sur-
faces, buried in the bed of inert particles and aligned with the flow direction.
Figure 6.7a sketches a section through a packed bed along which liquid is
flowing, close to a flat wall, part of which (0 \ x \ L) is slightly soluble. Liquid
flow will be taken to be steady, with uniform average interstitial velocity u, and if
the concentration of solute in the liquid fed to the bed is C0 and the solubility of the
solid in the wall is C*, a mass transfer boundary layer will develop, across which
the solute concentration drops from C ¼ C ; at y ¼ 0; to C ! C0 ; for large y.
The question of how large is meant by a ‘‘large y’’ needs some clarification.
Obviously, if L were only of the order of a few particle diameters, and u were
large, the concentration of solute would fall to C0 over a distance of less than one
particle diameter. In that case, flow in the bulk of the packed bed would have little
influence on the mass transfer process, which would be dominated by diffusion in a
thin layer of liquid, adjacent to the soluble surface. Already for large L and low u,
the thickness of the mass transfer boundary layer will grow from zero, at x = 0,
to a value of several particle diameters, at x = L and the process of mass transfer
will then be determined by a competition between advection and dispersion in the
bulk of the bed. Now, it is well known (e.g. [55]) that the voidage of a packed bed
(and therefore the fluid velocity) is higher near a containing flat wall, but in the
case of Guedes de Carvalho and Delgado [24] experiments it may be considered
that such a non-uniformity will have negligible effect. For one thing, we work with
bed particles of between 0.2 and 0.5 mm and therefore the region of increased
voidage will be very thin. Furthermore, because the inert particles making up the
bed indent the soluble surface slightly, as dissolution takes place (and this slight
indentation is easily confirmed when the piece of soluble solid is removed from
Fig. 6.7 a Flow along u

soluble slab, b Mass transfer (C = C0)
boundary layer
x=0 y y
Y
Packed bed x
y
x
x=L
(a) (b)
within the bed), there is in fact virtually no near wall region of higher voidage.
Confirmation of these assumptions is given by the results of the experiments
described below.
Taking a small control volume inside this boundary layer (see Fig. 6.7b), with side
lengths dx; dy and unity (perpendicular to the figure), it is possible to perform a mass
balance on the solute, for the steady state. If the boundary layer is thin, compared to
the length of the soluble slab, axial dispersion is likely to be negligible, since the
surface y = 0, 0 \ x \ L, is a surface of constant concentration (C = C*).
Noting that the surface y ¼ 0; 0\ x \L; is a surface of constant concentration,
along which @ 2 C=@x2 ¼ 0 and axial dispersion will be negligible, for a boundary
layer which is thin in comparison with the length of the soluble slab. (A conser-
vative criterion for this approximation to be valid is L=d [ 20). For a slab the
equation of diffusion in one dimension is
@C @ 2C
u ¼ DT 2 ð6:26Þ
@x @y
to be solved with
C ¼ C0 x ¼0 y[0 ð6:27aÞ
C ¼ C x[0 y¼0 ð6:27bÞ
C ! C0 x[0 y!1 ð6:27cÞ

The solution is
!
C C0 y
¼ erfc pffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:28Þ
C C0 2 DT x=u
and the flux of dissolution at any point on the slab surface may be obtained from
(6.28) as

@C DT 1=2
N ¼ DT e ¼ ðC C0 Þe ð6:29Þ
@y y¼0 px=u
The instant rate of solid dissolution over the whole slab surface may now be
calculated by integration of the local flux; taking a width b along the surface of the
solid, perpendicular to the flow direction, there results
ZL 1=2
4DT
n¼ N b dx ¼ ðC C0 Þe b L ð6:30Þ
pL=u
0
and it is useful to define the coefficient

1=2
n 4DT
k¼ ¼ e ð6:31Þ
ðbLÞðC C0 Þ pL=u
This result shows how the measurement of the rate of dissolution of the solid,
which is directly related to the average mass transfer coefficient, may be used to
determine the coefficient of radial dispersion in the bed.
A simple way of checking the result in Eq. 6.31 is afforded by the predicted
proportionality between k and the inverse square root of L. Experiments performed
by Coelho and Guedes de Carvalho [12] with a wide range of slab lengths, both for
the dissolution of benzoic acid in water and the sublimation of naphthalene in air,
confirm the general validity of the above theory, provided that the approximate
criterion

L ud
0:62 ð6:32Þ
d Dm
is observed, where Dm is the molecular diffusion coefficient of the solute. When

the above criterion is not observed, the near wall film resistance to diffusion will
have to be taken into account and approximate ways of doing this are described by
Coelho and Guedes de Carvalho [12].
The similarity between the result given by Eq. 6.32 and that obtained by
Higbie [31], for gas–liquid mass transfer by surface renewal, is striking.
Equation 6.24a–c simply states that the average mass transfer coefficient, for the
soluble wall, is that corresponding to surface renewal with a time of contact
tc ¼ L=u and an apparent diffusion coefficient DT .
6.2.2.4 Mass Transfer From a Cylinder Aligned With the Flow
For practical reasons, it proves simpler to perform experiments in which the dis-
solving solid is a cylinder, aligned with the flow direction and it is important to
know the theoretical expressions relating the average mass transfer coefficient with
the coefficient of dispersion, DT ; for that situation.
Fortunately, under appropriate conditions, easy to reproduce in the laboratory, the
thickness of the mass transfer boundary layer is small in comparison with the radius
of the dissolving cylinder and under such circumstances, the analysis presented
above, for dissolution from a flat surface, is still applicable with good accuracy.
However, there are instances in which this simplification is not valid and an
exact solution may be worked out in cylindrical co-ordinates, as shown by Coelho
and Guedes de Carvalho [12].
The resulting expression for k is cumbersome to evaluate, but for small values
of the parameter hc ¼ DT tc =a2 ; where tc ¼ L=u is the time of contact between
liquid and solid, a good approximation is
pffiffiffi pffiffiffi
4DT 1=2 p 1=2 1 p 3=2
k¼e 1þ hc h c þ hc ð6:33Þ
ptc 4 12 32
For higher values of hc, up to hc =0.4, the first four terms may be used, instead
of the infinite series on the right hand side of Eq. 6.33, with an error of less than
1% in k.
6.2.2.5 Experiments
Experiments were performed on the dissolution of individual cylinders of

2-napthol, buried in beds of sand through which a metered stream of distilled water
(which had been previously deaerated under vacuum) was forced to flow steadily,
as sketched in Fig. 6.8. The beds of sand were contained in a stainless steel column
500 mm long and 100 mm in internal diameter, with the cylinder of 2-napthol
placed co-axially inside it, all in vertical alignment. The cylinders of 2-napthol had
a diameter of 20 mm and a length of 250 mm and they were tightly mounted in
alignment between two rods of stainless steel (each 20 mm in diameter and
100 mm long). These two metal rods fulfilled the double purpose of covering the
top and bottom of the 2-napthol cylinder and straightening the flow field, upstream
and downstream from it; they also provided the points of support for alignment
with the stainless steel column. Near the bottom of the stainless steel column, a
perforated plate, covered with fine stainless steel wire mesh, was used to support
the bed of sand and the cylinder inside it. The preparation and assemblage of the
cylinders of 2-napthol followed closely the method detailed by Coelho and Guedes
de Carvalho [12] for the preparation of cylinders of benzoic acid. The use of
2-napthol instead of benzoic acid was determined by the need to work at high
temperatures and the relevant properties of 2-naphtol.
Fig. 6.8 Diagram of experimental set-up
Before any new series of runs, the cylinder buried in the sand (and when
required, the sand itself) had to be replaced. In order to do that, ball valves B1 and
B2 were closed and disconnected from the upstream and downstream piping,
respectively, so as to allow the test column and the associated copper coil (and
valves B1 and B2 in the ends) to be lifted from the thermostatic bath. Following
that, the free ends of valves B1 and B2 were connected to plastic tubing that could
be supplied with distilled water or directed to the drain. After removing the lid of
the test column, distilled water was forced up through the bed of sand, so as to
fluidise it slightly and allow the cylinder to be replaced without difficulty. As the
water flow was stopped, to allow the sand particles to settle, the column was
vibrated for a few seconds to give a good compaction of the bed. The lid of the test
column was placed back in position and the stainless steel column (with attached
copper coil and valves B1 and B2) was again immersed in the silicone oil bath.
After connection to the main water circuit, valves B1, B2, and valve B3 were fully
open and the flowrate of water fed to the top of the column was adjusted by means
of valve N. The silicone oil bath was then heated to the required temperature of
operation. A pre-heater helped warm up the water feed to near the temperature of
the bath and an ice water bath was used to cool the liquid stream leaving the test
column, before it passed the valve regulating the flow. In this way it was possible
to perform experiments at temperatures up to the normal boiling point of water
(and above it, if necessary).
The concentration of 2-naphtol in the outlet stream, Cout ; was continuously
measured by means of a UV/VIS Spectrophotometer, set at 274 nm, and when
steady state was reached, the rate of dissolution of the solid could be found from
n ¼ QCout ; where Q is the measured volumetric flow rate of water.
For the second example, a rig was built as sketched in Fig. 6.9, that allowed
experiments to be performed over a wide range of temperatures. Two large res-
ervoirs, of about 0.06 m3 internal volume, were used to store distilled water and a
Fig. 6.9 Sketch of experimental set-up for measurement of radial dispersion coefficient
dilute solution of NaCl in distilled water with a concentration of salt of approxi-

mately 1.5 kg/m3; both reservoirs were connected, at the top, to a source of air at
4 bar. The two reservoirs were kept in a thermostatic bath that was set at the
temperature of the experiment, or at 353 K, when the experiments were performed
above that temperature. The liquid held in each reservoir had been initially
deaerated by stripping under vacuum, at room temperature, to avoid subsequent
liberation of gas bubbles in the liquid streams going through the packed bed. In the
test column, care was taken to have a good alignment of the axes of the inlet and
outlet tubes of the core stream with that of the packed bed. The apparatus was kept
in vertical alignment to avoid ‘‘settling’’ of the packing to any one side. Liquid
flow was driven by the difference in pressure between the reservoirs and the exit
sections, at atmospheric pressure. Valves V1 and V2 were used to adjust the outlet
flowrate of each stream, to give equal velocity in the core tube and in the annulus,
at the outlet; by acting slightly on valves E1 or E2 it was then possible to also
impose equal velocities in the inlet streams to the bed. At the higher temperatures,
the outlet streams had to be cooled before reaching valves V1 and V2, to avoid
vaporization downstream of these valves, due to de-pressurization. The conduc-
tivity meter cell was built on a 30 mm nylon rod, 75 mm long, having a 10 mm
hole drilled along its axis. Two platinum wires, 0.5 mm in diameter, crossed the
wall of this nylon tube along opposite ends of one diameter of the mid cross
section and they were glued in place, leaving a distance of about 8 mm between
their tips. The platinum electrodes were connected to three 1 kX resistances (to
within 0.01%) to form a Wheatstone bridge that was connected to a 9.30 V source,
so that the liquid flowing through the cell acted as the variable resistance. The
output of this conductivity meter was connected to a micro-computer and care was
taken to calibrate the meter at the operating temperatures, whenever fresh solutions
were introduced in the reservoirs. After allowing steady state to be reached, the
rate of solute transfer from the core stream to the annulus is given simply as
n ¼ v C0 C ; where m is the volumetric flowrate of the core stream and C is the
average concentration of salt in that stream, at the exit. Care was taken to check
that the value of n agreed with n ¼ x C x ; where x is the volumetric flowrate of the
stream going through the annulus and Cx the corresponding average salt con-
centration, at the outlet.
6.3 Measurement of Solubility at Different Temperatures
Solubility is perhaps the most fundamental of all chemical phenomena. The sig-
nificance of what dissolves what, to what extent, at what temperature and pressure,
and the effects of other species, was recognized at a very early stage. In more
recent times the importance of solubility phenomena has been acknowledged
throughout science. For example, in the environment, solubility phenomena
influence the weathering of rocks, the creation of soils, the composition of natural
water bodies and the behaviour and fate of many chemicals.
The characteristic ability of water to behave as a polar solvent changes when
water is subjected to high temperatures and pressures. As water becomes hotter, its
molecules seem much more likely to interact with non-polar molecules. For
example, at 300°C (and high pressure) water has dissolving properties very similar
to acetone, a common organic solvent.
Also, solid–liquid and solid–gas mass transfer investigations with Newtonian or
non-Newtonian fluids are frequently made by following the rate of dissolution of a
low solubility solute. In all researches, accurate solubility data are required.
On mass transfer investigations in porous media, as in studies of dispersion
coefficients and solute transport, the most common solutes used are benzoic acid,
2-naphthol, naphthalene, salicylic acid and succinic acid with water or air
(see [56]). In these experiments, knowledge of accurate solubility data at different
temperatures is very important, i.e., for low solubility solutes.
The experiment proposed is simple and inexpensive, and it provides an accurate
method for the measurement of solubilities of solid solutes in liquids and gases.
Consider a vertical column of length L; containing a packed bed of soluble spherical
particles of diameter d1 : If liquid flow is steady, with a uniform volumetric flowrate
Q; if the concentration of solute in the liquid fed to the bed is c0 and the solubility of
the solid particle is c ; a mass transfer boundary layer will develop.
In the analysis of results of experiments of dissolution of soluble spherical
particles in liquid flow, the equation for dissolution rate is given by,
@c
Q ¼ kSL ðc c0 Þ ð6:34Þ
@x
where SL is the active surface area per unit length and k is the average mass
transfer coefficient. Given constant flowrate, uniformly distributed particles and
isothermal conditions, Eq. 6.34 is integrated between the inlet and outlet condi-
tions of the bed, x ¼ 0 to x ¼ L and c ¼ c0 to c ¼ c: The following equation
results,

c c0 kSL
¼ 1 exp L ð6:35Þ
c c0 Q
In order to guarantee that the outlet stream is saturated, it’s important to observe
the approximate criterion ðc c0 Þ=ðc c0 Þ [ 0:999 (error less that 0.1%). The
number of soluble spheres presented in a packed bed is given by,
Vcolunn 3 ð1eÞD2 L
n¼ ¼ ð6:36Þ
Vparticle 2 d13
and the general validity of the above theory holds, provided that the approximate
criterion
6ð1 eÞkL
[ 6:908 ð6:37Þ
u 0 e d1
If the criterion of Eq. 6.37 is to be satisfied, it is important to know the value of
the average mass transfer coefficient, k; so as to be able to estimate the interstitial
velocity of liquid, u0 :
6.3.1 Mass Transfer Around a Buried Soluble Sphere
For the propose of analysis, let as consider the situation of a slightly soluble sphere
of diameter d1 ð¼2aÞ buried in a bed of inert particles of diameter d ðwith d
d1 Þ,
packed uniformly (void fraction e) around the spheres. The packed bed is assumed
to be ‘‘infinite’’ in extent and a uniform interstitial velocity of liquid, u0 ; is
imposed, at a large distance from the spheres.
In order to obtain the flow field in the vicinity of the buried sphere, Darcy’s law,
u = K grad p; is coupled with the continuity equation, div u ¼ 0; and Laplace’s
equation, r2 / ¼ 0; is obtained for the flow potential / ¼ k p:
In terms of spherical coordinates (r, h), the potential and stream functions are,
respectively (see [14]),

1 a
3
/ ¼ u0 1 þ r cos h ð6:38Þ
2 r
a
3
u0
w¼ 1 r 2 sin2 h ð6:39Þ
2 r
and the velocity components are

a
3
@/
ur ¼ ¼ u0 cos h 1 ð6:40Þ
@r r
6.3 Measurement of Solubility at Different Temperatures 143

1 @/ 1 a
3
uh ¼ ¼ u0 sin h 1 þ ð6:41Þ
r @h 2 r
Making use of the potential and stream lines, it is possible to perform a material
balance on the solute in a differential element of a ‘‘stream tube’’ to obtain (see [11])

@c @ @c @ @c
¼ DL þ DT x2 ð6:42Þ
@/ @/ @/ @w @w
where x is the distance to the flow axis, and DL and DT are the longitudinal and
transverse dispersion coefficients, respectively.
The boundary conditions to be observed in the integration of Eq. 6.42 are:
(1) the solute concentration is equal to the background concentration, c0 ; far away
from the sphere; (2) the solute concentration is equal to the equilibrium concen-
tration, c ¼ c ; on the surface of the sphere and (3) the concentration field is
symmetric about the flow axis.
For very low fluid velocities, dispersion is the direct result of molecular dif-
fusion, with DT ¼ DL ¼ D0m ; and the numerical solution presented by Carvalho
et al. [11] applies. Those authors suggest that their results are well approximated
(with an error of less than 1%) by

D0m 4 0 2=3 4 0 1=2
k¼e 4 þ ðPe Þ þ Pe ð6:43Þ
d1 5 p
where Pe0 ¼u0 d1 =D0m is the Peclet number for the soluble sphere.
Now, by substituting the average mass transfer coefficient, given by Eq. 6.43,
into Eq. 6.37, the following expression is obtained for the approximate validity
criterion of theory developed above:
! 2
1 p p d1
1þ 0þ [ ð6:44Þ
Pe0 Pe 5ðPe0 Þ1=3 ð1 eÞL
Finally, with Eq. 6.44 we could predict the volumetric flowrates that guarantee
saturation in the outlet stream, Q ¼ Pe0 pD2 eD0m =ð4d1 Þ: However, an important
aspect to consider is the dependence of Q on the effective molecular diffusion
coefficient, D0m : Fortunately, values of D0m increase with temperature, and the value
of D0m ; at room temperature or lower, is a good estimate.
6.3.2 Experimental Set-Up
Experiments were performed on the dissolution of spheres of benzoic acid,

2-naphthol and salicylic acid (6.0 mm of internal diameter), buried in beds of sand
Fig. 6.10 Sketch of experimental set-up
(0.496 mm average particle diameter) through which water was steadily forced
down, at temperatures in the range 293–373 K.
A stainless steel tube (21 mm i.d. and 200 mm long) was used to hold the bed
of soluble solid spheres in an upright position while a metered stream of distilled
water was fed to the top of the column, as sketched in Fig. 6.10. Near the bottom
of the stainless steel column, a perforated plate, covered with fine wire mesh, was
used to support the bed.
The distilled water was initially dearated, under vacuum, to avoid liberation of
gas bubbles in the rig, at high temperature. The test column was immersed in a
silicone oil bath kept at the desired operating temperature by means of a ther-
mosetting bath head (not represented in Fig. 6.10). The copper tubing feeding the
distilled water to the column at a constant metered rate was partly immersed in a
pre-heater and it had a significant length immersed in the same thermosetting bath
as the test column; the copper tubing leaving the test column was immersed in a
chillier to cool the outlet stream before reaching the UV analyser.
The water flowrate was then adjusted to the required value, Q; and the
concentration of solute in the outlet stream was continuously monitored by means
of a UV/VIS Spectrophotometer (set at 274 nm, for 2-naphtol, at 226 nm, for
benzoic acid and 292 nm, for salicylic acid).
The solubility of the solutes studied in water was calculated from the steady
state average concentration of solute, cout ; in the outlet stream (refrigerated to
6.3 Measurement of Solubility at Different Temperatures 145
room temperature), as c ¼ ð1 þ Q1 =QÞcout ; where Q and Q1 were the measured

volumetric flowrates.
The spheres of solutes studied were prepared from p.a. grade material, which
was molten and then poured into moulds made of silicone rubber. Wherever any
slight imperfections showed on the surface of the spheres, they were easily
removed by rubbing with fine sand paper. Using callipers three measurements
were made of the diameter of each sphere along three perpendicular directions.
6.4 Measurement of Tortuosity in Porous Media
The two major properties to describing porous media and the associated mass
transfer phenomena are the permeability coefficient (flow phenomena) and the
effective diffusion (mass transfer phenomena). However, both coefficients are
functions of the characteristics of porous media, namely the media porosity and
tortuosity (see [17, 18]).
The tortuosity of a packed bed is an important parameter that describes pore
connectivity and fluid transport, however, it is difficult to determine experimentally.
Normally, tortuosity is calculated through measured values of the porosity and the
experimentally determined effective diffusion coefficient. However, tortuosity var-
ied with the particle volume fraction and the particle size ratio in the mixture.
In recent years, nuclear magnetic resonance (NMR) has been used to determine
both diffusion and tortuosity coefficients (see e.g. [22, 36]), with significant
advantages, but it’s a very expensive method.
The experiment proposed is simple and inexpensive, and it provides an accurate
method for the measurement of the tortuosity in packed beds.
Tortuosity can be defined as the ratio of the distance actually travelled by a
tracer through the pore space, Le, to the straight-line distance between the two
points, L, of the porous media with solute concentrations C1 and C2 : When
ignoring that the ‘‘real’’ length (or effective) of diffusion trajectory is sinuous the
representative equation of the diffusion law it should be written as:
C 1 C2
Q ¼ D0m e A ð6:45Þ
L
where D0m is the ‘‘effective’’ molecular diffusion coefficient. However, the appli-
cability of Ficks first law to the ‘‘real’’ length of diffusion in a packed bed,
Le, result as
C 1 C2
Q ¼ Dm e A ð6:46Þ
Le
where Dm is the molecular diffusion coefficient. Now, by substituting Eq. 6.45 into
6.46 the following relation is obtained:
Dm
D0m ¼ ð6:47Þ
Le =L
The ratio Le =L being the tortuosity factor and it is usually represented by

s [7, 18]. In this work an original method for measurement of tortuosity in packed
beds is presented.
Consider a vertical column containing a packed bed of inert particles, and filled
with liquid to some level above the top of the packed bed (liquid ‘‘pool’’). If a
concentrated salt solution is then poured into this liquid ‘‘pool’’, with uniform
concentration Cp0 ; the tracer will gradually penetrate down the packed bed. The
concentration of tracer in the liquid ‘‘pool’’ decrease gradually in the time, until a
uniform concentration of tracer, equilibrium, is reached in the whole liquid.
Analysis of the process of salt diffusion may be made in analogy with the
process of diffusion from a stirred solution of limited volume, described by
Crank [13]. The concentration of the dissolver salt in the packed bed, C; will vary
according to (Fick’s Second Law)
@C @ 2C
¼ D0m 2 ð6:48Þ
@t @x
subject to the following initial and boundary conditions,
t¼0 0xL C¼0 ð6:49aÞ
@C
t[0 x¼0 ¼0 ð6:49bÞ
@x
@C Vp @Cp
t[0 x ¼ L D0m e ¼ ð6:49cÞ
@x A @t
where Vp is the volume of liquid in the ‘‘pool’’, above the bed of inerts and A is the
cross section area of the packed bed.
The solution of Eq. 6.48 with the boundary conditions (6.49a–c) could be
obtained through the use of Fourier transform. However, the solution given by
Crank [13] is conveniently expressed as the ratio between the amount of tracer in
the bed of inert particles at any time, Mt ; and the corresponding amount of tracer
after a sufficiently long time, equilibrium, M1 :
X1
Mt 2að1þaÞ D0m q2n t
¼ 1 exp ð6:50Þ
M1 n¼0
1 þ a þ a2 q2n L2
where a is the ratio between the volume in the ‘‘pool’’ and the volume in the
packed bed ða¼Vp =VÞ and qn are the non-zero positive roots of
tanðqn Þ ¼ a qn ð6:51Þ
6.4 Measurement of Tortuosity in Porous Media 147
As smaller is the value of D0m t=L2 more terms in the series, in Eq. 6.50, are
needed for a given accuracy. When more than three or four terms are need it is
better to use an alternative form of solution. For most values of a; the simplest
expression is
" 0 0 0:5 #
Mt Dm t D t
¼ ð1 þ aÞ 1 exp 2 2 erfc 2m 2 ð6:52Þ
M1 La L a
The amount of tracer in the packed bed, at any time t; could be expressed by the
following equation,

Mt ¼ Vp Cp0 Cp ðtÞ ð6:53Þ
where Cp0 is the initial tracer concentration in the ‘‘pool’’; and the equilibrium
amount of tracer in the packed bed of inerts is given by

M1 ¼ V ðC1 C0 Þ ¼ Vp Cp0 Cp1 ð6:54Þ
where C0 is the initial concentration of tracer in the bed of inerts (C0 ¼ 0) and C1
and Cp1 are the equilibrium salt (tracer) concentration (t ¼ 1), in the packed bed
and in the liquid ‘‘pool’’, respectively (C1 ¼ Cp1 ). So, the equilibrium tracer
concentration in packed bed could be determined as,
C0 V þ Cp0 Vp Cp0 Vp
C1 ¼ ð6:55Þ
Vp þ V Vp þ V
and, for experimental data, the expression of the ratio between the amount of tracer
at any time t and the amount of tracer in equilibrium is given by:

Mt Vp þ V Cp0 Cp Vp þ V Cp0 Cp
¼ ð6:56Þ
M1 V Cp0 C0 VCp0
In each experiment the value of a is constant and the values of Cp was measured
at different times in order to determine the corresponding values of Mt =M1 ; from
the appropriate representation of Eq. 6.52 or 6.54; and the value of D0m t=L2 cor-
responding to each value of Mt =M1 could be determined. For each experimental
data point a plot of D0m t=L2 vs: t; was organized and the best straight line through
the points and the origin determined. The plot gives a straight line with slope of
D0m t=L2 and intercept of origin, to yield the experimental value of effective
molecular diffusion coefficient, D0m : Finally, the value of tortuosity was obtained
by s ¼ Dm =D0m .
Fig. 6.11 Sketch of

experimental set-up
conductivity meter cell
mechanical stirrers
Personal
computer “liquid pool”
(data logger) L
Packed bed
x
6.4.1 Experimental Example
All measurements of tortuosity were performed in a transparent acrylic column,

0.10 m in diameter, was kept at a constant temperature using a thermostatically
controlled water-bath (see Fig. 6.11).
The packed beds used in our experiments were beds of silica sand with average
diameter of 0.496, 0.297, 0.219 and 0.110 mm. The silica sand was washed, dried
and sieved in closely sized batches, for the experiments.
The dried silica sand was placed back inside the acrylic column at L length, and
the packed bed was carefully compacted. Afterwards, distilled water, with volume
V, was flowed in the ‘‘open space’’ of the packed bed; and a dilute salt solution
flowed carefully on top of the packed bed, with volume Vp (liquid ‘‘pool’’).
A mechanical stirrer was used to homogenize the salt concentration in liquid
‘‘pool’’, with very slow rotation to avoid dispersion by forced convection.
The liquids used in the experiments were both distilled water and a dilute
solution of NaCl in distilled water with a concentration of salt in the range of
0.05–2.0 M. The use of very dilute salt solutions and the care taken in equalizing
the temperature of both liquids is needed to avoid dispersion by natural
convection.
For each experiment the fraction of liquid volume in the packed bed, k, was
determined. The value of a¼Vp =V used in the experiments is largely a matter of
personal choice. The data in the present work refer to a 1:5.
To prevent evaporation the column are closed with a rubber cap. However, it is
impossible to prevent some evaporation during the long execution time of the test.
The change of concentration in liquid ‘‘pool’’ was measured by monitoring
decreases in the electrical conductance as a function of time. The output of this
conductivity meter was connected to a micro-computer and care was taken to
calibrate the conductivity cell.
6.5 Mass Transfer Around Active Solids 149
6.5 Mass Transfer Around Active Solids
There are several situations of practical interest, both in nature and in man made
processes, in which there is fluid flow through a bed of inert particles, packed
around a solid mass that contacts with the moving fluid. Examples may be found in
diverse fields, such as dilute catalyst fixed bed reactors, fluidised bed combustion,
ore leaching and water contamination by buried waste. In such processes there is
an interplay between diffusion, convection and dispersion and a detailed system-
atic study of the problem has been given by Coelho and Guedes de Carvalho [12]
for transfer from buried flat surfaces and for transfer from buried spheres. Both
these references present accurate solutions for certain limiting situations, namely
those of low and high fluid velocity (more precisely, low and high Peclet
numbers).
When the mass transfer process occur in a porous media with a fluid flowing
around the soluble particle, and at low fluid velocities (as typically observed in
underground flow), the assumption of thin boundary layer is not legitimate, and the
theoretical analysis developed by Coelho and Guedes de Carvalho [12] is not
applicable. Therefore, it is necessary to employ numerical methods for a correct
analysis of the mass transfer process in more general situations.
Flow around a buried sphere is an important model situation in many processes
and in a recent work Carvalho et al. [11] treated the problem numerically, so as to
cover the entire range of values of Peclet and Schmidt numbers.
Flow along buried cylindrical surfaces and flat surfaces are also important
model situations, and were investigated theoretically and numerically by Alves
et al. [3], yielding results for a wide range of values of Peclet number, aspect ratio
of soluble solid mass and Schmidt number.
Over the last years, our focus has been driven preferentially to the analysis of
the mass transfer of particles with spherical geometry. This work is a contribution
for the study of mass transfer of soluble particles with different geometries buried
in inert particles with smaller diameter. Additionally, a simple approximate
method is presented to obtain concentration contours plots for solute distribution
around and downstream of the buried surfaces (with different geometries).
The following sections present a detailed description of the mass transfer and
dispersion process around a soluble solid particle with different shapes (sphere,
cylinder or a plane surface aligned with the flow, cylinder in cross-flow, prolate
spheroid and a oblate spheroid) buried in a packed bed of smaller inert particles
with uniform voidage, with a moving fluid with constant interstitial velocity.
6.5.1 Mass Transfer From a Soluble Flat Slab
Figure 6.12 sketches a section through a packed bed along which liquid is flowing,
close to a flat wall, part of which (0 \ z \ L) is soluble. Liquid flow is assumed
Fig. 6.12 Flow through

packed bed near soluble flat
surface
steady, with uniform average interstitial velocity u. If the concentration of solute in

the liquid fed to the bed is c0 and the solubility of the solid wall is c*, a mass
transfer boundary layer will develop, across which the solute concentration drops
from c = c*, at y = 0, to c ? c0 for large y.
If we restrict the analysis to those situations for which the mass transfer
boundary layer extends over several particle diameters and if a small control
volume is considered, inside this boundary layer, with side lengths dz, dy and
unity, a steady state material balance on the solute leads to
@c @2c @2c
u ¼ D0m 2 þ D0m 2 ð6:57Þ
@z @y @z
To integrate Eq. 6.57 it is convenient to define the following dimensionless
variables:
c c0
C¼ ð6:58aÞ
c c0
y
Y¼ ð6:58bÞ
L
z
Z¼ ð6:58cÞ
L
uL
Pe0sf ¼ ð6:58dÞ
D0m
where Pe0sf represents the Peclet number (based on the length, L, of the soluble
slab), and D0m is the effective molecular diffusion coefficient, defined as the ratio
between the molecular diffusion coefficient and the tortuosity, s; of the packed bed
ðD0m ¼ Dm =sÞ.
In terms of dimensionless variables, Eq. 6.57 becomes
@C @ 2 C @ 2 C
pe0sf ¼ þ ð6:59Þ
@Z @Z 2 @Y 2
and the appropriate boundary conditions are
C!0 Z ! 1 8Y ð6:60aÞ
C=1 0Z1 Y¼0 ð6:60bÞ
@C
¼0 Z \0 _ Z [ 1 Y¼0 ð6:60cÞ
@Y
C!0 8Z Y ! þ1 ð6:60dÞ
C!0 Z ! þ1 Y0 ð6:60eÞ

Equation 6.59 is to be solved numerically, subjected to the boundary conditions
(6.60a–6.60e), over the ranges of Pe0sf of practical interest.
6.5.1.1 Discretisation
Equation 6.59 was solved numerically, using a finite-difference method in a non-

uniform grid similar to that adopted by Carvalho et al. [11]. A second-order central
differencing scheme was adopted for the discretisation of the diffusive terms on the
right hand side of Eq. 6.59, and the convective term, on the left hand side of
Eq. 6.59, was discretised using the CUBISTA high-resolution scheme of Alves
et al. [3], which preserves boundedness, even for highly advective flows.
The discretised equation resulting from the finite-difference approximation of
Eq. 6.59 reads:
Ciþ1=2; j Ci1=2; j Ciþ1; j ðDZi Þ Ci; j ðDZi þ DZiþ1 Þ þ Ci1; j ðDZiþ1 Þ
pe0sf ¼
ðDZi þ DZiþ1 Þ=2 DZi DZiþ1 ðDZi þ DZiþ1 Þ=2

Ci; jþ1 DYj Ci; j DYj þ DYjþ1 þ Ci; j1 DYjþ1
þ
DYj DYjþ1 DYj þ DYjþ1 =2
ð6:61Þ
The Ciþ1=2;j and Ci1=2;j values are interpolated from the known grid node
values using the CUBISTA high-resolution scheme (HRS), in order to ensure
numerical stability and good precision. The normalised variable approach (NVA)
of Leonard [34] was adopted for implementations of the HRS, in which a general
differencing scheme up to third order of accuracy can be expressed as
Ciþ1=2;j ¼ f ðCi1;j ; Ci;j ; Ciþ1;j Þ ð6:62Þ
The NVA uses an appropriate upwind biased normalisation, and Eq. 6.62 can
be rewritten in compact form as
_ _
C iþ1=2;j ¼ f ðC i;j Þ ð6:63Þ
where
_ Ck;j Ci1;j
Ck;j ¼ ðfor k ¼ i 1; i; i þ 1=2; i þ 1Þ ð6:64Þ
Ciþ1;j Ci1;j
_ _
since, by definition of Eq. 6.64, C i1;j ¼ 0 and C iþ1;j ¼ 1:
The CUBISTA scheme is represented in the context of the NVA by the fol-
lowing piecewise linear functions [3]:
8 _
> 7 _ 3
>
> C 0 C i;j \
> 4 i;j
> 8
>
>
<3C _ 3 3 _ 3
_ i;j þ Ci;j
Ciþ1=2;j ¼ 4 8 8 4 ð6:65Þ
>
> 1 _ 3 3 _
>
> C i;j þ \C i;j 1
>
> 4 4 4
>
: _
Ci;j elsewhere
The resulting system of equations was solved iteratively using the successive
over-relaxation (SOR) method [20], and the implementation of the boundary
conditions was carried out in the same way as described in our previous work [11].
For the situation under study an orthogonal mesh is adequate and care was taken to
ensure proper refinement in the regions where high concentration gradients are
expected. The computational domain was defined according to the flow conditions
(typically for small Pe0sf ; longer meshes are needed) and during mesh refinement,
the number of nodes along each direction was doubled, thus halving the mesh sizes
along each direction. This systematic procedure allows direct use of Richardson’s
extrapolation technique in order to obtain very accurate results [20].
6.5.1.2 Numerical Results
The numerical solution of Eq. 6.59 gives the concentration field and from it, the
rate of dissolution of the slab, n, is obtained integrating the diffusion/dispersion
flux over the whole slab surface. This integral is evaluated numerically, for each
100
Numerical solution
Eq. (6.68)
10
Shsf'/ε
0.1
0.001 0.01 0.1 1 10 100 1000 10000
Pesf'
Fig. 6.13 Dependence of Sh0sf =e on Pe0sf when DT ¼ DL ¼ D0m throughout, for slab flat
set of conditions, from the discretised concentration field that is obtained through
the numerical solution of Eq. 6.59. It is convenient to express the rate of disso-
lution in terms of a Sherwood number, Sh0sf ¼ kL=D0m ; where k ¼ n=½Sðc c0 Þ is
the mass transfer coefficient for the soluble slab and A is the exposed area of the
soluble solid.
In the analysis of the results of the numerical computations we only consider
the situation of low Pe0p ð¼ ud=D0m Þ values, diffusional regime.
For low Pe0p values, dispersion is the direct result of molecular diffusion and
DT ffi DL ffi D0m : Fig. 6.13 presents the results obtained numerically, for the situ-
ation of dispersion dominated by molecular diffusion, and two asymptotes can be
observed:
1=2
Sh0sf 2 0 1=8
! Pesf for Pe0sf ! 0 ð6:66Þ
e p
and
1=2
Sh0sf 4 0 1=2
! Pesf for Pe0sf ! 1 ð6:67Þ
e p
Taking the geometric mean of those two asymptotes

Sh0sf 2 0 1=4 4 0 1=2
¼ Pe þ Pesf ð6:68Þ
e p sf p
one observes that it does not deviate more than 3% from the numerical values.

packed bed near soluble
cylinder aligned with flow
6.5.2 Mass Transfer From a Soluble a Cylinder Aligned

With Flow
Figure 6.14 sketches a dissolving cylinder in a packed bed, aligned with the flow
direction. In this case the mass transfer boundary layer forms around the cylinder
and has axial symmetry.
Considering a small control-volume inside the boundary layer and performing a
steady-state material balance on the solute, one obtains
2
0 @C @2C L 1 @ @C
Pec ¼ þ4 R ð6:69Þ
@Z @Z 2 d1 R @R @R
where R ¼ 2r=dc ; and dc represents the diameter of the soluble cylinder. Equa-
tion 6.62 was solved numerically with boundary conditions
C!0 Z ! 1 R1 ð6:70aÞ
C=1 0Z1 R=1 ð6:70bÞ
@C
¼0 Z \0 _ Z [ 1 R=1 ð670cÞ
@R
C!0 8Z R ! þ1 ð6:70dÞ
C!0 Z ! þ1 R1 ð6:70eÞ
over the ranges of Pe0c ð¼ uL=D0m Þ and L=dc of practical interest.

100
10 1/ 2 3/ 2
Sh' L 4 L
6
0 d1 3 d1
(Sh' / )0 Sh' 4 L
3/ 4
3 d1
1 1/ 4
Sh' L
6
d1
numerical solution
asymptote for high L /d 1
asymptote for low L /d 1
0.1
0.01 0.1 1 10 100
L /d 1
Fig. 6.15 Dependence of ðSh0 =eÞ0 on L=d1
1000
L /d c =0 L /d c =0.01 L /d c =0.1
L /d c =0.5 L /d c =1 L /d c =2
L /d c =5 L /d c =10 Eq. (6.73)
100
Shc'/ε
10
0.1
0.001 0.01 0.1 1 10 100 1000 10000
Pec'
Fig. 6.16 Dependence of Sh0c =e on Pe0c for different values of L=dc

For low Pe0p values ðsay, Pe0p \0:1Þ; the values of Sh0c =e ¼ kL=D0m e were obtained
by integrating the flux along the soluble surface, in a similar way to that described
for the flat surface. The first point to be mentioned is that the lower set of points in
Fig. 6.16 (those for L=dc ¼ 0) were not obtained by solving Eq. 6.69 directly.
Indeed, for L=dc ! 0 (with finite L) the physical situation depicted in Fig. 6.15
degenerates into mass transfer from a flat surface. The plot for L=dc ¼ 0 reveals
the two asymptotes obtained for the dissolution of a slab in a packed bed (see
Eqs. 6.66 and 6.67. Equation 6.67 is the result obtained by Coelho and Guedes de

packed bed near soluble 0
sphere
d1 = 2R
u r
u ur
0’
u0
Carvalho [12] who argued that it would be also the asymptote for L=dc [ 0: This is
shown to be the case in Fig. 6.16 and it is not surprising, since for thin concen-
tration boundary layer, the curvature of the cylinder is not relevant.
For each value of L=dc [ 0; the plot of Sh0c =e vs. Pe0c has a horizontal
0
asymptote, for Pec ! 0: This limiting value of Sh0c =e (for each value of L=dc ) may

be conveniently represented by Sh0c =e 0 : to emphasize that it corresponds to the
situation of pure molecular
diffusion with no flow. The numerical results indicated
a dependence of Sh0c =e 0 on L=dc and again, two asymptotes are revealed:
1=4
Sh0c L L
! 6 for low ð6:71Þ
e 0 dc dc
and
3=4
Sh0c 4L L
! for high ð6:72Þ
e 0 3 dc dc
Having disclosed the above asymptotes, a general approximate expression for
Sh0c =e; that has the correct asymptotic behaviour, was found to be [4]
2 31=2
2 0 1=4 4 0 L 1=2 4 L 3=2
pe þ pec þ 6 þ
sh0c 6
6 p c p dc 3 dc
7
7
¼6 1=3 7 ð6:73Þ
e 4 5 0 5=9 L L 5
þ pec pe0c 2=9 2
3 dc dc
This equation does not deviate more than 2% from the numerical values, over the
entire range of values of Pe0c and L=dc x represented in Fig. 6.16. The approximate
conditions of validity of these equation are: dc =d 50 and L=dc 100:
6.5.3 Mass Transfer From a Soluble Sphere
Consider a soluble sphere of diameter d1 ðd1 ¼ 2aÞ; buried in a bed of inert particles
of diameter d (with d
d1 ), packed uniformly (void fraction e) around the sphere
(see Fig. 6.17). The packed bed is assumed to be ‘‘infinite’’ in extent and a uniform
interstitial velocity of liquid, u0 ; is imposed, at a large distance from the sphere.
Darcy’s law, u ¼ K grad p; is assumed to hold and if it is coupled with the
continuity relation for an incompressible fluid, div u ¼ 0; Laplace’s equation
r2 / ¼ 0 is obtained for the flow potential, / ¼ K p; around the sphere. In terms
of spherical coordinates (r, h), the potential and stream functions are, respectively
[14],

@c @ 0 @c @ 0 2 @c
¼ Dm þ Dm x ð6:74Þ
@/ @/ @/ @w @w
where x is the distance to the flow axis ð¼ r sin hÞ.

To integrate Eq. 6.74, with the auxiliary Eqs. 6.75 and 6.76, it is convenient to
define the following dimensionless variables:

1 a
3
/ ¼ u0 1 þ r cos h ð6:75Þ
2 r
a
3
u0
w¼ 1 r 2 sin2 h ð6:76Þ
2 r
and the velocity components are

a
3
@/
ur ¼ ¼ u0 cos h 1 ð6:77Þ
@r r

1 @/ 1 a
3
uh ¼ ¼ u0 sin h 1 þ ð6:78Þ
r @h 2 r
The analysis of mass transfer is based on a steady state material balance on the
solute crossing the borders of an elementary volume, limited by the potential
surfaces / and / þ d/; and the stream surfaces w and w þ dw: The resulting
equation is [4],
c c0
C¼ ð6:79aÞ
c c0
r
<¼ ð6:79bÞ
a
2 1=2
u ur þ u2h
U¼ ¼ ð6:79cÞ
u0 u0
4 /
U¼ ð6:79dÞ
3 u0 d1
w
W¼ ð6:79eÞ
u0 d12
Equation 6.78 may be re-arranged to

@C @ 4 @C @ 3 <2 sin2 h @C
¼ þ ð6:80Þ
@U @U 3Pe0s @U @w 16 Pe0s @w
The boundary conditions to be observed in the integration of Eq. 6.74 are: (1)
the solute concentration is equal to the background concentration, c0 ; far away
from the sphere; (2) the solute concentration is equal to the equilibrium concen-
tration, c = c*, on the surface of the sphere and (3) the concentration field is
symmetric about the flow axis:
U ! 1; W 0 C ! 0 ð6:81aÞ
U ! þ1; W 0 C ! 0 ð6:81bÞ
(
1 / 1 C ¼ 1
W¼0 @C ð6:81c; dÞ
j/j [ 1 ¼0
@W
W ! þ1; all U C!0 ð6:81eÞ
Equation 6.80 was solved numerically, subjected to the boundary conditions
(6.81a–e), over the ranges of Pe0s ð¼ u0 d1 =D0m Þ of practical interest.
For low values of Pe0p ; dispersion is the direct result of molecular diffusion and the
numerical solution obtained by Guedes de Carvalho and Alves [23] applies. Those
authors showed that the numerical results are well described (with an error of less
than 1% in Sh0s ¼ kd1 =D0m ) by (see Fig. 6.18)

Sh0s 4 2=3 4 0 1=2
¼ 4 þ Pe0s þ pes ð6:82Þ
e 5 p
100
Numerical solution
Eq. (6.82)
Shs'/ε
10
1
0.001 0.01 0.1 1 10 100 1000 10000
Pes'
Fig. 6.18 Dependence of Sh0s =e on Pe0s for soluble sphere

packed bed near soluble 0
cylinder in cross flow
d1 = 2R
u r
u ur
0’
u0
6.5.4 Mass Transfer From a Cylinder in Cross Flow
Consider a slightly soluble cylinder, of diameter d1 ; buried in a packed bed of inert

spherical particles of diameter d ðd
d1 Þ and exposed to fluid flow perpendicular
to its axis, with uniform interstitial velocity, u0 ; at a large distance from the
cylinder (see Fig. 6.19).
Assuming Darcy’s law, Laplace’s equation is obtained for the flow potential
around the cylinder. For a sufficiently long cylinder (assumed to be of ‘‘infinite’’
length) the flow field is bi-dimensional and in polar coordinates, the potential and
stream functions are expressed as
a
2
/ ¼ u0 1 þ r cos h ð6:83Þ
r
a
2
w ¼ u0 1 r sin h ð6:84Þ
r
and the velocity components are given by

a
2
@/
ur ¼ ¼ u0 cos h 1 ð6:85Þ
@r r
a
2
1 @/
uh ¼ ¼ u0 sin h 1 þ ð6:86Þ
r @h r
To formulate the mass transfer problem we take the concentration of the dif-
fusing species to be c* on the surface of the cylinder and c0 at a large distance from
it, in the approaching stream. The resulting concentration field will have axial
symmetry and the differential equation for mass transfer may be derived from a
mass balance on the elementary volume.
The analysis of mass transfer is based on a steady state material balance on the
solute crossing the borders of an elementary volume, limited by the potential
surfaces / and /þd/; and the stream surfaces w and w þ dw: The material mass
balance on the solute may be expressed as

@c @ @c @ @c
¼ D0m þ D0m ð6:87Þ
@/ @/ @/ @w @w
The assumption of steady state is acceptable for a solid that is taken to be
slightly soluble and the use of the dispersion coefficients makes sense if the
boundary layer extends over several particle diameters.
To integrate Eq. 6.87, with the auxiliary Eqs. 6.83 and 6.84, it is convenient to
define the dimensionless variables:
c c0
C¼ ð6:88aÞ
c c0
r
<¼ ð6:88bÞ
R
2 1=2
u ur þ u2h
U¼ ¼ ð6:88cÞ
u0 u0
/
U¼ ð6:88dÞ
u0 dc
100
Numerical solution
Eq. (6.91)
10
Shcf'/ε
0.1
0.001 0.01 0.1 1 10 100 1000 10000
Pecf'
Fig. 6.20 Dependence of Sh0cf =e on Pe0cf ; for a soluble cylinder in cross flow
w
W¼ ð6:88eÞ
u0 dc
u0 dc
Pe0cf ¼ ð6:88fÞ
D0m
where Pe0cf represents the Peclet number based on the diameter of the soluble
cylinder, dc : Equation 6.87 may then be rewritten as

@C @ 1 @C @ 1 @C
¼ þ ð6:89Þ
@U @U Pe0cf @U @w Pe0cf @w
The boundary conditions to be observed in the integration of Eq. 6.89 are: (1)
the solute concentration is equal to the background concentration, c0 ; far from the
cylinder; (2) the solute concentration is equal to the equilibrium concentration,
c ¼ c ; on the surface of the cylinder and (3) the concentration field is symmetric
about w ¼ 0:
U ! 1; W 0 C!0 ð6:90aÞ
U ! þ1; W 0 C ! 0 ð6:90bÞ
(
1 U 1 C ¼ 1
W¼0 @C ð6:90c; dÞ
jUj [ 1 ¼0
@W
W ! þ1; all U C!0 ð6:90eÞ
Equation 6.89 has to be solved numerically and the method developed and
described in detail by Alves et al. [3] was adapted, to obtain the solute concen-
tration field around the dissolving cylinder.
Fig. 6.21 The prolate z

spheroidal coordinates
system
L’ = const.
c
x
For very low Pe0p ; dispersion is the direct result of molecular diffusion, with
DT ffi DL ffi D0m ; and Sherwood number is only function of Pe0 : Numerical solu-
tions were worked out for the range of Pe0 between 10-3 and 104. The numerical
values of Sh0 =e obtained are shown as dots in the plot of Fig. 6.20.
An improved approximation was found to describe the values of Sh0cf =e obtained
numerically, Eq. 6.91, which is represented as a full line in the same figure:

Sh0cf 2 0 1=4 32 0 1=2
¼ 2 Pecf þ 3 Pecf ð6:91Þ
e p p
and the values of Sh0cf =e calculated from this equation differ at most by 2% from
the corresponding numerical solution obtained in the present work.
6.5.5 Mass Transfer From a Prolate Spheroid
In many practical situations it is often required to consider operations in which

there are physico-chemical interactions between a solid particle and the fluid
flowing around it. In the treatment of these operations it is common practice to
assume the soluble particle to be spherical, because the treatment of irregular
shapes could only be done by numerical methods.
The surface of the prolate spheroid ðh ¼ h0 Þ is described by (see Fig. 6.21)
x2 þ y2 z2
þ 2¼1 ð6:92Þ
a2 c
Since r2 ¼ x2 þ y2 ; Eq. 6.92 can be written as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
z ¼ c 1 ðr=aÞ2 ð6:93Þ
The surface area S and volume V of a prolate spheroid are given by

!
c=a pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 1
S ¼ 2pa 1 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sin 1 a =c 2 2 ð6:94Þ
1 a2 =c2
4
V ¼ pa2 c ð6:95Þ
3
where e ¼ 1 a2 =c2 is the ellipticity, and where e ¼ 0 corresponds to a sphere.
The dimensional Cartesian coordinates (x, y, z) are related to the prolate spheroidal
ones (h, g, b) through the equations [37]
x ¼ L0 sinh h sin g cos b ð6:96aÞ
y ¼ L0 sinh h sin g sin b ð6:96bÞ
z ¼ L0 cosh h cos g ð6:96cÞ

where L0 is the focal distance L0 ¼ c2 a2 and the coordinates range are:

0 h\1; 0 g p and 0 b 2p:
6.5.5.1 Analytical Results
For high Peclet numbers, with DT ffi DL ffi D0m ; our the theory is based on the
assumption that the inert particles in the bed are packed with uniform voidage,
e; and that the flow may be approximated everywhere by Darcy’s law,
u ¼ K grad p: Furthermore, if the fluid is treated as incompressible, mass con-
servation leads to div u ¼ 0; Laplace’s equation is obtained r2 / ¼ 0.
Darcy’s law is strictly valid only for laminar flow through the packing, but
according to Bear [7] it is still a good approximation for values of the Reynolds
number (based on superficial velocity) up to 10; which for beds with e 0:4 is
equivalent to Re 25; the upper limit for the validity of this analysis.
When a solid prolate spheroid is immersed in a packed bed of significantly
smaller particles, through which fluid flows with uniform interstitial velocity u0 ;
far from the spheroid, the solution of Laplace’s equation and the corresponding
stream function, in terms of spheroidal coordinates (h, g, b), are [2]
2 3
6 cosh h coth1 ðcosh hÞ 1 7
/ ¼ u0 L0 cos g6
4cosh h
7 ð6:97Þ
1 cosh h0 5
coth ðcosh h0 Þ
sinh2 h0
2 3
L2 6 sinhhcoth1 ðcoshhÞ cothh7
w ¼ u0 sinhh6
4 sinhh cosð2gÞsinhh ð1 cosð2gÞÞ 7
4 coshh0 5
coth1 ðcoshh0 Þ
sinh2 h0
ð6:98Þ
The stream and potential functions are related to the dimensionless velocity
components ðuh ; ug Þ by Batchelor, [5]
1 @/ 1 @w
uh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:99Þ
L0 sinh h þ sin g @h L02 sinh h þ sin g sinh h sin g @g
2 2 2 2
1 @/ 1 @w
ug ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:100Þ
L0 sinh h þ sin g @g L02 sinh h þ sin g sinh h sin g @h
2 2 2 2
resulting the following velocity components

2 3
u0 cos g 6 sinh h coth1 ðcosh hÞ cothðhÞ7
uh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 6
4sinh h 7 ð6:101Þ
sinh2 h þ sin2 g 1 cosh h0 5
coth ðcosh h0 Þ
sinh2 h0
2 3
u0 sin g 6 cosh h coth1 ðcosh hÞ 1 7
ug ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 6
4cosh h 7 ð6:102Þ
sinh2 h þ sin2 g 1 cosh h0 5
coth ðcosh h0 Þ
sinh2 h0
The tangential velocity at the surface of the prolate spheroid ðh ¼ h0 Þ can be
found by

1 @/
ug0 ¼ p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:103Þ
L0 sinh2 h þ sin2 g @g h¼h0
and the resulting expression is
u0 sin g
ug0 ¼ 0:5 ð6:104Þ
1=e2 1 þ sin2 g 1=e ð1=e2 1Þ tanh1 e
Figure 6.22 shows the adimensional tangential surface velocity, ug0 =u0 ; of a
prolate spheroid as a function of g, for different values of the eccentricity, e. Note
that, for the case of a sphere, e 0; the well-known result of ug0 ¼ 1:5u0 sin g is
2.0
e= 0
e = 0.5
1.6 e = 0.7
e = 0.9
e = 0.99
1.2
u η0 /u 0
0.8
0.4
0.0
0 π/8 π/4 3π/8 π/2 5π/8 3π/4 7π/8 π
η
Fig. 6.22 The adimensional tangential surface velocity, ug0 =u0 ; as a function of g, for different
values of the eccentricity, e
obtained, for potential flow over the surface of the sphere. On the other hand, for a
slender prolate, e ! 1; as expected ug0 =u0 ! 1.
A convenient way of expressing the differential mass balance on the solute is to
take a control volume along a stream tube, between two nearby potential surfaces.
The resulting expression, for convection with molecular diffusion, is

@C @ 0 @C @ 0 2 @C
¼ Dm þ Dm x ð6:105Þ
@/ @/ @/ @w @w
For high values of the Peclet number the concentration boundary layer will be
thin and the first term on the r.h.s. of Eq. 6.105 may be neglected. After some
algebraic manipulation and a suitable change of variables, it is then possible to
obtain
@C @ 2 C
¼ ð6:106Þ
@n @w2
where n is defined by
Zg
n¼ L03 sinh2 h0 sin2 g0 ðsinh2 h0 þ sin2 g0 Þ0:5 ug D0m dg0 ð6:107Þ
0
with x ¼ L0 sinh h0 sin g: The boundary conditions for Eq. 6.105, in our problem,
are
C ¼ C0 n ¼ 0 w[0 ð6:108aÞ
C ¼ C n[0 w¼0 ð6:108bÞ
C ! C0 n[0 w!1 ð6:108cÞ

and the corresponding solution is

C C0 w
¼ 1 erf pffiffiffi ð6:109Þ
C C0 2 n
The value of n varies over the surface of the spheroid. Now, for potential flow
ug is given by Eq. 6.104 over the surface of the spheroid ðh ¼ h0 Þ and the integral
in (6.107) is

c3 e3 ð1 e2 Þ 2 1
n ¼ u0 D0m cos g þ cos 3
g ð6:110Þ
e ð1 e2 Þ tanh1 e 3 3
The flux of solute at any point on the surface of the spheroid is

@C D0m e @C
N ¼ D0m e ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
@b L sinh h0 þ sin g @h h¼h0
0 2 2
ð6:111Þ
0 0 @C
¼ Dm e ug L sinh h0 sin g
@w w¼0
pffiffiffiffiffiffiffiffiffi
and from (6.104) it may be shown that ð@C=@wÞw¼0 ¼ 1= ðpnÞ ðC C0 Þ.
The rate of dissolution of the spheroid in the region 0\g\g1 will then be
Zg1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
nðg1 Þ ¼ N2pL02 sinh2 h0 þ sin2 g sinh h0 sin g d g
0
Zn
pffiffiffiffiffi pffiffiffi
¼ 2peðC C0 Þ ð1= pzÞdz ¼ 4 p e n1=2 ðC C0 Þ ð6:112Þ
0
with nðg1 Þgiven by (6.110). In particular, the total rate of dissolution of the
spheroid, nT ; may be obtained taking g1 ¼ p: By definition, the average mass
transfer coefficient, k, is
k ¼ nT = ½S ðC C0 Þ ð6:113Þ
the resulting expression for k (from (6.112) and (6.113)) is
pffiffiffi 1=2
4e p 4 c3 ð1 e2 Þe3 u0 D0m
k¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ! ð6:114Þ
1=ð1 e2 Þ 1 3 e ð1 e2 Þ tanh1 e
2pa2 1þ sin ðeÞ
e
Fig. 6.23 The oblate
θ = const.
spheroidal coordinates
system
a
L’
z
x
c
It is convenient to express the rate of dissolution in terms of the Sherwood
number, Sh0ps ¼ kdeq =D0m ; with deq ¼ 2ða2 cÞ1=3 ; and the expression obtained, after
some algebraic manipulation is
rffiffiffiffiffiffiffiffiffiffiffiffi 1=2
sh0ps 4 0 2 e3 2
¼ Peps 1 2 1=6
e p 2
3 e ð1 e Þ tanh e ð1 e2 Þ þ ð1e Þ sin1 e
1=3
e
ð6:115Þ
where Pe0ps ¼ u0 deq =D0m is the Peclet number. It is expected that Eq. 6.115 does
not differ by more than 10% from the exact solution to obtain numerically.
For the special case of a sphere, e 0; the result of Sh0s =e ¼ ½4 Pe0 =p1=2 is
obtained, which corresponds to the asymptotic behaviour for thin concentration
layer (high values of Peclet number) when dispersion is constant and tend to D0m
over the surface of the sphere.
6.5.6 Mass Transfer From an Oblate Spheroid
The surface area S of an oblate spheroid is given by

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!!
2 ðc=aÞ2 1 þ 1 c2 =a2
S ¼ 2pa 1 þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ln pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:116Þ
2 1 c2 =a2 1 1 c2 =a2
where e ¼ 1 c2 =a2 is the ellipticity, and for e ¼ 0 corresponds to a sphere.
Figure 6.23 describes the oblate spheroidal coordinate system.
The dimensional Cartesian coordinates (x, y, z) are related to the oblate sphe-
roidal ones (h, g, b) through the equations
x ¼ L0 cosh h sin g cos b ð6:117aÞ
y ¼ L0 cosh h sin g sin b ð6:117bÞ

z ¼ L0 sinh h cos g ð6:117cÞ

where L0 ¼ a2 c2 is the focal distance and the coordinates are
0 h\1; 0 g p and 0 b 2p:
6.5.6.1 Analytical Results
For high values of Peclet numbers, the theory used is based on the assumption that
the inert particles in the bed are packed with uniform voidage, e; and that the fluid
flow may be approximated everywhere by Darcy’s law, u ¼ K grad p; as in the
case of a prolate spheroid.
When a solid oblate spheroid is immersed in a packed bed of significantly
smaller particles, through which fluid flows with uniform interstitial velocity u0 ;
far from the spheroid, the solution of Laplace’s equation and the corresponding
stream function, in terms of spheroidal coordinates (h, g, b), are [2]
2 3
6 sinh h cot1 ðsinh hÞ 1 7
/ ¼ u0 L0 cos g6
4sinh h
7 ð6:118Þ
1
sinh h0 5
cot ðsinh h0 Þ
cosh2 h0
2 3
cosh h cosð2gÞ cosh h
L2 6 1 7
6 7
w ¼ u0 cosh h6 ð1 cosð2gÞÞ cosh h cot ðsinh hÞ tanh h 7 ð6:119Þ
4 4 sinh h 0 5
cot1 ðsinh h0 Þ 2
cosh h0
The stream and potential functions are related to the dimensionless velocity
components and resulting the following velocity components
2 3
u0 cos g 6 cosh h cot1 ðsinh hÞ tanhðhÞ7
uh ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 6
4 cosh h 7 ð6:120Þ
cosh2 h sin2 g 1
sinh h0 5
cot ðsinh h0 Þ
cosh2 h0
2 3
u0 sin g 6 sinh h cot1 ðsinh hÞ 1 7
ffi 6sinh h
ug ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 7 ð6:121Þ
2 2 4 sinh h0 5
cosh h sin g cot1 ðsinh h0 Þ
cosh2 h0
The tangential velocity at the surface of the oblate spheroid ðh ¼ h0 Þ can be
found by
6
e=0
e = 0.5
5
e = 0.7
e = 0.9
4 e = 0.99
u η 0 /u 0
0
0 π/8 π/4 3π/8 π/2 5π/8 3π/4 7π/8 π
η
Fig. 6.24 The adimensional tangential surface velocity, ug0 =u0 ; as a function of g, for different
values of the eccentricity, e

1 @/
ug0 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
L0 cosh2 h sin2 g @g h¼h0
u0 sin g ð6:122Þ
¼ 0:5 " rffiffiffiffiffiffiffiffiffiffiffiffi2ffi rffiffiffiffiffiffiffiffiffiffiffiffiffi!#
1 2 1e 1 e2
2
sin g 2
þ e12 cot1
e e e2
Figure 6.24 shows the adimensional tangential surface velocity, ug0 =u0 ; of a
prolate spheroid as a function of g, for different values of the eccentricity, e.
Note that, for the case of a sphere, e 0; the well-known result of ug0 ¼
1:5u0 sin g is obtained, for potential flow over the surface of the sphere. On the
other hand, for a slender oblate, e ! 1; as expected ug0 =u0 ! 1:
A convenient way of expressing the differential mass balance on the solute is to
take a control volume along a stream tube, between two nearby potential surfaces.
The resulting expression, for convection with molecular diffusion, is given by
Eq. 6.99, where n is defined by
Zg
n¼ L03 cosh2 h0 sin2 g0 ðcosh2 h0 sin2 g0 Þ1=2 ug D0m d g0 ð6:123Þ
0
0
with x ¼ L cosh h0 sin g. The boundary conditions for are given by Eqs. 6.108a–c
and the corresponding solution is

C C0 w
¼ 1 erf pffiffiffi ð6:124Þ
C C0 2 n
The value of n varies over the surface of the spheroid. Now, for potential flow
ug is given by Eq. 6.115 over the surface of the spheroid ðh ¼ h0 Þ and the integral
in (6.125) is

0 e3 c3 =ð1 e2 Þ 2 1 3
n ¼ u0 Dm pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi cos g þ cos g
e þ e3 þ 1 e2 cot1 ð 1=e2 1Þ 3 3
ð6:125Þ
The flux of solute at any point on the surface of the spheroid is

D0m e @C @C
ffi ¼ D0m e ug L0 cosh h0 sin g
N ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 2
L cosh h0 sin g 2 @w w¼0 @h h¼h0
ð6:126Þ
pffiffiffiffiffiffiffiffiffi
and from (6.126) it may be shown that ð@C=@wÞw¼0 ¼ 1= ðpnÞ ðC C0 Þ.
The rate of dissolution of the spheroid in the region 0\g\g1 will then be
Zg1 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
02
pffiffiffi
nðg1 Þ ¼ N2p L cosh2 h0 sin2 g cosh h0 sin g d g ¼ 4 p e n1=2 ðC C0 Þ
0
ð6:127Þ
with nðg1 Þgiven by (6.125). In particular, the total rate of dissolution of the
spheroid, nT ; may be obtained taking g1 ¼ p: By definition, the average mass
transfer coefficient, k, is given by Eq. 6.106 and the resulting expression for k is
pffiffiffi !1=2
4e p 2 e3 u0 D0m c3 =ð1 e2 Þ
k¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 e2 1þe 3 e þ e3 þ 1 e2 cot1 ð 1=e2 1Þ
2pa2 1 þ ln
2e 1e
ð6:128Þ
It is convenient to express the rate of dissolution in terms of the Sherwood
number, Sh0os ¼ kdeq =D0m ; with deq ¼ 2ða2 cÞ1=3 ; and the expression obtained is
rffiffiffiffiffiffiffiffiffiffiffiffi !1=2
Sh0os 4 0 4 e3 ð1 e2 Þ1=3
¼ Pe pffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
e p os 3 e þ e3 þ 1 e2 cot1 ð 1=e2 1Þ 1 e2 1þe
1þ ln
2e 1e
ð6:129Þ
where Pe0os ¼ u0 deq =D0m is the Peclet number.

References 171
References
1. Akehata, T., Sato, K.: Flow distribution in packed beds. Chem. Eng. Jpn. 22, 430–436 (1958)
2. Alassar, R.S., Badr, H.M.: Analytical solution of oscillating inviscid flow over oblate
spheroids with spheres and flat disks as special cases. Ocean Eng. 24, 217–225 (1997)
3. Alves, M.A., Oliveira, P.J., Pinho, F.T.: A convergent and universally bounded interpolation
scheme for the treatment of advection. Int. J. Num. Meth. Fluids 41, 47–75 (2003)
4. Alves, M.A., Delgado, J.M.P.Q., Guedes de Carvalho, J.R.F.: Mass transfer from cylinders
and plane surfaces buried in packed beds in alignment with the flow direction. Chem. Eng.
Sci. 61, 1174–1183 (2006)
5. Batchelor, G.K.: An Introduction to Fluid Dynamics. Cambridge University Press,
Cambridge (1997)
6. Baetsle, L.H.: Migration of radionuclides in porous media. In: Duhamed Pergamon, A.M.F.
(ed.) Progress in Nuclear Energy, Series XII, pp. 707–730. Health Physics, Elmsford (1969)
7. Bear, J.: Dynamics of Fluids in Porous Media. Dover Publications, New York (1972)
8. Blackwell, R.J.: Laboratory studies of microscopic dispersion phenomena. Soc. Petrol. Eng. J.
225, 1–8 (1962)
9. Bruinzeel, C, Reman, GH, van der Laan, ETh Axial dispersion. Proceedings of the 3th
Congress of the European Federation of Chemical Engineering, Olympia (1962)
10. Carslaw, H.S., Jaeger, J.C.: Conduction of Heat in Solids, 2nd edn. Oxford University Press,
Oxford (1959)
11. Carvalho, J.R.F.G., Delgado, J.M.P.Q., Alves, M.A.: Mass transfer between flowing fluid and
sphere buried in packed bed of inerts. AIChE J. 50, 65–74 (2004)
12. Coelho, M.A.N., Guedes de Carvalho, J.R.F.: Transverse dispersion in granular beds. Part I:
Mass transfer from a wall and the dispersion coefficient in packed beds. Chem. Eng. Res.
Des. 66, 165–177 (1988)
13. Crank, J.: The Mathematics of Diffusion, 2nd edn. Oxford University Press, Oxford (1975)
14. Currie, I.G.: Fundamental Mechanics of Fluids. McGraw-Hill, New York (1993)
15. Danckwerts, P.V.: Continuous flow systems. Chem. Eng. Sci. 2, 1–13 (1953)
16. Dorweiler, V.P., Fahien, R.W.: Mass transfer at low flow rates in a packed column. AIChE J.
5, 139–144 (1959)
17. Dullien, F.A.L.: Single phase flow through porous media and pore structure. Chem. Eng.
J. 10, 1–34 (1975)
18. Dullien, F.A.L.: Porous Media: Fluid Transport and Pore Structure. Academic, San Diego
(1979)
19. Fahien, R.W., Smith, J.M.: Mass transfer in packed beds. AIChE J. 1, 28–37 (1955)
20. Ferziger, J.H., Peric, M.: Computational Methods for Fluid Dynamics. Springer, Berlin
(1996)
21. Fuller, E.N., Schettler, P.D., Giddings, J.C.: A new method for prediction of binary gas-phase
diffusion coefficients. Ind. Eng. Chem. 58, 19–27 (1966)
22. Gibbs, S.J., Lightfoot, E.N., Root, T.W.: Protein diffusion in porous gel filtration
chromatography media studied by pulsed field gradient NMR spectroscopy. J. Phys. Chem.
96, 7458–7462 (1992)
23. Guedes de Carvalho, J.R.F., Alves, M.A.M.: Mass transfer and dispersion around active
sphere buried in a packed bed. AIChE J. 45, 2495–2502 (1999)
24. Guedes de Carvalho, J.R.F., Delgado, J.M.P.Q.: Lateral dispersion in liquid flow through
packed beds at Pem \ 1400. AIChE J. 46, 1089–1095 (2000)
25. Gunn, D.J.: Mixing in packed and fluidised beds. Chem. Eng. 219, CE153–CE172 (1968)
26. Gunn, D.J., Pryce, C.: Dispersion in packed beds. Trans. IChemE 47, T341–T350 (1969)
27. Han, N.W., Bhakta, J., Carbonell, R.G.: Longitudinal and lateral dispersion in packed beds:
effect of column length and particle size distribution. AIChE J. 31, 277–288 (1985)
28. Harleman, D.R.F., Rumer, R.: Longitudinal and lateral dispersion in an isotropic porous
medium. J. Fluid Mech. 16, 1–12 (1963)
29. Harrison, D., Lane, M., Walne, D.J.: Axial dispersion of liquid on a column of spheres. Trans.
IChemE 40, 214–220 (1962)
30. Hiby, J.W., Schummer, P.: Zur messung der transversalen effektiven diffusion in
durchstromten fullkorpersaulen. Chem. Eng. Sci. 13, 69–74 (1960)
31. Higbie, S.: The rate of absorption of a pure gas into a still liquid during short periods of
exposure. Trans. AIChE J. 31, 365–389 (1935)
32. Hunt, B.: Dispersive sources in uniform groundwater flow. J. Hydraul. Div.–Proc. Am. Soc.
Civil Eng. 104, 75–85 (1978)
33. Klinkenberg, A., Krajenbrink, H.J., Lauwerier, H.A.: Diffusion in a fluid moving at uniform
velocity in a tube. Ind. Eng. Chem. 45, 1202–1208 (1953)
34. Leonard, B.P.: Simple high-accuracy resolution program for convective modelling of
discontinuities. Int. J. Num. Meth. Fluids 8, 1291–1318 (1988)
35. Marrero, T.R., Mason, E.A.: Gaseous diffusion coefficients. J. Phys. Chem. Ref. Data 1, 3–25
(1972)
36. Mair, R.W., Hurlimann, M.D., Sen, P.N., Schwartz, L.M., Patz, S., Walsworth, R.L.:
Tortuosity measurement and the effects of finite pulse widths on xenon gas diffusion NMR
studies of porous media. Mag. Reson. Imag. 193, 345–351 (2001)
37. Moon, P., Spencer, D.E.: Field theory handbook (2nd ed). Springer, Berlin (1971)
38. National Research Council: International critical tables–Volume 5. McGraw-Hill
International Editions, New York (1929)
39. Ogata, A., Banks, R.B.: A solution of differential equation of longitudinal dispersion in
porous media. US Geolog. Surv. Prof. Pap. 411-A:7–12 (1961)
40. Perkins, T.K., Johnston, O.C.: A review of diffusion and dispersion in porous media. Soc.
Petrol. Eng. J. 2, 70–84 (1963)
41. Perry, R.H., Green, D.: Perry’s chemical engineer’s handbook (6th ed). McGraw-Hill
International Editions, New York (1984)
42. Pinto, A.M.F.R., Guedes de Carvalho, J.R.F.: Transverse dispersion in granular beds. Part III:
Mass transfer around particles dispersed in granular beds of inerts and the combustion of
carbon particles in beds of sand. Trans. IChemE 68, 503–509 (1980)
43. Plautz, D.A., Johnstone, H.F.: Heat and mass transfer in packed beds. AIChE J. 1, 193–199
(1955)
44. Reid, R.C., Prausnitz, J.M., Poling, B.E.: The properties of gases and liquids (4th ed).
McGraw-Hill International Editions, New York (1988)
45. Rifai, M.N.E., Kaufman, W.J., Todd, D.K.: Dispersion phenomena in laminar flow through
porous media. University of California, Sanitary Engineering Report 3. Inst. Eng. Res. Series
90, 1–157 (1956)
46. Robbins, G.A.: Methods for determining transverse dispersion coefficients of porous media in
laboratory column experiment. Water Resour. Res. 25, 1249–1258 (1989)
47. Roemer, G., Dranoff, J.S., Smith, J.M.: Diffusion in packed beds at low flow rates. Ind. Eng.
Chem. Fund. 1, 284–287 (1962)
48. Sherwood, T.K., Pigford, R.L., Wilke, C.R.: Mass Transfer. McGraw-Hill International
Editions, Tokyo (1975)
49. Sinclair, R.J., Potter, O.E.: The dispersion of gas in flow through a bed of packed solids.
Trans. IChemE 43, T3–T9 (1965)
50. Stephenson, J.L., Stewart, W.E.: Optical measurements of porosity and fluid motion in
packed beds. Chem. Eng. Sci. 41, 2161–2170 (1986)
51. Sun, N.Z.: Mathematical modelling of groundwater pollution. Springer, New York (1996)
52. Taylor, G.: Dispersion of soluble matter in solvent flowing slowly through a tube. P. Roy Soc.
A –Math. Phy. 219, 186–203 (1953)
53. Uno, S.: Mass Transfer from naphthalene slabs. PhD Dissertation, Illinois Institute of
Technology, Chicago (1958)
54. Van Genuchten, M.Th., Alves, W.J.: Analytical solutions of the one-dimensional convective-
dispersive solute transport equation. Tech. Bull. U.S. Dep. Agric. 1661, 149–165 (1982)
References 173
55. Vortmeyer, D., Schuster, J.: Evaluation of steady flow profiles in rectangular and circular
packed beds by a variational method. Chem. Eng. Sci. 38, 1691–1699 (1983)
56. Wakao, N., Funazkri, T.: Effect of fluid dispersion coefficients on particle-to-fluid mass
transfer coefficients in packed beds. Chem. Eng. Sci. 33, 1375–1384 (1978)
Chapter 7
Applications and Examples
7.1 Contaminant Plume Sizes Associated to Different

Active Solids
There are several situations of practical interest, both in nature and in manmade
processes, in which there is fluid flow through a bed of inert particles, packed
around a solid mass that contacts with the moving fluid. Examples may be found in
diverse fields, such as dilute catalyst fixed bed reactors, fluidized bed combustion,
ore leaching and water contamination by buried waste. In such processes there is
an interplay between diffusion, convection and dispersion and a detailed system-
atic study of the problem has been given by Coelho and Guedes de Carvalho [16]
for transfer from buried flat surfaces and for transfer from buried spheres. Both
these references present accurate solutions for certain limiting situations, namely
those of low and high fluid velocity (more precisely, low and high Peclet
numbers).
When the mass transfer process occurs in a porous media with a fluid flowing
around the soluble particle, and at low fluid velocities (as typically observed in
underground flow), the assumption of thin boundary layer is not legitimate, and the
theoretical analysis developed by Coelho and Guedes de Carvalho [16] is not
applicable. Therefore, it is necessary to employ numerical methods for a correct
analysis of the mass transfer process in more general situations.
Flow around a buried sphere is an important model situation in many processes
and in a recent work Guedes de Carvalho et al. [31] treated the problem numer-
ically, so as to cover the entire range of values of Peclet and Schmidt numbers.
Flow along buried cylindrical surfaces and flat surfaces are also important
model situations, and were investigated theoretically and numerically by Alves
et al. [8], yielding results for a wide range of values of Peclet number, aspect ratio
of soluble solid mass and Schmidt number.
Over the last years, our focus has been driven preferentially to the analysis of
the mass transfer of particles with spherical geometry. This work is a contribution

176 7 Applications and Examples
for the study of mass transfer of soluble particles with different geometries buried
in inert particles with smaller diameter. Additionally, a simple approximate
method is presented to obtain concentration contours plots for solute distribution
around and downstream of the buried surfaces (with different geometries).
The following sections present a detailed description of the mass transfer and
dispersion process around a soluble solid particle with different shapes (sphere,
cylinder or a plane surface aligned with the flow, cylinder in cross-flow, prolate
spheroid and a oblate spheroid) buried in a packed bed of smaller inert particles
with uniform voidage, with a moving fluid with constant interstitial velocity.
7.1.1 Concentration Profiles From a Soluble Flat Slab
In many processes, both in industry and in the environment, there is a physical/

chemical interaction between a lump of solid matter buried in a granular bed of
small inert particles, and the fluid flowing around it, through the interstices in the
bed. In such processes, a region of higher solute/reactant concentration develops in
the immediate vicinity of the ‘‘active mass’’, with consequent solute migration
away from it, by diffusion and convection in the moving fluid.
In groundwater flow applications the velocities are low and small values of
Peclet number are therefore observed; solute dispersion is then determined by
molecular diffusion.
The concentration distribution in the wake of a soluble solid particle immersed
in a granular bed of inert particles, through which fluid flows with ‘‘uniform
velocity’’, has been obtained, for solute transport by both advection and diffusion.
General expressions, based in a theoretical analysis and using the expressions
obtained for the mass transfer rate, n, are presented to estimate contaminant
‘‘plume’’ sizes downstream of the polluting source.
In this model situation, we only consider the steady state conditions when the
plane surface releases solute to the flowing liquid at a constant rate, n. The value of
n depends on the concentration gradients next to the surface of the plane surface
and these depend on the process variables. According to Eq. 6.68, the following
equation
1=2
D0 2 0 1=4 4
n ¼ m e bL ðc c0 Þ Pesf þ Pe0sf ð7:1Þ
L p p
gives n within 3% of the exact value, over the entire range of values of Pe0sf ; if
Pe0p \0:1:
Imagine a situation of continuous injection of solute, at a rate n, at the point
ðx ¼ 0; y ¼ 0Þ; in the porous medium through which fluid flows with uniform
interstitial velocity u0 : It is assumed that the volume of additional fluid injected
into the main fluid, with the solute, is negligible.
7.1 Contaminant Plume Sizes Associated to Different Active Solids 177
Fig. 7.1 Concentration

profiles obtained with
Eq. 7.3, at long distances
from the flat slab
Making DT ffi DL ffi D0m a diffusion/convection cloud is again formed and the

corresponding concentration contour plots are given by the solution for the
‘‘continuous point source’’ [100] of two-dimensional solute transport,
(sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi )
n=b u0 x u20 ðx2 þ y2 Þ
c¼ exp K0 ð7:2Þ
2ep D0m 2 D0m 4 D0m 2
where b is the depth and K0 is the modified Bessel function of second kind
and zero order. In our case, the problem of mass transfer from a soluble slab (with
length L) immersed in a granular bed of inerts through which fluid flow with
uniform interstitial velocity and with n from Eq. 7.1, one obtains after
re-arrangement:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 0 1=4 4 0
Pe þ Pesf 0 (sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
)
p sf p Pesf x Pe0sf 2 x 2 y 2
C¼ exp K0 þ ð7:3Þ
2p 2 L 4 L L
This equation may be used to estimate the region of propagation of a given

concentration level.
Figure 7.1 show the concentration contour plots obtained, taking Pe0 ¼ 1; 000
as an example, for low values of C. As the value of C decreases, the distance of the
contour surfaces to the solid soluble particle increase and the solution for the
‘‘continuous point source’’ approach to the ‘‘exact’’ solution, obtained numerically.
This result was confirmed by Gurdes de Carvalho et al. [31], who analysed the
accuracy of the ‘‘continuous point source’’ solution, for different ranges of C, in
the case of an active sphere immersed in a packed bed. The authors showed that for
the higher values of C the concentration contours obtained with ‘‘continuous point
source’’ solution deviates considerably from the ‘‘exact’’ solution, obtained
numerically. However, for low values of C and if a correct value of n is used, true
coincidence is observed between the numerical results and the ‘‘continuous point
source’’ solution.
7.1.2 Concentration Profiles From a Cylinder Aligned With Flow
The analytical solution for a continuous point source has also been derived by
Wexler [100], solving the three-dimensional solute-transport equation from a point
source. The solution is given by
" #
n u0 ½x ðx2 þ y2 Þ1=2
c¼ exp : ð7:4Þ
4epðx2 þ y2 Þ1=2 D0m 2D0m
with DT ffi DL ffi D0m and a good estimate for n is required. The point source is
located at the point ðx; yÞ ¼ ð0; 0Þ: In our case, the problem of mass transfer
around a soluble cylinder (with length L and diameter d1) immersed in a granular
bed of inerts through which fluid flow with uniform interstitial velocity, the mass
flux rate is expressed as,
0 D0m
n ¼ e kpd1 Lðc c0 Þ ¼ Shc epd1 Lðc c0 Þ ð7:5Þ
L
0
with Sherwood number, Shc ; given by Eq. 6.73. Making use of the dimensionless
variables (please see Eq. 6.58), Eq. 7.4 results, after re-arrangement:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1=2 3=2 1=3ffi
2 0 1=4 4 0 L 4L 5 0 5=9 L 0 2=9 L
Pe þ Pec þ 6 þ þ Pec Pec 2
p c p d1 3 d1 3 d1 d1
C¼ h i 1=2
4ðL=d1 Þ ðx=LÞ2 þðy=LÞ2
0 h i 1=2
Pec
exp x=L ðx=LÞ2 þðy=LÞ2 þ
2
ð7:6Þ
Figure 7.2 show the concentration contour plots obtained, taking Pe0c ¼ 1; 000;
as an example, for low values of C.
7.1.3 Concentration Profiles From a Soluble Sphere
In this model situation, we investigate the steady state conditions when a soluble
sphere releases solute to the flowing fluid, with uniform interstitial velocity u0, at a
constant rate, n. Again, it is assumed that the volume of additional fluid injected
into the main fluid, with the solute, is negligible.
A diffusion/convection cloud is formed and the corresponding concentration
contour plots for the ‘‘continuous point source’’, at the point ðx ¼ 0; y ¼ 0Þ; are
also given by Eq. 7.4. If n from Eq. 6.82 is substituted in Eq. 7.4, the resulting
equation may be re-arranged to read

from the cylinder
1=2
4 4
4 þ Pe0s 2=3 þ Pe0s 0 h i1=2
5 p Pes 2 2
C¼ h i1=2 exp ðx=aÞ ðx=aÞ þ ðy=aÞ
4
2 ðx=aÞ2 þ ðy=aÞ2
ð7:7Þ
Figure 7.3 show the concentration contour plots obtained, taking Pe0s ¼ 1; 000;
as an example, for low values of C.
7.1.4 Concentration Profiles From a Cylinder in Cross Flow
If a slightly soluble cylinder, buried in a packed bed, is exposed to uniform fluid

flow with velocity u0, it will then release solute at a rate n, given by Eq. 6.91.
A diffusion/convection cloud is formed and the corresponding concentration
contour plots for the ‘‘continuous point source’’, at the point ðx ¼ 0; y ¼ 0Þ; are
also given by Eq. 7.4, and the substitution of n=b from Eq. 6.91 into 7.4 gives then
1=2
2 0 1=4 32 0 (sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pe þ 3 Pecf ffi)
p2 cf p Pe0cf x Pe0cf2 x2 y 2
C¼ exp K0 þ
2 4 a 16 a a
ð7:8Þ
Concentration contours given by Eq. 7.8 are shown in Fig. 7.4, for Pe0cf ¼
1; 000 (note that for d1 ¼ 2 m, u0 ¼ 5 104 mm/s and D0m ffi 109 m2 =s; there
results Pe0cf ffi 1; 000). From a detailed study of a similar nature, for flow past a
sphere, it is reasonable to expect that values of the downstream reach of each
contour surface (defined as the value of x=a; for y=a ¼ 0; for that surface),

from the sphere
predicted by Eq. 7.8, will differ by less than 10% from the exact value, if
Pe0 [ 500 and C\0:05: The accuracy of the prediction will improve as C is
decreased.
7.1.5 Concentration Profiles From a Prolate Spheroid
If a soluble prolate spheroid, buried in a packed bed, is exposed to uniform fluid

flow with uniform interstitial velocity u0, it will then release solute at a rate n,
given by
rffiffiffiffiffiffiffiffiffiffiffiffi 1=2
0 4 0 2 e3 2pa
nT ¼ eDm Peps 1
ðC C0 Þ ð7:9Þ
p 2
3 e ð1 e Þ tanh e ð1 e2 Þ1=6
with Sherwood number, Sh0ps ; given by Eq. 6.115. Making use of the dimensionless
variables, Eq. 7.4 results, after re-arrangement:
1=2 1=2
1 4 0
2 e3
Pe
C C0 2 p ps
3 e ð1 e2 Þ tanh1 e

¼ h i 1=2
C C0
4 ðx=deq Þ2 þ ðy=deq Þ2 ð7:10Þ
0 h i 1=2
Peps
exp x=deq ðx=deq Þ2 þ ðy=deq Þ2
2
Fig. 7.4 Concentration 12 Line 1

profiles obtained with 2 3 4 5
C: 0.005 0.006 0.007 0.01 0.02
from the cylinder in cross 10
1
flow Pecf´=1000
8
2
y/a
6
3
4
4
2
5
0
0 20000 40000 60000
x/a
7.1.6 Concentration Profiles From an Oblate Spheroid
If a slightly oblate spheroid, buried in a packed bed, is exposed to uniform fluid

flow with uniform interstitial velocity u0, it will then release solute at a rate n,
given by
rffiffiffiffiffiffiffiffiffiffi !1=2
4 0 4 e3
nT ¼eD0m Pe pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
p 3 e þ e3 þ 1 e2 cot1 ð 1=e2 1Þ
ð7:11Þ
2pa2 ð1 e2 Þ1=3
ðC C0 Þ
deq
with Sherwood number, Sh0os ; given by Eq. 6.129. Making use of the dimensionless
variables, Eq. 7.4 results, after re-arrangement:
1=2 !1=2
1 4 0 4 e3
2 p Peos
pffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
C C0 3 e þ e3 þ 1 e2 cot1 ð 1=e2 1Þ
¼ h i 1=2
C C0 ð7:12Þ
4 ðx=deq Þ2 þ ðy=deq Þ2
0 h i 1=2
Peos 2 2
exp x=deq ðx=deq Þ þ ðy=deq Þ
2
Figure 7.5 show the concentration contour plots obtained, taking Pe0os ¼ 1; 000
and e ¼ 0:5 as an example, for low values of concentration, C. As the value of
C decreases, the distance of the contour surfaces to the solid soluble particle
increase and the solution for the ‘‘continuous point source’’ approach to the
‘‘exact’’ solution, possible to obtain numerically (i.e. if a correct value of n is used,
true coincidence is observed).
Fig. 7.5 Concentration C = 0.001 0.0005 0.0002 0.0001

Line: 1 2 3 4
from the oblate spheroid
y / d eq Pe’=1000
e = 0.5
3
2
1
x /d eq
The problem of mass transfer between a buried solid with different shapes and
the fluid flowing along it in a granular bed, lends itself to a simple full theoretical
analysis, under an appropriate set of conditions.
The elliptic partial differential equations resulting from a differential mass
balance have been solved numerically over a wide range of values of the relevant
parameters. It was found that the resulting Sherwood values are in good agreement
with experimental results.
Results of the exact numerical solutions and analytical solutions were also used
to predict the solute migration from an active solid particle buried in a packed bed
of inert particles, through which fluid flows with uniform velocity. The concen-
tration contour surfaces were obtained using an analytical solution of continuous
injection of solute at a point source in a uniform stream and the proposed corre-
lations for the mass transfer rate developed numerically or analytically.
7.2 Rising Damp in Building Walls
Rising damp, a world-wide phenomenon, is a major cause of decay in masonry

materials such as stone, brick and mortar. The conservation of historic buildings
has become important nowadays and has developed significantly in recent years.
Rising damp in historic buildings may be considered one of the most important of
all the different manifestations of dampness, leading to the destruction of stone
materials due to frost/defrost cycles and the presence of salts associated with
crystallization (Künzel [61] argues that in many cases increased salt content is
7.2 Rising Damp in Building Walls 183
interpreted as rising damp), decay in elements in contact with stone walls, such as
wooden beams, wooden ceilings and other finishings, and insanitary conditions
caused by excess humidity associated with the development of fungus and mould.
Rising damp, which originates in the ground and permeates porous materials by
capillarity, is one of the main causes of decay in old buildings, particularly those
containing thick walls built from different materials. Having identified the particular
characteristics of those buildings, it is important to recognize the limitations of
traditional technologies and to investigate new solutions for the phenomenon. The
traditional techniques used to deal with this kind of problem (such as watertight
barriers, the injection of hydrofuge products, etc.) sometimes prove ineffective or are
very expensive, justifying the need to find new approaches (see [34]).
The Building Physics Laboratory (LFC), at the University of Porto-Faculty of
Engineering (FEUP) has been developing important experimental research into the
problems of rising damp. In recent years the principles for a treatment technique
called the ‘‘Wall Base Ventilation System to Treat Rising Damp–HUMIVENT
device’’ has been validated and characterised in experiments. The technique
consists of circulating air around the base of very thick walls built from different
materials, with a relative humidity saturation distance. Wall base ventilation
increases evaporation, which reduces the level of the damp front. This is possible
only when the groundwater is lower than the base of the wall [17].
In previous research it was observed that the ‘‘wall base ventilation device’’
reduces the level of the wet front. Following this, a series of experiments were
carried out to explore boundary conditions, geometry, engine speed, etc., and it
was possible to observe that a hygro-regulated system is essential in controlling
possible condensation inside the system. A hygro-regulated system is a mechanical
system controlled by probes placed at the inlet and at the outlet. In accordance
with pre-programmed criteria, the probes will switch the system on/off.
7.2.1 Rising Damp Theory
The phenomenon of rising damp can be analysed using the sharp front model
developed by Hall and Hoff [39]. This model describes the relationship between
sorptivity, S, wall thickness, b, evaporation potential, e, and capillary rise
heights, h.
Considering the absorption inflow and evaporation loss to be balanced and
disregarding gravitational forces (see Fig. 7.6), the following differential equation
is obtained [40]
di S
b ¼ b t1=2 eh ð7:13Þ
dt 2
where i is the volume of liquid absorbed per unit cross section in time t [36].
Considering that i ¼ hw h; the following expression is obtained
Fig. 7.6 Sketch of sharp b

front model
e e
h
dh S 1=2 eh
¼ t ð7:14Þ
dt 2hw bhw
with i ¼ St1=2 ¼ hw h; results
dh S2 1 eh
¼ 2 ð7:15Þ
dt 2hw h bhw
The analytical solution of Eq. 7.15 is
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

b 2e
h¼S 1 exp t ð7:16Þ
2ehw bhw
with the equilibrium height, h1 ; given by

rffiffiffiffiffiffiffiffiffiffi
b
h1 ¼ S ð7:17Þ
2ehw
7.2.2 The Wall Base Ventilation System
The wall base ventilation system (HUMIVENT device) consists of ventilating the
base of walls using a natural ventilation process or by installing a hygro-regulated
mechanical ventilation device (see Fig. 7.7). In terms of analysis, we considered a
wall base ventilation system along which air is flowing close to the ‘‘saturated’’
wall ð0\z\LÞ:
In laminar flow, the transport of moisture and temperature across the boundary
layers is controlled by molecular diffusion. The Reynolds number, Re, may be
interpreted as the ratio of the flow of destabilising forces to stabilising forces
e e
e e
e e
eH eH eH eH
e e
w ater w ater w ater
Configuration 1 Configuration 2 Configuration 3
Fig. 7.7 Diagram of the different boundary conditions studied
(viscosity). Stable laminar flow is thus characterized by low Re values. Laminar

flow occurs at Reynolds numbers of less than approximately 3 9 105. In our study
Re\105 ; in all experiments.
If the analysis is restricted to situations in which the moisture transfer boundary
layer is thin and if a small control volume is considered inside this boundary layer,
with side lengths dz, dy and unity (perpendicular to the plane of the figure), a
steady state material balance in the solute leads to
oc o2 c o2 c
u ¼ Dm 2 þ D m 2 ð7:18Þ
oz oy oz
where Dm is the molecular diffusion coefficient, in the cross stream and in the
stream wise directions. If the boundary layer is thin compared to the length of the
saturated wall, the last term on the right-hand side of Eq. 7.18 is likely to be
negligible. For the situation outlined in Fig. 7.6, the boundary conditions are,
c ¼ c0 z ¼ 0 y[0 ð7:19aÞ
c ¼ c z[0 y¼0 ð7:19bÞ
c ! c0 z[0 y!1 ð7:19cÞ
and the analytical solution is given by [21]

!
c c0 y
¼ erfc pffiffiffiffiffiffiffiffiffiffiffiffiffi ð7:20Þ
c c0 2 Dm z=u
where c0 is the bulk concentration of water vapour, c* is the equilibrium con-

centration of water vapour, and u is the air velocity. The flux of evaporation, N, at
the saturated wall surface may be obtained from (7.20) as

oc Dm 1=2
N ¼ Dm ¼ ðc c0 Þ ð7:21Þ
oy y¼0 p z=u
Finally, it is possible to obtain the total rate of solid evaporation, n, over the
whole wall surface
ZL 1=2
4Dm
n¼ N hH dx ¼ ðc c0 ÞhH L ð7:22Þ
pL=u
0
where N ¼ Dm ðoc=oyÞy¼0 and hH is the width of HUMIVENT system.

If we consider a rising damp steady state, involving a wall buried in sand and a
wall base ventilation system (HUMIVENT) located on both sides above the base
(see Fig. 7.7), Eq. 7.17 produces
bS2
¼ eðh1 hs hH Þ þ eH hH þ es hs ð7:23Þ
2hw h1
with eH (related to HUMIVENT) given by [34]
1=2
n ðc c0 Þ 4Dm
eH ¼ ¼ ð7:24Þ
L:hH qw qw pL=u
and the steady-state height of rise given by

es hs e H hH
h1 ¼ 1 þ 1
e 2 e 2
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
bS 2 es 2 h 2 e 2 h2 es eH es eH hs hH
s H H
þ þ 1 þ 1 þ 1 þ e
2hw e e 4 e 4 e e 2 2
ð7:25Þ
Noting that the evaporation potential related with sand placed on both sides of
the wall above its base, es, will be negligible (sand water saturated) in comparison
with the other evaporation rates, e and eH, and Eq. 7.25 reduces then to
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
hs eH hH bS2 h2 eH 2 h2
H e H hs hH
h1 ¼ þ 1 þ þ sþ 1 þ 1
2 e 2 2hw e 4 e 4 e 2
ð7:26Þ
and for eH ¼ es ¼ 0; Eq. 7.26 reduces to

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
hs hH bS2 h2 h2 hs hH
h1 ¼ þ þ þ sþ Hþ ð7:27Þ
2 2 2hw e 4 4 2
Finally, to estimate the evaporation potential, e, many existing formulae can be

applied [53]. In this study we used the Penman–Monteith equation [34] which, given that
the latitude of Porto is 40.15, results in an evaporation potential of 0.0017 mm/min.
7.2.3 Numerical Simulation
The simulation programs to evaluate changes in the moisture content and temperature
inside walls are essential instruments for simulating the wall’s behaviour in the
presence of moisture, depending on the internal and external climatic conditions [12].
The calculation program used in the numerical simulations for the experimental
and analytical validations was WUFI-2D, developed by Fraunhofer Institute for
Building Physics. The governing equation for moisture transport is [44]
dw o/
¼ rðDu r/ þ dp r/ps Þ ð7:28Þ
d/ ot
and the governing equation for heat transport is
dH oT
¼ rðkT rTÞ þ hv rðdp rups Þ ð7:29Þ
dT ot
where dH/dT is the heat storage capacity of the moist building material, dw/d/ is
the moisture storage capacity, kT is the thermal conductivity, D/ is the liquid
conduction coefficient, dp is the water vapour permeability, hv is the evaporation
enthalpy of the water, ps is the water vapour saturation pressure, T is the tem-
perature and / is the relative humidity.
As hygrothermal modelling offers a powerful tool for predicting heat and moisture
transport through multi-layer building assemblies. In this work, the hygrothermal
model was used to compare the results of three case studies (monolithic wall) under
different natural conditions (see Fig. 7.7). The laboratory climate was modelled by a
sine wave varying between 19C, 50% RH in winter and 21C, 70% RH in summer,
which corresponds to a normal moisture load (mean values of 60% RH and tem-
perature of 20C). The initial moisture content of the used materials was the equi-
librium moisture at 80% RH. The simulation periods began on January 1.
The vapour diffusion thickness value used was zero (no coating) and the interior
heat transfer coefficient was constant and equal to 8 W/m2K. The exterior heat
transfer coefficient only contained the convective part and was considered inde-
pendent from the wind (constant value of 17 W/m2K).
7.2.4 Rising Damp Analysis
7.2.4.1 In the Laboratory
The physical model selected consisted of a prismatic wall 1.58 m high, 2.00 m
long and 0.20 m thick, waterproofed on the two upper sides (measuring
Table 7.1 Properties of the limestone sample [93]

Limestone
Bulk density, q (kg/m3) 2155 ± 14
Heat capacity, cp (J/kgK) 1,000
Porosity, e (%) 19.7
Thermal conductivity, kT (W/mK) 1.33 ± 0.025
Vapor diffusion resistance factor, l (-) Dry cup: 41 ± 2.2
Wet cup: 29 ± 3.1
Moisture storage function, w (kg/m3) HR (%) Adsorption Desadsorption
4.0 0.521 0.951

11.2 0.593 1.150
34.8 0.872 1.239
58.6 1.043 1.628
76.3 1.237 2.360
80.0 1.334 –
84.2 1.584 –
92.1 2.381 4.144
Capillary transport coefficient, Dw (m2/s) w80% = 1.7 kg/m3 6.6 9 10-11
wsat = 188 kg/m3 6.2 9 10-8
Water absorption coefficient, A (kg/m2s1/2) 0.024 ± 0.002
Free-water saturation, wf (kg/m3) 177 ± 1.8
Fig. 7.8 Position of the relative humidity and temperature probes [93]
1.58 9 0.20 m2) to prevent moisture entering from this direction. The limestone
walls (see Table 7.1) were placed in a reservoir approximately 2.20 9 2.50 9
0.50 m3 made from cement blocks. To assess moisture transfer inside the walls,
probes were inserted at different heights and depths to measure relative humidity
and temperature, as sketched in Fig. 7.8. These probes were then connected to a
data acquisition and recording system.
The configurations tested are shown in Fig. 7.7. In Configuration 1 the base of
the wall is immersed up to a height of 8 cm and, as we wished to assess the effect
of the wall base ventilation system–HUMIVENT, in Configurations 2 and 3, a
ventilation box was placed on both sides of the wall. The only difference between
these configurations was that in Configuration 3 we measured the behaviour of a
Table 7.2 Experi ment Configuration 1 Configuration 2 Configuration 3

values (average) for relative
humidity and temperature /0 (%) /0 (%) /s (%) /0 (%) /s (%)
inlet in the wal l and system 60.0 57.8 60.1 58.2 60.3
outlet T0 (8C) T0 (8C) Ts (8C) T0 (8C) Ts (8C)
20.0 22.6 22.4 23.0 22.8
Table 7.3 Comparison h? (mm) Limestone

between experimental,
analytical and numerical Conf. 1 Conf. 2 Conf. 3
results (obtained with an error Experiments 455–540 380–455 455–540
lesser than 5%) WUFI 2D (/) 550 – 490
WUFI 2D (w) 520 – 490
Analytical solution 510 380 451
/eq (%) 60
weq (kg/m3) 1.047
wall with both sides buried in sand to a height of 45 cm. The sand was saturated
(100% RH) during the tests. To avoid turbulence, an air velocity of 0.083 m/s was
used and the relative humidity and temperature values for the experiments are
presented in Table 7.2.
The climatic conditions inside the laboratory during the experiments were
considered constant at a temperature of 238C and a relative humidity of 58%.
Under these conditions and considering an air velocity inside the laboratory of
1 cm/s, the evaporation potential was given by [87]
ðc c0 Þ Re1=2 Sc1=3

e ¼ Dm 0:646 ð7:30Þ
qw L
The resulting evaporation potential value is 0.46 9 10-4 mm/min. The evapo-
ration potential associated with the HUMIVENT system, eH, is given in Eq. 7.24 as
2.17 9 10-4 mm/min. If these values are inserted into Eq. 7.25, a steady state height
of rise of 379.5 mm is obtained for Configuration 2 and a steady flow rate through the
wall of about 27.1 L/year per metre length of wall. This value is very similar to the
experimental results obtained, of between 380 and 455 mm (as sketched in Fig. 7.8,
position 6 and 7–average value of 417.5 mm), and significantly different from the
steady state height of rise obtained without the use of the HUMIVENT system, which
was h1 ¼ 540 mm in Configuration 1 (experimental value between 455 and
540 mm—Fig. 7.8, position 7 and 8—as shown in Table 7.3).
In the analyse of our experimental results, the following expression was used
for water vapour diffusivity in air [56]
1:5
273:15 þ Tð CÞ
Dm ðm2 =sÞ ¼ 2:23 105 ð7:31Þ
273:15
16 6
Q (kg)
14 3 5
C (g/m )
12 4
10
C (g/m3 )
3
8
Q (kg)
2
6
1
4
2 0
0 -1
(a)
-2 -2
9-Jul 29-Jul 18-Ago 7-Set 27-Set 17-Out 6-Nov
Time (days)
16 6
Q (kg)
14 3 5
C (g/m )
12 4
10
C (g/m3 )
3
8
Q (kg)
2
6
1
4
2 0
0 -1
(b)
-2 -2
9-Jul 29-Jul 18-Ago 7-Set 27-Set 17-Out 6-Nov
Time (days)
Fig. 7.9 Quantity of water vapour extracted and concentration differential (a) in Configuration 2
and (b) in Configuration 3
Table 7.3 shows the experimental results for the first series of laboratory tests,
i.e. the rising damp values for the steady height of rise under different boundary
conditions. The results of the experiment show that the presence of a wall base
ventilation system on both sides prevents the damp front (see results) and that the
simple analytical model proposed by Hall and Hoff [39] clearly describes the rising
damp front when compared with the experiment results. During the four months of
experimental research, in laboratory tests on Configurations 2 and 3 the new
treatment technique extracted approximately 11.8 and 14.5 kg of water respec-
tively (see Fig. 7.9a, b).
The simulation program provided the water content, the relative humidity and
the temperature at all the points of the pre-defined grid and their evolution for the
duration of the simulation. It also gave the heat and moisture flows along its
(a) (b)
1000
1000
910
910
766
766
623
623
550
520
485
485
490
356
356
8
RH (-) w (kg/m3) 240 RH (-) w (kg/m3)

240
0.95-1.00 <1.0500 0.95-1.00 <1.0500
0.90-0.95 1.0500-1.0495 0.90-0.95 1.0500-1.0495
143 143
0.85-0.90 1.0495-1.0490 0.85-0.90 1.0495-1.0490
0.80-0.85 1.0490-1.0485 0.80-0.85 1.0490-1.0485
67 67
0.75-0.80 0.75-0.80 1.0485*1.0483
1.0485*1.0483
0.70-0.75 0.70-0.75 1.0483-1.0480
1.0483-1.0480
18 18 0.65-0.70
0.65-0.70 1.048-10477 1.048-10477
0.60-0.65 0.60-0.65 1.0477-1.0474

1.0477-1.0474
0 0.55-0.60 1.0474-1.0470 0.55-0.60 1.0474-1.0470
0
Fig. 7.10 Variation of rising damp front with relative humidity and water content for
(a) Configuration 1 and (b) Configuration 3
surfaces. Figure 7.10a, b show the variation of rising damp front with relative
humidity and water content for different boundary conditions. These figures also
show that the numerical results are in accordance with the experimental and
analytical values obtained (see Table 7.3).
The research group is currently studying whether these values may have adverse
consequences in terms of the long-term deterioration of the wall materials [99].
As we understand how salts dissolve in water and their behaviour in terms of
crystallization/dissolution, the target, in future, will be to scale the new rising damp
treatment system to avoid crystallization/dissolution problems.
7.2.4.2 In-Field
The hygro-regulable system was installed in a church in Northern Portugal

(see Fig. 7.11a). Inside the building, two hygro-regulable mechanical ventilation
subsystems were installed (see Fig. 7.11b). In the Southside subsystem, air
was admitted through grids located inside the building, and was extracted into
the cloister. Extraction was controlled by a hygro-regulable engine of variable
speed [26]. The system had two probes for measuring relative humidity and
Fig. 7.11 (a) Church in Northern Portugal; (b) Wall base ventilation system and (c) Data
acquisition and recording system
temperature, two transmitters, a control module and a data acquisition system for
recording results (see Fig. 7.11c).
The system installed in the church initially began operating whenever the rel-
ative humidity at the exit was 5% higher than the relative humidity at the entrance.
The idea was to admit dry air comparing to the air inside the system. This criterion
was found to be inadequate, as it meant that the system was operating at periods
when condensation occurred inside it. Consequently, a new criterion was proposed
with a view to optimizing the system, based upon the difference in vapour pressure
(DP) at the exit and entrance. The system now began functioning whenever the
DP was positive.
Figure 7.12a, b shows the functioning periods of the ventilator, in a north and
south wall, and the importance of HUMIVENT device in rising damp treatment,
by the quantity of water vapour extracted during the five months analyzed
(272.3 kg in the north wall and 170.2 kg in the south wall).
280 1000
Q (kg)
240 P (Pa) (a)
600
200
P (Pa)
160 200
Q (kg)
120
-200
80
-600
40
ON
0 OFF -1000
10-Mai 30-Mai 19-Jun 9-Jul 29-Jul 18-Ago 7-Set 27-Set
Time (days)
280 600
Q (kg)
(b)
240 P (Pa)
400
200
P (Pa)
200
160
Q (kg)
120 0
80
-200
40
ON
0 OFF -400
10-Mai 30-Mai 19-Jun 9-Jul 29-Jul 18-Ago 7-Set 27-Set
Time (days)
Fig. 7.12 Quantity of water vapour extracted, pressure differential and functioning of system:
(a) North wall and (b) South wall
The entry of air with very low relative humidity could generate the crystalli-
zation of salts existing in the building materials, threatening its durability. For this
reason, the relative humidity value at the entrance had to be limited. The relative
humidity scores recorded, which range from 60 to 95%, are not considered to
present a risk of salt crystallization/dissolution inside the system, consider the salts
detected in those area. However, the problem might arise in another type of
external climate or with other salts type [99]. In that case the HUMIVENT device
must also limit the inlet relative humidity value.
In conclusion, the mechanisms of moisture transfer are complex, particularly in
terms of rising damp in historic buildings. As rising damp is one of the main causes of
decay in these buildings, it is important to study the factors related to this phenomenon.
The traditional techniques currently used to minimise rising damp are not
effective or too expensive, in particular when dealing with walls of considerable
thickness and heterogeneous materials. Laboratory tests on 20 cm thick limestone
walls show that an efficient way of treating rising damp in historic buildings is to
ventilate the wall base using the HUMIVENT device.
The analytical model of rising damp predicts steady height of rise figures that
are consistent with laboratory observations, which provide a good validation of the
analysis. Using this model it is possible to scale the HUMIVENT system, optimise
it and predict its behaviour. The numerical simulation results are similar to those
obtained experimentally and analytically, which provide a qualitative validation of
the results obtained.
7.3 Bubbles and Drops in Porous Structures
Transport of mixtures is encountered in many industrial processes and engineering

applications. More specifically, one of the most interesting cases is that of a dis-
continuous phase (droplets) situated within a flowing continuous aqueous phase.
A typical example of this may be found in the oil industry, and is related to uncon-
ventional behavior found in some heavy oil reservoirs where a disconnected gas
phase (gas bubbles) inside a connected liquid phase (heavy oil) is observed over
periods of time long enough to consider that such a phase distribution governs flow
dynamics [29, 42]. Qualitatively speaking, the aspects of such a transport process are
rather well understood. In general, droplets are transferred by convection and
diffusion while their population is strictly governed by local hydrodynamic
phenomena [85]. It is very difficult to obtain a globally valid description for the
consequent transient (dynamic) behavior of the microstructure of such a mixture due
to the high geometrical complexity of such a configuration. For example, the domain
size could grow because the domains may coalesce under flow conditions and a
break-up process sets in when they are large enough to be affected.
Renewed interest in this field originates from theoretical investigations on the
flow in porous materials using conventional computations [30, 73, 89], the Lattice
Boltzmann Method [46, 68, 69] and pore network simulators [13, 24]. These
approaches focus mainly on flow dynamics for multiphase and multi-component
fluids rather than the microstructure formation where limited research has been
done [59, 60, 64, 70, 98].
The exemplar application presented here aims to describe this micro-structure
especially for mixtures flowing in a homogeneous porous medium under laminar
conditions. In particular, the scope of this theoretical study is to investigate the
physical/geometrical parameters and the flow conditions which could assure the
mixture’s homogeneity (water/droplets, i.e. continuous/discontinuous phase) in
terms of mean droplet size. Focus is on the domain sizes formed during this process
and the approach taken here is the so-called ‘Sc concept’, i.e. the ‘moments of
distribution’, for droplet size description [52], where an arbitrary number of
moments of the domain size distribution is used to describe the microstructure. In the
same study, the method was also described in relation to experiments on bubble
coalescence. Subsequently, Klahn et al. [57] and Agterof et al. [3] used the same
method to analyze experiments on the transient behavior of emulsion droplet break-
up in an impeller flow, and to analyze the behavior of double emulsions. All these
examples demonstrated the method’s power, and will be explained in detail below.
7.3 Bubbles and Drops in Porous Structures 195
In this section the Sc method is extended to phenomena related to porous

structures. The problem examined here is the laminar flow of a biphasic mixture in
a homogeneous porous medium. The domain size for the two phases not only
depends on the flow itself but also on the break-up and coalescence processes
taking place because of the flow. The method’s fundamental idea is that the
evolution of the moments of a distribution can be analyzed using a transport
equation including a convective term, which can be coupled to the local flow
characteristics, which should not be affected by the presence of droplets. This
assumption is valid only for droplet sizes small enough to follow the bulk flow and
adequately assure the homogeneity of the flowing mixture (see Eq. 7.38) and
relative discussion). The source terms in the transport equation describe the local
phenomena (break-up and coalescence) that affect particle size distribution. The
main advantage of this approach is that analytical expressions (including proba-
bility parameters) could be derived for the characteristic magnitudes, leading to
successful simulations through common and rather simple CFD techniques.
7.3.1 Moments of Distribution (Sc)
The domain (droplet) size distribution can be described by a collection of moments

of the distribution as [3, 52, 57]:
Z1
Sc ¼ n dc PðdÞ dd ð7:32Þ
0
where c is an integer indicating the specific moment, n is the total number density,
and P(d) is the droplet size distribution. Some of these moments can be related
directly to physical properties of the dispersion. More precisely [3, 52]:
Z1
S0 ¼ n PðdÞ dd ¼ n ð7:33Þ
0
is the total number of droplets (number density),

Z1
S2 ¼ n d2 PðdÞ dd ð7:34Þ
0
is the specific interfacial area, and

Z1
6u
S3 ¼ n d 3 PðdÞ dd ¼ ð7:35Þ
p
0
is an expression of the volume fraction, u.

It can be easily shown that for two different c values, the corresponding Sc
functions can produce a useful characteristic diameter dc1 c2 according to the relation:
R1 c ! 1 c c
1
0 d 1 PðdÞ dd c1 c2 Sc1 1 2

dc1 c2 ¼ R1 ¼ for c1 6¼ c2 ð7:36Þ
c2 Sc2
0 d PðdÞ dd
From geometrical and physical point of view, the most interesting dc1 c2 is the
area-weighted mean droplet diameter, d32, given as:
R1
S3 d½d 2 PðdÞ dd
d32 ¼ ¼ 0R 1 2 ð7:37Þ
S2 0 d PðdÞ dd
It is worth noticing that this diameter is a spatially averaged size which here is
considered to represent the actual droplet size.
The main advantage of the Sc function is that it satisfies the transport
equation [52]. Starting from the well-known population balance equation and
following the mathematical route presented by Kamp et al. [52], the following
transport equation can be derived:
onPðdÞ
þ r ðunPðdÞÞ ¼ G E ð7:38Þ
ot
where u is the local velocity vector and G, E are the generation and the extinction
terms, respectively. These terms are strongly related to droplet creation (break-up)
and death (coalescence).
After multiplying each term by dc and subsequently integrating over all particle
sizes, the following equation can be obtained:
2 1 3 2 1 3
Z Z Z1
o4
d nPðdÞdd5 þ r 4n d uPðdÞdd5 ¼
c c
ðG EÞdc dd ð7:39Þ
ot
0 0 0
By reversing the order of differentiation and integration, Eq. 7.39 can also be
written as a transport equation:
oSc ðcÞ
þ r ðuc Sc Þ ¼ hi ð7:40Þ
ot
where:
R1
0 PðdÞuðdÞd c dd
uc ¼ R1 ð7:41Þ
c
0 PðdÞd dd
Expression (7.41) equals the local velocity vector, u, if the droplets have the
same velocity as the flowing continuous bulk phase, an assumption adopted here. It
is assumed that the local flow field is independent of the droplets, thus their mean
diameter is too small to affect the homogeneity of the flowing phase. Even for
porosities close to unity, the mean droplet size must be lower than 10-6 m for
creeping or laminar flow conditions [64] to satisfy the above assumption.
ðcÞ
In Eq. 7.40, hi is the source term related to droplet break-up and coalescence,
expressed in terms of the moments of the distribution. For the majority of engi-
neering applications, transport equation (7.41) can be simplified to:
o Sc ð cÞ
þ u rSc ¼ hi ð7:42Þ
ot
When the flow problem is uncoupled from the transport one, the velocity field u
can be obtained from typical CFD flow simulations.
7.3.2 Determination of Source Terms
Since the governing phenomena in the fluid phase are the droplet break-up and
coalescence processes, the source term of the previous equation can be expressed as:
ðcÞ ðcÞ ðcÞ
hi ¼ hbr þ hcl ð7:43Þ
ðcÞ ðcÞ
where hbr and hcl are the respective source terms that can be modeled explicitly.
Break-up occurs when droplet diameter becomes larger than a critical value,
dcr . This value is related to the critical capillary number as follows:
2 r Cacr
dcr ¼ ð7:44Þ
gc c_
where Cacr is the critical capillary number, c_ is the local velocity gradient, gc is the
viscosity of the continuous phase, and r is the interfacial tension. This critical
diameter also depends on viscosity ratio and flow type, as discussed by Grace [32]
and Stone [90].
The break-up source term, in its generic form, can be written as [49]:
Z1
ðcÞ
hbr ¼ Kbr ðdÞDSbr
c ðdÞnPðdÞdd ð7:45Þ
0
where Kbr is the break-up rate, and DSbr c ðdÞ is the change in Sc due to a single
break-up event in a droplet of size d. An expression for DSbrc ðdÞ follows from the
conservation of dispersed phase volume when droplet-fragments are formed [101]:
3c

c
DSbr
c ðdÞ ¼ d Nf ðdÞ
3 1 ð7:46Þ
where Nf is the number of fragments formed. Since the break-up rate Kbr is
conversely analogous to the break-up time sbr, the break-up source term becomes:
Z1
ðcÞ dc 3c
hbr ¼ Nf ðdÞ 1 n PðdÞ dd
3 for d [ dcr ð7:47Þ
sbr ðdÞ
0
Coalescence depends not only on collisions but also on coalescence probability,

thus the generic form of the coalescence source term is [15]:
Z1 Z1
ðcÞ
hcl ¼ Kcl ðd; d0 ÞDScl 0 2 0
c ðd; d Þn PðdÞPðd Þdddd
0
ð7:48Þ
0 0
where Kcl is the collision rate, whose detailed description follows. The change in
0
Sc due to a single coalescence event, DSclc ðd; d Þ ; between two droplets of diam-
0
eters d and d ; respectively, can be expressed as [15]:
2 3
0 3 !c=3 0 c
0 c4 d d 5
DScl
c ðd; d Þ ¼ d 1þ 1þ ð7:49Þ
d d
or, by involving a uniform distribution with an equivalent mean diameter deq:

c c=3
DScl
c ¼ deq ð2 2Þ ð7:50Þ
Consequently, the coalescence source term becomes [3]:

2
ðcÞ c=3 6u c6
hcl ¼ ð2 2Þ Kcl ðdeq Þ deq ð7:51Þ
p
The coalescence rate Kcl ðdeq Þ is typically given as [15]:
Kcl ðdeq Þ ¼ Kcoll ðdeq Þ Pcoal ðdeq Þ ð7:52Þ
where Pcoal ðdeq Þ is the coalescence probability of a single collision event, which is
a function of the equivalent diameter, deq, and Kcoll ðdeq Þ is analogous to the
product kcoll urel ðdeq Þ ; representing the Smoluchowski collision rate [3].
Finally, the coalescence source term is given as:
2
ðcÞ c=3 6u c4
hcl ¼ ð2 2Þ kcoll urel ðdeq Þ Pcoal ðdeq Þ deq ð7:53Þ
p
7.3.3 Simulations
Our research team conducted a series of simulation experiments during last

15 years to study the above theoretical approach at laboratory scale. The simu-
lations were carried out using a FORTRAN code especially developed for this
Fig. 7.13 Flowchart of the simulation algorithm
work. Calculations were performed on a simple PC equipped with an Intel Duo

Core processor and 2 GB of RAM, using Windows XP operating system. On
average, the calculations took about 12 h.
To adequately simulate the above problem, the algorithmic procedure shown in
Fig. 7.13 was used. The procedure followed was: firstly the characteristics of the flow
in a porous medium were calculated using standard CFD methods, and subsequently
the local velocity field obtained was used in a transport equation to analyze the evo-
lution of the domain size distribution. Although the algorithm aims at a steady-state
solution, the intermediate results before reaching steady-state conditions could be used
to study transient behavior of the mean droplets, by taking into account only the steady-
state flow field. For the numerical solution of the transport boundary value problems, a
non-uniform finite differences scheme with up winding was used for discretization in
space, with the resulting linear systems of equations solved again using SOR.
To define a realistic domain for the solution of the flow and transport problem, a
porous medium was constructed in the form of a spherical particle assemblage.
Specifically, representation of the biphasic domains considered was achieved by
the random deposition of spheres of a given radius in a box of specified dimensions
as presented in Fig. 7.14, depicting a sample medium of e = 0.72. The structure
was digitized and the phase function was determined to obtain the specified
porosity. The size of the digitized domains was 102 9 102 9 102 and the length
of the simulation box was approximately ten times the sphere radius.
To obtain the velocity field it is necessary to solve the Stokes equations:
rp ¼ lr2 u ð7:54Þ
ru¼0 ð7:55Þ
Fig. 7.14 Stochastically

constructed 3-D porous
medium. (Sample medium
of e = 0.72)
where u, p, and l are the velocity vector, pressure field and fluid viscosity, respec-
tively. The procedure for solving the three dimensional Stokes flow problem is rather
complicated in the vicinity of a porous structure [51, 54]. In the present study, a
staggered marker-and-cell (MAC) mesh was involved, the pressure was defined at
the cell center, and the velocity components were defined along the corresponding
face boundaries. The resulting linear system of equations was solved by a successive
over-relaxation (SOR) method. An initial estimate of p was determined by solving a
Laplace equation. Next, the velocity vector u was calculated from the corresponding
momentum balance and the continuity equation r u ¼ 0: The pressure was cor-
rected using an artificial compressibility equation of the form:
dp
¼ br u ð7:56Þ
dt
Essentially, the method adds an artificial density time derivative that is related
to the pressure by an artificial equation of state p = bq, where b is an artificial
compressibility factor. Similar to the compressible momentum equation, c = b1/2
is an artificial speed of sound and for stability reasons during the iterative pro-
cedure, its magnitude should be such that the respective artificial Mach number,
!1=2
R X
2
M ¼ max ui is low (M 1), where R is the relevant Reynolds
c D i
number. In the limiting case of R ? 0, which is the present case, any finite value
of b should meet this criterion. Thus, b = 1 was chosen here, although it has been
evident that the exact value cannot have any effect on the final (steady state) results
since at steady state the artificial density time derivative is zero. The above steps
were repeated until convergence was reached. This numerical scheme to determine
the velocity field has been widely validated in terms of both the velocity field and
the corresponding permeability [2, 20].
Regarding boundary conditions, one of the six surfaces of the simulated domain
was considered an inlet, while the opposite surface was assigned as an outlet. The
other four domain surfaces were assumed to be spatially periodic, both in geo-
metrical and physical (flow and transport) terms.
The differential equations for the flow-field were integrated with the following
initial and boundary conditions:
uðr; t ¼ 0Þ ¼ 0 ð7:57Þ
Pðr; t ¼ 0Þ ¼ P0 ð7:58Þ
uðr ¼ void space of the inlet; tÞ ¼ uin ð7:59Þ
r uðr ¼ void space of the inlet; tÞ ¼ 0 ð7:60Þ
Pðr ¼ void space of the inlet; tÞ ¼ Pin ð7:61Þ
Pðr ¼ void space of the outlet; tÞ ¼ P0 ð7:62Þ
uðr ¼ void/solid interfaces; tÞ ¼ 0 ð7:63Þ

Initial conditions (7.57) and (7.58) indicate a quiescent mixture in the void
space of the porous medium, while boundary conditions (7.59–7.61) impose a
uniform (plug) flow at the inlet. Obviously, non-zero values appear only on the
surfaces of the void space, since velocity and pressure equal zero at the solid
phase. Boundary condition (7.62) imposes the necessary pressure gradient for the
Stokes flow, while Eq. 7.63 expresses the non-slip condition at the fluid/solid
interfaces throughout the porous medium.
The transport equation (7.42) was integrated with the following initial and
boundary conditions:
Sc ðr; t ¼ 0Þ ¼ 1016 ð7:64Þ
Sc ðr ¼ void space of the inlet; tÞ ¼ ðSc Þin ð7:65Þ
r Sc ðr ¼ void space of the inlet; tÞ ¼ 0 ð7:66Þ
Sc ðr ¼ void space of the outlet; tÞ ¼ ðSc Þout ð7:67Þ
r Sc ðr ¼ void/solid interfaces; tÞ ¼ 0 ð7:68Þ

Fig. 7.15 Evolution of the 10.00

average droplet size for two
droplet size x 10 (m)

different sets of Sc values
7
1.00
0.10 d 32
d 41
0.01
0.1 1.0 10.0 100.0 1000.0
time (s)
The initial condition (7.64) indicates a very low initial droplet population, while
boundary conditions (7.65–7.67) impose a gradient for the Sc. Equation 7.68
expresses neutral behavior at the fluid/solid interfaces, while spatial periodicity is
again assumed for the domain’s other outer surfaces. Note that the above condi-
tions are independent of the c values.
7.3.4 Algorithm Validation and Findings
To examine the validity of the solution scheme, the algorithm was tested against its
consistency. More precisely, different sets of Sc values should lead to similar
distribution parameters, i.e. to the same mean droplet size, by using Eq. 7.36. Such
a test was performed by carrying out simulations for two specific sets of Sc values,
namely S1, S2, S3 and S4. These distributions lead to mean droplet size through
13
S3 S4
d32 ¼ and d41 ¼ ; which should theoretically be the same. The mesh and
S2 S1
time step size were the same in both cases.
Figure 7.15 depicts this comparison and the two curves represent the evolution
of d32 and d41 through time. Since the results are very similar, this plot proves the
self-consistency of Sc modeling.
Since droplet size is represented by d32, the solution scheme is hereafter used
only for c = 2 and c = 3, thus the relevant source terms become:
Z1 qffiffiffiffiffiffiffiffiffiffiffiffi
ð 2Þ d2 3
hbr ¼ Nf ðdÞ 1 n PðdÞ dd ð7:69Þ
sbr ðdÞ
0
2
ð2Þ 6u
hcl ¼ 0:740079 kcoll urel ðdeq Þ Pcoal ðdeq Þ deq ð7:70Þ
p
ð3Þ
hbr ¼ 0 ð7:71Þ
ð3Þ
hcl ¼ 0 ð7:72Þ
Table 7.4 Parameters used in the simulations

Parameter Units Expression/value Reference/comment
3
Nf ðdÞ (–) d [32]
2:69f1 ðkÞ
dcr
sbr ðdÞ (s) 2gc d [32]
f2 ðkÞ
r
f1 ðkÞ, f2 ðkÞ (–) • 1 for the current case
• Other values/expressions for
the specific case studied
dcr (m) Eq. 7.44 Calculated at each point and time
step
r (N m-1) 0.0728 Since the current work is purely
theoretical, the continuous
phase is supposed to be water
Cacr (–) 1 [72]
gc (centipoises) 0.89 Since the current work is purely
theoretical, the continuous
phase is supposed to be water
PðdÞ (–) 2 ! [52]
1 ln d ln d
pffiffiffiffiffiffi exp
dr^ 2p 2r^2
r
^ (–) 0.1 [15, 52]
d (m) 3 x 10-8 [15, 52]
kcoll urel ðdeq Þ (m s-1) 0:6667u [15, 52]
Pcoal ðdeq Þ (–) Follows the log-normal [15, 52]
distribution accordingly to
the droplets’ diameters
deq (m) 3c 1
[3]
6u
pSc
u (–) 4 3 1 Calculated at each time step
3 pðd32 Þ 0:030:030:02 n
-3
n (m ) whole algorithm and Eq. 7.33 Calculated once from the solution
for S0
For the results presented here, the simulation’s parameters are presented in
Table 7.4.
The time evolution of the average characteristic droplet size, d32 , is presented in
Fig. 7.16 up to the critical time where the system reaches the steady-state situation.
The maximum size is observed at t = 550 s, mainly due to competition between
the hydrodynamic phenomena, i.e. break-up and coalescence. At t = 900 s, the
droplet size distribution levels off taking a rather constant value for longer time
scales.
This dynamic behavior of the system is clarified in Fig. 7.17 which shows the relative
significance of the source terms. The individual
n source
o terms have beenn normalized
o by
ðcÞ ðcÞ ðcÞ ðcÞ ðcÞ ðcÞ
their maximum value, i.e. hbr ¼ hbr =MAX hbr ; hcl ¼ hcl =MAX hcl ; to fit the
same graph. However, their initial values are non-zero because the model assumes
Fig. 7.16 Characteristic 10.0

average droplet size (d32)
droplet size x 107 (m)

7.5
5.0
2.5
0.0
0 100 200 300 400 500 600 700 800 900 1000
time (s)
Fig. 7.17 Relative 1.0

break-up
significance of source terms
normalized source terms (-)
0.8 coalescence
0.6
0.4
0.2
0.0
0 100 200 300 400 500 600 700 800 900 1000
time (s)
that droplet size evolution also initiates from a non-zero value (see Fig. 7.16).
As shown in Fig. 7.17 during the first 550 s, coalescence dominates over break-up as
the particles are quite small, therefore, droplet size increases. At t = 550 s the
contributions of break-up and coalescence are balanced, thus particle size reaches a
peak. After 550 s, break-up becomes more significant than coalescence, thus particle
size decreases until t = 900 s. It is worth noting that the time period considered in the
simulation is critical for result validity. Obviously, the simulation time is strongly
dependent on the specific medium’s geometry and the flow characteristics. The
simulation time for the above presented results was t = 1,000 s for the given porosity
(e = 0.72) and velocity field (u = 1 m/s). A more detailed study of the influence of
these parameters follows.
Although porosity has not been directly included in the specified transport
equations, it is a crucial parameter for the microstructure transport phenomena
since it affects significantly the local flow field and mass transport regime through
the void space in the porous medium. The effect of porosity is presented in
Figs. 7.18 and 7.19. The time instant when droplet diameter reaches the maximum
value, tmax, is presented in Fig. 7.18 as a function of the medium’s porosity. It is
notable that tmax varies little with porosity, and this is because droplet size is much
smaller than mean pore diameter, leaving therefore the droplets unaffected by the
presence of the solid phase, even for the lowest porosity value. Furthermore, lower
porosity values correspond to higher coalescence probabilities because the lower
Fig. 7.18 Effect of porosity 700

on the time when MAX{d32}
is observed 600
time (s)
500
400
300
0.4 0.5 0.6 0.7 0.8 0.9 1.0
porosity (-)
Fig. 7.19 MAX{d32} as a 10

function of porosity
8
MAX[d 32 ] x 10 7 (m)
0
0.4 0.5 0.6 0.7 0.8 0.9 1.0
porosity (-)
the porosity, the narrower the path for a bubble to escape without any collision
and, therefore, the higher the coalescence source term. In addition, break-up is not
ðcÞ
affected significantly by porosity, because hbr is non-zero only for diameters
larger than a critical diameter value, whose order of magnitude is 10-7 m which
corresponds to porosity values of order of magnitude 10-5 for feasible charac-
ðcÞ
teristic lengths. Actual porosity values range from 0.5 to 0.9, assuring that hbr is
not practically affected by the porosity. Therefore, the only influence on decreasing
ðcÞ
porosity is the enhancement of hbr which corresponds to earlier times for maxi-
mum diameters with lower porosity. However, this influence on tmax does not
correspond to any significant change in the absolute value of maximum diameter,
as shown in Fig. 7.19. Indeed, maximum values of source terms remain rather
unaffected by porosity, leading to a fairly constant maximum diameter, despite the
fact that the time when these maximum values occur varies with porosity, as
discussed previously.
Finally, Figs. 7.20 and 7.21 show the effect of flow intensity on droplet size. As
depicted in Fig. 7.20, an increase in inlet velocity value corresponds to delayed
appearances of maximum diameters. Indeed, as velocity increases i.e. convection
dominates diffusion, it is easier for droplets to collide, therefore, the coalescence
source term increases. Furthermore, a very low increment of the break-up source
term is also expected because the higher the velocity, the easier for a droplet to
Fig. 7.20 Effect of inlet 700

velocity magnitude on the
time when MAX{d32} is 600
observed
time (s)
500
400
300
0 .1 1.0 10.0
inlet velocity magnitude (m/s)
Fig. 7.21 MAX{d32} as a 10

function of inlet velocity
magnitude 8
MAX[d 32 ] x 10 7 (m)
0
0.1 1.0 10.0
inlet velocity magnitude (m/s)
reach and interfere with the solid phase. As the phenomenon dominantly influ-
enced by the flow is coalescence, maximum diameter is observed later for higher
convective regimes. Again, and for the same reasons as those discussed above, this
does not seem to affect significantly the maximum diameter (see Fig. 7.21).
7.4 Fluid Flows Through Porous Media in Fuel Cells
During the last decades, gradually increasing global energy demands, rising
environmental and health concerns, and depletion of natural resources, have
intensified the need to discover new energy production technologies. These new
technologies should utilize current fuels more effectively by raising the efficiency
of the energy production units, and be more environmentally friendly by reducing
CO2 emissions. Fuel cells seem to be one potential answer to this goal and offer
energy policy makers energy independence under fully environmentally friendly
conditions.
Fuel cells are electrochemical devices that directly convert chemical energy
from gaseous fuels into electricity by overcoming Carnot limitations. Theoreti-
cally, they can reach high efficiencies and, unlike batteries, are able to provide a
continuous supply of electric power when replenished with fuel. Furthermore, fuel
cells are very compact units without moving parts, therefore ensuring silent
operation, fewer material design constraints and reasonable portability.
7.4 Fluid Flows Through Porous Media in Fuel Cells 207
Of the fuel cells available, SOFCs present considerable advantages due to their
high operating temperature. These high temperatures favor electrochemical reac-
tion kinetics permitting the use of low cost metal catalysts and they promise higher
efficiencies when recovering the high energy waste heat (electrical efficiencies of
45–50% can be achieved). Furthermore, SOFC operation allows external or even
internal reforming reactions to occur, thus fuels such as carbon monoxide and
hydrocarbons considered poisonous to low temperature fuel cells, can be used with
minimal fuel processing [1, 33, 48, 55, 76]. Despite these benefits, they do present
some drawbacks regarding the materials used caused mainly by the high tem-
peratures, such as thermal expansion of the different fuel cell components and
stability issues due to mechanical deficiencies and catalyst deactivation. However,
much effort has been given to fabricating new, cost effective materials with
improved physical and electrochemical characteristics that would enhance both
SOFC performance and stability [95].
Due to the variety of their advantages, SOFCs are considered ideal candidates
for various applications. As mentioned previously, they operate at high tempera-
tures and the waste thermal energy (waste heat) produced can be recovered in
combined heat and power systems (CHP) to produce electricity and power in
large-scale distributed power generation systems as well as in small-scale domestic
heat and power production units, taking full advantage of the fuel used. SOFCs are
also considered to be useful as auxiliary power units (APUs) for various electrical
systems in cooking and transportation, such as in vehicle air conditioning and
portable electronics, i.e. cell phones and personal computers. In addition, SOFCs
can be used for the chemical cogeneration of electricity and chemical compounds
with the use of appropriate materials. They could also be an alternative choice for
remote distributed power generation either in areas with no grid supply, such as
isolated islands, or when local power production is necessary, such as small power
units [19, 63, 67, 84].
According to their temperature operation, SOFCs can be classified into low-,
intermediate- and high-temperature types with operating temperatures between
773–823 K, 823–1,073 K and 1,073–1,273 K, respectively. Regarding geometri-
cal aspects and design, one of the most commonly used geometries is the tubular
one, initially introduced by Siemens-Westinghouse [86]. The operation principal
here is that the tubular fuel cell comprises a unique tube, the air is supplied
internally, and the fuel is thought to flow in the surroundings. Another design is the
planar geometry in which the PEN assembly is surrounded by the collectors
[43, 74]. According to the flow direction of air and fuel, planar geometry SOFCs
can be subdivided into (1) co-flow planar SOFCs, where air and fuel flow in the
same direction, (2) counter-flow, where they flow in opposite directions, and (3)
cross-flow, where they flow perpendicular to each other [25]. Tubular and planar
geometries can be both electrolyte or electrode (anode or cathode) supported
[4, 92].
Fig. 7.22 A typical SOFC

configuration
7.4.1 Solid Oxide Fuel Cell Configuration
A typical SOFC consists of an anode and a cathode compartment between which a

catalyst (electrolyte) layer exists. Air flow enters one gas channel and the other
channel is continuously supplied by a feeding mixture (usually a hydrogen-rich
mixture containing some higher hydrocarbons with carbon monoxide and dioxide).
The electrolyte has high ionic and minimum electronic conductivity and it pre-
vents fuel crossover and mixing of the gas supplies. The fuel is electrochemically
oxidized by the oxygen ions, O2 ; formed at the cathode electrode by the oxygen
reduction reaction. These ions migrate through the electrolyte to the anode. It is
evident that the same number of electrons must be transferred through the elec-
trolyte in order to preserve equilibrium between the electrodes. The electrons
released at the anode compartment flow via an external electrical circuit. In most
cases, the principle operation of a typical fuel cell incorporates the formation of
water from hydrogen and oxygen, and the production of electricity and heat.
A typical configuration of an SOFC is presented in Fig. 7.22.
7.4.2 Electrochemical and Surface Reactions
The most significant advantage of SOFCs compared to other fuel cell types, is that
they allow wide fuel flexibility due to high operational temperature and the
materials used. As a result, apart from hydrogen (H2), SOFCs can be fed with
carbon monoxide (CO) and synthesis gas (H2 and CO of various ratios), as well as
natural gas and hydrocarbon fuels (coal fuels). The electrochemical reactions take
place at the interface between the catalyst, the ionic conductor, and the gas phase–
the so-called three-phase boundary layer (TPB). The electrochemical oxidation
reactions of H2 and CO occurring at the anode electrode can be described by:
H2 þ O2 ! H2 O þ 2e ð7:73Þ
CO þ O2 ! CO2 þ 2e ð7:74Þ

Experiments have shown that the electrochemical oxidation rate of H2 is sig-
nificantly higher (1–2.5 times) than that of CO [18, 71, 91], thus a large number of
existing models neglect the contribution of CO in the production of current
density.
With the use of appropriate anode cermets, such as Cu-YSZ and Ni-YSZ, direct
oxidation of hydrocarbons and/or alcohols is possible in SOFCs [47, 77, 107].
In general, assuming that the fuel can be expressed as Cn Hm Op where p = 0 or
p = 1 denoting hydrocarbon or alcohol fuels, respectively, the electrochemical
oxidation reactions can be expressed as:
Cn Hm þ ð2n þ 0:5mÞO2 ! nCO2 þ 0:5mH2 O þ ð4n þ mÞe ; where p ¼ 0

ð7:75Þ
Cn Hm O þ ð2n þ 0:5m 1ÞO2 ! nCO2 þ 0:5mH2 O þ ð4n þ m 2Þe ; where p ¼ 1

ð7:76Þ
Atmospheric air (humidified or not) and, rarely, pure O2 is often introduced into
the cathode gas channel. At the cathode electrode, the electrochemical oxygen
reduction taking place can be written as:
0:5O2 þ 2e ! O2 ð7:77Þ

and in case of Cn Hm Op fuels the reaction is:
ðn þ 0:25mÞO2 þ ð4n þ mÞe ! ð2n þ 0:5mÞO2 ; where p ¼ 0 ð7:78Þ
ðn þ 0:25m 0:5ÞO2 þ ð4n þ m 2Þe ! ð2n þ 0:5m 1ÞO2 ; where p ¼ 1

ð7:79Þ
Apart from the electrochemical reactions, in practice, numerous homogeneous
or heterogeneous chemical reactions may appear in an SOFC. The most common
reactions included in simulation models are presented below. In general, the water
gas shift reaction (WGS):
CO þ H2 O $ CO2 þ H2 ð7:80Þ
is applied when both hydrogen and carbon monoxide are present, and is usually
assumed to be in equilibrium.
To effectively use hydrocarbon fuels they should be partially reformed to H2
and CO by the internal or external steam reforming reactions as follows [1]:
Cn Hm þ nH2 O ! nCO þ ð0:5m þ nÞH2 ; where p ¼ 0 ð7:81Þ
Fig. 7.23 Simulated SOFC geometry
Cn Hm O þ ðn 1ÞH2 O ! nCO þ ð0:5m þ n 1ÞH2 ; where p ¼ 0 ð7:82Þ

Additionally, the presence of hydrocarbons may lead to carbon formation. To
study the adverse effects of carbon on catalyst deactivation, the Boudouard reac-
tion [7, 41, 66, 96, 104]
2CO $ C þ CO2 ð7:83Þ
and cracking reactions (where applicable)
Cn Hm $ nC þ 0:5mH2 ð7:84Þ
may be introduced.
7.4.3 Transport Phenomena in Gas Channels
The fundamental transport phenomena occurring in gas channels are flow and heat
transfer, mass transport and charge transfer. These are described by the equations
of continuity, momentum and species (neutral or charged) conservation equation.
Therefore, the velocity field, temperature profile, gas composition and electric
potential distribution of the fuel cell can be calculated. For better insight on the
processes occurring, fuel cell geometry can be divided into two general regions.
The first is the non-porous region, which refers to the gas channels, and the second
includes all the porous parts of the fuel cell i.e. anode and cathode electrode, the
dense electrolyte and even the current collectors. A two-dimensional cut of the fuel
cell is depicted in Fig. 7.23, where the dimensions have been chosen in accordance
with Ramakrishna et al. [83].
Firstly, for non-porous regions the mass conservation equation (continuity
equation) can be written as:
oq
þ r ðqUÞ ¼ 0 ð7:85Þ
ot
where U is the velocity vector, q is the fluid density, and t is time.
Velocity in fuel cell gas channels is usually quite low, so the assumption of
laminar flow can be considered accurate. As a result, for compressible fluids and
neglecting gravitational effects, the momentum equation is:
o
ðqUÞ þ r ðqU UÞ ¼ rP þ r s ð7:86Þ
ot
where P is the pressure, and s is the shear stress tensor.
By neglecting radiation, the energy conservation equation becomes:
o oP _
ðqhÞ þ r ðqUhÞ ¼ r ðkrTÞ þ þQ ð7:87Þ
ot ot
where h is the total enthalpy of the fluid, k is the thermal conductivity of the
mixture, and Q_ describes the additional thermal sources due to exothermic/
endothermic reactions. The mass fraction of each gas species can be computed by
the species conservation equation as follows:
o
ðqYi Þ þ r ðqUYi Þ ¼ r Ji þ x_ i ð7:88Þ
ot
where Yi are the mass fractions of the ith chemical species, x_ is the production/
destruction rate of the i th chemical species in gas phase due to homogeneous
reactions, and Ji is the species i mass diffusion flux.
Several approaches are used in the literature to calculate the mass diffusion flux,
but this is beyond the scope of this analysis. The simplest model used to calculate
the species diffusion flux is Fick’s law:
Ji ¼ qDi rYi ð7:89Þ
where Di is the mass diffusion coefficient of species i in the mixture.

Although several models for the estimation of diffusivity have been presented
in the literature, in the present study the mass diffusion coefficient for gas channels
is obtained through the Schmidt number, Sc, according to the relation:
l
Di ¼ ð7:90Þ
Sc
where l is the viscosity of the gas species.
The above differential equations are all strongly coupled and should be integrated
along with the appropriate boundary conditions, which are application dependant.
7.4.4 Transport Phenomena in Porous Media
Porous materials are widely used in SOFC applications and particularly at the
anode and cathode electrodes, since they present large catalytic surface areas.
All the transport phenomena described above should be modified slightly to
incorporate the porous nature of the electrodes and simulate the transport processes
since porous material structure implies mass transport limitations.
Analytically, the mass conservation equation in a porous medium can be

written as:
o
ðeqÞ þ r ðeqUÞ ¼ 0 ð7:91Þ
ot
where e is the porosity of the medium representing the volume occupied by the
pores to the total volume of the porous media.
In porous regions, the momentum conservation equation becomes:
o e2 l
ðeqUÞ þ r ðeqUUÞ ¼ erP þ r ðesÞ þ U ð7:92Þ
ot j
where l is the viscosity of the fluid, and j is the permeability representing the
square of the volume to surface area ratio of the porous material. Note that the last
term of Eq. 7.92 represents Darcy’s law and describes the superficial velocity in
the porous medium [23].
The energy conservation equation inside a porous medium, written in terms of
enthalpy, takes into account thermal convection and conduction, species diffusion
contributions, and chemical reaction (neutral or electrochemical) effects on tem-
perature. It can be written as:
! 2
o X
Ng
dP ji j
ðeqhÞ þ r ðeqUhÞ ¼ r keff rT þ J i hi þ esrU þ e ji g þ
ot i¼1
dt r
ð7:93Þ
where keff is the effective thermal conductivity of the mixture, Ng is the total
number of gas species in the system, and hi is the enthalpy of the ith species. The
temperature gradient is significantly affected inside the porous material by the
electrochemical reactions and consequently both Joule heating and electrical work
affect energy transfer. This effect is described by the last two terms of the energy
transfer equation.
By taking into account the combination of porous and solid parts of the porous
medium, the effective thermal conductivity of this medium can be defined as [35]:
1
keff ¼ 2kS þ ð7:94Þ
e 1e
þ
2kS þ kF 3kS
where kF is the thermal conductivity of fluid parts of the porous medium, and kS is
the thermal conductivity of solid parts of the porous medium.
The conservation equation of the i th gas species is given by the relation:
o
ðeqYi Þ þ r ðeqUYi Þ ¼ r Ji þ x_ i ð7:95Þ
ot
where Fick’s model can also be applied, given as:

Ji ¼ qDi;eff rYi ð7:96Þ
where Di;eff is the effective mass diffusion coefficient of species i.

To consider the porosity, e, and tortuosity, s, of the porous medium, a number
of correlations can be found in the literature such as the Daggan model. However,
the most applicable is the Bruggeman correlation, defined as [79]:
Di;eff ¼ es Di ð7:97Þ
To calculate species diffusion fluxes, apart from Fick’s model applied here,
several other models cam be used such as the Stefan-Maxwell model for multi
component systems, and the Dusty Gas model (DGM), which incorporates both the
Stefan-Maxwell formulation and the Knudsen diffusion [11, 94].
Regarding charge conservation in conducting materials, the sum of all current
flows should be zero based on electro-neutrality. Thus, the current conservation
equation is:
ri¼0 ð7:98Þ
where i is the current density vector

However, based on Newman and Tabias [75], charge transport consists of elec-
tronic and ionic phase transports. During electrochemical reactions electrons are
either transferred from the pores (ionic phase) to the solid region (electronic phase) or
vice versa, i.e. electron transfer is expressed as the transfer current, jT, where:
r iF ¼ r iS ¼ jT ð7:99Þ
where iF is the current density vector flowing through the pores (ionic phase), and
iS is the current density vector flowing through the solid parts of the porous
medium (electronic phase). By applying Ohm’s law, the transfer current yields:
r ðrF ruF Þ ¼ r ðrS ruS Þ ¼ jT ð7:100Þ
where rF and rS are the ionic phase conductivity and the solid phase conductivity,
respectively, and uF and uS are the ionic potential of the fluid and the electric
potential of the solid, respectively.
For the non-conducting electrolyte Eq. 7.100 becomes:
r ðrF ruF Þ þ r ðrS ruS Þ ¼ 0 ð7:101Þ
and the current transfer jT can be defined through the Butler-Volmer equation as:
YN
j0 VA aa F ac F
jT ¼ QN aie exp g exp g ½Ci aie ð7:102Þ
i¼1 Ci;ref
RT RT i¼1
where i.e indexes the electrochemical reactions, j0 is the exchange current density,
aa and ac are the anodic and cathodic charge transfer coefficients as determined by
the Tafel plots respectively, N is the total number of reacting species, ½Ci is the
near wall molar concentration of the ith reacting species or expressed in mass
Yi
fractions as ½Ci ¼ q , where Mi is the molar weight of the i th species, ½Ci;ref
Mi
is the molar concentration at a reference state at the inlet, and ak is the concen-
tration exponent. The overpotential, g, can be expressed as the potential difference
between the solid phase and the porous phase potential:
g ¼ uS uF ð7:103Þ
In surface reactions occurring at porous/catalyst interfaces, the volumetric
reaction rate (production/destruction of a gas species i) can be computed assuming
a balance between the reaction flux and the diffusion flux on the surface [50]. This
rate is given as:

Yi YP;i A
x_ ¼ qDi ð7:104Þ
d V eff
where YP,i is the mass fraction of the ith species in the pore fluid, ½A=V eff denotes
the effective surface to volume ratio of the catalyst and represents the catalyst load,
and d is the diffusion length scale. For electrochemical reactions, the production/
destruction rate is expressed through the current transfer by the relation:
jT
x_ ¼ ðbPR bR Þ ð7:105Þ
F
where bPR and bR are the stoichiometric coefficients of the products and reactants,
respectively.
7.4.5 Simulations
Planar SOFC geometry was used for the three-dimensional approach. This
geometry consists of seven separated volumes: the fuel and air channel where the
corresponding mixtures are introduced, the porous anode and cathode electrodes
where the reactions occur, the dense electrolyte through which oxygen ions
migrate to reach the anode electrode, and finally the anode and cathode contacts. In
the present study, both H2 and CO were assumed to react with oxygen ions, in
accordance with Eqs. 7.73 and 7.74. Regarding the Butler-Volmer equation
(Eq. 7.102), the exchange current density was set to j0 ¼ 1011 Am3 for the anode
and j0 ¼ 1010 Am3 for the cathode, where the charge transfer coefficients for both
electrodes were aa ¼ 0:7 and ac ¼ 0:7. The electrochemical reactions were
assumed to occur only in the porous electrodes and precisely on the three-phase
boundary layer. It was also assumed that no other reactions occurred either on
the porous interface or in the bulk phase. The feeding stream was assumed to be a
reformate mixture of 9.6% H2, 26.0% CO, 2.16% CO2 and 42.8% H2O on w.t.%
basis, and typical atmospheric air composition was applied at the cathode channel
(23.3% O2 and 76.7% N2 on w.t. basis).
Regarding the boundary conditions, the mass flow rates for the anodic and
cathodic mixtures were assumed to be constant and equal to 3.0 9 10-7 kg/s and
4.0 9 10-6 kg/s, respectively, while pressure of 1 atm was set at the inlets and
outlets. Considering no accumulation, zero mass flux was set at the walls and
outlets. Additionally, preheated mixtures of 1,173 K entered the fuel cell, while
zero heat flux was set at all the other boundaries. Finally, the constant value of zero
overpotential was set to the anode contact, while the overpotential of the cathode
contact was set to -0.7 V.
Both electrodes were modeled as isotropic porous media and their physical
characteristics are as follows [58, 83]: porosity was e ¼ 0:4 for the anode, e ¼ 0:01
for the electrolyte, and e ¼ 0:5 for the cathode; permeability was j ¼ 1012 m2 for
the anode and cathode, and j ¼ 1018 m2 for the electrolyte; solid phase electrical
conductivity was rS ¼ 10; 000 Xm1 for the anode, rS ¼ 1020 Xm1 for the
electrolyte, and rS ¼ 7; 700 Xm1 for the cathode; ionic phase electrical con-
ductivity was rF ¼ 10 Xm1 for both the electrodes and the electrolyte.
To calculate the density, q, it was assumed that the gases followed the ideal gas
law, and the kinetic theory of gases was used to estimate viscosity, l. Specific
heat, cp, was derived by fittings to experimental JANNAF curves, and mass dif-
fusivity was calculated by imposing 0.7 to the Schmidt number.
To achieve residual values less than 10-4 for all the quantities and assuming
steady state conditions, the numerical solution of the transport problems was
obtained by the commercial package CFD-ACE+ by ESI-Group, which is based
on the finite volume method, together with all the appropriate boundary conditions.
The three-dimensional fuel cell was discretized in space using a structured grid of
33, 516 cells.
The typical distributions for the main physical quantities (velocity, temperature
mass fractions and overpotential) obtained for constant gas inlet mixture and
temperature for both fuel and air mixtures are presented in Figs. 7.24, 7.25, 7.26,
7.27, 7.28, 7.29. These are contour plots in a two-dimensional cut in the middle
plane of the fuel cell. Figure 7.24 shows the developed velocity profile, which was
parabolic for both flow channels, satisfying the non-slip conditions applied on the
walls. Due to the mass flow rates imposed, high velocity values can be seen in
the air channel and lower velocity values in the fuel channel, while zero velocities
are observed in the porous media. The highest value observed was approximately
5 m/s, which can be considered quite high, but necessary to satisfy the energy
demands of the system in terms of cooling, and maintain active chemical reactions.
Fig. 7.24 Velocity profile
Fig. 7.25 Mass fraction profiles for (a) hydrogen and (b) carbon monoxide
The molar fractions of the major reactants and products along the fuel cell are
presented in Figs. 7.25, 7.26. Figure 7.25 shows that hydrogen and carbon mon-
oxide are consumed at the anode channel and anode electrode. Furthermore, their
fractions decrease with distance, due to the electrochemical reaction controlling
the system. The consumption of both species is accompanied by relative pro-
duction of both CO2 and H2O, in accordance with the relative reaction rates.
A small depletion of oxygen can be observed at the cathode and the cathode
channel (Fig. 7.27). This can be attributed to its participation in the electro-
chemical reactions. The O2 gradient is fairly small since the air mass flow rate is
high enough to avoid oxygen depletion that would lower fuel cell performance,
while the highly convective regime does not favor the electrochemical reactions.
Temperature varies along the fuel cell length from 1,173 K imposed at the inlet
boundary to 1,256 K at the outlet (Fig. 7.28). This temperature variation is
strongly affected by the electrochemical reactions occurring at the anode electrode.
Hydrogen and carbon monoxide electro oxidation reaction is exothermic, thus a
certain amount of heat is released in the cell and transferred locally due to con-
vection and conduction. Consequently, temperature increase can be observed
along the length of the cell and it is evident that convectional effects prevail at the
air channel due to the developing velocities. Finally, the over potential profile is
depicted in Fig. 7.29. The over potential at the anode electrode is higher compared
to that at the cathode electrode due to fuel depletion at the anode channel. It should
Fig. 7.26 Mass fraction profiles for (a) carbon dioxide and (b) steam
Fig. 7.27 Mass fraction profile for oxygen
also be mentioned that over potential increases along the anode electrode and this
rise is followed by the reduction of active species mass fractions i.e. H2 and CO.
7.5 Multi Phase Transport in Porous Media
Numerous physicochemical processes exist where an aqueous phase coexists and/

or interacts with a non-aqueous liquid phase in a porous medium. Many of these
processes have significant importance in terms of industrial and technological
applications, e.g. tracer transport in petroleum reservoirs, chemical contamination
of soils and aquifers, long-term interaction between liquefied or chilled foods with
packaging materials. The description of such processes often relies on mathe-
matical models as experimentation can be expensive or difficult.
In the presence of porous media, the need for a realistic description of the
structure of a porous medium significantly increases the mathematical complexity
of a model. However, elements of the microstructure must be captured when
moving from the pore level to the macroscopic level, where process performance
must often be studied. A number of techniques for up scaling from the pore-scale
to the macroscopic scale in porous media have been developed. Of specific interest
here is volume-averaging [14, 78, 105, 106], which is very useful when pore scales
Fig. 7.28 Spatial distribution for temperature
Fig. 7.29 Overpotential profile
and macroscopic scales are separated. Starting from the relevant differential
equations at the pore level and using the spatial averaging theorem, after several
mathematical manipulations based on scale separation [102, 103], one is lead to
unit-cell problems to estimate macroscopic quantities [80, 81]. Spatial averaging
techniques allow the integral effect of the influence of pore geometry on transport
to be captured in the mathematical formulation.
Typically, macroscopic models involve an exchange term to describe the rate of
mass transfer between the two phases. The majority of existing theoretical works
derived such models based on the assumption of the mass exchange equilibrium
between the aqueous and non-aqueous liquid phases [27, 28, 45, 62]. This assump-
tion of infinitely fast diffusion in the non-aqueous phase, simplifies mathematical
modeling and consequent simulations [37, 82, 88]. It is necessary to determine the
validity of this assumption. When the assumption is relaxed, the resulting unit-cell
problem includes mass transport between the two phases (the ‘‘aqueous’’ and the
‘‘non-aqueous’’ liquids), thus accounting for inter-phase diffusion and partitioning.
Its solution provides expressions required for the macroscopic mass-transfer coef-
ficient. Various model pore geometries are used to obtain representative results. An
ab initio calculation is also presented, in which the Stokes equations are solved in
typical pore geometries and compared with the volume-averaging results.
7.5.1 Theoretical Background
Consider a multi-phase domain consisting of a flowing aqueous phase (b-phase),

an immobile non-aqueous liquid phase (c-phase) and a solid phase (r-phase), as
schematically depicted in Fig. 7.30.
7.5 Multi Phase Transport in Porous Media 219
σ-phase
γ-phase
β-phase
volume averaging space

Fig. 7.30 Schematic depiction of a typical representative volume
The assumption of a motionless c-phase is a good approximation for several

applications, such as water or gas tracer transport in hydrocarbon reservoirs where
the oil phase rests practically immobile (e.g. residual saturation), slow dissolution
of NonAqueous Phase Liquids (NAPLs) in underground porous formations, and
packing porous pellets with a wetting fluid. To relax this assumption, one should
backtrack to techniques (e.g. Lattice-Boltzmann) capable of solving the flow
problem while simultaneously determining the dynamic distribution of fluid phases
in the considered porous domain. This approach, however, requires significant
computational resources [10]. A partitioning tracer is advected by the flowing
b-phase and partitioned in the immobile phase into which it is diffusing. It is
assumed that the solid phase is physicochemically neutral, i.e. the tracer is neither
adsorbed nor reacts with the r-phase. The governing processes in the b-phase are
diffusion and advection, and diffusion in the c-phase. The mass exchange at the bc
interface is characterized by the diffusion and partitioning properties of the tracer.
The macroscopic modeling of the processes is typically described by the following
advection–dispersion-reaction equations:

o Cb 2
1
eb þ hvi r Cb ¼ eb Db r Cb a Cb Cc ð7:106Þ
ot K

o Cc 2
1
ec ¼ e c D c r C c þ a Cb Cc ð7:107Þ
ot K
where eb, ec denote the volume fraction of the b- and c-phase, respectively, v is the
velocity vector in the b-phase, Db ; Dc are the macroscopic dispersion coefficients
in the b- and c- phase, respectively, and K is the partitioning coefficient. Brackets
denote spatial averages.
The knowledge of several parameters (e.g., of the mass-transfer coefficient and
the dispersion tensor) is required to solve the above macroscopic equations. As
previously noted, volume averaging can provide these expressions when scale
separation exists. The main theoretical work in this field is that of Quintard and
Whitaker [82], which presents methods for the calculation of the mass-transfer
coefficient and the dispersion tensor under the assumption of infinitely fast dif-
fusion in the c-phase. This corresponds to an almost constant concentration profile
in the c-phase and permits decoupling of the mass transport problems in the two
phases. Mass exchange between the two phases has been also modeled by various
researchers, considering either a controlling diffusive process macroscopically
described by first order kinetics, or an advection process [5, 9, 22, 38, 65, 96].
What is proposed here is a more rigorous extension that considers non-equilibrium
partitioning of the tracer and mass transfer in both phases. Three-dimensional
model pore geometries were considered in the following numerical illustrations,
where the immobile c-phase may be distributed as either wetting films or non-
wetting aggregates (blobs). The objective is to estimate the mass transport coef-
ficient and investigate the effect of different structural and physicochemical
parameters under non-equilibrium partitioning. Otherwise, the volume averaging
approach is conventional and consists of the algorithm presented in Sect. 5.5.
7.5.2 Formulation of the Problem
The pore-level transport of the tracer in the b-phase is described by the diffusion–
advection equation
oCb
þ r ðvCb Þ ¼ Db r2 Cb ð7:108Þ
ot
where Cb is concentration, t is time, v is the fluid velocity, and Db is the diffusivity
in the b-phase. Since the c-phase is assumed to be immobile, the diffusion equation
describes the transport of the tracer in that phase:
oCc
¼ Dc r2 Cc ð7:109Þ
ot
where Cc and Dc are concentration and diffusivity in the c-phase, respectively.
Zero-flux boundary conditions apply to the solid–liquid interfaces:
nbr rCb ¼ 0 at Abr ð7:110Þ
ncr rCc ¼ 0 at Acr ð7:111Þ

as the transported species do not absorb or react at the solid. At interface Abc
between trapped and flowing phases, the following conditions apply:
Cc ¼ KCb ð7:112Þ
Db nbc rCb ¼ Dc nbc rCc ð7:113Þ
where K is the partitioning coefficient. Equation 7.112 describes the partition

equilibrium, while (7.113) expresses the flux continuity at the interface. This mass
exchange description is more general than in previous works [82] and does not
assume fast diffusion in the c-phase as used by Quintard and Whitaker [82],
De Smedt and Wierenga [88] and Gvirtzam et al. [37].
Following the volume averaging procedure [82], local concentrations and
velocities are next decomposed into interstitial averages and fluctuations:
b
Cb ¼ Cb þCb0 ð7:114Þ
c
Cc ¼ Cc þCc0 ð7:115Þ
b
vb ¼ vb þv0 b ð7:116Þ
which are subsequently substituted into the governing differential equations.

Invoking separation of scales to discard small terms, linearizing and following
Quintard and Whitaker [82], the representation below can be easily obtained when
assuming an isotropic medium of uniform porosity and constant volume fractions:

b 1 c b
Cb0 ¼ bb r Cb þsb Cc Cb ð7:117Þ
K

0
c 1 c b
Cc ¼ bc r Cc þsc C c Cb ð7:118Þ
K
where bb, bc, sb and sc are closure variables satisfying specific boundary value
problems.
7.5.3 The Closure Problems
It can be readily shown that variable sb of the b-phase satisfies the boundary value
problem [82]:
vb rsb ¼ Db r2 sb e1
b a ð7:119Þ
with boundary conditions:

sc
sb ¼ 1 þ at Abc ð7:120Þ
K
Db nbc rsb ¼ Dc nbc rsc at Abc ð7:121Þ
nbr rsb ¼ 0 at Abr ð7:122Þ
and the compatibility condition:

sb ¼ 0 ð7:123Þ
Similarly, variable sc of the c-phase satisfies the following problem:
Dc r2 sc ¼ e1
c a ð7:124Þ
ncr rsc ¼ 0 at Acr ð7:125Þ

sc ¼ 0 ð7:126Þ
where the mass transfer coefficient, a, is given by:

Z
Db
a¼ nbc rsb dA ð7:127Þ
V
Abc
The above equation can be simplified by introducing the transformation sb ¼

1 þ awbs and sc ¼ awcs ; and in dimensionless form:
sb ¼ 1 þ a fb ð7:128Þ
sc ¼ a fc ð7:129Þ
al2b
where the dimensionless mass transfer coefficient is defined as a ¼ ; and lb is a
Db
characteristic length of the b-phase. Then, in dimensionless notation, the boundary
value problems read as follows: In the b-phase:
Peb v rfb ¼ r2 fb e1

b ð7:130Þ
nbr rfb ¼ 0 at Abr ð7:131Þ
hvib lb
where Peb ¼ : In the c-phase:
Db
0 ¼ dr2 fc þ e1
c ð7:132Þ
ncr rfc ¼ 0 at Acr ð7:133Þ

fc ¼ 0 ð7:134Þ
Dc
where d ¼ : The two problems are coupled at their interface through the
Db
condition:
1
fb ¼ f at Abc ð7:135Þ
K c
In this notation, the mass transfer coefficient becomes:
eb
a ¼ ð7:136Þ
fb
One final substitution will allow for additional insight. Define:

/b ¼ dKec fb ð7:137Þ
and
/c ¼ dec fc ð7:138Þ
Then, the two boundary value problems take the canonical form:
Peb u r/b ¼ r2 /b K in thebphase ð7:139Þ
nbr r/b ¼ 0 at Abr ð7:140Þ
dKec
where K ¼ and
eb
0 ¼ r2 /c þ 1 in the cphase ð7:141Þ
ncr r/c ¼ 0 at Acr ð7:142Þ
/b ¼ /c at Abc ð7:143Þ

/c ¼ 0 ð7:144Þ
The mass-transfer coefficient is simply:

dKeb ec
a ¼ ð7:145Þ
/b
The above scaling relations are checked against analytical solutions and numerical
simulations below. Two different unit cell geometries are used: the first involving
two parallel phases of infinite extent, and the second involving a unit cube with a
disordered phase distribution.
Fig. 7.31 A two- y

dimensional unit-cell for
parallel phases
εβ
y = εγ+
2
lγ
γ-phase
εβ
y=
2
y=0 β-phase lβ L
x
εβ
y=-
2
σ-phase lσ
In the numerical simulations, the velocity field was computed numerically by

solving the Stokes equations:
rp ¼ lr2 v ð7:146Þ
rv¼0 ð7:147Þ
v¼0 at Abr ð7:148Þ
where v, p, and l are the velocity vector, pressure field and fluid viscosity,
respectively. The procedure for solving the three-dimensional Stokes flow problem
involves discretization in terms of cubic elements and was presented in Sect. 3.5.
To numerically solve the transport boundary value problems, a non-uniform finite
differences scheme with upwinding was used for discretization, with the resulting
linear systems of equations solved again using Successive Over-Relaxation
method (SOR).
The solution for simplified geometry of parallel phases assuming one-dimen-
sional flow is presented first. In addition, it is considered that diffusion is signif-
icant only in the direction perpendicular to flow. These assumptions correspond to
a Pe-independent Type II closure problem similar to that of Quintard and Whitaker
[82] for sufficiently fast diffusion. The unit cell is shown Fig. 7.31 and was used to
calculate properties of the Taylor-Aris dispersion problem for a passive tracer.
The boundary value problem can be solved analytically in this geometry,
yielding the following expression for the dimensionless mass transfer coefficient:
eb ec
a ¼ ð7:149Þ
A1 þ A2 þ A3
where the length L of the unit cell is used as characteristic length. (Note that in this
2
^ lb
notation a ¼ a ). The coefficients are, respectively:
L
Fig. 7.32 Analytical and 10

εβ =0.6
numerical results of the
dimensionless mass transfer εγ =0.2
coefficient for parallel phases 1 Analytical solution
Numerical solution
αL2/Dβ
0.1
0.01
1E-3
1E-4 1E-3 0.01 0.1 1 10 100 1000
Λ
" #
1 eb 3 e3
b 2
A1 ¼ þ ec e ð7:150Þ
6K 2 8 b
" 2 e2
#
1 eb eb b
A2 ¼ 1þ þ ec ec e2b ð7:151Þ
2K 2ec 2 4

e4b ec eb eb ec
3
A3 ¼ ec þ ð7:152Þ
3 4 2K
In the limiting case of infinitely fast diffusion in the c-phase ðd ! 1 ) K !
e4b ec
1Þ; the asymptotic results are A1 ! 0; A2 ! 0 and A3 ! ; thus the dimen-
3
3
sionless mass transport coefficient is â ! eb : This result coincides with the ana-
lytical mass transport coefficient for parallel phases found by Quintard and
Whitaker [82] and shown to be independent of Peclet number.
Figure 7.32 shows the dimensionless mass transfer coefficient plotted as a
dKec
function of the governing dimensionless parameter K ¼ ; calculated using
eb
both the analytical approach and numerical simulation. The analytical and
numerical solutions are almost identical. For low values of K, the mass transfer
coefficient increases with K (i.e. with increasing values of d and/or K) in a linear
fashion. At higher values, a reaches a constant value and becomes practically
independent of K. The spatial distribution of the dimensionless scalar sb is shown
in Fig. 7.33. In the limiting case of K ! 1 ) d ! 1 the calculated sb values are
identical to those found by Quintard and Whitaker [82] for fast diffusion in the
c-phase with the same geometry.
Further investigation of the validity of the asymptotic scaling results in this
case, is portrayed in Fig. 7.34, where the dimensionless mass transfer coefficient is
eb
plotted against K for various values of the fraction eb, with the ratio kept
ec
constant.
Fig. 7.33 Numerical results 1.2

Λ infinite
for sb for parallel phases 1.0 Λ =10/3
Λ =1/3
0.8
Λ =0
0.6
0.4
sβ
0.2
0.0
-0.2
-0.4
-0.6
-0.4 -0.2 0.0 0.2 0.4
y
Fig. 7.34 The effect of K on (a)

the mass transfer coefficient, 1.E+02
defined based either on the
characteristic length L (a) or εβ/ε =3
1.E+01
the characteristic length lb
(b) for various values of the
1.E+00
volume fraction and for ε β = 0.15
αL2/Dβ
parallel phases. Static ε β = 0.3

conditions 1.E-01 ε β = 0.6
1.E-02
1.E-03
1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03
Λ
(b)
1.E+01
1.E+00
εβ/ε =3
1.E-01 ε β = 0.15
αlβ /Dβ
ε β = 0.3
2
ε β = 0.6
1.E-02
1.E-03
1.E-04
1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 1.E+01 1.E+02 1.E+03
Λ
In Fig. 7.34a, the length of the cell, L ¼ lr þ lb þ lc ; is used as the character-

istic length. The mass transfer coefficient âis shown to decrease linearly with eb.
This result is in accordance with previous investigations (e.g. see Eq. 47e in [82])
although a* is expected to increase linearly with eb. The explanation lies in the
Fig. 7.35 A typical three-dimensional unit cell with the immobile fluid phase configuration in
the form of (a) blobs, or (b) films. Volume fractions are eb = 0.31, ec = 0.12. The immobile
phase is also plotted separately for clarity
different characteristic lengths used to non-dimensionalize the mass transfer

l
coefficient. It is not difficult to show that for parallel phase geometry it is Lb ¼ eb ,
*
hence a ¼ âe2b : Thus, Fig. 7.34b shows the corresponding plot for a for various
values of eb. Where a* ? 3eb when d ! 1 , K ! 1, as expected.
More realistic geometries are considered in Fig. 7.35 where a typical periodic
three-dimensional unit cell is shown, assumed to represent regularly packed
granular porous media.
The solid grains (r-phase) are located on the eight edges of the cell. For the
immobile c-phase, two different cases are considered:
• The immobile phase does not wet the solid, hence it is distributed in the pore
space in the form of blobs.
• The immobile phase wets the solid, hence it is distributed in the pore space in
the form of films.
Fig. 7.36 The effect of 60

blobs Dβ / Dγ =0
Peclet number on the mass films
transfer coefficient for the 50
limiting case of sufficiently

40
fast diffusion in the c-phase
for the three-dimensional unit
αL2/Dβ
30
cell of Fig. 7.35
20
10
0
1 10 100
Pe
Fig. 7.37 The effect of (a)

Peclet number on the curve of 100
Pe=100
the mass transfer coefficient

as a function of K. Two
10
values are shown, one low Pe=1
(Pe = 1) and another high
(Pe = 100). Three 1
αL2/Dβ
dimensional unit cells with films

volume fractions of blobs
0.1
(a) eb = 0.31 and ec = 0.12,
and (b) eb = 0.39 and
0.01
ec = 0.04
1E-3
1E-4 1E-3 0.01 0.1 1 10 100 1000 10000
Λ
(b)
100 Pe=100
10
1
Pe=1
films
αL 2/Dβ
blobs
0.1
0.01
0.001
0.0001 0.001 0.01 0.1 1 10 100 1000 10000
Λ
In both cases the ‘‘aqueous’’ phase (b-phase) flows under Stokes flow condi-

Dc
tions. Initially, the case of infinitely fast diffusion in the c-phase d ¼ !1
Db
is considered. The effect of the pore-scale Peclet number on the dimensionless
mass transfer coefficient is shown in Fig. 7.36 for d ? ?.
The mass transfer coefficient is nearly constant at low Peclet numbers (of the
order of less than unity), and starts increasing when the Peclet number increases.
The wetting condition affects the value of the mass transfer coefficient: the film
(wetting) configuration of the immobile phase yields mass transfer coefficients
always larger than those corresponding to blobs (non-wetting). This is a result of
the higher surface area per unit volume available in films compared to those of
blobs for the same volumetric fractions. In addition, it should be noted that these
results are in excellent agreement with the theoretical investigation of Ahmadi
et al. [6].
In Fig. 7.37, the effect of the dimensionless parameter K on the mass transfer
coefficient is shown for two different values of Peclet number. Again, the mass
transport coefficient increases linearly with K at low K values, and reaches a
constant value at high K. As before, the mass transfer coefficient is higher when the
immobile phase is wetting (films) rather than non-wetting (blobs). The coefficient is
independent of the Peclet number for low K values. When eb and ec vary, the
resulting trends are qualitatively similar, as observed by comparing Fig. 7.37a, b.
It is important to note that in these simulations, the dry porosity remains constant
and equal to 0.43, ec varied from 0.12 to 0.4 and, consequently, eb varied from 0.31
to 0.39. It is evident that as ec decreases so does the mass transfer coefficient.
References
1. Achenbach, E., Reinsche, E.: Methane/steam reforming kinetics for solid oxide fuel cells.
J. Power Sources 52, 238–299 (1994)
2. Adler, P.M., Jacquin, C.J., Quiblier, J.A.: Flow in simulated porous media. Int. J. Multiphase
Flow 16, 691–712 (1990)
3. Agterof, W.G.M., Vaessen, G.E.J., Haagh, G.A.A.V., Klahn, J.K., Janssen, J.J.M.:
Prediction of emulsion particle sizes using a computational fluid dynamics approach.
Colloid Surf. B 31, 141–148 (2003)
4. Aguiar, P., Adjimana, C.S., Brandona, N.P.: Anode-supported intermediate temperature
direct internal reforming solid oxide fuel cell. I: model-based steady-state performance.
J. Power Sources 138, 120–136 (2004)
5. Ahmadi, A., Quintard, M., Whitaker, S.: Transport in chemically and mechanically
heterogeneous porous media V: two-equation model for solute transport with adsorption.
Adv. Water Resour. 22, 59–86 (1998)
6. Ahmadi, A., Aigueperse, A., Quintard, M.: Calculation of the effective properties describing
active dispersion in porous media: from simple to complex porous media. Adv. Water
Resour. 24, 423–438 (2001)
7. Ahmed, K., Foger, K.: Kinetics of internal steam reforming of ethane on Ni/YSZ based
anodes for solid oxide fuel cells. Catal. Today 63, 479–487 (2000)
8. Alves, M.A., Delgado, J.M.P.Q., Guedes de Carvalho, J.R.F.: Mass transfer from cylinders
and plane surfaces buried in packed beds in alignment with the flow direction. Chem. Eng.
Sci. 61, 1174–1183 (2006)
9. Bekri, S., Thovert, J.F., Adler, P.M.: Dissolution and deposition in fractures. Eng. Geol. 48,
283–308 (1997)
10. Bekri, S., Adler, P.M.: Dispersion in multiphase flow through porous media. Int.
J. Multiphase Flow 28, 665–697 (2002)
11. Bird, R.B., Stewart, W., Lightfoot, E.N.: Transport Phenomena. Wiley, New York (1960)
12. Bomberg, M.: Moisture flow through porous building materials. In: Report no. 52, Division
of Building Technology, Lund Institute of Technology, Lund, Sweden (1974)
13. Bravo, M.C., Araujoc, M., Lago, M.: Pore network modeling of two-phase flow in a liquid-
(disconnected) gas system. Physica A 375, 1–17 (2007)
14. Carbonell, R.G., Whitaker, S.: Heat and mass transfer in porous media. In: Bear, J.,
Carpacioglu, M.Y. (eds.) Fundamentals of Transport Phenomena in Porous Media. Martinus
Nijhoff Publ, Dordrecht (1984)
15. Chesters, A.K.: The modeling of coalescence processes in fluid-liquid dispersions. Chem.
Eng. Res. Des. 69, 259–270 (1991)
16. Coelho, M.A.N., Guedes de Carvalho, J.R.F.: Transverse dispersion in granular beds: Part
II–mass transfer from large spheres immersed in fixed or fluidised beds of small inert
particles. Chem. Eng. Res. Des. 66, 178–189 (1988)
17. Colombert, R.: L’Humidité des bâtiments anciens; causes et effets; Diagnostic et remèdes,
Editions du Moniteur, Paris (1975)
18. Costa-Nunes, O., Gorte, R.J., Vohs, J.M.: Comparison of the performance of Cu-CeO2-YSZ
and Ni-YSZ composite SOFC anodes with H2, CO and syngas. J. Power Sources 141,
241–249 (2005)
19. Costamagna, P., Magistri, L., Massardo, A.F.: Design and part-load performance of a hybrid
system based on a solid oxide fuel cell reactor and a micro gas turbine. J. Power Sources 96,
352–368 (2001)
20. Coutelieris, F.A., Kainourgiakis, M.E., Stubos, A.K.: Low Peclet mass transport in
Interface Sci. 264, 20–29 (2003)
21. Crank, J.: The mathematics of diffusion, 2nd edn. Oxford University Press, UK (1975)
22. Dagan, G., Lessoff, S.: Solute transport in heterogeneous formations of bimodal
conductivity distribution: 1. Theory. Water Resour. Res. 37, 465–472 (2001)
23. Dullien, F.A.: Porous Media: Fluid Transport and Pore Structure. Academic Press,
California (1979)
24. Fenwick, D.H., Blunt, M.: Three-dimensional modeling of three phase imbibition and
drainage. Adv. Water Resour. 21, 121–143 (1998)
25. Ferguson, J.R., Fiard, J.M., Herbin, R.: Three-dimensional numerical simulation for various
geometries of solid oxide fuel cells. J. Power Sources 58, 109–122 (1996)
26. Freitas, V.P., Guimarães, A.S.: Characterization of a hygro-regulated wall base ventilation
system for treatment of rising damp in historical buildings. In: Proceedings of the 2nd
Nordic Symposium on Building Physics, Copenhagen, Denmark, pp. 911–919 (2008)
27. Fried, J.J., Muntzer, P., Zilliox, L.: Groundwater pollution by transfer of oil-hydrocarbons.
Ground Water 17, 586–594 (1979)
28. Geller, J.T., Hunt, J.R.: Mass transfer from non-aqueous phase organic liquids in water-
saturated porous media. Water Resour. Res. 29, 833–845 (1993)
29. Gerami, S., Pooladi-Darvish, M.: Predicting gas generation by depressurization of gas
hydrates where the sharp-interface assumption is not valid. J. Pet. Sci. Eng. 56, 146–164
(2007)
30. Goldschmidt, M.J.V., Weijers, G.G.C., Boerefijn, R., Kuipers, J.A.M.: Discrete element
modelling of fluidised bed spray granulation. Powder Technol. 138, 39–45 (2003)
31. Guedes de Carvalho, J.R.F., Delgado, J.M.P.Q., Alves, M.A.: Mass transfer between
flowing fluid and sphere buried in packed bed of inerts. AIChE J. 50, 65–74 (2004)
32. Grace, H.P.: Dispersion phenomena in high-viscosity immiscible fluid systems and
application of static mixers as dispersion devices in such systems. Chem. Eng. Commun. 14,
225–277 (1982)
33. Grgicak, C.M., Green, R.G., Giorgi, J.B.: SOFC anodes for direct oxidation of hydrogen and
methane fuels containing H2S. J. Power Sources 179, 317–328 (2008)
References 231
34. Guimarães, A.S., Delgado, J.M.P.Q., de Freitas, V.P.: Mathematical analysis of the
evaporative process of a new technological treatment of rising damp in historic buildings.
Build. Environ. 45, 2414–2420 (2010)
35. Gurau, V., Liu, H., Kakac, S.: Two-dimensional model for proton exchange membrane fuel
cells. AIChE J. 44, 2410–2422 (1998)
36. Gummerson, R.J., Hall, C., Hoff, W.D.: Water movement in porous building materials-III.
A sorptivity test procedure for chemical injection damp proofing. Build. Environ. 16,
193–199 (1981)
37. Gvirtzam, H., Paldor, N., Magaritz, M., Bachmat, Y.: Mass exchange between mobile
freshwater and immobile saline water in the unsaturated zone. Water Resour. Res. 24,
1638–1644 (1988)
38. Gwo, J.P., O’Brien, R., Jardine, P.M.: Mass transfer in structured porous media: embedding
mesoscale structure and microscale hydrodynamics in a two-region model. J. Hydrol. 208,
204–222 (1998)
39. Hall, C., Hoff, W.D.: Water transport in brick, stone and concrete. Taylor and Francis,
New York (2002)
40. Hall, C., Hoff, W.D.: Rising damp: capillary rise dynamics in walls. Proc. R. Soc. A Math.
Phy. 463, 1871–1884 (2007)
41. He, H., Hill, J.M.: Carbon deposition on Ni/YSZ composites exposed to humidified
methane. Appl. Catal. A Gen 317, 284–292 (2007)
42. Heibel, A.K., Scheenen, T.W.J., Heiszwolf, J.J., van As, H., Kapteijn, F., Moulijn, J.A.: Gas
and liquid phase distribution and their effect on reactor performance in the monolith film
flow reactor. Chem. Eng. Sci. 56, 5935–5944 (2001)
43. Hirschenhofer, J.H., Stauffer, D.B., Engleman, R.R., Klett, M.G.: Fuel Cell Handbook, 4th
edn. Business/Technology Books, Orinda (1997)
44. Holm, A., Kunzel, H.M.: Two-dimensional transient heat and moisture simulations of rising
damp with WUFI-2D. In: Proceedings of the 2nd International Conference on Building
Physics, Leuven, Belgium, pp. 363–3677 (2003)
45. Hunt, J.R., Sitar, N., Udell, K.S.: Non-aqueous phase liquid transport and cleanup I: analysis
of mechanisms. Water Resour. Res. 24, 1247–1258 (1988)
46. Inamuro, T.: Lattice Boltzman methods for viscous fluid flows and for two-phase fluid
flows. Fluid Dyn. Res. 38, 641–659 (2006)
47. Irvine, J.T.S., Sauvet, A.: Improved oxidation of hydrocarbons with new electrodes in high
temperature fuel cells. Fuel Cells 1, 205–210 (2001)
48. Jamsak, W., Assabumrungrat, S., Douglas, P.L., Laosiripojana, N., Suwanwarangkul, R.,
Charojrochkul, S., Croiset, E.: Performance of ethanol-fuelled solid oxide fuel cells: proton
and oxygen ion conductors. Chem. Eng. J. 133, 187–194 (2007)
49. Janssen, J.J.M., Boon, A., Agterof, W.G.M.: Influence of dynamic interfacial properties on
droplet break-up in simple shear flow. AIChE J. 40, 1929–1939 (1994)
50. Jaouen, F., Lindbergh, G., Sundholm, G.: Investigation of mass-transport limitations in the
solid polymer fuel cell cathode. J. Electrochem. Soc. 149, A437–A447 (2002)
51. Kainourgiakis, M.E., Kikkinides, E.S., Stubos, A.K.: Diffusion and flow in porous domains
constructed using process-based and stochastic techniques. J. Porous Mat. 9, 141–154
(2002)
52. Kamp, A.M., Chesters, A.K., Colin, C., Fabre, J.: Bubble coalescence in turbulent flows: a
mechanistic model for turbulence-induced coalescence applied to microgravity bubbly pipe
flow. Int. J. Multiphase Flow 27, 1363–1396 (2001)
53. Kay, A.L., Davies, H.N.: Calculating potential evaporation from climate model data: a
source of uncertainty for hydrological climate change impacts. J. Hydrol. 358, 221–239
(2008)
54. Kikkinides, E.S., Burganos, V.N.: Permeation properties of three-dimensional self-affine
reconstructions of porous materials. Phys. Rev. E 62, 6906–6915 (2000)
55. Kim, T., Ahn, K., Vohs, J.M., Gorte, R.J.: Deactivation of ceria-based SOFC anodes in
methanol. J. Power Sources 164, 42–48 (2007)
56. Kimball, B.A., Jackson, R.D., Reginato, R.J., Nakayama, F.S., Idso, S.B.: Comparison of
field-measures and calculated soil-heat fluxes. Soil Sci. Soc. Am. Proc. 40, 18–25 (1976)
57. Klahn, J.K., Janssen, J.J.M., Vaessen, G.E.J., de Swart, R., Agterof, W.G.M.: On the escape
process during phase inversion of an emulsion. Colloid Surf. A 210, 167–181 (2002)
58. Klein, J.M., Bultel, Y., Pons, M., Ozil, P.: Modeling of a solid oxide fuel cell by methane:
analysis of carbon deposition. J. Fuel Cell Technol. 7, 425–434 (2007)
59. Kostoglou, M., Karabelas, A.J.: Evaluation of zero order methods for simulating particle
coagulation. J. Colloid Interface Sci. 163, 420–431 (1994)
60. Kostoglou, M., Karabelas, A.J.: Induced pulsing in trickle beds-characteristics and
attenuation of pulses. Chem. Eng. Sci. 60, 6584–6595 (2005)
61. Kunzel, H.M.: Simultaneous heat and moisture transport in building components; one and
two dimensional calculation using simple parameters. Dissertaion, University of Stuttgart
(1995)
62. Lam, A.C., Schechter, R.S., Wade, W.H.: Mobilization of residual oil under equilibrium and
non-equilibrium conditions. Soc. Petrol. Eng. J. 23, 781–790 (1983)
63. La O’, G.J., In, H.J., Crumlin, E., Barbastathis, G., Shao-Horn, Y.: Recent advances in
micro devices for electrochemical energy conversion and storage. Int. J. Energy Res. 31,
548–575 (2007)
64. Lehr, F., Millies, M., Mewes, D.: Bubble-size distributions and flow fields in bubble
columns. AIChE J. 48, 2426–2443 (2002)
65. Lessoff, S., Dagan, G.: Solute transport in heterogeneous formations of bimodal
conductivity distribution: 2. Applications. Water Resour. Res. 37, 473–480 (2001)
66. Lin, Y.B., Zhan, Z.L., Liu, J., Barnett, S.A.: Direct operation of solid oxide fuel cells with
methane fuel. Solid State Ionics 176, 1827–1835 (2005)
67. Lu, N., Li, Q., Sun, X., Khaleel, M.A.: The modeling of a standalone solid-oxide fuel cell
auxiliary power unit. J. Power Sources 1661, 938–948 (2006)
68. Luo, L.S., Girimaji, S.S.: Lattice Boltzmann model for binary mixtures. Phys. Rev. E 66,
035301/1–035301/4 (2002)
69. Luo, L.S., Girimaji, S.S.: Theory of the lattice Boltzmann method: two-fluid model for
binary mixtures. Phys. Rev. E 67, 036302/1–036302/11 (2003)
70. Luo, L.S., Svensen, H.F.: Theoretical model for drop and bubble breakup in turbulent
dispersions. AIChE 42, 1225–1233 (1996)
71. Matsuzaki, Y., Baba, Y., Sakurai, T.: High electric conversion efficiency and
electrochemical properties of anode-supported SOFCs. Solid State Ionics 174, 81–86 (2004)
72. Mingzhe, D., Dullien, F., Jun, Z.: Characterization of water-flood saturation profile histories
by the ‘complete’ capillary number. Transp. Porous Media 31, 213–237 (1998)
73. Mousavi, S.M., Jafari, A., Yaghmaei, S., Vossoughi, M., Sarkomaa, P.: Computer
simulation of fluid motion in a porous bed using a volume of fluid method: application in
heap leaching. Miner. Eng. 19, 1077–1083 (2006)
74. Nagel, F.P., Schildhauer, T., Biollaz, S.M.A., Wokaun, A.: Performance comparison of
planar, tubular and Delta8 solid oxide fuel cells using a generalized finite volume model.
J. Power Sources 184, 143–164 (2008)
75. Newman, J.S., Tabias, C.W.: Theoretical analysis of current distribution in porous
electrodes. J. Electrochem. Soc. 109, 1183–1191 (1962)
76. Ni, M., Leung, D.Y.C., Leung, M.K.H.: Mathematical modeling of ammonia-fed solid oxide
fuel cells with different electrolytes. Int. J. Hydrogen Energy 33, 5765–5772 (2008)
77. Park, S., Cracium, R., Vohs, J.M., Gorte, R.J.: Direct oxidation of hydrocarbons in a solid
oxide fuel cell I. Methane oxidation. J. Electrochem. Soc. 146, 3603–3605 (1999)
78. Plumb, O.A., Whitaker, S.: Diffusion, adsorption and dispersion in heterogeneous porous
media: the method of large scale averaging. In: Cushman, J.H. (ed.) Dynamics of Fluids in
Hierarchical Porous Media. Academic Press, New York (1990)
79. Pulkrabek, W.W.: Engineering Fundamentals of the Internal Combustion Engine, 2nd edn.
Pearson Prentice Hall, New Jeresey (2003)
References 233
80. Quintard, M., Whitaker, S.: One and two equation models for transient diffusion processes
in two-phases systems. Adv Heat Transfer 23, 369–464 (1993)
81. Quintard, M., Whitaker, S.: Transport in ordered and disordered porous media: volume
averaged equations, closure problems and comparison with experiments. Chem. Eng. Sci.
48, 2537–2564 (1993)
82. Quintard, M., Whitaker, S.: Convection, dispersion and interfacial transport of
contaminants: homogeneous porous media. Adv. Water Res. 17, 221–239 (1994)
83. Ramakrishna, P.A., Yang, S., Sohn, C.H.: Innovative design to improve the power density
of a solid oxide fuel cell. Power Sources 158, 378–384 (2006)
84. Riensche, E., Achenbach, E., Froning, D., Haines, M.R., Heidug, W.K., Lokurlu, A., von
Andrian, S.: Clean combined-cycle SOFC power plant–cell modelling and process analysis.
J. Power Sources 86, 404–410 (2000)
85. Sheng, J.J., Hayes, R.E., Maini, B.B., Torkite, W.S.: Modelling foamy oil flow in porous
media. Transp. Porous Media 35, 227–258 (1999)
86. Singhal, S.C.: Advances in solid oxide fuel cell technology. Solid State Ionics 135, 305–313
(2000)
87. Skelland, A.H.P.: Diffusional Mass Transfer. Wiley, New York (1974)
88. de Smedt, F., Wierenga, P.J.: A generalized solution for solute flow in soils with mobile and
immobile water. Water Resour. Res. 15, 1137–1141 (1979)
89. Štěpánek, F., Ansari, M.A.: Computer simulation of granule microstructure formation.
Chem. Eng. Sci. 60, 4019–4029 (2005)
90. Stone, H.A.: Dynamics of drop deformation and break up in viscous fluids. Annu. Rev.
Fluid Mech. 26, 65–102 (1994)
91. Sukeshini, A.M., Habibzadeh, B., Becker, B.P., Stoltz, C.A., Eichhorn, B.W., Jackson, G.S.:
Electrochemical oxidation of H2, CO and CO/H2 mixtures on patterned Ni anodes on YSZ
electrolytes. J. Electrochem. Soc. 153, A705–A715 (2006)
92. Suwanwarangkul, R., Croiset, E., Pritzker, M.D., Fowler, M.W., Douglas, P.L., Entchev, E.:
Modelling of a cathode-supported tubular solid oxide fuel cell operating with biomass-
derived synthesis gas. J. Power Sources 166, 386–399 (2007)
93. Torres, M.I.M., Freitas, V.P.: Treatment of rising damp in historical buildings: wall base
ventilation. Build. Environ. 42, 424–435 (2007)
94. Tseronis, K., Kookos, I.K., Theodoropoulos, C.: Modelling mass transport in solid oxide
fuel cells anodes: a case for multicomponent dusty gas model. Chem. Eng. Sci. 63,
5626–5638 (2008)
95. Tsipis, E.V., Kharton, V.V.: Electrode materials and reaction mechanisms in solid oxide
fuel cells: a brief review II. Electrochemical behavior vs. materials science aspects. J. Solid
State Electrochem. 12, 1367–1391 (2008)
96. Vogel, T., Gerke, H., Zhang, R., Ge-nuchten, M.V.: Modeling flow and transport in a two-
dimensional dual-permeability system with spatially variable hydraulic properties.
J. Hydrol. 238, 78–89 (2000)
97. Walters, K.M., Dean, A.M., Zhu, H.Y., Kee, R.J.: Homogeneous kinetics and equilibrium
predictions of coking propensity in the anode channels of direct oxidation solid-oxide fuel
cells using dry natural gas. J. Power Sources 123, 182–189 (2003)
98. Wang, T., Wang, J., Jin, J.: A novel theoretical breakup kernel function of bubble/droplet in
a turbulent flow. Chem. Eng. Sci. 59, 2593–2595 (2003)
99. Watt, D., Colston, B.: Investigating the effects of humidity and salt crystallization on
medieval masonry. Build. Environ. 35, 737–749 (2000)
100. Wexler, E.J.: Analytical solutions for one-, two-, and three-dimensional solute transport in
ground-water systems with uniform flow. In: U.S. Geological Survey Techniques of Water-
Resources Investigations, Book 3, Chap. B7, p. 190 (1992)
101. Wieringa, J.A., van Dieren, F., Janssen, J.J.M., Agterof, W.G.M.: Droplet break-up
mechanism during emulsification in colloid mills at high dispersed fraction. Inst. Chem.
Eng. 74, 554–562 (1996)
103. Whitaker, S.: Simultaneous heat, mass and momentum transfer in porous media: A theory of
drying. Adv. Heat Transf. 13, 119–203 (1977)
104. Xu, J., Froment, G.F.: Methane steam reforming and water-gas shift: I intrinsic kinetics.
AIChE J. 35, 88–96 (1989)
105. Zanotti, F., Carbonell, R.G.: Development of transport equations for multiphase systems I:
general development for two-phase systems. Chem. Eng. Sci. 39, 263–278 (1984)
106. Zanotti, F., Carbonell, R.G.: Development of transport equations for multiphase systems II:
application to one-dimensional axi-symmetric flows of two-phases. Chem. Eng. Sci. 39,
279–297 (1984)
107. Zhu, H., Kee, R.J., Pillai, M.R., Barnett, S.A.: Modeling electrochemical partial oxidation of
methane for cogeneration of electricity and syngas in solid–oxide fuel cells. J. Power
Sources 183, 143–150 (2008)
Index
A E
Adsorption—heterogeneous reaction— Effective mass diffusion
desorption mechanism, 107 coefficient, 213
Adsorption efficiency, 88 Effective thermal conductivity, 212
Area-weighted mean droplet Electrochemical reactions, 209
diameter, 196 Evaporation potential, 183
Artificial compressibility Experimental methods, 3
equation, 35
F
B Fick’s law, 40, 211
Boudouard reaction, 210 Fuel cells, 207
Break-up, 197
Brinkman law, 24
Bruggeman correlation, 213 G
Bubbles, 194 Gas flow, 55
Building materials, 193
H
C Happel’s sphere-in-cell
Characteristic time, 107 model, 29
Coalescence, 197 High-resolution scheme, 151
Concentration, 7
Continuity equation, 26
Convection-diffusion equation, 88 I
Cracking reaction, 210 Instantaneous adsorption, 90
CUBISTA, 151
Current conservation
equation, 213 K
Cylinder-in-cell model, 29 Kuwabara’s sphere-in-cell model, 27
D L
Darcy law, 24 Laminar liquid flow, 52
Droplets, 194 Langmuir’s isotherm, 90

Advanced Structured Materials 20, DOI: 10.1007/978-3-642-27910-2,
236 Index
M Representation of 3-D Sphere

Mach number, 35 Assemblages, 33
Mass transport coefficient, 88 Reynolds number, 2
Mass-transfer coefficient, 115
Moisture transfer, 185
Moments of the distribution, 195 S
Sharp front model, 183
Sherwood number, 89
N Sinusoidal cell, 8
Numerical methods, 150 SOFC, 207
Sphere-in-cell model, 6
Spheroid-in-cell model, 8
O Steam reforming reaction, 209
Ohm’s law, 213 Stokes equation, 26
Overpotential, 214
U
P Unit-bed element, 8
Peclet number, 88
Permeability, 25
Phase function, 23 V
Volume-averaging procedure, 114
R
Realistic sorption mechanism, 90 W
Relative humidity, 183 Water gas shift reaction, 209

Transport Processes in Porous Media

Uploaded by

Copyright:

Available Formats

Transport Processes in Porous Media

Uploaded by

Document Information

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

Transport Processes in Porous Media

Uploaded by

Copyright:

Available Formats

Advanced Structured Materials

For further volumes:

ISSN 1869-8433 e-ISSN 1869-8441

Library of Congress Control Number: 2011946089

Ó Springer-Verlag Berlin Heidelberg 2012

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

What is a ‘‘porous medium’’? Although several definitions have been proposed,

be intelligible to readers who have some previous knowledge of fundamentals of

2 Fundamentals of Porous Structures . . . . . . . ....... ......... 5

3 Flow in Porous Media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4 Transport Phenomena in Porous Structures . . . . . . . . . . . . . . . . . 39

5 Modeling of Transport Processes in Porous Materials . . . . . . . . . . 87

5.4.5 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

6 Experimental and Numerical Investigation of Mass Transport

7 Applications and Examples. . . . . . . . . . . . . . . . . . . . . . . . ...... 175

7.3 Bubbles and Drops in Porous Structures. . . . . . . . . . . . . . . . . . 194

1.1 Why Porous Materials?

What is a porous material? The term ‘‘porous’’ is used to describe a complex

F. A. Coutelieris and J. M. P. Q. Delgado, Transport Processes in Porous Media, 1

‘‘porosity’’. Generally speaking, any solid material which contains cavities,

1.2 Aims and Scopes

1.3 Book Contents and Structure

different temperatures, as well as estimation of tortuosity in porous media.

The understanding of multiphase flow in porous media is of great importance in many

2.2 Cell Models

Mathematical modeling of transport processes in porous media is a powerful tool

F. A. Coutelieris and J. M. P. Q. Delgado, Transport Processes in Porous Media, 5

GRANULAR POROUS MEDIUM

The first introduction of sphere-in-cell model was by Cunningham [18]. When

GRANULAR POROUS MEDIUM

Fig. 2.2 Schematic representation of sinusoidal cell concept

The models described above all considered either spherical or cylindrical

POROUS MEDIUM 2-D UNIT CELL

1-D UNIT CELL 3-D UNIT CELL

Fig. 2.3 Various unit cells

2.3 Digital Reconstruction (Lattice Boltzmann Method)

Transport in porous media is a topic related to many technological and environ-

fluid–solid systems and the difficulty in representing accurately the complicated

2.3.1 Porous Media Generation

Accurate numerical simulation of fluid flow in porous media requires detailed

computational resources. It is thus important to construct models that are able to

2.4 Stochastic Modeling

Fig. 2.4 Example of generated porous media

where De is the diffusion coefficient, C is the mean solute concentration, u is the

where K is the isotropic hydraulic conductivity, e is the porosity supposed to be

2.4.1 The Pore Space

very useful for validating numerical reconstruction techniques of the microstruc-

Fig. 2.6 Comparison between different 2D cross-sections of a Fontainebleau sandstone [6],

Another technique is the use of object-based modelling whereby the actual

2.4.2 Transport Properties

When estimating single phase transport properties like absolute permeability it is

For multiphase flow it is possible to use Lattice-Boltzmann techniques

As mentioned in Chap. 1, a porous medium is a particulate phase (usually solid) that

where x is the position vector from an arbitrary origin.

F. A. Coutelieris and J. M. P. Q. Delgado, Transport Processes in Porous Media, 23

Experimental and theoretical investigations are normally associated with the

3.1.1 Macroscopic Description

the permeability of a porous media is defined as 1 Darcy, if a fluid of 1 cP viscosity,

where parameter b depends on the gas studied.