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
Visit to download the full and correct content document:
https://ebookmeta.com/product/applied-mathematical-modeling-and-analysis-in-rene
wable-energy-1st-edition-manoj-sahni/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Mathematical Modeling and Applied Calculus Joel Kilty
https://ebookmeta.com/product/mathematical-modeling-and-applied-
calculus-joel-kilty/
Renewable Hydrogen Renewable Energy and Renewable

Hydrogen APAC Markets Policies Analysis 1st Edition
Matthieu Jaunatre
https://ebookmeta.com/product/renewable-hydrogen-renewable-
energy-and-renewable-hydrogen-apac-markets-policies-analysis-1st-
edition-matthieu-jaunatre/
Mathematical Analysis and Analytical Modeling 1st

Edition Ivan Stanimirovic■
https://ebookmeta.com/product/mathematical-analysis-and-
analytical-modeling-1st-edition-ivan-stanimirovic/
Windfall Profit in Portfolio Diversification An

Empirical Analysis of the Potential Benefits of
Renewable Energy Investments An Empirical Analysis of
the Potential Benefits of Renewable Energy Investments
1st Edition Frederik Bruns
https://ebookmeta.com/product/windfall-profit-in-portfolio-
diversification-an-empirical-analysis-of-the-potential-benefits-
of-renewable-energy-investments-an-empirical-analysis-of-the-
Renewable Energy 1st Edition Rahman
https://ebookmeta.com/product/renewable-energy-1st-edition-
rahman/
Applied Modeling Techniques and Data Analysis 1:

Computational Data Analysis Methods and Tools 1st
Edition Yiannis Dimotikalis
https://ebookmeta.com/product/applied-modeling-techniques-and-
data-analysis-1-computational-data-analysis-methods-and-
tools-1st-edition-yiannis-dimotikalis/
Advanced Mathematical Modeling with Technology Advances

in Applied Mathematics 1st Edition Fox William P Burks
Robert E
https://ebookmeta.com/product/advanced-mathematical-modeling-
with-technology-advances-in-applied-mathematics-1st-edition-fox-
william-p-burks-robert-e/
Renewable Energy Systems : The Earthscan Expert Guide

to Renewable Energy Technologies for Home and Business
1st Edition Dilwyn Jenkins
https://ebookmeta.com/product/renewable-energy-systems-the-
earthscan-expert-guide-to-renewable-energy-technologies-for-home-
and-business-1st-edition-dilwyn-jenkins/
Topics in Applied Mathematics and Modeling 1st Edition

Oscar Gonzalez
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
Applied Soft Computing and Embedded System Applications

in Solar Energy
Rupendra Kumar Pachauri, J. K. Pandey, Abhishek Sharma, Om Prakash
Nautiyal, Mange Ram
Differential Equations in Engineering

Research and Applications
Edited by Nupur Goyal, Piotr Kulczycki, and Mangey Ram
Sustainability in Industry 4.0

Challenges and Remedies
Edited by Shwetank Avikal, Amit Raj Singh, Mangey Ram
Applied Mathematical Modeling and Analysis in Renewable Energy

Edited by Manoj Sahni and Ritu Sahni
For more information about this series, please visit: https://www.routledge.

com/Mathematical-Engineering-Manufacturing-and-Management-Sciences/
book-series/CRCMEMMS
Applied Mathematical
Modeling and Analysis
in Renewable Energy
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
2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN
© 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.
ISBN: 9780367746988 (hbk)

ISBN: 9780367747008 (pbk)
ISBN: 9781003159124 (ebk)
DOI: 10.1201/9781003159124
Typeset in Times
by KnowledgeWorks Global Ltd.
Contents
Preface......................................................................................................................vii
Acknowledgments......................................................................................................ix
About the Editors.......................................................................................................xi
Contributors............................................................................................................ xiii
PART I Mathematical Modeling

and Simulation Results
Chapter 1 Mathematical Modeling of Transitional Fluid Phase Flows in

Coal Bed Methane Reservoir................................................................3
Subhashini Nainar and Suresh Kumar Govindarajan
Chapter 2 AI-Based Detection and Localization of Gastrointestinal

Polyps by Using Deep Learning, Transfer Learning
and the Fusion of These Techniques................................................... 21
Yogesh Chaudhari, Umme Salma Pirzada, and Darshee Baxi
Chapter 3 Mathematical Modeling of an EOQ for a Multi-item Inventory

System with Selling Price and Price Break Sensitive Demand.......... 35
Abhijit Barman and Pijus Kanti De
Chapter 4 Fair Allocation of Items: A Comprehensive Study............................. 47

Manisha Bansal and Purnima Bindal
Chapter 5 Hierarchical Demand Response Controller........................................ 57

Dima Kayyali, Hussam Nosair, Amit V. Sant,
and Hannah Michalska
PART II G
eneralized Mathematical
Ideas and Their Applications
Chapter 6 A General Class of Polynomials Inspired by a General

Lagrange Inversion Pair Due to Gessel and Stanton........................... 79
Manisha Dalbhide-Ubale
v
vi Contents
Chapter 7 Squeezing Graphs.............................................................................. 101

Ved Suthar
Chapter 8 Finding the Surface Area and Volume of the Hyperspheres

Using Simple Calculus...................................................................... 125
Shantanu, Ritu Sahni, and Manoj Sahni
PART III Mathematical Modeling in Renewable Energy
Chapter 9 Analysis of RSM Method for Optimization of Ultrasound-

Assisted KOH Catalyzed Biodiesel Production from Waste
Cotton-Seed Cooking Oil.................................................................. 133
Suvik Oza, Pravin Kodgire, and Surendra Singh Kachhwaha
Chapter 10 Energy Data Analysis of an Educational Institution

in India.............................................................................................. 149
Chandana Sasidharan and Aman Aggarwal
Chapter 11 Factors to Consider: A Review of Smart Grid

Implementation in India.................................................................... 163
Atmiya Patel, Vipul N. Rajput, Kartik S. Pandya,
and Dipayan Guha
Chapter 12 Integration and Modeling of Small-Scale Pumped Storage.............. 181

Jyoti Gupta and Arun Kumar
Index....................................................................................................................... 193
Preface
Mathematics is the backbone for the development of scientific and technical fields. It
makes our life orderly and prevents chaos. This is a subject which provides power of
reasoning, creativity, spatial thinking, and critical thinking; basically, it provides the
way to solve worldly problems related to day-to-day marketing, maintaining credit
card bills, electricity bills, banking, medical insurance, business, computer simula-
tion, education system, statistical data in various cases, engineering problems, bio-
logical systems, and more. The mathematical techniques become the strength of
everyone who works in the field of engineering, sciences, business, medical profes-
sion, politics, etc. to deal with various problems and perform different tasks. This
book highlights the latest mathematical techniques, novel research in the area of
mathematical modeling, and also their applications. The purpose of this book is to
provide both the latest techniques discovered and the advance study of various physi-
cal phenomena. Each of its chapters contain some advanced technique for solving
practical applications. It mainly focuses on the mathematical modeling of renewable
energy, its benefits, and its impact on environment and society. Due to increases in
population, we have a shortage of the energy generated by fossil fuel; society is fac-
ing challenges in electricity generation. So, there is a need to generate power from
other sources so that we can have a better quality of life. Many technologies are
developed for the generation of energy, including solar, wind, hydro power energy,
biomass, biofuel energy, etc., which are longlasting natural resources. This book
contains mathematical modeling of various renewable energy systems, and provides
methods for creating integrated environments through effective energy management
and control. Basically, it provides a brief overview of new emerging technologies
for the generation and management of energy systems. It is a collection of novel
advancements, both with conceptual and mathematical works containing various
physical problems, a methodology for generations and smart storage of renewable
energy.
This book is written considering readers with mathematical backgrounds, and
with a basic knowledge of physics, chemistry, engineering, statistics, fluid mechan-
ics, thermodynamics, heat transfer, and renewable energy. It is important for scien-
tists, researchers, students, teachers, and more who have the scope to modify these
results, and can use it for making their projects by introducing new results in this
area. Teachers can use these results for explaining applications of various math-
ematical tools. As many chapters are based on modeling of energy storage, this book
also benefits both manufacturing and energy industries.
In conclusion, the text explores various challenging areas dealing with various
physical problems, their modeling, and novel, advanced mathematical techniques
used in day-to-day life and in the energy sector. It paves the way to dealing with
other problems using advanced techniques mentioned in different chapters, and so, it
is very useful in present and future scenarios. We hope that all readers benefit from
this book and succeed with the great effort made in this book.
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.
Dr. Manoj Sahni

Dr. Ritu Sahni
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.
Ritu Sahni has been an assistant professor,

Department of Physical Sciences, Institute of
Advanced Research, Gandhinagar for the last
six years. Previously, she worked at Jaypee
Institute of Information Technology, Noida
and Navrachna University, Vadodara. Her total
teaching and research experience is approxi-
mately 12 years. She has published more than
35 research papers in various international jour-
nals and conference proceedings. She is a Life
Member of various prestigious societies such as
Indian Science Congress, Indian Mathematical
Society, and more. Currently, she is working in the area of fixed-point theory and
fuzzy modeling.
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
Pijus Kanti De Arun Kumar

Department of Mathematics Department of Hydro and Renewable
National Institute of Technology Energy
Silchar, India Indian Institute of Technology
Roorkee, India
Suresh Kumar Govindarajan
Indian Institute of Technology Hannah Michalska
(IIT)-Madras McGill University
Chennai, India Montreal, Canada
xiii
xiv Contributors
Subhashini Nainar Vipul N. Rajput

Indian Institute of Technology Dr. Jivraj Mehta Institute of Technology
(IIT)-Madras Anand, India
Chennai, India
Manoj Sahni
Hussam Nosair Pandit Deendayal Energy University
New York Independent System (PDEU)
Operator Gandhinagar, India
New York, USA
Ritu Sahni
Suvik Oza Institute of Advanced Research (IAR)
Chemical Engineering Department Gandhinagar, India
Center for Biofuel and Bioenergy
Studies Amit V. Sant
Pandit Deendayal Petroleum University Pandit Deendayal Petroleum
Gandhinagar, India University
Gandhinagar, India
Kartik S. Pandya
Charotar University of Science and Chandana Sasidharan
Technology Teri School of Advanced Studies
Anand, India New Delhi, India
Atmiya Patel Shantanu

Dr. Jivraj Mehta Institute of Technology Indian Institute of Science Education
Anand, India and Research (IISER)
Pune, India
Umme Salma Pirzada
School of Engineering and Technology Ved Suthar
Navrachana University Indus University
Vadodara, India Ahmedabad, India
Part I
Mathematical Modeling
and Simulation Results
1 Mathematical Modeling
of Transitional Fluid
Phase Flows in Coal
Bed Methane Reservoir
Subhashini Nainar
and Suresh Kumar Govindarajan
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.
1.1.1 Conceptual Model

A three-dimensional coal formation is shown. It comprises cleat network interwoven
between porous matrices. The gas is adsorbed on the coal grains contrary to being
present in pores as observed in the conventional oil/gas reservoirs. Face cleats are
also shown in the figure.
The three-dimensional coal formation (Fig. 1.1a) is reduced to a two-dimensional
model (Fig. 1.1b). The radius is varied dynamically with production time. The study
length is the limiting length used for the analysis. Payzone width is used in the com-
putation of drained volume of gas.
The two-dimensional model is further reduced to one-dimensional single cleat
flow analysis (Fig. 1.1c). It is divided into nodes along the study length. The number
of nodes increases with time. The varying wellbore pressure is indicated at the initial
node (Fig. 1.1d).
The time versus drained distance is plotted for the cleat. With an increase in pro-
duction time, the drained distance increases away from the wellbore center, and so
does the number of nodes.
The coal formation is heterogeneous and assumed to be isotropic. One-
dimensional model is studied. The flow behavior of a single cleat and the associated
coal matrix around is studied. The equations are framed, considering all the phases
starting from single-phase water flow, then multiphase gas and water flow, followed
by single-phase gas flow. The verified and validated code for the dewatered coal is
further extended to include the single-phase gas flow equations.
1.2 PERMEABILITY MODEL
k n = k i × e −(3× cfn ×(σ −σ i )n ) (1.1)

(Li et al., 2017)
Mathematical Modeling of Transitional Fluid Phase Flows 5
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.
   −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)
1.2.2 Variable Coefficients

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 
gpiv
qgiv = (1.9)
t + dt
Bwn = e(.0000026 ×(14.7− Pn )) (1.10)

cfn =
(
cfi × 1 − e −α (σ −σ i )n ) (1.11)
α (σ − σ i )n
(Robertson, 2006)
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.
1.3 SINGLE-PHASE WATER FLOW

The water in the matrix is ignored since dewatered cleat water is sufficient to lower
down the reservoir pressure to desorption pressure. Overburden stress is constant,
and only the fluid pressure varies with time. Overburden stress is vertical, and the
sides of reservoir length are assumed to be constrained horizontally. Any deforma-
tion in the vertical direction exerts stress horizontally. The horizontal stress includes
only shrinkage stress, when there is no injection of supercritical carbon dioxide
(sCO2), and includes both swelling and shrinkage stresses, when sCO2 is injected.
Dependent variablesa: Pnt + dt Independent variable: Xinv t + dt , Ltw+ dt , t

Variable coefficients: k nt+ dt , cfnt + dt, ϕ nt+ dt, Bwt+ndt , wpivt+ dt , qw t + dt
iv
Constant coefficients: Sw, h, dt, dx, T
Note:
a Unknown.
1.3.1 Mathematical Model

Water flow through cleats:
φiv ( P ) × cfiv ( P ) × Swiv ( P ) ∂P k iv ( P ) ∂2 P

× = × 2 (1.14)
Bwiv ( P ) ∂ t µwiv ( P ) × Bwiv ( P ) ∂x
(Derived from mass continuity equation)
Equation 1.14 is a non-linear partial differential equation applicable in case of water

flow in the cleats of coal formation. The equation is diffusive and, thus, parabolic.
Since the equation is parabolic dominant, the solution converges after some time.
The equation is non-linear due to the presence of six variable coefficients.
1.3.2 Initial and Boundary Conditions

1. Boundary condition to the right (Dirichlet): P( N ) t = Pi
2. Initial condition (Dirichlet): Pn 0 = Pi
3. Boundary condition at left on the first node (Dirichlet): P0 t = Pwf
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.
1.4 MULTIPHASE GAS AND WATER FLOW
Dependent variablesa: Pnt + dt , Swn t + dt Independent variable: Xinv t + dt , Ltw+ dt , t

Variable coefficients: k nt+ dt , cfnt + dt, ϕ nt+ dt, Bwt+ndt , Constant coefficients: h, dt, dx, T
ct nt+ dt , cl nt+ dt , wpivt+ dt , gpivt+ dt , k rg t + dt, k rw t + dt, Sgn t + dt
Note:
a Unknown.
1.4.1 Mathematical Model

Water flow through cleats:
µwiv ( P ) × φiv ( P )  ∂P ∂Sw 

 ct iv ( Sw , P ) × Swiv × + 
k iv ( P ) × k rw ( Sw )  ∂t ∂t 
2
∂2 P dP 
= + cl iv ( Sw , P ) ×   (1.15)
∂x 2  dx 
2
dP 
cl iv ( Sw , P ) ×   ≈ 0 ( Since compressibility of water is negligible ) (1.16)
 dx 
It can be written as:

µwiv ( P ) × φiv ( P )  ∂P ∂Sw  ∂2 P
 ct iv ( P ) × Swiv × + = (1.17)
k iv ( P ) × k rw ( Sw ) ∂t ∂ t  ∂x 2
Gas flow through cleats:
Mass continuity equation:
(
∂ 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 )
FIGURE 1.2 The coding procedure is followed with the use of assumptions and calculation
of dependent variables for single-phase water flow.
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.
1.4.2 Initial and Boundary Conditions

Initial condition: Pressure at initial time (t = Nt; time period of single-phase
water flow) is equal to previous time nodal pressures.
Boundary condition: Same as for single-phase water flow stage.
1.4.3 Methodology to Solve for Multiphase Flow

IMPES
To solve for pressure at node ‘n’ at time ‘t+dt’ implicitly and to proceed to obtain
for the value of water saturation explicitly at node ‘n’ for time ‘t+dt.’
Steps:
µwiv × k rg
1. Step 1: Multiply equation 1.6 by µgiv × k rw
, we get:
µw iv × Bgiv  ct iv × φiv × Sg iv ∂P φiv ∂Sg VL × PL ∂P 

 × + × + × 
k iv × k rw  Bgiv ∂ t Bgiv ∂ t ( P + PL )2 ∂ t 
(1.23)
µw iv × k rg ∂ P µw × k rg dP 
2
× cl iv ×  
2
= × 2 + iv
µg iv × k rw ∂x µgiv × k rw  dx 
2. Step 2: Add equations 1.17 and 1.23.
µwiv × φiv  ∂P ∂Sw 

 ct iv × Swiv × + 
k iv × k rw  ∂t ∂t 
µwiv × Bgiv  ct iv × φiv × Sg iv ∂P φiv ∂Sg VL × PL ∂P 

+  × + × + ×  (1.24)
k iv × k rw  Bg iv ∂ t Bgiv ∂ t ( P + PL )2 ∂ t 
∂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 
µwiv × φiv ∂P µwiv

× ct iv × Swiv × + × ct iv × φiv × Sgiv
k iv × k rw ∂ t k iv × k rw
∂P µwiv × φiv ∂Sw µwiv ∂Sg
× + × + × φiv ×
∂ t k iv × k rw ∂t k iv × k rw ∂t
(1.25)
µw × Bgiv VL × PL ∂P ∂2 P µwiv × k rg
+ iv × = = +
k iv × k rw (P + PL ) 2 ∂ t ∂ x 2 µgiv × k rw
∂2 P µwiv × k rg Sgiv  dP 
2
× + × ×  
∂ x 2 µgiv × 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 
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 
Equation 1.26 is the water-gas saturation relation. Equation 1.27 is splitting

equation 1.25 into time and distance terms. Equation 1.29 is the decoupled
form of non-linear partial differential equation (PDE) in terms of p and respec-
tive coefficients.
3. Step 3: Substitute ‘P’ and linearize equation 1.29.
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 
(v nt+ dt − v nt ) = G × (v nt ++dt1 − 2 × v nt+ dt + v nt +−dt1 )

(1.36)
+ H × cl iv × ( P [ N ] − P [ 0 ]) × (v nt ++dt1 − v nt +−dt1 )
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.
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.
1.4.5 Water Saturation to be Solved Explicitly

Using Equation 1.17
Equation 1.42 is the water saturation equation, and equation 1.45 is the gas saturation
equation.
µwiv ( P ) × φiv ( P )   P t + dt − Pnt   Swt+ dt − Swt  

 cfiv ( Sw , P ) × Swiv ×  n  +   
k iv ( P ) × k rw ( Sw )   dt dt 
(1.40)
 P t + dt − 2 × Pnt + dt + Pnt −+1dt 
=  n +1 
 dX 2
(Swt+ dt ) × (1 + cfiv × ( Pnt + dt − Pnt ))

 k × k rw  Pnt ++1dt − 2 × Pnt + dt + Pnt −+1dt  
=  Swt + dt × iv × (1.41)
 µwiv × φiv  dX 2
 

(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.
of dependent variables for multiphase gas and water flow.
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.
1.5 VERIFICATION AND VALIDATION

Verification:
The permeability and porosity relation, as mentioned in the reference journal, is
pressure-dependent. Therefore, to verify the code with semi-analytic (semi-numerical)
results, the following empirical relationships are used:
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.
1.5.1 Verification and Validation Results

In the plots of cumulative production and the rate of gas and water against the
number of days, data (semi-numerical model) could not be possible as the ini-
tial input values of some variables differ. The missing initial input data for the
calculation obtained from different works done by the authors such as Clarkson
and Qanbari et al. (2016), Zhang et al. (2008), and Robertson (2006) are used.
These works are referred to for similar coal formation (Manville). From the
above verification and validation plots, cumulative water produced shows a trend
similar to both field values and semi-analytic numerical values from the refer-
ence paper (Fig. 1.4(a)). Water is produced both during the single-phase and
multiphase flow stages. Initially, the increase in production is approximately
linear since water saturation is assumed constant throughout the single-phase
flow. With the onset of multiphase flow, the plot becomes non-linear since water
produced decreases with a decrease in water saturation in cleats. Cumulative gas
(Fig. 1.4(b)) increases continuously with time due to an increase in gas saturation
in cleats. The increase is non-linear because there is no proportional relation of
production with gas saturation, and is also dependent on other variables. The
water rate shows a linearly increasing trend during the initial stages of produc-
tion, followed by a reducing trend when gas flow starts (Fig. 1.4(c)). Due to the
decreasing water rate in multiphase flows, the cumulative water produced, dur-
ing this period, depicts a non-linear increase. The gas rate increases with time
following the dewatering process (Fig. 1.4(d)). The code for the single-phase
water followed by multiphase water and gas flow is verified and validated and
used for further calculations.
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).
FIGURE 1.4 (Continued)
1.6 SINGLE-PHASE GAS FLOW
∂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 
of dependent variables for single-phase gas flow.
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.
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 ) .
cf : Cleat compressibility ( psi −1 ) .

ct: Total compressibility ( psi −1 ) .
cl: Fluid compressibility ( psi −1 ) .
ε L : Langmuir volumetric strain.
σ : Effective horizontal stress ( psi ) .
ν : Poisson’s ratio of coal.
Bw : Fomation volume factor of water ( stb rb
).
µw : Water viscosity (cp).
Sw : Water saturation.
N : Number of points along X direction.
t: Time one dt before present investigating time ( t + dt ) ( days ) .
Xinv: Drained distance at time ( t + dt )(ft).
wp: Cumulative water produced ( stb ) .
ct: Total compressibility ( psi −1 ) .
h: Thickness of formation ( ft ) .
Nt: Total transient phase time ( days ) .
dt: Time increment within time Nt ( days ) .
dx: Space interval in distance of xinv ( ft ) .
qw: Water flow rate ( stb / d ) .
gp: Cumulative gas produced ( Mscf ).
qg: Gas flow rate ( Mscf / d ).
Bg: Fomation volume factor of gas ( stb rb
).
µg: Gas viscosity (cp).
Sg : Gas saturation.
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
2.1 INTRODUCTION: BACKGROUND AND DRIVING FORCES

As per the World Health Organization (WHO) statistics, all around 1,800,000 indi-
viduals are diagnosed and around 862,000 individuals lost their life in the year
2018 globally due to colorectal cancer (CRC). Among the various diagnosis meth-
ods, colonoscopy is the most reliable norm for screening of CRC. The petite forma-
tion of cells and enlargements of lesions in the inward covering of colon or rectum
DOI: 10.1201/9781003159124-2 21
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
(7–53%), and 20–30% of adenomas could be missed during screening colonoscopies.

The ADR performance targets are 25% in general, ≥30% for men and ≥20% for
women [3].
Automated examination for detection and localization of polyps may signifi-
cantly increase ADR and control the high missed diagnosis rate. A few extra
procedures and apparatuses are under analysis for improving an endoscopist’s
capability to recognize adenomas with the objective to expand productivity and
lessening costs. The convolutional neural network (CNN)-based diagnosis sys-
tems adapt exceptionally fast for the computer-aided diagnosis (CAD) for detec-
tion, localization and grouping of polyps. The CAD helps image investigation that
junctions both expanded polyp recognizable proof and histopathologic separation
without adjustments to the imaging instruments or the standard method. The point
when CNN is implemented in fusion with feature extractor and transfer learn-
ing techniques, the general execution of the CAD system is improved in terms
of learning time and accuracy [4]. We have made an attempt to implement the
transfer learning using Visual Geometry Group (VGG)16 and VGG19 techniques
to further improve the performance of deep learning architecture implemented
using CNN architecture and to study its effect on performance with respect to the
number of epochs.
2.2 DEEP NEURAL NETWORK (DNN)

The DNN is designed by cascading multiple layers of artificial neural network
(ANN). The application of deep learning techniques to recognize objects and
their use in the field of computer vision have been profoundly studied since 2000,
and the outcomes are well accepted [5]. Deep learning algorithms are designed
to learn visual feature hierarchies that can be implemented using either unsuper-
vised learning algorithms (if labeled data samples are relatively low) or super-
vised learning algorithms (if labeled data samples are relatively sufficient). The
successes of supervised learning CNN-based architectures demonstrated in 2012
in ImageNet competition garnered the growing attention of computer vision com-
munity that outperformed other approaches in terms of recognition accuracy. The
work published in ref. [6] validates the use of deep learning for auto-detection of
polyps during colonoscopy, and the results are promising though the authors have
cautioned related to the risk of major complications remain similar to conven-
tional colonoscopy.
A CNN architecture is perfectly suitable for visual applications as its design and
development is inspired by the natural mechanism of ‘visual perception and atten-
tion’ [7]. The discovery by Hubel & Wiesel in 1959 [8] further motivated Kunihiko
Fukushima who designed the neocognitron – a neural network-based self-organizing
model to mimic visual pattern recognition mechanism in 1980 [9]. The work LeNet-5
published by Yann Le Cun in refs. [10] and [11] was capable of recognizing hand-
written character and classifying when two-dimensional images are given as an
input to the model. Zhang X. et al. [12] have discussed the use of CNN architecture
for polyp detection and reported the finding that the implantation can help in reduc-
ing the polyp miss rate.
2.3 TRANSFER LEARNING

The professor and data scientist Andrew Ng stated in NIPS 2016 that after super-
vised learning, transfer learning will be the next driver of commercial success
of machine learning (ML). Transfer learning relaxes the hypothesis that the
training data must be independent and identically distributed with the test data,
which motivates us to use transfer learning to solve the problem of insufficient
training data.
Transfer learning is normally an approach wherein a model trained on one
problem is tuned in a way to work on other similar problem. The transfer of
learning in CNNs showed great potential for celiac disease classification based
on endoscopic images [13]. The authors used four CNNs that were pre-trained on
the ImageNet database. Three different transfer learning strategies were explored
to classify the endoscopic images of the celiac disease. Full fine-tuning of the
CNNs achieved the highest classification accuracies, although the small amount
of training data available led to overfitting [14]. The Fig. 2.2 shows the difference
in difference in knowledge transfer in (a) conventional machine learning and in
(b) transfer learning.
FIGURE 2.2 Conventional machine learning (a) vs transfer learning and (b) in progress.
2.4 POLYPS AND COMPUTER-AIDED DIAGNOSIS

In ref. [15], the authors studied the feasibility of polyp recognition and classification
of two types of polyps – hyperplasias and adenomas. The study uses image segmen-
tation and is based on the use of visual feature of vascularization for computer-based
classification. The approaches designed with feature extraction and that are based
on more polyp features like shape, size and color of polyps will lead to improved
diagnosis. Figure 2.3 shows the growth of polyps from stage 1 to stage four of malig-
nancy. The larger polyp is in metastasis stage as it spreads to nearby tissues of its
origin.
2.5 DATASET AND EXPERIMENTAL SETUP

For the experimental setup, the public dataset CVC-Clinic DB has been explored
[16]. The polyp positive and negative images are prepared using annotation. The
cropped images are grouped for the purpose of training, testing and validation.
In each of these categories, the images are further sub-grouped as positive and
negative.
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.)
TABLE 2.1
Description of Dataset
Sr. No. Particular Details

1 Original images 612 frames extracted from colonoscopy
videos
2 Polyp mask 612 images corresponding each to
original images
Monochrome image representing polyp
with ‘white’ and ‘black’ otherwise
3 Sequences 29
4 Dimensions 384 by 288
5 Size of RGB image 324 kB
6 Size of polyp mask 108 kB
7 Size of database 258 MB
Vision Challenge. The CVC-Clinic DB database consists of two sets of images:

(1) colored Red Green Blue (RGB) images and (2) corresponding monochrome polyp
masks. The description of dataset is provided in Table 2.1.
Figure 2.4 shows 3 random images taken from the dataset along with the
corresponding monochrome mask for each of these images.
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.)
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.
2.5.3 CNN Architecture

The CNN architecture is designed (as shown in Fig. 2.6) for the training and testing
of the samples from dataset. The CNN model has been implemented with Keras. A
CNN architecture has three main parts:
• A convolutional layer to extracts features

• A pooling layer to reduce the dimensionality with no loss of features or
patterns
• A fully connected (dense) layer to generate a prediction
Three systems are designed to detect and locate the polyp as below:
• Using CNN architecture with layers as discussed in Fig. 2.7

• Using VGG16 pre-trained model
• Using VGG16 pre-trained model with fusion of fully connected (FC) layer
TABLE 2.2
Number of Data Samples in Each Set
Sr. No. Set Subset Samples Total Percentage

1 Train Positive 1820 3640 36.11 72.22
2 Negative 1820 36.11
3 Validation Positive 600 1200 11.90 23.81
4 Negative 600 11.90
5 Testing Positive 100 200 1.98 3.97
6 Negative 100 1.98
FIGURE 2.6 Major blocks of proposed system for polyp detection.
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.
2.5.4 VGG16 and ND VGG19

Various pre-trained CNN models are available that are trained on ImageNet dataset
inside the Keras library. These pre-trained models can be utilized for prediction,
feature extraction and fine-tuning. Few pre-trained models are listed below:
• VGG16
• VGG19
FIGURE 2.7 Sequential representation of CNN architecture.

• ResNet50
• Inception V3
• Xception
• GoogLeNet
In comparison to VGG16, VGG19 performs better with more memory require-

ment. VGG16 and VGG19 models are designed with convolution layers, max
pooling layers and fully connected layers, with total 16 and 19 layers, respec-
tively. With the reference to the work published in ref. [17], VGG16 and VGG19
are explored and also modified [18] to test the performance for the diagnosis
of gastrointestinal polyps. The diagrammatic representation of (a) VGG16 and
(b) VGG19 is shown in Fig. 2.8.
FIGURE 2.8 Layered architecture of (a) VGG16 and (b) VGG19.

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.
2.6 RESULTS AND DISCUSSION

In the processing of input images, the precaution is taken due to varying dimen-
sionality of samples in training and validation set. We have experimented with the
following ten different implantations as detailed in Table 2.3.
After execution of the above model, we obtained the results as shown in Fig. 2.9.
The models are arranged in the increasing order of accuracy. It is observed during
experimentation that as the number of layers in a model increases, the accuracy also
increases, with the exception at model number 5. The structure of model numbers
8, 9 and 10 is similar, and improvement is achieved by reducing the drop rate. For
the VGG implantation, we have used the batch size of 64 and trained the network
for same number of epochs, i.e., 100. The training and validation loss (VL) results
obtained for VGG16 implementation are shown in Fig. 2.5.
For VGG16 implementation, the overfitting is observed as VL is far greater than
training loss (TL) as shown in Fig. 2.10(a). The drop rate is set to 0.92 to reduce
overfitting and the obtained result is shown in Fig. 2.10(b).
From Table 2.4, we can observe that the difference between VL and TL is sig-
nificantly reduced by increasing the drop rate. The third implementation of VGG16
with training of last convolutional block with fully connected layer is underfitting for
more number (150) of epochs. With fine-tuning and working with hyperparameters
like ‘learning rate’ and ‘momentum’, better results are obtained with model accuracy
of 93.35.
TABLE 2.3
Details of Different Combinations of Architectures Implemented
Convolutional Number of Filters Number of

Sr. No. Layers Used Kernel Size Epochs
1 1 32 0 0 5×5 100
2 1 32 0 0 3×3 100
3 1 64 0 0 3×3 100
4 2 64 128 0 3×3 100
5 1 128 0 0 3×3 100
6 2 32 64 0 3×3 100
7 3 32 32 64 3×3 100
8 3 64 64 128 3×3 100
9 3 64 64 128 3×3 100
10 3 64 64 128 3×3 100
Another random document with
no related content on Scribd:
of the stomach are disgorged, or the excrement voided, either of
which is adroitly caught by this foul freebooter of the sea before it
reaches the water.
A hazy moonless night, with a sou’-easterly breeze and drizzling
rain—given these conditions, at this season of the year we have
numerous visits of various birds, members of the autumnal migratory
flight. Making straight for the light, they dash themselves against the
heavy plate-glass of the lantern; many of them are thus killed and
swept by the wind into the sea. Others, again, arrive with more
caution, and though taken in the hand and thrown clear of the tower
invariably return, and remain fluttering against the glass till daylight
reveals to them the futility of their exertions in that direction. The
most numerous of these visitors are the redwings and fieldfares, but
blackbirds, larks, starlings, wheatears, finches, tits, etc., may be met
with in the course of the season. It is somewhat startling, when on
watch in the lightroom, to hear the thud with which they strike. The
woodcock, owing to his rapid flight, strikes hardest of all, and the
other extreme is met with in the smallest of our British birds, the tiny
gold-crested wren, whose presence on the lantern is announced by
a feeble tinkling sound, which a robust butterfly might easily imitate.
The heavier birds do not always strike with impunity; instances have
occurred where ducks have gone clean through the lantern to the
derangement of the revolving gear of the light, the splintered glass
bringing the machinery to a dead stop. An incident of this nature
happened a few years ago at Turnberry Lighthouse, on the Ayrshire
coast, the intruder in this case being a curlew or whaup. A storm-
pane is considered a necessary adjunct to every lightroom, and is
always held in readiness to be shipped in case of such emergency.
At some shore stations it is customary on the approach of a
favourable night, during the migratory period, to keep the cats
indoors to prevent them mangling the expected catch. In one
particular instance the birds collected of a morning filled an ordinary
clothes-basket, and a few nights later included five wild geese, which
were secured out of a large flock that came to grief on the dome.
An hour before daybreak on the 22nd it appeared as if we were
about to suffer a bombardment, and that daylight was to witness the
commencement of hostilities. No less than seven torpedo-boat
destroyers were seen creeping close up to the Rock, their low black
hulls scarcely discernible in the feeble light, and not until daylight
disclosed the white ensign were we assured of their intentions. A
little later they were joined by three gunboats and, after some clever
manœuvring, formed into three lines, the gunboats occupying the
centre. They then steamed away in the direction of the Firth of Forth.
Two hours later other three gunboats passed us, going in the same
direction, escorted by four destroyers, and followed shortly after by a
solitary gunboat. Extremely interesting it was to witness the precision
and dexterity of their movements as they swung into their respective
positions for the advance, their semaphores all the while going like
windmills. Again, on the 24th, about 11 a.m., a fleet of about a dozen
battleships, headed by a dispatch boat, was seen moving in stately
procession from the Tay, evidently bound for the Forth.
We have had several heliographic communications from our
shore station in Arbroath during the month, and providing there is
sunshine there is now no difficulty in transmitting messages to the
Rock by this means. Four years ago the late Dr Russell, Arbroath,
while on a professional visit to the shore station, for which he was
medical attendant, witnessed our initial attempts in this direction,
and, convinced of the feasibility of the method, urged upon us, in his
characteristically vigorous style, the necessity for persevering in our
attempts, at the same time predicting that it would ultimately prove
successful. Little did we then dream it was soon to become the
means of conveying the sorrowful intelligence of this estimable
gentleman’s death.
NOVEMBER 1901.
Boisterous weather prevailing for the greater part of this month, we

have been closely confined to the house. Our connection with the
amphibia being so extremely remote completely disqualifies us from
enjoying our usual “constitutional,” the grating, even at low water,
being occasionally swept by the heavy seas. Our winter boarders,
the eider ducks, have been reinforced, on the morning of the 14th—
somewhat later than usual—by the arrival of a flock of long-tailed
ducks. These, with the eiders, will keep us company till April again
calls their attention to domestic affairs. Our relief, which was due on
the night of the 11th, was effected just in time; had it been delayed
another day a “missed relief” would probably have been recorded.
The morning after brought a severe north-easterly gale, which
precluded all possibility of making a landing during the three
succeeding days. That is usually the time allotted by the steamer in
the attempt. Should she fail to make a landing on the third day, we
are abandoned for another fortnight, minus the time engaged in the
attempt. As our stock of fresh provisions is generally consumed by
the time the relief is due, a missed relief means a fortnight’s regime
of “hard tack” and “beef embalmed,” of which during the winter
months we have a three months’ reserve stock on hand in case of
such emergencies. Fortunately, this is not of common occurrence;
during the past six years but three reliefs have been missed, and
only one in the preceding ten. This speaks much for the ability and
skill of those concerned in the handling of the boats, for during the
winter months the landings were until recently effected in darkness,
and an exciting scene it was to see the two boats buffeting their way
through the foaming channels, with jutting rocks so close on either
side that an oar’s length deviation would entail serious disaster. A
powerful searchlight has of recent years been added to the
equipment of the relieving steamer, and is of much advantage in the
guidance of the boats, though it has the peculiarity of grossly
exaggerating the tempestuous appearance of the sea. The sea,
which on the evening of the relief was comparatively calm, was the
next day rolling down on us like a solid wall, and viewed from the
balcony in all its magnificent grandeur what a puny, frail, unstable
structure our habitation seemed in comparison. Each succeeding
wave seemed imbued with the sole motive of accomplishing our
destruction, and as they struck and sliced away on either side in two
mighty crescents of hissing foam, blinded our kitchen windows
seventy feet above the rock. Clashing together again to leeward with
a roar, as if incensed at our stubborn resistance, they drive their way
furiously along the remaining portion of the reef in foam-capped
ridges, and where the cross seas meet them the spray is flung high
in the air from their points of intersection. The appearance of the reef
at this stage, as seen from our elevation, is of a number of
rectangular enclosures, each about the size of an ordinary bowling-
green, with well-defined walls, the whole under a heavy coating of
snow, with each corner marked by a snow-laden tree. At high
water—the sea having flowed about twelve or fifteen feet on the
building by that time—the waves, generally unbroken, slip past
harmlessly; an hour before or after high water is when we
experience the heaviest shocks, for then the depth of water is such
that the waves are arrested by the rock when close to the tower, and
their whole volume flung violently against the building. The effect of
such weather on the tower must be felt to be understood. The
nearest description I can give of the seas striking is as if a log of
wood were hurled by each sea, striking end on, and a short, sharp,
tremulous motion—sufficient to rattle the crockery in the kitchen
cupboard—is imparted to the tower by each impact. This tremor is
more particularly felt when the gale subsides and the heavy swell
sets in, for when the gale is at its height, the seas are so broken and
tossed about that their assaults are but feeble in comparison with
those of the long curly-headed combers of the after-swell. The bell-
shaped formation of the base of the tower is admirably adapted for
withstanding the assaults of the sea, and is built solid to a height of
thirty feet, above which the seas never strike, though I have seen the
spray carried right over our balcony, a hundred feet from the rock.
That the building remains to all appearance as intact as when
completed, almost a century ago, speaks volumes for the skill and
ingenuity displayed in its erection. In weather such as I have
described we are as completely cut off from outside assistance as
though we were at the North Pole; indeed, it is doubtful if there is
another situation—save similar ones, of course—where men could
live so comfortable and unconcerned and yet remain for the time
being so completely “ungetatable.”
DECEMBER 1901.
As a consequence of the stormy weather which has been prevailing

here of late, we have been visited by numerous “Travellers.” This
may seem strange considering the inclemency of the season, but
stranger still when it is known that our reception of them is fiercely
hostile, and our duty only considered accomplished when we have
completely annihilated them. Huge boulders of hard red sandstone,
sometimes weighing over three tons; these are our “Travellers,” and
their appearance on the Rock is at once resented and their speedy
removal effected by blasting and hammering whenever the tide and
weather permits. This is absolutely necessary, for if allowed to
remain lying in the boat tracks they constitute a serious danger at
relief times, besides the possibility of their carrying away portions of
our cast-iron grating, which occasionally does happen in spite of all
precautions. Where they come from is a mystery; ever since the
tower was built they have been in evidence. Although composed of
the same material, the Rock itself does not suffer any apparent
diminution, nor can their original abode be located even at the lowest
tides. Many of them carry a crop of seaweed and tangles, and have
their angularities rubbed down and water-worn; none of them,
however, bear any trace of recent detachment, but probably from
their similarity of structure they at some remote period formed a part
of the reef. They generally effect their entrance from the south side
of the reef during the prevalence of a heavy ground swell. This side
of the reef forms a steep declivity, sloping to 35 fathoms at a
distance of ¾ mile, while at a similar distance on the north side the
depth, though not exceeding 11 fathoms, presents a more
precipitous barrier to these wanderers of the deep. A dull, rumbling
noise, distinctly audible in the light room, announces their presence
at the base of the tower, and at low water a dotted line of chips and
abrasions marks their passage across the Rock to where they are
again hurled to the depths. Others, again, may bring up in some
sheltered corner, where, if not considered dangerous, they may
remain a fixture for years.
An instance occurred recently where one was wheeled against
our grating after occupying a safe position for many years. Those
that take up positions in the boat tracks are of course assailed at the
earliest opportunity, an operation which generally entails a bit of
submarine mining on our part. The reef consists of hard, red
sandstone, arranged in irregular layers, with a dip of 15 degrees
towards the south-east and extends in a north-easterly and south-
westerly direction, having an area of about 500 yards by 100 yards
considered dangerous to shipping. The north-east end, on which the
Lighthouse is built, is slightly higher, and has an area of about 140
yards by 70 yards, the highest portions of which do not exceed 10
feet above the lowest tides. The geological formation of the Bell
Rock is similar to that of the Redhead, in Forfarshire, and can be
traced northward through Rossshire, while in the opposite direction
the shores of Berwick present the same features, and continues as
far as Cumberland. Soundings prove the existence of a ridge or
shallower part of the sea bottom extending a considerable way in
these directions, and as the adjacent coasts present ample evidence
of the sea having at some remote period in the world’s history
occupied a much higher level, the theory that the Bell Rock did not
always occupy the isolated position it now does, but stretched
continuously from the Red Head to Berwick, damming the waters of
the Forth and Tay, appears highly tenable. Possibly our present day
“Travellers” are, through some great seismic disturbance, wandering
evidences supporting this theory.
An item of interest to Arbroath Freemasons is the laying of the
foundation stone of the Bell Rock Lighthouse, on the 10th July 1808,
with Masonic honours, by the builder, Robert Stevenson, who, in his
own words, applied the square, the level, and the mallet, and
pronounced the following benediction:—“May the Great Architect of
the Universe complete and bless this building,” on which three hearty
cheers were given and success to the future operations was drunk
with the greatest enthusiasm. Another interesting feature of that
period was the existence of the “Pressgang,” which, owing to our war
with the Northern Powers, was considered necessary. Centres were
established at Dundee, Aberdeen, and Arbroath, and were the
means of rendering the Lighthouse operations popular with seamen,
as they stood protected from impressment while in that employment.
Prior to this there was a tendency among seamen to shun the works
on account of the hazardous nature of the undertaking. As the
impress officers were exceedingly active in their duty, it was found
necessary to furnish each seaman engaged in the operations at the
Rock with a “ticket,” descriptive of his person, to which was attached
a silver medal, emblematical of the Lighthouse Service. On one side
of the medal was a figure of the Bell Rock Lighthouse, and on the
other the word “Medal,” referring to the Admiralty protection, and a
description of the person by the engineer. One of these medals is at
present in possession of an Arbroath gentleman, and is said to be
the only one in existence. The following is a copy of one of the
“tickets,” taken from “Stevenson’s Bell Rock Lighthouse”:—
Bell Rock Workyard,
Arbroath, 31st March 1808.
“John Pratt, seaman, in the service of the Honourable the
Commissioners of the Northern Lighthouses, aged 35 years,
5 feet 8 inches high, black complexion, and slightly marked
with the smallpox.”
(Signed) Robert Stevenson,
Engineer for Northern Lighthouses.
Obverse.
“The Bearer, John Pratt, is serving on board of the ‘Sir
Joseph Banks’ tender and craft, employed at the erection of
the Bell Rock Lighthouse.”
The signature of the Master of the tender.
(Signed) David Taylor.
The signature of the bearer (Signed) John Pratt.
Notwithstanding these precautions, so rigorous were the impress
officers that they actually pressed a Bell Rock seaman named
George Dall, while on a visit to some friends near Dundee, in July
1810, and this despite the fact of his having the protection medal and
ticket in his possession. These proofs the officer chose to ignore,
holding that a seaman only stood protected on board the ship to
which the Admiralty protection had been granted, or in a boat
belonging to the ship. This was absurd, as it was impossible for each
man to carry the ship’s protection with him. However, Dall was kept a
prisoner, and only on the representations of the Lighthouse
Commissioners did the Dundee Magistrates order his release.
JANUARY 1902.
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.
Signs of uneasiness and unrest are now apparent amongst our

winter boarders, the eiders and long-tailed ducks. Taking wing on the
slightest provocation, they wheel aimlessly round the Rock, and
instead of their usual steady persistence in diving for a living, they
seem quite discontented with their lot, and plainly making up their
minds to desert us for the summer. Advances by the males are as
yet met with scornful rebuffs by their less showy plumaged partners,
but soon a mutual understanding will be arrived at, and before the
month closes they will have gone house-hunting, eiders possibly to
the Isle of May, while the long-tails, being migratory, seek their
homes in the frozen North. It seems a strange anomaly that the less
robust looking longtail should choose such rigorous latitudes for the
rearing of its brood, while the sturdy “dunter,” swathed in his arctic
coat, should elect to stay at home. On the other hand, we have been
visited on hazy nights by numbers of larks and thrushes returning to
our shores, after wintering in “Norroway ower the faem.” These
members of the spring migratory movement often come to grief on
our lantern, and when one considers the number of lighthouses
round our coasts, it will be understood that the death-roll from this
cause alone must be extremely high. Designed to save life, we
unwittingly lure our feathered friends to their destruction.
A couple of seals have been sporting round our door of late, and
they also exhibit signs of exuberance in keeping with the season. At
high water they come quite close to the tower, and their antics are
seen to advantage from our balcony. Rolling over each other, they
make for the bottom, gliding along the rocks like hounds hunting in
couples; then with a rush they are on the surface, floating bolt
upright, with their muzzles almost touching, staring with their large,
expressive eyes into each other’s face. An almost human touch was
given to their play by one taking the head of the other between his
fore-flippers, as if about to salute him, or more likely her, in the
orthodox fashion. One was seen the other morning in possession of
a large fish, while a number of gulls sat at a safe distance round him,
waiting for the fragments when the feast should begin. By the way he
glared at them, he was evidently annoyed at their presence. Sinking
for a few seconds, he appeared on the surface minus the fish. This
was evidently intended as a ruse, and meant to imply that he had
lost it; but the gulls seemed to know better, and kept their position.
Diving, he made his appearance some distance off, this time with the
fish in his mouth, only to find himself, to his annoyance, again the
centre of wistful expectations. Presuming these gulls to be up-to-
date birds, their exulting cacklings might be literally rendered—“You
better begin, Mister Phoca; it’s no use trying, you know; you can’t
possibly dewett us!” At least, the seal seemed to think so, for he
there and then opened the banquet with a rip of his teeth that
distributed the offal amongst the hungry cordon.
The rocks become at this season of the year invested with a
slippery coating of algæ, which renders it extremely difficult to
maintain one’s footing, and also necessitates repeated applications
of hot lime to our gratings in order to render them passable. Myriads
of minute whelks, no larger than turnip seed, strew the rocks and
crunch under foot as we walk, while great patches of mussel spawn
delight the heart of the more venturesome of the white whelks—a
prospecting party who will doubtless communicate the promising
state of the commissariat to their fellows still in winter quarters.
Fishing in the Rock pools has been tried for the first time this
season, and resulted in the capture of a solitary “cobbler.” It may be
a month hence before we meet with any success.
This month has been extremely mild, though the hills behind
Arbroath are still seen to carry portions of their winter coat, while the
higher ranges inland are completely snow-capped. On a clear day
our view is limited by Tod Head, about twenty-five miles to the north,
and St Abb’s Head thirty miles south of us. The coast-line presents a
uniform flatness, which becomes monotonous in comparison with the
more picturesque raggedness of the West Coast. A most
conspicuous feature in the landscape in the vicinity of Arbroath is the
clump of trees on the summit of the Law Hill—a landmark well known
to navigators, and easily discernible, as it stands sharply defined
against the sky-line. Arbroath, when not enveloped in smoke, is
clearly seen, and with the aid of our telescope the after-church
promenaders can be distinguished on the Protection Wall, or
wending their way towards the Victoria Park.
APRIL 1902.
The extremely low tides prevalent at the opening of the month

enabled us to extend our hunting grounds somewhat further than
usual, and also to reach and demolish several “travellers” which the
heavy seas had hurled into the boat tracks, thus constituting a
serious danger at relief times. Quite a forest of luxuriant tangles now
cover the lower lying portion of the reef. Their dripping blades appear
on the surface, scintillating in the brilliant sunshine like so many
diamonds, till the receding tide permits the warm sun to rob them of
their freshness, their beauty vanishing in a perceptible vapour,
leaving them flaccid and inert till the returning tide restores their
pristine beauty. The badderlock or henware is here also in great
profusion, and usually selects a position the reverse of peaceful,
being generally found where the wash of the seas is most constant.
Of rapid growth, they attain a great length, some measuring fully
sixteen feet; one we had under observation was seen to increase a
foot in length in six weeks time. Owing to hazy weather we had a
number of compulsory visitors to dinner yesterday. Seated outside
our kitchen window was a party of fog-bound travellers, consisting of
a pigeon, a starling, a wagtail, a robin, and a couple of wheatears.
The starling was sitting bunched up by himself, preserving a stolid
indifference at his enforced detention, and appeared to treat the
animated expansion and flirting of the wheatears’ tails as undue
levity, unbecoming their sorrowful predicament. The beautiful black-
throated wagtail is all alertness, and the slightest movement on our
part sends him circling round the Rock till, unable to sight the land,
he is fain to regain his resting place. The pigeon has been here a
week now, and evidently has no intention of leaving. Should the
window be left open he makes bold enough to enter, although but
the other day he gave us a somewhat dramatic illustration of the
proverbial hen on the hot “griddle” by rehearsing a fandango on the
top of our cooking range, a position from which he had to be forcibly

Applied Mathematical Modeling and Analysis in Renewable Energy 1St Edition Manoj Sahni Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Copyright:

Available Formats

You might also like

Applied Mathematical Modeling and Analysis in Renewable Energy 1St Edition Manoj Sahni Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Applied Mathematical Modeling and Analysis in Renewable Energy 1St Edition Manoj Sahni Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Applied Mathematical Modeling and

Analysis in Renewable Energy 1st

Mathematical Modeling and Applied Calculus Joel Kilty

Renewable Hydrogen Renewable Energy and Renewable

Mathematical Analysis and Analytical Modeling 1st

Windfall Profit in Portfolio Diversification An

Applied Modeling Techniques and Data Analysis 1:

Advanced Mathematical Modeling with Technology Advances

Renewable Energy Systems : The Earthscan Expert Guide

Topics in Applied Mathematics and Modeling 1st Edition

Applied Soft Computing and Embedded System Applications

Differential Equations in Engineering

Sustainability in Industry 4.0

Applied Mathematical Modeling and Analysis in Renewable Energy

For more information about this series, please visit: https://www.routledge.

2 Park Square, Milton Park, Abingdon, Oxon, OX14 4RN

ISBN: 9780367746988 (hbk)

PART I Mathematical Modeling

Chapter 1 Mathematical Modeling of Transitional Fluid Phase Flows in

Chapter 2 AI-Based Detection and Localization of Gastrointestinal

Chapter 3 Mathematical Modeling of an EOQ for a Multi-item Inventory

Chapter 4 Fair Allocation of Items: A Comprehensive Study............................. 47

Chapter 5 Hierarchical Demand Response Controller........................................ 57

Chapter 6 A General Class of Polynomials Inspired by a General

Chapter 7 Squeezing Graphs.............................................................................. 101

Chapter 8 Finding the Surface Area and Volume of the Hyperspheres

PART III Mathematical Modeling in Renewable Energy

Chapter 9 Analysis of RSM Method for Optimization of Ultrasound-

Chapter 10 Energy Data Analysis of an Educational Institution

Chapter 11 Factors to Consider: A Review of Smart Grid

Chapter 12 Integration and Modeling of Small-Scale Pumped Storage.............. 181

Dr. Manoj Sahni

Ritu Sahni has been an assistant professor,

Pijus Kanti De Arun Kumar

Subhashini Nainar Vipul N. Rajput

Atmiya Patel Shantanu

1.1.1 Conceptual Model

1.2 PERMEABILITY MODEL

k n = k i × e −(3× cfn ×(σ −σ i )n ) (1.1)

1.2.2 Variable Coefficients

Bwn = e(.0000026 ×(14.7− Pn )) (1.10)

1.3 SINGLE-PHASE WATER FLOW

Dependent variablesa: Pnt + dt Independent variable: Xinv t + dt , Ltw+ dt , t

1.3.1 Mathematical Model

φiv ( P ) × cfiv ( P ) × Swiv ( P ) ∂P k iv ( P ) ∂2 P

Equation 1.14 is a non-linear partial differential equation applicable in case of water

1.3.2 Initial and Boundary Conditions

1.4 MULTIPHASE GAS AND WATER FLOW

Dependent variablesa: Pnt + dt , Swn t + dt Independent variable: Xinv t + dt , Ltw+ dt , t

1.4.1 Mathematical Model

µwiv ( P ) × φiv ( P )  ∂P ∂Sw 

It can be written as:

1.4.2 Initial and Boundary Conditions

1.4.3 Methodology to Solve for Multiphase Flow

µw iv × Bgiv  ct iv × φiv × Sg iv ∂P φiv ∂Sg VL × PL ∂P 

2. Step 2: Add equations 1.17 and 1.23.

µwiv × φiv  ∂P ∂Sw 

µwiv × Bgiv  ct iv × φiv × Sg iv ∂P φiv ∂Sg VL × PL ∂P 

PART I Mathematical Modeling

1.4.5 Water Saturation to be Solved Explicitly