Professional Documents
Culture Documents
Applied Mathematical Modeling and Analysis in Renewable Energy 1St Edition Manoj Sahni Online Ebook Texxtbook Full Chapter PDF
Applied Mathematical Modeling and Analysis in Renewable Energy 1St Edition Manoj Sahni Online Ebook Texxtbook Full Chapter PDF
Applied Mathematical Modeling and Analysis in Renewable Energy 1St Edition Manoj Sahni Online Ebook Texxtbook Full Chapter PDF
https://ebookmeta.com/product/mathematical-modeling-and-applied-
calculus-joel-kilty/
https://ebookmeta.com/product/renewable-hydrogen-renewable-
energy-and-renewable-hydrogen-apac-markets-policies-analysis-1st-
edition-matthieu-jaunatre/
https://ebookmeta.com/product/mathematical-analysis-and-
analytical-modeling-1st-edition-ivan-stanimirovic/
https://ebookmeta.com/product/renewable-energy-1st-edition-
rahman/
https://ebookmeta.com/product/applied-modeling-techniques-and-
data-analysis-1-computational-data-analysis-methods-and-
tools-1st-edition-yiannis-dimotikalis/
https://ebookmeta.com/product/advanced-mathematical-modeling-
with-technology-advances-in-applied-mathematics-1st-edition-fox-
william-p-burks-robert-e/
https://ebookmeta.com/product/renewable-energy-systems-the-
earthscan-expert-guide-to-renewable-energy-technologies-for-home-
and-business-1st-edition-dilwyn-jenkins/
https://ebookmeta.com/product/topics-in-applied-mathematics-and-
modeling-1st-edition-oscar-gonzalez/
Applied Mathematical
Modeling and Analysis
in Renewable Energy
Mathematical Engineering, Manufacturing, and Management
Sciences
Series Editor: Mangey Ram
Professor, Assistant Dean (International Affairs), Department of Mathematics,
Graphic Era University, Dehradun, India
The aim of this new book series is to publish the research studies and articles that
bring up the latest development and research applied to mathematics and its appli-
cations in the manufacturing and management sciences areas. Mathematical tool
and techniques are the strength of engineering sciences. They form the common
foundation of all novel disciplines as engineering evolves and develops. The series
will include a comprehensive range of applied mathematics and its application in
engineering areas such as optimization techniques, mathematical modeling and
simulation, stochastic processes and systems engineering, safety-critical system
performance, system safety, system security, high assurance software architecture
and design, mathematical modeling in environmental safety sciences, finite element
methods, differential equations, reliability engineering, etc.
Non-Linear Programming
A Basic Introduction
Nita H. Shah and Poonam Prakash Mishra
Edited by
Manoj Sahni and Ritu Sahni
First edition published 2022
by CRC Press
6000 Broken Sound Parkway NW, Suite 300, Boca Raton, FL 33487-2742
and by CRC Press
© 2022 selection and editorial matter, Manoj Sahni and Ritu Sahni; individual chapters, the contributors
CRC Press is an imprint of Taylor & Francis Group, LLC
Reasonable efforts have been made to publish reliable data and information, but the author and pub-
lisher cannot assume responsibility for the validity of all materials or the consequences of their use.
The authors and publishers have attempted to trace the copyright holders of all material reproduced in
this publication and apologize to copyright holders if permission to publish in this form has not been
obtained. If any copyright material has not been acknowledged please write and let us know so we may
rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or here-
after invented, including photocopying, microfilming, and recording, or in any information storage or
retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, access www.copyright.com
or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. For works that are not available on CCC please contact mpkbookspermissions@tandf.
co.uk
Trademark notice: Product or corporate names may be trademarks or registered trademarks and are
used only for identification and explanation without intent to infringe.
DOI: 10.1201/9781003159124
Typeset in Times
by KnowledgeWorks Global Ltd.
Contents
Preface......................................................................................................................vii
Acknowledgments......................................................................................................ix
About the Editors.......................................................................................................xi
Contributors............................................................................................................ xiii
PART II G
eneralized Mathematical
Ideas and Their Applications
vii
Acknowledgments
The material in this book reflects the research of many authors. We are immensely
grateful to all the authors and express our sincere appreciation for their contribu-
tions. We also acknowledge our colleagues at Pandit Deendayal Energy University
(formerly known as Pandit Deendayal Petroleum University) for their encouragement
and support. We also thank the editorial team at Taylor & Francis who accepted the
manuscript and then guided us through many stages that are necessary for shaping
this book. We are indebted to many people for their invaluable assistance in writing
this book. We also express our gratitude to the reviewer of the chapters.
We are also thankful to our family members for their support, patience, encour-
agement, and all possible help provided by them while we were engaged in preparing
this manuscript. We would like to especially acknowledge our son, Mohit Sahni;
without him and his support, we would not be able to complete this work. This book
would not have been completed if God had not provided us with enough patience
during this period. We shall be highly grateful to the Almighty.
We have tried our best to make this book an error-free text. Nevertheless, any
suggestions, comments, and feedback for further improvement of the book will be
gratefully accepted.
ix
About the Editors
Manoj Sahni has been working as an associate
professor and head, Department of Mathematics,
Pandit Deendayal Energy University, Gandhinagar
for the last seven years. He has more than 17
years’ experience in academic and research
fields. During the last 17 years, he has worked
with Jaypee Institute of Information Technology,
Noida, Navrachna University, Vadodara, and
PDEU, Gandhinagar. During this period, he has
published more than 60 research papers in various
international journals and conference proceed-
ings. He is a Life Member of various societies such
as Indian Science Congress, Indian Mathematical
Society, Allahabad Mathematical Society, and more. He has successfully coordi-
nated various events such as international conferences, expert lectures, seminars,
workshops, and more. He is the reviewer of various reputed international journals.
His research areas include continuum mechanics, functionally graded materials,
fuzzy sets, and numerical methods.
xi
Contributors
Aman Aggarwal Dipayan Guha
Teri School of Advanced Studies Motilal Nehru National Institute of
New Delhi, India Technology
Prayagraj, India
Manisha Bansal
Indraprastha College for Women Jyoti Gupta
University of Delhi Department of Mechanical
Delhi, India Engineering
Indian Institute of Technology
Abhijit Barman
Kanpur, India
Department of Mathematics
National Institute of Technology
Surendra Singh Kachhwaha
Silchar, India
Mechanical Engineering
Darshee Baxi Department
School of Science Center for Biofuel and Bioenergy
Navrachana University Studies
Vadodara, India Pandit Deendayal Petroleum
University
Purnima Bindal Gandhinagar, India
P.G.D.A.V. College
University of Delhi Dima Kayyali
Delhi, India McGill University
Montreal, Canada
Yogesh Chaudhari
Computer Science and Engineering Pravin Kodgire
Navrachana University Chemical Engineering Department
Vadodara, India Center for Biofuel and Bioenergy
Studies
Manisha Dalbhide-Ubale Pandit Deendayal Petroleum
Indus University University
Ahmedabad, India Gandhinagar, India
xiii
xiv Contributors
CONTENTS
1.1 Introduction....................................................................................................... 3
1.1.1 Conceptual Model..................................................................................4
1.2 Permeability Model...........................................................................................4
1.2.1 Auxiliary Equations...............................................................................6
1.2.2 Variable Coefficients.............................................................................6
1.3 Single-Phase Water Flow...................................................................................7
1.3.1 Mathematical Model.............................................................................. 7
1.3.2 Initial and Boundary Conditions...........................................................7
1.3.3 Numerical Model................................................................................... 8
1.4 Multiphase Gas and Water Flow........................................................................8
1.4.1 Mathematical Model.............................................................................. 8
1.4.2 Initial and Boundary Conditions......................................................... 10
1.4.3 Methodology to Solve for Multiphase Flow........................................ 10
1.4.4 Numerical Model................................................................................. 13
1.4.5 Water Saturation to be Solved Explicitly Using Equation 1.17........... 13
1.5 Verification and Validation.............................................................................. 15
1.5.1 Verification and Validation Results..................................................... 15
1.6 Single-Phase Gas Flow.................................................................................... 17
1.6.1 Numerical Model................................................................................. 19
1.7 Conclusion....................................................................................................... 19
Nomenclature............................................................................................................ 19
References.................................................................................................................20
1.1 INTRODUCTION
Coal formation under average reservoir pressure and temperature contains a specified
volume of gas adsorbed per unit ton of coal formation. Extracting methane from an
unconventional coal formation involves fluid flow stages such as dewatering, multiphase
flow, and gas flow. Every fluid flow stage includes either one or many fluid phases.
DOI: 10.1201/9781003159124-1 3
4 Applied Mathematical Modeling and Analysis in Renewable Energy
Very few earlier works have provided for equations covering all fluid flow stages.
The study mainly considers stress dependency of cleat permeability and sorption
strains in the coal when modeling. At each period, stress dependency is either a
function of cleat compressibility or sorption strain or both. Work done by Jishan Liu
et al. (2011) focused on the multiphase fluid flow stage alone, while Nie et al. (2011)
gave equations only for gas flow but included stress dependency effects. Lili (2012)
started analysis after the dewatering stage, and therefore, omitted the single-phase
water flow. Clarkson and Qanbari (2016) semi-analytically demonstrated equations
for multiphase and single-phase gas. The literature is a reference for verification and
validation. It has assumed cleat permeability to be pressure dependent.
The objective of the present work is to understand how an equation is written
for different flow stages when fluid is assumed to flow through a cleat in the one-
dimensional unconventional coal gas reservoir. Having modeled the fluid flow stages
mathematically, they need to be numerically solved to obtain values of dependent
variables. Here, an effort has been made to explain the numerical methodology
adopted in solving the transition effects between the various fluid flow stages.
FIGURE 1.1 (a) Three-dimensional coal formation comprising cleat and matrix. (b) Two-
dimensional coal formation with study length, pay zone thickness, and radius varies dynami-
cally. (c) One-dimensional model and the varying drained distance for the changing wellbore
pressure. (d) One-dimensional model divided as nodes along study length with the drained
distances increasing for each time increment.
6 Applied Mathematical Modeling and Analysis in Renewable Energy
−ν E × εL Pn Pi
(σ − σ i )n = 0.36 × × ( Pn − Pi ) + × −
(1 − ν ) 3 × (1 − ν ) ( Pn + PL ) ( Pi + PL )
− (1 + ν ) 2 × E × ε L Pn Pi
+ 0.64 × × ( P − P ) + × −
3 × (1 − ν ) 9 × (1 − ν ) ( Pn + PL ) ( Pi + PL )
n i
(1.2)
(Li et al., 2017)
k
φn = φi × √3 n (1.3)
ki
(porosity-permeability cubic relation)
1.2.1 Auxiliary Equations
Sw + Sg = 1 (1.4)
k rg + k rw = 1 (1.5)
φi × Swi φiv ( P ) × Swiv
wpiv = ( 4 × L W × X iv × h ) × − (1.6)
( 5.615 × Bwi ) 5.615 × Bwiv ( P ) ( )
(Clarkson and Qanbari, 2016)
wpiv
qw iv = (1.7)
t + dt
gpiv =
( 4 × L W × X iv × h ) × φi × Sgi − φiv ( P ) × Sg (Sw ) + VL × P − VL × Piv (1.8)
iv
1000 (B )
gi
(B giv ( P )) P + PL Piv + PL
(Clarkson and Qanbari, 2016)
gpiv
qgiv = (1.9)
t + dt
cfn =
(
cfi × 1 − e −α (σ −σ i )n ) (1.11)
α (σ − σ i )n
(Robertson, 2006)
Mathematical Modeling of Transitional Fluid Phase Flows 7
Sgn
cl n = (1.12)
Pn
ct n = cl n + cfn (1.13)
Equations 1.1–1.3 represent the empirical relations for stress-dependent cleat perme-
ability and porosity. The sum of fluid saturations add up to unity, while the fluid rela-
tive permeability values also sum up to one as shown in equations 1.4 and 1.5. The
fluid flow rate is shown in equations 1.7 and 1.9, while the material balance is shown
in equations 1.6 and 1.8 for water and gas, respectively. Equations 1.10–1.13 are the
empirical fluid properties.
Note:
a Unknown.
1.3.3 Numerical Model
The coefficients in the equation are discretized using backward time central
space (BTCS) method to obtain a tri-diagonal matrix of form AX=B. Tri-diagonal
matrix algorithm (TDMA) approach is used to solve the pressure at each node at
the next time level.
A flowchart (Fig. 1.2) explains the complete procedure followed from defining
the dependent variable to calculating them for a single-phase water flow in cleat. The
variable is average reservoir pressure. The iterative loop solves for the non-linearity
in the flow equation. Both reservoir and fluid properties mentioned in the list vary
with time.
Note:
a Unknown.
2
dP
cl iv ( Sw , P ) × ≈ 0 ( Since compressibility of water is negligible ) (1.16)
dx
(
∂ ρg (VPL+ ×PLP) +
ρg ×φiv ( P )× Sg ( Sw )
Bgiv ( P ) )= k iv ( P ) × k rg ( Sw )
(
× ∇ ρg∇P ) (1.18)
∂t µg iv ( P ) × Bgiv ( P )
Mathematical Modeling of Transitional Fluid Phase Flows 9
FIGURE 1.2 The coding procedure is followed with the use of assumptions and calculation
of dependent variables for single-phase water flow.
10 Applied Mathematical Modeling and Analysis in Renewable Energy
µ g × Bgiv
iv
×
(
∂ ρg (VPL+ ×PLP) +
ρg ×φiv × Sg
Bgiv ) = ∇ ( ρ ∇P )
g (1.19)
k iv × k rg ∂t
2
∂2 P dP
× ρg + ρg × cl iv ×
∂x 2 dx
=
µg iv × Bgiv
× (
∂ ρg ×φiv × Sg
Bgiv ) + ∂( ρ VL × P
g ( P + PL ) ) (1.20)
k iv × k rg ∂t ∂t
µgiv ( p ) × Bgiv ( P )
k iv ( p ) × k rg ( sw )
ct iv ( sw , P ) × φiv ( P ) × sg ( sw ) ∂p φiv ( P ) ∂sg VL × PL ∂p (1.21)
iv
× + × + 2 ×
Bgiv ( P ) ∂ t Bgiv ( P ) ∂ t ( P + PL ) ∂t
2
∂2 p dp
= + cl iv ( Sw , P ) ×
∂x 2 dx
∂P V × PL φiv ( P ) ∂Sg
× ct iv ( Sw , P ) × φiv ( P ) × Sg iv ( Sw ) + L + ×
∂t ( P + PL )2 Bgiv (, P ) ∂t (1.22)
k iv ( P ) × k rg ( sw ) ∂2 P dP
2
=
µg iv ( P ) × Bgiv ( P ) ∂x 2
+ cl iv ( S w , P ) ×
dx
Equation 1.17 is the water flow equation deduced from the original equation 1.15
after neglecting the compressibility term (equation 1.16). Equation 1.22 is derived
from initial equation 1.18 for gas flow through stages of differentiation (equations
1.19–1.21), considering gas compressibility. It is the non-linear partial differential
equation applicable for combined gas and water flow in the cleats of coal formation.
The equation is both diffusive and advective. It is highly non-linear due to the sec-
ond degree advective part, and thus, is hyperbolic dominant. Implicit pressure and
explicit saturation (IMPES) are used to solve the pressure-saturation coupling. The
non-linear term is linearized and solved using the TDMA approach.
Steps:
µwiv × k rg
1. Step 1: Multiply equation 1.6 by µgiv × k rw
, we get:
∂2 P µw iv × k rg ∂2 P µwiv × k rg Sg iv dP
2
= + × + × ×
∂x 2 µg iv × k rw ∂x 2 µg iv × k rw Piv dx
dSwiv dSgiv
+ =0 (1.26)
dt dt
µwiv VL × PL ∂P
φiv × ct iv + Bgiv × ×
k iv × k rw ( P + PL )2 ∂t
(1.27)
µw × k rg ∂2 P µwiv × k rg Sgiv dP 2
= 1 + iv × + × ×
µgiv × k rw ∂x 2 µgiv × k rw Piv dx
12 Applied Mathematical Modeling and Analysis in Renewable Energy
µwiv VL × PL µwiv × k rg
φiv × ct iv + Bgiv × 2 = b; =a (1.28)
k iv × k rw ( P + PL ) µg iv × k rw
2
∂P ∂2 P Sg dP
b× = (1 + a ) × 2 + a × iv × (1.29)
∂t ∂x Piv dx
P = →+ v (1.30)
P
2
∂(→+ v) ∂2 (→+ v) Sg d (→+ v)
b× P
= (1 + a ) × P
+ a × iv × P (1.31)
∂t ∂x 2 Piv dx
2
∂(→) ∂(v ) ∂2 (→+ v) Sgiv d (→ )
d (v )
b× P
+ = (1 + a ) × P
+ a × × P
+ (1.32)
∂ t ∂ t ∂x 2 Piv dx dx
∂(→) ∂2 (→)
∂(v ) ∂2 (v)
b×
p
+ = (1 + a ) × p
+
∂ t ∂t
∂x 2 ∂x 2
(1.33)
d (→) 2 d (→) 2
sgiv d (v ) d (v )
× +2×
p p
+a × × +
piv dx dx dx dx
Sg iv d (→)
∂(v ) ∂2 (v ) d (v ) (1.34)
b× = (1 + a ) × + a × × 2 × P
×
∂ t
∂x 2
Piv dx dx
(
(v nt+ dt − v nt ) ) (v nt ++dt1 − 2 × v nt+ dt + v nt +−dt1 )
b× = (1 + a ) ×
dt dX 2
(1.35)
d (→)
+ a × cl iv × 2 ×
p
×
(v nt ++dt1 − v nt +−dt1 )
dx 2 × dX
Mathematical Modeling of Transitional Fluid Phase Flows 13
µwiv VL × PL µwiv × k rg
φiv × ct iv + Bg iv × 2 = b; =a (1.37)
k iv × k rw ( P + PL ) µg iv × k rw
H=
a × dt
; G=
((1 + a ) × dt ) (1.38)
( N ) × b × dx 2 ( b × dX 2 )
( (
v nt+ dt × (1 + 2 × G ) − v nt ++dt1 G + H × cl iv × ( P [ N ] − P [ 0 ]) ))
(1.39)
( )
− v nt +−dt1 G − H × cl iv × ( P [ N ] − P [ 0 ]) = v nt
Pressure at next time level is defined by the pressure at present time and
error (equation 1.30). Equation 1.30 is substituted into equation 1.29 and
the steps from Equations 1.31–1.38 are followed. Equation 1.39 is the final
form of PDE to be solved using the TDMA approach to solve ∈ v nt+ dt , and
equate the value with the average reservoir pressure at node n and next
time level.
1.4.4 Numerical Model
The non-linear PDE is linearized. The coefficients in the linearized equation dis-
cretized using BTCS. The dependent variable at the next time step is solved using
the TDMA approach.
(Swt+ dt )=
(S + dt ×
t
w
k iv × k rw
µw ×φiv
iv
× ( Pnt ++1dt − 2 × Pnt + dt + Pnt +−1dt
dX 2 ) × 0.0158) (1.42)
(1 + cf × ( P
iv
t + dt
n − Pnt ) )
Sgt+ dt = (1 − Swt+ dt ) (1.43)
A flowchart (Fig. 1.3) explains the complete procedure followed from defining the
dependent variables to calculating them for a multiphase water and gas flow in cleat.
FIGURE 1.3 The coding procedure is followed with the use of assumptions and calculation
of dependent variables for multiphase gas and water flow.
Mathematical Modeling of Transitional Fluid Phase Flows 15
The variables are average reservoir pressure and water saturation. Two iterative loops
are present. The flow equation consists of the coupled pressure-saturation term. The
first iterative loop solves the functions of pressure, while the relative permeabilities
are solved in the second loop.
φn = φi × e( 0.0003×( Pn − Pi )) (1.44)
3
φ
kn = ki × n (1.45)
φi
Equations 1.44 and 1.45 represent the pressure-dependent cleat porosity term,
and the permeability derived from it using the cubic relation, respectively.
FIGURE 1.4 (a) Validating numerically obtained cumulative water results with field values
and verifying the same with semi-numerical results values of reference Clarkson et al. (2016).
(b) Validating numerically obtained cumulative gas results with field values. (c) Validating
numerically obtained water rate results with field values and verifying the same with semi-
numerical results values of reference Clarkson et al. (2016). (d) Validating numerically
obtained gas rate results with field values and verifying the same with semi-numerical results
values of reference Clarkson et al. (2016).
Mathematical Modeling of Transitional Fluid Phase Flows 17
∂P V × PL φiv ( P ) ∂Sg
× ct iv ( Sw , P ) × φiv ( P ) × Sgiv ( Sw ) + L + ×
∂t ( P + PL )2 Bgiv ( P ) ∂t (1.46)
k iv ( P ) × k rg ( sw ) ∂ P dP
2
+ cl iv ( Sw , P ) ×
2
=
µgiv ( P ) × Bgiv ( P ) ∂x 2 dx
Sgiv = 1; k rg = 1; k rw = 0 (1.47)
∂P V × PL k iv ( P ) ∂2 P dP
2
× ct iv ( P ) × φiv ( P ) + L = + cl ( P ) × (1.48)
( P + PL )2 µgiv ( P ) × Bgiv ( P ) ∂x 2 dx
iv
∂t
Substituting cl iv ,
∂P V × PL k iv ( P ) ∂2 P 1 dP 2
× ct iv ( P ) × φiv ( P ) + L 2 = + × (1.49)
∂t ( P + PL ) µgiv ( P ) × Bgiv ( P ) ∂x 2 Piv dx
VL × PL k iv ( P )
φiv × ct iv + 2 = b; =a (1.50)
( L)
P + P µ giv ( P ) × Bgiv ( P )
∂P ∂2 P 1 dP 2
×b= a× 2 + × (1.51)
∂t ∂x Piv dx
2
∂P ∂2 P a dP
×b=a× 2 + × (1.52)
∂t ∂x Piv dx
18 Applied Mathematical Modeling and Analysis in Renewable Energy
FIGURE 1.5 The coding procedure is followed with the use of assumptions and calculation
of dependent variables for single-phase gas flow.
Mathematical Modeling of Transitional Fluid Phase Flows 19
Equation 1.46 is the gas flow equation, the same as equation 1.22. Equation 1.47
shows the conditions of single-phase gas flow that modifies equations 1.46–1.48.
Equation 1.49 is the final equation for gas flow, while equation 1.50 is the splitting of
equation 1.49 into a and b terms. It is the non-linear partial differential equation. It
is highly non-linear due to the second degree advective part, and thus is hyperbolic
dominant. IMPES is used to solve the pressure-saturation coupling. The non-linear
term is linearized and solved using the TDMA approach. Equations 1.51 and 1.52
represent the PDE and the coefficients.
1.6.1 Numerical Model
The methodology of linearization and solving linearized equations by BTCS is the
same as that used for the multiphase flow.
A flowchart (Fig. 1.5) explains the complete procedure followed from defining the
dependent variable to calculating them for a single-phase gas flow in the cleat. The
variable is average reservoir pressure. The iterative loop solves for the non-linearity
in the flow equation. Both reservoir and fluid properties mentioned in the list vary
with time.
1.7 CONCLUSION
Every fluid flow PDE consists of the mass and momentum conservation equation. It
is non-linear or linear equation in nature. If non-linear, then the cause is understood
and solved numerically. A fully implicit numerical method, finite difference, is used
to convert the partial differential equation to an algebraic form. Iterative methods, in
case of single-phase water flow, are adopted to solve for the non-linearity. IMPES is
used to decouple pressure-saturation, followed by the methodology to linearize the
pressure terms in case of multiphase flow problems. Pressure term linearization is
used in PDE for a single-phase gas flow equation.
NOMENCLATURE
i ( suffix ): Initial value of variable.
iv ( suffix ) : Averaged value of variable at time t in the area of Xinv.
n ( suffix ) : n th nodal values of the parameter of points N.
t ( suffix ): t th values of the parameter of points Nt.
w( suffix ): Water phase.
g( suffix ): Gas phase.
pwf : Flowing well bore pressure ( psi ) .
P: Pressure at previous time level ( psia ) .
k: Permeability ( mD ) .
φ : Fracture porosity.
PL : Langmuir pressure ( psi ) .
VL : Langmuir Volume ( Mscf ) .
b: A constant value related to desorption pressure.
E: Young’s modulus of coal ( psi ) .
20 Applied Mathematical Modeling and Analysis in Renewable Energy
REFERENCES
Li, C., Wang, Z., Shi, L., Feng, R., (2017). ‘Analysis of Analytical Models Developed under the
Uniaxial Strain Condition for Predicting Coal Permeability during Primary Depletion.’
Energies, Vol. 10, No. 11, pp. 1849.
Clarkson, C.R., Qanbari, F. (2016). ‘A Semi-analytical Method for Forecasting Wells
Completed in Low Permeability, Undersaturated CBM Reservoirs.’ Journal of Natural
Gas Science and Engineering, Vol. 30, pp. 19–27, doi.org/10.1016/j.jngse.2016.01.040.
Lili, X. (2012). ‘Numerical well testing of coal bed methane reservoir.’ PhD thesis, Heriot-
Watt University.
Jishan, L., Chen, Z., Elsworth, D., Qu, H., Chen, D. (2011). ‘Interactions of Multiple Processes
during CBM Extraction: A Critical Review.’ International Journal of Coal Geology,
Vol. 87, No. 3, pp. 175–189, doi.org/10.1016/j.coal.2011.06.004.
Nie, R., Ying, F., Meng, J., Chun, G., Yong, L. (2011). ‘Modeling Transient Flow Behavior
of a Horizontal Well in a Coal Seam.’ International Journal of Coal Geology, Vol. 92,
pp. 54–68, doi.org/10.1016/j.coal.2011.12.005.
Zhang, H., Liu, J., Elsworth, D. (2008). ‘How Sorption Induced Matrix Deformation
Affects Gas Flow in Coal Seams: A New FE Model.’ International Journal of Rock
Mechanics and Mining Sciences, Vol. 45, No. 8, pp. 1226–1236, doi.org/10.1016/
j.ijrmms.2007.11.007.
Robertson, P.E. (2006).‘Measurement and modeling of sorption-induced strain and perme-
ability changes in coal,’ PhD thesis, Colorado School of Mines.
2 AI-Based Detection
and Localization of
Gastrointestinal Polyps
by Using Deep Learning,
Transfer Learning
and the Fusion of
These Techniques
Yogesh Chaudhari, Umme Salma Pirzada,
and Darshee Baxi
CONTENTS
2.1 Introduction: Background and Driving Forces................................................ 21
2.2 Deep Neural Network (DNN)......................................................................... 23
2.3 Transfer Learning............................................................................................24
2.4 Polyps and Computer-Aided Diagnosis...........................................................25
2.5 Dataset and Experimental Setup.....................................................................25
2.5.1 Dataset.................................................................................................25
2.5.2 Data Preparation..................................................................................26
2.5.3 CNN Architecture............................................................................... 27
2.5.4 VGG16 and ND VGG19......................................................................28
2.6 Results and Discussion.................................................................................... 30
2.7 Conclusion and Future Work........................................................................... 32
References................................................................................................................. 32
DOI: 10.1201/9781003159124-2 21
22 Applied Mathematical Modeling and Analysis in Renewable Energy
progresses to malignant tumor that gradually leads to CRC. These protuberances are
termed as polyp, which when missed or ignored, can prompt CRC. Almost 90% of
CRC cases result from the unnoticed development of adenomatous polyps. For effec-
tive diagnosis and treatment, early identification of colonic polyps utilizing colonos-
copy is necessary. The ADR is the effective assessment parameter to measure the
performance of endoscopist [1]. The colored portion in the Fig. 2.1 shows the portion
of intestine examined in the colonoscopy.
The 1993 milestone National Polyp Study [2] showed that the frequency of
colorectal malignant growth could be decreased by colonoscopy of adenomatous
polyps, which has been affirmed in many follow-ups and check-ups. Adenoma detec-
tion rate (ADR) – the expertise level of screening colonoscopies – has become a key
quality measure. Higher ADRs are related with lower post-colonoscopy colorectal
tumors and lower colorectal malignant growth mortality. In spite of the fact that ADR
ought to be perfect, studies show that ADR shifts generally among colonoscopists
FIGURE 2.1 Anatomy of the gastrointestinal system. (©2017 American Cancer Society,
Inc., Surveillance Research.)
AI-Based Detection and Localization of Gastrointestinal Polyps 23
FIGURE 2.2 Conventional machine learning (a) vs transfer learning and (b) in progress.
AI-Based Detection and Localization of Gastrointestinal Polyps 25
2.5.1 Dataset
The dataset utilized in the present work is CVC-Clinic DB that consists of frames
extracted from videos captured through colonoscopy and is published for Endoscopic
FIGURE 2.3 Polyp (colorectal cancer) growth. (©2005, Terese Winslow, U.S. Government
has certain rights.)
26 Applied Mathematical Modeling and Analysis in Renewable Energy
TABLE 2.1
Description of Dataset
2.5.2 Data Preparation
The images prepared using annotation are further processed with image augmenta-
tion to create transformed images to increase the data samples. These images are
FIGURE 2.4 Top row – RGB polyp frames extracted from sequence and bottom row – cor-
responding monochrome polyp mask. (CVC-Clinic DB database.)
AI-Based Detection and Localization of Gastrointestinal Polyps 27
FIGURE 2.5 Data hierarchy – (a) superset consisting of three subsets and (b) each subset
with two more subsets.
grouped into three sets, i.e., training, validation and testing, each with two subsets:
positive and negative.
The details of number of samples according to data hierarchy given in Fig. 2.5 are
shown in Table 2.2.
Three systems are designed to detect and locate the polyp as below:
TABLE 2.2
Number of Data Samples in Each Set
The polyp positive and polyp negative images are cropped from original images
using semi-automated approach using the script for annotations, and hence the polyp
masks are not used in the processing. The cropped images are given as an input to
image augmentation to generate three more files with rotation of parent file in angles
90°, 180° and 270°.
In the architecture, 64 filters are used with each convolution kernel of size
3 × 3, and to reduce spatial dimensionality, the pooling kernels used are of size
2 × 2. The stride is used to enhance the performance of the architecture. Out of
available activation functions such as step, sigmoid, tanh, ReLu, ELU and Leaky
ReLu, the current implementation makes use of ReLu. To prevent an overfitting in
the model, dropout technique is incorporated.
• VGG16
• VGG19
• ResNet50
• Inception V3
• Xception
• GoogLeNet
The layers are termed as ‘3 × 3 conv, X’, where ‘conv’ are convolutional layers
with 3 × 3 filters and X represents number of filters. MaxPool layer reduces dimen-
sions by the factor of 2. At the end, FC layer is with 4096 and 1000 units. Softmax
outputs one of a 1000 classes from labeled data provided by ImageNet.
TABLE 2.3
Details of Different Combinations of Architectures Implemented
A ramble round the rocks at low water just now discloses a scene of
bareness quite in keeping with the season of the year. The upper
surface of the higher lying rocks is as bare as a street pavement,
and only an occasional patch of acorn barnacles remains of the
encrustation with which they were invested during the summer. The
white whelk, so much in evidence here, have all gone into winter
quarters, and underneath projecting ledges and in sheltered nooks
they may be seen in myriads, their position being so judiciously
chosen as to be completely protected from the heavy north-east
seas. So closely are they wedged together that were a given space
to be cleared it would be found almost impossible to replace them in
the same area. Detaching one from its anchorage, it seems quite
dormant and inert, and appears to have lost the alacrity with which,
in summer, they withdraw themselves into their shells, and only with
apparent difficulty is the operculum or door of their domicile closed
against intruders. To witness the continual thumping and pounding to
which the Rock is subjected during the winter, one is surprised to
find that life in any form should continue to exist under such
conditions. A close search reveals exceedingly minute forms of life.
Here in this stony basin, originally but a shallow depression in which
a stone had lodged, and by the swirling action of the seas converted
to its present shape, with its sediment of broken shells, is a small
crab, so small indeed that a split pea might easily conceal him. He is
not a youngster either, but fully adult, in proof of which we have
frequently found them, in the proper season, with their spawn
attached. Deep in his little pit he seems quite immune from the
furious seas that tumble overhead as the tide makes. Numbers of
small white-banded whelks, which one may easily crush between the
fingers, maintain their position on the base of the tower, despite the
constant swirl of waters, though they may be detached with a flick of
the finger.
Vegetation now exists only at low-water mark; above that, broken
tangle roots, or, to be more correct, the claspers are seen still
adhering to the rocks, the tangles themselves having been shorn
clean from their moorings. Away towards the south-west, in the
deeper water, a boat may float among whole groves of storm-torn
tangles as they flaunt their tattered banners in the frosty sunlight,
suggestive of leafless trees in a winter landscape. Over the recently
emptied contents of the cook’s slop-pail a flock of gulls are circling
and screaming, actually hustling each other in their attempts to
capture anything edible. A solitary “black-back” is seen amongst the
noisy crowd, and as he swoops at some tempting morsel, his black,
beady eye watches our every movement with suspicion. What a
handsome bird he is as he swings past within a few feet of us, the
back and wings presenting a dead black appearance in startling
contrast with the immaculate whiteness of the fan-shaped tail and
the remainder of the body. Despite his handsome appearance, he is
a veritable vulture, and nothing comes amiss to him in the way of
food, be it fish, flesh or fowl. Frequently I have seen them make a
meal of a wounded duck, and once witnessed in Orkney a tug-of-war
between two of them for the possession of a dead lamb, resulting,
thanks to its decomposed state, in an equal division.
More gruesome meals are credited to them by those who have
witnessed their proceedings on a wreck strewn shore where loss of
life had been involved. A terror also on the grouse moors, they
devour both eggs and young, and even the sitting grouse herself is
not safe from him. One can scarcely credit such a sweeping
indictment against this handsome bird, but the proofs are all too
plain. Consequently we find him outside the pale of the Wild Birds
Protection Act, an Ishmael among his kind, whom any man may slay
when and wherever found. Except when harrying the eider ducks of
their legitimate spoil, he may be seen riding gracefully, head to wind,
in front of our kitchen window, with his weather eye always lifting in
our direction. A hand thrust from the window is sufficient invitation,
he is up at once, and the smallest morsel tossing among the foaming
breakers does not escape his keen eye. How gracefully he floats
back to his former position, lighting on the surface like a fleck of
foam. What a contrast to the eiders, who, when changing their
fishing ground, wing their way with such rapid wing beats as to give
one the impression that they are barely able to support themselves,
and finally strike the water with an awkward splash, reminding one of
the somewhat inelegant term with which boys designate a bad
dive—a “gutser.” Should a flock of eiders be fishing to leeward of the
tower, an amusing sight may be witnessed if advantage be taken,
while they are under water, of pouring a little paraffin oil from the
balcony, so that it will drift in their direction. No sooner does the head
of the first emerge in the greasy track of the oil than he is conscious
of something unusual having taken place. Flippering hither and
thither with outstretched neck, he becomes quite excited, and each
as he bounces to the surface joins in the commotion, frequently
colliding with each other. Finally, with loud cacklings, the whole flock
takes wing, evidently in high dudgeon at the insult offered to their
olfactory organs.
Sea pheasant is the name by which the long tailed duck is known
in some localities, and as we watch a flock of them crossing the reef
in full flight the synonym is at once apparent. In style of flight and
shape, to the long tail feathers, they are similar to the pheasant, but
only half the size, with beautiful plumage of black and white. Here
they are known as “candlewicks,” their call notes needing but little
stretch of the imagination to be rendered “Here’s a candlewick,”
repeated several times in shrill falsetto, which on a quiet day
becomes somewhat annoying as it clamorously floats through our
bedroom window. Some queer visitors we have here at times in the
way of birds. Once we captured a large owl dosing sleepily in one of
our windows. During the week of his captivity he would not deign to
partake of any food we offered him. Coming off watch one night I
took one of a flock of larks which were making suicidal attempts to
pierce the plate glass of the lantern. Placing it in the room where the
owl was roosting, it fluttered to the window, when, like a flash of
lightning and equally as noiseless, from the other side of the room
the owl came crash against the glass, a few feathers later on
testifying his appreciation of this form of dietary.
FEBRUARY 1902.
Piercing cold weather here of late, with a good deal of frost and
occasional snow showers. No matter how heavy the snowfall may be
here we only see it falling, as it does not lie long round our doors,
and only when our gaze is directed Arbroathwards—which, you may
be sure, is not seldom—are we reminded of its occurrence. The
close of last month saw our barometer taxed to its utmost
intelligence, and though a tenth higher would have seen its limit,
nothing of a phenomenal nature was noted. The solan geese or
gannets, which are pretty much in evidence here during the breeding
season, foraging for their families on the Bass Rock, gradually
disappeared, till during the month of November not one was to be
seen. A solitary one was seen in the first week of December, and
since then the number sighted has gradually increased, till in the
middle of the present month, as many as eight in one string were
counted winging their way southward. The Bass Rock, Ailsa Craig,
and the outlying stacks of lonely St Kilda, are said to be the only
breeding places of these birds in Scotland. At the beginning of the
past century they were considered a dainty article of food by the
Edinburgh gentry, and the Bass Rock was rented for the purpose of
supplying the market, the birds selling at the rate of half-a-crown a-
piece. I have seen it stated that the modus operandi of these birds
when engaged in fishing is to flit along the surface till fish are
sighted, when they rise to a high altitude, close their wings, and drop
hawk-like on their prey. This, I venture to think, is scarcely correct.
My experience is that when flitting near the surface if fish are sighted
they are invariably struck at without rising to a higher elevation. It is a
well known fact that objects under water are more easily
distinguished from a height than from near the surface, so that it may
be taken for granted that the higher these birds are flying when in
pursuit of prey the deeper the fish are swimming. Again, when diving
from a high altitude, the wings are kept rigidly outspread, and as the
tail is never seen spread rudder-like, as in the case of the hawk, any
deviation from their line of descent is controlled by the long narrow
wings, and only when nearing the “plunge” are they partially closed.
For the past fortnight we have had the company of a solitary seal.
His fishing does not seem to be very successful, either in quantity or
quality, as the only catch we have seen him negotiating was a saithe
the length of a man’s forearm. Playing with it as a cat would a
mouse, he would allow it to swim feebly for some distance, then
diving he would bring it to the surface, till latterly, with a toss of his
head and a thrust with his fore flipper, he quite disembowelled it, an
act of charity which the screaming gulls were not slow to appreciate.
Although so long here he has not been seen to rest on the rocks;
indeed, I only once saw one ashore here, and as we had a
somewhat amusing experience with him it will perhaps bear relating.
For several days it was seen, as the tide fell, to rest in one particular
place a few yards from the base of the tower. Our outer door opens
outwards, and is always closed at night, not that we are afraid of
burglars, but merely to prevent the entrance of the seas, and for our
own general comfort. The opening of this door always alarmed the
seal, and sent him into the water instanter. Dropping a line from the
balcony at low water, we made the end of it fast within a few feet of
his accustomed resting place. Next day, as the tide fell and the rocks
began to appear, he was seen to take up his former position,
yawning lazily as he rolled from side to side in the sunshine. Fixing a
four ounce charge of tonite to our electric cable, we quietly lowered it
down the line we had already made fast till within about six feet from
where he lay, apparently in blissful ignorance of what was happening
overhead. When yawning at his widest, we, by means of our
magneto-exploder, fired the charge, and, well—he stopped yawning
and went away! and his going was about the smartest thing I ever
witnessed. The force of the explosion, being unconfined, merely
tilted him on his side, but quickly recovering himself he flopped into
the water and shot seaward through the gully like a flash, a black line
under water denoting his course. Rounding the outer end of the
gully, he doubled back on the outside of the reef, and when opposite
his original position, made his appearance on the surface, a very
much startled seal. His aspect was quite comical as he stood, so to
speak, on his tip-toes evidently investigating the cause of his hurried
departure.
Several schools of porpoises have been seen this month,
presumably in pursuit of herring. To anyone who has seen these
animals gambolling in front of a ship’s bows when travelling at her
best, the ease with which they maintain their distance is a matter of
surprise—always on the point of being run down, but ever ahead,
snorting playfully as if in derision at the possibility of their being
overtaken by their lumbering follower. Off the island of Anticosta, in
the Gulf of St Lawrence—where these animals attain a size several
times larger than those of our home waters, and are of a cream
colour—I had an interesting view of their manner of suckling their
young. I have seen it stated that the mother by muscular
compression expels the nutritive fluid, which is absorbed by the
young one as it floats to the surface. The operation appeared to me
to be one of actual contact. The young one—which, by the way, is of
a slatey-blue colour—snuggling as close as possible to the mother
as she lay somewhat on her side on the surface, all the while
exhibiting the tenderest solicitude for her offspring. Truly the one
touch of nature which makes the whole world kin. It is surprising to
learn the evolution these animals have undergone in order to
accommodate themselves to their altered circumstances. Land-
dwellers at one stage of the world’s history, but acquiring a taste for
fish, they gradually became aquatic in their habits, dispensing with
such portions of their anatomy as were no longer necessary, while
developing others more appropriate to their new sphere of existence,
till, like their big brother the whale, from being a four-footed animal
they became quite fish like in appearance, even to the cultivation of
a dorsal fin, though still possessing rudimentary traces of their
former construction. Change is apparent on every hand in the plan of
nature; ages were necessary for the evolution of our present day
horse from his five toed ancestors; and after all it does not seem so
very startling when the transformation is enacted before our very
eyes in a few short stages, as in the case of the common frog, from
the gill breathing tadpole to the lung breathing adult. More startling it
is to learn that man himself was at one time a gill breather, and, as
biologists affirm, still exhibits traces of gill clefts at one stage of his
embryonic development.
MARCH 1902.