Computational Models in

Engineering
Edited by Konstantin Volkov
Computational Models in
Engineering
Edited by Konstantin Volkov

Published in London, United Kingdom

Supporting open minds since 2005
Computational Models in Engineering
http://dx.doi.org/10.5772/intechopen.77484
Edited by Konstantin Volkov

Contributors
Faham Tahmasebinia, Chengguo Zhang, Ismet Canbulat, Samad M.E. Sepasgozar, Onur Vardar, Serkan
Saydam, Chen Chen, Jeongho Kim, Ganesh Anandakumar, Andrey Kurkin, Andrey Kozelkov, Valentin
Efremov, Sergey Dmitriev, Vadim Kurulin, Dmitry Utkin, Md Sakib Hasan, Mst Shamim Shawkat, Sherif
Amer, Nicole McFarlane, Syed Islam, Garrett Rose, Mohammed Albadi, Yuri Menshikov, Dimitris
Spiridakos, Nicholaos Markatos, Despoina Karadimou

© The Editor(s) and the Author(s) 2020

The rights of the editor(s) and the author(s) have been asserted in accordance with the Copyright,
Designs and Patents Act 1988. All rights to the book as a whole are reserved by INTECHOPEN LIMITED.
The book as a whole (compilation) cannot be reproduced, distributed or used for commercial or
non-commercial purposes without INTECHOPEN LIMITED’s written permission. Enquiries concerning
the use of the book should be directed to INTECHOPEN LIMITED rights and permissions department
(permissions@intechopen.com).
Violations are liable to prosecution under the governing Copyright Law.

Individual chapters of this publication are distributed under the terms of the Creative Commons
Attribution 3.0 Unported License which permits commercial use, distribution and reproduction of
the individual chapters, provided the original author(s) and source publication are appropriately
acknowledged. If so indicated, certain images may not be included under the Creative Commons
license. In such cases users will need to obtain permission from the license holder to reproduce
the material. More details and guidelines concerning content reuse and adaptation can be found at
http://www.intechopen.com/copyright-policy.html.

Notice
Statements and opinions expressed in the chapters are these of the individual contributors and not
necessarily those of the editors or publisher. No responsibility is accepted for the accuracy of
information contained in the published chapters. The publisher assumes no responsibility for any
damage or injury to persons or property arising out of the use of any materials, instructions, methods
or ideas contained in the book.

First published in London, United Kingdom, 2020 by IntechOpen

IntechOpen is the global imprint of INTECHOPEN LIMITED, registered in England and Wales,
registration number: 11086078, 7th floor, 10 Lower Thames Street, London,
EC3R 6AF, United Kingdom
Printed in Croatia

British Library Cataloguing-in-Publication Data

A catalogue record for this book is available from the British Library

Additional hard and PDF copies can be obtained from orders@intechopen.com

Computational Models in Engineering

Edited by Konstantin Volkov
p. cm.
Print ISBN 978-1-78923-869-3
Online ISBN 978-1-78923-870-9
eBook (PDF) ISBN 978-1-78985-400-8
 We are IntechOpen,
the world’s leading publisher of
Open Access books
Built by scientists, for scientists

4,700+ 120,000+ 135M+

Open access books available International authors and editors Downloads

Our authors are among the

151
Countries delivered to
Top 1%
most cited scientists
12.2%
Contributors from top 500 universities

Selection of our books indexed in the Book Citation Index

in Web of Science™ Core Collection (BKCI)

Interested in publishing with us?

Contact book.department@intechopen.com
Numbers displayed above are based on latest data collected.
For more information visit www.intechopen.com
 Meet the editor

Dr. Volkov is a senior lecturer in thermofluids at the Kingston

University (London, UK). He holds a PhD in fluid mechanics.
After completion of his PhD, Dr Volkov worked at the Bal-
tic State Technical University (Russia), University of Central
Lancashire (UK), and University of Surrey (UK). His areas of
expertise cover multidisciplinary areas: from design and optimi-
zation of energy systems to fundamental problems focused on
modelling and simulation of turbulent multi-phase flows. He is a Chartered Engi-
neer and member of the Institute of Physics, Institution of Mechanical Engineers,
Combustion Institute, and Higher Education Academy. He is the author of more
than 120 scientific papers and a member of editorial boards and scientific commit-
tees of a number of journals and conferences.
Contents

Preface XIII

Chapter 1 1
Technology of 3D Simulation of High-Speed Damping Processes
in the Hydraulic Brake Device
by Valentin Efremov, Andrey Kozelkov, Sergey Dmitriev,
Andrey Kurkin, Vadim Kurulin and Dmitry Utkin

Chapter 2 17
Three-Dimensional Finite Element Analysis for Nonhomogeneous
Materials Using Parallel Explicit Algorithm
by Ganesh Anandakumar and Jeongho Kim

Chapter 3 35
A New Concept to Numerically Evaluate the Performance of
Yielding Support under ImpulsiveLoading
by Faham Tahmasebinia, Chengguo Zhang, Ismet Canbulat,
Samad M.E. Sepasgozar, Onur Vardar, Serkan Saydam
and Chen Chen

Chapter 4 47
Criteria for Adequacy Estimation of Mathematical Descriptions
of Physical Processes
by Yuri Menshikov

Chapter 5 67
Power Flow Analysis
by Mohammed Albadi

Chapter 6 89
Modeling Emerging Semiconductor Devices for Circuit Simulation
by Md Sakib Hasan, Mst Shamim Ara Shawkat, Sherif Amer,
Syed Kamrul Islam, Nicole McFarlane and Garrett S. Rose

Chapter 7 115
Mathematical Modeling of Aerodynamic Heating and Pressure
Distribution on a 5-Inch Hemispherical Concave Nose in
Supersonic Flow
by Dimitrios P. Spiridakos, Nicholas C. Markatos and
Despoina Karadimou
 Preface

In many fields, modelling and simulation are integral and therefore essential to
business and research. Modelling and simulation provide the capability to enter
fields that are either inaccessible to traditional experimentation or where carrying
out traditional empirical inquiries is prohibitively expensive. The role of modelling
and simulation in engineering and various applications such as computational fluid
dynamics, finite element analysis, and others has been underestimated for a long
time. However, with a growing complexity of application scenarios as well as
numerical algorithms and hardware architectures, the need for sophisticated
methods from numerical analysis and computer science has become more impor-
tant. The growth in computing power has revolutionized the use of realistic math-
ematical models in science and engineering, and numerical analysis is required to
implement these detailed models of the world.

A computer simulation is the execution of a model, represented by a computer

program that gives information about the system being investigated. The simulation
approach of analyzing a model is opposed to the analytical approach, where the
method of analyzing the system is purely theoretical. As this approach is more
reliable, the simulation approach gives more flexibility and convenience. Modelling
and simulation offer people the chance to develop an understanding of their prob-
lem domain by building a simulation of the problem space in which they are
interested.

The modelling and simulation identify more key stages in a successful simulation
cycle: implementation, exploration and visualization, validation, and embedding.

The book provides and discusses different examples from engineering, as well as
where and how numerical methods contribute to more efficient simulation envi-
ronments. There are 7 chapters in the book, covering different aspects of modelling
and simulation in engineering and technology.

In Chapter 1, a three-dimensional simulation technology for physical processes in

concentric hydraulic brakes with a throttling-groove partly filled hydraulic cylinder
is considered. The technology is based on the numerical solution of a system of
Navier-Stokes equations. Free surface tracking is provided by the volume of fluid
method. The results of hydraulic brake simulations in the counter-recoil regime are
reported and compared to experimental data. The performance of the hydraulic
brake is studied as a function of the fluid mass and firing elevation of the gun.

Chapter 2 addresses the behavior of functionally graded solids under dynamic

impact loading within the framework of linear elasticity using the parallel explicit
algorithm. Numerical examples are presented that verify the dynamic explicit finite
element code and demonstrate the dynamic response of graded materials. A three-
point bending beam made of epoxy and glass phases under low velocity impact is
studied. Finite element modeling and simulation discussed herein can be a critical
tool in helping to understand the physics behind the dynamic events.
Chapter 3 discusses the case studies that revealed premature failures of stiffer
elements prior to utilising the full capacity of more deformable elements within the
same system. From a design perspective, it is important to understand that the
dynamic-load capacity of a ground support system depends not only on the capacity
of its reinforcement elements but also, and perhaps most importantly, on their
compatibility with other elements of the system and on the strength of the connec-
tions. The failure of one component of the support system usually leads to the
failure of the system.

Chapter 4 estimates criteria for mathematical descriptions in the form of ordinary

differential equations. Adequate mathematical descriptions can increase the objec-
tivity of the results of mathematical modeling for future use. These descriptions
make it possible and reasonable to use the results of mathematical modeling to
optimize and predict the behavior of physical processes. Interrelations between
criteria are considered. The proposed criteria are easily transferred to mathematical
descriptions in algebraic form.

In Chapter 5, the power flow model of a power system is built using the relevant
network, load, and generation data. Outputs of the power flow model include
voltages at different buses, line flows in the network, as well as system losses. These
outputs are obtained by solving nodal power balance equations. Since these equa-
tions are nonlinear, iterative techniques are commonly used to solve this problem.
The chapter will provide an overview of different techniques used to solve the
power flow problem.

Chapter 6 shows how different approaches can be adopted to model three emerging
semiconductor devices namely, silicon-on-insulator four gate transistor, single
photon avalanche diode, and insulator-metal transistor device.

In Chapter 7, the results of numerical simulation of heat transfer on a 5-inch

hemispherical concave nose at a Mach number of 2 are reported and compared with
the available experimental data. Different turbulence models and different
discretization schemes are also examined.

The book promotes open discussion between research institutions, academia, and
industry from around the globe on research and development of enabling technol-
ogies. The book covers many aspects of theory and practice, which deliver essential
contributions and provide their input and support to the cooperative efforts.

Dr. Konstantin Volkov

MEng. MSc. PhD. DSc. CEng. MIMechE. MinstP. FHEA.,
Department of Mechanical and Automotive Engineering,
School of Engineering and the Environment,
Faculty of Science, Engineering and Computing,
Kingston University,
London, United Kingdom

XIV
Chapter 1

Technology of 3D Simulation of
High-Speed Damping Processes in
the Hydraulic Brake Device
Valentin Efremov, Andrey Kozelkov, Sergey Dmitriev,
Andrey Kurkin, Vadim Kurulin and Dmitry Utkin

Abstract

This chapter describes a three-dimensional simulation technology for physical

processes in concentric hydraulic brakes with a throttling-groove partly filled
hydraulic cylinder. The technology is based on the numerical solution of a system of
Navier–Stokes equations. Free surface tracking is provided by the volume of fluid
(VOF) method. Recoiling parts are simulated by means of moving transformable
grids. Numerical solution of the equations is based on the finite-volume
discretization on an unstructured grid. Our technology enables simulations of the
whole working cycle of the hydraulic brake. Results of hydraulic brake simulations
in the counter-recoil regime are reported. The results of the simulations are com-
pared with experimental data obtained on JSC “KBP” test benches. The calculated
and the experimental sets of data are compared based on the piston velocity as a
function of distance. The performance of the hydraulic brake is studied as a func-
tion of the fluid mass and firing elevation of the gun.

Keywords: hydraulic brake, Navier-stokes equations, multi-phase flow,

volume of fluid, moving rigid body, turbulent flow

1. Introduction

This chapter describes our three-dimensional simulation technology for phys-

ical processes in hydraulic brakes with a throttling-groove partly filled hydraulic
cylinder. The technology is based on the numerical solution of a system of Navier-
Stokes equations with an additional transport equation to track the working fluid/
free volume interface by the volume of fluid (VOF) method [1, 2]. As a solver, we
use an iterative algorithm, PISO [1, 3], combined with a SLAE solver based on the
algebraic multigrid method [4]. To improve the accuracy of solution, we use
interface capturing schemes, HRIC [1, 4], near the interface between the phases.
Moving parts are simulated by means of moving deforming meshes [5], the
motion of which is represented in the initial equations with the Lagrange-Eulerian
approximation [6]. Our technology enables modeling of the whole working cycle
of the hydraulic brake.
To demonstrate the performance of our technology, we are considering simu-
lation of a hydraulic brake with a throttling-groove partly filled hydraulic cylin-
der. The simulation outputs are compared with experimental data obtained on the

1
Computational Models in Engineering

test benches at JSC “KBP.” The calculated and the experimental sets of data are
compared based on the piston velocity as a function of displacement. The perfor-
mance of the hydraulic brake is studied as a function of the fluid mass and
hydraulic brake angle.

2. Problem definition

The hydraulic brake serves to absorb the energy of moving parts. Its structure
consists of a case with a grooved inner surface, a ring piston mounted directly on
the outer surface of the barrel and having holes matching the guide rods with
stranded counter-recoil springs uniformly distributed around the barrel in the
hydraulic brake chamber, and front and rear packings with circular-cross-section
rubber O-rings to seal the hydraulic brake chamber filled with hydraulic fluid
(Figure 1).
The hydraulic brake starts to react as soon as there is no more free space in the
recoil chamber (space between the piston and the rear packing); the fluid is sprin-
kled through the gaps from the recoil chamber to the counter-recoil chamber (space
between the piston and the front packing) at a growing rate and stops the moving
parts.
Recoil of the moving parts is countered by the energy stored by the counter-
recoil springs while braking. In the counter-recoil phase, the fluid from the counter-
recoil chamber is sprinkled through the gaps into the recoil chamber providing the

Figure 1.
(a) General layout of the hydraulic brake: (a) general layout (1—hydraulic fluid cylinder with guides and
springs, 2—case, 3—piston, 4—packings); (b) general layout without case (5—guides); and (c) piston;
(b) schematic of operation.

2
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

required counter-recoil braking law. Both for recoil and for counter recoil, the areas
of fluid flow in the hydraulic brake are generated by the gaps between the piston
and the grooved case surface defining the law of motion, and between the piston
holes and the guide rods.
Numerical simulation of the hydraulic brake includes the following physical
processes: turbulent flow of fluid encountering sudden contractions, expansions,
and axisymmetric gaps; multi-phase flow with different equations of state (it is
reasonable to use constant-density approximation for fluid and perfect gas law for
air); and flow with moving solids.
The scope of this work included the development of a mathematical model and
numerical method to simulate these physical processes.

3. Mathematical model and numerical scheme

Simulation of hydraulic brake operation involves multi-phase flow simulation

with two phases. Each phase can have its individual equation of state. Let us assume
that the flow is isothermal within one operation cycle and that the phases have the
same field of velocity. These assumptions allow us to construct our numerical model
using the VOF method [1, 7]. Considering the assumptions above, let us describe
the flow of matter by a system of equations, comprised of the mass and momentum
equations and the volume fraction transfer equation:
8 ∂ρ ∂
>
> þ ðρui Þ ¼ 0,
>
> ∂t ∂x
>
> i
< ∂ρu ∂ � � ∂p ∂
i
þ ρui uj ¼ � þ τij þ ρg i , (1)
>
> ∂t ∂x i ∂x i ∂x i
>
> ∂ρ αξ � �
>
: ξ þ ∂ ui ρξ αξ ¼ 0,
>
∂t ∂xi
where t is the time, ui = {u1, u2, u3} = {u, v, w} is the velocity, xi is the spatial
vector component, τij is the viscous stress tensor, gi is the gravity acceleration
vector, ξ is the phase index, αξ is the volume fraction of the ξ-th phase, and ρ is the
resulting density defined as an average over all the phases:

N
ρ ¼ ∑ ρξ αξ (2)
ξ¼1

where N is the number of phases.

The volume fraction transfer equation is written in terms of the transfer of the
quantity αξ:
� �
∂αξ ∂ � � αξ ∂ρξ ∂ρξ
þ ui αξ ¼ � þ ui (3)
∂t ∂xi ρξ ∂t ∂xi

Summing Eq. (3) over all the phases and considering the equality ∑ξ αξ ¼ 1, we
obtain the following mass equation with respect to the velocity divergence:
� �
∂ui αξ ∂ρξ ∂ρξ αξ dρξ
¼ �∑ þ ui ¼ �∑ (4)
∂xi ξ ρξ ∂t ∂xi ξ ρξ dt

The momentum equation is written in a half-divergent form. This stabilizes the

iterative procedure of numerical solution of this equation [3, 4]:

3
Computational Models in Engineering

∂ui ∂   ∂   ∂p ∂τij
ρ þ ρui uj � ui ρuj ¼ � þ þ ρg i (5)
∂t ∂xj ∂xj ∂xi ∂xj

An important feature of this problem is that we need to consider the motion of

moving parts, which travel to a distance exceeding the half length of the hydraulic
brake during computation. To incorporate their motion, we decided to use a
topology-preserving mesh deformation method. To prevent unnecessary cell defor-
mations, the mesh nodes located at fixed elements (case and guide rods) follow the
moving piston preserving their geometry. The mesh deformation was implemented
by the IDW method [5].
To allow for the mesh motion in Eq. (1), we have rewritten the equations
of volume fraction transfer and momentum of the phases in a moving frame of
coordinate in accordance with the known law [6]:

d∗ φ ∂φ ∂φ
¼ þ vi (6)
dt ∂t ∂xi
∗
where ddtφ is the substantial derivative φ with respect to the moving frame of
reference and vi is the mesh displacement velocity vector. Using Eq. (3), the volume
fraction transfer equation can be written as:
 ∗ 
d∗ αξ ∂αξ ∂ui αξ d ρξ ∂ρξ
þ ðui � vi Þ þ αξ ¼� þ ðui � vi Þ (7)
dt ∂xi ∂xi ρξ dt ∂xi
d∗ α
Here dt ξ denotes the substantial derivative on the moving mesh. The momen-
tum equation is also defined with respect to the moving frame of reference consid-
ering Eq. (3):

d∗ ui ∂    ∂    ∂p ∂τij
ρ þ ρui uj � vj � ui ρ uj � vj ¼ � þ þ ρg i (8)
dt ∂xj ∂xj ∂xi ∂xj

The mass equation is defined with respect to the velocity in the moving frame of
reference:
 ∗ 
∂ðui � vi Þ αξ d ρξ ∂ρξ
¼ �∑ þ ðui � vi Þ (9)
∂xi ξ ρξ dt ∂xi

This form of the equations is easy to implement within the framework of finite-
volume discretization [8–11]. The chosen way of taking into account the mesh
motion is optimal, because it does not require topology reconstruction, though
conservative with respect to the major quantities.
The experimental data on the hydraulic brake operation make it possible to
estimate the Reynolds number, which equals Re = 105–106 in the cylindrical gap
during recoil of the moving parts. The most reasonable way to include turbulent flow
components with such a Reynolds number is to use the RANS approach based on the
solution of the Reynolds-averaged system Eq. (2) [8, 9]. The averaged system of
equations is closed using the SST turbulence model, which has proven its efficiency in
real-life problem simulations [11–13]. In the SST model, the (k-ε) model is formulated
in terms of (k-ω) and is focused on resolving small-scale turbulence in the outer flow
region. A (k-ω) model intended to describe large-scale turbulence is used in the
boundary layer. The combination of these models together is carried out with the help
of a function that ensures the proximity of the total model to the (k-ε) model far from
the solid walls and to the (k-ω) model in the near-wall flow area.

4
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

The resulting system of equations is solved by numerical integration on the

finite-volume mesh. The relationship of pressure and velocity while solving Eq. (1)
is determined by the SIMPLE algorithm [2]. The SIMPLE algorithm allows to find
the pressure field for closure of the continuity equation. Let us write the equation of
conservation of momentum at time discretization according to the Euler scheme:

∂ � nþ1 n � ∂ � nþ1 �
j
uinþ1 � ui ∂pnþ1
ρ þρ ui uj ¼ � þ τ (10)
Δt ∂xj ∂xi ∂xj ij

where n is the solution from the previous iteration, and j is the solution from
the previous time step. To solve this equation, the pressure and velocity are
supposed to have the form:
(
unþ1
i ¼ uni þ u∗i ,
(11)
pnþ1 ¼ pn þ αp ð pnþ1 � pn Þ ¼ pn þ αp δpnþ1 :

Here 0 ≤ αp ≤ 1 is the parameter of relaxation.

Substitution of the first expression in Eq. (11) into Eq. (10) yields a preliminary
estimate of the velocity value in the next step from the equation:

∂ � ∗ n� ∂ � ∗�
j
u∗i u ∂pn
ρ þρ ui uj � τij ¼ ρ i � (12)
Δt ∂xj ∂xj Δt ∂xi

The molecular and turbulent components of the tensor of tangential stress in

Eq. (12) are also calculated using u∗i . In the second stage, the full speed value at the
(n + 1)th iteration is calculated by adjustment using the correction to pressure:

∂ðδpnþ1 Þ
unþ1
i ¼ u∗i � Δt (13)
∂xi
The pressure correction is found from Eq. (13) using the continuity condition
for unþ1
i . Taking the derivative of both sides of Eq. (13), we obtain the Poisson
equation for pressure:
� �
∂ ∂ðδpnþ1 Þ 1 ∂u∗i
¼ (14)
∂xi ∂xi Δt ∂xi

This iterative procedure allows to obtain fields of velocity and pressure satisfy-
ing the system of Eq. (1).
Discretization of the equations is performed by a finite-volume technology. Let
us show it on the example of the equation of transfer of a scalar quantity φ:

∂ρφ ∂ � � ∂ ∂φ
þ ρφuj ¼ τj þ Q , τj ¼ μ (15)
∂t ∂xj ∂xj ∂xj

The time discretization of Eq. (15) can be carried out using one of the known
schemes, for example, the Euler scheme, which is used in the work to solve non-
stationary problems using the RANS approach:
� �jþ1
ρ jþ1 φ jþ1 � ρ j φ j ∂ � � ∂
þ ρφuj � τj � Q ¼0 (16)
Δt ∂xj ∂xj

Here j is the number of the time step.

5
Computational Models in Engineering

Let us integrate Eq. (16) over volume and proceed to integration over the area
for the convective and diffusion terms:
ð jþ1 jþ1 þ þ ð
ρ φ � ρ jφ j ∂φ
dV þ ρφuj dSj � μ dSj � QdV ¼ 0 (17)
Δt ∂xj
Vp Sp Sp Vp

For approximation on a finite-volume grid, the convective term can be written

in the form:
þ
ρφuj dSj ≈ ∑ ρk φk uj, k Sj, k ≈ ∑ ρk φk F k (18)
k k
Sp

where Fk is the volume flow through the face k. The value of φk on the face k is
determined by the applied convective scheme.
The discrete analog of the diffusion term can be written in the following form:
þ � � � �
∂φ ∂φ ∂φ
μ dSj ≈ ∑ μ Sj, k ¼ ∑ μk j Sk j (19)
∂xj k ∂xj k k ∂nk k
Sp

where nk is the
� normal
� to face k.
∂φ
The value of ∂xi k on the face k is found by linear interpolation on the values of
the gradient in the cells, which are determined by one of the known methods, for
example, the Gauss method:
� �
∂φ 1
¼ ∑ φ Sk, i (20)
∂xi P V P k k
Using this discretization, Eq. (15) is replaced by a system of linear algebraic
equations written for each calculation cell:

AP φP þ ∑ Akint φMk ¼ Ri, P (21)

int
kint

Time discretization of the equations is performed by a three-layer second-order

scheme [10]. Discretization of the convective terms in the equation of motion,
equation of transfer of turbulent parameters, is performed by the upwind scheme
LUD [10], and in the volume fraction transfer equation, by the HRIC scheme [1],
which prevents excessive numerical diffusion of the phase interface. The force of
gravity was included using a bulk force fitting algorithm [14].
These methods and models have been implemented in the software package
LOGOS. LOGOS is a 3D multi-physics code for convective heat and mass transfer,
aerodynamic and hydrodynamic simulations on parallel computers [11, 13–18].
LOGOS has been successfully verified and demonstrated with sufficiently high
efficiency on a number of various hydrodynamic tests, including propagation of
gravity waves on a free surface (tsunami) [2, 14, 18] and industry-specific simula-
tions [11, 15]. Speedup of computations on highly parallel computers is provided by
an original implementation of the algebraic multigrid method [11, 19].

4. Numerical simulation of hydraulic brake operation

Let us consider the counter-recoil phase in the operation of the hydraulic brake,
when the moving parts together with the piston (see Figure 1b) are driven by the

6
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

springs from their rightmost position to the leftmost one resisting the hydrody-
namic force acting on the fluid side.
To solve the problem numerically, an unstructured hexahedral-dominant mesh
was constructed using the pre- and post-processing tools of the LOGOS. The mesh is
refined near the piston and in the gaps between the piston and the case, where the
flow of fluid is particularly strong. Figure 2 shows mesh fragments.
When constructing the mesh, we placed the piston in the middle between its
rightmost initial and leftmost end positions of counter recoil. This minimizes mesh
deformation during computation. The problem was simulated in the unsteady
setup; the time step was 5 � 10�4 t0, where t0 is the total time of piston counter-
recoil motion. The forces acting on the piston, its velocity and displacement were
calculated at the end of each time step after finding the solution of the Navier–
Stokes equation.
During computation, the piston together with the moving parts is accelerated by
the springs and reaches its maximum velocity during the second half of its travel.
The piston then enters the grooved region of the case with a decreasing gap, as a
result of which the hydrodynamic resistance grows and slows down the moving
parts. The computation ends when the piston is at dm = 0.01 m from its end
position, which is attributed to the maximum admissible mesh deformation in the
counter-recoil chamber.
When the piston is moving, in addition to the force exerted by the springs and
hydrodynamic resistance, the moving parts are exposed to friction arising in the
packings, between the guides and the piston, and inside the stranded springs during
their compression or tension. The resultant of the forces exerted by the springs and
friction can be expanded into two components. The first does not depend on the
velocity of the moving parts and can be easily measured experimentally by pulling
the moving parts at a slow rate. The second depends on the velocity of the moving
parts and is mostly related to friction in the turns of the stranded springs. Empirical
estimates show that the second component of the resultant force can be expressed
empirically as

F din
mp ¼ ηuM (22)

where u is the velocity, M is the mass of the moving parts, η is an empirical

coefficient, which can vary in the range of η = 0–5 s�1 for the springs we use.
Computations were conducted with η = 0, 3, 4, and 5. Figure 3 shows the field of
volume fraction of the fluid at different times for η = 0 s�1.
At t = 0.2 t0, where t0 is the total time of counter recoil, a wave of fluid rises
upstream of the piston. Pressure in the counter-recoil chamber grows. Air is actively
bled over through the upper part of the gap. A flow of fluid emerges in the bottom
part of the gap. Its rate is 10–15 times slower than the rate of the air flow (Figure 4).

Figure 2.
Mesh fragments.

7
Computational Models in Engineering

Figure 3.
Field of volume fraction of the fluid at different times.

Figure 4.
Field of velocity amplitude (t = 0.2 t0).

As the piston continues to move, the fluid is sprinkled through the gaps at a
growing rate. At t = 0.95 t0, most of the fluid is already pressed out of the counter-
recoil chamber; it forms a nearly homogeneous gas–liquid mixture in the opposite
chamber. Pressure in the counter-recoil chamber reaches P = 150 P0, where
P0 = 1 atm is the initial pressure.
Fluid velocity in the gap grows as high as V = 50–70 m/s (Figure 5). After the
fluid has flown over through the gap, the bulk gas content in the mixture increases
steeply as a result of a pressure drop. This leads to a further growth of mixture
velocity to V = 90 m/s (Figure 5). This effect is less pronounced in the bottom part
of the gap because of the lower gas content in the mixture sprinkled through the
gap.
Figure 6 shows a plot of piston velocity as a function of displacement for
different values of η.
On the whole, our results match the experimental data both qualitatively and
quantitatively. All the simulated cases consistently predict the state of piston accel-
eration. In the high-velocity phase of piston motion, the parameter η starts to play a
significant role. The best agreement with experiment is provided by η = 3 s1. This
does not mean that the value above is close to the actual one, but is indicative of the
fact that our numerical model, input data, and mesh are capable of describing the
process of piston motion with required accuracy, if we use this value of η in our
computational model.

8
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

Figure 5.
Field of velocity amplitude (t = 0.95 t0).

Figure 6.
Piston velocity as a function of displacement.

The counter-recoil phase occurs at zero angle (the hydraulic brake is placed
horizontally). Let us consider the dependence of piston motion on the accuracy of
horizontal positioning of the hydraulic brake. For this purpose, let us carry out
calculations with angles α = 1° and α = 1°. The calculation results are shown in
Figure 7.
The plots demonstrate that the piston velocity grows with increasing angle. This
is quite predictable, because the amount of fluid in the counter-recoil chamber
decreases as the angle increases. One should also note that the gravity force acting
on the moving parts in the direction opposite to that of their motion at α > 0 or in
the direction of their motion at α < 0 has no measurable effect on the result.
The angle of the hydraulic brake also controls the static pressure in the counter-
recoil chamber (Figure 8). The first pressure maximum is attributed to a decrease
in the volume fraction of air in the counter-recoil chamber as the piston moves, and
the subsequent drop, to a growth in the gap size between the piston and the grooved
case region. As the angle increases, the initial amount of fluid in the counter-recoil
chamber decreases, and the maximum shifts toward the range of greater piston

9
Computational Models in Engineering

Figure 7.
Piston velocity as a function of piston displacement for different angles.

Figure 8.
Static pressure in the counter-recoil chamber as a function of piston displacement.

displacements, which allows the piston to speed up to a higher velocity. This

increase in the piston velocity leads to a higher pressure maximum in the region of
strong slowdown (the second pressure maximum in Figure 6). Thus, an increase in
the amount of fluid in the counter-recoil chamber leads to a higher and earlier first
pressure maximum and a lower second pressure maximum, which is quite consis-
tent with process physics.
The results of piston motion analysis as a function of initial fluid level in the
hydraulic brake (h = h0, h = h0 + 0.01 m, h = h0–0.01 m) are shown in Figure 9.
The fluid level has a considerable effect on the piston velocity. A 1-cm increase
in the level results in a 10% drop in the average piston velocity, which is explained

10
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

by a decrease in the force of hydrodynamic resistance acting on the moving parts.

Note that the slopes of the velocity plot are indicative of the same trend: as the fluid
level in the counter-recoil chamber rises, the first pressure maximum increases and
the second one decreases.
An important characteristic of the counter-recoil piston motion is piston velocity
Vextr at 159 mm to the counter-recoil end (ΔZ = 153 mm), where the shell extractor
is located, the performance of which depends on the velocity of the moving parts.
Another important characteristic is piston velocity close to the end position Vend,
which will govern the force exerted by the piston on the packing in the counter-
recoil chamber. The values of these characteristics are presented in Table 1.
The table demonstrates that a change in the fluid level within 1 cm leads to a 15%
change in the velocity Vextr. A change in the angle to 1 degree results in a 5% change
in the velocity. Note that the level and angle variations have a minor effect on the
velocity Vend, the oscillations of which do not exceed 2–3%.
The results demonstrate that the governing quantity, which controls the perfor-
mance of the hydraulic brake, is the initial fluid level in the counter-recoil chamber.
An increase in the fluid level results in a lower average piston velocity, smaller Vextr
and Vend, shift and rise in the first pressure maximum, and drop in the second
pressure maximum. The process pattern, however, does not depend on the reason

Figure 9.
Piston velocity as a function of displacement for different fluid levels.

No. α [degree] h – h0 [sm] Vextr [m/s] Vend [m/s]

1 0.0 1.0 3.64 1.74

2 0.0 0.0 3.21 1.69

3 0.0 1.0 2.78 1.65

4 1.0 0.0 3.05 1.66

5 1.0 0.0 3.38 1.72

Table 1.
Piston motion characteristics.

11
Computational Models in Engineering

of the rise in the fluid level in the counter-recoil chamber: both general increase in
the initial amount of fluid and negative angle of the hydraulic brake result in the
same trends.

5. Conclusion

In this chapter, we have presented a three-dimensional numerical modeling

technology for physical processes occurring in hydraulic brake devices. The tech-
nology is based on the numerical solution of a system of Reynolds-averaged Navier–
Stokes equations. To track the free surface, we use the volume of fluid (VOF)
method. Moving parts are simulated by means of moving deforming meshes.
Numerical solution of the equations is based on the finite-volume discretization,
which is used to model the counter-recoil phase in hydraulic brake operation. This
technology enables simulations on three-dimensional unstructured meshes. The
technology is implemented based on the program package LOGOS, which provides
high parallel efficiency of simulations.
The technology has been used to model the counter-recoil phase in hydraulic
brake operation. Results of studying the effect of the parameter related to friction,
hydraulic brake angle, and initial fluid level are reported. Our numerical experi-
ments have demonstrated that the initial fluid level in the counter-recoil chamber is
the governing quantity in hydraulic brake operation. An increase in the fluid level in
the counter-recoil chamber as a result of pouring more fluid or placing the hydraulic
brake at a negative angle results in the same trend in the change in pressure and its
maximums.

Acknowledgements

This research has been funded by grants of the President of the Russian Federa-
tion for state support of research projects by young doctors of science (MD-
4874.2018.9) and state support of the leading scientific schools of the Russian
Federation (NSh-2685.2018.5), and supported financially by the Russian Founda-
tion for Basic Research (project No. 16-01-00267).

Conflict of interest

The authors declare that they have no conflict of interest.

12
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

Author details

Valentin Efremov1, Andrey Kozelkov2, Sergey Dmitriev3, Andrey Kurkin3*,

Vadim Kurulin2 and Dmitry Utkin2

1 Joint-Stock Company “Instrument Design Bureau named after Academician A.G.

Shipunov”, Tula, Russia

2 Russian Federal Nuclear Center—All-Russian Research Institute of Experimental

Physics, Sarov, Russia

3 Nizhny Novgorod State Technical University n.a. R.E. Alekseev,

Nizhny Novgorod, Russia

*Address all correspondence to: aakurkin@gmail.com

© 2018 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

13
Computational Models in Engineering

References

[1] Ubbink O. Numerical prediction of Berlin, Heidelberg, New York: Springer-

two fluid systems with sharp interfaces Verlag; 1991. p. 401
[thesis]. London: Imperial College of
Science, University of London; 1997 [9] Fletcher CAJ. Computational
techniques for fluid dynamics. In:
[2] Kozelkov AS, Kurkin AA, Pelinovsky Specific Techniques for Different Flow
EN, Tyatyushkina ES, Kurulin VV, Categories. Vol. 2. Berlin, Heidelberg,
Tarasova NV. Landslide-type tsunami New York: Springer-Verlag; 1998. p. 496
modelling based on the Navier-stokes
equations. Science of Tsunami Hazards. [10] Jasak H. Error analysis and
2016;35:106-144 estimation for the finite volume method
with applications to fluid flows [thesis].
[3] Yatsevich SV, Kurulin VV, Rubtsova London: Imperial College of Science,
DP. On the Application of the PISO University of London; 1996
Algorithm to Molecular-Immiscible
Fluid Dynamics Problems. VANT, Ser. [11] Kozelkov AS, Kurulin VV, Lashkin
Matematicheskoe Modelirovanie SV, Shagaliev RM, Yalozo AV.
Fizicheskikh Protsessov. Vol. 1. 2015. Investigation of supercomputer
pp. 16-29 capabilities for the scalable numerical
simulation of computational fluid
[4] Khrabry AI, Smirnov EM, Zaytsev dynamics problems in industrial
DK. Solving the convective transport applications. Computational
equation with several high-resolution Mathematics and Mathematical Physics.
finite volume schemes. Test 2016;56:1506-1516. DOI: 10.1134/
computations. In: Proceedings of the 6th S0965542516080091
International Conference on
Computational Fluid Dynamics [12] Menter FR, Kuntz M, Langtry R.
(ICCFD-6); 12–16 July 2010; Russia. Ten years of experience with the SST
Berlin, Heidelberg: Springer-Verlag; turbulent model. In: Hanjalic K, Nagano
2011. pp. 535-540 Y, Tummers M, editors. Turbulence,
Heat and Mass Transfer 4. Danbury:
[5] Luke E, Collins E, Blades E. A fast Begell House Inc; 2003. pp. 625-632
mesh deformation method using explicit
interpolation. Journal of Computational [13] Kozelkov A, Kurulin V, Emelyanov
Physics. 2012;231:586-601. DOI: V, Tyatyushkina E, Volkov K.
10.1016/j.jcp.2011.09.021 Comparison of convective flux
discretization schemes in detached-eddy
[6] Loitsyansky LG. Liquid and Gas simulation of turbulent flows on
Mechanics. Vol. 840. Moscow: Nauka; unstructured meshes. Journal of
1987 Scientific Computing. 2016;67:176-191.
DOI: 10.1007/s10915-015-0075-7
[7] Rusche H. Computational fluid
dynamics of dispersed two-phase flows [14] Efremov VR, Kozelkov AS, Kornev
at high phase fractions [thesis]. London: AV, Kurkin AA, Kurulin VV, Strelets
Imperial College of Science, University DY, et al. Method for taking into
of London; 2002 account gravity in free-surface flow
simulation. Computational Mathematics
[8] CAJ F. Computational techniques for and Mathematical Physics. 2017;57:
fluid dynamics. In: Fundamental and 1720-1733. DOI: 10.1134/
General Techniques. 2nd ed. Vol. 1. S0965542517100086

14
Technology of 3D Simulation of High-Speed Damping Processes in the Hydraulic Brake Device
DOI: http://dx.doi.org/10.5772/intechopen.83358

[15] Deryugin YN, Zhuchkov RN,

Zelenskiy DK, Kozelkov AS, Sarazov
AV, Kudimov NF, et al. Validation
results for the LOGOS multifunction
software package in solving problems of
aerodynamics and gas dynamics for the
lift-off and injection of launch vehicles.
Mathematical Models and Computer
Simulations. 2015;7:144-153. DOI:
10.1134/S2070048215020052

[16] Kozelkov AS, Kurkin AA,

Legchanov MA, Kurulin VV,
Tyatyushkina ES, Tsibereva YA.
Investigation of the application of RANS
turbulence models to the calculation of
nonisotermal low-Prandtl-number
flows. Fluid Dynamics. 2015;50:501-513.
DOI: 10.1134/S0015462815040055

[17] Kozelkov AS, Kurkin AA, Kurulin

VV, Lashkin SV, Tarasova NV,
Tyatyushkina ES. Numerical modeling
of the free rise of an air bubble. Fluid
Dynamics. 2016;51:709-721. DOI:
10.1134/S0015462816060016

[18] Kozelkov AS, Kurkin AA,

Pelinovsky EN, Kurulin VV,
Tyatyushkina ES. Numerical modeling
of the 2013 meteorite entry in Lake
Chebarkul, Russia. Natural Hazards and
Earth System Sciences. 2017;17:671-683.
DOI: 10.5194/nhess-17-671-2017

[19] Volkov KN, Kozelkov AS, Lashkin

SV, Tarasova NV, Yalozo AV. A parallel
implementation of the algebraic
multigrid method for solving problems
in dynamics of viscous incompressible
fluid. Computational Mathematics and
Mathematical Physics. 2017;57:
2030-2046. DOI: 10.1134/
S0965542517120119

15
Chapter 2

Three-Dimensional Finite Element

Analysis for Nonhomogeneous
Materials Using Parallel Explicit
Algorithm
Ganesh Anandakumar and Jeongho Kim

Abstract

This chapter addresses the behavior of functionally graded solids under dynamic
impact loading within the framework of linear elasticity using parallel explicit
algorithm. Numerical examples are presented that verify the dynamic explicit finite
element code and demonstrate the dynamic response of graded materials. A three-
point bending beam made of epoxy and glass phases under low-velocity impact is
studied. Bending stress history for beam with higher values of material properties at
the loading edge is consistently higher than that of the homogeneous beam and the
beam with lower values of material properties at the loading edge. Larger bending
stresses for the former beam may indicate earlier crack initiation times, which were
proven by experiments performed by other researchers. Wave propagation in a 3D
bar is also investigated. Poisson’s ratio and thickness effects are observed in the
dynamic behavior of the bar. Finite element modeling and simulation discussed
herein can be a critical tool to help understand physics behind the dynamic events.

Keywords: parallel algorithm, nonhomogeneous materials, dynamic response,

wave propagation

1. Introduction

Functionally graded materials (FGMs) are materials characterized by smooth

variation in composition, microstructure, and properties. These materials have
emerged with the need to enhance material performance and to meet specific
functions and applications [1, 2]. The material gradation concept has been utilized in
various applications [3–16]. The concept of FGMs has been applied to thermal barrier
structures and wear- and corrosion-resistant coatings and also used for joining
dissimilar materials [17]. FGMs are subjected to harsh thermal/mechanical/dynamic
environments. Thin-walled plates and shells, which are used in reactor vessels,
turbines, and other machine parts, are susceptible to buckling failure, large
deflections, or excessive stresses induced by thermomechanical loading. Functionally
graded coatings on these structural elements may help reduce these failures [18].
The dynamic response of FGMs has been investigated [19]. Reddy and Chin [20]
carried out nonlinear transient thermoelastic analysis of FGMs by using plate ele-
ments for moderately large rotations. Gong et al. [21] studied the elastic response of

17
Computational Models in Engineering

FGM shells subjected to low-velocity impact. Rousseau and Tippur [22–24] studied
the dynamic fracture of a three-point bending beam made of epoxy/glass phases
under low-velocity impact. Cheng et al. [25] used a peridynamic model to investigate
dynamic fracture of FGMs. Lindholm and Doshi [26] looked into a slender bar with
free ends and obtained a solution that is synthesized from eigenfunctions by using the
principle of virtual work. Karlsson et al. [27] developed a Green’s function approach
for 1D transient wave propagation in composite materials. Santare et al. [28] used
graded finite elements to simulate elastic wave propagation in graded materials.
Chakraborty and Gopalakrishnan [18] developed a spectral element to study wave
propagation behavior in FGM beams subjected to high-frequency impact loads.
The development of parallel computers and improvements in computer hard-
ware has been of significant importance [29]. A dynamic event has a key charac-
teristic that the incident pulse duration is very small in the order of microseconds
that makes the frequency content of pulse very high (in the order of kHz). Thus, all
the higher-order modes participate in the dynamic response, and the finite element
mesh must be very fine enough to capture the small wavelengths [18]. This makes
the system size enormously large. But in an explicit analysis, storage of large matri-
ces is avoided with the added advantage of not requiring to solve linear algebraic
equations [29]. Explicit methods are also conditionally stable, and so the time step
size must be below a critical value that is dependent on the size of the smallest finite
element [30]. Due to mesh size constraint, extremely small time steps (0.01–100 μs)
are needed to solve explicit problems. This can be overcome by parallel computing.
Krysl and Belytschko [30] investigated to parallelize an existing object-oriented
code written in C where a Parallel Virtual Machine (PVM) was used for the
necessary communication. Krysl and Bittnar [31] and Rao [32, 33] adopted the
Message Passing Interface (MPI) standard for the implementation of the exchange
algorithm, which is used in this chapter.
This chapter addresses the behavior of functionally graded materials under
dynamic impact loading within the framework of linear elasticity using MPI-based
parallel explicit algorithm. Numerical examples are presented that verify the
dynamic explicit finite element code and demonstrate the dynamic response of
graded materials. Finite element modeling and simulation can be a critical tool to
help understand physics behind the dynamic events.

2. Finite element formulation for dynamic analysis

The governing equation for structural dynamics is given by [34]

Z Z
ðσ : δE þ ρu∙δu
€ ÞdΩ � Text ∙δu dΓ ¼ 0 (1)
Ω Γ ext

where Ω is domain volume, Γ is the boundary with a normal vector n, u is the

displacement vector, T is the traction vector, σ is the Cauchy stress tensor, u€ is the
acceleration vector, ρ is the mass density, Γ ext represents the boundary surface on
which external traction Text is applied, and E is the Green strain tensor. For a finite
element of volume V and surface area S, the work balance becomes
Z Z n
fδugT fFgdV þ fδugT f∅gdS þ ∑ fδugTi fpgi
i¼1
Z � � (2)
¼ fδugT ρfu€g þ fδugT cfu_ g þ  fδεgT fσ g dV

18
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

where {F} and {Φ} represent prescribed body forces and surface tractions, f pgi
� �
and fδugi fδεgi represent prescribed concentrated loads and virtual displacements
and strains at a total of n points, and c is a damping coefficient. The finite element
discretization provides
n o n o
fug ¼ ½N �fdg; fu_ g ¼ ½N � d_ ; fu€g ¼ ½N � d€ ; fεg ¼ ½B�fdg (3)

in which N are the shape functions, the nodal displacement vectors fdg are
functions of time, and [B] is the matrix of shape function derivatives. The equation
of motion is
� � � �
€ þ ½C� D_ þ ½K �fDg ¼ fF g
½M� D (4)

where the consistent mass (m), damping (c), and stiffness (k) matrices are given by
Z Z Z
T T
½M� ¼ ϱ ðxÞ½N � ½N �dV ½C� ¼ cðxÞ½N � ½N �dV ½K � ¼ ½B�T ½DðxÞ�½B�dV (5)

where mass density, damping coefficient, and constitutive matrices are func-
tions of spatial positions.

3. Explicit method, stability, and mass lumping

Dynamic response is evaluated at discrete instants of time separated by time

increments Δt. At the nth time step, the equation of motion for linear dynamic
problems is given by [34]

½M�fu€gn þ ½C�fu_ gn þ ½K �fugn ¼ fF ext gn (6)

Discretization in time is accomplished using finite difference approximations of

time derivatives. Explicit algorithms use a difference expression of the general form:
� �
fugnþ1 ¼ f fugn ; fu_ gn ; fu€gn ; fugn�1 ; … (7)

which contains only historical information on its right-hand side. This difference
expression is combined with the equation of motion at time step n.
The stability of the explicit finite element scheme is governed by the Courant
condition [34], which provides a critical time step size beyond which the results
become unstable. It is given by

Δt ≤ le =Cd (8)

where le is the shortest distance between two nodes in the mesh and the dilata-
tional wave speed Cd is expressed as

EðxÞð1 � νðxÞÞ
C2d ¼ (9)
ð1 þ νðxÞÞð1 � 2νðxÞÞρðxÞ

Since Cd varies due to material gradation, the desired time step for the finite
element implementation is chosen based on the largest dilatational wave speed and
is maintained constant throughout the simulation. The mass matrix used in the
explicit method described above is typically lumped in order to avoid matrix

19
Computational Models in Engineering

inversion. The HRZ technique [35] is used to lump the mass matrix in which only
the diagonal terms are retained and scaled so as to preserve the total mass:
!,� �
n n
mlii ¼ mii ∑ mij ∑ mii (10)
j¼1 i¼1

where ml is the lumped element mass matrix, m is the consistent mass matrix,
and n is the number of degrees of freedom in the element.

4. Explicit parallel FEA using MPI

A master-slave approach in the parallel finite element code is used here to solve
wave propagation-type problems using the Message Passing Interface (MPI)
standard [36]. Figure 1 shows the procedure used in the parallel execution of the

Figure 1.
Time-integration procedure in the parallel execution of the explicit Newmark-β method [37] using MPI
standard [36].

20
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

explicit Newmark-β method. The first step is to partition the structure (FE mesh)
into number of sub-domains. The stiffness matrices are calculated for elements in
the local processor. The stiffness matrix needs not be assembled in an explicit
analysis as the algorithm is implemented on an element basis only. The parallel
explicit time integration analysis of finite elements from onetime step to another
consists in obtaining the unknown velocity and displacement of master and slaves,
sending the effective force vector from slaves to the master for assembly and
sending back the assembled force vector to the slaves for calculation of acceleration
vector. The communication between the slaves and the master processor is done
using MPI-based function calls as shown in the figure. The entire force vector at
each processor is being sent to the master processor for assembly, and the master
processor returns the assembled force vector to the slaves in the same manner.
These steps are repeated for the rest of the time integration. Local field quantities
like element stresses and strains can be calculated separately on each processor
without any communication.

5. Example 1: 3D graded beam under velocity impact

A 3D three-point bending beam under velocity impact is studied to understand

the influence of material gradation on the bending behavior in a 3D space. The
beam is a real FGM system made of glass/epoxy phases. The dynamic fracture
experiments of the linearly graded specimens have been conducted by Rousseau
and Tippur [37–40].
In this paper, we investigate the 3D dynamic behavior of a beam. Using 3D
explicit finite element code, we can simulate the 3D dynamic behavior of the beam.
This 3D finite element study offers enhanced knowledge of the dynamic behavior of
this material system and understanding of the stress field in homogeneous and graded
materials, which help to predict fracture initiation times in various graded specimens.
Consider a three-point bending specimen under velocity impact load at the top
as shown in Figure 2(a). Due to the symmetry of the geometry and the loading
conditions, one-fourth of the beam is modeled as shown in Figure 2(b).
Figure 2(c) and 2(d) shows the 3D FE mesh of the one-fourth model and a zoom of
the FE mesh at the left bottom corner, respectively. The mesh is refined along the
loaded edge with a uniform element size of 92.5 μm, and the stress history at point
P(0, 0.2 W) is retrieved, where W is the beam depth. Point P is of significance
because it corresponds to the location of the crack tip in the dynamic fracture
analysis of the cracked beam [38, 41].
Three material gradation cases are considered for the dynamic analysis of the
three-point bending beam: a homogeneous beam (Homog, E2 = E1), a beam stiffer
at the impacted surface (StiffTop, E2 > E1), and a beam softer at the impacted
surface (StiffBot, E2 < E1), where subscripts 1 and 2 denote bottom and top surfaces
of the beam, respectively. For the graded beam, Young’s modulus (E) is varied
linearly between 4 GPa and 12 GPa, and the mass density (ρ) is varied linearly
between 1000 kg/cm3 and 2000 kg/cm3. These material properties correspond to
values used in [42]. For the homogeneous beam, the mass density (ρ) is taken as the
average of the graded beam counterparts, and Young’s modulus (E) is calculated
such that the equivalent E/ρ value equals that of the graded specimen in an average
sense, i.e.,
Z W
E 1 Eð yÞ
¼ dy (11)
ρ W 0 ρð yÞ

21
Computational Models in Engineering

Figure 2.
Epoxy/glass beam subjected to velocity impact: (a) geometry and boundary conditions of the three-point bending
specimen and (b) one-quarter model with symmetric boundary conditions. Stress values are retrieved at
P(0, 0.2 W). (c) 3D FE mesh of the one-quarter model. The FE mesh contains 14,085 15-node wedge elements
and 44,827 nodes. (d) Zoom of the FE mesh near the left bottom corner, i.e., at x = 0.

Poisson’s ratio is 0.33 for all the three cases. The average dilatational wave speed
(Cd(avg)) is defined as
Z W
1
Cd ðavg Þ ¼ Cd ðyÞdy (12)
W 0

where Cd is calculated assuming plane stress behavior given by

EðyÞ
C2d ¼ (13)
ð1 � ν2 ðyÞÞρðyÞ

The dilatational wave speed (Cd) is calculated assuming plane stress behavior to
compare results with those obtained by Zhang and Paulino [42]. The difference in
Cd(avg) for each material gradation is marginal with 2421.5 m/s for the homogeneous
beam and 2418.4 m/s for the graded beams. Hence, the dilatational wave speed is
assumed constant as 2421.5 m/s, and the results are normalized based on this
average dilatational wave speed. Upon impact on the top surface of the beam,
compressive stress waves are generated and propagate toward the bottom surface
and reflect back as tensile waves from the bottom surface.
A detailed stress history analysis is performed to obtain the influence of material
gradation on the bending behavior of the beam. Figure 3 shows the bending stress
(σx) history of point P due to impact velocity of 5 m/s for the three cases mentioned
above with normalized time t0 = t*Cd/W as the x-axis and stress (MPa) as the y-axis.
We see that material gradation leads to considerable change in the stress field at
point P. The stress wave takes normalized time of about 0.7 to reach the point P,
which is different from 0.8 in Zhang and Paulino [42]. This is due to 3D effects. In a
3D medium, stress waves travel at different speeds when compared to a 2D
medium. At this time, the point P undergoes compressive stress until the

22
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

normalized time of around 1.1. This compressive stress is due to the Poisson’s ratio
(ν) effect. In a separate simulation of the beam under similar loading conditions, the
Poisson’s ratio of the beam was made zero, and we found that the point P
underwent tensile stress throughout the simulation (result not shown). From this
instant, i.e., after normalized time of 1.1, the point P undergoes monotonically
increasing tensile stress. The maximum tensile stress is consistently attained in the
StiffTop beam, and the minimum in the StiffBot beam and the Homog beam
undergoes stresses in between these two. This type of behavior may be intuitively
observed as the material properties (E, ρ) of a StiffTop beam at point P are lower
than the StiffBot and Homog beams and hence may undergo higher stresses. We
also found that for the line load, the stress component σx along the thickness (z)
direction at point P(y = 0.2 W) does not vary considerably.
Figure 4 shows the stress history of component σy at point P for different
material gradations due to impact velocity of 5 m/s. We see that there is a consid-
erable effect of the gradation on this stress behavior, too. The first stress wave
approaches point P at normalized time of 0.7 due to the compressive velocity
loading. At this point, there is a sudden increase in σy stress magnitude (compres-
sive) until a normalized time of 1.25. The stress wave reaches the bottom surface of
the beam and reflects as tensile waves at which point there is a gradual decrease in
the compressive stress until the normalized time of 2.0. Between normalized times
2.0 and 2.7, the stress behavior does not change considerably, although there are
small variations, which may be because of the discrete wave reflections from the
edges that occur in finite configurations, to form a plateau in the stress field. The
stress wave, after reflecting from the top surface, becomes compressive and leads to
an increase in compressive stress between normalized times 2.7 and 3.2. Between
the normalized times of 3.2 and 6, the stress behavior tends to repeat itself as seen
between 0.7 and 3.2, although with larger variations of increase and decrease. The
results plotted in Figures 13 and 14 are consistent with the plane stress simulation
results obtained by Zhang and Paulino [42] and Rousseau and Tippur [38].
Figure 5 shows the 2D contour plot of the stress component σx at a specific time
instant of 90 μs (t0 � 5.9) for the three beams subjected to the impact velocity of
5 m/s. We firstly see that the stress pattern is considerably different among the

Figure 3.
Stress history σx at location P (0, 0.2 W) for homogeneous and graded beams with linearly varying E and ρ,
subjected to velocity impact (V = 5 m/s) at the top.

23
Computational Models in Engineering

Figure 4.
Stress history σy at location P (0, 0.2 W) for homogeneous and graded beams with linearly varying E and ρ,
subjected to velocity impact (V = 5 m/s) at the top.

beams considered. Further, at regions close to the loading, compressive stress is

concentrated for all the beams, although the size of the compressive zone differs
considerably. The StiffTop beam has the largest compressive region at the loading
region followed by the Homog beam and the StiffBot beam. At regions close to the
loading for the StiffTop beam, we see that the material properties are relatively
higher than the other cases and hence the compressive load gets transferred to
nearby regions. For the StiffBot case, the compressive region is localized in a small
region because of the compliance at the top. Also, we see that the compressive
region extends the most in the x direction in the same order as mentioned above.
We observe that the size of the tensile zone near the bottom of the beam (i.e., x = 0,
y = 0) varies in the same order as the compressive zone at the loading region. The
neutral axis location varies for each of the graded beams considered which may be
the reason behind the larger tensile region being developed for StiffTop beam when
compared to the other two cases. For the dynamic loading considered, the neutral
axis shifts in time; nevertheless, the overall stress pattern will be similar to those
shown in Figure 5. With respect to crack initiation in such graded beams (assuming
that the crack lies parallel to the applied loading), we know that the σx stress
component dominates the most. We see that the StiffTop beam is consistently
subjected to larger tensile stresses than other two cases at point P and will lead to
earlier crack initiation.
Figures 6 and 7 show the σx and σy plots for the different graded beams
subjected to 5 m/s and 10 m/s velocities of impact. We see that the stress behavior
increases linearly (gets doubled almost) for V = 10 m/s when compared to the case
of V = 5 m/s for all the gradation cases.
We have so far investigated the 2D dynamic behavior of a beam under the line
load using 3D finite element method. Using 3D explicit finite element code, we can
simulate the 3D dynamic behavior of the beam. We apply the impact point load of
V = 10 m/s at a central node at the top edge of the beam. Note that this cannot be
simulated using 2D finite elements in Zhang and Paulino [41]. The stress history
results are obtained at point P and compared against those shown in Figure 7.

24
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

Figure 5.
Stress contour (σx MPa) at t = 90 μs due to impact velocity of 5 m/s. (a) StiffBot beam (E2 < E1),
(b) homogeneous beam (E2 = E1), and (c) StiffTop beam (E2 > E1).

Figures 8 and 9 show the comparison of σx and σy, respectively, for the three
beams under impact velocity of V = 10 m/s applied throughout the thickness (line
load) and at one central node (point load). We see that the stresses for the latter
case (point load) are much lower than the former case (line load). Maximum tensile
stress (σx) at point P due to point load is experienced by the StiffTop beam followed
by the Homog beam and the StiffBot beam, similar to the line load case. The stress

25
Computational Models in Engineering

Figure 6.
Stress history σx at location P(0, 0.2 W) for homogeneous and graded beams subjected to two different velocities.

Figure 7.
Stress history σy at location P(0, 0.2 W) for homogeneous and graded beams subjected to two different velocities.

values at point P across the z (thickness) direction do not change noticeably (result
not shown).

6. Example 2: wave propagation in 3D bar

Wave propagation in a fixed-free 3D bar with homogeneous and graded mate-

rials in the vertical direction (y) is simulated to illustrate the relevance of material
gradation and also to verify the parallel explicit 3D finite element code. Consider a

26
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

Figure 8.
Stress history σx at location P(0, 0.2 W) for homogeneous and graded beams subjected to impact velocity of
10 m/s as a line load and point load (see the insert). Thick and thin lines correspond to line load and point load,
respectively. Solid, dashed, and dash-dot lines correspond to StiffTop, Homog, and StiffBot beams.

Figure 9.
Stress history σy at location P(0, 0.2 W) for homogeneous and graded beams subjected to impact velocity of
10 m/s as a line load and point load (see the insert). Thick and thin lines correspond to line load and point load,
respectively. Solid, dashed, and dash-dot lines correspond to StiffTop, Homog, and StiffBot beams.

fixed-free bar in Figure 10(a). The square bar is of length L = 1 m and height
H = 0.05 m. A step loading is used (Figure 10(b)) at the free end of the bar. The
material considered is steel for the homogeneous bar and steel and alumina for the
graded bar. The properties of steel and alumina are given in Table 1. For the graded
bar, the lower surface of the bar is made of alumina, and the upper part is made of
steel, and the material properties vary linearly from alumina to steel.

27
Computational Models in Engineering

Figure 10.
(a) Schematic of the 3D bar and (b) step load.

Material E (GPa) ν ρ (kg/m3) Cd (m/s)

Steel 210 0.31 7800 6109

Alumina 390 0.22 3950 10,617

Table 1.
Material properties of steel and alumina.

Explicit time integration method is used for analyzing wave propagation type of
problems since it avoids storage of large matrices and does not require the solution
of global systems of equations, but is limited by its conditional stability [43]. The
time step size has to be smaller than a critical value which is directly dependent on
the largest frequency of the FE discretization (smallest element). Higher-order
modes participate in the bar response which in turn requires refined element size.
To capture the transient response, the time step size needs to be small enough in
order to also track the change of applied loads, and adequate element size can be
estimated according to the Courant condition [34]. Considering material gradation
in the y direction, the FE mesh is discretized into 300 � 15 � 15 quads, each quad
divided into four 15-node wedge elements, totaling 270,000 elements and 725,836
nodes. The 1D analytical solution for the motion of a bar with step force loading at
the free end is obtained using a finite sine transform method given by [44]

P0 8P0 L ∞ ð�1Þn�1  πx  πc 

uðx; tÞ ¼ x� 2 ∑ 2
sin ð2n � 1Þ cos ð2n � 1Þ t (14)
EA π EA n¼1 ð2n � 1Þ 2L 2L

where P0, E, A, L, and c represent the applied force magnitude, the Young’s
modulus, cross-sectional area, length, and dilatational wave speed of the bar,
respectively. The first term on the right side of Eq. (14) is the static solution,
whereas the series represents a superposition of normal modes of amplitudes
inversely proportional to (2n � 1)2. The axial stress can be easily obtained by
differentiating displacement Eq. (14).
Results are presented for the 3D wave propagation in a bar under step loading
considering homogeneous and graded material properties. Figures 11 and 12 show

28
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

Figure 11.
Normalized longitudinal stress history at three locations (see the insert) of the homogeneous bar subjected to step
loading using 1D analytical solution given in Eq. (14).

Figure 12.
Normalized longitudinal stress history at three locations (see the insert) of the homogeneous bar subjected to step
loading obtained using finite element analysis.

the normalized longitudinal stress at three locations along the x-axis obtained using
the exact solution and the finite element method, respectively. For the exact solu-
tion, we see small oscillation of the stress solution at x = 0 at t = 1 which is due to
finite series of n used in Eq. (14). Poisson’s ratio (ν) is taken as zero for the present
finite element simulation. From the figures, we see that the overall stress behavior
of the finite element simulation is similar to the exact solution, although there are

29
Computational Models in Engineering

some oscillations. These oscillations are seen at times when there is a sudden
increase or decrease in stress due to wave reaching a certain location in the bar. This
oscillation effect may be due to discrete wave reflections that occur in finite con-
figurations and also due to the finite element discretization.
Figures 13 and 14 show the normalized longitudinal stress at six locations
(see the insert) for the functionally graded bar along z = 0.025 m and z = 0 m,

Figure 13.
Normalized longitudinal stress history of six points (see the insert) on a graded bar (z = 0.025 m) subjected to
step loading. Solid and dashed lines indicate points at x = 0 and x = 0.5 L, respectively. Thick, intermediate-
thick, and thin lines indicate alumina-rich side, midplane, and steel-rich side points, respectively.

Figure 14.
Normalized longitudinal stress history of six points (see the insert) on a graded bar (z = 0 m) subjected to step
loading. Solid and dashed lines indicate points at x = 0 and x = 0.5 L, respectively. Thick, intermediate-thick,
and thin lines indicate alumina-rich side, midplane, and steel-rich side points, respectively.

30
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

respectively. The stress history at x = L is not plotted as the behavior is similar to the
homogeneous case. We see that the stress wave gets distorted in time because of the
material gradation. The stress wave at x = 0.5 L and mostly at x = 0 differs highly
across the thickness direction due to difference in wave speeds. As expected, the
alumina side undergoes higher stresses than the steel side, more so at the fixed end
than at other x locations.

7. Conclusions

Dynamic and wave propagation behavior of 3D functionally graded continua is

investigated using explicit finite element formulations. Three numerical examples
are presented. The first example is on dynamic analysis of a three-point bending
beam made of epoxy and glass phases under low-velocity impact. Material grada-
tion considerably affects the dynamic stress behavior of the beam. Tensile stress is
maximum for StiffTop beam at the imaginary crack-tip location for stress compo-
nent (σx) indicating that crack initiation will occur earlier for this type of beam
which was also verified by experiments conducted by Rousseau and Tippur [40].
The second example is on 3D wave propagation behavior of a homogeneous and
functionally graded bar under sinusoidal and step loading. Stress results for the
homogeneous bar with zero Poisson’s ratio obtained using the finite element
method match very well with the exact solution. Poisson’s ratio and material con-
straints in a 3D continuum affect the stress behavior for both homogeneous and
functionally graded bars. In the functionally graded bar, the alumina side consis-
tently experienced more stress than the steel side. Finite element modeling and
simulation discussed herein can be a critical tool to help understand physics behind
the dynamic events.

Author details

Ganesh Anandakumar and Jeongho Kim*

Department of Civil and Environmental Engineering, University of Connecticut,
Storrs, CT, USA

*Address all correspondence to: jeongho.kim@uconn.edu

© 2018 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

31
Computational Models in Engineering

References

[1] Ootao Y, Kawamura R, Tanigawa Y, deformation-induced martensitic

Imamura R. Optimization of material transformation. Journal of Materials
composition of nonhomogeneous Science Letters. 1993;12(5):326-328
hollow circular cylinder for thermal
stress relaxation making use of neural [9] Koike Y. Graded-index and single
network. Journal of Thermal Stresses. mode polymer optical fibers. In: Chiang
1999;22:1-22 LY, Garito AG, Sandman DJ, editors.
Electrical, Optical, and Magnetic
[2] Ilschner B. Processing- Properties of Organic Solid State
microstructure-property relationships Materials. Vol. 247. Pittsburgh, PA:
in graded materials. Journal of the Materials Research Society Proceedings;
Mechanics and Physics of Solids. 1996; 1992. p. 817
44(5):647-656
[10] Udupa G, Rao SS, Gangadharan KV.
[3] Hirano T, Teraki J, Yamada T. On the Functionally graded composite
design of functionally gradient materials: An overview. Procedia
materials. In: Yamanouochi M, Koizumi Materials Science. 2014;5:1291-1299
M, Hirai T, Shiota I, editors.
Proceedings of the First International [11] Miyamoto Y, Kaysser WA, Rabin
Symposium on Functionally Gradient BH, Kawasaki A, Ford RG. Functionally
Materials. Japan: Sendai; 1990. pp. 5-10 Graded Materials: Design, Processing
and Applications. Berlin, Germany:
[4] Gupta A, Shirvanimoghaddam K. Springer; 2013
Functionally graded materials: A review
of fabrication and properties. Applied
[12] Kieback B, Neubrand A, Riedel H.
Materials Today. 2016;5:223-245
Processing techniques for functionally
graded materials. Materials Science and
[5] Liew KM, Lei ZX, Zhang LW.
Engineering A. 2003;362(1-2):81-106
Mechanical analysis of functionally
graded carbon nanotube reinforced
[13] Hart NT, Brandon NP, Day MJ,
composites: A review. Composite
Shemilt JE. Functionally graded
Structures. 2015;120:90-97
cathodes for solid oxide fuel cells.
Journal of Materials Science. 2001;36:
[6] Sola A, Bellucci D, Cannillo V.
1077-1085
Functionally graded materials for
orthopedic applications – an update on
design and manufacturing. [14] Barthel K, Rambert C. Thermal
Biotechnology Advances. 2016;34(5): spraying and performance of graded
504-531 composite cathodes as SOFC-
component. Materials Science Forum.
[7] Osaka T, Matsubara H, Homma T, 1999;308-311:800-805
Mitamura S, Noda K. Microstructural
study of electroless-plated CoNiReP/ [15] Holtappels P, Bagger C. Fabrication
NiMoP double-layered media for and performance of advanced multi-layer
perpendicular magnetic recording. sofc cathodes. Journal of the European
Japanese Journal of Applied Physics. Ceramic Society. 2002;22:41-48
1990;29(10):1939-1943
[16] Gerk C, Willert-Porada M.
[8] Watanabe Y, Nakamura Y, Fukui Y, Development of graded composite
Nakanishi K. A magnetic-functionally electrodes for the SOFC. Materials
graded material manufactured with Science Forum. 1999;308-311:806-815

32
Three-Dimensional Finite Element Analysis for Nonhomogeneous Materials Using Parallel…
DOI: http://dx.doi.org/10.5772/intechopen.82410

[17] Suresh S, Mortensen A. materials. Composite Structures. 2015;

Fundamentals of Functionally Graded 133:529-546
Materials. London: IOM
Communications Ltd; 1998 [26] Lindholm US, Doshi KD. Wave
propagation in an elastic
[18] Chakraborty A, Gopalakrishnan J. A nonhomogeneous bar of finite length.
spectrally formulated finite element for Journal of Applied Mechanics.
wave propagation analysis in Transactions ASME. 1965;32:135-142
functionally graded beams.
International Journal of Solids and [27] Karlsson A, Otterheim H, Stewart R.
Structures. 2003;40(10):2421-2448 Transient wave propagation in
composite media: Green’s function
[19] Chiu TC, Erdogan F. One- approach. Journal of the Optical Society
dimensional wave propagation in a of America A. Optics and Image Science.
functionally graded elastic medium. 1993;10:886-895
Journal of Sound and Vibration. 1999;
222(3):453-487 [28] Santare MH, Thamburaj P, Gazonas
GA. The use of graded finite elements in
[20] Reddy JN, Chin CD. the study of elastic wave propagation in
Thermomechanical analysis of continuously nonhomogeneous
functionally graded cylinders and plates. materials. International Journal of Solids
Journal of Thermal Stresses. 1998;26(1): and Structures. 2003;40:5621-5634
593-626
[29] Ekevid T, Wiberg N-E. A
[21] Gong SW, Lam KY, Reddy JN. The comparison of parallel implementation
elastic response of functionally graded of explicit DG and central difference
cylindrical shells to low-velocity impact. method. Communications in Numerical
International Journal of Impact Methods in Engineering. 2002;18:
Engineering. 1999;22:397-417 585-597

[22] Rousseau C-E, Tippur HV. [30] Krysl P, Belytschko T. Object-

Compositionally graded materials with oriented parallelization of explicit
cracks normal to the elastic gradient. structural dynamics with PVM.
Acta Materialia. 2000;48(16): Computers and Structures. 1998;66
4021-4033 (2-3):259-273

[23] Rousseau C-E, Tippur HV. Dynamic [31] Krysl P, Bittnar Z. Parallel explicit
fracture of compositionally graded finite element solid dynamics with
materials with cracks along the elastic domain decomposition and message
gradient: Experiments and analysis. passing: Dual partitioning scalability.
Mechanics of Materials. 2001;33: Computers and Structures. 2001;79:
403-421 345-360

[24] Rousseau C-E, Tippur HV. [32] Rao ARM. Explicit nonlinear
Evaluation of crack tip fields and stress dynamic finite element analysis on
intensity factors in functionally graded homogeneous/heterogeneous parallel
elastic materials: Cracks parallel to computing environment. Advances in
elastic gradient. International Journal of Engineering Software. 2006;37:701-720
Fracture. 2002;114:87-111
[33] Rao ARM. A parallel mixed time
[25] Cheng Z, Zhang G, Wang Y, Bobaru integration algorithm for nonlinear
F. A peridynamic model for dynamic dynamic analysis. Advances in
fracture in functionally graded Engineering Software. 2002;33:261-271

33
Computational Models in Engineering

[34] Bathe K-J. Finite Element [43] Rao ARM. Explicit nonlinear
Procedures. New Jersey: Prentice-Hall; dynamic finite element analysis on
1996 homogeneous/heterogeneous parallel
computing environment. Advances in
[35] Hilber HM, Hughes TJR, Taylor RL. Engineering Software. 2006;37:701-720
Improved numerical dissipation for time
integration algorithms in structural [44] Meirovitch L. Analytical Methods in
dynamics. Earthquake Engineering and Vibrations. New York: The Macmillan
Structural Dynamics. 1977;5:283-292 Company; 1967

[36] MPI-2: Extension to the Message

Passing Interface. Knoxville, Tennessee:
University of Tennessee; 1997

[37] Rousseau C-E, Tippur HV.

Compositionally graded materials with
cracks normal to the elastic gradient.
Acta Materialia. 2000;48(16):
4021-4033

[38] Rousseau C-E, Tippur HV. Dynamic

fracture of compositionally graded
materials with cracks along the elastic
gradient: Experiments and analysis.
Mechanics of Materials. 2001a;33:
403-421

[39] Rousseau C-E, Tippur HV.

Influence of elastic gradient profiles on
dynamically loaded functionally graded
materials: Cracks along the gradient.
International Journal of Solids and
Structures. 2001b;38:7839-7856

[40] Rousseau C-E, Tippur HV.

Evaluation of crack tip fields and stress
intensity factors in functionally graded
elastic materials: Cracks parallel to
elastic gradient. International Journal of
Fracture. 2002;114:87-111

[41] Zhang Z, Paulino GH. Cohesive

zone modeling of dynamic fracture in
homogeneous and functionally graded
materials. International Journal of
Plasticity. 2005;21(6):1195-1254

[42] Zhang Z, Paulino GH. Wave

propagation and dynamic analysis of
smoothly graded heterogeneous
continua using graded finite elements.
International Journal of Solids and
Structures. 2007;44:3601-3626

34
Chapter 3

A New Concept to Numerically

Evaluate the Performance of
Yielding Support under Impulsive
Loading
Faham Tahmasebinia, Chengguo Zhang, Ismet Canbulat,
Samad M.E. Sepasgozar, Onur Vardar, Serkan Saydam
and Chen Chen

Abstract

The dynamic capacity of a support system is dependent on the connectivity and

compatibility of its reinforcement and surface support elements. Connectivity
refers to the capacity of a system to transfer the dynamic load from an element to
another, for example, from the reinforcement to the surface support through plates
and terminating arrangements (split set rings, nuts, etc.), or from a reinforcement/
holding element to others via the surface support. Compatibility is related to the
difference in stiffness amongst support elements. Load transfer may not take place
appropriately when there are strong stiffness contrasts within a ground support
system. Case studies revealed premature failures of stiffer elements prior to utilising
the full capacity of more deformable elements within the same system. From a
design perspective, it is important to understand that the dynamic-load capacity of
a ground support system depends not only on the capacity of its reinforcement
elements but also, and perhaps most importantly, on their compatibility with other
elements of the system and on the strength of the connections. The failure of one
component of the support system usually leads to the failure of the system.

Keywords: fully grouted rock bolts, yielding support, coal burst, shear
dynamic loading

1. Introduction

Currently, rock/cable bolts and steel mesh are widely used as temporary and
permanent support element in tunnelling, underground excavations, and the
surface slope stability. A combination of the rock/cable bolts and steel mesh can
provide a robust system, which is known as a combined or yielding support. A
combined support can transfer the applied external load from the damaged exterior
to the confined interior of a rock mass. The load transferring mechanism between
the unstable exterior and interior of a rock mass by means of the combined support
system is still of the grey areas in designing the support systems. The understanding
the load transferring mechanism between the engaged elements in a rock mass will

35
Computational Models in Engineering

be more complicated when it comes to designing effective elements against rock/

coal burst. Ref. [1] defined ground support as a combination of surface support and
reinforcement systems. Ref. [1] also mentioned that the words support and rein-
forcement are usually used interchangeably. Another concept to recall the support
system as a membrane system is determined by [2]. Ref. [2] classified the surface
support or containment support as the application of the reaction of the forces at
the face of the excavation where it can include the techniques and smart devices to
establish both local and global supports. These techniques and devices included fill,
timber, steel sets, shotcrete, and the steel mesh. Ref. [3] defined rock reinforcement
as an enhancement of the overall rock mass properties within the rock mass, and
contain all techniques and devices that perform within the rock mass, including
rock bolts, cable bolts, and ground anchors. Dynamic support is defined as ground
support systems in which it would be available to resist against sudden energy
release and dynamic loading so as to uphold the strength of an underground exca-
vation. A number of researchers including [4–8] focused on studying the shear
behaviour of rock bolts when it comes to the high stress and jointed rock mass
condition. Ref. [9] classified the procedure for rock mass damage mechanisms by
including the sudden volume expansion or bulking of the rock due to fracturing of
the rock mass around an excavation. Rock mass bulking is a major cause of damage
to support burst-prone condition.

2. Overview of research on yielding support

Ansell [10] carried out further tests on another type of energy-absorbing rock
bolt made of soft steel. The aim of the test was to determine the strain rate effects
on the yield stress and ultimate strength of the steel bars under dynamic loading
conditions. The results demonstrated that the strain rate has a pronounced influ-
ence on the yield stress of steel bars. Ansell [10] concluded that the yield stress, the
ultimate strength, and the energy absorbing capacity increased with the increasing
loading rate, and the soft steel bars had a delayed plastic yielding when subjected to
the dynamic loading. The impact was plastic during the retardation of the movable
components of the testing machine; it was large at the beginning but dropped
dramatically as an exponential curve. Ansell [10] also proposed that a rock bolt used
under dynamic loading should have high dynamic-yield capacity so that it can resist
high peak forces when experiencing impact loading. There are two methods com-
monly used to test energy absorption capacity of rock bolts under dynamic loading:
free fall of a mass and momentum transfer. Plouffe et al. [11] performed the free fall
method in the laboratory, where dynamic loading was simulated by dropping a
mass over a certain distance onto the impact plate attached to the lower tube. The
load and the displacement at the split were recorded by a load cell under the plate
and a differential extensometer, respectively. The performance of bolts under
dynamic loading was then analysed in terms of loads, displacements, and energy
dissipated. The results showed that the potential and kinetic energies were almost
equal to each other, and not all the energy was transferred to the support elements.
The dissipated energy results in noise and permanent deformation of the domed
plate. An experiment was carried out by [12] to estimate the energy absorption
capacity of a new rock bolt, named D bolt, which has a large capacity for load
bearing and deformation. Both static pull tests and dynamic drop tests were
performed to evaluate the characteristics of the D bolt. For dynamic tests, boreholes
were simulated by a split steel tube, which was placed in a jig to align it with a drill.
During the test, a mass was dropped from a certain height onto a plate connected to
the D bolt, and the impact load and the plate load were then measured during the

36
A New Concept to Numerically Evaluate the Performance of Yielding Support under Impulsive…
DOI: http://dx.doi.org/10.5772/intechopen.79643

dropping process. The results indicated that the D bolt’s impact peak load capacity
was slightly larger than the static tensile strength of the bolt. The D bolt absorbed a
large amount of energy along its entire length. The bolt was equally loaded in each
deformable section which worked alone to stop rock expansion. Ghadimi et al. [13]
developed an analytical model to calculate shear stress in a fully grouted rock bolt in
jointed rock mass. The model considered the bolt profile and jump plane under pull
test conditions. The results demonstrated that shear stress from the bolt to the rock
decreased in an exponential rate. This decrease in shear stress is determined by bolt
profiles such as the height, width and spacing of rib, resin thickness, material, and
jointed properties. Jiang et al. [14] proposed another analytical model of the grouted
rock bolt. The coupling behaviour of the rock bolt and rock mass was discussed
from the point of displacement. Based on the analysis, the initial force in rock mass
was controlled by its displacement. Another finding was that each bolt has at least
one neutral point whose position is influenced by the displacement in the rock mass
around the tunnel. The axial force of a rock bolt and the shear stress at the interface
between the bolt and rock mass were affected by the length of the rock bolt, the
internal radius of the tunnel, and the characteristics of the rock mass. A similar
analytical model was also proposed by [15], in which the stress of distribution and
variation of fully grouted rock bolts in surrounding rock mass were investigated and
analysed. Based on the stress equilibrium in a tiny area of a rock bolt and the shear
stress transfer mechanism of the interface between the rock bolt and rock mass, the
axial displacement differential equation of the rock bolt was developed. Aziz et al.
[16] researched the load transfer capacity of fully grouted rock bolts under an
elastoplastic rock mass condition and proposed an analytical solution to determine
the axial load along the rock bolts. In this model, the rock bolts were assumed to be
elastic material and the surrounding rock mass was considered as an elastoplastic
material. According to the model analysis, the load transfer capacity of a fully
grouted rock bolt can be expressed as a function of various parameters of the
surface condition, including bolt length, shear stiffness of the interfaces, in situ
stress, and rock displacement along the bolt. Rock bolts are part of the reinforce-
ment support system in underground excavations. It is important to evaluate
and understand the performance of bolts under varying loads. Laboratory
testing methods, especially the double shear test, can be used to determine the
mechanical properties of rock bolts under static loading. Tests for rock bolts
under dynamic loading are used to determine the energy absorption capacity,
which is an important parameter for describing the dynamic performance of a
rock bolt. Numerical modelling by using simulation software and theoretical
analysis can also be used to analyse the behaviour of rock bolts in a jointed rock
mass. Since experimental investigation of cable bolts under dynamic loading is very
complicated and requires sophisticated facilities, numerical modelling can be used
as a reliable method.

3. Numerical procedure

The proposed three-dimensional Finite Element model is developed using the

commercial software ABAQUS, due to its ability to deal with complex contact
problems. In fact, one of the main difficulties in the modelling of steel and concrete
members with ABAQUS is in the convergence issues which need to be addressed
due to the extensive number of contacts required to be implemented between the
cable bolt and the concrete boxes. The proposed numerical model is developed to
predict the structural response of the cable bolts using solid elements for all com-
ponents. For this purpose, the 8-node linear brick element (C3D8R) with a reduced

37
Computational Models in Engineering

integration and hourglass control is adopted, which is the element with three tran-
sitional degrees of freedom (Figure 1).
One of the main difficulties in the modelling of steel and concrete members with
ABAQUS is in the convergence issues which need to be addressed due to the
extensive number of contacts required to be implemented between the cable bolt
and the concrete blocks. The proposed numerical model is developed to simulate the
structural response of the cable bolts using solid elements for all components. For
this purpose, the 8-node linear brick element (C3D8R) with a reduced integration
and hourglass control is adopted, which is the element with three transitional
degrees of freedom. The base model in this chapter is similar to the author’s previ-
ous work [17], where preliminary numerical modelling and analytical study were
conducted to examine cable behaviour. This study provides further parametric
analysis on bolts behaviour under different conditions. The adequacy and stability
of a Finite Element model to describe the behaviour of a cable bolt is strongly
influenced by the definition of adequate contact properties between the concrete
and steel components, as suggested by [17]. The presence of a large number of
contact regions, especially when dealing with cable bolt models, significantly
increases the complexity of the analysis due to the nonlinear and discontinuous
nature of cable bolts. For this reason, the Finite Element simulations carried out as
part of this study are implemented with ABAQUS/Explicit, to avoid convergence
difficulties. The interface behaviour between the concrete box and the cable bolt is
modelled using the surface-to-surface interaction available in ABAQUS. In particu-
lar, a HARD contact property is adopted in the direction normal to the interface
plane. For the tangential behaviour, the PENALTY option is used with a friction
coefficient of 0.5 between the cable bolts and the grout. The behaviour of the cable
bolts is described using a linear elastic relationship up to yielding, followed by a
strain-hardening behaviour. The concrete in compression is described with an ini-
tial linear elastic range up to 40% of its compressive strength after which it is
represented by the Concrete Damage Plasticity model available in ABAQUS. Its

Figure 1.
An example of 3-D finite element models under static and dynamic loading.

38
A New Concept to Numerically Evaluate the Performance of Yielding Support under Impulsive…
DOI: http://dx.doi.org/10.5772/intechopen.79643

tensile behaviour is initially linear elastic followed by a softening response after

cracking is initiated.

4. Analytical and numerical solution

Plate is one of the important elements which may influence the overall perfor-
mance of the combined support. To date, a little attention was devoted by the
industry to evaluate the performance of the plate in the roof and rib blot system.
The strength of the head plate can be part of a roof or rib bolt system component.
However roof bolting system that uses pre-tension for reinforcing purposes may not
always achieve full encapsulation, the head plate is in fact a vital component of the
system. Thus, at this stage, different and critical buckling modes in a steel plate with
hole in the centre of the plate were presented. A number of researchers [18–22]
theoretically and experimentally reported buckling of square plates with central
circular holes subjected to compression in the plane. The presented theoretical
methods by most of the researchers [18–21] calculate the critical loads in a plate
with hole in the centre were extracted from the minimum energy method [23].
However, the available methods buckling and post-buckling solutions, which are
mathematically discussed, are limited to small hole sizes, and are not able to study
the effects of different boundary conditions of the plate on the strengths of buckling
plates with holes of arbitrary size. By inventing powerful computers and advantag-
ing from simulating Finite Element software, it is now possible to calculate the
stresses of post-buckling for rectangular plates with any aspect ratio, any shapes
and sizes of cuts, and in different boundary conditions. Figure 2 illustrates the
applied boundary conditions and the applied pressure on the top and bottom edges.
Initially, a fundamental method is presented for computing the compressive buck-
ling load of a simply supported elastic rectangular plate having a central circular
hole.
Subsequently, a three-dimensional Finite Element model is developed to com-
prehensively compute the critical buckling load in a rectangular plate having a
central circular hole. The developed models will be extended to simulate the post-
buckling analysis in the future research reports. An energy method to determine the
buckling load of r rectangular plates of constant thickness under compressive loads
is developed by Timoshenko [17]. A review of this derivation shows that the
method is also applicable to plates of variable thickness. In this case, Timoshenko’s
integrals I1, I2 are [17]:
ðð (� �2 " � 2 �2 #)
∂2 ω ∂2 ω ∂2 ω ∂2 ω ∂ω
I1 ¼ D þ � 2 � ð1 � μ Þ � � � dxdy (1)
∂x2 ∂y2 ∂x2 ∂y2 ∂x∂y
Surface
ðð " � �2 � �2 #
σx ∂ω σy ∂ω τxy ∂ω ∂ω
I2 ¼ h � þ � þ2� � � dxdy (2)
S ∂x S ∂y S ∂x ∂y
Surface

where.
h : is the plate thickness (function of x and y), x, y : is the rectangular coordinates
with origin at centre of plate and x-axis in direction of load, and ω : is the lateral
deflection of plate.

E � h3
D¼ (3)
12 � ð1 � μ2 Þ

39
Computational Models in Engineering

Figure 2.
A simply supported plate subjected to compression.

where D : is the flexural rigidity of plate (function of x and y); μ ¼ 0:3 : is the
Poisson’s ratio; σ x : is the tensile stress in x direction; σ y : is the tensile stress in y
direction; τxy : is the shear stress; and S : is the tensile stress in x direction far from
hole.
This study investigates the effect of holes on the critical buckling of the web
plates of studs subjected to pure compressive load along the length of the plate. The
web of the studs is considered to be simply supported along the edges that intersect
with the flanges. The buckling analysis is performed using the commercially avail-
able finite element software ABAQUS/Standard. In the current research, the general
purpose of three-dimensional, stress/displacement, reduced integration with hour-
glass control, shell element S4R (available in the ABAQUS/Standard element
library), is considered to model the plates. S4R is involved in four nodes (quadri-
lateral), with all six active degrees of freedom per node. S4R allows transverse shear
deformation, and the transverse shear becomes very small as the shell thickness
decreases.

4.1 Simulation of the behaviour of the steel mesh under dynamic

impact loading

A same trend to simulated dynamic behaviour of the steel mesh due the applied
dynamic loading was taken into account. Thus, the mesh steel reinforcement sizes
20 mm diameter which they arranged by 100 mm distance centre to centre of the
steel bars were tested under free fall of the dropped hammer. The same dropped
hammer, which was used in the last section, was considered for the current simula-
tion. It is indicated that a 110 kg hammer was dropped at a velocity of 0.2 m/s on
top of the steel reinforcement.

40
A New Concept to Numerically Evaluate the Performance of Yielding Support under Impulsive…
DOI: http://dx.doi.org/10.5772/intechopen.79643

Figures 3 and 4 illustrates the simulated experimental set up which was used to
simulate the structural behaviour of the steel mesh under impact loading. It should
be noted that steel mesh can play a significant role as a part of the yielding support
in a coal mine, as it can mitigate the effect of the destructive released kinetic energy
due to a possible coal burst. In the coal mines, it was observed that both rock bolt
and cable bolt might be losing the initial bond stiffness at the early stages of the
applied dynamic loading due to the failure and separation of the anchored zone in
the cable and rock bolt inside embedded coal. The anchorage length in a post-
tensioned member and the magnitude of the transverse forces (both tensile and
compressive), that act perpendicular to the longitudinal prestressing force, depend
on the magnitude of the prestressing force and on the size and position of the

Figure 3.
The testing set up for simulating the behaviour of the steel mesh under impact loading.

Figure 4.
Simulation of the behaviour of the steel mesh under dynamic impact loading.

41
Computational Models in Engineering

anchorage hooks. Both single and multiple anchorages are commonly used in the
coal mining. Prestressing force anchors transfer large forces to the coal in concen-
trated areas. Furthermore, coal is a very brittle material. This can cause localised
bearing failure or split open the end of members. Thus, the steel mesh can consid-
erably reduce the effect of the induced dynamic loading due to the coal burst. In the
current simulation, the tensile stress for the steel mesh was fy = 500 MPa and the
ultimate stress for the steel mesh fu = 700 MPa was taken into account. The post-
failure of the steel mesh which may result in the rupturing of the steel bars was also
defined. The ductile damage function was determined to simulate the post-failure
of the steel mesh. Also, the rupturing strain εrupture = 0.3% was assumed. Weld
properties of the steel mesh can also influence the overall deformation and energy
absorption of the yielding support.

4.2 Simulation under dynamic shear loading

After calibrating the numerical models under static loading, the structural
behaviour of the simulated models under dynamic loading was also studied. Since
preparing the laboratory experiments to simulate the behaviour of cable bolts under
dynamic loading is demanding, a validated and novel numerical simulation was
developed. In order to simulate the behaviour of the cable bolts under impact
loading, a 110 kg mass at velocity of 0.2 m/s was dropped on top of the concrete
blocks. Figure 5 presents the structural behaviour of the cable bolts under impact
loading. As illustrated, the momentum energy from the dropped mass would ini-
tially be transferred to the concrete surfaces. The transmitted energy due to the
impulsive loading will reach the cable bolt. Figure 5 demonstrates the failure
process of the cable bolt under the impact loading starting with the initial
deformed shape followed by the brittle shear failure. The computation time was

Figure 5.
Cable bolt under impact loading.

42
A New Concept to Numerically Evaluate the Performance of Yielding Support under Impulsive…
DOI: http://dx.doi.org/10.5772/intechopen.79643

around t = 5e-3 seconds, which is very short to simulate the effect of the impulsive
loading.

5. Conclusion

A Finite Element model to predict the structural behaviour of cable bolts under
both static and dynamic loading was presented. The models were initially calibrated
against the experiments available in the literature. By taking the advantage of
numerical modelling, some of the localised structural performance of cable bolts
under both static and dynamic loading can be observed. By computing the area
under curves of the load–displacement curves under both static and dynamic load-
ing, the amount of the absorbed energy in each cable bolt can be estimated. The
results indicate that numerical modelling can be used to assess the rock bolt behav-
iour under dynamic loading for designing the ground support in burst-prone con-
ditions. Numerical techniques can predict an acceptable estimation of the reaction
of cable bolts under impact or impulse loading. The most recognised advantages of
numerical modelling are that following the initial calibration stage, variety of load-
ing conditions can be modelled and the behaviour of bolts can be studied in greater
detail compared with laboratory tests. The current simulations can be taken into
account as a cost-effective replacement instead of using drop test facilities. As
indicated, drop test facilities are very complicated and expensive. Also, it needs to
hire highly skilful technical staffs to calibrate the measurement tools including data
acquisition system and the load cell. Subsequently, extracting the induced data after
performing the impact test, individually, when it comes to remove the effect of the
inertia forces, is one of the significant tasks. In the second section of the current
paper, extensive parametrical studies from the developed Finite Element models
will be undertaken. The main purpose of doing the following parametrical studies is
to determine realistic and technical design guidelines for the combined support
against impulsive dynamic loading.

Acknowledgements

This research was undertaken with the assistance of resources and services from
the National Computational Infrastructure (NCI), which is supported by the Aus-
tralian Government.

43
Computational Models in Engineering

Author details

Faham Tahmasebinia1,2,3*, Chengguo Zhang1, Ismet Canbulat1,

Samad M.E. Sepasgozar2, Onur Vardar1, Serkan Saydam1 and Chen Chen3

1 School of Minerals and Energy Resources Engineering, University of New South

Wales, Sydney, NSW, Australia

2 Faculty of Built Environment, The University of New South Wales, Sydney,

NSW, Australia

3 School of Civil Engineering, The University of Sydney, Sydney, NSW, Australia

*Address all correspondence to: faham.tahmasebinia@sydney.edu.au

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

44
A New Concept to Numerically Evaluate the Performance of Yielding Support under Impulsive…
DOI: http://dx.doi.org/10.5772/intechopen.79643

References

[1] Windsor C. Rock reinforcement and Geotechnical Engineering. 2012;

systems. International Journal of Rock 4(3):215-227
Mechanics and Mining Sciences. 1997;
34(6):919-951 [10] Ansell A. Dynamic testing of steel
for a new type of energy absorbing rock
[2] Stacey T. Review of membrane bolt. Journal of Constructional Steel
support mechanisms, loading Research. 2006;62(5):501-512
mechanisms, desired membrance
performance, and appropriate test [11] Plouffe M. Anderson T. Judge K.
methods. Journal of the Southern Rock bolts testing under dynamic
African Institute of Mining and conditions at CANMET-MMSL. In:
Metallurgy. 2001;101(7):343-351 Proceedings 6th international
symposium on ground support in
[3] Thompson A, Villaescusa E, Windsor mining and civil engineering
C. Ground support terminology and construction, Cape Town. S. Afr. Inst.
classification: An update. Geotechnical Min. Metall. Symposium Series S; 2008
and Geological Engineering. 2012;30(3):
553-580 [12] Li CC. A new energy-absorbing bolt
for rock support in high stress rock
[4] Bjurstrom S. Shear strength of hard masses. International Journal of Rock
rock joints reinforced by grouted Mechanics and Mining Sciences. 2010;
untensioned bolts. In: Proc. 3rd Cong. 47(3):396-404
ISRM, Denver. Vol. 2. 1974.
pp. 1194-1199 [13] Ghadimi M, Shahriar K, Jalalifar H.
A new analytical solution for the
[5] Haas C. Analysis of rock bolting to displacement of fully grouted rock bolt
prevent shear movement in fractured in rock joints and experimental and
ground. Mining Engineering. 1981; numerical verifications. Tunnelling and
33(6):698-704 Underground Space Technology. 2015;
50:143-151
[6] Egger P, Zabuski L. Behaviour of
rough bolted joints in direct shear tests. [14] Cai Y, Esaki T, Jiang Y. An analytical
In: 7th ISRM Congress. International
model to predict axial load in grouted
Society for Rock Mechanics and Rock
rock bolt for soft rock tunnelling.
Engineering; 1991
Tunnelling and Underground Space
Technology. 2004;19(6):607-618
[7] Ferrero AM. The shear strength of
reinforced rock joints. International
[15] Ma S, Nemcik J, Aziz N. An
Journal of Rock Mechanics and Mining
analytical model of fully grouted rock
Science and Geomechanics Abstracts.
bolts subjected to tensile load.
1995;35(7):889-905
Construction and Building Materials.
2013;49:519-526
[8] Hongwei S, Yanyan D, Jing Y.
Numerical simulation on bolted rock
[16] Aziz N, et al. Shear strength
joint shearing performance. Mining
properties of Hilti plain and indented
Science and Technology (China). 2010;
strand cable bolts. 2014
20(3):460-465

[9] Kaiser PK, Cai M. Design of rock [17] Tahmasebinia F, et al. Numerical
support system under rockburst and analytical simulation of the
condition. Journal of Rock Mechanics structural behaviour of fully grouted

45
Computational Models in Engineering

cable bolts under impulsive loading.

International Journal of Mining Science
and Technology. 2018;28(5):807-811

[18] Levy S, Woolley RM, Kroll WD.

Instability of simply supported square
plate with reinforced circular hole in
edge compression. Journal of Research.
1947;(39):571-577

[19] Kumai T. Elastic stability of the

square plate with a central circular hole
under edge thrust. In: Proc. Japan Nat.
Cong. Appl. Mech. 1951

[20] Schlack A. Elastic stability of

pierced square plates. Experimental
Mechanics. 1964;4(6):167-172

[21] Kawai T, Ohtsubo H. A method of

solution for the complicated buckling
problems of elastic plates with
combined use of Rayleigh-Ritz’s
procedure in the finite element method.
In: Proceedings of the 2nd Air Force
Conference on Matrix Methods in
Structural Mechanics. Wright-Patterson
Air Force Base OH; 1968

[22] Yu W-W, Davis CS. Cold-formed

steel members with perforated
elements. Journal of the Structural
Division. 1973;99(10):2061-2077

[23] Timoshenko SP, Gere JM. Theory of

Elastic Stability. Courier Corporation;
2009

46
Chapter 4

Criteria for Adequacy Estimation

of Mathematical Descriptions of
Physical Processes
Yuri Menshikov

Abstract

In this chapter, adequacy estimation criteria for mathematical descriptions in the

form of ordinary differential equations were proposed. Adequate mathematical
descriptions can increase the objectivity of the results of mathematical modeling for
future use. These descriptions make it possibly reasonable to use the results of
mathematical modeling to optimize and predict the behavior of physical processes.
Interrelations between criteria are considered. The proposed criteria are easily
transferred on mathematical descriptions in algebraic form.

Keywords: mathematical simulation, adequate descriptions, criteria of adequacy,

applications

1. Introduction

Mathematical modeling (simulation) of physical processes is an important tool

for the study of the environment.
Mathematical modeling is a means of studying the real objects, processes, or
systems by replacing the real objects on the mathematical models, which are more
comfortable to study with the aid of computers.
The mathematical model is an approximate representation of real-world objects,
processes, and systems, expressed in mathematical terms. In this case, a significant
feature of original are saved from researcher’s point of view.
First of all, some definitions and some concepts are given for the convenience of
exposition.
A mathematical model of a real object will be called as mathematical dependen-
cies and connections between the elements of a mathematical model. These ele-
ments are chosen on the basis of the interests of the researcher himself and the
ultimate goals of study of the object. Usually, dependencies and relationships have
forms of differential equations, integral equations, algebraic connections, etc.
The functions of external influences and external loads that are present in the
mathematical model of the object in the form of symbols will be called as models of
external loads.
The initial conditions, boundary conditions, and other conditions for the math-
ematical model will be called as additional conditions.
The totality of the mathematical model of the object, models of external influences,
and additional conditions will be called as a mathematical description of the object.

47
Computational Models in Engineering

The study of the behavior of the mathematical model of an object under the
influence of models of external loads and additional conditions will be called as
mathematical modeling or mathematical simulation.
The practical significance of the results of mathematical modeling or simulation
of physical processes depends on the degree of coincidence of the results of mathe-
matical modeling of the selected mathematical description of the real process with
experimental data [1]. Such property of a mathematical description of a physical
process is usually called adequacy.

1.1 Preliminary definition

A mathematical description will be called as an adequate mathematical description

(AMD) of the process under study if the results of mathematical modeling (simu-
lation) using this description coincide with experimental data with the accuracy of
experimental measurements.
The definition of adequacy will be clarified later for some types of mathematical
models.
If the coincidence of the results of mathematical modeling with experiment is
bad, then further use of these mathematical descriptions is problematic.
It is note that authors of works on mathematical modeling concern seldom
questions of adequacy of the constructed mathematical description of process to
real measurements [2–5]. Sometimes, such adequacy is proved by the real facts;
sometimes, authors refer to results of other authors; and sometimes, there have not
been any arguments.
The considered situation requires formation of some uniform approach to this
problem, common methodological approach, general algorithms, and common
criteria of estimation of adequacy degree.

• Currently, there are two main approaches to the problem of constructing an

adequate mathematical description [1, 6–8]: for a mathematical model with a
priori chosen structure and inaccurate parameters, a model of external
influence is determined, which together with the mathematical model of the
process provide the adequacy condition (coincidence with experiment);

• a model of external loads is given a priori and then parameters of mathematical

model or of its structure are selected, in such a way that results of
mathematical simulation match up with experiment.

Having a comparison of the results of mathematical modeling with experimental

data in definition of adequate mathematical description ensures the objectivity of
the results of the synthesis of a mathematical description. In the literature, this
approach is called as an identification method: estimation of the parameters of an
adequate mathematical description based on the results of measurements of the
characteristics of the state of the physical process [9, 10].
Mathematical models of physical processes can be presented as systems of
ordinary differential equations, systems of partial differential equations, algebraic
relations, integral equations, etc.
It should be noted that in many works, the accuracy of the results of mathemat-
ical modeling is several times lower than the accuracy of experimental data.
In the given work, the mathematical models of physical processes described only
by the system of the ordinary differential equations will be examined [2, 3]. Such
idealization of real processes or dynamic systems is widely used in various areas for
the description of control systems [11], as well as of mechanical systems with the

48
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

concentrated parameters [5, 12], economic processes [13], biological processes [14],
ecological processes [15], etc. In some works with the help of such systems, human
emotions are simulated [16].
Many problems investigated in the given work, have place for other types of
mathematical models of physical processes, for example, for mathematical models
in the form of the partial differential equations.
The chapter proposes several criteria for checking the adequacy of the
constructed mathematical descriptions for cases when the mathematical model of
the physical process is represented by a system of differential equations.
The author hopes that the offered criteria of adequacy will be useful in a con-
struction of the adequate mathematical descriptions of real physical processes.

2. Criteria of adequacy mathematical description of quantitative type

Consider the specified criteria for mathematical descriptions in the form of a

system of differential equations.
For simplicity, we select the physical processes with mathematical models in the
form of linear system of ordinary differential equations [17–19]:

x_ ðtÞ ¼ C xðtÞ þ D zðtÞ, (1)

where C, D—matrices with constant coefficients, which are given approxi-

mately, x ¼ ðx1 ; x2 ; …xn ÞT —vector-function variables, characterized the state of
process (ð�ÞT —a mark of transposition), zðtÞ ¼ ðz1 ðtÞ; z2 ðtÞ; …zm ðtÞÞT —vector-
function of external load; and x ∈ X, z ∈ Z, X, Z are normalized functional spaces.
We assume that state variables xi ðtÞ, 1 ≤ i ≤ n of system (1) correspond to some
real characteristics of process which is under investigation x ~i ðtÞ, 1 ≤ i ≤ n.
By mathematical description of the physical process, we mean the set of the
system of Eq. (1), the vector of the external loads functions zðtÞ ¼ ðz1 ðtÞ;
z2 ðtÞ; … ; zm ðtÞÞT and the initial conditions xðt0 Þ ¼ x0 . In other words, a mathemati-
cal description is a collection of mathematical models, models of external influences,
and initial conditions.
The process of solving the system of differential Eq. (1) under the influence of
selected models of external loads zðtÞ ¼ ðz1 ðtÞ; z2 ðtÞ; … ; zm ðtÞÞT , taking into account,
the initial conditions xðt0 Þ ¼ x0 , is usually called mathematical modeling or mathe-
matical simulation.
An adequate mathematical description of a physical process of such type with
respect to all variables x1 ðtÞ, x2 ðtÞ, :…, xn ðtÞ of quantitative type will be called as the
mathematical description for which the results of mathematical simulation of vari-
ables x1 ðtÞ, x2 ðtÞ, :…, xn ðtÞ coincide with the results of experimental measurements
xex ex ex
1 ðtÞ, x2 ðtÞ, :…, xn ðtÞ of the characteristic x1 ðtÞ, x2 ðtÞ, :…, xn ðtÞ with the accuracy of
the experiments δ1 , δ2 , : :, δn :
 
 xi ðtÞ � xex ðtÞ  ≤ δi , 1 ≤ i ≤ n: (2)
i X

In practice, the measurement of the characteristics of state variables is limited to

only one or two components. We formulate a refined definition of the adequacy of a
mathematical description for the case of a single variable.
An adequate mathematical description of a physical process of such type with
respect to the variable xk ðtÞ, 1 ≤ k ≤ n (ALMDxk ) of quantitative type will be called
as the mathematical description for which the results of mathematical simulation of

49
Computational Models in Engineering

a variable xk ðtÞ coincide with the results of experimental measurements xex

k ðtÞ of the
characteristic xk ðtÞ with the accuracy of the experiment δk :
 
 xk ðtÞ � xex ðtÞ  ≤ δk : (3)
k X

For the rest of the variables, coincidence with experiment is not determined.
Adequate mathematical descriptions are similarly determined in the case of several
measurements of state variables. The metrics of comparison in this case is deter-
mined by the objectives of specific studies.
In [12, 20], which were considered before, coincidence with experiment is 10
times below the accuracy of experiment.
The criteria of mathematical description adequacy of quantitative type, which are
offered in the given chapter, can be used for other types of mathematical descriptions
of physical processes, for example, for mathematical descriptions in the form of the
partial differential equations [21]. They have many common features.
It can be shown that there are an infinite set adequate mathematical descriptions
for the same physical experiment.
In addition, qualitatively, different physical processes can have adequate math-
ematical descriptions for the same experiment.
There exist two approaches to problem of construction of adequate mathemati-
cal description of quantitative type [22, 23]:

1. Mathematical model of process of type (1) is given a priori with inexact

parameters and then the models of external loads were determined for which
the results of simulation coincide with experiment [22, 23];

2. Some models of external loads are given a priori and then mathematical model
of process of type (1) is chosen for which the results of simulation coincide
with experiment [6–8].

Now, we will consider the synthesis of adequate mathematical description of

quantitative type in the frame of first approach analyzing the process with the
concentrated parameters, for which the motion is described by ordinary differential
equations of n-order (1).
We assume that some functions of state x1 ðtÞ, x2 ðtÞ, …xr ðtÞ, r ≤ n in system (1)
are obtained from experiment and presented by graphs. Besides, we suppose that
some functions of external loads, for example, z1 ðtÞ, z2 ðtÞ, …zl ðtÞ, l ≤ m are
unknown. According to first approach, it is necessary to develop the construction of
such model of external load component, which is characterized by the functions of
state x1 ðtÞ, x2 ðtÞ, …xr ðtÞ of mathematical model (1), and will coincide with experi-
mental measurements x ~1 ðtÞ, x
~2 ðtÞ, …~
xr ðtÞ with inaccuracy of initial data. Such
mathematical model of process behavior together with obtained model of external
load can be considered as adequate mathematical description of quantitative type of
process.
Such method of obtaining of mathematical models of external loads (functions
z1 ðtÞ, z2 ðtÞ, …zl ðtÞ, l ≤ m) is determined in literature as a method of identification
[9, 10]. By the way, physical reasons of occurrence of such external loads are not
being taken into account. They are only functions, which in combination with
mathematical model (1) provide results of modeling, which coincide with experi-
ment with the given accuracy.
Consider an example of a mathematical description that satisfies the criterion of
the adequacy of a quantitative type for all variables x1 ðtÞ, x2 ðtÞ, :…, xn ðtÞ.

50
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

Figure 1.
Kinematic scheme of the main mechanical line of rolling mill.

2.1 Vibrations in the main mechanical line of the rolling mill

Now, we consider in detail, the problem in which the dynamics of the main
mechanical lines of rolling mills is investigated [24, 25]. One variant of the kine-
matic scheme of it is presented in Figure 1. (a) where the engine is marked by label
(1), the coupling is marked by label (2), gears is marked by label (3), driving shafts
is marked by label (4), operational barrels is marked by label (5).
The four-mass model with weightless elastic connections is chosen as mathe-
matical model of dynamic system of the main mechanical line of the rolling mill
[24, 25]. The system of vibrations equations is obtained from the Lagrang’s
equations of second kind:

€ 12 þ ω2 M12 � c12 M23 � c12 M24 ¼ c12 Meng ðtÞ;

M 12
ϑ2 ϑ2 ϑ1

€ 23 þ ω2 M23 � c23 M12 þ c23 M24 ¼ c23 MU rol ðtÞ;

M (4)
23
ϑ2 ϑ2 ϑ3

€ 24 þ ω2 M24 � c24 M12 þ c24 M23 ¼ c24 ML rol ðtÞ:

M 24
ϑ2 ϑ4 ϑ4

Here, the following designations were accepted: Meng —moment of engine,

ϑi —moments of inertia of the concentrated weights, cik —rigidity of the appropriate
elastic connection, MU L
rol , Mrol —moments of technological resistance put to the upper
and lower operational barrels accordingly, and Mik —moments of elasticity forces,
which are applied to shafts between mass ϑi and ϑk ; ω2 ik ¼ ðϑi ϑk Þ � 1 cik ðϑi þ ϑk Þ:

51
Computational Models in Engineering

Actually, the constructed mathematical model may correspond to real process

and may not. It is necessary to check up correctness of the constructed mathemat-
ical model. For this purpose, the data of experiment are used. If the results of
mathematical modeling coincide with results of experiment (with accuracy of mea-
surements), then mathematical description of process is considered as adequate to a
reality in the quantitative sense. In other words, the mathematical description
corresponds to real process.
The information related to the real motion of the main mechanical line of rolling
mill was obtained by an experimental way [23, 24, 26]. Such information is being
understood as availability of functions M12 ðtÞ, M23 ðtÞ, M24 ðtÞ. The records of func-
tions M12 ðtÞ, M23 ðtÞ, M24 ðtÞ of a given process are shown in Figure 2.
It is obvious, that the results of mathematical modeling of system (4) depend
directly on character of change of external loads, which is applied to operational
barrels of the rolling mill and external impact of the engine Meng , MU L
rol , Mrol . Some-
times, it is possible to pick up such loadings Meng , MU L
rol , Mrol in which the results of
mathematical modeling M12 ðtÞ, M23 ðtÞ, M24 ðtÞ coincide with experiment (Figure 2).
If such choice is possible, then mathematical model (4) combined with the
found loads Meng , MU L
rol , Mrol will give adequate mathematical description of real
process. It is necessary to note that in many papers analyzing the problem of
mathematical modeling with the use of system of a differential Eq. (4) together
with functions Meng , MU L
rol , Mrol are determined as mathematical model of process.
Coincidence is understood as coincidence with the accuracy of experimental
measurements.

Figure 2.
The records of functions M12 ðtÞ, M23 ðtÞ, M24 ðtÞ.

52
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

According to this approach, it is necessary to construct such models of external

loads MU L
rol , Mrol on system (4), for which the functions M12 ðtÞ, M23 ðtÞ, M24 ðtÞ of
elastic moments in the links of the model (solution of the system(4)), coincide with
the corresponding experimental functions of the moments of elastic forces in the
links of the main line of the rolling mill (Figure 2).
Consider the construction of an adequate mathematical description within the
framework of the first approach. To construct, for example, a model Murol , which
corresponds to the moment of the external load to the upper work roll, consider the
second equation in the system (4). The solution of this equation has the form
ðt
c23
_ �1
M23 ðtÞ ¼ M23 ð0Þ cos ω23 t þ M 23 ð0Þ ω23 sin ω23 t þ Murol ðτÞ sin ω23 ðt � τÞ dτ
ϑ3 ω23
0

ðt
�1
þc23 ðϑ2 ω23 Þ ½M12 ðτÞ � M24 ðτÞ� sin ω23 ðt � τÞ dτ
0

or
ðt
sin ω23 ðt � τÞ Murol ðτÞ dτ ¼ F ðtÞ, (5)
0

where
� � ��
FðtÞ ¼ ϑ3 ω23 ðc23 Þ � 1 M23 ðtÞ � M23 ð0Þ cos ω23 t þ M
_ 23 ð0Þ ω � 1 sin ω23 t
23

ðt
�1
� ϑ3 ðϑ2 Þ ½M12 ðτÞ � M24 ðτÞ� sin ω23 ðt � τÞ dτ:
0

We will assume that function F ðtÞ in (5) belongs to the normalized space L2 ½0; T �
(½0; T � is the period of time at which the function Murol is studied) and the solution
Murol of Eq. (5) belongs to the normalized functional space C½0; T �.
Let us rewrite the equation (5) in the more compact form

A p z ¼ uδ , (6)

where z is the searched element, uδ is the given element, which belong, respec-
tively, to the functional spaces C½0; T � and L2 ½0; T �, Ap is the integral operator. In
this case, z ¼ Murol ðtÞ, uδ ¼ F ðtÞ.
Since the right-hand side F ðtÞ of the integral Eq. (5) is determined from the
experiment, it is natural to assume that instead of the exact right-hand side
uex ¼ Fex ðtÞ of the Eq. (6), some approximation of it is given uδ ¼ F ðtÞ:
k uδ � uT kL2 ½0;T � ¼ k F ðtÞ � FT ðtÞkL2 ½0;T � ≤ δ, δ is given.
The set of possible solutions Q δ ⊂ C½0; T � of Eq. (6) consists of elements that
correspond to the equation with given accuracy:
n � � o
Q δ ¼ z : � Ap z � uδ �L2 ½o;T� ≤ δ :

Each function in the set Q δ together with the given mathematical model (6)
provides an adequate mathematical description of the physical process.
In this case, the problem of identifying model of external load in the rolling mill
is considered as the inverse of the synthesis problem [17].

53
Computational Models in Engineering

Figure 3.
The results of identification (functions MU L
rol ðtÞ, Mrol ðtÞ) for case of rolling that were shown on Figure 2.

In this chapter, oscillograms of the moments of the forces of elasticity in the

links of the main line of rolling mill 1150, obtained in [24, 25], are used. A copy of
this oscillogram is shown in Figure 2. The value of T is chosen equal to 0.48 s.
When synthesizing the model of external load on the lower work roll of the
state, it is necessary to use the last differential equation in the system (4).
In Figure 3, the graphs of the models of external loads on the upper and lower
working rolls for rolling case shown on Figure 2.
Thus, models of external loads MU L
rol ðtÞ, Mrol ðtÞ were obtained that together with
the mathematical model (4) and the initial conditions yield simulation results that
coincide with the experimental measurements with the accuracy of the experiment.
In other words, the mathematical model (4), models of external loads and initial
conditions give an adequate mathematical description of quantitative type for all vari-
ables of physical process.
Now, we consider another example of astrodynamical processes mathematical
description, which has the property of adequacy of quantitative type in only one variable.
Based on theoretical analysis of mathematical vortex model of planetary sys-
tems, the analytical expression for planetary distances in the prevailing planetary
systems was obtained. These distances are functions of the coordinates of the
centers of vortical rings of primary planetary vortex. Comparison of theoretical and
real distances planets of the Solar system show their good agreement.
Known in the cosmogonic theories of the solar system, the law of Titsius-Bode
(1772) of planetary distances rn

rn ¼ 0, 4 þ 0, 3 � 2n а:о: (7)

is a successful empirical approximation of the real sequence of distances rn of

planets with number n from to Sun. In this case, the first planet (Mercury)

54
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

corresponds to the value n ! � ∞, Venus—n ¼ 0, the Earth—n ¼ 1, etc., and the

conditional unformed planet between Mars and Jupiter must be attributed to the
value of n ¼ 3. Despite the excellent conformance of this law to the average number
of planets, the law (7) for the first and distant planets of Neptune and Pluto is not
fulfilled [27].
In the twentieth century, some attempts were made [28, 29] to theoretically
obtain the law of planetary distances, but in the basis of these theories, the authors
had to impose new arbitrary hypotheses. Schmidt [28] introduces a hypothetical
function of the distribution of kinetic moments in the masses of the primary nebula,
and for the simplest functions it receives a quadratic law, a geometric progression,
and others. Kuiper [29] deduces his law on the basis of the theory of tidal stability
using the concept of “critical density of Rosh.” However, the law it received give the
distance between planets, which differ on several orders from the real distance.
Below, based on the mathematical vortical model of the formation of planetary
systems [30, 31], the analytical law of planetary distances for any planetary systems
was obtained, which gives a good agreement with real distances in the solar system,
which has another form compared with (7).

2.2 The theory of planetary vortex

The general picture and the basic relations in the primary vortex explosion,
which creates stars and their planetary systems, is constructed in [30] on the basis
of a separate exact solution of the Euler hydrodynamic equations for spherical eddy
currents [30]. The main physical feature of this axisymmetric spatial flow, called
the planetary vortex [30], is the presence of a vortex dipole in the center of the
vortex dipole, which flows through a moving, twisted stream of outer space, and
the interaction of these motions generates vortical flow of a planetary vortex [30].
Using the method of integrating the complete nonlinear system of Euler’s
hydrodynamic equations was introduced and flow functions Ψð y; θÞ constructed.
The function of flow in spherical coordinates (r, θ, φ) [30] is constructed.
The planetary vortex described above as a complicated vortex flow is the initial
stage of the formation of a star planetary system from the primary nebula that has
fallen into the vortex region. Further prolonged evolution of this vortex to the state
of the planetary system is characterized by a variety of complex physical processes
such as: collision, accretion, accumulation of massive bodies, and their gravitation;
the formation of a massive star and its light and gravitational action; mutual grav-
itational and resonant influence of system structures, etc. [27].
Since the forces of gravitation, collision, and others acting between parts of a
single vortex are internal, they do not change its integral physical invariants.
In [30, 31], the modern planetary distances were calculated and they are shown
in the graphs (Figure 4). As we can see, the theoretical curve in the entire range of
distances is almost equidistant from the curve of real distributions of distances in
the solar system with deviations in both directions of the order of 20%.
Finally, the technique developed here for the calculation of the primary vortex
parameters can be applied in the reverse direction to determine according to the
data of several open planets of the main parameter of the planetary system and the
establishment of the method of work [38] of the complete structure of new
exoplanetary systems: the number of vortex planets, their distances from the star,
angular velocities, etc. This will give astronomers-observers reasoned data for the
search for new, yet open exoplanets in stellar planetary systems, which have already
opened 2–6 planets [32, 33].
Thus, the mathematical model of the process of formation of planetary systems,
which describes interplanetary distances well, is constructed.

55
Computational Models in Engineering

Figure 4.
Planetary distances in the solar system as functions of planetary numbers.

The development of this mathematical model does not take into account impor-
tant physical factors that have a significant effect on the behavior of interplanetary
matter. To such factors, it is necessary to attribute, first of all, gravitational inter-
action and heat flows. Because of this, one should not expect a good coincidence of
the real characteristics of the physical process with the results of mathematical
modeling. This situation occurs when the criterion of adequacy of the qualitative
type is not met (see Section 2). Consequently, the coincidence of the results of
mathematical modeling (with the adequate mathematical description constructed
only one or two of the system’s variables) with all the main characteristics of the
physical process is an exception, as a rule, and it can take place only if the adequacy
of a quality type is performed.

3. Criteria of adequacy of mathematical description of qualitative type

We will consider what prospects of adequate descriptions are valid for further
use and what goals should be selected as the creation of adequate mathematical
descriptions.
It will be useful to address to classical works in this area. In work [34], the
following statement was done: “…the imitation modeling is the creation of experi-
mental and applied methodology which aimed at the use of it for a prediction of the
future behavior of system.”
So, the adequate mathematical descriptions are intended for the forecast of
behavior of real process at first. It is possible with the aid of adequate mathematical
modeling to predict behavior of real process in new conditions of operation. For
example, it is possible to test more intensive mode of operations of the real machine
without risk of its destruction. Such tool (adequate mathematical description)
allows determining the optimum parameters of real process.
Let us now consider the conditions under which it is possible to further use ade-
quate mathematical descriptions for “…a prediction of the future behavior of system.”
Obviously, the structure of system (1), its parameters, and the specific type of
external influences are determined by the properties of a real physical process.
Let the selected structure of the mathematical model of the physical process
 T
include parameters p ¼ p1 ; p2 ; …; pk (e.g., the mass of the elements, the stiffness
of the elastic elements, etc.), which are reflecting the actual physical characteristics.

56
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

The structure of the mathematical model also includes dependencies that reflected
real physical patterns and dependencies of the process under study.
For the purpose of further substantiated use of mathematical descriptions, it is
necessary to require that there is a one-to-one correspondence between the compo-
nents of the vector parameter p of mathematical description and the actual physical
elements. In addition, it is necessary to require that the interconnections between
the parameters of a mathematical model comply with the physical laws of the
process being studied, and the main external loads had been included. This impor-
tant correspondence will be called the main correspondence (MC). The execution of
the MC believed the fulfillment of the criterion of adequacy of the qualitative type. In
other words, a mathematical description of a physical process satisfies the criterion
of adequacy of a qualitative type if the main correspondence is fulfilled.
An additional requirement for the implementation of the MC is explained firstly
by the fact that the quantitative agreement of the results of mathematical modeling
with a specific experiment is possible for mathematical descriptions of qualitatively
different physical processes due to the selection of parameters of mathematical
descriptions.
The implementation of the MC leads to the fact that the models of external
influences obtained by the method of identification will correspond to the real
external influences on the physical process. At least, these models will not contra-
dict the physical meaning. If we return to the example of the synthesis of an
adequate mathematical description of the process of mechanical oscillations in the
main line of the rolling mill, then it can be argued that the MC is being executed. By
virtue of this, the obtained models of external influences have a reasonable physical
interpretation (do not contradict the physical meaning). The external load smoothly
increases from zero to a steady-state value (see Figure 3).
When fulfilling the adequacy of a mathematical description of a qualitative type,
it becomes possible to argue that a mathematical description that satisfies two
criteria of adequacy will retain its useful properties for other experiments in the
future under small changes in the conditions of the physical process. In other words,
this description can be used for “a prediction of the future behavior of system.” An
example of such a successful application could be further mathematical modeling
using an adequate mathematical description for the rolling mill [23].
In the second example, given in Section 1, the main correspondence is not
fulfilled, and therefore, the application of the obtained results in the new conditions
will not be justified.
The algorithm for constructing an adequate mathematical description of a qual-
itative type cannot be formalized, as in the case of the adequacy of a mathematical
description of a quantitative type. The process of constructing such a description
mainly depends on subjective factors, such as the scientific tasks of studying the
physical process using mathematical modeling methods.
In some cases, it is impossible to perform a check of mathematical description
adequacy of a quantitative type due to lack of experimental data in principle. Let us
give an example of a mathematical description that satisfies only the criterion of
adequacy of a qualitative type.
An important and relatively new field of applications of methods of mathemat-
ical modeling is a tectonic processes study [35]. This work presents a complete
algorithm required for the successful application of mathematical modeling
methods of operations, which does not include the verification operation. Consider
this algorithm in more detail.
It is widely known that earthquakes predicting is challenging and unsolved still
(but it can be happen in future). Even where earthquakes have unambiguously

57
Computational Models in Engineering

occurred within the parameters of a prediction, statistical analysis has generally

shown these to be not better than lucky guesses. Now, there are hundreds of well-
known earthquakes precursors and a number of theories to explain their origin.
However, the problem of earthquake prediction in many of its aspects still remains
open.
Utsu studied theoretically the relation between the size of aftershock activity
and the magnitude of the main shock [36]. Independence of the occurrence of main
shocks has been assumed in many models, some chapters discuss the migration of
large earthquakes and casual relationship between seismic activities in different
geophysical regions. Trigger models assume a series of primary events (main
shocks) distributed completely random in time. Each of these primary events may
generate secondary series of events. Epidemic-type model can be considered as
birth and death process.
Utsu proposes a new model that takes into account the influence of strain
solitary waves as a “trigger” of some shocks and appropriate methods of forecasting.
Authors analyze the 2011 Japan earthquake. These studies show that solitary waves
can be generated as aftershocks hypocenters at the Moho surface.
Significant amount of works have been devoted to research of solitons in solids.
For example, considering structural-phenomenological approach one distinguishes
damaged environment with microstructure, Kosser’s continuum with limited traf-
fic, Leru’s continuum, environment with deformities [37], grainy environment,
which has the soliton solutions of motion equations. It is known that the pulse
perturbations in rocks are different from seismic waves of harmonic type.
Some researchers [36–38] have considered generalizations of the traveling wave
solutions. The necessary and sufficient conditions for the existence of appropriate
solutions were obtained.
Let us consider an anisotropic elastic medium. By [39], it follows that relations
between stress and strain in Hooke’s law contain 21 free coefficients. The system of
motion equations in this case has the form:

∂2 u
ρ ¼ ððc11 ; c66 ; c55 ; 2c16 ; 2c15 ; 2c56 Þ; ΘuÞ
∂t2
(8)
þ ððc16 ; c26 ; c45 ; c12 þ c66 ; c14 þ c6 ; c46 þ c25 Þ; ΘvÞ

þ ððc15 ; c46 ; c35 ; c14 þ c56 ; c13 þ c55 ; c36 þ c45 Þ; ΘwÞ,
∂2 v
ρ ¼ ððc16 ; c26 ; c45 ; c66 þ c12 ; c56 þ c14 ; c25 þ c46 Þ; ΘuÞ
∂t2
(9)
þ ððc66 ; c22 ; c44 ; 2c26 ; 2c46 ; 2c24 Þ; ΘvÞ

þ ððc56 ; c24 ; c34 ; c46 þ c25 ; c36 þ c45 ; c23 þ c44 Þ; ΘwÞ,
∂2 w
ρ ¼ ððc15 ; c46 ; c35 ; c14 þ c56 ; c13 þ c55 ; c45 þ c36 Þ; ΘuÞ
∂t2
(10)
þ ððc56 , c24 , c34 , c46 þ c25 , c45 þ c36 , c23 þ c44 Þ, ΘvÞ

þ ððc55 ; c44 ; c33 ; 2c45 ; 2c35 ; 2c34 Þ; ΘwÞ,

where u, v, and w are the displacements along the respective axes of a Cartesian
coordinate
 2system, ρðx1 ; x2 ; x3 Þ is the density, 
∂2 ∂2 2 2 2  
Θ ¼ ∂x1 2 ; ∂x2 2 ; ∂x3 2 ; ∂x∂2 ∂x1 ; ∂x∂3 ∂x1 ; ∂x∂2 ∂x3 , C ¼ cij i, j¼1, 6 is matrix of elastic
∂

constants.

58
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

There, the following concept was introduced. Authors found the solutions of
(8)–(10) as follows:

ðu; v; wÞT ¼ ðψ u ðtÞ; ψ v ðtÞ; ψ w ðtÞÞT W ðx; y; z; tÞ, (11)

� �
where W ð:Þ ¼ exp � gðx �ε1x ðtÞÞ � gðy �ε2y ðtÞÞ � gðz �ε3z ðtÞÞ , g ð:Þ ∈ G is the class of
~ ~ ~

positive-definite, unimodal, twice continuously differentiated functions with a mini-

mum at zero, and for which the second derivative is not a constant, ψ u ðtÞ, ψ v ðtÞ,
ψ w ðtÞ are the functions which determined the amplitude of the relevant
perturbations, ε1 , ε2 , ε3 are constants which determined the localization
perturbations, x ~ ðtÞ, ~y ðtÞ, ~z ðtÞ are the functions which determined the trajectory of
solitary waves.
In [39], authors got the necessary and sufficient conditions of existent solutions
of the system (8)–(10) in the form (11). In [40], the various crystal systems for the
existence of the appropriate type of motion equations solutions are studied.
The main hypothesis of method of earthquakes forecast is that a single shock
causes the appearance of one or more solitary waves that move from the hypocenter
of the earthquake. Each wave passing through the zone of accumulation of seismic
energy, causing a new earthquake can in turn generate new solitons. The method of
prediction involves the separation from the total population of earthquake sub-
sequences caused by the same soliton and the construction of a hypothetical trajec-
tory of the solitons. Knowing the distance between individual impulses along the
trajectory of the soliton can estimate its speed. Knowing some point of its trajectory,
it is possible to make an assessment of the trajectory. With the rate and trajectory of
each soliton, one can estimate its position at any time. Having information about the
position of the soliton at some time can determine the “soliton component” of shock
probability at this time.
As initial data, we consider a sequence of the form: ðx1 ; t1 ; μ1 Þ, ðx2 ; t2 ; μ2 Þ, …,
ðxk ; tk ; μk Þ, where x1 , x2 , …, xk are earthquake hypocenters, μ1 , μ2 , …, μk are magni-
tudes, and t1 , t2 , …, tk are times of shocks.
Let a trajectory of the soliton be described parametrically: xðtÞ ¼ ðx1 ðtÞ; x2 ðtÞÞ,
vðtÞ is speed. Then, we have:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
dxðtÞ
vðtÞ ¼ ¼ ðx1 0 ðtÞÞ2 þ ðx2 0 ðtÞÞ2 ,
dt
tð
iþ1 t ð qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
iþ1

vðtÞdt ¼ ðx1 0 ðtÞÞ2 þ ðx2 0 ðtÞÞ2 dt ¼ lðxðti Þ; xðtiþ1 ÞÞ,

ti ti

tiþ1
ð
lðxðti Þ; xðtiþ1 ÞÞ
vi ¼ ¼ vðtÞdt=ðtiþ1 � ti Þ,
tiþ1 � ti
ti

where lðxðti Þ, xðtiþ1 Þ is the length of the curve corresponding to time interval
ðti ; tiþ1 Þ.
Assume that the speed of the soliton is monotone—decreasing function. If the
motion occur in the region of constant density, the ratio will be implemented:
v1 ¼ v2 ¼ … ¼ vk � 1 . In the field of variable density: v1 > v2 > … > vk � 1 .
x3 � x2
Then, we consider the approximate speed ~v1 ¼ xt22 � x1
� t1 , v2 ¼ t3 � t2 , …,
~
~vk � 1 ¼ xtkk � xk � 1
� tk � 1 :
If a density is constant, then ~v1 þ ε1 ¼ ~v2 þ ε2 ¼ …~vk � 1 þ εk � 1 , where
ε1 , ε2 , …εk � 1 is the curvature parameters. If a density is variable, then:

59
Computational Models in Engineering

~v1 > ~v2 > … > ~vk  1 . This is criteria for the identification a subsequence of individual
solitons trajectory.
In Figure 5, the results of seismic process analysis that occurred on the Japanese
islands for 3 days before the earthquake of magnitude 8.9 (occurred on March 11,
2011) are shown. Here, the numbers from 0 to 12 mark epicenters of the foreshock,
the epicenter of the main shock indicates maximum circle radius (it is near to the
epicenter of the foreshock number 1). Curves and straight lines marked the soliton
trajectory, which are calculated using a special software. The calculations take into
account the hypothetical rate of the solitary waves and their possible reflection from
the areas with a high density of rocks.
As you can see, the foreshock is arranged so that a large number of possible
waves pass through the region, where there was a maximum magnitude shock.
Clearly, traced kind of a soliton with a focusing effect is at the point where there
was the main shock.
Thus, here it proposed the mathematical model of the process of earthquake
sequences, taking into account the impact of slow solitary wave soliton type as a
“trigger” to some shocks. The proposed theory allows us to construct forecasts when
geophysics seismic process is similar to that which occurred on Japanese islands in
2011.
The considered example of a mathematical description cannot be checked for
the adequacy of a quantitative type due to objective reasons. But this description
meets the criterion of adequacy of the qualitative type and so the results of mathe-
matical modeling do not contradict the physical meaning.
Further, consider an example of a mathematical description, the adequacy of
which cannot be fundamentally assessed.
Methods of mathematical modeling penetrate recently into many nontraditional
areas of human activity such as the study and modeling of emotions [16]. Let us
consider in more detail the peculiarities of the application of methods of mathe-
matical modeling in this field.

Figure 5.
Results of seismic process analysis that occurred on the Japanese islands on March 11, 2011 for 3 days before the
earthquake of magnitude 8.9.

60
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

The mathematician Rinaldi investigated as first Petrarch’s emotional cycle and

established an ODE model, starting point for the investigations in two directions:
mapping the mathematical model to a suitable modeling concept, and trying to
extend the model for love dynamics in modern times.
A control-oriented approach observes emotions and inspiration as states fading
over time-behaving like a transfer function approaching a steady state. This obser-
vation suggests a modeling approach by transfer functions. Both model approaches
allow an easy extension to modern times.
In literature, two special contributions can be found:

• Love affairs and differential equations by Strogatz [41], —harmonic oscillators

making reference to Romeo and Juliet;

• Laura and Petrarch: an Intriguing Case of Cyclical Love Dynamics by Rinaldi

[42]—presenting a nonlinear ODE with cyclic solutions.

Both contributions start directly with nonlinear oscillations, observing a certain

historic emotional behavior of prominent couples. Laura group at Vienna Univer-
sity of Technology tries to consider general modeling concepts for emotional rela-
tions, which cover or coincide with Petrarch’s emotional cycle, in case of
appropriate parameterization.
Following a suggestion of Strogatz [41] here examines a sequence of dynamical
models involving coupled ordinary differential equations describing the time-
variation of the love or hate displayed by individuals in a romantic relationship. The
models start with a linear system of two individuals and advance to love triangles,
and finally to include the effect of nonlinearities, which are shown to produce chaos.
An obvious difficulty in any model of love is defining what is meant by love and
quantifying it in some meaningful way.

3.1 Simple linear model

Strogatz [40] considers a love affair between Romeo and Juliet, where R(t) is
Romeo’s love (or hate if negative) for Juliet at time t and J(t) is Juliet’s love for
Romeo.
The simplest model is linear with

dRðtÞ
¼ a RðtÞ þ b J ðtÞ,
dt
dJ ðtÞ
¼ c RðtÞ þ d J ðtÞ, (12)
dt
where a and b specify Romeo’s “romantic style,” and c and ad specify Juliet’s
style. The parameter a describes the extent to which Romeo is encouraged by his
own feelings, and b is the extent to which he is encouraged by Juliet’s feelings. The
resulting dynamics are two-dimensional, governed by the initial conditions and the
four parameters, which may be positive or negative.
A similar linear model has been proposed by Rinaldi [42] in which a constant
term is added to each of the derivatives in (12) to account for the appeal (or
repulsion if negative) that each partner presents to the other in the absence of other
feelings. Such a model is more realistic since it allows feelings to grow from a state
of indifference and provides an equilibrium not characterized by complete apathy.
However, it does so at the expense of introducing two additional parameters. While
the existence of a non-apathetic equilibrium may be very important to the

61
Computational Models in Engineering

individuals involved, it does not alter the dynamics other than to move the origin of
the RJ state space.
Romeo can exhibit one of four romantic styles depending on the signs of a and b,
with names adapted from those suggested by Strogatz [40] and his students:

1. Eager beaver: a > 0, b > 0 (Romeo is encouraged by his own feelings as well as
Juliet’s).

2. Narcissistic nerd: a > 0, b < 0 (Romeo wants more of what he feels but retreats
from Juliet’s feelings).

3. Cautious (or secure) lover: a < 0, b > 0 (Romeo retreats from his own feelings
but is encouraged by Juliet’s).

4.Hermit: a < 0, b < 0 (Romeo retreats from his own feelings as well as Juliet’s).

Note that for the mathematical description (12), the criterion of the adequacy of
the quantitative type and the criterion of the adequacy of the qualitative type
cannot be checked due to the specificity of the process under study. Therefore, the
further use of the results of mathematical modeling is unreasonable. However, there
are no obstacles to nontraditional interpretations of the results of mathematical
modeling of emotional processes.

4. Conclusion

The proposed adequacy criteria for mathematical descriptions in the form of

ordinary differential equations make it possible to reasonably use the results of
mathematical modeling to optimize and predict the behavior of physical processes.
The proposed criteria are easily transferred on mathematical descriptions in
algebraic form [43].
Criteria for the adequacy of mathematical descriptions in the form of partial
differential equations are currently missing in the literature. However, some criteria
for the adequacy of mathematical descriptions can be transferred to the specified
descriptions. For example, the criterion of adequacy of the qualitative type can be
transferred almost unchanged.

Author details

Yuri Menshikov
University of Dnipro, Dnipro, Ukraine

*Address all correspondence to: menshikov2003@list.ru

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

62
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

References

[1] Menshikov Y. About adequacy of International Conference on Authomatical

mathematical modeling results. In: Control, Authomatics. Odessa, Ukraine:
Proceedings of the International ОNМА; 2008. pp. 774-777
Conference “Simulation-2008”; Kiev,
Ukraine. 2008. pp. 119-124 [9] Ikeda S, Migamoto S, Sawaragi Y.
Regularization method for identification
[2] Soeiro N et al. Experimental of distributed systems. In: Proceedings of
measurements and numerical modelling IY a Symposium IFAC on Identification
of vibroacoustics behaviour of a vehicle and Evaluation of Parameters of Systems.
gearbox. In: Proceedings of the Vol. 3. Tbilisi, Мoscow: USSR, Preprint;
International Eleventh International 1976. pp. 153-162
Congress on Sound and Vibration; St.
Petersburg, Russia. 2004. pp. 2821-2828 [10] Gelfandbein J, Kolosov L.
Retrospective identification of
[3] Alexik M. Modelling and perturbations and interferences.
identification of eye-hand dynamics. Science; Moscow. 1972. 232 p
Simulation Practice and Theory. 2000;8:
25-38. ISSN 0928-48693 [11] Krasovskij N. Theory of motion
control. Science; Moscow. 1968. 476 p
[4] Pop O, Valle V, Cottron M.
Numerical and experimental study of
[12] Sarmar I, Malik A. Modeling,
the plastic in the vicinity of the crack tip
analysis and simulation of a pan tilt
by the optical caustics method. In:
platform based on linear and nonlinear
Proceedings of XXI International
systems. In: Proceedings of
Congress of Theoretical and Applied
International Conference IEEE/ASME
Mechanics; Warsaw, Poland. 2004. 24 p
MESA, China. 2008. pp. 147-152
[5] Lino P, Maione B. Multi-Domain and
[13] De la Fuente A. Mathematical
Control in Mechatronics: The Case of
Methods and Models for Economists.
Common Rail Injection Systems.
London: Cambridge University Press;
Vienna, Austria: In-Teh is Croatian
2000. ISBN: 0521585295, 0521585120,
Branch of I-Tech and Publishing KG;
9780521585293
2008. pp. 234-245. DOI: 10.5772/5825

[6] Stepashko V. Method of critical [14] Tillack J et al. Software framework

dispersions as analytical tools of theory for modeling and simulation of dynamic
of inductive modeling. Problems of metabolic and isotopic systems. In:
Control and Information. 2008:40(2): Proceedings of 6-th International
4-22. DOI: 10.1615/JAutomatInfScien. Vienna Conference on Mathematical
v40.i3.20 Modelling, Vienna, Full Papers CD
Volume. 2009. pp. 769-778
[7] Gubarev V, Method of iterative
identification of multidimensional systems [15] Perminov V. Mathematical
by uncertain data. Part I. Theoretical modelling of large forest fires initiation.
aspects, Problems of Control and In: Proceedings of 6-th International
Information. 2008;38(9):8-26. DOI: Vienna Conference on Mathematical
10.1615/J Automat Inf Scien.v38.i9.20 Modelling, Vienna, Full Papers CD
Volume. 2009. pp. 1165-1172
[8] Gukov O. Algorithms of iterative
identification of many-dimensional [16] Breitenecker F et al. Laura and
systems. In: Proceedings of XV Petrarca—True emotions vs. modelled

63
Computational Models in Engineering

emotions. In: Proceedings of 6-th Ukraine: Science and Osvita; 2009.

International Vienna Conference on 188 p. ISBN 978-966-8736-05-6
Mathematical Modelling, Vienna, Full
Papers CD Volume, Vienna Univ. of [24] Menshikov Y. Identification of
Technology. 2009. pp. 46-69. ISBM moment of technological resistance on
978-3-901608-35-3 rolling. Journal of Differential Equations
and their Applications in Physics. 1976;
[17] Menshikov YL. Synthesis of 1:22-28
adequate mathematical description as
solution of special inverse problems. [25] Menshikov Y, Yah G. Synthesis of
European Joumal of Mathematical the moment of technological resistance
Sciences. 2013;2(3):256-271. DOI: on a rolling mill. Journal of Dynamics
10.4236/jmp.2015.614217 and Strength of Heavy Machines. 1983;
7:33-39
[18] Bogaevski V, Povzner A. Equations
in partial derivatives. In: Algebraic [26] Bondar V et al. Some results of an
Methods in Nonlinear Perturbation experimental study of the main line of
Theory. Applied Mathematical Sciences. blooming 1150. Journal of Dynamic
Vol. 88. New York, NY: Springer; 1991. Strength and Stability of Elements of
DOI: 10.1007/978-1-4612-4438-7_5 Large Machines. 1973:3-17

[19] Kalman R. Mathematical linear [27] Alfven H. The structure and

description of dynamical systems. SIAM evolutionary history of the solar system.
Journal on Control and Optimization. Astrophysics and Space Science. 1973;
1963;1(2):152-192. Available from: h 21(1):117-176. DOI: 10.1007/BF00642197
ttp://www.siam.org/journals/ojsa.php
[28] Schmidt O. The origin of the earth
[20] Schulmeister K, Kozek M. and planets. Science (Moscow). 1962.
Modelling of lateral dynamics for an 279 p
endless metal process belt. In:
Proceedings of 6-th International [29] Kuiper GP. On the origin of the solar
Vienna Conference on Mathematical system. In: Hynek JA, editor. Article in
Modelling, Vienna, full Papers CD Astrophysics. New York, Toronto,
Volume. 2009. pp. 1965-1973 London: McGraw-Hill; 1951

[21] Porter W. Modern Foundations of [30] Perehrest V. The planetary vortex

Systems Engineering. New York/ and the Laplace and Weizsäcker
London: The Macmillan hypotheses. Visnik Dnipropetr
Company/Collier-Macmillan Limited; University. Mechanics. 2009;17(5):
1970. DOI: 10.1126/ 113-124
science.157.3792.1030-a
[31] Perehrest V. About the structure of
[22] Menshikov Y. Synthesis of adequate planetary viiors and laws of law and
mathematical description for dynamical wrapping. Visnik Dnipropetr
systems. In: Proceedings of International University. Mechanics. 2010;18(14):
Conference on Mathematical Modelling 110–118
in Applied Sciences. ICMMAS’2017;
24-28 July 2017. St. Petersburg-Russia: [32] NASA “Sun Fact Sheet”. Available
SPbPU; 2017. 2 p from: http://nssds.gfsc.nasa.gov/plantary
2009
[23] Menshikov Y, Polyakov N.
Handbook of Identification of Models of [33] Available from: http://exoplanets.
External Influences. Dnepropetrovsk, org/exotable/exotable.html

64
Criteria for Adequacy Estimation of Mathematical Descriptions of Physical Processes
DOI: http://dx.doi.org/10.5772/intechopen.82725

[34] Shanon R. Systems Simulation—The [43] Menshikov Y. Features of

Art and Science. New Jersey: Prentice- parameters identification of algebraic
Hall, Inc., Englewood Cliffs; 1975. ISBN mathematical models. International
10.0138818398 Journal of Engineering and Innovative
Technology. 2014;4(5):5. ISSN:
[35] Vere-Jones D. Some models and 2277-3754
procedures for space-time point
processes. Environmental and
Ecological Statistics. 2008;16(2):
173-195. DOI: 10.1007/s10651-007-
0086-0

[36] Utsu T. Representation and analysis

of the earthquake size distribution: A
historical review and some new
approaches, Mechanics. 2010;18(14):
110-118. DOI: 10.1007/s000240050276

[37] Erofeev V. Wave Processes in Solids

with Microstructure. Moscow: Moscow
University; 1999. 255 p

[38] Turbal Y. The trajectories of self-

reinforsing solitary wave in the gas disc
of galaxies. In: Proceedings of the 3rd
International Conference on Nonlinear
Dynamic. 2010. pp. 112-118

[39] Turbal Y. Necessary and sufficient

conditions for the existence of the
motion equations solutions for
anisotropic elastic bodies in the form of
solitary waves. Problems of Applied
Mathematics and Mathematical
Modeling. 2012;12:78-86. DOI: 10.1155/
2013/537520

[40] Turbal Y. Elastic anisotropy from

the standpoint of the solitary waves
existence like δ-Soliton. Problems of
Mechanics. 2012;18:76-90

[41] Strogatz SH. Love affairs and

differential equations. Mathematics
Magazine. 1988;61:35. DOI: 10.2307/
2690328

[42] Rinaldi S. Laura and Petrarch: An

intriguing case of cyclical love
dynamics. SIAM Journal on Applied
Mathematics. 1998;58(4):1205-1221.
DOI: 10.1007/s11071-015-2351-3

65
Chapter 5

Power Flow Analysis

Mohammed Albadi

Abstract

Power flow, or load flow, is widely used in power system operation and
planning. The power flow model of a power system is built using the relevant
network, load, and generation data. Outputs of the power flow model include
voltages at different buses, line flows in the network, and system losses. These
outputs are obtained by solving nodal power balance equations. Since these
equations are nonlinear, iterative techniques such as the Newton-Raphson, the
Gauss-Seidel, and the fast-decoupled methods are commonly used to solve this
problem. The problem is simplified as a linear problem in the DC power flow
technique. This chapter will provide an overview of different techniques used to
solve the power flow problem.

Keywords: power flow, load flow, iterative techniques, Newton-Raphson,

Gauss-Seidel, fast-decoupled, DC power flow

1. Problem formulation

Power flow analysis is a fundamental study discussed in any power system

analysis textbook such as [1–6]. The objective of a power flow study is to calculate
the voltages (magnitude and angle) for a given load, generation, and network
condition. Once voltages are known for all buses, line flows and losses can be
calculated. The starting point of solving power flow problems is to identify the
known and unknown variables in the system. Based on these variables, buses are
classified into three types: slack, generation, and load buses as shown in Table 1.
The slack bus is required to provide the mismatch between scheduled generation
the total system load including losses and total generation. The slack bus is com-
monly considered as the reference bus because both voltage magnitude and angles
are specified; therefore, it is called the swing bus. The rest of generator buses are
called regulated or PV buses because the net real power is specified and voltage
magnitude is regulated. Most of the buses in practical power systems are load buses.
Load buses are called PQ buses because both net real and reactive power loads are
specified.
For PQ buses, both voltage magnitudes and angles are unknown, whereas for PV
buses, only the voltage angle is unknown. As both voltage magnitudes and angles
are specified for the Slack bus, there are no variables that must be solved for. In a
system with n buses and g generators, there are 2(n-1)-(g-1) unknowns. To solve
these unknowns, real and reactive power balance equations are used. To write
these equations, the transmission network is modeled using the admittance matrix
(Y-bus).

67
Computational Models in Engineering

Bus type Voltage (jV i j ∠ δi ) Real power Reactive power

Magnitude Angle Generation Load Net (Pi ) Generation Load Net (Q  i )

Slack/ Specified Specified Unknown Specified Unknown Unknown Specified Unknown

Swing

Generator/ Specified Unknown Specified Specified Specified Unknown Specified Unknown

Regulated/
PV

Load/PQ Unknown Unknown Specified Specified Specified Specified Specified Specified

Table 1.
Type of buses in the power flow problem.

2. Admittance matrix and power flow equation

The admittance matrix of a power system is an abstract mathematical model of

the system. It consists of admittance values of both lines and buses. The Y-bus is a
square matrix with dimensions equal to the number of buses. This matrix is sym-
metrical along the diagonal.
2 3
Y 11 ⋯ Y 1n
6 7
Y¼4 ⋮ ⋱ ⋮ 5 (1)
Y n1 ⋯ Y nn

The values of diagonal elements (Y ii ) are equal to the sum of the admittances
connected to bus i. The off-diagonal elements (Y ij ) are equal to the negative of the
admittance connecting the two buses i and j. It is worth noting that with large
systems, Y-bus is a sparse matrix.
n
Y ii ¼ ∑ yij (2)
j¼0
j 6¼ i

Y ij ¼ Y ji ¼ �yij (3)

The net injected power at any bus can be calculated using the bus voltage (V i ),
neighboring bus voltages (V j ), and admittances between the bus and its
neighboring buses (yij ) as shown in Figure 1.
� �
Ii ¼ V i yi0 þ ðV i � V 1 Þyi1 þ ðV i � V 2 Þyi2 þ … þ V i � V j yij

Rearranging the elements as a function of voltages, the current equation

becomes as follows:
� �
Ii ¼ V i yi0 þ yi1 þ yi2 þ :: þ yij � V 1 yi1 � V 2 yi2 � … � V j yij

Ii ¼ V i ∑ yij � ∑ yij V j ¼ V i Y ii þ ∑ Y ij V j
j¼0 j¼1 j¼1
j 6¼ i j 6¼ i j 6¼ i

The power equation at any bus can be written as follows:

Si ¼ Pi þ jQ  i ¼ V i Ii ∗

68
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

Figure 1.
Net injected power.

Or

Si ∗ ¼ Pi � jQ  i ¼ V i ∗ Ii

Substituting the expression on the current in Si ∗ equation results in the following

formula:
0 1 0 1
B C B C
Si ∗ ¼ V i ∗ B C ∗B C
@V i ∑ yij � ∑ yij V j A ¼ V i @V i Y ii þ ∑ Y ij V j A
j¼0 j¼1 j¼1
j 6¼ i j 6¼ i j 6¼ i

Real and reactive power can be calculated from the following equations:
8 0 19 8 0 19
>
> >
> >
> >
>
< B C= < B C=
∗B C ∗ B C
Pi ¼ Re V i @V i ∑ yij � ∑ yij V j A ¼ Re V i @V i Y ii þ ∑ Y ij V j A
>
> j¼0 j¼1 >
> >
> j¼1 >
>
: ; : ;
j 6¼ i j 6¼ i j 6¼ i
8 0 19 8 0 19
>
> >
> >
> >
>
< B C = < B C=
∗B ∗
Q  i ¼ �Im V i @V i ∑ yij � ∑ yij V j C B C
A> ¼ �Im>V i @V i Y ii þ ∑ Y ij V j A>
>
> j¼0 j¼1 > > j¼1 >
: ; : ;
j 6¼ i j 6¼ i j 6¼ i

Or
n � �� � � �
Pi ¼ ∑ jV i j�V j ��Y ij � cos θij � δi þ δj (4)
j¼1

n � �� � � �
Q  i ¼ � ∑ jV i j�V j ��Y ij � sin θij � δi þ δj (5)
j¼1

And the current (Ii) can be written as a function of the power as follows:

Pi � jQ  i
¼ V i ∑ yij � ∑ yij V j ¼ V i Y ii þ ∑ Y ij V j (6)
Vi∗ j¼0 j¼1 j¼1
j 6¼ i j 6¼ i j 6¼ i

69
Computational Models in Engineering

Figure 2.
Impedance diagram.

Figure 3.
Admittance diagram.

Example 1: Admittance matrix formation.

For the below 4-bus system in Figure 2, the admittance matrix is constructed by
converting all impedances in the system into admittances as shown in Figure 3.
Then, diagonal and off-diagonal elements are calculated using Eqs. (2) and (3).
2 3 2 3
�j7:5 j4 j2:5 0 7:5 ∠ � 90° 4 ∠ 90° 2:5 ∠ 90° 0
6 j4 �j7:75 0 7 6
j2:5 7 6 4 ∠ 90° 7:75 ∠ � 90° 0 2:5 ∠ 90° 7
6 7
Y¼6 7¼6 7
4 j2:5 0 �j4:5 j2 5 4 2:5 ∠ 90° 0 4:5 ∠ � 90° 2 ∠ 90° 5
0 j2:5 j2 �j4:5 0 2:5 ∠ 90° 2 ∠ 90° 4:5 ∠ � 90°

3. Gauss-Seidel technique

The Gauss-Seidel (GS) method, also known as the method of successive dis-
placement, is the simplest iterative technique used to solve power flow problems. In
general, GS method follows the following iterative steps to reach the solution for the
function f ðxÞ ¼ 0:

• Rearrange the function into the form x ¼ g ðxÞ to calculate the unknown
variable.

70
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374
� �
• Calculate the value g x½0� based on initial estimates x½0� .
� �
• Calculate the improved value x½1� ¼ g x½0� .

• Continue
� ½kþ1solving �for improved values until the solution is within acceptable
�
limits x � ½k� �
� x ≤ ϵ.

The rate of convergence can be improved using acceleration factors by modify-

ing the step size α.
h � � i
x½kþ1� ¼ x½k� þ α g x½kþ1� � x½k� (7)

In the context of a power flow problem, the unknown variables are voltages at
all buses, but the slack. Both voltage magnitudes and angles are unknown for
load buses, whereas voltage angles are unknown for regulated/generation buses.
The voltage V i at bus i can be calculated using either equations:
!
1 Pi sch � jQ  i sch
Vi ¼ þ ∑ yij V j (8)
∑ j ¼ 1 yij Vi∗ j

j 6¼ i
0 1
sch sch
1 B BP i � jQ  i C
Vi ¼ � ∑ Y ij V j C (9)
Y ii @ Vi∗ j¼1
A
j 6¼ i

where yij is the admittance between buses i and j, Y ij is the Y-bus element, Pi sch
the net scheduled injected real power, Q  i sch is the net scheduled injected reactive
power, and V i ∗ is the conjugate of V i . The net injected quantities are the sum of the
generation minus load. Typically, the initial estimates of V i ¼ 1 ∠ 0° .
The iterative voltage equation is as follows:
!
½kþ1� 1 Pi sch � jQ  i sch ½k or kþ1�
Vi ¼ � ∑j 6¼ i Y ij V j (10)
Y ii V i ∗ ½k�

Or
!
1 Pi sch � jQ  i sch
V i ½kþ1� ¼ þ ∑ yij V j ½k or kþ1� (11)
∑j ¼ 0 yij Vi∗ j 6¼ i

Both real and reactive powers are scheduled for the load buses, and Eq. 6 is used
to determine both voltage magnitudes and angles (jV i j ∠ δ) for every iteration
(V i ½kþ1� ).
For regulated buses, only real power is scheduled. Therefore, net injected
reactive power is calculated based on the iterative voltages (V i ½kþ1� ) using either
equations:
8 0 19
>
> >
>
< B n n C=
½kþ1� ∗ ½k� B ½k� ½k or kþ1� C
Q  i ¼ �Im V i @V i ∑ yij � ∑ yij V j A> (12)
>
> j¼0 j¼1 >
: ;
j 6¼ i

71
Computational Models in Engineering

n � �½k or kþ1� � � � �
Q  i ½kþ1� ¼ � ∑ jV i j½k� �V j � �Y ij � sin θij � δi ½k� þ δj ½k� (13)
j¼1
� �
where jV i j and �V j � are the magnitudes of the voltage at buses i and j, respec-
tively. δi and δj are the associated angles. yij is the admittance between buses i and j.
� �
�Y ij � is the magnitude of the Y-bus element between the two buses; and θij is the
corresponding angle.
Since the voltage magnitude (jV i j) is specified at regulated/PV buses, Eqs. (8) or
(9) will be used to determine the voltage angles only. To achieve this, two options
can be used:

1. When using the polar form (jV i j ∠ δi ), discard the iterative voltage magnitude
and keep the iterative angle.

2. When using the rectangular form (Rei þ j Imi ), discard the real part (Rei ) and
keep the imaginary part (Imi ) of the iterative voltage. The new real part
(Rei new ) can be calculated from the specified magnitude (jV i j) and the iterative
imaginary part.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Rei new
¼ jV i j2 � Imi 2 (14)

The iterative process stops when the voltage improvement reaches acceptable
� �
limits: �V i ½kþ1� � V i ½k� � ≤ ϵ.
Example 2: Gauss-Seidel power flow solution.
Figure 4 below shows a 3-bus system. Perform 2 iterations to obtain the voltage
magnitude and angles at buses 2 and 3. Impedances are given on 100 MVA base.
Solution:
The admittance values of the transmission network and the injected power in
per unit at buses 2 and 3 are calculated as shown in Figure 5. Note that net injected
power at the load bus is negative while that of the PV bus is positive. Per units
values are obtained by diving actual values (MW and MVAR) by the base (100
MVA).
Iteration #1: assume V 2 ½0� ¼ 1:00 ∠ 0° and V 3 ½0� ¼ 1:03 ∠ 0° .
!
½1� 1 P2 sch � jQ 2 sch ½0�
V2 ¼ ∗ þ y21 V 1 þ y23 V 3
y21 þ y23 V 2 ½0�
� �
1 �2 þ j0:5 ° °
¼ þ ð5 � j15Þ1:02 ∠ 0 þ ð15 � j50Þ1:00 ∠ 0
ð5 � j15Þ þ ð15 � j50Þ 1:00 ∠ 0°
V 2 ½1� ¼ 1:0120 � j0:0260 ¼ 1:0123 ∠ � 1:4717°

As Q3 is not given, it is calculated based on the latest available information using

Eqs. (12) or (13).
8 0 19
>
> >
>
< B n n C =
½1� ∗ ½0� B ½0� ½k or kþ1� C
Q 3 ¼ �Im V 3 @V 3 ∑ y3j � ∑ y3j V j A>
>
> j¼0 j¼1 >
: ;
j 6¼ i
n � � � h i�o
Q 3 ½1� ¼ �Im V 3 ∗ ½0� V 3 ½0� y31 þ y32 � y31 V 1 þ y32 V 2 ½1�

72
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

Figure 4.
3-Bus power system.

Figure 5.
Power flow input data.

� � �
Q 3 ½1� ¼ �Im 1:03 ∠ 0° 1:03 ∠ 0° ð10 � j40 þ 15 � j50 Þ � ð10 � j40Þ1:02 ∠ 0°
þð15 � j50Þð1:0120 � j0:0260Þ�Þg
¼ �Imf1:7201 � j0:9373g ¼ 0:9373 pu

Now that Q 3 ½1� is calculated, the voltage V 3 ½1� can be calculated:

!
½1� 1 P3 sch � jQ 3 ½1� ½1�
V3 ¼ ∗ þ y31 V 1 þ y32 V 2
y31 þ y32 V 3 ½ 0�
� �
1 1:5 � j0:9373 °
¼ þ ð 10 � j40Þ1:02 ∠ 0 þ ð 15 � j50 Þð 1:0120 � j0:0260 Þ
10 � j40 þ 15 � j50 1:03 ∠ 0°
¼ 1:0294 � j0:0022 pu

Since the magnitude of V 3 is specified, we retain the imaginary part of V 3 ½1� and
calculate the real part using Eq. (14).
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Rei new ¼ 1:032 � 0:00222 ¼ 1:03 pu

73
Computational Models in Engineering

Therefore,

 V 3 ½1� ¼ V 3 ½1� ¼ 1:03 � j0:0022 ¼ 1:03 ∠ � 0:1226° pu

Iteration #2: considering V 2 ½1� ¼ 1:0120 � j0:0260 and V 3 ½1� ¼ 1:03 � j0:0022.
!
1 P2 sch � jQ 2 sch
V 2 ½2 � ¼ ∗ þ y21 V 1 þ y23 V 3 ½1�
y21 þ y23 V 2 ½1 �
� �
1 �2 þ j0:5
¼ þ ð5 � j15Þ1:02 ∠ 0° þ ð15 � j50Þð1:03 � j0:0022Þ V 2 ½2�
ð5 � j15Þ þ ð15 � j50Þ 1:0120 � j0:0260
¼ 1:0115 � j 0:0270 ¼ 1:0119 ∠ � 1:5273°

Q 3 ½2� calculation is given below:

n � � � h i�o
Q 3 ½2� ¼ �Im V 3 ∗ ½1� V 3 ½1� y31 þ y32 � y31 V 1 þ y32 V 2 ½2�

Q 3 ½2� ¼ �Imfð1:03 � j0:0022Þðð1:03 � j0:0022Þð10 � j40 þ 15 � j50 Þ

� ���
� ð10 � j40Þ1:02 ∠ 0° þ ð15 � j50Þð1:0115 � j 0:0270Þ
¼ 1:0000 pu

The voltage V 3 ½2� is calculated as follows:

!
1 P3 sch � jQ 3 ½2�
V 3 ½2� ¼ ∗ þ y V
31 1 þ y V
32 2
½2�
y31 þ y32 V 3 ½1 �
� �
1 1:5 � j1 °
¼ þ ð 10 � j40 Þ1:02 ∠ 0 þ ð 15 � j50 Þ ð 1:0115 � j 0:0270 Þ
10 � j40 þ 15 � j50 1:03 ∠ 0°
¼ 1:0298 � j0:0030 pu

Only imaginary value of the calculated V 3 ½2� is retained and the real part is
calculated based on the retained imaginary values and the scheduled jV 3 j:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Rei new ¼ 1:032 � 0:00302 ¼ 1:03 pu

Therefore,

V 3 ½2� ¼ 1:03 � j0:003 ¼ 1:03 ∠ � 0:1644° pu

The iterative solution is presented in Table 2.

Iteration V 2 ½k� V 3 ½k�

1 1:0123 ∠ � 1:4717 ° 1:03 ∠ � 0:1226°

2 1:0119 ∠ � 1:5273° 1:03 ∠ � 0:1644°

3 1:0119 ∠ � 1:5598° 1:03 ∠ � 0:1846°

4 1:0119 ∠ � 1:5750° 1:03 ∠ � 0:1941°

5 1:0118 ∠ � 1:5823° 1:03 ∠ � 0:1986°

6 1:0118 ∠ � 1:5857 ° 1:03 ∠ � 0:2008°

Table 2.
Gauss-Seidel iterative solution.

74
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

4. Newton-Raphson technique

The Newton-Raphson (N-R) technique, also known as the method of successive

approximation, is based on Taylor’s expansion approximation. The unknown x in
the function f ðxÞ ¼ c can be determined using Taylor’s expansion approximation.
Starting with an initial estimate x½0� , the deviation from the correct solution is Δx½0� .
Applying Taylor’s expansion, the function can be written as follows:
� �
f x½0� þ Δx½0� ¼ c (15)
� � � � � � 1 00 � � 2 1 000 � � 3
f x½0� þ Δx½0� ¼ f x½0� þ f 0 x½0� Δx½0� þ f x½0� Δx½0� þ f x½0� Δx½0� þ … ¼ c (16)
2! 3!
00 � � 2
Assuming Δx½0� is small, the higher order terms (2!1 f x½0� Δx½0� þ
1
00 0 � � 3
3! f x½0� Δx½0� þ …) are neglected and the function can be approximated by the
first two terms.
� � � � � �
f x½0� þ Δx½0� ≈ f x½0� þ f 0 x½0� Δx½0� ¼ c (17)

Based on x½0� , the deviation from the correct solution can be iteratively
calculated.
� � � �
c � f x½0� Δf x½0�
Δx½0� ¼ ¼ (18)
f 0 ðx½0� Þ f 0 ðx½0� Þ

The improved solution can be calculated iteratively.

x½1� ¼ x½0� þ Δx½0� (19)

The iterative process is stopped when the mismatch between

� scheduled
� and
½k� � ½k� � � ½k� �
calculated value (Δf ¼ c � f x ) is within acceptable limits�Δf � ≤ ϵ.
In the context of power flow problems, unknown variables x are both voltage
magnitude and angles (jV i j ∠ δi ) at load buses, as well as voltage angles (δi ) at
regulated buses. The scheduled (specified) quantities (c) are both net real (Pi sch )
and reactive power (jQ  i sch ) values at load buses and real power at generation buses
as shown in Table 1. The iterative values of reactive power are calculated using
Eqs. (12) or (13). Similarly, the iterative values of real power are calculated using
Eqs. (20) or (21):
8 0 19
>
> >
>
< B n n C=
∗ ½k� B ½k�
Pi ½kþ1� ¼ Re V i @V i ∑ yij � ∑ yij V j ½k or kþ1� C
A> (20)
>
> j¼0 j¼1 >
: ;
j 6¼ i

n � �½k or kþ1� � � � �
Pi ½kþ1� ¼ ∑ jV i j½k� �V j � �Y ij � cos θij � δi ½k� þ δj ½k� (21)
j¼1

The Newton-Raphson power flow formulation can be written as follows:

" # " # " # " # " # " #
½k� δi ½k� Pi sch Pi ½k� Pi sch Pi ½k� ΔPi ½k�
x ¼ ,c¼ ,f ¼ , and � ¼ .
jV i j½k� Q  i sch Q  i ½k� Q  i sch Q  i ½k� ΔQ i ½k�

75
Computational Models in Engineering

When more variables are used, the derivative f 0 is replaced by partial derivatives
with respect to different variables. The partial derivative matrix is called the Jaco-
bian matrix.
2 3
∂Pi ∂Pi " #
6 ∂δi ∂jV i j 7 J J PjV j
f0 ¼ 6 7 ¼ Pδ (22)
4 ∂Q ∂Q  i 5 J Qδ J Q jV j
 i
∂δi ∂jV i j

Therefore, the Newton-Raphson power flow formulation can be solved using the
below equation:

" # " # " # 2 3" #

sch ½k� ½k� ∂Pi ½k� ∂Pi ½k�
½k�
Pi Pi ΔPi ∂δi ∂jV i j
5 Δδi
� ¼ ¼4 (23)
Q  i sch Q  i ½k� ΔQ i ½k� ∂Q  i ½k� ∂Q  i ½k �
ΔjV i j½k�
∂δi ∂jV i j

To solve for the deviation, the inverse of the Jacobian matrix is required for
every iteration.
" # " #�1 " #
Δδ½k� J Pδ ½k� J PjV j ½k� ΔPi ½k�
¼ (24)
ΔjV i j½k� J Qδ ½k� J Q jV j ½k� ΔQ i ½k�

Then, new values are calculated:

" # " # " #
δi ½k� δi ½k�1� Δδi ½k�
¼ þ (25)
jV i j½k� jV i j½k�1� ΔjV i j½k�

The iterative process stops

" when
# the mismatch between calculated and sched-
ΔPi ½k�
uled quantities is within ≤ ϵ.
ΔQ i ½k�
Example 3: Newton-Raphson power flow solution.
Solve the power flow problem shown in Figure 3 using the Newton-Raphson
technique. Perform two iterations.
Solution:
The first step in Newton-Raphson Power Flow technique is building Y-bus using
Eqs. (2) and (3).
2 3
15 � j55 �5 þ j15 �10 þ j40
6 7
6 7
Y ¼ 6 �5 þ j15 20 � j65 �15 þ j50 7
4 5
�10 þ j40 �15 þ j50 25 � j90
2 3
57:01 ∠ � 74:74° 15:81 ∠ 108:43° 41:23 ∠ 104:04°
6 7
6 7
¼ 6 15:81 ∠ 108:43° 68:01 ∠ � 72:90° 52:20 ∠ 106:70° 7
4 5
41:23 ∠ 104:04° 52:20 ∠ 106:70° 93:41 ∠ � 74:48°

Since the unknown variables are δ2 , δ3 , and jV 2 j; the scheduled quantities are
P2 sch , P3 sch , and Q 2 sch , the following problem formulation can be written.

76
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374
2 3
2 3 2 3 ∂P2 ½k� ∂P2 ½k� ∂P2 ½k� 2 3
sch ½ k�
P2 6 P2 ∂δ2 ∂δ3 ∂jV 2 j 7 Δδ2 ½k�
6 sch 7 6 ½k� 7 6 ∂P3 ½k� ∂P3 ½k� ∂P3 ½k� 76
7
½k� 7
4 P3 5 4 P3 5 6
� ¼ 6 ∂δ2 ∂δ3 ∂jV 2 j 74 Δδ3 5
4 5 ½k�
Q 2 sch Q 2 ½ k� ∂Q 2 ½k� ∂Q 2 ½k� ∂Q 2 ½k � Δ j V 2 j
∂δ2 ∂δ3 ∂jV 2 j

To calculate the Jacobian matrix elements, P2 , P3 , and Q 2 equations are obtained

using (4) and (5).

n � �� � � �
P2 ¼ ∑ jV 2 j�V j ��Y 2j � cos θ2j � δ2 þ δj
j¼1

¼ jV 2 jjV 1 jjY 21 j cos ðθ21 � δ2 þ δ1 Þ þ jV 2 j2 jY 22 j cos ðθ22 Þ þ jV 2 jjV 3 jjY 23 j cos ðθ23 � δ2 þ δ3 Þ
∂P2
¼ jV 2 jjV 1 jjY 21 j sin ðθ21 � δ2 þ δ1 Þ þ jV 2 jjV 3 jjY 23 j sin ðθ23 � δ2 þ δ3 Þ
∂δ2
∂P2
¼ �jV 2 jjV 3 jjY 23 j sin ðθ23 � δ2 þ δ3 Þ
∂δ3
∂P2
¼ jV 1 jjY 21 j cos ðθ21 � δ2 þ δ1 Þ þ 2jV 2 jjY 22 j cos ðθ22 Þ þ jV 3 jjY 23 j cos ðθ23 � δ2 þ δ3 Þ
∂jV 2 j
n � �� � � �
P3 ¼ ∑ jV 3 j�V j ��Y 3j � cos θ3j � δ3 þ δj
j¼1

¼ jV 3 jjV 1 jjY 31 j cos ðθ31 � δ3 þ δ1 Þ þ jV 3 jjV 2 jjY 32 j cos ðθ32 � δ3 þ δ2 Þ þ jV 3 j2 jY 33 j cos ðθ33 Þ
∂P3
¼ �jV 2 jjV 3 jjY 32 j sin ðθ32 � δ3 þ δ2 Þ
∂δ2
∂P3
¼ jV 3 jjV 1 jjY 31 j sin ðθ31 � δ3 þ δ1 Þ þ jV 3 jjV 2 jjY 32 j sin ðθ32 � δ3 þ δ2 Þ
∂δ3
∂P3
¼ jV 3 jjY 32 j cos ðθ32 � δ3 þ δ2 Þ
∂jV 2 j
n � �� � � �
Q 2 ¼ �∑ jV 2 j�V j ��Y 2j � sin θ2j � δ2 þ δj
j¼1

¼ �jV 2 jjV 1 jjY 21 j sin ðθ21 � δ2 þ δ1 Þ � jV 2 j2 jY 22 j sin ðθ22 Þ � jV 2 jjV 3 jjY 23 j sin ðθ23 � δ2 þ δ3 Þ
∂Q 3
¼ jV 2 jjV 1 jjY 21 j cos ðθ21 � δ2 þ δ1 Þ þ jV 2 jjV 3 jjY 23 j cos ðθ23 � δ2 þ δ3 Þ
∂δ2
∂Q 3
¼ �jV 2 jjV 3 jjY 23 j cos ðθ23 � δ2 þ δ3 Þ
∂δ3
∂Q 3
¼ �jV 1 jjY 21 j sin ðθ21 � δ2 þ δ1 Þ � 2jV 2 jjY 22 j sin ðθ22 Þ � jV 3 jjY 23 j sin ðθ23 � δ2 þ δ3 Þ
∂jV 2 j

Iteration #1: assume V 2 ½0� ¼ 1:00 ∠ 0° and V 3 ½0� ¼ 1:03 ∠ 0° .

2 ½1� 3 2 3
P2 �0:5500
6 7 6 7
The calculated quantities:4 P3 ½1� 5 ¼ 4 0:5665 5
Q ½1� �1:8000:
2 2sch 3 2 3
P2 �2
6 7 6 7
The scheduled quantities:4 P3 sch 5 ¼ 4 1:5 5
Q 2 sch �0:5

77
Computational Models in Engineering

2 33 2 2 3 2 3
ΔP2 ½1��2 �0:5500 �1:4500
6 7 6 7 6 7 6 7
The mismatch power matrix:4 ΔP3 ½1� 5 ¼ 4 1:5 5 � 4 0:5665 5 ¼ 4 0:9335 5
ΔQ 2 ½1� �0:5 �1:8000 1:3000
The Jacobian matrix (J) elements for iteration # 1 are as follows:
2 3
66:8 �51:5 19:45
6 7
J ¼ 4 �51:5 93:52 �15:45 5
�20:55 15:45 63:2

Newton-Raphson formulation is as follows:

2 3 2 32 3
�1:4500 66:8 19:45 �51:5 Δδ2 ½1�
6 7 6 76 7
4 0:9335 5 ¼ 4 �51:5 �15:45 54 Δδ3 ½1� 5
93:52
1:3000 �20:55 15:45 63:2 ΔjV 2 j½1�
2 3 2 3�1 2 3 2 3
Δδ2 ½1� 66:8 �51:5 19:45 �1:4500 �0:0279 rad
6 7 6 7 6 7 6 7
4 Δδ3 ½1� 5 ¼ 4 �51:5 93:52 �15:45 5 4 0:9335 5 ¼ 4 �0:0033 rad 5
ΔjV 2 j½1� �20:55 15:45 63:2 1:3000 0:0123 pu
2 ½1� 3 2 ½0� 3 2 3
δ2 δ2 Δδ2 ½1�
6 ½1� 7 6 ½0� 7 6 7
4 δ3 5 ¼ 4 δ3 5 þ 4 Δδ3 ½1� 5
jV 2 j½1� jV 2 j½0� ΔjV 2 j½1�
2 ½1� 3 2 3 2 3 2 3 2 3
δ2 0 �0:0279 rad �0:0279 rad �1:5986°
6 ½1� 7 6 7 6 7 6 7 6 7
4 δ3 5 ¼ 4 0 5 þ 4 �0:0033 rad 5 ¼ 4 �0:0033 rad 5 ¼ 4 �0:1891° 5
jV 2 j½1� 1:00 0:0123 pu 1:0123 pu 1:0123 pu

Iteration #2: Consider V 2 ½1� ¼ 1:0123 ∠ � 1:5986° and V 3 ½1� ¼ 1:03 ∠ � 0:1891° .
2 ½ 2� 3 2 3
P2 �2:0109
6 7 6 7
The calculated quantities: 4 P3 ½2� 5 ¼ 4 1:5202 5
Q 2 ½ 2� �0:4621
2 3 2 3 2 3 2 3
ΔP2 ½2� �2 �2:0109 0:0109
6 7 6 7 6 7 6 7
The mismatch power matrix: 4 ΔP3 ½2� 5 ¼ 4 1:5 5 � 4 1:5202 5 ¼ 4 �0:0202 5
ΔQ 2 ½2� �0:5 �0:4621 �0:0379
The Jacobian matrix elements are calculated as follows:
2 3
67:07 �51:74 18:26
6 7
J ¼ 4 �52:50 94:49 �14:18 5
�22:51 16:91
65:34
2 3 2 3�1 2 3
Δδ2 ½2� 67:07 �51:74 18:26 0:0109
6 7 6 7 6 7
4 Δδ3 ½2� 5 ¼ 4 �52:50 94:49 �14:18 5 4 �0:0202 5
ΔjV 2 j½2� �22:51 16:91 65:34 �0:0379
2 3 2 3
Δδ2 ½2� 0:1272 � 10�3 rad
6 ½2� 7 ¼ 6 7
4 Δδ3 5 4 �0:2154 � 10�3 rad 5
ΔjV 2 j½2� �0:4810 � 10�3 pu

78
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

2 3 2 3 2 3
δ2 ½2� δ2 ½1� Δδ2 ½2�
6 ½2� 7 6 ½1� 7 6 7
4 δ3 5 ¼ 4 δ3 5 þ 4 Δδ3 ½2� 5
jV 2 j½2� jV 2 j½1� ΔjV 2 j½2�
2 ½2� 3 2 3 2 3
δ2 �0:0279 rad 0:1272 � 10�3 rad
6 ½2� 7 6 7 6 7
4 δ3 5 ¼ 4 �0:0033 rad 5 þ 4 �0:2154 � 10�3 rad 5
jV 2 j½2� 1:0123 pu �0:4810 � 10�3 pu
2 ½2� 3 2 3 2 3
δ2 �0:0277 rad �1:5871°
6 ½ 2� 7 6 7 6 7
4 δ3 5 ¼¼ 4 �0:0035 rad 5 ¼ 4 �0:2005° 5
jV 2 j½2� 1:0118pu 1:0118 pu

It is worth noting that the mismatch between calculated and scheduled quanti-
ties diminishes very quickly.
2 3 2 3 2 3 2 3
ΔP2 ½1� �1:4500 ΔP2 ½2� 0:0109
6 7 6 7 6 7 6 7
4 ΔP3 ½1� 5 ¼ 4 0:9335 5and4 ΔP3 ½2� 5 ¼ 4 �0:0202 5
ΔQ 2 ½1� 1:3000 ΔQ 2 ½2� �0:0379
The iterative solution is presented in Table 3.

Iteration V 2 ½k� V 3 ½k�

1 1:0123 ∠ � 1:5986° 1:03 ∠ � 0:1891°

2 1:0118 ∠ � 1:5871° 1:03 ∠ � 0:2005°

3 1:0118 ∠ � 1:5871° 1:03 ∠ � 0:2005°

Table 3.
Newton-Raphson iterative solution.

5. Fast-decoupled technique

In high voltage transmission systems, the voltage angles between adjacent buses
are relatively small. In addition, X=R ratio is high. These two properties result in a
strong coupling between real power and voltage angle and between reactive power
and voltage magnitude. In contrary, the coupling between real power and voltage
magnitude, as well as reactive power and voltage angle, is weak. Considering adja-
cent buses, real power flows from the bus with a higher voltage angle to the bus
with a lower voltage angle. Similarly, reactive power flows from the bus with a
higher voltage magnitude to the bus with a lower voltage magnitude.
Fast-decoupled power flow technique includes two steps: (1) decoupling real
and reactive power calculations; (2) obtaining of the Jacobian matrix elements
directly from the Y-bus.
� � " #� �
ΔP J Pδ 0 Δδ
¼ (26)
ΔQ 0 J Q jV j ΔjV j

or

½ΔP� ¼ ½J Pδ �½Δδ� (27)

79
Computational Models in Engineering

h i
½ΔQ � ¼ J Q jV j ½ΔjV j� (28)

Next step is to obtain J Pδ and J Q jV j from the Y-bus as flowing:

� �
ΔP�1
½Δδ� ¼ �½B0 � (29)
jV j
h 00 i�1 �ΔQ �
½ΔjV j� ¼ � B (30)
jV j
� 00 �
Where the ½B0 � and B are relevant imaginary part of the Y-bus matrix elements.
� 00 �
½B0 � is related to the buses at which real power is scheduled (δ is unknown) and B is
related to the buses at which reactive power is scheduled (jV j is unknown).
� � � �
Bij ¼ �Y ij � sin θij (31)

The fast-decoupled technique requires more iterations to converge compared to

the Newton-Raphson power flow formulation, especially if X=R ratio is not high.
Another advantage of this method is that the Jacobian matrix has constant term
elements which are obtained and inverted once at the beginning of the iterative
process.
Example 4: Fast-decoupled power flow solution.
Solve the power flow problem shown in Figure 3 using fast-decoupled power
flow technique. Perform two iterations.
Solution:
� 00 �
The first step in fast-decoupled power flow technique is obtaining ½B0 � and B
from the Y-bus.
2 3
15 � j55 �5 þ j15 �10 þ j40
6 7
Y¼6
4 �5 þ j15 20 � j65 �15 þ j50 7
5
�10 þ j40 �15 þ j50 25 � j90

Since the unknown voltage angles are δ2 and δ3 , elements for ½B0 � are obtained
from the Y-bus (intersection of columns numbers 2 & 3 and rows numbers 2 & 3)
as follows:
� �
0 �65 50
½B � ¼
50 �90
� 00 �
Since the unknown voltage magnitude is jV 2 j at bus 2, B contains one element
only (intersection of the column and the row number 2):

½B0 � ¼ ½�65�

The fast-decoupled power flow formulation becomes as follows:

2 3
ΔP2
� �
Δδ2 �1 6 jV 2 j 7 � � h i
¼ �½B0 � 6 7 and½ΔjV 2 j� ¼ � B0 0 �1 ΔQ 2
Δδ 4 ΔP3 5 jV 2 j
3
jV 3 j
Or

80
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374
2 3
ΔP2 ½k�
" # 6 ½k�1� 7 h i
Δδ2 ½k� �1 6 jV 2 j 7 � � h ½ k�
i
¼ �½B0 � 6 7 and ΔjV 2 j½k� ¼ � B00 �1 ΔQ½2k�1�
Δδ3 ½k� 6 ½k� 7 jV 2 j
4 ΔP3 5
jV 3 j½k�1�
Iteration #1: assume V 2 ½0� ¼ 1:00 ∠ 0° and V 3 ½0� ¼ 1:03 ∠ 0° .
The calculated quantities:
2 3 2 3
P2 ½1� �0:5500
6 ½1� 7 6 7
4 P3 5 ¼ 4 0:5665 5
Q 2 ½1� �1:8000

The mismatch power matrix:

2 3 2 3 2 3 2 3
ΔP2 ½1� �2 �0:5500 �1:4500
6 7 6 7 6 7 6 7
4 ΔP3 ½1� 5 ¼ 4 1:5 5 � 4 0:5665 5 ¼ 4 0:9335 5
ΔQ 2 ½1� �0:5 �1:8000 1:3000

Calculation of δ2 and δ3 :
2 3
ΔP2 ½1� 2 3
" # 6 7 " #�1 �1:4500
Δδ2 ½1�
6 jV 2 j½0� 7
0 �1 6
�65 50 6 1:00 7
¼ �½B � 6 7¼� 6 7
7 4 0:9335 5
Δδ3 ½1� 4 ΔP3 ½1� 5 50 �90
jV 3 j½0� 1:03
" # " °
#
�0:0254 rad �1:4569
¼ ¼
�0:0041 rad �0:2324°
" # " # " # � � " # " #
δ2 ½1� δ2 ½0� Δδ2 ½1� 0 �1:4569° �1:4569°
¼ þ ¼ þ ¼
δ3 ½1� δ3 ½0� Δδ3 ½1� 0 �0:2324° �0:2324°

Calculation of jV 2 j:
h i � �
½1� �1 1:3
ΔjV 2 j ¼ �½�65� ¼ 0:0200
1:00
h i
ΔjV 2 j½1� ¼ jV 2 j½0� þ ΔjV 2 j½1� ¼ 1:00 þ 0:0200 ¼ 1:020 pu

Iteration #2: considering V 2 ½1� ¼ 1:020 ∠ � 1:4569° and V 3 ½0� ¼ 1:03 ∠ � 0:2324° .
The calculated quantities:
2 3 2 3
P2 ½2� �1:6671
6 ½2� 7 6 7
4 P3 5 ¼ 4 1:2133 5
Q 2 ½2� �0:0239

The mismatch power matrix:

2 3 2 3 2 3 2 3
ΔP2 ½2� �2 �1:6671 �0:3329
6 7 6 7 6 7 6 7
4 ΔP3 ½2� 5 ¼ 4 1:5 5 � 4 1:2133 5 ¼ 4 0:2867 5
ΔQ 2 ½2� �0:5 �0:0239 �0:4761

81
Computational Models in Engineering

Calculation of δ2 and δ3 :
2 3
ΔP2 ½2� 2 3
" # 6 ½1� 7 � ��1 �0:3329
Δδ2 ½2� 0 �1 6 jV 2 j 7
6 7 �65 50 6 1:02 7
6 7
¼ �½B � 6 ½2� 7
¼� 4 0:2867 5
Δδ3 ½2� 4 ΔP3 5 50 �90
jV 3 j½1� 1:03
" # � � " #
Δδ2 ½2� �0:0046 rad �0:0300°
¼ ¼
Δδ3 ½2� 0:0005 rad 0:0286°
" # " # " # " # " # " #
δ2 ½2� δ2 ½1� Δδ2 ½2� �1:4569° �0:0300° �1:7213°
¼ þ ¼ þ ¼
δ3 ½2� δ3 ½1� Δδ3 ½2� �0:2324° 0:0286° �0:2021°

Calculation of jV 2 j:
h i � �
½ 2� �1 �0:4761
ΔjV 2 j ¼ �½�65� ¼ �0:0072
1:02
h i
jV 2 j½2� ¼ jV 2 j½1� þ ΔjV 2 j½2� ¼ 1:02 � 0:0072 ¼ 1:0128 pu

The remaining five iterations are shown in Table 4. It shows that this method
converges slower than Newton-Raphson method.

Iteration V 2 ½k� V 3 ½k�

1 1:0200 ∠ � 1:4569° 1:03 ∠ � 0:2324°

2 1:0128 ∠ � 1:7213° 1:03 ∠ � 0:2021°

3 1:0111 ∠ � 1:6019° 1:03 ∠ � 0:2029°

4 1:0118 ∠ � 1:5759° 1:03 ∠ � 0:2026°

5 1:0119 ∠ � 1:5877 ° 1:03 ∠ � 0:2027 °

Table 4.
Fast-decoupled iterative solution.

6. DC power flow technique

DC power is an extension to the Fast-decoupled power flow formulation. In DC

power flow method, the voltage is assumed constant at all buses; therefore,
theðΔjV j, ΔQ ) equation is neglected. The (Δδ, ΔP) equation can be further simpli-
fied to a linear problem that does not require iterative solution:

�½B�½δ� ¼ ½P� (32)

Or

½δ� ¼ �½B��1 ½P� (33)

Example 5: DC power flow.

Solve the power flow problem shown in Figure 3 using the DC power flow
technique.

82
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

Solution:
The first step in DC power flow technique is obtaining ½B� from the Y-bus.
2 3
15 � j55 �5 þ j15 �10 þ j40
6 7
Y ¼ 4 �5 þ j15 20 � j65 �15 þ j50 5
�10 þ j40 �15 þ j50 25 � j90

For this problem, Since the unknown voltage angles are δ2 and δ3 , elements for
½B� are obtained from the Y-bus as follows:
� �
�65 50
½B� ¼
50 �90
" # " #�1 " # " #�1 " #
δ2 �65 50 P2 �65 50 �2
¼� ¼�
δ3 50 �90 P3 50 �90 1:5
" # " °
#
�0:0313 rad �1:7958
¼ ¼
�0:0007 rad �0:0428°

7. Slack bus power and losses calculations

The main objective of power flow calculations is to determine the voltages

(magnitude and angle) for a given load and generation conditions. Once voltages
are known for all buses, slack bus power, as well as line flows and losses, can be
calculated. The slack bus real and reactive power are calculated using Eqs. (4) and
(5), respectively. Overall system losses are the difference between generation and
load.

SL ¼ SGen � SLoad (34)

Specific branch losses are calculated using branch power flow. For example, the
losses in the line i – j are the algebraic sum of the power flow.

SL ij ¼ Sij þ Sji (35)

Sij and Sji are defined as follows:

Sij ¼ V i Iij ∗ (36)

Sji ¼ V j Iji ∗ (37)

The current between buses Iij is a function of the voltages and the admittance
between V i and V j . It is worth noting that Iji ¼ �Iij
� �
Iij ¼ V i � V j yij (38)

Example 6: Slack bus power and losses.

For the 3-bus system shown in Figure 3, the voltages at buses 2 and 3 were
iteratively calculated: V 1 ¼ 1:02 ∠ 0° , V 2 ¼ 1:0118 ∠ � 1:5871° and
V 3 ¼ 1:03 ∠ � 0:2005°

83
Computational Models in Engineering

a. Calculate the slack bus power.

b.Calculate the total system losses.

c. Calculate individual branch losses.

Solution:
The polar form of the Y-bus is used.
2 3
57:01 ∠ � 74:74° 15:81 ∠ 108:43° 41:23 ∠ 104:04°
6 7
Y ¼ 4 15:81 ∠ 108:43° 68:01 ∠ � 72:90° 52:20 ∠ 106:70° 5
41:23 ∠ 104:04° 52:20 ∠ 106:70° 93:41 ∠ � 74:48°

a. The slack bus power

The slack bus power can be calculated using real and reactive power equations:
n � �� � � �
P1 ¼ ∑ jV 1 j�V j ��Y 1j � cos θ1j � δ1 þ δj
j¼1

¼ jV 1 j2 jY 11 j cos ðθ11 Þ þ jV 1 jjV 2 jjY 12 j cos ðθ12 � δ1 þ δ2 Þ þ jV 1 jjV 3 jjY 13 j cos ðθ13 � δ1 þ δ3 Þ
¼ 0:5195 pu
n � �� � � �
Q 1 ¼ �∑ jV 1 j�V j ��Y 1j � sin θ1j � δ1 þ δj
j¼1

¼ �jV 1 j2 jY 11 j sin ðθ11 Þ � jV 1 jjV 2 jjY 12 j cos ðθ12 � δ1 þ δ2 Þ � jV 1 jjV 3 jjY 13 j sin ðθ13 � δ1 þ δ3 Þ
¼ �0:4572 pu

b.The total system losses

The total real power losses can be calculated using the total net injected real a
power at all buses:

PL ¼ P1 þ P2 þ P3
PL ¼ P1 þ P2 þ P3 ¼ 0:5195 � 2 þ 1:5 ¼ 0:0195 pu

Similarly, the reactive power losses can be calculated.

QL ¼ Q1 þ Q2 þ Q3

However, Q 3 is unknown and can be calculated based on the given bus voltages:
n � �� � � �
Q 3 ¼ �∑ jV 3 j�V j ��Y 3j � sin θ3j � δ3 þ δj
j¼1

¼ �jV 3 jjV 1 jjY 31 j sin ðθ31 � δ3 þ δ1 Þ � jV 3 jjV 2 jjY 32 j sin ðθ32 � δ3 þ δ2 Þ � jV 3 j2 jY 33 j sin ðθ33 Þ
¼ 1:0220 pu

Once Q 2 is calculated, the total reactive power losses are calculated:

Q L ¼ Q 1 þ Q 2 þ Q 3 ¼ �0:4572 � 0:5 þ 1:0220 ¼ 0:0648 pu

84
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

Figure 6.
Power flow results.

c. Branch losses

To calculate the losses in line 1–2, SL 12 is calculated by summing S12 and S21 .
These flows are calculated as follows: S12 ¼ V 1 I12 ∗ and S21 ¼ V 2 I21 ∗
 
I12 ¼ ðV 1 � V 2 Þy12 ¼ 1:02 ∠ 0° � 1:0118 ∠ � 1:5871° ð�5 þ j15Þ ¼ 0:4635 þ j0:0121 pu
I21 ¼ ðV 2 � V 1 Þy21 ¼ �I12 ¼ �0:4635 � j0:0121 pu
S12 ¼ V 1 I12 ∗ ¼ 1:02 ∠ 0° � ð0:4635 � j0:0121Þ ¼ 0:4728 � j 0:0123 pu
S21 ¼ V 2 I21 ∗ ¼ 1:0118 ∠ � 1:5871° � ð�0:4635 þ j0:0121Þ ¼ �0:4685 þ j0:0252 pu
SL 12 ¼ S12 þ S12 ¼ ð0:4728 � j 0:0123 Þ þ ð�0:4685 þ j0:0252Þ ¼ 0:0043 þ j0:0129pu

Similarly, power flow and losses in other branches are calculated.

Line 1–3:

S31 ¼ �0:0456 þ j0:4494

S13 ¼ 0:0467 � j 0:4449
SL 13 ¼ 0:0011 þ j 0:0045

Line 2–3:

S23 ¼ �1:5315 � j0:5252

S32 ¼ 1:5456 þ j0:5722
SL 23 ¼ 0:0141 þ j 0:0470

Total losses:

SL ¼ SL 13 þ SL 23 þ SL 12 ¼ 0:0195 þ j0:0648 pu

The generation and load at different buses is shown in Figure 6.

8. Conclusions

Power flow analysis, or load flow analysis, has a wide range of applications in
power systems operation and planning. This chapter presents an overview of the
power flow problem, its formulation as well as different solution methods. The

85
Computational Models in Engineering

power flow model of a power system can be built using the relevant network, load,
and generation data. Outputs of the power flow model include voltages (magnitude
and angles) at different buses. Once nodal voltages are calculated, real and reactive
power flows in different network branches can be calculated. The calculation of
branch power flows enables technical loss calculation in different network
branches, as well as the total system technical losses.
Power flow analysis is performed by solving nodal power balance equations.
Since these equations are nonlinear, iterative techniques such as the Gauss-Seidel,
the Newton-Raphson, and the fast-decoupled power flow methods are commonly
used to solve this problem. In general, the Gauss-Seidel method is simple but
converges slower than the Newton-Raphson method. However, the latter method
required the Jacobian matrix formation of at every iteration. The fast-decoupled
power flow method is a simplified version of the Newton-Raphson method. This
simplification is achieved in two steps: 1) decoupling real and reactive power cal-
culations; 2) obtaining of the Jacobian matrix elements directly from the Y-bus
matrix. The DC power method is an extension to the fast-decoupled power flow
formulation. In DC power flow method, the voltage is assumed constant at all buses
and the problem becomes linear.

Acknowledgements

The author would like to acknowledge the financial support and thank the Rural
Area Electricity Company and Sultan Qaboos University for sponsoring the publi-
cation of this chapter under project number CR/ENG/ECED/16/02.

Author details

Mohammed Albadi
Sultan Qaboos University, Muscat, Oman

*Address all correspondence to: mbadi@squ.edu.om

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

86
Power Flow Analysis
DOI: http://dx.doi.org/10.5772/intechopen.83374

References

[1] Bergen AR. Power Systems Analysis.

Delhi, India: Pearson Education India;
2009

[2] Glover JD, Sarma MS. Power System

Analysis and Design: With Personal
Computer Applications. Toronto, ON,
Canada: International Thomson
Publishing Company; 1994

[3] Glover JD, Sarma MS, Overbye T.

Power System Analysis & Design, SI
Version. Noida, India: Cengage
Learning; 2012

[4] Grainger JJ, Stevenson WD,

Stevenson WD. Power System Analysis.
New York, USA; 2003

[5] Gross CA. Power System Analysis.

Vol. 1. New York: Wiley; 1986

[6] Saadat H. Power System Analysis.

Vol. 232. Singapore: WCB/McGraw-Hill;
1999

87
Chapter 6

Modeling Emerging
Semiconductor Devices for
Circuit Simulation
Md Sakib Hasan, Mst Shamim Ara Shawkat, Sherif Amer,
Syed Kamrul Islam, Nicole McFarlane and Garrett S. Rose

Abstract

Circuit simulation is an indispensable part of modern IC design. The significant

cost of fabrication has driven researchers to verify the chip functionality through
simulation before submitting the design for final fabrication. With the impending
end of Moore’s Law, researchers all over the world are looking for new devices with
enhanced functionality. A plethora of promising emerging devices has been pro-
posed in recent years. In order to leverage the full potential of such devices, circuit
designers need fast, reliable models for SPICE simulation to explore different appli-
cations. Most of these new devices have complex underlying physical mechanism
rendering the model development an extremely challenging task. For the models to
be of practical use, they have to enable fast and accurate simulation that rules out
the possibility of numerically solving a system of partial differential equations to
arrive at a solution. In this chapter, we show how different modeling approaches
can be used to simulate three emerging semiconductor devices namely, silicon- on-
insulator four gate transistor(G4FET), perimeter gated single photon avalanche
diode (PG-SPAD) and insulator-metal transistor (IMT) device with volatile
memristance. All the models have been verified against experimental /TCAD data
and implemented in commercial circuit simulator.

Keywords: silicon-on-insulator (SOI), multiple-gate transistor, G4FET,

semiconductor device, spline interpolation, macromodel, silicon photomultiplier,
single photon avalanche diode, circuit design, insulator metal transition, IMT,
compact model, SPICE, neuron, Mott transition

1. Introduction

Circuit simulation is an indispensable part of modern IC design. The significant

cost of fabrication has driven researchers to verify the chip functionality through
simulation before submitting the design for final fabrication. With the impending
end of Moore’s Law, researchers all over the world are now looking for new devices
with enhanced functionality and a plethora of promising emerging devices has been
proposed in recent years. To leverage the full potential of such devices, circuit
designers need fast, reliable models for SPICE simulation to explore different appli-
cations. Most of these new devices have complex underlying physical mechanism
rendering the model development an extremely challenging task. To be of practical

89
Computational Models in Engineering

use, the model has to enable fast and accurate simulation, which rules out the
possibility of numerically solving differential equations underlying the physics of
semiconductor devices to arrive at a solution. In this chapter, we show how differ-
ent approaches can be adopted to model three emerging semiconductor devices,
namely silicon-on-insulator four-gate transistor (G4FET), single-photon avalanche
diode (SPAD), and insulator-metal transistor (IMT) device with volatile
memristance.

2. Background

The amazing technological advancements in the semiconductor industries dur-

ing last 60 years have largely shaped the modern world we live in. However, bulk
silicon devices are now faced with some fundamental physical limits and innovative
researchers and scientists all over the world have been diligently exploring new
devices and computing paradigms to continue the breathtaking pace of technology
advancement. The widespread use of a technology in circuit design is heavily
dependent upon good SPICE models for CAD (computer-aided design) tools that
are now ubiquitous in circuit design. Sophisticated models for existing semicon-
ductor devices integrated with CAD tools have enabled designers worldwide to
design innovative and reliable circuits which are in a large part responsible for the
technology boom of the last several decades. In this chapter, we will be discussing
three relatively new devices highlighting the works and challenges involved in
model development and SPICE simulation.
The first device is SOI (silicon-on-insulator) four-gate transistor known as
G4FET. Here, we outline its operating mechanism, different approaches towards
modeling and highlight two particular approaches, namely multivariate cubic spine
interpolation model and MOS-JFET macromodel. The second device is a single
photon avalanche diode (SPAD). We review and discuss the most recent develop-
ments in SPAD modeling with particular emphasis on a new class of SPAD known
as PGSPAD (perimeter gated SPAD). The third device is insulator metal transition
(IMT) device. Here, we focus on a very recent IMT model to demonstrate how
proper simplification of device physics can lead to a suitable SPICE model
facilitating circuit design.

3. Silicon-on-insulator (SOI) four gate field effect transistor (G4FET)

G4FET was first reported in 2002 [1]. It has four independent gates, two of
which provide vertical MOS (metal-oxide-semiconductor) field-effect action
whereas the other two gates provide lateral junction field effect transistor (JFET)
functionality. The unique G4FET structure can be leveraged to design circuits for
different analog, mixed-signal and digital applications with significantly reduced
transistor counts. Some of these have already been experimentally demonstrated
including LC oscillators and Schmitt trigger circuit with adjustable hysteresis using
negative differential resistance [2], four quadrant analog multipliers [3], adjustable
threshold inverters, real time reconfigurable logic gates and DRAM cell [4], and
universal and programmable logic gate capable of highly efficient full adder design
[5]. G4FET inspired multiple state electrostatically formed nanowires have already
been used for threshold logic functions [6] and high-sensitivity gas sensing [7].
Another potential application is to leverage the tunable tent map shape characteris-
tics of the voltage controlled G4NDR circuit [2] to build chaotic logic gates for
hardware security applications [8, 9].

90
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

3.1 Device structure

One particular advantage of G4FET is that standard SOI process without signif-
icant modification can be used to fabricate these devices. The carrier conduction
through the channel is modulated using four independent gates. The conventional
source and gates become junction gates (JG1 and JG2) to provide lateral JFET
functionality and the top and bottom gates provide vertical MOS functionality.
Figure 1 shows the 3-D schematic structure of a p-channel G4FET. It is evident that
no specialized fabrication procedure is necessary for this device.

3.2 Conduction mechanism

By putting two highly doped body contacts on the opposite sides of the tradi-
tional channel, we can convert a conventional n-channel SOI MOSFET into a p-
channel G4FET. What used to be n + doped source and drain now act as lateral
JFET-like gates. These gates can modulate the width of the conduction channel
according to applied bias. The functionality of the conventional MOS gate is
achieved through the top oxide gate. The substrate along the buried oxide provides
an additional bottom-gate functionality. A high positive voltage on top gate creates
an inversion layer of electrons near the top surface. As we reduce the top gate
voltage, the top surface gradually changes from inversion to depletion, and eventu-
ally, with high negative voltage, it will enter the accumulation region. Similar effect
due to bottom-gate can be observed for the surface near the buried oxide interface.
In order to ensure ohmic contact with the conducting channel, body contacts on the
opposite side of the channel are highly doped. Through this process, an inversion-
mode, n-channel MOSFET can be transformed into an accumulation/depletion-
mode p-channel G4FET. A similar mirror transformation will convert a classic SOI
p-channel MOSFET into an n-channel G4FET. The current from drain to source
follows in a direction perpendicular to a traditional MOSFET. Similarly, the geo-
metrical features are also swapped, i.e., the length and width of the classic MOSFET
now become the width and length of the new G4FET, respectively. As outlined here,
there are three possible paths for current conduction, namely (1) top surface near
gate oxide interface, mostly controlled by top gate bias, (2) bottom surface near

Figure 1.
Schematic device structure of a p-channel G4FET.

91
Computational Models in Engineering

buried oxide interface, mostly controlled by bottom gate bias, and (3) volume
conduction away from vertical oxide interfaces, mostly controlled by lateral gates.

3.3 State-of-the-art G4FET models

Different approaches have been adopted to understand and model the operation
of G4FET. Extraction methods for threshold voltage, mobility, and subthreshold
swing in the linear region were demonstrated in [10]. The experimental results
from a partially-depleted (PD)-SOI G4FET showed the dependence of these
parameters on different gate biases. The charge coupling between front, back, and
lateral junction-gates was considered and a 2-D analytical relationship for the fully-
depleted body potential was derived in [11]. Here, a closed form front-interface
threshold voltage expression was derived as a function of the back and the lateral
gate voltages for different back interface conditions such as accumulation, depletion
and inversion.
A very interesting application of G4FET is the formation of quantum wire. The
quantum wire can be electrostatically formed when the conducting channel is
surrounded by depletion regions induced by vertical MOS and lateral JFET gates
[12]. In this unique conduction mechanism named depletion-all-around (DAA),
majority carriers flow in the volume of the silicon film far from the silicon/oxide
interfaces. The control of lateral gates on the conduction channel can be adjusted by
changing biases on the vertical gates. There is a reduced sensitivity of the channel to
the oxide and interface defects, low subthreshold swing, high gm/ID ratio, high
mobility, low noise, and high immunity to ionizing radiation [13]. A low frequency
noise model combining volume and surface noise sources was presented in [14] and
effects of gate oxide and junction nonuniformity on the DC and low-frequency
noise performance were investigated in [15].
A charge sheet model has been proposed to analyze the transistor characteristics
of fully-depleted G4FETs [16]. Here, surface accumulation behavior, drain current
and gate capacitance of fully-depleted G4FET are modeled analytically. In [17], a
mathematical model is developed to determine the subthreshold swing of thin-film
fully-depleted G4FET. Based on the exact solution of the Poisson equation, a new
two-dimensional model of potential and threshold voltage for the fully-depleted
G4FET was developed in [18]. Another mathematical model is developed in [19] to
determine the 3-D potential distribution of a fully-depleted G4FET.
Several numerical methods have also been used for modeling G4FET.
Multidimensional Lagrange and Bernstein polynomial approaches have been used in
[20] for modeling and SPICE implementation. The authors in [21] have reported
single multivariate regression polynomial approach for modeling the device opera-
tion across different operating regimes. We will focus on two new techniques in the
next two sections. The first one is based on multivariate cubic spline interpolation
method for DC modeling first reported in [22]. The second one is a MOS-JFET
macromodel suitable for DC, AC and transient simulation [23].

3.4 Multivariate cubic spline interpolation model

Developing a interpolating polynomial f(x) requires fitting a polynomial to a set

of chosen data points such as (x0, y0), (x1, y1) … (xm, ym). It can be written as,

yi ¼ f ðxi Þ; i ¼ 1, 2, …:, m (1)

Spline based interpolating polynomial is a piecewise low-degree polynomial

which solves some of the problems associated with a single high degree interpolant.

92
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

If the dataset is regular and monotonic, this interpolation method can get rid of an
important problem of single high degree interpolating polynomial known as Runge’s
phenomenon, i.e., oscillation between interpolation nodes, which can be cata-
strophic close to the boundary. With the suitable choice of spline, it is also possible
to reduce error while maintain continuity of the function and its first and second
order derivatives. For these reasons, multivariate cubic spline polynomials are used
for the model development [24].
Given n data points, (xk, yk), k = 1, 2,…, n with distinct xk’s, there is a unique
polynomial in x of degree less than n whose graph passes through the points. If two
successive points are (xk, yk) and (xk + 1, yk + 1) then the kth interval between these
two points can be interpolated using splines. The interval index k is such that, xk ≤ x
< xk + 1. The local variable, s is s = x – xk. The first divided difference is
δk = (yk + 1 � yk)/(xk + 1 � xk). Let hk denote the length of kth subinterval, i.e.,
hk = xk + 1� xk; then, δk = (yk + 1 � yk)/hk. Let dk denote the slope of the interpolant at
xk, i.e., dk = P0 (xk). The cubic spline polynomial will be the following function on
the interval xk ≤ x ≤ xk + 1, written in terms of local variables s = x – xk and h = hk.

3hs2 � 2s3 h3 � 3hs2 þ 2s3 s2 ð s � h Þ sð s � h Þ 2

PðxÞ ¼ y kþ1 þ y k þ dkþ1 þ dk (2)
h3 h3 h2 h2

The cubic polynomial in Eq. (2) satisfies four conditions for smooth interpola-
tion; two on the function values and two on the derivative values so that,

Pðxk Þ ¼ yk , Pðxkþ1 Þ ¼ ykþ1 , P0 ðxk Þ ¼ dk, P0 ðxkþ1 Þ ¼ dkþ1

In order to ensure continuous second derivative, we can add an extra constraint

which gives rise to the following equation,

hk dk�1 þ 2ðhk�1 þ hk Þdk þ hk�1 dkþ1 ¼ 3ðhk δk�1 þ hk�1 δk Þ (3)

If knots are equally spaced, Eq. (4) becomes

dk�1 þ 4dk þ dkþ1 ¼ 3δk�1 þ 3δk (4)

Following the above-mentioned procedure at every interior knot xk (k = 2,. .,

n � 1), results in n � 2 equations with n unknown dk’s. At the interval boundary,
“not-a-knot” method is used, i.e., a single cubic polynomial is used on first two
subintervals, x1 ≤ x ≤ x3, and last two subintervals, xn�2 ≤ x ≤ xn. Now, these two
boundary conditions give two more equations resulting in n unknowns with n
linear equations. Then dk’s are estimated by solving this system of linear algebraic
equations.
For cases involving multiple independent variables, tensor product formulation
can be used to extend the univariate analysis to multiple dimensions. In each
respective dimension, value at a desired point is interpolated according to a cubic
interpolation of the values at nearby knot points. In this section, we are using VDS
(drain-source voltage), VJG (common junction-gate voltage), VTG (top-gate
voltage) and VBG (bottom-gate voltage) as our independent variables for spline
interpolation of drain current. Moreover, the geometric features such as W (width)
and L (length) may be used as independent variables to include the effect of device
geometry.
An n-channel partially-depleted SOI (PDSOI) G4FET was developed in a device
simulator (TCAD Sentaurus from Synopsys) and a cubic spline model was formu-
lated based on the generated data. In Figure 2, drain current versus drain voltage is

93
Computational Models in Engineering

Figure 2.
Comparison between isolines of test data and cubic spline model for different junction-gate voltages ranging
from �4 V to 0 V in 1 V increment arranged from bottom to top [22].

plotted for different values of junction-gate voltages ranging from 0 to �4 V. The

bottom gate/substrate bias, VBG is 0 V and top-gate bias, VTG is 2.25 V, respectively.
Both junction gates are tied together for simplicity. From the graph and the values
of corresponding mean relative error, it is clear that the model fits reasonably well
for all the isolines. More details on this method can be found in [22].

3.5 MOS-JFET macromodel

G4FET has also been called MOSJFET [1] since it combines both metal-oxide-
semiconductor field-effect transistor (MOSFET) and junction field-effect transistor
(JFET) actions in a single silicon island. As mentioned earlier, the vertical gates
provide MOS functionality and the lateral gates add JFET functionality. By applying
different junction-gate biases, it is possible to modulate the threshold voltages of
top- and bottom-gates. As a result, G4FET can be thought of as a combination of
two MOSFETs primarily responsible for top and bottom surface conduction and a
dual-gate JFET that is primarily responsible for volume conduction. The top-gate
threshold voltage is VTH and the bottom-gate voltage causing the onset of accumu-
lation and inversion at the bottom-gate are V acc inv
BG and V BG , respectively. Some of the
terms used in the model are introduced below:
Junction-gate, top gate and bottom gate capacitances are CJG ¼ εSi=w ,
Cox1 ¼ εox =tox1 and Cox2 ¼ εox =tox2 , respectively. Three constants based on device
pffiffi γCJG = pffiffi
2� p2ffi � 2� p2ffi �
geometry, α, β and γ are defined as, α ¼ 2 2tSi
, β ¼ 1þαCJG =Cox1 , γ ¼ 2 2t
.
tanh W Cox2 sinh W Si
� � � �
E qN W 2
Other terms include φF ¼ �V T ln Nnid , φb ¼ 2g þ V T ln Nnid , V P ¼ φb � 8εd Si .
Here, W = width of the transistor, tsi = silicon film thickness, tox1 = top oxide
thickness, tox2 = buried oxide thickness, VT = kT/q is the thermal voltage, Nd = donor
concentration in the body, ni = intrinsic carrier concentration, εsi = permittivity of
silicon, and εox = permittivity of silicon dioxide.
The onset voltage of accumulation and inversion for the bottom-gate, V acc BG and
inv
V BG , can be expressed [27] as,

94
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

CJG  
V acc
BG ¼ V FB2 þ ðγ � αÞ V JG � V P (5)
Cox2
 
CJG CJG  
V inv
BG ¼ V FB2 þ 1 þ α 2φF � ðγ � αÞ ðV P Þ þ 1 þ γCJG =Cox2 V JG (6)
Cox2 Cox2

The back gate may be accumulated, depleted or inverted. When the bottom-gate
is in inversion, i.e., V BG < V inv
BG ,
   
CJG CJG  
V TH ¼ V FB1 � γ ð2φF þ V P Þ � α V JG � V P (7)
Cox1 Cox1

When the bottom-gate is depleted, i.e., V inv acc

BG < V BG < V BG ,
 
CJG CJG  
V TH ¼ V FB1 � βðV BG � V FB2 Þ þ ðγ � αÞ þβ V JG � V P (8)
Cox1 Cox1

When the bottom-gate is in accumulation, i.e., V BG >V acc

BG ,
 
CJG  
V TH ¼ V FB1 þ ðγ � αÞ V JG � V P (9)
Cox1

Here, VFB1 and VFB2 are the flat band voltages of the top-gate and the bottom
gates, respectively.
Based on the above relationships among different gates, a macromodel
subcircuit is created combining conventional SPICE Level-1 MOSFET and the
JFET models which follow from the quadratic FET model of Schichman and Hodges
[25]. It should be noted that the model can be improved by using higher-level
MOSFET models (Level-2/3) and BSIM/BSIMSOI models for devices with deep
submicron geometry. Here, the goal of this work, with simple first order MOSFET
and JFET models, is to show the feasibility of the macromodel and demonstrates
how it can capture the essential physics underlying the complex interaction
between multiple gates. Here, the top conduction is modeled using a MOSFET and
the volume conduction is modeled using a JFET. However, instead of a constant
threshold MOSFET, the subcircuit allows for threshold voltage modification based
on multiple gate biases using the relationship described above. The transient simu-
lation results for the first analog multiplier circuit configuration in [3] is shown in
Figure 3. The simulation result using the macromodel shows good matching with
the experimental result [3].

Figure 3.
Transient simulation results for product of a 20 Hz, 1 Vp-p sinusoidal-wave with 500 Hz, 1 Vp-p square-wave
(W = 0.3 μm � 10, L = 2.4 μm, VDD = 3.5 V, VSS = �3.5 V, Ibias = 35 μA, Vbias1 = 2 V, Vbias2 = �2.5 V,
RL = 100 kΩ) using analog multiplier; (a) input, (b) output [3].

95
Computational Models in Engineering

4. Single photon avalanche diode (SPAD)

Single photon avalanche diodes (SPADs) based optical detectors have gained
interest for use in a wide range of applications such as biochemical analysis,
imaging and light ranging applications [26–32]. The main causes of the increased
popularity are the exceptional level of miniaturization and portability, overall high
performance and low costs due to the integration of SPADs with mixed-signal
readout circuits in standard complementary metal-oxide-semiconductor (CMOS)
technology [26–32]. The capability to detect single photon and provide sub-
nanosecond time resolution along with the low-power and high-speed CMOS read-
out circuits have made SPADs superior to other optical detectors in high perfor-
mance weak optical signal detection applications. In order to correctly simulate
the behavior of the SPAD device and the performance of its readout circuit during
the design phase, a suitable and comprehensive model of the SPAD is required. In
this work, we review and discuss the most recent developments in SPAD modeling
to predict the real behavior of the CMOS SPAD systems before fabrication
reducing the design cycle.

4.1 SPAD theory and operation

A single photon avalanche diode (SPAD) is a p-n junction, which is biased above
the breakdown voltage. The device operates in the so-called Geiger mode. With the
increase of reverse bias voltage beyond a critical potential, the current increases
rapidly. The rapid increase in current is due to the free carriers, which have gained
enough energy to ionize the fixed lattice atoms and free other carriers through
high-energy collisions. This rapid multiplication process results in a sudden large
avalanche current. The reverse voltage above which this multiplication process
occurs is called breakdown voltage [26–32]. In order to respond to an incident
photon, impact ionization is necessary for a SPAD. A high electric field is required
to break the covalent bonds and generate the free carriers, which take part in the
avalanche process. The probability that carriers gain the threshold energy is a
function of the local electric field and previous states of the carrier. This probability
determines the ionization rates. Figure 4 illustrates impact ionization causing an
avalanche current in the presence of a high electric field. A number of free carriers
is generated at each step, quickly multiplying and generating a large current [10].
Free carriers are also generated in absence of photon due to inherent noise
processes such as thermal generation, minority carrier diffusion, and band-to-band

Figure 4.
Impact ionization causing avalanche current in existence of high electric field where a number of free carriers
are multiplied in each step.

96
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

tunneling. Dark count rate is the effective noise for SPADs caused by these
non-photon driven carrier generations. The large avalanche current must be
quenched to protect the long-term use of the device. An external circuit, typically a
passive ballast resistor or a transistor, is employed to quench the large avalanche
current. As the current increases the voltage drop across the quenching resistor
increases. As a result, the bias voltage across the SPAD reduces below the break-
down voltage ending the avalanche process. The SPAD is then recharged and ready
to detect photon event induced avalanche. The avalanche gain is large enough to be
almost infinite in this Geiger mode of operation. The generation of an electron-hole
pairs causes the device to be in an “ON” or “OFF” state. Since the generation of an
avalanche gives a current or voltage spike, the device exhibits an inherently digital
operation [26–32]. Figure 5 shows the Geiger mode operation of SPAD and
quenching of the avalanche current. Here, Vbr represents the breakdown voltage
and Vs is the applied voltage. During the current flow, some carriers may be
trapped in the deep-levels and are released after a characteristics lifetime later.
The released carrier may trigger a new avalanche process when the SPAD is ready
to detect another photon. The new avalanche process associated with trapped
carriers during the first avalanche pulse causes after-pulses and it reduces detector
performance [26].
SPADs fabricated in CMOS provides the benefits of commercial CMOS process
such as low cost, miniaturization, and improved performance [27–32]. However,
due to the planar nature of the junction in standard CMOS process, electric field
distributions exhibit maxima at the diodes’ edge [27–32]. Since SPADs operate in a
high reverse bias region, they are more vulnerable to breakdown at the edges.
Therefore, full volumetric breakdown cannot be achieved as the diode periphery
undergoes breakdown earlier compared to the other parts of the diode. As a result,
the active area is decreased and the photon detection efficiency is significantly
compromised. This premature breakdown is one of the major problems of SPADs
implemented in standard CMOS process [27–32]. It has been previously proved that
the perimeter gated SPAD (PGSPAD), a p-n junction incorporating a poly-silicon
gate surrounding the periphery, suppresses the premature breakdown effectively
[29, 31, 33–39]. The applied voltage at the gate terminal modulates the electric field
making it uniform throughout the junction preventing the premature breakdown.

4.2 State-of-the-art SPAD models

Various SPAD models have been presented in literature to predict the actual
behavior and choose suitable parameters during the design phase to ensure the

Figure 5.
Geiger mode operation of SPAD; the avalanche current is quenched using a quenching resistor.

97
Computational Models in Engineering

optimal performance for CMOS SPAD systems in existing and new applications
[29, 35–47]. State of the art SPAD modeling approaches can be grouped in three sub
groups named as model based on device physics, circuit simulation model, and
model based on information theory. The main features of the models of each group
along with the limitations are discussed below.

4.2.1 Models based on device physics

The SPAD models in this group have been developed to simulate the SPADs at
the semiconductor level based on purely quantum mechanical or semi-classical
representation of the transport properties of semiconductor structure underlying
the SPAD device [29, 35–37]. The key operational features such as breakdown
voltage, avalanche gain, and noise have been generated through the model.
In [35], a new compact numerical model to simulate the forward and reverse dc
diode characteristics has been presented. The model is based on the solution of 1D
steady-state hole continuity equation in the depletion layer of a p-n junction. The
developed dc model includes the physical mechanisms such as band-to-band
tunneling, trap-assisted tunneling, Shockley-Read-Hall recombination, and ava-
lanche breakdown.
The model presented in [36], focuses on breakdown state stability studying the
SPAD transient behavior based on physical and numerical approach. At first, the
spatial steady-state free charge carrier distribution during breakdown is computed.
Then the temporal change of the distribution is considered. It explores the depen-
dence of the probability for the avalanche process on the quenching resistance. It
aids designers to accurately choose the optimal passive quenching resistance for
minimizing the overall dead time value.
The reported model in [37] investigates the lateral spreading of the avalanche in
Geiger mode reach-through avalanche photo diode, which is the main cause of
avalanche dynamics and the time jitter. The model estimates the photon assisted
effect based on the Poisson equation and the time-dependent semiconductor conti-
nuity equation to ensure the avalanche dynamics. A Monte Carlo simulation of the
avalanche current rise has been developed in the model.
The effect of structural effect on the breakdown voltage and the localization of
the avalanche region within p+/n-well junction was determined in model [29]. The
reported 2-D numerical model is based on Poisson, hole and electron current conti-
nuity equation coupled with carrier generation rate equation due to impact ioniza-
tion. The model highlights the regions most susceptible to breakdown through the
determination of the spatial maxima of carrier generation rates and establishes that
the generation rate reaches its peak at the perimeter and the surface of the junction
causing perimeter breakdown. Finally, it demonstrates that full volumetric break-
down can be achieved using perimeter breakdown suppression techniques.

4.2.2 Circuit simulation model

Circuit simulation models aim to represent the equivalent circuit representation

of the SPAD for circuit level designs and simulation. These models provide a first
order estimation of performance at the circuit level and allow simulation of readout
circuits based on the electrical characteristics of SPAD [31, 33, 38–41]. The readout
delays, time constant for both quenching and reset can be calculated using the
circuit simulation model.
Traditional SPAD basic model reported in [38] includes voltage sources, resistors
and capacitors to represent the diode characteristics (Figure 6). In the model, the
diode resistance, RD, includes both space charge resistance and resistance of neutral

98
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

Figure 6.
Traditional SPAD basic model where closing the transistor switch imitates the avalanche triggering [38].

regions crossed by the avalanche current. CAC represents the junction capacitance
and CAS and CCS represents the stray capacitances from anode and cathode respec-
tively. The closure of NMOS switch simulates the triggering of an avalanche due to
photon absorption or other phenomenon including thermal generation or band-
band-band tunneling.
Figure 7 shows the improved version, which includes the triggering, the self-
sustaining process, and the self-quenching of the avalanche by incorporating of
current-voltage controlled switches [39]. In order to represent the nonlinear I-V
characteristics above breakdown, the model includes nonlinear voltage generator,
VSPAD, to generate piecewise linear curve with different slopes.
An accurate model for SPADs [40] exploits the behavioral description of the
sensor with biasing circuits and aims to provide optimal biasing circuit design
through SPAD simulation. In order to represent the SPAD above breakdown and
with the avalanche triggering and self-quenching mechanisms, a complete SPICE
circuit model for SPAD has been presented in [41]. The SPAD is implemented using
a piecewise non-linear approximation and modeled through a non-linear series
resistance above breakdown. The model also includes the forward region and the
secondary breakdown due to edge-junction or punch- through effects (Figure 8).
SPAD stochastic phenomena affect the switching behavior of the device and
define the transition from no-avalanche to avalanche (turn-on) and vice versa. In
order to provide a more realistic simulation platform, more comprehensive models
[34, 42–45] include stochastic nature of Geiger mode operation based on theoretical
equations of dark count rate, breakdown probability, signal-to-noise ratio and after-
pulsing probability.
The behavioral model presented in [42], simulates the noise model due to dark
count rate and after pulsing based on fabricated component for the first time. The
model improves the previous circuit model [41] by adding the noise sources of
avalanche photodiode. The model has been developed using specter active and
passive components to represent the unwanted after-pulse and dark count events.
At first, the parasitic capacitors of the sensors (anode-cathode capacitor, anode-
bulk and the cathode-bulk capacitors) are included to represent the dynamic

99
Computational Models in Engineering

Figure 7.
Improved SPAD model [39]; switch STRIG is used to trigger the detector through “photon” input, SSELF is used to
include self-sustaining and self-quenching of the avalanche, and VSPAD represents the nonlinear I-V curve.

Figure 8.
Complete SPAD model [41] with switches STRIG and SSELF representing avalanche triggering and self-
sustaining; two additional branches are included to represent forward and reverse biasing.

behavior. Three different branches representing forward, reverse and effects of

noise behavior are incorporated to simulate the I-V behavior. Each branch is con-
trolled by an ideal voltage controlled switch. Finally, two noise sources, represented
with two corresponding branches are added to simulate the noise effect (Figure 9).
An accurate behavioral model reported in [43] models the major statistical
behaviors of SPADs including the turn-off probability, dark count rate and after-
pulse phenomena. One the major focus of the model was to explore the dependence
of the device capacitance on the reverse bias.

100
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

Figure 9.
SPAD behavioral model including I-V behavior and noise sources for dark counts and after-pulsing [42].

SPAD is unresponsive to any subsequent photons after detecting the first photon
(during avalanche, quenching and resetting cycle) until it is reset back to undergo
another avalanche. This time duration is known as dead time. In order to accurately
interpret the SPAD measurements, dead time needs to be taken into account espe-
cially for large arrays of CMOS SPAD based detectors. In [44], a dead time model
was reported for externally reset SPADs. Here, time distribution of incidents counts
is assumed to follow Poisson distribution.

4.2.3 Model based on information theory

In order to simulate the behavior of SPAD’s array and predict the behavior at
system level design, model based on information theory include a combination of
the previously mentioned models [47, 48]. These system levels models are helpful
to simulate the operation of multiple SPADs including both high-density and low-
density arrays. Various operational aspects of SPAD arrays such as array breakdown
voltage non-uniformity, pixel cross-talk can be investigated using these models.
Some models are able to study the performance at single-pixel-level [48]. They
consider the SPAD front-end as an information channel. Then the information
capacity and mutual information can be calculated from the information channel
as a function of excess bias voltage, dark count rate, and after pulsing probability.
The information theoretic model reported in [47] includes a channel-capacity
metric to evaluate the performance of SPADs as a detector in laser communication
systems for optimizing its performance with respect to the device structure and
operating voltage.

4.3 State-of-the-art PGSPAD models

A perimeter gated SPAD (PGSPAD) is a SPAD with an additional polysilicon

gate terminal and its effects need to be properly modeled. PGSPAD models were
reported in the literatures to predict the Geiger mode response to a photon; simulate
the static and dynamic I-V characteristics considering the effect of applied voltage
at additional gate terminal [31, 33, 34, 48]. The equivalent circuit models for
PGSPAD [31, 33] simulate the static and dynamic behavior of PGSPADs with the
additional gate terminal. These models are based on the improved SPAD
equivalent model [41]. The PGSPAD model simulates the avalanche in similar way
to SPAD model [41]. In addition, they include a dependent voltage source to
represent the dependence of breakdown voltage (Vbr) on gate voltage (VG) and the
reverse biased resistance as a variable resistance depending on gate bias voltage

101
Computational Models in Engineering

Figure 10.
Equivalent circuit model of perimeter gated single photon avalanche diode (PGSPAD) where terminals A, C, S,
and G represents the anode, cathode, substrate and gate terminals [31].

and reverse bias voltage (Figure 10). Two parallel branches are incorporated to
simulate the forward and reverse bias operation. Model parameters and threshold
voltage of the switches are extracted from the fabricated PGSPAD device. From the
empirical data, it was found that the breakdown voltage depends on zero-bias
breakdown voltage and the gate voltage. The empirical relation [31] of Vbr, reverse
branch resistance (RR), forward branch resistance (RF) with gate voltage (VG)
and anode-cathode biasing (VCA) are,

V br ¼ γV G þ V br0 (10)
RR ¼ P exp ðQ ðV CA � V br0 þ V 1 ÞÞ (11)
RF ¼ R exp ðSV AC Þ (12)

where γ is a fitting constant obtained from experimental perimeter gate voltage

versus breakdown voltage, P, Q , R, and S are constants determined experimentally,
V br0 is the zero bias breakdown voltage.
An improved PGSPAD SPICE model was reported in [34] to simulate the sto-
chastic behavior of PGSPAD. The model correctly simulates the dark count rate
with variation of gate voltage and excess bias voltage, the applied reverse biased
voltage in excess of breakdown voltage. The model includes the dark current noise
due to non-photon driven carrier generation mechanisms and the spectral response
with the additional gate terminal. It uses Shockley-Hall-Read equation and tunnel-
ing current equation to calculate the thermal carrier generation rate and carrier
generation rate due to band-to-band tunneling respectively. Finally, the model
established that the band-to-band tunneling mechanism depends on the gate bias
voltage through the electric field. The PGSPAD model presented in [48] evaluates
the performance of PGSPAD using the information theory principles. The model
estimates the optimization of the information rate of the PGSPAD by applying the
asymmetric communication channel model to the device.

102
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

5. IMT

Insulator Metal Transition (IMT) devices are volatile memristors that have
recently spurred significant interest in the research community [49]. In crossbar
memory arrays, IMT devices are often leveraged as selector devices owing to their
Back-End-Of-Line (BEOL) compatibility and high ON/OFF ratio [50, 51]. On the
neuromorhpic front, the inherent switching dynamics of IMT devices have been
shown to mimic the Integrate-And-Fire neurons which alleviates the need for
complex CMOS circuitry, thus, enabling higher integration density [52, 53]. In
addition, volatile memristors with short-term memory [54–56] can also be used as
synapses for neuromorphic application.

5.1 Device operation

Several studies have shown that temperature is the main driver of phase transi-
tion [57, 58] while other works argue that the electric field is the main driver of
resistive switching while the Joule heating plays a secondary role [59]. A more
thorough study about IMT’s switching mechanism is reported in [60, 61] which
show that Joule heating is not sufficient to instigate switching and that an Electric
field assisted phase transition is more likely. The work in [60] proposes that a
certain threshold voltage is required to effect a phase transition which decreases as
temperature increases.

5.2 Motivation behind compact model

In [52], an IMT device model that captures the positive feedback between the
temperature and electric field was presented. Here, the relationship between the
device resistance and the device temperature was implemented using a look up
table. In [62], another device model was proposed based on band theory. The IMT
device is modeled as a low bandgap semiconductor. Increasing the device tempera-
ture results in decreasing the bandgap, thus, increasing carrier concentration. This
increase in carrier concentration ultimately results in decreasing the device’s resis-
tance. A model is also presented which captures the change in the thermal conduc-
tivity with temperature. The thermal conductivity model along with the band gap
model are then solved in a self-consistent fashion to effect a temperature dependent
phase transition. The model in [62] was implemented in Sentaurus TCAD simulator
wherein the built-in electrothermal models and finite element drift-diffusion model
where leveraged. This model, similar to the previous one, is best used in a TCAD
simulation flow and not SPICE level simulators. The lack of a physics-based com-
pact model, however, has hindered circuit designers from exploring the full poten-
tial of IMTs in circuit applications which is the motivation behind the development
of the following simplified compact model.

5.3 Simplified compact model

This work focuses on an IMT compact model originally proposed in [63]. The
model describes the IMT device as a memristor with the device’s local temperature
being the state variable. The proposed model is simulated using Specter from
Cadence and exhibits a close match to experimental data and electro-thermal simu-
lations based on the models in [52, 62].
As the current flows through the device, the device’s temperature increases until
it reaches a critical temperature at which point it transitions from a high resistance

103
Computational Models in Engineering

state to a low resistance state. As the temperature decreases, the resistance relaxes
back to its default high resistance state. The memristive dynamics of the IMT can be
expressed as follows [64].

V ¼ RðxÞI (13)

dx
¼ g ðx, V Þ (14)
dt

where (13) and (14) represent the port and state equations, respectively, and x is
the state variable. The proposed model consists of two governing equations: (a) the
resistance change equation that corresponds to the port equation and (b) the tem-
perature evolution equation which corresponds to the state equation, with the
temperature being the state variable such that x = T(t).
The behavior of the resistance change versus temperature is described by two
thermistor states for the high resistance state and the low resistance state and a
sigmoid function capturing the transition from a high resistance state to a low
resistance state. Since the thermistance behavior depicts a linear relationship
between the resistance and the temperature in the Log-Linear domain, one can
readily model the two thermistors as exponential functions of the temperature such
that RLRS ¼ RLRSF e�BLRS ðT ðtÞ�T F Þ and RHRS ¼ RHRS0 e�BHRS ðTðtÞ�T 0 Þ . RLRSF is the low resis-
tance state defined at temperature T F (a reference temperaure) and RHRS0 is the
high resistance state defined at the ambient temperature T 0 . BLRS and BHRS are the
negative temperature coefficients of the thermistors which can be readily extracted
from the slope of the thermistance versus temperature plot. This modeling frame-
work, however, requires clipping of the RLRS and RHRS at some minimum and
maximum values to avoid any unphysical behavior in circuit simulation. Clipping,
however, often uses conditional statements, which hampers the “smoothness” of
the model yielding potential convergence difficulties in circuit simulation. Hence,
the model equations are reformulated such that RLRS and RHRS smoothly approach to
RLRSF and RHRS0 at high and low temperatures, respectively.
This relationship between the resistance and the temperature can be described as
follows:
 1=A
RLRS ¼ RLRSF 1 þ K A
LRS (15)

K HRS
RHRS ¼ RHRS0  1=A (16)
1 þ KA
HRS

ðRHRS � RLRS Þ
RIMT ¼ RLRS þ T ðtÞ�T c
(17)
1þe Tx

where K LRS ¼ e�BLRS ðTðtÞ�T F Þ and K HRS ¼ e�BHRS ðT ðtÞ�T 0 Þ

The fitting parameter, T x , captures the sharpness of the resistance transition. T c
is the critical temperature which is around 340 K in the case of VO2 devices [53].
RLRS and RHRS are the low resistance state and high resistance state, respectively. A
is a control parameter that governs how the two thermistors approach the asymp-
totes [65]. While the model might seem complicated at first glance, the principal
equations are simple exponential functions. This formulation is employed to abide
by compact modeling practices as suggested in [65, 66].
The temperature evolution dynamics are described by the compact thermal
model in [67] as shown in (4):

104
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

dT ðtÞ T ðtÞ � T 0
Cth ¼ V IMT IIMT � (18)
dt Rth

where VIMTIIMT is the Joule heating, Cth and Rth are the effective thermal
capacitance and the effective thermal resistance, respectively, and T0 is the
ambient temperature. This model assumes that the device stays at an effective
temperature T(t) and exchanges heat with the ambient environment at an
ambient temperature T0.
The model predictions closely follow the experimental data as well as electro-
thermal simulation [52] as shown in Figures 11–15. Figures 11 and 12 capture the
resistance transition around the critical temperature which is about 340 K in VO2
devices fitted against experimental data from [52]. Figure 13 depicts the hysteresis
in the I-V domain (a characteristic of memristors) exhibited by the IMT device as
shown in [52, 53] and fitted against the experimental data from [52]. Figures 14 and
15 capture the time dependence of temperature and resistance evolution, respec-
tively, fitted against electrothermal simulations from [52]. Three voltage levels,
based on the values used in [52] were applied across the IMT device: 1.4 V (blue),
1.6 V (red) and 1.8 V (green). One can readily observe in Figure 14 that the local
temperature of the device saturates at a higher temperature value for higher

Figure 11.
Model (solid line) fitting against experimental data (marker), sample 1.

Figure 12.
Model (solid line) fitting against experimental data (marker), sample 2.

105
Computational Models in Engineering

Figure 13.
Model fitting showing pinched hysteresis.

Figure 14.
Model (solid line) fitting against electro-thermal simulations (markers); plotting the device local temperature
against time for three applied voltage values.

Figure 15.
Model (solid line) fitting against elector-thermal simulations (markers); plotting the device resistance against
time for three applied voltage values.

106
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

Figure 16.
Schematic of the proposed neuron circuit.

voltages due to increased Joule heating. In Figure 15, higher voltages result in a
faster transition due to higher rate of joule heating.

5.4 Application in neuron design

Figure 16 depicts a simple neuron based on the proposed circuit in [52]. First,
the input voltage spikes are fed through the synaptic network. The dot product
multiplication is then executed between the input spikes and the synaptic weights
via Ohm’s law such that the accumulated sum follows Isum ¼ ∑i V i Gi . Isum is then fed
to the neuron which is represented by the parallel combination of the IMT device
and the capacitor. The core of the neuron is the IMT device which switches from
RHRS to RLRS when the device’s temperature exceeds the critical temperature
resulting in a current spike. The current spike is converted to a voltage pulse via a
CMOS inverter. A spike generator is then added to generate an output voltage spike
with a pulse width controlled by the RC time constant of the RC network preceding
the output buffer. Figure 17 depicts the simulation results of the neuron circuit.

Figure 17.
Simulation of the proposed neuron.

107
Computational Models in Engineering

The essence of neuron oscillation rests in the IMT device alternating between
RHRS and RLRS. At RHRS, the steady state temperature exceeds the critical tempera-
ture and, accordingly, the neuron fires. At RLRS, the steady state temperature drops
below the critical temperature, the neuron resets and the process is repeated for the
next inputs.
At steady state (dT
dt ¼ 0Þ, the solution to the differential equation in (18) can be
described as follows:

Tss ¼ T0 þ Rth I2IMT RIMT

where T ss is the steady state temperature of the IMT device. Hence, according
to the aforementioned explanation, the oscillation condition can be expressed as
follows:

Rth I2IMT RLRS < Tc � T0 < Rth I2IMT RHRS

Inequality (7) establishes the oscillation condition for the IMT-based neuron as a
function of device parameters such as Rth , RLRS , RHRS and T c and circuit variables
such as IIMT which is a function of the amplitude of the voltage spike and the series
resistance with the IMT device including the synapse resistance, diode ON resis-
tance and the IMT series resistance. The neuron typically operates in three phases:
(I) accumulation wherein the inputs from the synapse networks are summed, (II)
firing when the accumulated value reaches the neuron’s threshold and (III) refrac-
tory period wherein the neuron is idle.
In CMOS neurons, two operational amplifiers are required for accumulation
(integration) and firing (comparison). In addition, a feedback circuit is often
employed to implement the refractory period. These operational amplifiers, besides
entailing all the complexities of analog design, are also area consuming and power
hungry. On the other hand, the IMT device can provide the accumulation function
through the heating of the device, fire through device transition and a refractory
period during device cooling should the device be placed in such configuration. A
typical CMOS neuron, such as the work in [68], requires more than 20 transistors
while the proposed IMT-based neuron only requires 7 transistors.

6. Conclusion

Computer aided design and simulation plays a major role in the advancement of
semiconductor industry. A multitude of exciting new devices has emerged in recent
years. There is a growing need in the industry and academia for fast and reliable
compact models of these emerging devices to enable useful circuit design leveraging
their unique capabilities. In this chapter, we show how different approaches can be
adopted to model three emerging semiconductor devices namely, silicon- on- insu-
lator four gate transistor (G4FET), single photon avalanche diode (SPAD) and
insulator-metal transistor (IMT) device with volatile memristance. All the models
have been verified against experimental/TCAD data and implemented in commer-
cial circuit simulator. The ideas developed in this chapter can also be transferred to
model other emerging devices.

108
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

Author details

Md Sakib Hasan1*, Mst Shamim Ara Shawkat1, Sherif Amer1, Syed Kamrul Islam2,
Nicole McFarlane1 and Garrett S. Rose1

1 Department of Electrical Engineering and Computer Science, University of

Tennessee, Knoxville, TN, USA

2 Department of Electrical Engineering and Computer Science, University of

Missouri, Columbia, MO, USA

*Address all correspondence to: mhasan4@vols.utk.edu

© 2019 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

109
Computational Models in Engineering

References

[1] Blalock BJ, Cristoloveanu S, Dufrene [8] Majumder MB, Hasan MS, Uddin M,
BM, Allibert F, Mojarradi MM. Rose GS. Chaos computing for
International journal of high speed mitigating side channel attack. In: IEEE
electronics and systems. The Multiple- International Symposium on Hardware
Gate MOS-JFET Transistor. 2002;12(2): Oriented Security and Trust (HOST).
511-520 2018

[2] Akarvardar K, Chen S, Vandersand J, [9] Shanta AS, Majumder MB, Hasan
Blalock BJ, Schrimpf RD, Prothro B, MS, Uddin M, Rose GS. Design of a
et al. Four-gate transistor voltage- Reconfigurable Chaos Gate with
controlled negative differential enhanced functionality space in 65nm
resistance device and related circuit CMOS. In: 2018 IEEE 61st International
applications. In: Proc. IEEE Midwest Symposium on Circuits and
International SOI Conference. 2006. Systems (MWSCAS), Windsor. 2018
pp. 71-72
[10] Akarvardar KDBCS, Blalock BJ,
[3] Akarvardar K, Chen S, Blalock B, Higashino T, Mojarradi MM, Kolawa E.
Cristoloveanu S, Gentil P, Mojarradi M. Multi-bias dependence of threshold
A novel four-quadrant analog multiplier voltage, subthreshold swing, and
using SOI four-gate transistors (G4- mobility in G4-FETs. In: Proc.
FETs). In: Proceedings of ESSCIRC. ESSDERC. 2003
2005
[11] Akarvardar K, Cristoloveanu S,
[4] Akarvardar K, Blalock B, Chen S, Gentil P. Analytical modeling of the
Cristoloveanu S, Gentil P, Mojarradi two-dimensional potential distribution
MM. Digital circuits using SOI four-gate and threshold voltage of the SOI four-
transistor. In: 8th International gate transistor. IEEE Transactions on
Conference on Solid-State and Electron Devices. 2006;53(10):2569-
Integrated Circuit Technology 2577
Proceedings. 2006
[12] Akarvardar K, Cristoloveanu S,
[5] Fijany A, Vatan F, Mojarradi M, Gentil P, Blalock BJ, Dufrene B,
Toomarian B, Blalock B, Akarvardar K, Mojarradi MM. Depletion-all-around in
et al. The G4-FET: A universal and SOI G4-FETs: A conduction mechanism
programmable logic gate. In: with high performance. In: Proceedings
Proceedings of the 15th ACM Great of the 30th European Solid-State
Lakes symposium on VLSI. 2005 Circuits Conference. 2004

[6] Friedman JS, Godkin A, Henning A, [13] Akarvardar K, Cristoloveanu S,

Vaknin Y, Rosenwaks Y, Sahakian AV. Gentil P, Schrimpf RD, Blalock BJ.
Threshold logic with electrostatically Depletion-all-around operation of the
formed nanowires. IEEE Transactions SOI four-gate transistor. IEEE
on Electron Devices. 2016;63(3):1388- Transactions on Electron Devices. 2007;
1391 54(2):323-331

[7] Henning A, Swaminathan N, Godkin [14] Tejada JAJ, Rodríguez AL, Godoy A,
N, Shalev G, Amit I, Rosewaks Y. Villanueva JAL, Gómez-Campos FM,
Tunable diameter electrostatically Rodríguez-Bolívar S. A low-frequency
formed nanowire for high sensitivity gas noise model for four-gate Field-effect
sensing. Nano Research. 2015;8(7):2206- transistors. IEEE Transactions on
2215 Electron Devices. 2008;55(3):896-903

110
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

[15] Tejada JAJ, Rodríguez AL, Godoy A, [23] Hasan MS, Mahbub I, Islam SK,
Rodríguez-Bolívar S, Villanueva JAL, Rose GS. A MOS-JFET macromodel of
Marinov O, et al. Effects of gate oxide SOI four-gate transistors (G4FET) to aid
and junction nonuniformity on the DC innovative circuit design. In: IEEE 13th
and low-frequency noise performance Dallas Circuits and Systems Conference
of four-gate transistors. IEEE (DCAS), Dallas. 2018
Transactions on Electron Devices. 2012;
59(2):459-467 [24] Moler CB. Numerical Computing
with MATLAB. 2004
[16] Sayed S, Khan MZ. Analytical
modeling of surface accumulation [25] Shichman H, Hodges DA. Modeling
behavior of fully depleted SOI four gate and simulation of insulated-gate field-
transistors (G 4-FETs). Solid-State effect transistor switching circuits. IEEE
Electronics. 2013;81:105-112 Journal of Solid-State Circuits. 1968;3(3):
285-289
[17] Sayed S, Hossain MI, Khan MZR. A
subthreshold swing model for thin-film [26] Cova S, Ghioni M, Lotito A, Rech I,
fully depleted SOI four-gate transistors. Zappa F. Evolution and prospects for
IEEE Transactions on Electron Devices. single-photon avalanche diodes and
2012;59(3):854-857 quenching circuits. Journal of Modern
Optics. 2004;51:1267-1288
[18] Chiang TK. A new two-dimensional
analytical model for the fully-depleted
[27] Niclass C, Rochas A, Besse P,
SOI four-gate transistor. In: 10th IEEE
Charbon E. Design and characterization
International Conference on Solid-State
of a CMOS 3-D image sensor based on
and Integrated Circuit Technology,
single-photon avalanche diodes. IEEE
Shanghai. 2010
Journal of Solid-State Circuits. 2005;40:
[19] Sayed S, Hossain MI, Huq R, Khan 1847-1854
MZR. Three dimensional modeling of
SOI four gate transistors. In: IEEE [28] Gersbach M, Maruyama Y, Niclass
Nanotechnology Materials and Devices C, Sawada K, Charbon E. A room
Conference, Monterey. 2010 temperature CMOS single photon
sensor for chemiluminescence
[20] Hasan MS, Rahman T, Islam SK, detection. In: Miniaturized Systems for
Blalock B. Numerical modeling and Chemistry and Life Sciences. 2006
implementation in circuit simulator of
SOI four-gate transistor (G4FET) using [29] Dandin M, Akturk A, Nouri B,
multidimensional Lagrange and Goldsman N, Abshire P.
Bernstein polynomial. Microelectronics Characterization of single-photon
Journal. 2017;65:84-93 avalanche diodes in a 0.5μm standard
CMOS process—Part 1: Peimeter
[21] Hasan MS, Islam S. DC modeling of breakdown suppression. IEEE Sensors
SOI four-gate transistor (G4FET) for Journal. 2010;10(11):1682-1690
implementation in circuit simulator
using multivariate regression [30] Stoppa D, Mosconi D, Pancheri L,
polynomial. IET Circuits, Devices & Gonzo L. Single-photon avalanche diode
Systems;2018 CMOS sensor for timeresolved
fluorescence measurements. IEEE
[22] Hasan MS, Islam SK, Blalock BJ. Sensors Journal. 2009;9(9):1084-1090
Modeling of SOI four-gate transistor (G
4 FET) using multidimensional spline [31] Dandin M, Habib MHU, Nouri B,
interpolation method. Microelectronics Abshire P, McFarlane N.
Journal. 2018;76:33-42 Characterization of single-photon

111
Computational Models in Engineering

avalanche diodes in a 0.5 μm standard for circuit simulations. IEEE Photonics

CMOS process—Part 2: Equivalent Technology Letters. 2007;19(23):1922-
circuit model and Geiger mode readout. 1924
IEEE Sensors Journal. 2016;16(9):3075-
3083 [40] Mita R, Palumbo G, Fallica P.
Accurate model for single photon
[32] Palubiak DP, Deen M. CMOS avalanche diodes. IET Circuits, Devices
SPADs: Design issues and research and Systems. 2008;2(2):207-212
challenges for detectors, circuits, and
arrays. IEEE Journal of Selected Topics [41] Zappa F, Tosi A, Mora AD, Tisa S.
in Quantum Electronics. 2014;20(6): SPICE modeling of single photon
409-426 avalanche diodes. Sensors Actuator A,
Physics. 2008;153(2):207-212
[33] Habib MHU, Mamun KAA,
McFarlane N. A SPICE model for [42] Arbat A, Dièguez A, Gascon D,
perimeter-gated single photon Trenado J, Garrido L. Avalanche
avalanche diode. In: International photodiodes for high energy particle
Midwest Symposium on Circuits and tracking in 130 nm CMOS technology.
Systems. 2014 In: 16th IEEE International Conference
on Electronics, Circuits and Systems—
[34] Shawkat MSA, Habib MHU, (ICECS 2009). 2009
McFarlane N. An analog CMOS silicon
photomultiplier using perimeter-gated [43] Giustolisi G, Mita R, Palumbo G.
single-photon avalanche diodes. IEEE Behavioral modeling of statistical
Transactions on Circuits and Systems I. phenomena of single-photon avalanche
2018;65(11):3830-3841 diodes. International Journal of Circuit
Theory and Applications. 2012;40(7):
[35] Hurkx GAM, Graaff HCD, 661-679
Kloosterman WJ, Knuvers MPG. A new
analytical diode model including [44] Chick S, Coath R, Sellahewa R,
tunneling and avalanche breakdown. Turchetta R, Leitner T, Fenigstein A.
IEEE Transactions on Electron Devices. Dead time compensation in CMOS
1992;39(9):2090-2098 single photon avalanche diodes with
active quenching and external reset.
[36] Javitt M, Savuskan V, Merhav T, IEEE Transactions on Electron Devices.
Nemirovsky Y. Revisiting single photon 2014;61(8):2725-2731
avalanche diode currentvoltage
modeling and transient characteristics. [45] Kolb KE. Signal-to-noise ratio of
Journal of Applied Physics. 2014;115 Geiger-mode avalanche photodiode
(20) single-photon counting detectors.
Optical Engineering. 2014;53(8):
[37] Lacaita A, Cova S, Spinelli A, Zappa 081804-1 -081804-9
F. Photon-assisted avalanche spreading
in reach-through photodiodes. Applied [46] Field RM, Lary J, Cohn J, Paninski
Physics Letters. 1993;62(6):606-608 L, Shepard KL. A low noise, single-
photon avalanche diode in standard 0.13
[38] Tisa S, Zappa F, Tosi A, Cova S. μm complementary metal-oxide-
Electronics for single photon avalanche semiconductor process. Applied Physics
diode arrays. Sensors and Actuators A. Letters. 2010;97(21)
2007;140:113-122
[47] Ramirez DA, Hayat MM, Torres SN,
[39] Mora AD, Tosi A, Tisa S, Zappa F. Saleh BEA, Teich MC. Information-
Single-photon avalanche diode model theoretic criterion for the performance

112
Modeling Emerging Semiconductor Devices for Circuit Simulation
DOI: http://dx.doi.org/10.5772/intechopen.85873

of single-photon avalanche photodiodes. systems. In: IEEE Biomedical Circuits

IEEE Photonics Technology Letters. and Systems Conference (BioCAS).
2005;17(10):2164-2166 2018

[48] Gu J, Habib MHU, McFarlane N. [56] Hasan MS, Schuman CD, Najem JS,
Perimeter gated single photn avalanche Weiss R, Skuda N, Belianinov A, et al.
diodes: An information theoretic Biomimetic soft-material synapse for
assessment. IEEE Photonics Technology neuromorphic computing: From device
Letters. 2016;28(6):701-704 to network. In: IEEE Dallas Circuits and
Systems Conference, Dallas. 2018
[49] Zhou Y, Ramanathan S. Mott
memory and Neuromorphic devices. [57] Ruzmetov D, Senanayake SD,
Proceedings of the IEEE. 2015;103(8): Narayanamurti V, Ramanathan S.
1289-1310 Correlation between metal-insulator
transition characteristics and electronic
[50] Son M, Lee J, Park J, Shin J, Choi G, structure changes in vanadium oxide
Jung S, et al. Excellent selector thin films. Physical Review B. 2008;77
characteristics of Nanoscale VO2 for (19)
high-density bipolar ReRAM
applications. IEEE Electron Device [58] Freeman E, Kar A, Shukla N, Misra
Letters. 2011;32(11):1579-1581 R, Engel-Herbert R, Schlom D, et al.
Characterization and modeling of
[51] Gao L, Holbert KE, Yu S. Total metal-insulator transition (mit) based
ionizing dose effects of gamma ray tunnel junctions. In: IEEE 70th Annual
radiation on NbO x-based selector Device Research Conference (DRC).
devices for crossbar array memory. 2012
IEEE Transactions on Nuclear Science.
2017;64(6):1535-1539 [59] Ruzmetov D, Gopalakrishnan G,
Deng J, Narayanamurti V, Ramanathan
[52] Lin J, Annadi A, Sonde S, Chen C, S. Electrical triggering of metal-
Stan L, Achari K, et al. Low-voltage insulator transition in nanoscale
artificial neuron using feedback vanadium oxide junctions. Journal of
engineered insulatorto-metal-transition Applied Physics. 2009;106(8)
devices. In: IEEE International Electron
Devices Meeting (IEDM). 2016 [60] Yang Z, Hart S, Ko C, Yacoby A,
Ramanathan S. Studies on electric
[53] Chen P, Seo J, Cao Y, Yu S. Compact triggering of the metal-insulator
oscillation neuron exploiting metal- transition in vo2 thin films between 77 K
insulator-transition for neuromorphic and 300 K. Journal of Applied Physics.
computing. In: Proceedings of the ACM 2011;110(3)
International Conference on Computer-
Aided Design. 2016 [61] Gopalakrishnan G, Ruzmetov D,
Ramanathan S. On the triggering
[54] Najem JS, Taylor GJ, Weiss RJ, mechanism for the metal–insulator
Hasan MS, Rose G, Schuman CD, et al. transition in thin film vo2 devices:
Memristive Ion Channel-Doped Electric field versus thermal effects.
Biomembranes as Synaptic Mimics. ACS Journal of Materials Science. 2009;
Nano. 2018 44(19):5345-5353

[55] Weiss R, Najem JS, Hasan MS, [62] Karda K, Mouli C, Ramanathan S,
Schuman CD, Belianinov A, Collier P, Alam MA. A self-consistent,
et al. A soft-matter biomolecular semiclassical electrothermal modeling
memristor synapse for neuromorphic framework for Mott devices. IEEE

113
Computational Models in Engineering

Transactions on Electron Devices. 2018;

65(5):1672-1678

[63] Amer S, Hasan MS, Adnan MM,

Rose GS. Spice modeling of insulator
metal transition: Model of the critical
temperature. IEEE Journal of the
Electron Devices Society. 2018

[64] Chua L. Everything you wish to

know about memristors but are afraid to
ask. Radioengineering. 2015;24(2)

[65] Tsividis Y, McAndrew C. Operation

and Modeling of the MOS Transistor.
Oxford Univ. Press; 2011

[66] McAndrew CC, Coram GJ,

Gullapalli KK, Jones JR, Nagel LW, Roy
AS, et al. Best practices for compact
modeling in verilog-a. IEEE Journal of
the Electron Devices Society. 2015;3(5):
383-396

[67] Chen P-Y, Yu S. Compact modeling

of RRAM devices and its applications in
1T1R and 1S1R Array design. IEEE
Transactions on Electron Devices. 2015;
62(12):4022-4028

[68] Chakma G, Adnan MM, Wyer AR,

Weiss R, Schuman CD, Rose GS.
Memristive mixed-signal neuromorphic
systems: Energy-efficient learning at the
circuit-level. IEEE Journal on Emerging
and Selected Topics in Circuits and
Systems. 2018;8(1)

114
Chapter 7

Mathematical Modeling of
Aerodynamic Heating and
Pressure Distribution on a 5-Inch
Hemispherical Concave Nose in
Supersonic Flow
Dimitrios P. Spiridakos, Nicholas C. Markatos and
Despoina Karadimou

Abstract

During the design of an aircraft, a significant parameter that is taken into

consideration is aerodynamic heating. Aerodynamically induced heating affects
both the structure of the aircraft and its vulnerability to heat-seeking missiles in
modern warfare. As a result, the ability to calculate efficiently the heating produced
as well as the pressure distribution in such flows is crucial. Therefore, in this present
study, the PHOENICS CFD code as modified by DRA Farnborough in order to
calculate heat transfer and pressure measurement on a 5-inch hemispherical con-
cave nose at a Mach number of 2.0 is evaluated and customized in order to produce
faster and more accurate results. Apart from numerical alterations, different turbu-
lence models are being examined as well as different discretization schemes.
Numerical solutions show improvement up to 6% in comparison with the original
model, both in terms of convergence rate and in terms of agreement with the
available experimental data. With the new modeling suggested in the present work,
the significance of both the discretization scheme and the choice of the turbulence
modeling is demonstrated for the flows under consideration. The use of a high-
order discretization scheme is suggested for more acute areas of the body modeled
in order to improve results further.

Keywords: aerodynamic heating, turbulence modeling, hypersonic flow,

discretization scheme, PHOENICS

1. Introduction

Aerodynamic heating has been a major study issue for a long time now and
many investigations have been conducted by researchers in order to achieve high-
quality simulation results, both on the theoretical and the numerical level. Experi-
mental studies of the above phenomenon are rare, as conducting such research is
prohibitively expensive.

115
Computational Models in Engineering

Aerodynamic heating is a phenomenon worth studying as its significance lies in

two major scientific fields. At first, aerodynamic heating is the most significant
parameter of the emitted infrared radiation from aircraft, a fact that is of direct
interest in military studies. According to statistics, the majority of aerial vehicles
downed in modern warfare was due to weapons based on detecting Infrared signa-
ture level of those vehicles. Their percentage is as high as 89% of total aircraft shot
downs between 1967 and 1993, according to Thompson J. et al. [1]. Apart from the
emitted infrared radiation, the phenomenon of aerodynamic heating has to be taken
into consideration when designing aerospace vehicles that fly at speeds comparable
to that of sound. Ultrasonic aircraft and aircraft that re-enter the atmosphere can
develop temperatures high enough that have to be carefully considered when
designing the aircraft, in order to avoid unpleasant situations.
Aerodynamic heating is a phenomenon that takes place mostly due to heat from
the warm part of the boundary layer to the body’s surface. However, when
approaching the speed of sound, aerodynamic heating phenomena are enhanced by
the radiation emitted from molecules that are being decomposed and ionized due to
the temperature rising behind the shockwave. At speeds greater than 5000 m/sec,
the emitted radiation becomes significant and plays a measurable role in the phe-
nomenon. Finally, another way that aerodynamic heating takes place concerns
aircraft that fly in the higher atmospheric layers, where the distance between the air
molecules is comparable to the aerial vehicle dimensions. However, the present
study is focused only on the convection phenomena that lead to aerodynamic
heating, as the speeds examined do not exceed Mach numbers of 2.
As mentioned above, the significance of aerodynamic heating is a major issue in
both the preliminary design of aerial vehicles and the calculation of the infrared
radiation emitted from the vehicle in flight. Apart from experimental methods, which
may offer reliable results but require enormous budgets, numerous simulations have
been carried out via means of computational mechanics. Computational mechanics
simulations approach the problem mainly through inviscid-viscous methods thus
promising reduced computational time. AEROHEAT and INCHES are two of the
simpler methods to approach the phenomenon of aerodynamic heating [2]. Both
methods use the axisymmetric analog concept that allows axisymmetric boundary
layer techniques to be applied to three-dimensional flows, provided that the surface
streamlines are known. AEROHEAT calculates approximate surface streamlines based
solely on the body geometry. INCHES uses an approximate expression for the scale
factor in the windward and leeward planes, which describes the spreading of surface
streamlines. An empirical relation then generates circumferential heating rates.
DRA Farnborough developed the SAPPHIRE software that is capable of calcu-
lating the surface temperatures and the characteristics of the flow field for a given
geometry and defined flight characteristics. The above software is a series of codes
that use PHOENICS [3] as the main CFD solver. The main advantage of CFD
techniques is that aerodynamic heating can be related to other phenomena, as well,
such as conjugating heat transfer and providing more detailed results. In order for
SAPPHIRE to provide reliable results on the thermal trace, aerodynamic heating of
the body has to be calculated with an error that does not exceed 5% [4]. For this
purpose, a PHOENICS-based code was developed that has the possibility to validate
results with available experimental data. The purpose of the present work is to make
the model more time efficient and more accurate, producing results closer to the
available experimental data.
According to finite volume or codes with FEM, attempts have been made, with
acceptable results, by means of conjugating CFD and FEM. Murakami et al. [5] used
CFD coding for the Navier-Stokes equations to calculate the outer flow, while the
finite element method was used to perform the thermal analysis via equations
describing unsteady heat conduction for a 3D field. The results were impressive, as
116
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

they were very close to the experimental data measured in a wind tunnel experiment.
Lu Jianwei and Wang Qiang [6] in order to calculate the emitted IR radiation suggested
a division of the computational grid into two regions: the fluid and the solid. The fluid
region is modeled by the Navier-Stokes equations, while turbulence is modeled by the
RNG k-ε model. In the solid region, the energy transfer equation is used. In a similar
study for calculating the emitted IR radiation, Marlene Johansson and Mats Dalenbring
[7] developed the SIGGE software. SIGGE uses inputs of temperature, pressure, and
species distribution that have been roughly estimated beforehand, together with the
grid of the geometry being studied. The Navier-Stokes equations are solved by the
VOLSOL solver, while turbulence is described by the k-ε model.
Apart from CFD methods, attempts have been made to simulate phenomena that
involve aerodynamic heating via other methods as well. Rodrigo C. Palharini and
Wilson F. N. Santos [8] used the Direct Simulation Monte Carlo (DSMC) method to
simulate a two-dimensional steady supersonic flow in an orthogonal cavity. Main
interests of this study were the characteristics of the aerodynamic surface, such as
heat transfer coefficients, etc. Another method introduced by Georgia Tech
Research Institute [9] named GTSIG (Georgia Tech Signature Inviscid Model)
involves the calculation of temperature distribution by creating thermal meshes and
partial differential equations. Finally, Christopher J. Riley et al., in their study [2],
developed a 3D technique without viscous calculations that has the ability to calcu-
late the degree of surface heating. Surface streamlines are calculated from the
inviscid solution and the axisymmetric analog. Then, they are used in combination
with an equation for convective heat transfer, to calculate the total surface heat
transfer. The results of this method are satisfying as they are close to experimental
data and the results of both the Navier-Stokes and the VSL equation solutions.

2. Mathematical modeling

2.1 Governing equations

In order to describe the phenomena under investigation, use is made of the

conservation equations describing a three-dimensional, single phase, turbulent,
steady state flow. The above equations are formulated in such a manner as to
incorporate the physical modeling of the case study under the following assumptions:

• One phase, steady-state flow

• Compressible fluid

• Newtonian fluid

• Absence of chemical reaction

It is also clear that because of the general geometries considered, the equations must
be modified in a way to be suitable for use in a body-fitted coordinate (BFC) system.

2.1.1 Modification of conservation equations from Cartesian coordinate system to BFC

!
Assuming that X ¼ ðX 1 , X 2 , X 3 Þ the Cartesian coordinate system, where Xj are
! !
the Cartesian Coordinates and ij the direction unit vectors, the velocity V can be
expressed as:
117
Computational Models in Engineering

3
X
! ! ! ! !
V¼ uj ij ¼ u1 i þ u2 j þ u3 k:
j¼1

Furthermore,

X3
! ! ∂
∇¼ ij (1)
i¼1
∂xj

Assume a non-orthogonal coordinate system, defined as ξη ¼ ξη ðX 1 , X 2 , X 3 Þ,

η = 1,2,3, where ξη are the curvilinear coordinates at any point and X1, X2, X3 its
Cartesian coordinates. Assume ! g η to be the non-orthogonal unit vectors along the
directions of the curvilinear coordinates; then the components of the velocity vector
tangent to the unit vectors !g η can be defined as:

3
X
!
Vη ¼ !
gη V ¼ g η,j U j , η ¼ 1, 2, 3 (2)
j¼1

where Uj are the Cartesian components and the unit tangent vectors alongside
the three directions are:

� � X
! .� ! �
3
! ! !
gη ¼ X ,ξη �X ξη � ¼ g η,j i j , η ¼ 1, 2, 3 (3)
j¼1

!
where X is the position vector of a point and the index ξη denotes the partial
derivative to the curvilinear coordinate ξη.

2.1.2 Conservation of momentum

In order to obtain the resolute [10] of the momentum equation in each of the
coordinate directions, we form the dot product of tangent unit vectors with the
momentum equation (Eq. (1)). The momentum equation, considering the assump-
tions and the physics of this simulation, is derived as follows:
h�! !� !i ! ! !
ρ ∇V V � ∇ � ~τ ¼ �∇p þ F , (4)

where
" � �#
! �! !� ! X 3 3
X ∂ ∂uj !
∇ � ~τ ¼ ∇μ∇ V þ μ iλ (5)
λ¼1 j¼1
∂xj ∂xλ

In order to obtain the resolute of the momentum equation in each of the coordi-
nate directions, the dot product of the tangent unit vector and the above momen-
tum equation is formed:
h�! !� !i ! ! � ! !� ! �! !� !
!
gn ρ ∇V V � !g n ∇ � ~τ ¼ �g! ∇ p $ !
g ρV � ∇ V � !
g �
!
∇ � μ∇ V � g n � Sμ
n n n
!
¼ �g!∇p n
(6)
P3 hP � �i!
3 ∂u
where Sμ ¼ λ¼1
∂
j¼1 ∂xj μ ∂xλj iλ

118
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

From vector calculus, using the identity AxðBxCÞ ¼ BðA � CÞ � CðA � BÞ, we have:
! �! !� � ! !�
!
�
! !� ! ! �! ! � !
�! !�
∇ g n � ρV ¼ ρV � ∇ g n þ g n � ∇ ρV þ ρV x ∇x g n þ g n x ∇ x ρV (7)

Also,

! ! �! ! � !
� ! �! !�� !
V � g n x ∇xV ¼ � g n � V x ∇xV V (8)
!
� ! !� ! ! jV j2 ! �! ! �
n
V�∇ V ¼∇ � V x ∇xV (9)
2
! ! � ! ! � � ! ! � � ! ! � ! !
V � V x ∇x g n ¼ V xV � ∇x g n ¼ 0, since V xV ¼ 0 (10)

Thus,
!
! ! jV n j2 ! �! !� !
gn � ∇ ¼ V � gn � ∇ � V
2

Forming the dot product of the above equation and (Eq. (7)), we have:
� ! !� ! ! � ! !�! � ! !� !
!
ρV � ∇ V n ¼ V � ρV � ∇ g n þ g n � ρV � ∇ V (11)

Also, for a scalar variable φn:

! � ! � ! !
∇ � ρV φn ¼ ρV � ∇φn (12)

and
�! !� ! �! !� ! ! �! !�
! ! !
∇ � μ∇ V � g n ¼ g n � ∇ � μ∇ V þ V � ∇ � μ∇ g n (13)

Using Eqs. (11)–(13), we can write (Eq. (6)), as follows:

! � ! � �! !� ! ! ! � ! !� ! � �! !�
∇ � ρV V n � ∇ � μ∇ V n � !gn � Sμ ¼ �gn ∇p þ V ρV � ∇ !g n � V ∇� μ∇ ! g n (14)

and
� ! !� ! � ! !� � ! !� !
V � ∇ Vn ¼ V V � ∇ !gn þ !
gn V � ∇ V

From Eqs. (1) and (3), the partial derivative in the direction ξn is:

X3
∂ ! ∂
¼!
gn � ∇ ¼ gn,j (15)
∂ξn j¼1
∂x j

With inverse

X3
∂ ∂
¼ Gn,j (16)
∂xn j¼1
∂ξj

where Gn,j the elements of the inverse matrix of the elements gn,j.
119
Computational Models in Engineering

2.1.3 Conservation of mass

Since the case study is for steady flow, the conservation of mass equation is
hence simplified in the following form for the Cartesian coordinate system:
� !� � !� � !�
∂ ρV ∂ ρV ∂ ρV
divðρV Þ ¼ þ þ ¼0
∂x ∂y ∂z

Using body-fitted coordinates, the continuity equation must be solved in terms

of the velocity components along the coordinate directions. These components, un,
P !
would be V ¼ 3n¼1 g n un and are calculated from the velocity resolutes by solving
the simultaneous equations

3
X � �
! !
Vn ¼ g n � g j uj , n ¼ 1, 2, 3: (17)
j¼1

Velocity components in terms of the resolutes are

3
X
un ¼ Gn,j V j , n ¼ 1, 2, 3 (18)
j¼1

where
2 3
ð1 � c2 Þ
ðbc � aÞ ðac � bÞ
6 � � 7
Gn,j=σ ¼ 6 ðbc � aÞ
4 1 � b2 ðab � cÞ 7
5 (19)

ðac � bÞ ðab � cÞ ð1 � a2 Þ

where

a ¼ g1 � g2 , b ¼ g1 � g3 , c ¼ g2 � g3 (20)

and
� � ��
σ ¼ 1 � a2 þ b2 þ c2 � 2abc (21)

Using Eqs. (1) and (18), the following expression for the velocity is derived:

3
X 3 3 !
! ! X X !
V¼ in Gn,j V j ¼ Q jVj (22)
n¼1 j¼1 j¼1

where

3
�! X !
Qj ¼ Gn,j i n (23)
j¼1

!
the vector used to obtain the resolutes (projections) of the velocity vector V .
From Eqs. (1), (16), and (20):

120
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

X3 !
! ∂
∇¼ Qj (24)
j¼1
∂ξ j

!� ! �
Finally, the continuity equation can be written as ∇ ρV ¼ 0 where

3
X
! !
V¼ g n un (25)
n¼1

2.2 Turbulence modeling

In order to evaluate the use of different turbulence models in the cases under
consideration, a parametric study was accomplished by trying three “popular”
turbulence models: (i) standard k-ε model, (ii) RNG k-ε model, and (iii) k-ω model.
The k-ε model simulation was chosen in order to reproduce the original simulation
accomplished by Steve Rooks [4]. In order to take it one step further and include
the effect of the swirling of the flow, another simulation was performed using the
RNG k-ω model. Last but not least, the k-ω turbulence model was used as well, as
preliminary studies proved its suitability for wall-bounded flows.

2.2.1 The general ϕ-equation

By generalizing Eq. (14), the appropriate equation for any scalar variable can be
provided [11]:
! !
∇ � Q ¼ Sφn (26)

where
! ! �!!
Q ¼ ρV φn � Γ φn ∇φn (27)

and φ a general scalar variable like H (enthalpy), k, ε, etc. Regarding k (turbu-

lence kinetic energy) and ε (turbulence dissipation rate):

Sk ¼ G � ρε (28)

Sε ¼ C1 ε=kG � C2 ρε2=κ (29)

� �
G ¼ μ ∂uj=∂xi þ ∂ui=∂xj ∂uj=∂xj , in Cartesian � tensor form (30)

where ui, uj the velocity components in the direction of the Cartesian coordi-
nates xi, xj with i = 1,2,3 and j = 1,2,3 [10]. It must also be noted that the effective
exchange coefficients for k, ε are as follows:

μl þ μt ¼ μ (31)

k2
μt ¼ ρCμ (32)
ε
Γ k ¼ ðμl þμt Þ=σk,t (33)

Γ ε ¼ ðμl þμt Þ=σ ε,t (34)

121
Computational Models in Engineering

where μl is the laminar viscosity; C1, C2, and Cμ are turbulence model constants;
and σk,t and σε,t are the turbulence Prandtl numbers for k, ε [10].
In total seven differential equations are solved, namely, the continuity equation,
the three momentum equations, the enthalpy equation, and the two equations for
the turbulence variables.

2.2.2 The near-wall region

Special attention was given to the treatment of the near-wall region, as it is of

high importance for the type of flows studied and the 2-equation models used to
simulate turbulence do not, in general, produce results of high accuracy in near-wall
regions. The near-wall region was treated via a log-law wall function that also
played a major role in deciding the computational grid as well as ensuring compu-
tational economy since there is no need to solve throughout the viscous sub-layer.
The use of this method was introduced by Launder and Spalding [12], and the
formula used is:

ln ðEY þ Þ U
¼ (35)
k Uτ

where
U: the absolute velocity parallel to the wall in the first node of the grid
Uτ: the friction velocity
k: the Von Karman constant
E: the surface’s roughness parameter
Y+: the dimensionless distance from the wall
The above function is applied to nodes whose dimensionless distance from the
wall is between 30 and 130. In this region of the boundary layer, the effects of
turbulence and viscosity are equally important.

2.3 Boundary conditions

Because of the selection of body-fitted coordinates to design the grid of the

problem, the velocity must be broken down to its components and transferred from
the Cartesian coordinate system to the BFC used in the present study.
At entrance, a uniform flow is applied of Mach 2 speed. The rate of flow is
consistently calculated for each control volume. The kinetic energy of turbulence
and its rate of dissipation are considered constant with values of 21.78 m2/s2 (10% of
the mean velocity squared) and 2.765*105 m2/s3 (assuming turbulence viscosity
1000 times the laminar value), respectively (Figure 1).
The outlet boundary conditions describe the fact that there is no alteration in the
rate of mass flow, as there is neither mass creation nor destruction. As for the
turbulence kinetic energy, turbulence dissipation rate, and the enthalpy, Neumann
conditions prevail, suggesting there is no alteration of values regarding these

Figure 1.
The plane of inlet boundary conditions specification.

122
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

Figure 2.
The plane of outlet boundary conditions specification.

properties along the boundary (Figure 2). Since the boundary conditions must be
transported in the BFC system, the Jacobian of the transformation used is:

∂x ∂y ∂x ∂y
J¼ �
∂g 1 ∂g 2 ∂g 2 ∂g 1

and the general form of the functions:

� �
∂f 1 ∂f ∂y ∂f ∂y
¼ �
∂x J ∂g 1 ∂g 2 ∂g2 ∂g 1
� �
∂f 1 ∂f ∂x ∂f ∂x
¼ �
∂y J ∂g 2 ∂g 1 ∂g 1 ∂g 2

Finally, the boundary condition at the northern boundary is prescribed external

pressure. Like the boundary conditions at the outlet, the kinetic energy of turbu-
lence, its dissipation rate, and the enthalpy are described by Neumann boundary
conditions, implying absence of factors capable of altering their values along the
northern boundary.

2.4 Computational grid and method used

2.4.1 Computational grid generation

The study for grid independence came down to the same results as the one
proposed by Steve Rooks [4]. The computational grid used is one with dimensions
48*45*1 and has been proved to lead to grid-independent results [4]. Briefly, the
grid divides the computational field in three regions along the x-axis and three along
the y-axis, as one can see in the following figure (Figure 3).
Along the y-axis the three sub-regions correspond to the area below the lip, the
area of the lip, and finally the area above the lip. A most important role in the
dimensions of the grid played the fact that a denser or a sparser grid would result in
improper use of the wall functions, as it would affect the dimensionless distance Y+.

2.4.2 Computational methods used

In order to solve the equations, the finite-volume method is employed. As for

regular grids, the finite-volume equations can be derived by integrating over a
control volume in a system with body-fitted coordinates. The use is made of the
Gauss theorem that transforms a volume integral to a surface one:
ð ð ð � ! � ð � �!! �
! ! ! ! ! !
∇ � Q dV ¼ n � Q dS ¼ n � ρV φη dS � n � Γ φη ∇φη dS (36)
ΔV S S S

123
Computational Models in Engineering

Figure 3.
The computational grid.

!
where n is the outward vector drawn normal to any of the six surfaces defining
ΔV and S denotes the enclosing the volume surface area [10]. The first term
represents convection, where properties are regarded as uniform over the cell faces,
except in respect of the velocities for which the face-center values are stored. The
values prevailing at the cell face are determined by using upwind interpolation. As
for the second term, the one representing diffusion the property gradients and the
transport properties that they multiply are uniform over cell faces. In order to
calculate the gradients, it is supposed that properties vary linearly and also the
transport properties are arithmetic averages of those on either side of the cell faces.
In source terms, the nodal values are supposed to prevail over the whole of the cell
volume.
Any dependent variable’s partial differential equation is represented by a
coupled set of algebraic equations of the following form:
X
AP φP ¼ Ai φi þ Sφ (37)
i

where the Ai’s represent the influence of convection and diffusion, Sφ is the
source term, and P refers to the control volume under consideration. Σi indicates the
summation over the neighboring nodes. All the above equations are derived
through work that cannot be presented here due to limited space [10].

2.4.3 Numerical simulation

The numerical simulations in this work are realized by the fluid dynamics
package PHOENICS. The SIMPLEST algorithm is used in the iterative procedure,
whereas the equations are solved via the finite-volume method. The momentum
equations are solved using the initial estimate of the pressure field, using velocities
that do satisfy momentum but not in the general continuity. For each cell, the
equation of (inlet-outlet) is formed, and an equation of pressure correction is solved
where the coefficients are A (d(vel))/dp and the sources are the errors in continuity.
Afterwards, the pressure field is corrected alongside with the velocity field.

124
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

The iterative procedure continues until the errors in the equations of momentum
and continuity are acceptable.

2.5 Relaxation methods used

In order to improve convergence and eliminate great fluctuations in values of

variables in consecutive iterations that could lead to the divergence of the solution,
relaxation techniques are used. In the present study, two techniques are used: the
linear relaxation and the false time-step relaxation. Linear relaxation is used for the
variables of pressure, enthalpy, turbulence kinetic energy, and turbulence dissipa-
tion rate as it is more suitable for scalar variables in the following form: φnew = φold +
α(φnew  φold). In the case of the vector variables of U1 and V1, false time-step
relaxation is used as it is more suitable for the velocity variables in the following
form ρVp/δt(Φ*  Φρ) = β(Φ*  Φρ) where VP is the cell volume, Φ* the previous
iteration value of variable Φ, Φρ the value of φ in the last iteration, and δtfalse the
false time step, generally of the order of the cell residence time.

2.6 Discretization schemes used

In the original coding, the Hybrid discretization scheme was used. However, in
order to investigate further the most efficient way to simulate the phenomenon,
runs were also carried out using the upwind discretization scheme that, according to
literature, is more efficient when describing directional flows but introduces more
“false-diffusion” errors.

2.7 Modifications in the original coding

In order to improve the simulation results both in terms of convergence speed

and agreement with the experimental data, numerous modifications were made to
the original coding. These modifications include the calculation of the diffusion
coefficient by means of the harmonic mean and not the numerical mean, and the
calculation of reference kinematic viscosity for laminar flow was accomplished via
Sutherland’s law. Moreover, the initial values of the variables for enthalpy and
velocity were not chosen randomly, as in the original coding but were calculated
based on the simulation’s data so as to help the convergence procedure with more
“realistic” initial variable values. At the northern boundary, a boundary condition
that suggests constant mass flow was used so as to ignore the effects of diffusion.
Finally, in opposition to the initial coding, where false time-step relaxation was used
for the variables of enthalpy, turbulence kinetic energy, and turbulence dissipation
rate, in the modified version, for these variables, linear under-relaxation was used.

2.8 Different case studies

In terms of this present research, many simulations were carried out, and results
of the most relevant eight of them were brought into comparison in order to decide
which method is more efficient in simulating the phenomenon studied by Markley
et al. [13]. The first simulation carried out in terms of this research was the repro-
duction of the simulation carried out by Steve Rooks [4] in an up-to-date version of
the PHOENICS. The other seven include the modified original version, described in
the above paragraph, although no alteration in either the turbulence model or the
discretization scheme was made. The other six were the combinations of the three
turbulence models (k-ε, RNG k-ε, and k-ω) with the two discretization schemes
(Hybrid and Upwind) studied in this research paper.

125
Computational Models in Engineering

3. Results

In order to examine the validity of the results of the simulation, a comparison

had to be made with the available experimental data. The available experimental
data are for temperature, as that was the main point of interest at the experiment
conducted by NACA [13], measured behind the lip, and high enough so as the flow
would be unaffected by boundary layer phenomena. As a result, measurements
were taken along the line Y = 18 and behind the region dictated by cell X = 20.
The first simulation carried out in terms of this research was the reproduction of
the simulation carried out in [4] using our modified version of the PHOENICS code,
so as comparison can be carried out. The modifications made to the original coding
that led to improved results, both in terms of agreement with the experimental data
and in terms of decreased convergence time, include the calculation of the diffusion
coefficient by means of the harmonic mean rather than the numerical mean and the
calculation of reference kinematic viscosity for laminar flow that was accomplished
via Sutherland’s law. Moreover, the initial values of the variables for enthalpy and
velocity were not chosen randomly, as in the original coding, but were calculated
based on the simulation’s data so as to help the convergence procedure with more
“realistic” initial variable values. At the northern boundary, a boundary condition
that suggests constant mass flow was used so as to ignore the effects of diffusion.
Finally, different to the initial coding, where false time-step relaxation was used for
the variables of enthalpy, turbulence kinetic energy, and turbulence dissipation
rate, in the modified version, for these variables, linear relaxation was used.
For this initial modified version, no alterations were made as far as the turbu-
lence modeling and the discretization scheme are concerned. The initial modeling
with k-ε turbulence modeling and the Hybrid discretization scheme were used. As a
result of the above changes, the convergence of the coding was faster by 48.33% as
convergence was reached at 5427 iterations instead of 10,503 for the same level of
errors remaining, and there were even slight improvements in the agreement with
the experimental results as well.
The following diagram depicts the distribution of temperature along the x-axis
above the airfoil at the height of the line Y = 18, as mentioned above (Figure 4).
As seen in the above diagram, the two simulations differ only slightly, and it can
be noted that the modified version offers a marginal 0.04% better accuracy along
with the faster convergence. More precisely, at the point where the temperature
drop is noticed, which is the lip region, the accuracy is improved by 0.015%,

Figure 4.
Comparison diagram for temperature distribution along the x-axis.

126
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

whereas after the threshold of 5.3 cm, which is behind the lip region, the accuracy is
improved by 0.27%.
Likewise, the diagrams depicting the main velocity and density distribution
along the x-axis slightly differ from the diagrams of the initial simulation since the
mean fault is 0.5847 and 0.9%, respectively (Figures 5 and 6).
Regarding the distributions of temperature, velocity and density along the
Y-axis, defined by the line X = 20 (Figures 7–9), no major difference is noticeable in

Figure 5.
Comparison diagram for velocity distribution along the x-axis.

Figure 6.
Comparison diagram for density distribution along the x-axis.

Figure 7.
Comparison diagram for temperature distribution along the y-axis.

127
Computational Models in Engineering

Figure 8.
Velocity distribution along Y-axis.

Figure 9.
Density distribution along Y-axis.

their main part, as one can notice from Figures 7–9. The mean errors are 0.7, 3.06,
and 0.9%, respectively. Yet, reaching the northern boundary, the results from the
modified simulation differ from the initial one, as they do not alter the value they
have settled in following the exit of the boundary layer. In the initial simulation,
an increase can be spotted for the variables of temperature and density and a
decrease in the value of velocity. The increase in the temperature value along with
the decrease in velocity can be explained from the fact that enthalpy is assumed
constant. However, in the modified simulation, the steady mass flow boundary
condition neglects diffusion phenomena, and therefore there is stability in this
variable. This approaches the nature of the original experiment, as no factors able to
alter the values of these variables exist near the northern boundary.
Figures 10–13 present the contour plots for temperature, density, and velocity
distributions as well as the vector diagram of the velocity vectors.
In order to investigate further the problem and any possible aspects that may
improve the quality of the results produced, a parametric study was conducted. The
parameters studied were both the turbulence model and the discretization scheme.
Simulations were conducted using all three different turbulence models: k-ε, RNG
k-ε, and k-ω. Each of the above turbulence models was used in combination with
either the Hybrid discretization scheme or the Upwind.

128
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

Figure 10.
Contour plot of the temperature distribution on the computational field for the modified simulation
(k-ε hybrid).

Figure 11.
Contour plot of the velocity distribution on the computational field for the modified simulation (k-ε hybrid).

Figure 12.
Contour plot with the density distribution on the computational field for the modified simulation (k-ε hybrid).

In total, six different simulations were performed. The results were compared to
the available experimental data in order to evaluate their accuracy. The schematic of
the comparison can be seen in the diagram that follows (Figure 14).

129
Computational Models in Engineering

Figure 13.
Velocity vectors zoomed in the area of interest depicting the recirculations.

Figure 14.
Comparison diagram of parametric simulations and experimental data.

The simulation results can also be seen in numbers in the following table, where
the mean error is calculated (Table 1).
As one can see, the most successful simulation in terms of the overall mean error
is the one engaging the k-ω turbulence model in combination with the Upwind
discretization scheme. If we take a closer look, it is obvious that it is the only
simulation that seems to approach the steep temperature descent in the 4–5 cm
region. In that region, all simulations fail to keep the mean error below 5%; how-
ever, as one can see in the following table, the k-ω Upwind simulation achieves a
mean error of 5.6274% (Table 2).
The difficulty in simulating realistically the behind the lip region is possibly due
to heat conduction phenomena. In the simulations performed, no solution is run for
conjugate heat transfer, and given that the airfoil geometry is very thin, heat

k-ε k-ε RNG k-ε RNG k-ε k-ω k-ω

Hybrid Upwind Hybrid Upwind Hybrid Upwind

Overall mean 4.9529 3.5055 4.9729 3.5741 5.034 3.3418

error (%)

Table 1.
Overall mean error calculation.

130
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

k-ε k-ε RNG k-ε RNG k-ε k-ω k-ω

Hybrid Upwind Hybrid Upwind Hybrid Upwind

Mean error in region 8.42 6.763 8.4331 6.8564 8.0097 5.6274

4–5 cm (%)

Table 2.
Mean error calculation in the 4–5 cm region.

Figure 15.
Comparison diagram between simulation performed and experimental data in the behind the lip region.

k-ε k-ε RNG k-ε RNG k-ε k-ω k-ω

Hybrid Upwind Hybrid Upwind Hybrid Upwind

Mean error in behind the lip 6.5465 0.3348 6.6621 0.6050 8.0915 1.9930
region (%)

Table 3.
Mean error calculation in behind the lip region.

conduction taking place on the edge of the lip is speculated to be one among the
causes that lead to higher calculated temperatures than the experimental data.
When examining the region behind the lip, where the flow presents a more
stabilized form than the one when encountering the lip, the turbulence model that
achieves better accuracy is the RNG k-ε model with the Upwind discretization
scheme. This can be seen in both the diagram and the table that follow (Figure 15
and Table 3).
It must be noted that in all simulations performed, the discretization scheme that
produced the best results was the Upwind scheme. That would be justified by the
fact that the flow is basically unidirectional and is characterized by high velocity.
As a result, the diffusion phenomena are of very low intensity, with the convective
phenomena being the main terms in the discretization equation. In both k-ω and k-ε
simulations ran, the benefits from altering the initial code were still valid.

4. Recommendations

The calculation of aerodynamic heating phenomena includes many aspects that

can interfere with both the speed of convergence but the accuracy of the results
as well. Apart from the choice of proper initial value and other numeric factors,
attention has to be paid to the flow field characteristics in order to decide the
suitability of the turbulence model and the discretization scheme that will be
131
Computational Models in Engineering

used in order to achieve better results. Despite the fact that improvements for
the 4–5 cm region were also suggested here, future research could focus on
simulating the lip region, as this study’s main interest lied in the behind the lip
region. In order to achieve that, high-order discretization schemes could be used, like
the Van Leer scheme as well as high-order turbulence models, like the large eddy
simulation (LES) and the direct numerical simulation (DNS) or any combinations.

5. Conclusion

Heat transfer and pressure measurement on a 5-inch hemispherical concave

nose at a Mach number of 2.0 is investigated via the use of PHOENICS to evaluate
aerodynamic heating phenomena. In continuation of [4], the code is modified
in order to achieve better results both in terms of faster convergence but also in
terms of more accurate results as well. Finite-volume method is used, while
turbulence models and discretization schemes are studied as parameters in order
to optimize results. Before altering the above mentioned parameters, attempts
were made to improve the speed of convergence. Altering the values of input
variables in order to be closer to the problem considered as well as choosing
relaxation methods suitable for each variable type, scalar or vector, along with a
few other changes described above, managed to improve the convergence speed by
up 48.33%. Moreover, an improvement in boundary conditions of the northern
boundary so as to ignore dissipation phenomena corrected the results of all variables
at the northern region of the computational field, as their values were stabilized, as
expected. In the parametric study, the main observation was that the discretization
scheme plays a major role in the quality of the results, as the flow is unidirectional
and is characterized by high speed of 2.0 Mach. As a result, dissipation phenomena
are not comparable to convection, which plays the major role. The choice of the
Upwind discretization scheme seems to be more accurate, as it considers the fact
that the flow is unidirectional and provides results improved by more than 1.5% for
given turbulence model on the whole computational field. That improvement can go
up to 6% when discussing regions behind the lip, where the flow has been stabilized.
The parametric study regarding the turbulence model produced interesting
results, as well. In order to evaluate solely the impact of the turbulence model, the
discretization scheme used was the Hybrid, as in the original. As it appears, there is
not only one fitting turbulence model for this simulation, as their suitability changes
with regions. When studying the flow field in the lip region (4–5 cm), the k-ω
turbulence model gave the best results, improved by 0.5% when compared to the
original simulation performed with k-ε turbulence model. The improvement intro-
duced by the use of the k-ω model in that area is due to the fact that this model can
calculate better the effect of the created vortices. When both k-ω turbulence model
and Upwind discretization scheme were implied, the total improvement in that area
of interest went up to almost 3% (5.6274% discrepancy from the experimental data,
compared to 8.42%). However, the study showed that behind the lip (x > 5.2),
where flow has been stabilized, the use of the RNG k-ε turbulence model combined
with the Upwind discretization scheme improved the mean error by more than 6%
(0.3348% compared to 6.5465).

Acknowledgments

We would first like to thank our colleagues in DRA Farnborough, who

performed the original step in this modeling effort. Their inspired research was the

132
Mathematical Modeling of Aerodynamic Heating and Pressure Distribution on a 5-Inch…
DOI: http://dx.doi.org/10.5772/intechopen.91041

basis of this current research paper. Moreover, we would also like to thank Dr. Mike
Malin, Technical Support Manager of CHAM Ltd., whose assistance was truly
helpful to go over technicalities that appeared along the way of this present
research.
PHOENICS is licensed by CHAM Ltd., London, UK. The present model may
become available from the present authors.

Author details

Dimitrios P. Spiridakos1*, Nicholas C. Markatos1,2 and Despoina Karadimou1

1 National Technical University of Athens, Athens, Greece

2 Texas A&M University at Qatar, Qatar

*Address all correspondence to: jamesspir92@gmail.com

© 2020 The Author(s). Licensee IntechOpen. This chapter is distributed under the terms
of the Creative Commons Attribution License (http://creativecommons.org/licenses/
by/3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work is properly cited.

133
Computational Models in Engineering

References

[1] Thompson J, Birk AM, [10] Galea ER, Markatos NC. The
Cunningham M. Design of infrared mathematical modeling and computer
signature suppressor for the bell 205 simulation of fire development in
(UH-1H) helicopters, part 1: aircraft. International Journal of Heat
Aerothermal design. In: Seventh CASI and Mass Transfer. 1990;34:181-197
Propulsion Symposium. 1999
[11] Swanson BW, Solutions of the
[2] Riley CJ, DeJarnette FR. An Three-Dimensional Navier-Stokes
Engineering Aerodynamic Heating Equations in Non-orthogonal
Method for Hypersonic Flow. American Coordinates to Calculate the Flow in a
Institute of Aeronautics and Long Spiral Impeller, ASME Paper 82-
Astronautics; 1992 GT-268. 1982
[3] PHOENICS. VR Reference Guide: [12] Launder BE, Spalding DB: The
Documentation for Phoenics (TR 326), Numerical Computation of Turbulent
Version 2010. London, UK: CHAM; Flows. 1973
2010
[13] Thomas Markley J. Heat Transfer
[4] Rooks S. Preliminary Investigations
and Pressure Measurement on a 5-Inch
into the Use of PHOENICS for
Hemispherical Concave Nose at a Mach
Evaluation of Aerodynamic Heating
Number of 2.0. National Advisory
Phenomena. DRA Farnborough; 1996
Committee for Aeronautics; 1958
[5] Murakami K, Yamamoto Y,
Rouzand O. CFD Analysis of
Aerodynamic Heating for HYFLEX
High Enthalpy Flow Tests and Flight
Conditions. 24th International Congress
of the Aeronautical Sciences; 2004

[6] Jianwei L, Qiang W. Aircraft-skin

infrared radiation characteristics
modeling and analysis. Chinese Journal
of Aeronautics; 2009

[7] Johansson M, Dalenbring M.

Calculation of IR Signatures from
Airborne Vehicles. FOI, Swedish
Defence Research Agency; 2006

[8] Palharini RC, Santos WFN. Surface

Temperature Effects on the
Aerodynamic Surface Quantities of
Hypersonic Cavity Flow, Combustion
and Propulsion Laboratory (LCP).
National Institute for Space Research
(INPE); 2010

[9] Johnson KR, Rodriguez LJ. Technical

Reference Guide for GTSIG/TCM2.
Atlanta, GA: Georgia Tech Research
Institute; 1990

134
 Edited by Konstantin Volkov

The accurate prediction of multi-physical and multi-scale physical/chemical/mechanical

processes in engineering remains a challenging problem despite considerable work
in this area and the acceptance of finite element analysis and computational fluid
dynamics as design tools. This book intends to provide the reader with an overview
of the latest developments in computational techniques used in various engineering
disciplines. The book includes leading-edge scientific contributions of computational and
applied mathematics, computer science and engineering focusing on the modelling and
simulation of complex engineering systems and multi-physical/multi-scale engineering
problems. The following topics are covered: numerical analysis and algorithms, software
development, coupled analysis, multi-criteria optimization as they applied to all kinds of
applied and emerging problems in energy systems, additive manufacturing, propulsion
systems, and thermal engineering.

ISBN
ISBN978-1-78985-400-8
978-1-78923-869-3

Published in London, UK
© 2020 IntechOpen
© wacomka / iStock