Professional Documents
Culture Documents
MECHANICAL DESIGN,
MATERIALS AND
MANUFACTURING
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or
by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no
expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No
liability is assumed for incidental or consequential damages in connection with or arising out of information
contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in
rendering legal, medical or any other professional services.
MECHANICAL ENGINEERING THEORY
AND APPLICATIONS
MECHANICAL DESIGN,
MATERIALS AND
MANUFACTURING
SANDIP A. KALE
EDITOR
Copyright © 2019 by Nova Science Publishers, Inc.
All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted
in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying,
recording or otherwise without the written permission of the Publisher.
We have partnered with Copyright Clearance Center to make it easy for you to obtain permissions to
reuse content from this publication. Simply navigate to this publication’s page on Nova’s website and
locate the “Get Permission” button below the title description. This button is linked directly to the
title’s permission page on copyright.com. Alternatively, you can visit copyright.com and search by
title, ISBN, or ISSN.
For further questions about using the service on copyright.com, please contact:
Copyright Clearance Center
Phone: +1-(978) 750-8400 Fax: +1-(978) 750-4470 E-mail: info@copyright.com.
Independent verification should be sought for any data, advice or recommendations contained in this
book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to
persons or property arising from any methods, products, instructions, ideas or otherwise contained in
this publication.
This publication is designed to provide accurate and authoritative information with regard to the subject
matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in
rendering legal or any other professional services. If legal or any other expert assistance is required, the
services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS
JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A
COMMITTEE OF PUBLISHERS.
Additional color graphics may be available in the e-book version of this book.
Preface vii
Chapter 1 Finite Element Analysis of Three DoF Strain
Gauge Force Transducer 1
Ankur Jaiswal, H. P. Jawale and Atul Sane
Chapter 2 The Basic Principle of Calculation and Analysis of
the Defective Structure of Solids 17
V. I. Talanin, I. E. Talanin, V. V. Zhdanova,
D. I. Yakymchuk and A. V. Rybalko
Chapter 3 Advanced Methods Used in Molecular Dynamics
Simulation of Macromolecules 57
Hiqmet Kamberaj
Chapter 4 Study on Error in Sigmoidal Function Generation
of 4R Mechanism 135
Ankur Jaiswal and H. P. Jawale
Chapter 5 Identifying Transfer Vertex from the Adjacency
Matrix for Epicyclic Gear Trains 153
Mallu Chengal Reddy, Rudraraju Manish
and Yendluri Daseswara Rao
vi Contents
Sandip A. Kale
Editor
In: Mechanical Design, Materials … ISBN: 978-1-53614-791-9
Editor: Sandip A. Kale © 2019 Nova Science Publishers, Inc.
Chapter 1
ABSTRACT
Corresponding Authors Emails: jaiswalankur13@gmail.com; hpjawale@rediffmail.com.
2 Ankur Jaiswal, H. P. Jawale and Atul Sane
1. INTRODUCTION
Strain gauge based force and torque sensors are widely used in
measurement technology due to their stability at room temperature and
linear static characteristics. In such applications strain gauges are mounted
on the primary sensing elements of a desired cross-section. The applied
force or torque gets converted into displacement making the strain gauge to
change the resistance which is proportional to the applied force or torque.
There are different configurations of primary sensing elements that are
devised and calibrated to give desired sensitivity. However, inherently
these configurations also have moment components in supplement to the
force component. This makes the conversion of resistance into the readable
electrical signals complex. Additionally, due to the cross sensitivity,
analysis becomes tedious.
The performance of the strain gauge is a function of all other elements
like the position of the gauge, gauge factor, input voltage applied, bridge
connection and other circuitry to estimate the strain in terms of voltage.
Thus the location of dipicted strain gauges directly affects the
performance. Measurement of externally applied forces can be better
estimated with the accurate positioning of the strain gauge on the
transducer at the place of maximum strain.
The six-axis force sensor with four T-shaped bars and wrist force
sensor is seen to be analyzed using the FEM and analytical method. The
optimized design is used for high measurement sensitivities application
(Chao and Chen, 1997, Liu and Tzo, 2002). This chapter elaborates a new
frame/truss type of sensor body design. The unified criteria are based on
static and dynamic stiffness of the sensor and sensitivity for strain gauge.
The results have been compared with the previous design (Bayo and
Stubbe, 1989) on the basis of rigidity, force sensitivity and flexibility
parameters.
Finite Element Analysis of Three DoF Strain Gauge … 3
L
R
A (1)
R / R R / R
GF
L / L (2)
where,
ρ: Resistivity/Conductivity constant
L: Length of the sensor
A: Cross sectional area of sensor
R: Electrical resistance of the sensor
ε: Mechanical Strain
3. TRANSDUCER DESIGN
This chapter discusses a novel three DOF force transducer which has a
special shape that allows strategic placement of the strain gauges in Figure
6 Ankur Jaiswal, H. P. Jawale and Atul Sane
1. Because of such arrangements they only sense axial forces and ignore
the moments. This makes measurements and calculations very simple. The
simplest primary sensing elements of any force sensing transducer is
generally similar to a cantilever configuration. Combining three cantilevers
for each of the axis of Cartesian coordinate system in a single
configuration can function as a three DOF transducer for force sensing.
This design felicitates to study the application and results of uniaxial,
biaxial and tri-axial force loading with the means of pulleys and dead
weights directly on the transducer.
Figure 1. A CAD model describing the Force Transducer (green), Pulleys and dead
weight for loading (brown).
(a) (b)
Figure 2. CAD model depicting the complete assembly of FT1 and FT2.
Finite Element Analysis of Three DoF Strain Gauge … 7
There is also a new proposed design for the same transducer having
some alterations to the frame and the members so that the applied force on
the same region is more easily distributed in Figure 2. This helps in better
stress dispersion.
2
r x y z
2 2
(3)
y y
a cos a cos
2 2 r
x y z
2
(4)
x
a tan
z (5)
y r cos (7)
The material considered for design and analysis is Structural steel with
following in Table 1.
4.2 Meshing
Figure 4. Final Meshing of the geometry for the analysis of FT1 and FT2.
5. RESULTS
(a) (b)
(c) (d)
(e) (f)
6. OBSERVATIONS
16
Force in X- direction (FT1)
Maximum Principal Strain (*e5)
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Load (N)
4
X direction
3.5
Y direction
3 Z direction
Strain Difference
XYZ diection
2.5
1.5
0.5
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Load (N)
CONCLUSION
REFERENCES
Chapter 2
ABSTRACT
To describe the defective structure of semiconductor silicon, a triad
is created: physical plus mathematical models - computational algorithm -
program. This solution can be used both in studying the properties of
crystals and in industrial production. Such a solution is the basic principle
in the study of structural imperfections in any solid. A structural diagram
of the diffusion model for the formation of structural imperfections in a
crystal and a computational algorithm is presented.
Corresponding Author Email: v.i.talanin@mail.ru.
18 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
1. INTRODUCTION
∆𝐺(𝑅)
𝑓0 (𝑅)~𝑒𝑥𝑝 {− }
𝑘𝑇
𝑇𝑚2
𝑇(𝑡) =
𝑇𝑚 + 𝑉𝑔 𝐺𝑎 𝑡
~
f SiO2 I SiO2
O ~ (2)
~
f SiC t I SiC
0 ~
tc c
t
where is the dimensionless time. The time constants in (2) are
to
given by tO (nOcr ,0 )2 / g SiO
0
2
; tc (nccr ,0 )2 / g SiC
0
, where the critical growth
rates of the precipitates are defined as
0
g SiO2
NO0O exp( Gact
SiO2 0
/ kT ); g SiC NO0c exp(Gact
SiC
/ kT ) .
~ f SiO ~ f
f SiO2 0 2 ; f SiC SiC0
(3)
f SiO2 f SiC
The particles fluxes on right hand sides of system of equations (2) are
described as
~ BSiO ~ B
ASiO2 ( g~SiO2 d SiO2 )nOcr ,0 ~ 2 ; ASiC ( g~SiC d SiC )nOcr ,0 ~SiC ; (4)
O O
~ BSiO ~ B
ASiO2 ( g~SiO2 d SiO2 )nOcr ,0 ~ 2 ; ASiC ( g~SiC d SiC )nOcr ,0 ~SiC ; (5)
O O
~ ~
g~SiO2 d SiO2 g~SiC d SiC
BSiO2 ; BSiC (6)
2 2
g SiO g ~ d SiO ~ d
g~SiO2 0 2 ; g~SiC SiC
0
; d SiO2 0 2 ; d SiC SiC
0
(7)
g SiO2 g SiC g SiO2 g SiC
2uV p
rOcr (8)
kT ln( S0 Si i Sv v ) 6uV p
26 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
2uV p
rCcr (9)
kT ln( Sс Si i Sv v ) 6uV p
where So Co Coeg , Sc
Cc
, Si
Ci
, Sv
Cv
Cceg Cieg Cveg
formation the nuclei of new phase, and clusters growth, and stage of
coalescence (Talanin & Talanin 2004). This process is a first-order phase
transition.
The growth of clusters is accompanied by a significant decrease of the
supersaturation solid solution. Their number in the second stage does not
change. This process is described by a system of equations (Talanin &
Talanin 2011):
dN 0
k0 NN 0 g1 N1 ,
dt
N i ki N gi gi 1 N i 1 ki 1 NN i 1 ,
dN i (10)
dt
dN
N ki N i g i N i ,
dt i 0 i 1
the initial instant of time. Therefore, the average number of particles at the
nucleation centers can be represented in the form
28 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
iN i
N 0 N t
i i 0
(11)
N i 0
i Nc
C i, t ) 2
N t N E k i C i, t N t N E 2 k i C i, t (12)
t i 2i
k0 N N E it m ,
di
(13)
dt
dN t
k0 Nc1 N t N E N 0 mNc N t
(14)
dt
N t N E
1
exp Nc 1 N 0 N E k0t 1
(15)
N 0 N E
The Basic Principle of Calculation and Analysis ... 29
3bit
Rt 3 (16)
4
f R, t
f R, t dR / dt 0, f0 R f R,0 (16a)
t R
QTc cc t f R, t R3dR (17)
0
L
Tc t Tc 0 f R, t f R R dR
3
(18)
c p t
0
0
where expression (16) is the equation of continuity in the space of sizes for
the size distribution function of precipitates, expression (17) is the mass
balance equation, expression (18) is the equation accounting for the
amount of released heat, f 0 R is the initial size distribution function of
precipitates, Tc 0 is the initial temperature of the solid solution,
4 / 3 is the volume per atom in the precipitates of the new phase,
QTc is the total amount of the material in the precipitates and the
solution, cc t is the solute concentration in the solid solution, L is the
heat of the phase transition per atom of the precipitated phase, c p is the
heat capacity at constant pressure per unit mass of the solid solution, and
е is the density of the solid solution.
Equations (16) - (20) form a complete system of equations describing
the process of nonisothermal decomposition of the solid solution in the
coalescence stage. The solution of equations (16) - (20) is carried out by
the method developed in (Talanin & Talanin 2011). The simultaneous
nucleation and growth of new phase particles (oxygen and carbon
precipitates) during the cooling of silicon crystals after growth leads to a
strong interaction between the evolution processes of these two subsystems
of grown-in microdefects Absorption of increasing vacancy oxygen
precipitates leads to the emission of silicon atoms into the interstitial. The
intrinsic interstitial silicon atoms, in turn, interact with growing carbon
precipitates, which in the process of their growth supply vacancies for
growing oxygen precipitates. Such interaction leads to an accelerated
transition of the subsystems of oxygen and carbon precipitates to the
coalescence stage, in comparison with the independent evolution of these
two subsystems (Talanin & Talanin, 2011).
The Basic Principle of Calculation and Analysis ... 31
∞
𝐾1,2 (𝑟)
𝑈(𝑟) = 𝜆𝑘𝑇 ∫ 𝐾1,2 (𝑟)𝑒𝑥𝑝 (− ) 𝑑𝑟
𝑘𝑇
−∞
𝐾1,2 (𝑟)
𝜌(𝑟) = 𝜆𝑘𝑇𝑒𝑥𝑝 (− 𝑘𝑇
) (21)
(Vlasov, 1950) from the main criterion for the existence of the crystalline
state, and the crystallization condition can be written as follows:
4𝜋𝑁 ∞ ∗
∫ 𝐾1,2 (𝜌)𝜌2 𝑑𝜌
𝑘𝑇𝑚 0
=1 (22)
∗
where 𝑁 is the number of particles; 𝐾1,2 = −𝐾1,2 (Vlasov, 1978).
For the crystal case, Vlasov showed that the state space is determined
by the temperature changes (Vlasov, 1950). The density of complexes
distribution we defined as a function of crystal cooling temperature
𝑈1𝑚𝑖𝑛,2𝑚𝑖𝑛
𝜌(𝑇) = 𝜆𝑘𝑇𝑒𝑥𝑝 (− 𝑘𝑇
) (23)
where 𝑈1𝑚𝑖𝑛 and 𝑈2𝑚𝑖𝑛 are the minima of interatomic potentials in silicon-
oxygen and silicon-carbon complexes, respectively.
Vlasov showed that the spatial periodic distribution is one of the
particular states of particle motion (Vlasov, 1950). This solution was
carried out by Vlasov for an ideal (defect-free) crystal.
However, real crystals always contain structural imperfections. For
defective (real) crystals, we considered the possibility of pair formation of
an IPD-impurity with subsequent impurity precipitation (Talanin &
Talanin, 2016b). The potential of interatomic interaction is the basis for
describing the interaction between the atoms of matter. Analytic
consideration of the potentials of the interaction of two atoms is extremely
difficult, since after quantum-mechanical calculations they are represented
as functions with a large number of parameters. In this connection, model
potentials with a small number of parameters are used (Mazhukin et al.,
2014; Mattoni et al., 2007). To estimate the formation parameters of
silicon-carbon and silicon-oxygen complexes, we presented interaction in
form of Mee-Lennard-Jones potential.
To determine the characteristic numbers of silicon-oxygen (𝜆1 ) and
silicon-carbon (𝜆2 ) complexes, we use (22) for the number of particles in
𝐾1,2
the complex 𝑁 = 2 and 𝐾(𝜌) = 𝑘𝑇 (1 − 𝑒 − 𝑘𝑇 ). The calculation gave the
34 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
4. FORMATION OF SECONDARY
STRUCTURAL IMPERFECTIONS
theory and suggest the calculation of the formation of stable nuclei and the
kinetics of diffusion-limited growth of defects. The discrete-continuum
approximation (discrete-continuous model) involves a comprehensive
approach: the solution of discrete equations for the smallest defects and the
solution of the Fokker–Planck equations for large-sized defects (Brown
et al., 2001).
We calculated the formation of microvoids and interstitial dislocation
loops in accordance with the strict approximation of the model of PD
dynamics, provided that there is no high-temperature recombination of
IPDs. We confirm the homogeneous character of microvoid formation
process (Talanin & Talanin, 2010a). Microvoids are formed on the already
existing precipitates. The data from the computational experiment for
determining the concentration of vacancy microvoids correlate well with
the experimentally observed results (104-105 cm-3) (Kulkarni et al., 2004).
For dislocation loops, where the experiments give a value of ~ 106-107 cm-3
(Brown et al., 2001), the discrepancy reaches three orders of magnitude.
This may be due to the fact that, in contrast to microvoids, which are
formed only by the action of the coagulation mechanism, the formation of
dislocation loops occurs due to the action of both the coagulation
mechanism and the mechanism of prismatic extrusion (deformation
mechanism). The results of calculations show that the main contribution to
the formation of dislocation loops is made by the mechanism of prismatic
extrusion, when the formation of interstitial dislocation loops is associated
with the removal of stresses around the growing precipitate.
We know that the kinetics of high-temperature precipitation
encompasses three stages: the inception of a new phase, the growth of
sediment and the stage of fusion. The elastic interaction between IPDs and
impurity atoms causes the formation of precipitates. The transfer of excess
(scarce) matter from the sediment or vice versa is caused by elastic
deformations and associated mechanical stresses. During the growth of the
precipitate, its coherence with the matrix is lost. The formation and
displacement of dislocation loops is due to the structural relaxation of
precipitates.
The Basic Principle of Calculation and Analysis ... 37
a1 a2
(24)
a1
xx xy xz
xy yy yz pr , (25)
zz
zx zy
32
E pr J 2 R3pr , (26)
45 1
J b 2 RD 2 RD
Eloop ln 2 , (27)
2 1
f
where f is the radius of the core loop; b is the magnitude of the Burgers
vector. A critical value of the precipitate radius corresponds to a value at
which the loop is formed on the precipitate (Kolesnikova et al., 2007)
3b 1.08Rcrit
Rcrit ln , (28)
8 1 b
R(t ) Rcrit
2
j D(t ) t , (29)
M (t )
N (t ) , (30)
1 D(t ) t 2 Rcrit
2
The results obtained radically change the physical picture of our ideas
of defect formation in crystals. The entire process is responsible for high-
temperature precipitation, which occurs from the time the crystal is grown
until its end. Microvoids and dislocation loops are derived from the
emerging precipitate structure, their formation is determined by the
specific growth conditions of semiconductor silicon. The formed defect
structure, in turn, is the basis for the transformation processes that occur in
the crystal as a result of technological treatments when creating various
devices. The computational experiments carried out made it possible to
describe the process of defect structure formation.
For a theoretical description of the defect structure of semiconductor
silicon, we used several mathematical models with different levels of
complexity of the computational experiment. In the framework of the
diffusion model, they can be represented in the form of Figure 1
Each of them to some extent describes the details of the general picture
of physical phenomena occurring in the original system (the formation of
structural imperfections during the growth of semiconductor silicon).
The Basic Principle of Calculation and Analysis ... 41
The method that determines and calculates the growth defect structure
is based on a parameter that connects the main thermal conditions of
crystal growth Vg V
Ccrit . The condition g Ccrit in real crystals
Ga Ga
agrees to the V-shaped distribution of precipitates. In Figure 3 shows a
standard picture of the distribution of growth microdefects in dislocation-
free single crystals of silicon grown by the Czochralski method with
variable growth rate. The crystal diameter was 50 mm, the growth rate
varied from 𝑉𝑔𝑚𝑖𝑛 = 0.5 mm/min to 𝑉𝑔𝑚𝑎𝑥 = 3.0 mm/min.
Three areas of defect formation can be identified: region 1 (above the
V-shaped precipitates distribution), region 2 (V-shaped precipitates
distribution region) and region 3 (below the V-shaped precipitates
distribution). In small-sized silicon crystals in region 1, growth
microdefects (precipitates) of interstitial and vacancy types are formed at
comparable concentrations; in region 2 within the V-shaped distribution
the same defects as in region 1 are formed, and outside the V-shaped
distribution interstitial dislocation loops and predominantly interstitial
precipitates are formed. In crystals with a diameter of more than 80 mm in
44 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
vacancy. We will estimate the recombination parameters for high and low
temperature regions in germanium
High temperature area. The temperature dependence of the
configuration entropy for the defect model considered above can be
represented in the following form (Hodge, 1997):
CONCLUSION
In turn, defect engineering allows you to control the properties and quality
of the materials used.
REFERENCES
Cowern N.E.B., Simdyankin S., Ahn C., Bennett N.S., Goss J.P.,
Hartmann J.M., Pakfar A., Hamm S., Valentin J., Napolitani E.,
Salvador D., Bruno E., Mirabella S. (2013). “Extended point defects in
crystalline materials: Ge and Si.” Phys. Rev. Letters. 110: 155501.
Cristian J. W. (1965). “The theory of transformations in metals and
alloys.” London: Pergamon Press.
Ginzburg V. L., Landau L. D., Leontovich M. A. and Fok V. A. (1946).
“About inconsistency of works by A. A. Vlasov on general theory of
plasma and physics of solid body.” J. Exp. Theor. Phys. 16: 246-252.
Gosele U., Frank W., and Seeger A., (1980). “About the secret of self-
interstitial atoms in silicon.” J. Appl. Phys. 23: 361-367.
Gosele U., Frank W., and Seeger A., (1983). “An entropy barrier against
vacancy-interstitial recombination in silicon.” Solid State Commun.,
45: 31-33.
Hodge M.I. “Adam-Gibbs formulation of enthalpy relaxation near the glass
transition”. (1997). J. Res. Natl. Inst. Stand. Technol. 102: 195-202.
Hüger E., Tietze U., Lott D., Bracht H., Bougeard D., Haller E.E., and
Schmidt H. (2008). “Self-diffusion in germanium isotope multilayers
at low temperatures.” Applied Physics Letters 93: 162104.
Kolesnikova A. L., Romanov A. E., and Chaldyshev V. V. (2007).
“Elastic-energy relaxation in heterostructures with strained
nanoinclusions.” Physics of the Solid State. 49: 667-674.
Kulkarni M. S., Voronkov V. V. and Falster R. (2004). “Quantification of
defect dynamics in unsteady-state and steady-state Czochralski growth
of monocrystalline silicon.” J. Electrochem. Soc. 151: G663-G669.
Kulkarni M. S. (2005). “A selective review of the quantification of defect
dynamics in growing Czochralski silicon crystals.” Ind. Eng. Chem.
Res. 44: 6246-6263.
Lemke H., and Sudkamp W., (1999). “Analytical approximations for the
distributions of intrinsic point defects in grown silicon crystal.” Phys.
Stat. Sol. (a). 176: 843-865.
Mattoni A., Ippolito M. and Colombo L. (2007). “Atomistic modeling of
brittleness in covalent materials.” Phys. Rev. B, 76: 224103-224111.
54 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
Chapter 3
Abstract
Molecular dynamics (MD) simulations at atomic level have widely
been used in studying macromolecular systems, such as protein, DNA
and their complexes, mainly because the classical statistical mechanic’s
laws can explain different phenomena occurring at specified experimental
conditions. In this study, we will present the most advanced methods used
in the MD simulation of macromolecular systems. Furthermore, a discus-
sion of applications of these methods and perspective on developing new
approaches will be introduced. This study aims to review the methods
that are developed to enhance the conformation sampling of molecular
simulations, in particular, for observing rare events in complex molecular
systems. In the summary, we also present a discussion and perspective
∗ Corresponding Author Email: h.kamberaj@gmail.com
58 Hiqmet Kamberaj
on the methods described in this chapter and propose the new possible
improvement of these approaches, which could result in further enhance-
ment of conformation sampling.
1. Introduction
MD approach at atomic resolution is often used to study complex biomolecu-
lar systems (M. Karplus and J.A. McCammon, 2002), mainly because the clas-
sical statistical mechanic’s laws can explain different phenomena occurring
at specified experimental conditions (van Gunsteren et al., 2006). In partic-
ular, MD is used to study the internal fluctuations (A. Amadei et al., 1993;
M. Karplus and J.A. McCammon, 2002), protein folding dynamics (Rogal and
Bolhuis, 2008), transition path sampling (P. G. Bolhuis et al., 2002), protein-
DNA, protein-protein and protein-ligand complexes, and free energy calcula-
tions (S. L. Seyler and O. Beckstein, 2014). However, standard MD simula-
tion has limited time and size scale, which makes it difficult to study typi-
cal phenomena of macromolecular systems, such as slow conformation mo-
tions (Clarage et al., 1995; Palmer, 1982). Therefore, it has been argued else-
where (G. Ciccotti and E. Vanden-Eijnden, 2015; M. K. Transtrum et al., 2015)
that these limitations may be avoided by employing new statistical and compu-
tational approaches to be studied efficiently.
There have been different efforts in developing new approaches for en-
hancement of conformation sampling of simulations using MD technique, as
discussed elsewhere (van Gunsteren et al., 2006). These MD approaches are
used in many applications for lowering conformation transition barriers by in-
creasing the rate of rare events occurrence by introducing a bias that can be
rigorously removed a posterior, or even without the bias term. Different ap-
proached have been used to introduce bias during MD simulation, such as by
changing the shape of potential energy surface until a (quasi) flat landscape is
obtained (A. Piela et al., 1989; Hamelberg et al., 2004b; Hünenberger and van
Gunsteren, 1997; Laio and Parrinello, 2002a), using soft-core potential interac-
tions (Hünenberger and van Gunsteren, 1997), conformational flooding (Grub-
Advanced Methods Used in Molecular Dynamics ... 59
tions. For these reasons, the Reference System Propagator Algorithm (RESPA)
method is introduced to reduce computational efforts for simulations of such
system (Tuckerman and Berne, 1991a,b; Tuckerman et al., 1990, 1991). The
time-reversible forms of the RESPA methods have also been developed, named
r-RESPA, which have shown to be very stable concerning the order and stability
of numerical integrators (M.E. Tuckerman et al., 1992). The r-RESPA, which
will be discussed below in more details, uses Trotter factorization of the classi-
cal Liouville propagation operator (Creutz and Goksch, 1989; H.D. Raedt and
B.D. Raedt, 1983; Takahashi and Imada, 1984).
Following the discussion in literature (M.E. Tuckerman et al., 1992) (see
also Ref. (M.P. Allen and D.J. Tildesley, 1989)), for a system with f degrees of
freedom the Liouville operator, L, is defined as
f
∂ ∂
iL = {· · · , H} = ∑ ẋ j + ṗ j (1)
j=1 ∂x j ∂p j
where Cartesian coordinates are used with (x j , p j ) ≡ Γ the position and con-
jugate momenta of the system, ṗ j gives the force along the jth direction, and
{· · ·} represents the Poisson bracket of the system. L is a linear Hermitian oper-
ator of square integrable function on the phase space of Γ. The time propagation
operator as a function of L is defined by
which is a unitary: U(−t) = U −1 (t). The position and conjugate momenta state
point of the system at a given time t is defined as Γ(t) = U(t)Γ(0), which allows
determining one time step propagation as the following:
where ∆t = t/P is the size of a time step. Here, t is the total evolution time and
P are the number of integration points.
By splitting the Liouville operator into n different terms, like the following:
n
iL = ∑ iLk
k=1
Advanced Methods Used in Molecular Dynamics ... 61
and use the Trotter factorization scheme (H.F. Trotter, 1959), then the propaga-
tor becomes
(" #
n−1
U(t) = ∑ Uk (∆t/2) Un (∆t) (2)
k=1
" #)P
n−1
× ∑ Un−k(∆t/2) + O(t 3 /P2 )
k=1
decreases the number of calls for forces calculations, which reduces, in turn, the
overall computational time.
The basic idea of r-RESPA implementation, as discussed else-
where (M.E. Tuckerman et al., 1992; Minary et al., 2004; Tuckerman and Mar-
tyna, 2000), is on determining a reference system force Fs (r) for short range
interactions. Then, Eq. (1) can be written in the following form:
f
∂ ∂ ∂
iL = ∑ ẋ j + Fs (x j ) + Fl (x j ) (4)
j=1 ∂x j ∂p j ∂p j
f
∂
= iLs + ∑ Fl (x j )
j=1 ∂p j
where the operator exp (iLs ∆t) propagates the state vector using the short range
forces with a shorter time step δt (see Eq. 3). Here, this operator is factorized
using the Trotter formula (M.E. Tuckerman et al., 1992):
"
f
δt ∂
exp (iLs ∆t) = ∏ exp Fs (x j ) (6)
j=1 2 ∂p j
f
∂
× ∏ exp δtFs (x j )ẋ j
j=1 ∂x j
f #NMT S
δt ∂
× ∏ exp Fs (x j )
j=1 2 ∂p j
applied to an initial state (r(0), p(0)), it gives a solution for both position and
velocity similar to Verlet numerical integrator (M.E. Tuckerman et al., 1992).
Following the discussions in Refs. (M.E. Tuckerman et al., 1992;
M. Tuckerman and M. Parrinello, 1994; W.B. Street et al., 1978), for a Lennard-
Jones type of fluid, exists only the translational relaxation time characteristic.
In that case, the integration time step can easily be chosen. On the other hand,
for biomolecules, indeed there exists more than one time-scale. For example, in
addition to the translational and rotational relaxation times, there exists the time
characterizing intra-molecular motion, such as bond stretching, angle bending,
and dihedral angle motion. Furthermore, the inter-molecular motion, including
van der Waals and electrostatic interactions, is of the typical timescale of one or
more orders in magnitude larger than intra-molecular motion. In such cases, the
system is characterized by stiff nonlinear differential equations, which require
the use of a small enough time step to observe fast motion, if treated using one
time-scale.
Other systems that are characterized by more than one timescales are those
consisting of high-frequency oscillators interacting with a bath of slow mo-
tion (Tuckerman et al., 1990), and the systems consisting of large mass particles
(slow degrees of freedom) interacting with lighter ones (fast particles) (Tucker-
man et al., 1991).
The method is also used to treat systems coupled to a Nosé heat bath (Nosé,
1984; S. Nosé, 1984a; W.G. Hoover, 1985) used to keep temperature and/or
pressure fixed during MD simulations. Here, heat bath includes extra fast
degrees of freedom into the system, treated using multiple time stepping al-
gorithms (M.E. Tuckerman et al., 1992), typically, two or more time steps.
The method has been used by many molecular simulation software codes in
performing simulations of complex systems, for example, in CHARMM pro-
gram (B. R. Brooks et al., 2009).
However, the approach is limited by the so-called resonance phenomena,
which restricts the use of time steps higher than ∆t < 8 fs by r-RESPA in
MD simulations of biomolecular systems (Bishop et al., 1997; Ma et al., 2003;
Schlick et al., 1998). It must be noted that not just time-reversible integra-
tors, but also multiple time step symplectic integrators (Skeel et al., 1997)
show numerical instability limiting the use of large time steps (Wolfram, 2002).
According to Ref. (Schlick et al., 1998), the resonance phenomena is the re-
64 Hiqmet Kamberaj
which ensures that maximum total kinetic energy accumulated in each degree
of freedom is LkB T . In Eq. 7, M is the Nosé-Hoover chain length of ther-
(i)
mostats, ξk, j (k = 1, · · · , M and j = 1, · · · , L) are the thermostat velocities asso-
ciated with Lagrangian multiplier along the i degrees of freedom and η(i) is the
corresponding thermostat coordinate, which is used to control the accumulated
kinetic energy fluctuations. Fi is Newton’s force on the i degrees of freedom.
(i)
The thermostat forces Gk, j are defined as
(i) (i) (i)
Gk, j = Qk−1(ξk−1, j )2 − kB T, (k = 2, 3, · · · , M) (11)
p2i
Γi,1 = − kB T (13)
mi
2
(s)
Γi,k = Wi,k−1 Pi,k−1 − kB T, k = 2, · · · , M
Wi,k are thermostat masses determined by Eq. 12, as in Refs. (G. J. Martyna
et al., 1992; S. Nosé, 1984b).
Eqs. 7 can be numerically solved using the Liouville operator formalism and
Trotter factorization schemes as suggested elsewhere (Minary et al., 2004) (and
the references therein). The classical Liouville operator can be expressed as:
f
∂ ∂
iL = ∑ v̇d + q̇d (14)
d=1 ∂v d ∂q d
M L
(d) ∂ ∂
+ ∑ ∑ ξ̇k, j (d)
+ η̇(d)
k=1 j=1 ∂ξk, j ∂η(d)
M
(s) ∂ ∂
+ ∑ Ṗd,k (s)
+ s˙d,k
∂sd,k
k=1 ∂Pd,k
where
M
(d) ∂ ∂ ∂
iL1 = vd + η̇(d) (d) + ∑ ṡd,k (15)
∂qd ∂η k=1 ∂sd,k
Advanced Methods Used in Molecular Dynamics ... 67
(n)
!
(d) Fd (n) (n) (s) (n) ∂
iL2,n = (n)
− λd vd − Pd,1 vd (n)
md ∂vd
(n)
Γd,1 ∂ L
(n) (d) ∂
+ − ∑ λd ξ1, j n = 1, 2, · · · , Nd
Wd,1 ∂P(s) j=1 (d)
∂ξ1, j
d,1
M L G(d) M−1 L
(d) k, j ∂ (d) (d) ∂
iLNHC = ∑∑ (d) (d)
− ∑ ∑ ξk, j ξk+1, j (d)
k=2 j=1 Qk, j ∂ξk, j k=1 j=1 ∂ξk, j
L
(d) (d) ∂
− ∑ λb ξ1, j (d)
j=1 ∂ξ1, j
M M−1
Γd,k ∂ (s) (s) ∂
+∑ (s)
− ∑ Pd,k Pd,k+1 (s)
k=2 Wd,k ∂P k=1
d,k ∂P k
where Nd is the number of parts that the force on every degree of freedom can
be split, that is
Nd
(n)
Fd = ∑ Fd
n=1
Using the Trotter factorization scheme for classical Liouville operator, as sug-
gested in Ref. (Minary et al., 2004), then the approximation of true evolution
can be written as:
(t) h is1 −2
(t)
iL̃N δt
Γ(∆t) ≈ e d · · · eiL̃2 δt eiL̃1 δt (17)
68 Hiqmet Kamberaj
s −2
iL̃2 δt 2 iL̃Nd δt
×e · · ·e Γ(0)
where
(d) δt
eiL̃k δt = eiLNHC 2 (18)
(d) (d) (d)
iL2,1 δt2 iL1 δt ∑kn=1 iL2,n wn δt2
×e e e
(d)
iLNHC δt2
×e
time steps. Efforts should be made to also include the solvent as a part of the
system and check the efficiency of the method in the simulation of large macro-
molecular systems in the solvent. Future work should also focus on the compar-
ison of the efficiency of sampling conformation equilibrium space of such com-
plex systems using other methods discussed below or combining this method
with other enhanced sampling techniques.
exp (−β∆E)
with non-Boltzmann weights, which allows the system escaping from the local
minimum states. Here, ∆E represents the energy barrier height and β is the
inverse temperature of the simulation, β = 1/kB T .
In the canonical ensemble (characterized by fixed N, V , and T ), each state
point, (r, p), in the phase space is associated with a Boltzmann weight, which is
defined in terms of the Hamiltonian function H(r, p):
P(E, β) ∝ Ω(E)WB(E, β)
Emax = hEiT0
Then, for E ≤ Emax , a flat energy distribution is achieved, and for E > Emax, we
obtain the canonical ensemble distribution at T0 . At every MD time step, t, the
probability distribution weighting factor is given by:
W (t) (E, β0 ) = exp −β0 E (t) (E, β0 )
(t)
Then, a histogram N (t) (E) is accumulated for distribution Pmu (E) of potential
(t)
energy. Denoting by Emin the minimum energy value obtained until the t time
step. For the (t + 1) time step, the multicanonical potential energy is obtained
Advanced Methods Used in Molecular Dynamics ... 73
as
E, E ≥ Emax
1 (t)
(t) (t)
E mu (E, β 0 ) + ln N (E) −c ,
β0
(t+1) (t)
Emu (E, β0 ) = Emin ≤ E < Emax (28)
(t)
β(t+1) (Emin ) (t)
E − Emin
β0
(t+1) (t) (t)
+Emu (Emin , β0 ), E < Emin
where c(t) are used to ensure the continuity of energy function at E = Emax ,
determined as
1 (t)
c(t) = ln N (Emax)
β0
The MD simulation continues until a reasonably flat potential energy func-
tion is obtained, which is determined by comparing the values of energy for
all E < Emax and requiring to be of the same order of magnitude. After this
(t)
convergence is reached, Emin should be equal to the global minimum potential
energy function value. Note that during MD simulation, a polynomial or some-
times a cubic spline function (Y. Sugita and Y. Okamoto, 2000) is used to fit the
histograms each MD simulation time step (Nakajima et al., 1997).
Long MD simulation in a multicanonical ensemble is performed, after the
optimal weighting factor is obtained. Then, the ensemble average of any phys-
ical quantity, A , is determined using the Weighted Histogram Analysis Method
(WHAM) (Gallicchio et al., 2005), which is described in details, for the general
case, in Section 3.10..
Ω(E) = 1, ∀E
or in algorithmic scale
If the move is rejected, then E remains unchanged, and we modify the current
Ω(E) with the same factor γ. In the first publication (Wang and Landau, 2001),
the suggested initial value of factor γ is
γ = γ0 = e1 = 2.71828
which allows faster convergence of Ω(E) to the true density of states even for a
very large system. On the other hand, as discussed in Ref. (Wang and Landau,
2001), if γ0 is too small, then the convergence is extremely slow. However,
values of γ0 being too large will produce high statistical errors.
During the simulation a histogram H(E) is accumulated, representing
counts for every visited energy bin, E, by the system. After histogram becomes
flat in the sampled energy range, we say Ω(E) has converged to the true value
Advanced Methods Used in Molecular Dynamics ... 75
with an accuracy proportional to the factor ln(γ). Then, the modification factor
is decreased according to (Wang and Landau, 2001)
√
γ1 = γ0
At this moment, the histogram is reset, and the random walk sampling restarts.
Now, the density of states is multiplied by a smaller value of factor γ1 at each
step. The algorithm continues in this way, and each time that the histogram
becomes flat, the modification factor is decreased as (Wang and Landau, 2001)
√
γi+1 = γi
It can be seen that the accuracy is controlled by estimating the density of states
and the length of a simulation by a factor γ. In addition to γfinal , the accuracy of
estimating Ω(E) depends on the complexity and size of the system, criterion of
the flat histogram, and algorithm’s implementation (Wang and Landau, 2001).
In MD simulations WLM is implemented by modifying equations of motion
for the multicanonical method, as in Refs. (Hansmann et al., 1996; Nakajima
et al., 1997). Here, we purpose to modify equations of motion, given by Eqs. 27,
as the following
pi
q̇i = (33)
mi
β (s)
ṗi = Fi − λi pi − P1 pi
β0
(i) (i) 2
L Q1 (ξ1, j ) (i) M
(i)
η̇(i) = − ∑ ξ2, j − ∑ ξk, j
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L (34)
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
76 Hiqmet Kamberaj
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where
d lnΩ(E)
β0 = (35)
dE
with β and β0 being, respectively, the inverse simulation and multicanonical
temperatures. Note that the Wang-Landau application in MD simulation con-
sists in calculation of density of states Ω(E) from Eq. 31 or Eq. 32, then using
Eq. 35 to run MD simulation (T. Nagasima et al., 2007). In order to calculate
accurately Ω(E) and hence β0 , a histogram bin of energy distribution is esti-
mated, and the bin width will determine the accuracy of calculation of Ω(E)
and β0 since it defines the ruggedness of energy distribution (T. Nagasima et al.,
2007). To smooth the ruggedness, the energy distribution is approximated by
a Gaussian distribution, and then, WHAM can be used to estimate Ω(E) and
multicanonical inverse temperature.
The method has found application to Ising spin lattice systems (Landau
et al., 2004; Wang and Landau, 2001). It has been used to study the confor-
mation transitions of proteins using confined lattice models (Pattanasiri et al.,
2012), protein folding (Wüst and Landau, 2012), and optimizing temperature
distribution in replica exchange method (H. Kamberaj and A. van der Vaart,
2009).
q is an adjustable parameter taking real values and E0 is the system’s ground en-
ergy. Note that WT (E, β) > 0. Besides, for q → 1, the Boltzmann’s weight can
be obtained, and for q > 1, probability distribution has longer tails. The long
tails of the Tsallis distribution have inspired construction of generalized distri-
butions which will enhance the excursion towards regions with higher energy by
decreasing the magnitude of the force close to these regions. This increases the
rate of barrier crossing and hence allows the system escaping the local minimum
energy states (A. Karolak and A. van der Vaart, 2012; Andricioaei and Straub,
1997; H. Kamberaj and A. van der Vaart, 2007; J. Kim and J. E. Straub, 2009;
Sugita and Okamoto, 1999).
The aim of Tsallis statistical ensemble is to weight each state by a weighting
factor, WT (E, β) (Sugita and Okamoto, 1999):
∂Ueff (E, β)
F̃i = −
∂qi
∂Ueff (E, β)
= Fi
∂E
78 Hiqmet Kamberaj
1
= Fi
1 + β(q − 1)(E − E0 )
Fi is the Newton force on particle i (i = 1, 2, · · · , N). Then, the equations of mo-
tion describing a generalized canonical ensemble according to Tsallis statistics
can be given as the following:
pi
q̇i = (38)
mi
1 (s)
ṗi = Fi − λi pi − P1 pi
1 + β(q − 1)(E − E0 )
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑ ξ2, j − ∑ ξk, j
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
TSMD has successfully been employed to different molecular systems, such
as simulation of atomic clusters (Andricioaei and Straub, 1996, 1997),
protein folding (I. Fukuda and H. Nakamura, 2002; S. Jang et al., 2008;
U. H. E. Hansmann and Y. Okamoto, 1997; Y. Pak and S. Wang, 1999), and
molecular docking (Y. Pak and S. Wang, 2000). The approach has also been
Advanced Methods Used in Molecular Dynamics ... 79
ṗLbest
i = −γ1 u1 (qLbest
i − qi )
Gbest
ṗi = −γ2 u2 (qGbest
i − qi )
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑ ξ2, j − ∑ ξk, j
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
80 Hiqmet Kamberaj
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where the vector c = (c1 , c2 , · · · , cm )T characterizes the essential degrees of free-
dom in the system. Projection operator, P, transforms the real coordinates q to
the so-called collective coordinates c according to:
f
c j = ∑ Pi j qi
i=1
which are updated every time step. Here, qLbest is configuration vector with the
lowest value of the potential energy of the system and qGbest is configuration
vector of the final state of the system.
In Eq. 39 ui (i = 1, 2) denotes a uniformly distributed random number in
(0, 1), and γ1 and γ2 are adjustable parameters.
In Eq. 39, the δ function is given as:
Lbest
1, if U(q) < U qLbest
δ U(q) < U q =
0, otherwise
Advanced Methods Used in Molecular Dynamics ... 81
and
Gbest
1, if U(q) < U qGbest
δ U(q) < U q =
0, otherwise
The augmented dynamical system, which is given by Eq. (39) , sample an equi-
librium canonical distribution with conserved total energy given by:
f
p2i
Eext = ∑ (41)
i=1 2mi
f
2 !
pLbest
i
+ ∑ δ U(q) < U qLbest
i=1 2mi
f
2 !
pGbest
i
+ ∑ δ U(q) < U qGbest
i=1 2mi
| {z }
Etot,kin
h f 2 i
1 Lbest
2 Gbest
+U(q) +
2 ∑ 11 j
u γ q − q j + u γ
2 2 q j − q j
j=1
| {z }
Ubias
!
f M Qi,k ξ2i,k
+∑ ∑ + kB T si,k
i=1 k=1 2
| {z }
Ethermo
(si,k , ξi,k ), i = 1, 2, · · · , f ; k = 1, 2, · · · , M
In Eq. 41, Etot,kin is the total kinetic energy of augmented system, Ubias is the
total potential energy including bias term, and Ethermo is the thermostat energy.
82 Hiqmet Kamberaj
WHAM is used to recover the equilibrium canonical distribution of the real sys-
tem (H. Kamberaj, 2015, 2018). The augmented dynamical system is shown
to sample metastable (H. Kamberaj, 2018) and rare transition events (Hummer
and Szabo, 2010), and to enhance the conformation sampling (Andricioaei and
Straub, 1997; J. Kim and J. E. Straub, 2009). In Eq. (39), the first bias term
steers the system towards the state with the lowest energy, which has been vis-
ited at any instant time t and hence enhancing the local basin sampling. Besides,
the second bias term indicates the “information" about configuration with the
lowest energy ever visited, and hence enhancing the barrier crossing rate.
et al., 2005; R. Zhou, 2003; R. Zhou and B.J. Berne, 2002) that reductions in
system size may not accurately describe the structure and dynamics of the sys-
tem.
Other approaches, similar to REM, are also proposed. For instance, the
Hamiltonian Replica Exchange Method (HREM) (H. Fukunishi et al., 2002).
An HREM with biasing the backbone dihedral potentials yielded a reduction
in the number of replicas (K. Srinivasaraghavan and M. Zacharias, 2007). In
resolution HREM approach, which uses implicit solvent models only (E. Lyman
et al., 2006; P. Liu and G.A. Voth, 2007), in addition to different temperature
couplings, the replicas exchange a subset of configuration coordinates from a
coarse-grained model (E. Lyman et al., 2006).
Use of temperature scaling for the solvent-solvent and solvent-protein inter-
actions in REM has also shown to reduce the number of replicas (P. Liu et al.,
2005), which has further been improved by using the Tsallis biasing poten-
tial (H. Kamberaj and A. van der Vaart, 2007).
Efforts have also been made to optimize the distribution of tempera-
tures among the replicas as in Refs. (A. Kone and D.A. Kofke, 2005; Berg,
2004; C. Predescu et al., 2004; D.A. Kofke, 2002; Escobedo and Martinez-
Veracoechea, 2007; Gront and Kolinski, 2007; H.G. Katzgraber et al., 2006;
Li et al., 2007; Nadler and Hansmann, 2007; N. Rathore et al., 2005; Pre-
descu et al., 2005; Sabo et al., 2008; S. Trebst et al., 2004, 2006; Sugita and
Okamoto, 1999). Some of these methods (Escobedo and Martinez-Veracoechea,
2007; H.G. Katzgraber et al., 2006; Nadler and Hansmann, 2007; S. Trebst
et al., 2004, 2006) have particularly been important in studying the protein
folding/unfolding transitions, which represent a difficult case study in standard
REM because of the low rate of accepted swaps between replicas across the
transition temperature (Huang et al., 2007; S. Trebst et al., 2006).
To further increase the efficiency of REM, other approaches to REM have
also been proposed (H. Kamberaj and A. van der Vaart, 2009), which aims to
obtain a flat generalized probability distribution function in temperature space
using the Wang-Landau algorithm (F. Wang and D.P. Landau, 2001a,b). The
method addresses two problems of REM: it increases the probability of swap-
ping, and it decreases the bottleneck for exchange at the transition temperature.
Note that a WHAM is used for analyzing the data from all replicas (see
Section 3.10.).
Advanced Methods Used in Molecular Dynamics ... 85
where all the variables have the same meaning as in Eq. 39 for the replica α
and {cGbest
j }mj=1 is related to the global best coordinates qGbest corresponding to
configuration with the lowest energy among all replicas through the projection
operator P:
f
cGbest
j = ∑ Pi j qGbest
i
i=1
It has been shown elsewhere (H. Kamberaj, 2018) that the Eqs. 44 preserve
the detailed balance condition. Following Ref. (H. Kamberaj, 2018), assum-
ing that a Markovian chain of states is formed, the probability of obtaining a
trajectory in the configuration space of the replica k can be written as:
T −1
Pk (XkT ) = exp (−βk E(xk,0 )) ∏ π(xk,t → xk,t+1 ) (46)
t=0
with βk being the inverse temperature of the thermostat k. In Eq. 46 E(xk,t ) is the
total energy obtained for the configuration xk,t . Here, XkT represent T replicas
of the system:
The initial configurations of each replica are obtained from a canonically dis-
tributed with an initial unbiased energy of the system for replica k E(xk,0 ):
In Eq. 46, π(xk,t → xk,t+1 ) is the propagation probability at each time step,
which depend on the details of deterministic or stochastic dynamics. In general,
the Markovian transition probability π(xk,t → xk,t+1 ) can have any distribution
that conserves the Boltzmann distribution. Here, π(xk,t → xk,t+1 ) represents the
action characterized by augmented system given in Eq. (44), which produces a
Boltzmann distribution in the extended phase space of variables. In the general
case of the Newtonian dynamics, we can write:
p(xt → xt+1 ) = δ(xt+1 − Φ∆t (xt ))
where δ is the delta function and Φ∆t (xt ) is the discrete flow map of one time
step ∆t propagation operator. In this case, a trajectory can be generated using
an initial state sampled from some canonical distribution and then propagating
in time using usual Hamiltonian dynamics. Note that for Hamiltonian dynamics
is easy to find a time-reversible discrete flow map. On the other hand, when
dynamics are governed by Eq. (44), the structure is not symplectic, but still, it
is time reversible.
It is important to note that WHAM can be used for analyzing the data
from all replicas in the case of REM:SPMD simulation as presented in
Refs. (H. Kamberaj, 2015, 2018) (see Section 3.10.).
used in the replica exchange simulation. While outside this interval the potential
energy of multicanonical simulation is determined through extrapolation:
∂Emu (E, β0 )
(E − E1 )
∂E E1
+Emu (E1 , β0 ), E < E1
(0)
Emu (E) =
mu (E, β0 ), E
E 1 ≤ E ≤ EL (47)
∂E mu (E, β 0 )
(E − EL )
∂E EL
+Emu (EL , β0 ), E > EL
In multicanonical MD simulations, Eq. 25 (or Eq. 27) is used to govern New-
(0)
ton’s dynamics where Emu (E, β0 ) is replaced by Emu (E). Then, after the sim-
ulation run, the trajectories are analyzed using WHAM for a single run, as de-
scribed in the next section.
Eq. 47 can also be written as
T0 T0
(E − E1 ) + T0 S(E1 ) = E + constant,
T1 T1
E < E1 = hEiT1
(0)
Emu (E) = T0 S(E), E1 ≤ E ≤ EL (48)
T0 T0
(E − EL ) + T0 S(EL ) = E + constant,
TL TL
E > EL = hEiTL
and the dynamical equations of motion are defined here as:
pi,α
q̇i,α = (49)
mi
β1
Fi,α , E < E1 = hEiT1
β
β0 (E)
(α,s) α
ṗi,α = −λi,α pi,α − P1 pi,α + Fi,α , E1 ≤ E ≤ EL
β0
β
L Fi,α ,
E > EL = hEiTL
β0
(i,α) (i,α)
(i,α)
L Q1 (ξ1, j )2 (i,α) M (i,α)
η̇ =−∑ ξ2, j − ∑ ξk, j
j=1 kB Tα k=2
Advanced Methods Used in Molecular Dynamics ... 89
(i,α) (i,α) (i,α) (i,α) (i,α) (i,α)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi,α ξ1, j
j = 1, · · · , L
(i,α)
(i,α) Gk, j (i,α) (i,α)
ξ̇k, j = (i,α)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i,α)
(i,α) GM, j
ξ̇M, j = (i,α)
QM, j
j = 1, · · · , L
(α) (α,s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(α)
(α,s) Γi,k (α,s) (α,s)
Ṗi,k = (α)
− Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(α)
(α,s) Γi,M
Ṗi,M = (α)
Wi,M
From Eq. 23, the Boltzmann’s factor depends on temperature T and energy
E, and hence, scaling potential energy (and so the force) by a constant κ is
similar to scaling the temperature by 1/κ (Hansmann et al., 1996; R. Yamamoto
(0)
and W. Kob, 2000). Therefore, Emu given by Eq. 47 (or Eq. 48) generates a
canonical ensemble distribution at T = T1 , multicanonical ensemble distribution
for E1 ≤ E ≤ EL , and canonical ensemble distribution simulation at T = TL for
E > EL .
are needed for the final simulation run to guarantee an optimal distribution of
replicas among thermostats.
In MUCAREM, similar to REMUCA, short replica exchange MD simula-
tions are performed with L replicas and L thermostats, covering a temperature
range from T1 to TL . During this short simulation run, we can estimate the en-
ergy density of states Ω(E) for all range of energy using WHAM techniques.
After we define the density of states Ω(E), we can chose L pairs of thermostats
(m) (m) (m) (m)
with temperatures (TL , TH ), for m = 1, 2, · · · , L , where TL < TH . In prac-
tice, the temperatures are arranged such that ensure sufficient overlapping be-
(1) (L )
tween neighboring pairs. Here, we have TL = T1 and TH = TL , and
E (m) = hEi (m) ,
L TL
(50)
EH(m) = hEi (m) , m = 1, 2, · · · , L
TH
where
h i
(m+1) (m+1)
∆ = βm+1 Emu (E(q(i))) − Emu (E(q( j))) (53)
h i
(m) (m)
− βm Emu (E(q(i))) − Emu (E(q( j)))
Here, the multicanonical potential energies, E (m) (E(q( j))) and E (m+1)(E(q(i))),
have to be calculated since E (m) (E) has different values for m (Y. Sugita et al.,
2000).
Using the multiple-histogram reweighting method, the canonical distribu-
tion can be obtained (Ferrenberg and Swendsen, 1989; Kumar et al., 1992), as
presented in Section 3.10..
(0)
where U` is a reference minimum value of the potential energy U(q) at Tsal-
lis entropy parameter q` of replica ` associated with thermostat held at inverse
temperature β` . Note that in the limit when q` → 1, the Boltzmann’s weight of
standard MD simulation can be obtained:
WB(q) ∼ exp −β` (U −U (0) )
where
where Ueff,1 and Ueff,2 are the effective Tsallis potential energy of replica 1 and
2, respectively.
Close to the barrier regions, the magnitude of the force, because of low-
ering the barriers, is reduced for q larger than one (Andricioaei and Straub,
1997; Plastino and Anteneodo, 1997). Therefore, in these regions resistance on
the particles decreases and the barrier crossing rates increases. However, the
largest value of q has to be carefully determined after some preliminary test
runs depending on the system. It is empirically suggested (H. Kamberaj and
A. van der Vaart, 2007; U. H. E. Hansmann and Y. Okamoto, 1999) an upper
value as q = 1 + 1/ f , where f is the number of degrees of freedom.
A more general approach for optimization of q values has been suggested in
Ref. (van Giessen and Straub, 2005). Based on this approach, expression of ∆12
in Eq. 56 is written as
Zq1
1 1 1
∆12 = − dz (57)
kB T2(z) T1(z)
q2
where
−1
∂S
TS (U0 ) =
∂U U0
In Eq. (61) σq denotes the width of Tsallis probability density function at Gaus-
sian approximation given by (van Giessen and Straub, 2005):
0
TS (U) Ti 0 (U) −1
σq (U0) = − (62)
TS2 (U) Ti 2 (U) U0
where
∂TS ∂Ti
TS0 (U) = , T 0 (U) = .
∂U i ∂U
If TS (U) is assumed to be linear function of U around U0 , then TS0 (U) is a con-
stant. Furthermore, the equivalence between the microcanonical and canonical
ensembles indicates that this constant is 1/CV (T0 ), thus
then σq > σ0 (Eq. 63), and hence Tsallis probability density function, Pi (U),
has a broader distribution than canonical function at T0 . Whereas, for qi <
1, Pi (U) becomes narrower compare to canonical probability density function.
However, in both cases, Tsallis distribution has its maximum at stable point U0
as canonical distribution at T0 . For qi = 1 + qc , the Tsallis effective temperature,
T (U), it is tangential to canonical temperature TS (U) function at U0, and Pi (U)
is locally flat around U0 , indicating only marginal stability. Thus, the choice qi =
1 + qc generates the most delocalized Tsallis distribution for standard Tsallis
MD simulation run. For qi > qc, the local minimum of Pi (U), namely U0 , is an
unstable crossing point (van Giessen and Straub, 2005).
The WHAM is used to estimate the averages of unbiased system quanti-
ties at required temperature T0 (H. Kamberaj and A. van der Vaart, 2007). The
configuration probability density for each replica k (k = 1, 2, · · · , K) at inverse
temperature β` (` = 1, 2, · · · , L) is written as
1
P(U; qk, β` ) = q (64)
Zk`k
h i qk
(0) 1−qk
× Ω(U) 1 + β` (qk − 1)(U −U )
K L
nk` bias
Pm = Ωm ∑ ∑ Nk e−β (U ` m −F` ) (70)
k=1 `=1
Advanced Methods Used in Molecular Dynamics ... 97
where Nk is the total number of saved snapshots from the replica k. Pm can also
be approximated as (Chodera et al., 2007):
K
Hkm
Pm ≈ ∑ (71)
k=1 Nk
Then, the estimated value of the density of states Ω̂m is given as:
K
eff
∑ Hkm
k=1
Ω̂m = K L
(73)
bias −F )
∑ ∑ neff
k` e
−β` (Um `
k=1 `=1
M
1 bias
F` = − ln ∑ Ω̂m e−β`Um
β` m=1
From Eq. 73, Ω̂m depends on F` , and F` also depends on Ω̂m . Therefore,
F` and Ω̂m are usually determined iteratively from Eqs. 73, starting from some
98 Hiqmet Kamberaj
Ω̂m
σ2Ω̂ = K L
(74)
m
bias −F )
∑ ∑ neff
k` e
−β` (Um `
k=1 `=1
The estimated average value of any physical quantity A of the system at the
target inverse temperature β0 is computed by summing the weighted values from
all configurations:
K Nk
∑ ∑ Wkn(β0 )Akn
k=1 n=1
Â(β0 ) = (75)
K Nk
∑ ∑ Wkn(β0 )
k=1 n=1
The chain rule of error propagation is used to obtain the statistical error of
Â(β0 ) (Chodera et al., 2007):
2 2
2 hXi σX σY2 σ2XY
σ = + −2 (76)
hY i (hXi)2 (hY i)2 hXihY i
where
1 Nk
hXi = ∑ Wn(β0)An
Nk n=1
(77)
1 Nk
hY i = ∑ Wn(β0)
Nk n=1
(78)
Nk
gX
σ2X = ∑ (Wn(β0)An − hXi)2
Nk (Nk − 1) n=1
(79)
Nk
gY
σY2 = ∑ (Wn(β0) − hY i)2
Nk (Nk − 1) n=1
(80)
Advanced Methods Used in Molecular Dynamics ... 99
Nk
gXY
σ2XY = ∑ (Wn (β0 )An − hXi) (81)
Nk (Nk − 1) n=1
× (Wn (β0 ) − hY i)
4. Metadynamics Method
Metadynamics method has been developed by Alessandra Laio & Michele Par-
rinello (Laio and Parrinello, 2002b). The method consists on finding a limited
number of essential collective coordinates, ci (i = 1, 2, · · · , m), upon which the
free energy depends on F(c). At any moment of time t, the free energy sur-
face is explored based on the dynamical equations of motion determined by the
forces acting on the system:
∂F
fit = −
∂cti
In metadynamics method, a bias potential function Ubias(c) is constructed that
is added to the Hamiltonian function of system. This bias potential is written
as a sum of Gaussian distributions, which are added at any time t of the tra-
jectory in subspace expanded by collective coordinate as generated during the
MD simulation. The main effect of the bias potential is to not allow the system
visiting configurations that are already explored. The mathematical form of the
bias potential is (Barducci et al., 2011; Laio and Parrinello, 2002b)
!
Z t m 0 2
(ci (t) − c i (t ))
Ubias(c,t) = dt 0ω exp − ∑
0 i=1 2σ2i
where W and τ are the Gaussian height and the time interval of deposition,
respectively (Barducci et al., 2011). Here, σi is the Gaussian distribution width
of collective coordinate i. W , τ and σ are adjustable parameters to optimize the
algorithm (Laio and Parrinello, 2002b).
The main benefit of using metadynamics method is being able to escape
from the local minimum free energy metastable states, and hence it increases
the rate of sampling rare events. Besides, the metadynamics method allows
sampling of new reaction pathways after the system escapes local minimum
states (Barducci et al., 2011). In the metadynamics method, there is no need for
a priory knowledge of the exact topology of the free energy landscape.
After a certain long time, metadynamics technique will eventually give a
bias potential Ubias (Laio et al., 2005):
Ubias(c,t → ∞) = −F(c) +C
accumulation decreases with simulation time as 1/t and the deviations from the
equilibrium dynamics are small. Secondly, a convergence of the bias potential
is reached up to a constant value, C, though a complete compensation of the free
energy surface is not obtained:
∆T
Ubias(c,t → ∞) = − F(c) +C
T + ∆T
In long run simulation, the collective coordinates probability distribution be-
comes:
F(c)
P(c) ∝ exp −
kB (T + ∆T )
where for ∆T → 0, standard MD simulations are recovered, and for ∆T → ∞ we
have standard metadynamics. In contrast to standard metadynamics, in well-
tempered metadynamics, the bias potential converges to a finite value during
one run. For all other values of ∆T the extent of free energy surface exploration
is determined by adjusting ∆T (Barducci et al., 2011).
qref ref
i+1 = qi + wi
where wi is the window width. Alternatively, the experimental data can also be
used to determine optimized values of bias parameters (Mills and Andricioaei,
2008).
Ubias(q) = −F(q)
Adding this bias term to the potential yields an exactly flat energy surface, and
hence the resulting probability distribution is uniform along q. Usually, the
simulations start with an initial guess for Ubias(q), because F(q) is not known a
priory. Then, iteratively, Ubias(q) is improved until a complete uniform distribu-
tion is obtained in q space.
On the other hand, transition path sampling (Bolhuis et al., 2000; Dellago et al.,
1998), as an alternative method, does not require previous knowledge of tran-
sition states, but it relies on the calculation of isomorphic reversible work from
reactive flux correlation functions.
In this approach, L + 1 copies of the trajectories in phase space characterize
a path in space-time:
XL = {x0 → x1 → · · · → xL }
where β is the inverse simulation temperature and E(xt ) is the total energy at
configuration xt . Here, the initial configuration (at t = 0) is generated from a
canonical distribution (Dellago et al., 1998):
In Eq. (83), p(xt → xt+1 ) gives the transition probability for each time step,
which is determined based on the natural dynamics. Usually, any Markovian
transition probability p(xt → xt+1 ) is such that it should obey to the Boltzmann
distribution and is normalized (Dellago et al., 1998). Typically, two approaches
are proposed as a choice for p(xt → xt+1 ): the Markovian action and Langevin
action (Dellago et al., 1998). If the natural dynamics are governed by the New-
ton’s equations of motion for the Hamiltonian systems, then
where δ is the delta function and Φ∆t (xt ) is the discrete flow map of one-time
step propagation.
106 Hiqmet Kamberaj
Here, hA (x0 ) constraints the trajectory path to start in the region A (i.e., reactant)
and hB (xL ) constraints the trajectory path to stop in the region B (i.e., product).
Typically, L steps are used to take the system from states A to the state B, with
action defined by Eq. (84).
This approach has been used to probe the dynamics of folding pathways for
the C-terminal β-hairpin of protein G-B1 using MD simulation at room temper-
ature of protein in explicit solvent (Bolhuis, 2003).
It can be suggested for time propagation of the system to be governed by the
swarm particle-like dynamics given in the following form (i = 1, 2, · · · , f ):
pi
ẋi = (85)
mi
(s)
ṗi = Fi − λi pi − P1 pi
+ γ1 u1 (xLbest
i − xi ) + γ2 u2 (xGbest
i − xi )
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑ ξ2, j − ∑ ξk, j
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
Advanced Methods Used in Molecular Dynamics ... 107
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where xGbest represents the coordinates of final state B (product), i.e., xL . Then,
we can generate different trajectories starting in the region A (reactant) and bi-
ased towards the end in final region B. As advantage this method does not
postulate a priory knowledge of the transition state.
where α is used to adjust the depth of potential energy minimum and modu-
late local smoothness of the energy basins of Ubias. In this approach, the bias
term ∆U(r) raises the potential surfaces near the minimum states and leaves
unaffected surface points near the barriers.
Another form of the bias term ∆U(r) has also been proposed (de Oliveira
et al., 2008), such as
time scales of milliseconds for bovine pancreatic trypsin inhibitor protein em-
phasizing one of the method’s advantage for not needing prior knowledge of
free energy landscape or reaction coordinate.
c = (c1 , c2 , · · · , cm )T
where
x̄ = hxik
C = ET Λ−1 E (90)
where E and Λ are the matrix of eigenvectors and the diagonal matrix of eigen-
values, respectively. Projection of the Cartesian coordinates fluctuations along
the space spanned by eigenvectors E is
q = ET (x − x̄)
Advanced Methods Used in Molecular Dynamics ... 111
tal in designing the so-called flooding potential Vfl (c) chosen as a multivariate
Gaussian in order to fulfill the criteria discussed in Ref. (Grubmüller, 1995):
1 T
Vfl(c) = Efl exp − c Λfl c (93)
2
where Efl characterizes the strength of Vfl (c) and Λfl determines the shape of
Vfl (c) in conformation space, which is chosen to be
Λfl = Λc /γ2
112 Hiqmet Kamberaj
γ = (βEfl )1/2
robust concerning error compare to the potential energy function. Moreover, the
numerical integrator used in MD simulations should guarantee energy conser-
vation and stable trajectories in long time scale simulations (Gray et al., 1994;
Kamberaj, 2005). Large computer storage is also needed to save, analyze, and
better computer graphics to handle a large amount of data produced.
Coarse-grained models of macromolecular systems have probed problems
of biologically relevant size and timescale during simulations when combined
with computer power (Tozzini and McCammon, 2005). However, the coarse-
grained models of proteins remain demanding, because of the challenges in
building useful energy potential functions representing all the physics of in-
teractions (Kamberaj, 2011; Lange and Grubmüller, 2006; Stepanova, 2007).
The most tested coarse-grained model is the bead-spring model of polymers.
Solvent effects are added using Brownian dynamics (Ermak and McCammon,
1978) or Stokesian dynamics (Phung et al., 1996).
In this study, we attempted to give a big picture of the methods used to
enhance sampling in molecular dynamics simulations and thus being able to
simulate rare events for complex molecular systems. Also, we presented the ad-
vantages and limitations of each method. Our final message from this study
is that probing relevant time and size scales of (bio)physical and chemical
phenomena in macromolecular systems may need new statistical models of
data processing and computational theoretical models to allow studying them
efficiently (G. Ciccotti and E. Vanden-Eijnden, 2015; M. K. Transtrum et al.,
2015).
Acknowledgments
The author thanks International Balkan University for the support.
Conflict of Interest
The author declares that there is no conflict of interest regarding the publication
of this chapter.
114 Hiqmet Kamberaj
References
A. Amadei, A. B. M. Linssen, and H. J. C. Berendsen. Essential dynamics of
proteins. Proteins: Struct., Funct., Genet., 17:412–425, 1993.
A. Karolak and A. van der Vaart. Importance of local interactions for the sta-
bility of inhibitory helix 1 of Ets-1 in the apo state. Biophys. Chem., 165-166
(3):74–8, 2012.
B.A. Berg. Markov chain Monte Carlo simulations and their statistical analysis.
World Scientific, Singapore, 2004.
G. Bussi, A. Laio, and M. Parrinello. Equilibrium free energies from non equi-
librium metadynamics. Phys. Rev. Lett., 96:090601, 2006.
M. Falcioni and M. W. Deem. A biased Monte Carlo scheme for zeolite structure
solution. J. Chem. Phys., 110(3):1754, 1999.
F. Wang and D.P. Landau. Determining the density of states for classical sta-
tistical models: A random walk algorithm to produce a flat histogram. Phys.
Rev. E, 64:056101–16, 2001b.
G. Ciccotti and E. Vanden-Eijnden. The trees and the forest. Aims and objec-
tives of molecular dynamics simulations. Eur. Phys. J. Special Topics, 224:
2515–2518, 2015.
H. Kamberaj and A. van der Vaart. Multiple scaling replica exchange for the
conformational sampling of biomolecules in explicit water. J. Chem. Phys.,
127:234102–234109, 2007.
H. Kamberaj and A. van der Vaart. An optimised replica exchange method for
molecular dynamics simulations. J. Chem. Phys., 130:074904, 2009.
X. Huang, M. Hagen, B. Kim, R.A. Friesner, R. Zhou, and B.J. Berne. Replica
exchange with solute tempering: efficiency in large scale systems. J. Phys.
Chem. B, 111:5405–5410, 2007.
J. Kim and J. E. Straub. Optimal replica exchange method combined with Tsal-
lis weight sampling. J. Chem. Phys., 130:144114–11, 2009.
A. Laio and M. Parrinello. Escaping free-energy minima. Proc. Natl. Acad. Sci.
USA, 99:12562–12566, 2002a.
A. Laio and M. Parrinello. Escaping free-energy minima. Proc. Natl. Acad. Sci.
USA, 99(20):12562–12566, 2002b.
D. P. Landau, S.-H. Tsai, and M. Exler. A new approach to Monte Carlo simu-
lations in statistical physics: Wang-Landau sampling. Am. J. Phys., 72(10):
1294–1302, 2004.
X. Li, C.P. O’Brien, G. Collier, N.A. Vellore, F. Wang, and R.A. Latour. An im-
proved replica-exchange sampling method: Temperature intervals with global
energy reassignment. J. Chem. Phys., 127:164116, 2007.
M.E. Tuckerman, B.J. Berne, and G.J. Martyna. Reversible multiple time scale
molecular dynamics. J. Chem. Phys., 97(3):1990, 1992.
M.P. Allen and D.J. Tildesley. Computer simulation of liquids. Oxford Univer-
sity Press, 1989.
P. Liu and G.A. Voth. Smart resolution replica exchange: An efficient algorithm
for exploring complex energy landscapes. J. Chem. Phys., 126(4):045106–6,
2007.
P. Liu, B. Kim, R.A. Friesner, and B.J. Berne. Replica exchange with solute
tempering: A method for sampling biological systems in explicit water. Proc.
Natl. Acad. Sci. USA, 103(39):13749–13754, 2005.
R. Zhou. Free energy landscape of protein folding in water: explicit vs. implicit
solvent. Proteins: Struct., Funct., Bioinf., 53(2):148–161, 2003.
R. Zhou and B.J. Berne. Can a continuum solvent model reproduce the free
energy landscape of a β-hairpin folding in water? Proc. Natl. Acad. Sci.
USA, 99:12777–12782, 2002.
D. Sabo, M. Meuwly, D.L. Freeman, and J.D. Doll. A constant entropy increase
model for the selection of parallel tempering ensembles. J. Chem. Phys., 128:
174109, 2008.
Advanced Methods Used in Molecular Dynamics ... 129
V. Tozzini. Coarse-grained models for proteins. Curr. Opin. Struc. Biol., 15:
144, 2005.
V. Tozzini and J. McCammon. A coarse grained model for the dynamics of the
early stages of the binding mechanism of hiv-1 protease. Chem. Phys. Lett.,
413:123–128, 2005.
S. Wolfram. A new kind of science. Wolfram Media, Inc., Champaign, IL, 2002.
Chapter 4
ABSTRACT
This chapter presents an extended Freudenstein-Chebyshev
approximation method for sigmoidal function generation applied to four
and five precision points. The mathematical model for sigmoidal function
generation is presented and the structural error parameters are obtained.
The structural error between the generated function and desired function
is estimated. The least-square method is used for minimizing the
structural error. Comparison of the results of the least square method and
analytical method is carried out. The proposed methodology is
demonstrated for designing a four bar function generation planar
mechanism.
Corresponding Author Email: jaiswalankur13@gmail.com.
136 Ankur Jaiswal and H. P. Jawale
1. INTRODUCTION
1
F ( x)
1 e x
138 Ankur Jaiswal and H. P. Jawale
2. KINEMATIC SYNTHESIS
x 0 x n 1 x n 1 x 0 cos(2 j 1)
x( j ) .
2 2 2n (1)
L1 L
D1 D2 1
L4 , L2 ,
A
B
L3
L4 L2
ψ φ
L1
O4 O2
j i ( ) j
where:
1
L (1 s12 ) 2
R1 2
L1s1 ,
L2
R2
L1 ,
L2
R3
L4 s1 , (4)
L L L L
2 2 2 2
R4 1 2 3 4
2 L1 L4 s1 ,
140 Ankur Jaiswal and H. P. Jawale
1
(1 s12 ) 2
R5
s1
Equation (3) is the required design equation for synthesis of four bar
linkage up to four accuracy points.
In equation (3) and (4), there are five variables (R1, R2, R3, R4 and R5).
For four accuracy points, there are only four equations [29] and similarly
in five points case.
R1 R2 R5 (5)
R j Aj λB j
(6)
1
( A B ) ( A2 B1 ) 2 4 A1 B2 2
1 2
2 B2 (9)
The three link ratios, and two angles, as design variables in case of five
point synthesis. Let the crank and follower angles for five accuracy points
be φj and ψj (j = 1, 2, 3, 4, 5), then,
Study on Error in Sigmoidal Function Generation of 4R … 141
j i j j i ( ) j
and
where:
R1
s1 [1 1 s . 1 s
2
1
2
2
K 2 .s1 .s 2 ,
R2
s1 1 s 1 s
2
1
2
2
K 2 s1 s 2 ,
K 1 s1
R3
K 2 s 2 (11)
R4
K 1 . 1 s12
K 2 .s 2 (11)
K3
R5
K 2 s2 ,
R6
1 s 2
1
s2
Equation (10) is the required design equation for synthesis of four bar
linkage up to five accuracy points. Using compatibility condition for
linearization in equation (10) and (11) and applying gauss elimination
method is used to solve the non-linear equations. Derive the linear equation
(12) and (13) is to determine the design parameters of the mechanism.
142 Ankur Jaiswal and H. P. Jawale
R1 R4 R3 R2
R1 R3 R2 R4
A1 cos(1q 1q ) A2 sin(1q 1q ) A3 cos( )1q A4 sin( )1q
A5 cos 1q 0
(12)
L2
D1
L1
1
(1 s12 ) 2
D2
s1
L2
D3
L4 s1 (15)
L L L L
2 2 2 2
D4 1 2 3 4
2 L1 L4 s1
Study on Error in Sigmoidal Function Generation of 4R … 143
1 s
1
2 2
D5 1
L1 s1
D1 cos{ 1q 1q } D2 sin{ 1q 1q } D3 cos 1q
D4 sin 1q D5 D6 sin 1q cos 1q 0
(16)
where:
D1
s1 [1 1 s . 1 s
2
1
2
2
K 2 s1 s 2 ,
D2
s1 1 s12 1 s 22
K 2 s1 s 2 , (17)
D4
K1 . 1 s 2
1
K 2 .s 2
K3
D5
K 2 s2 ,
D6
1 s 2
1
s2
D1 D4 D2 D3
D6
D1 D3 D2 D4 (for five points) (19)
144 Ankur Jaiswal and H. P. Jawale
E Aq Bq
2 2
(20)
Aq 2 Bq 2
0 0
An , Bn (21)
The equations (22) & (23) and (24) & (25) determine the optimal
design parameters for four and five accuracy points respectively [29, 30].
The following Table 3.1 shows a comparative study of the four and
five point synthesis of sigmoidal function and their optimization using least
square technique. The results of synthesis for sigmoidal functions at four
and five accuracy points are summarized in Table 3.1. The structural errors
for sigmoidal functions are shown in Figure 2 and 3.
Study on Error in Sigmoidal Function Generation of 4R … 145
Figure 2. Plot of error for sigmoidal function generator with four precision points.
Table 3.1. Plot of error for sigmoidal function generator with four and
five accuracy points
Design Variables The results of four point The results of five point
synthesis by Sigmoidal synthesis by Sigmoidal
function [1/(1+e-x)] function [1/(1+e-x)]
Analytical Least Square Analytical Least Square
method Optimization method Optimization
Interval of ‘x’ -- 0x1 0x1 0x1 0x1
Range of φ , deg -- 90 90 90 90
Range of ψ , deg -- 60 60 60 60
Initial crank -- 15 9 73.1478 71.9893
angle, φi, deg
Link length L2/L1 3.0336 2.9389 -1.0389 -1.7078
ratio’s L3/L1 9.7791 6.3517 1.6085 1.6123
L4/L1 7.2282 8.8720 -1.7275 -1.0215
Follower angle, -- 51.2460 46.2122 61.6060 60.3716
ψj,deg
Maximum error -- 2.8614e-04 2.8033e-04 9.2931e-06 5.4204e-06
Maximum output -- 0.0391 0.0383 0.0013 7.4145e-04
error %
146 Ankur Jaiswal and H. P. Jawale
Figure 3. Plot of error for sigmoidal function generator with five precision points.
100
90
Precision Points
80
Input Angle ( )
70
60
50
40
30
20
10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Independent Variable (x)
90
80
Precision Points
70
Input Angle ( )
60
50
40
30
20
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Independent Variable (x)
CONCLUSION
In future work, the approach can be extended to cover five and six link
mechanisms and evolutionary optimization problems.
REFERENCES
Chapter 5
ABSTRACT
Corresponding Author Email:f20150428@hyderabd.bits-pilani.ac.in.
154 Mallu Ch. Reddy, Rudraraju Manish and Yendluri D. Rao
1. INTRODUCTION
Functional representation,
Graphical representation, and
Rotation graph
For example, if the EGT in Figure 2b is considered, the three FCs are
shown in Figure 3.
In a rotation graph of EGTs, the turning pair edges are removed and
each geared pair is labeled with the symbol of the associated transfer
vertex of the fundamental circuit (Tsai L. 1987). For example, for the
graphs of EGTs in Figure 2(a) and 2(b), rotation graphs are as shown in
Figures 4(a) and 4(b).
The fundamental rules followed (Buchsbaum, F. and Fruedenstein, F.
1970), (Ravisankar, R. and Mruthyunjaya, T. S.) while drawing a graph are
that, for n-links, epicyclic gear train having one degree of freedom, there
are n-vertices, n-1 turning pairs and n-2 gear pair edges. The subgraph
formed by deleting all the geared edges is a tree and a geared edge added to
the tree will form a fundamental circuit having one geared edge and several
turning edges.
The number of fundamental circuits will be equal to the geared edges
and a fundamental circuit containing only turning edges, is not permitted as
it violates the rotatability rule of an EGT. All vertices must at least be
connected to one turning edge. The level of a turning pair identifies its
location of its axis.
For example, for the graph of EGT shown in figure 2(b), the levels are
shown in Figure 3. For any fundamental circuit, the differential degree of
freedom should be the number of vertices minus two, with a minimum of
one.
Two turning pair edges at the same level must intersect at a common
vertex (Ravisankar R. and Mruthyunjaya, T. S.). All edges on one side of
the transfer vertex are at the same level and all edges on the other side are
at a different level.
Identifying Transfer Vertex from the Adjacency Matrix … 157
The adjacency matrix for the graph of EGT in figure 2 is given below.
Adjacency matrix of Figure 2(a):
158 Mallu Ch. Reddy, Rudraraju Manish and Yendluri D. Rao
0 1 1 1 2
1 0 2 2 0
1 2 0 0 1
1 2 0 0 0
2 0 1 0 0
0 1 1 2 1
1 0 2 1 0
1 2 0 0 0
2 1 0 0 2
1 0 0 2 0
0 10 9 8 6 33
10 0 7 6 8 31
9 7 0 1 7 24
8 6 1 0 6 21
6 8 7 6 0 27
0 9 8 8 4 29
9 0 5 9 5 28
8 5 0 8 4 25
8 9 8 0 8 33
4 5 4 8 0 21
From the hamming strings of the graphs, we can conclude that they are
not displacement isomers to one another.
After the structural isomers are eliminated, the remaining graphs are
tested for rotation analysis. In this, a computer code is run to identify
fundamental circuits and find out the transfer vertex of the graph. If a
160 Mallu Ch. Reddy, Rudraraju Manish and Yendluri D. Rao
graph has a single transfer vertex, those graphs are directly taken to be
rotationally non-isomorphic. If a graph has more than one transfer vertex
possible, those graphs need to be manually iterated to find the transfer
vertex. After finding out the transfer vertex, rotation graphs are drawn and
the corresponding rotation matrices are generated. These rotation graphs
are now matched with the earlier adjacency matrices. On finding a match,
the graph is said to be rotationally isomorphic and is eliminated. Since the
rotation matrices that do not find a match, rotation-hamming strings are
generated and the string is iterated for a match to be eliminated as a
rotational isomer. The graphs, which are unique, are said to be rotationally
non-isomorphic.
1.4. Example
0 9 7 7 1 24
9 0 6 10 8 33
7 6 8 0 6 27 136 33 31 27 24 21
7 10 8 0 6 31
1 8 6 6 0 21
Identifying Transfer Vertex from the Adjacency Matrix … 161
2. INTELLIGENT METHODS
The first graph of the given EGT is drawn and linkage adjacency
matrix is written using the rules mentioned above. Search for the gearing
pair in the adjacency matrix, in each FC. If a gear pair, denoted by
weightage 2, is in (i, j) position in the adjacency matrix in a FC, search in a
row i and row j for a set of turning pairs with weightage 1 having a
common vertex. The common vertex (link) is the Transfer Vertex (TV) for
that gear pair. In case there is no such common vertex (link), check for
other connecting pairs. If the number of turning pairs are more than two in
a FC, then the number of TVs are possible. In such cases where more than
one TV is possible, identify all the FCs in the EGT. Then assign levels to
each of the turning pairs without violating the fundamental rules of EGTs
(Ravisankar, R. and Mruthyunjaya, T. S.), (Tsai L. 1987). There by the TV
in that particular FC is identified.
2.1. Example
For example, in the adjacency matrix of Figure 2(a), one gear pair is at
position (1, 5). Then the search is for tuning pairs in rows 1 and 5. We find
that column 3 in each row has turning pair. That is, elements in the
adjacency matrix (1, 3) and (5, 3) are turning pairs (weightage 1).
Therefore, in the FC 1-5-3 as shown in Figure 5, it can be concluded that
vertex 3 is the TV for the Gear pair 1-5.
162 Mallu Ch. Reddy, Rudraraju Manish and Yendluri D. Rao
CONCLUSION
REFERENCES
Chapter 6
ABSTRACT
Corresponding Author Email: firozkhan2294@gmail.com.
166 Khan Mohammad Firoz, Garg Aman and H. D. Chalak
1. INTRODUCTION
2. MATERIAL PROPERTIES
between the CFRP sheet and beam-column joint in Model B is also defined
as tie connection. The interaction between reinforcement and concrete is
defined as an ideal, i.e., friction between the reinforcement and concrete is
infinite.
3. MODEL VALIDATION
4. CONVERGENCE STUDY
CONCLUSION
REFERENCES
Chapter 7
H. H. M. Darweesh
Department of Refractories, Ceramics and Building Materials,
National Research Centre, Cairo, Egypt
ABSTRACT
Corresponding Author Email: hassandarweesh777@gmail.com.
176 H. H. M. Darweesh
ABBREVIATIONS
1. NANOMATERIALS
1.1. Introduction
The efficiency of these gears depends on placing those gears teeth just
right in atomically precise positions. Also, the self assembling molecules
form micelles are very important. Micelles are aggregates of amphiphilic
molecules. Molecules with one end are soluble in water and the other end
that rejects it. These aggregates form spontaneously at a size that depends
on the concentration of the amphiphilic molecules in solution. The center
of the micelles acts as a chamber for chemical reactions (Darweesh, 1992;
Tseng and Nalwa, 2009; Ashby, 2005). Thus, it dictates the size of the
nanoparticles created (Figure 5).
Ultrafine particles are those particles that are having a grain size
distribution of nanometers (10-9 m). These are considered as nanostructured
materials (Nalwa, 2000; Drexler, 1991; Tseng and Nalwa, 2009), i.e., any
material comprised grains as clusters below 100 nm and/or layers of the
same dimensions had been seen as a nanostructure unit (Nalwa, 2000;
Poole and Owens, 2003; Ashby, 2005). The importance of those materials
is due to its too small size as: particles, grains or phases. Its high surface-
to-volume ratio, these materials shows unique mechanical, optical,
electrical and magnetic characteristics (Poole and Owens, 2003; Nalwa,
2000; Drexler, 1991; Ashby et al., 2009; Tseng and Nalwa, 2009). The
Nanomaterials, Ceramic Bulk and Bioceramics 181
1.3. Nanomaterials
Nanoscale TiO2 and nano scale SiO2 are being used as a filler in
several products as bioceramics as: bones, teeth or dental fillings (Nalwa,
2000; Poole and Owens, 2003; Sobolev and Ferrada-Gutiérrez, 2005;
Tseng and Nalwa, 2009; Ashby et al., 2009; Ashby et al., 2007; Callister,
2007).
Nanomaterials have very small size particles which are having 1-3
dimensions ≤ 100 nm or more. Nanomaterials can be nanoscales in one
dimension, e.g., surface films, two dimensions, e.g., fibers and three
dimensions, e.g., particles. They often exist in single, fused, aggregated
and/or agglomerated forms in spherical, tubular, and irregular shapes.
Conventional kinds of nanomaterials comprise nanotubes, dendrimers and
fullerenes. Nanomaterials have displayed different physical and chemical
characteristics from normal chemicals, i.e., nanosilver, carbon nanotube,
fullerene, photocatalyst, nanocarbon, nano-SiO2, nano-Fe2O3 and nano-
Al2O3 (Koch, 2002; Sobolev and Ferrada-Gutiérrez, 2005; Tseng and
Nalwa, 2009; Ashby, 2005; Ashby et al., 2009; Ashby et al., 2007;
Callister, 2007).
Figure 6. I: (0D: Spheres and clusters), II: (1D: Nanofibers, wires and rods), III: (2D:
Films, plates and network), V: (3D: Nanomaterials).
classified into:- Zero dimension (0D) as: clusters and/or cluster assemblies
and filaments; one dimension (1D) as: nanofibers, wires, multi-layers and
rods; two dimensions (2D) as: ultrafine-grained over-layers or buried
layers, films, plates, and networks and three dimensions (3D)
nanostructures as: nanomaterials, nanophase materials consisting of
equiaxed nanometer sized grains (Figure 6).
2. CERAMICS
The modern term ceramic was derived from an ancient Greek word
which referred to a burning or firing process. More recently, a ceramic
product was defined as an article made from clay with or without the
addition of either materials, shaped in the plastic state, dried and then fired
to give it the required strength and durability. Also, it may be defined as
one composed of inorganic, but non-metallic material (Jackson, 1969;
Kingery et al., 1975; Rayan and Radford, 1987). On this basis, ceramic
products are usually subdivided into four sections as follows:-
Al2O3, SiO2, and a flux of an alkaline nature e.g., Na, K, Li and Ca. The
fluxing agents as feldspar, talc, pyrophyllite, etc. could produce the liquid
or glassy phase during sintering while SiO2 favours the dimension control
of the product during firing (Nuttawat et al., 2013; Kim and Hwang, 2016;
Ke et al., 2016). The nanoceramics have unique properties due to its
smaller size particles and molecular structure. To construct these
nanostructures using a “three-step process” to develop these complex
structures, the nanoceramic Greer is currently used Al2O3, or aluminum
oxide, and its maximum compression is about 1 micron from a thickness of
50 nanometers. After its compression, it can revert to its original shape
without any damage on the structure and was coated with Al2O3, different
molecules could yield a different result (Sonjida, 2011; Vorrada, 2009;
Darweesh, 2015).
2.1. Synthesis
2.2. Sintering
2.3. History
hollow or perforated bricks, clay roofing tiles, draining and sewer pipes.
The used raw materials vary widely from easily worked highly plastic
clays to hard coal measure shales (Ke et al, 2016; Sonjida et al., 2011).
2.4.2. Refractories
Refractory products are those used in high temperature applications, in
kilns and furnaces of all types, in steel works and in other industries, in
glass works and other power stations. The raw materials used include fire
clays for lower temperature applications and oxides such as alumina
(Al2O3) and magnesia (MgO) for the higher temperature ones. The
manufacturing methods include extrusion and moulding, whereas dry
pressing techniques account for the great majority of the non-clay
refractories (Njindam et al., 2018; Sonjida, 2011).
But, the two cannot be divorced since the materials used will dictate to
a considerable extent, the most suitable manufacturing processes which are
available. The composition of the traditional ceramic units lies, in general
in definite limits. They are controlled by some factors: 1-Shape of the ware
which needed to the inclusion of a sufficient plastic material (clay). After
shaping, the clay also holds the ware together during drying. 2-Avoiding of
cracking during firing due to the incorporation of the non-clay materials
which do not shrink during firing as: flint, quartz and tiles grog. 3-The
quantity of fluxes so that a sufficient glassy material formed to hold it
together when fired and so it does not warp (Njindam et al., 2018;
Darweesh, 2015).
2.5.1. Plasticity
When a dry clay is mixed with water, there are two water contents
which could be recognized easily and simply by working the mass of the
194 H. H. M. Darweesh
clay/water body in the hands. Above the higher value of water content, the
mass becomes sticky and difficult to handle. Below the lower one, it tends
to crack and crumble. Accordingly, between these two values (upper and
lower plastic limits), the clay can usually be worked without any difficulty.
Therefore, the plasticity can be expressed as workability which enables a
material to be deformed continuously under a force exceeds certain
maximum values. Plasticity enables a material to be changed in shape
without rupturing by the application of an external force and to retain that
shape when the force is removed or reduced below a certain value, but
there is no a quantitative definition or units for plasticity (Njindam et al.,
2018; Vorrada et al., 2009; ASTM, 1980).
Many trials were carried out to find a relationship between the stress
applied and the resultant measured strain. The stress may be considered as
a force per unit area and the strain as the resulting deformation of the body.
Basically, there are three types of stress which may be applied as tensile,
compressive or shear forces. The property of a body to return to its original
size and shape after having been stretched, compressed or deformed, is
simply that stress is proportional to strain (Rayan and Radford, 1987) as
follows,
Stress
K (1)
Strain
dI
S (2)
I
Nanomaterials, Ceramic Bulk and Bioceramics 195
Where, S is the surface area of the clay sample (expressed as cm2/1000 g),
P is the measured plasticity index, A and B are constants.
The plotting of log S against B gave straight line graphs with intercept
on the log S axis giving the value of log B. In all samples, B was found to
be 1.8 x 106. Thus for P to have any value, S related to be greater than 1.8
x 106 cm2/100g (Rayan and Radford, 1987; Drews, 1983; Mostafa and
Darweesh, 1992). As the material becomes finer, not only does plasticity
increase, but also the amount of water needed to produce optimum
plasticity increases, and plasticity is exhibited over a wider range of water
content. Figure 9 indicates this behavior for various size fractions of clay.
Figure 9. Plasticity index as a function of water content for five size fractions of clay
sample.
196 H. H. M. Darweesh
where the greater the range is the higher plasticity. Figure 10 shows the
BCR compression plastometer by which the plasticity is measured. The
motor moves the lower plate upwards at a constant speed and when the
cam scale reads zero deformation, the upper plate just come into contact
with the top of the sample. As the cam turns further and the lower plate
rises, the sample is compressed between the upper and lower plates
(Drews, 1983; Chiang, 1997).
where, 1.8 is a factor used to permit for a change in the cross-sectional area
of the cylinder during deformation and is being equal to:
Care must be taken to ensure that the samples are of the correct length
and have parallel ends. Deformation of the sample cylinder must therefore
be avoided when unmoulding it from metal die. Often sex samples are
tested and the mean value is considered. Shrinkage of the shaped ceramic
articles takes place during drying and firing stages of manufacture. The
extent of these contraction is very important where not only will the size of
the finished product depend on their values, but also if they are too large,
or if they are occurring too rapidly, cracking or distortion of the article is
likely. So, the ceramist must recognize the wet-to-dry, dry-to-fired and
wet-to-fired processes. Each may be expressed on either a linear or a
volume basis depending on the method of measuring (Rayan and Radford,
1987; Drews, 1983; Rayan, 1978; Khalil, 2012) as the wet-to-Dry
contraction. In most pottery bodies, the clay content is primarily
responsible to determine the extent of the contraction though the ratio, size
and size distribution, particle shape of the non-clay component also play a
vital role. The fineness and plate-like character of the clay crystallites lead
not only to the desirable properties of the high plasticity and dry strength,
but also to high drying contraction. It is well known that a small drying
shrinkage is often desired in pottery products, where it permits the formed
product to shrink away from the plaster mould and facilitates its removal
from the mould. However, the excess shrinkage, specially anisotropic
shrinkage, i.e., different shrinkage values in different directions due
usually to a particle alignment that taking place during pugging, plastic-
making or slip casting, is responsible for a considerable rate of
manufacture losses. High drying shrinkage is also resulting in a loss of
dimensional accuracy and variations, which may give problems in
decoration, e.g., a lithograph is applied to the circumference of a plate, a
high dimensional accuracy and the moisture content of the body may be
kept low as possible (Rayan, 1978; Khalil, 2012), ASTM, 1980).
When a plastic body dries, water is removed from the surface and
replaced by water from the interior. As water content is reduced, particles
are drawn closer together. The particles eventually will touch each other
and though water still exists in the voids between particles, no further
shrinkage can take place on the removal of this water (Figure 11). If a
Nanomaterials, Ceramic Bulk and Bioceramics 199
piece of plastic clay or body lets to dry, values of the volume and the
corresponding moisture content are given at intervals, a plot of volume as a
function of moisture content is as shown in Figure 12. On drying, the
shrinkage takes place as water is removed. During this stage, water still
separate the particles, but the particles are moving closer together if water
is removed. The point at which there is no further removal of water is
known as the critical moisture content “CMC.” Its value often varies from
clay to clay and also from a ceramic body to another. This depends
essentially on clay to non-plastic ratio, particle size, distribution and
particle shape. So, this value is very important where drying from water
content above it will result in a shrinkage involving a risk of distortion or
cracking, and if below it, no shrinkage is involved.
Figure 12. The volume as a function of moisture content for a plastic body during
drying.
200 H. H. M. Darweesh
At moisture content higher than the CMC, the rate at which water is
lost will be constant under constant conditions. Since, drying involves only
evaporation of water from the surfaces of samples which is replaced by
water flowing from the interior of the surface. This is known as “Constant
rate period” (Rayan and Radford, 1987; ASTM, 1980; Konta, 1972). The
interval between the CMC and the dryness is known as “Falling rate
period” (Figure 13). Clay particles are a plate-like in nature and when
mixed with water to form a paste or slip, the particles are free to orientate
themselves when a force is applied. This orientation occurs to present the
least resistance to the applied force, e.g., in a pug, the particles will align
themselves in the direction of the force causing extrusion. The extruded
column of the clay will contain particles oriented so that their major axes
are parallel to the length of the column (Figure 14). Similar orientations
occur during plastic making and slip casting processes.
Figure 13. A plot of water removal rate as a function of moisture content for a plastic
body during drying.
Figure 14. Preferable orientation of the produced particles resulting from anisotropic
drying shrinkage.
Nanomaterials, Ceramic Bulk and Bioceramics 201
a) Clays
About 0ne kg of the clay is mixed with water to give a slurry which is
sieved through 120 mesh (China clay) or 80 mesh sieve (plastic clay). The
resulting suspension is poured on to a plaster of Paris batt and then let to
dry to the plastic condition. When sufficiently dry, the clay is removed
from the batt and thoroughly wedged by the hand till all air inclusions have
been removed, and a freshly cut surface (cut with a cheese wire) has a
homogeneous appearance. The test samples made for measuring its
shrinkage may be either slabs or, a hand-moulded in a plaster of Paris
moulds, or rods of a circular cross –section of about ½ in diameter. Rods
are generally extruded with the help of wad box, and may be used for
shrinkage and modulus of rupture. When the mechanism is operated, a
plastic clay is forced through a tapered section and ultimately through a ½
in diameter orifice to produce rods. Soon after extrusion, the rods are
placed in the grooves of a wooden pallet. The rods are then let to dry in air
for 24 hours after which drying is completed in a drier at 110 ºC. The
distance between the marks is then measured on each of the five rods and
the mean dry length is obtained (kingery, 1975; (Rayan, 1978, Todor,
1972). The linear wet-to-dry shrinkage is reported on the wet basis, i.e., the
% linear drying shrinkage (wet basis) is given by:
The moisture content of the rods at the beginning of drying should also
be quoted with these results. The rods should be then fired through the
production kiln in such a position as to receive an average firing
temperature, and a pyroscope (cone or ring) should be fired with the
samples. Again, the distance between marks is measured on the rods, and
the average fired length is calculated. It is suggested that the linear dry-to-
fired shrinkage is reported on the dry basis, i.e., the % linear firing
shrinkage (dry basis) is given by:
202 H. H. M. Darweesh
b) Ceramic Body
Concerning the shrinkage measurements on the body, the production
of a casting slip or a plastic body should be used as the starting material,
where the same procedures as for clay are used.
c) Volumetric Contraction
The more accurate measurement of both drying and firing shrinkages
could be obtained on the basis of volumes by using Doulton Densometer
(Figure 15). Two weights are often required during drying of sample, the
weight of the sample (w1) and the down thrust force (w2), which is equal
to the up thrust of mercury + the sample weight when the sample is
immersed in mercury and held immersed by the cradle.
Figure 15. Doulton Densometer for bulk density and volume shrinkage.
Nanomaterials, Ceramic Bulk and Bioceramics 203
d) Adsorbed Water
Plasticity is associated with adsorbed water films around each clay
particle or micelle. The electrical charges on clay particles enable them to
adsorb polar micelles like water at their surfaces. These adsorbed films
appear to act as lubricants, and are responsible for the plastic behavior. If
water in a clay paste is replaced by a liquid of lower polarity, e.g., alcohol,
the plasticity is much reduced since adsorption of the liquid is lessened.
204 H. H. M. Darweesh
e) Ionic Effects
All clay particles carry a negative charge, and hence have the ability to
attract and adsorb cations at their surfaces. These cations are not firmly
held and on treatment with suitable electrolyte solutions one type of
cations can be replaced by another, g. a clay having adsorbed Ca2+ ions at
its surface can have these replaced by Na+ ions on treatment with Na2CO3,
The Na-clay produced has very different physical properties than those
of Ca-clay. In a flocculated clay or body, the particles attract to each other
giving open packed particle arrangement requiring such water to produce
an optimum plasticity. Therefore, flocculation is used to increase the
workability of industrial tableware units. Such treatment suffers from a
disadvantage that unfired strength is reduced due to open particle packing
and reduced its bulk density (Rayan and Radford, 1987; Drews, 1983;
Rayan, 1978; Khalil, 2012).
g) Ageing of Souring
It is the practice of storing plastic clay or body for a time period prior
to its use, which commonly employed in the heavy clayware production,
but less so in white wares. The improvements in workability are essentially
due to two factors:
Nanomaterials, Ceramic Bulk and Bioceramics 205
When the clay dries from the plastic state or the dried unfired ware, its
strength has been developed. The dry strength of the clay depends on the
nature of the clay and the size of the clay particles. As a general rule, the
more plastic clays are the strongest. The construction usually takes place as
the clay dries due to the loss of water of plasticity. A little shrinkage is
desirable and excessive construction leads to problems such as warping
and cracking. A further construction always takes place during the firing of
clay article due to the melting of fluxes which are present in varying
amounts in the different clays (kingery, 1975; Mostafa and Darweesh,
1992). Darweesh, 2015 studied the dry and firing shrinkage of some
ceramic composites containing 0, 5, 10, 15, 20 and 25 wt. % cement kiln
dust waste (CKD) at 0, 1000, 1050, 1100 and 1150ºC to throw light on the
behavior of the samples on heating according to ASTM-Specification,
1980. Table 1and Figure 16 illustrate the dry and firing shrinkage as a
function of CKD content. He found that the dry and firing shrinkage
slightly increased with the increase of CKD content up to 25 wt. %, and
also with the increase of firing temperature (kingery, 1975; 1978; Khalil,
2012; ASTM, 1980; Mostafa and H. H. M. Darweesh, 1992; Chiang,
1997).
CKD content, wt %
Firing temperature, 0 5 10 15 20 25
ºC 0 0 0 0 0 0 0
1000 0.022 0.043 0.105 0.165 0.189 0.251
1050 0.106 0.152 0.286 0.482 0.535 0.603
1100 0.313 0.406 0.654 1.108 1.185 1.223
1150 0.473 .518 0.654 1.312 1.342 1.363
206 H. H. M. Darweesh
Figure 16. Dry and firing shrinkage of ceramic products with various CKD content.
h) Fired Color
The unfired color of the clay becomes another color after firing. For
example: many pottery clays in the dry state are brown or dark grey and
become almost white color by firing. A slip is a suspension of clay in
water. When 9 parts by weight of dry constituents are mixed with 10 parts
by weight of water, the resulting slip will have the consistency of a stick
cream. If we now add a small amount of two common alkaline materials
(CaCO3 and Na3SiO3), the slip will become quite fluid and it will be
possible to add a further 17-18 parts by weight of the dry materials before
the slip starts to thicken again without adding any more water. The process
of rendering a thick slip fluid in this manner is known as deflocculation
and the materials used to bring it about are called deflocculents.
Deflocculation of a pottery body is dependent on the clays present and is
very important in the production of a slip for the making process known as
casting. This method is used for hollowware, tea ports, jugs, etc., and in the
manufacture of sanitary wares (Jackson, 1969; Mostafa and Darweesh,
1992).
Nanomaterials, Ceramic Bulk and Bioceramics 207
Clay and Loams are sedimentary species; clay; silt and sand with low
alumina content from 16-23% and high proportion of iron compounds, are
used in the manufacture of colored ceramic tiles. They are usually divided
into two catigories: not-easily fusible pottery clays and loams, and
common pottery ones. It is subdivided by konta (Konta, 1981) into:
2.6.1. Fillers
This is a rather loose term used to describe non-plastic materials used
in body compositions, which are not fluxes. In practice, the filler of white
ware bodies is always silica, either in the form of flint of sand though few
bodies often used alumina. Fillers serve to open up the body facilitating the
drying process, reducing the drying shrinkage and plasticity. Moreover,
they are controlling the thermal expansion and contributing the whiteness
of the fired bodies. Fillers are often representing a large ratio of the total
body composition. So, the particle size distribution is very important in the
determination of the particle packing in the body. This is in turn affects
drying and firing shrinkage. Therefore, the fineness of the filler affects the
thermal expansion of the fired body. Fillers are high melting, chemically
resistant inorganic materials whose main function is to reduce a ceramic
body's tendency to warp or distort when fired to high temperatures. They
also play an important role in the determinaton of thermal expansion.
Quartz and/or flint are the most common fillers used in white wares and
other ceramic bodies. Though Al2O3 is an important ingredient in many
refractory and technical ceramic bodies, it is used in a few tableware
bodies (Jackson, 1969; kingery, 1975; Rayan and Radford, 1987; Mostafa
and Darweesh, 1992; Grim, 1962).
Sandstones are common rocks, but the sandstone pure enough to use in
white ware bodies are infrequent. Many are containing iron compounds,
which stain them brown or red colors (kingery, 1975; Mostafa and
Darweesh, 1992; Chiang, 1981).
2.6.3. Flints
Flints contain small amounts of water, minute air spaces and interstitial
amorphous SiO2 due to its mode of formation from the skeletons of
sponge-like organisms which dissolved in sea water and were later
deposited in chalky deposits. The fine microstructure of flints makes it
more reactive than quartz, and in particular, it converts during firing more
readily to cristobalite (another crystalline form of silica) which is
profoundly affects the thermal expansion of the body (kingery, 1975;
ASTM, 1980). Flint occurs as hard nodules in the middle and upper parts
of the chalk deposits often arranged in layers along bedding planes
separating the beds rather than evenly distributed throughout the chalk.
Therefore, flints were formed by precipitation from silica in solution. This
is confirmed by the fact that the white skin of flint gradually varies in
composition from CaCO3 to silica. In order to prepare flints for use, flints
pebbled from the sea shore or from chalk deposits are calcined up to
1100ºC to make it friable, and to facilitate its grinding to the required
fineness (45-50%less than 10 µ). During calcinations, flints easily shatter
due to expansion of air and water pockets within the structure. This
changes its color from black to white as an organic material is burnt off.
The specific gravity of flint is 2.62 reduced to 2.50 after calcination due to
the exfoliation of the flint on heating and to a lesser extent to the formation
of cristobalite (spec. gr. 2.3). About 18% of quartz is converted to
cristobalite during calcination process of flint. The calcined flint is wet
ground to the required size and used in a slop form in the body mix.
Grinding to the correct fineness is very important since it affects the extent
of conversion of quartz to cristobalite during firing of the body (kingery,
1975; ASTM, 1980; Darweesh, 1992). At all, flints of interest to the potter
are divided to:
210 H. H. M. Darweesh
5730 C
-Quartz -Quartz (7)
8700 C
-Quartz 2 -Tridymite (8)
14100 C
2 -Tridymite -Cristobalite (9)
220 2800 C
-Cristobalite -Cristobalite (10)
1630 C
2 -Tridymite 1 -Tridymite (11)
1170 C
1 -Tridymite -Tridymite (12)
2.6.4. Fluxes
A flux is a material which reduces the fusion temperature of the
materials or mixtures to which it is added. It is not strictly possible to
divide material to fluxes and refractory materials, since whether or not a
material acts as a flux depends not only on the material itself, but also on
the material to which it is added. It is good mention that as the alkali
content increases in a fluxing material, its fluxing action increases. The
F2O3 content of the fluxing material must be in its lowest content, or it will
impart its color to the fired ceramic body. Also, the grain size of the fluxes
is of great and vital importance (kingery, 1975; Rayan and Radford, 1987;
ASTM, 1980; Mostafa and Darweesh, 1992). Fluxes melt when the ware is
fired. On cooling, the glass solidifies and provides the bond that holds the
whole mass together.That is responsible for the strength of the pottery
ware. Cornish stone, feldspar and nepheline syenite are fluxes (kingery,
1975; Grim, 1962; Schuller and Jager, 1979).
2.6.4.1. Feldspars
It is essentially a mixture of two feldspars namely, orthoclase-potash
feldspar, KAlSi3O8 or K2O. Al2O3. 6SiO2 and albite-soda feldspar,
NaAlSi3O8 or Na2O. Al2O3. 6SiO2. Small amounts of mica and quartz are
usually present. The feldspar is often extracted by a normal quarrying or
mining with visual control over the extracted material at the quarry face.
Briefly, the method starts by crushing the rock to less than one inch by
means of jaw and gyratory crushers which followed by wet milling to pass
a 30 mesh sieve. The three stage froth flotation process which follows, first
removes the mica, garnet and other similar materials. Then, the feldspar is
frothed from the quartz. Finally, the separated feldspar undergoes a further
flotation to reduce the quartz as much as possible. The concentrated
feldspar and quartz which are separated by this method are then dried and
stored. Though feldspar contains about 5% quartz, quartz contains 6-8%
feldspar (kingery, 1975; Mostafa and Darweesh, 1992; Schuller and Jager,
1979).
Nanomaterials, Ceramic Bulk and Bioceramics 213
2.6.4.4. Talc
It is a magnesium silicate and a member of montmorillonite group
having the formula of 3MgO. 4SiO2. H2O. It is a very cheap source of
magnesia which is acting as a flux. The presence of Talc in bodies that
Nanomaterials, Ceramic Bulk and Bioceramics 215
2.6.5. Clyas
Clay minerals are often formed from the decomposition of igneous
rocks as granite which is composed mainly of roughly equal proportions of
potash mica (K2O. 3 Al2O3. 6SiO2. 2H2O), quartz (SiO2) and potash
feldspar (K2O. Al2O3. 6SiO2). Kaolinitic clay was the decomposition or
kaolinization of feldspar in the presence of air and water (Jackson, 1969;
kingery, 1975; Singer and Singer, 1971; ASTM, 1980; Mostafa and
Darweesh, 1992) as follows:
450500 C
Al2O3 .2SiO2 .H 2O Al2O3 .2SiO2 H 2O
0
(14)
2.6.6. Limestone
It is an important ingredient in wall and floor tile compositions due to
its capacity to redduce the moisture expansion of the resulting ceramic
units. This is a very critical charater specially in floor tiles (Kingry, 1975,
Mostafa and Darweesh, 1992). Limestone is often used at a range of 10-12
wt. % addition in tile products to lower the quantity of melting component
Nanomaterials, Ceramic Bulk and Bioceramics 217
700900 C
CaCO3 CaO CO2
0
(16)
Floor tiles are belonging to the 1st and 2nd groups with W. A. only
up to 6% (Rayan and Radford, 1987; Drews, 1983; Rayan, 1978).
There are two different types of floor tiles:
a) Floor Tiles
They are dense bodies, fully vitrified, colored frequently by the color
of the natural clays used in the batch or by colors, and can be obtained in
various colors as well as different textures. The extenal surfaces of floor
tiles are fine and soft. In some cases, surfaces are self-glazed. These are
characterized by their high resistance to abrasion and weather stains.
218 H. H. M. Darweesh
b) Quarry Tiles
The quarry tiles are thicker and extremaly bigger if compared with
floor tiles and are having a rough surface appearance with W. A. between
2-5%. These tiles are used in floor of public buildings, silos and wall
linings.
decrease volume and weight with retained strength and decrease the firing
temperature of the tiles (ASTM, 1980), the raw mixture of tiles contains
wastes from asbestos cement production was used.
Mass
Density
Volume (17)
For a vitreous object, there is only one weight and one volume
involved, but for a porous solid, there are three common volume
expressions as follows:
222 H. H. M. Darweesh
Weight D
Appearent (bulk ) Density (18)
Appearent volume S I
Weight
True Density
True volume (19)
Weight D
Appearent solid Density (20)
Appearent solid volume D I
where, D, S and I are the dry, soaked and immersed weights of the
material, respectively.
Nanomaterials, Ceramic Bulk and Bioceramics 223
W1 W2
W . A, % 100 (21)
W3
W3
B.D, g / cm3 (22)
W1 W2
W1 W3
A.P, % 100 (23)
W1 W2
where, W1, W2 and W3 are the saturated weight in air, suspended weight
and the dry weight, respectively. Darweesh et al., 2012 studied and
discussed W.A, B.D and A.P of some ceramic batches containing various
contents of CKD waste up to 25 wt. % (Figures 17-19). Table 4 shows the
chemical analysis of the starting raw materials used in this work. They
reported that the W.A and A.P decreased with firing temperature up to
1100ºC, but only decreased with CKD content up to 10 wt. %, and then
decreased with more addition of CKD content as shown in Figures 17 and
18, respectively, while the bulk density increased as shown in Figure 19.
224 H. H. M. Darweesh
Figure 17. Water absorption of ceramic products with different CKD contents up to 25
wt. %.
Nanomaterials, Ceramic Bulk and Bioceramics 225
Figure 18. Apparent porosity of ceramic products with various LCD contents up to 25
wt. %.
Figure 19. Bulk density of ceramic products with various LCD contents up to 25 wt.%.
226 H. H. M. Darweesh
3PL
F .S , MPa (24)
2bd 10.2
where, δ: flexural strength, MPa, P: the load of rupture, kg, L: span or the
distance between the two lower beams (5 cm), b: width of sample, cm and
d: thickness of sample, cm.
Figure 20. Schematic diagram of the bending strength, B: beam, S: span, W: width and
T: thickness.
Nanomaterials, Ceramic Bulk and Bioceramics 227
Figure 21. Flexural or bending strength of ceramic products with different contents of
CKD waste, %.
For a common practice, it often used two frits, i.e., Lead and Borax
frits or its oxides (PbO and B2O3). When these oxides are fritted together,
the lead does not generally reach satisfactory limits of insolubility.
However, lead borosilicate can be produced satisfactory if the PbO:B2O3
ratio is carefully selected (Rayan and Radford, 1987; Mostafa and
Darweesh, 1992; Mukhamedzhanov et al., 1991).
The glaze could be classified also due to the type of ware on which the
glazes applied as for Earthern wares, Sanitarywares and Porcelain glazes.
There is a wide range of glazes available within each group. This
classification has been broadened to include firing and maturing
temperatures noticing that in the following examples, glaze molecular
formulae have been given, where the oxides have been categorized as
basic, amphoteric, acidic. The sum of the basic oxides is one, where the
temperatures are referred to the maturing range (Jackson, 1969; Kingery,
1975; Rayan and Radford, 1987; Njindam et al., 2018; ASTM, 1980).
1- Majolica (900-1050ºC)
[0.7 PbO +0.3 CaO] +0.15 Al2O3 + [2.0 SiO2 + 0.3 B2O3]
2- Earthenware (1000-1150 C-Lead glaze
[0.4 PbO+0.3 CaO+0.3(Na,K)2O] 0.25 Al2O3 [2.5 SiO2+0.5 B2O3]
3- Leadless glaze
[0.55 CaO 0.3 (Na,K)2O] 0.30 Al2O3 [3.0 SiO2 + 0.8 B2O3]
4- Sanitaryware (1200-1250ºC)
[0.6 CaO+0.2 (Na,K)2O+0.2 ZnO] 0.35 Al2O3 [3.0 SiO2]
5- Hard paste porcelain (1400ºC)
[0.68 CaO+0.20 MgO+0.12 ZnO] 1.0 Al2O3 [10.0 SiO2]
Nanomaterials, Ceramic Bulk and Bioceramics 231
Colored bodies and slips have a few percent of the desired color added
to the bulk mix during preparation in the slip house. The major
disadvantage of using colored bodies is that shading of wares may take
place. A ceramic color seldom consists simply of a coloring oxide or
compound. This fact is understood by considering the following cobalt
blue color known as “Royal blue” or “Maz blue” (Jackson, 1969; Rayan,
1978; ASTM, 1980).
Cobalt oxide 45%, Whiting 10%, Flint 18%, Alumina 5%, Feldspar
22%. The color may be analyzed in terms of the function of its
components. Cobalt is the basic coloring compound giving the intense blue
color. CaCO3 acts as a modifying agent that modified the blue color to give
new blue tint and stability. Flint and Al2O3 act as a diluents, i.e., as filler
giving a slightly faint but more stable color. Feldspar which uses as a flux,
is always assisting in the sintering of color mixes through the calcination
process noticing that the underglazed colors need about 5% flux, but on
glaze colors need about 70%. Ceramic colors are usually prepared to fix
and carefully control specifications. Even so, it is difficult to produce
decorative wares of consistent quality on a commercial scale. It must
always remember that the final color depends mainly on the composition
of the glaze, firing temperature and kiln type. These are very important
particularly in sanitary ware industry, where the finished product can be
sold to an exact standard color, while the vitreous sanitary ware is
generally colored with opaque pastel glazes which have to match other
colored fittings, in the cloak or path rooms manufactured from quite
different materials. The metamerism phenomena must be taken into
consideration which is two color samples may appear to match, under a
given illumination, even though their spectrophotometric curves are not the
232 H. H. M. Darweesh
same (Figure 23). So, the sanitaryware colored glazes are formulated from
a color and compatible base glaze to an exact specification. If a different
illuminant is used, the samples may no longer match where the perceived
color depends essentially on the illuminant and the reflection curve at the
surface (Drews, 1983; Danto, 1979, Vauequelinet et al., 1980; Fiori and
Fabbri, 1980; Darweesh and El-Din, 2000). The color specification by CIE
system gives the chromaticity values or coordinates x, y and Y where these
values may be used in conjunction with the CIE chromaticity diagram
(Figure 24).
The difference between the contraction of the ceramic body and the
glaze during cooling often causes some faults. The thermal behavior or
expansion of ceramic products during firing can be measured by using
Ortom Automatic Dilatometer (Figure 25). In this type of instruments, the
temperature can be controlled in the range of 0.2-10ºC/min. and the rate of
heating is 3-5ºC/min. (Darweesh, 2016) investigated the thermal behavior
or expansion of some ceramic batches as 0, 5, 10, 15, 20 and 25% cement
kiln dust waste (CKD) using Orton Automatic Dilatometer to throw light
on the behavior of the samples on heating.
instrument. The test piece with parallel ends is preheated to 900ºC, cooled
and then pushed along the silica tube using thrust rod till it contacts the
silica stopping disc. The dial gauge is firmly clamped into position and the
needle adjusted so that a slight pressure is exerted on the thrust rod. The
outer ring on the dial gauge is moved to zero reading. The thrust rod
should move quite freely within the silica tube. The base of the apparatus
should be gently tapped till the dial gauge gives a consistent zero reading.
Figure 28 shows the typical thermal expansion of earthenware body and
low sol glaze. At 500ºC, the thermal expansion is 0.38% and 0.32% for
body and glaze, respectively. The difference of expansion is 0.06%.
Figure 26. The coefficient of linear thermal expansion as a function of CKD content
and its effect on the thermal expansion of ceramic products.
Nanomaterials, Ceramic Bulk and Bioceramics 235
Figure 27. Thermal expansion apparatus (A: Slate base, B: Silica tube, C: Stopping
disc, D: Fused silica rod, E: Invar extension rod, F: Dial gauge, G: Furnace, H: Test
specimen.
2.10. Applications
76, sandstone 15-20, chamotte 3-5 and bentonite 1-5 wt. % (Keijiro et al.,
1991).
The CKD waste is the main source of air pollution in many countries,
which resemble the present day problem, be a matter of great importance
due to the precipitation of huge amounts of it behind the kilns and its
characteristics such as: light weight and fine grains so that it can be
transported easily by air storms to many agriculture and population areas.
Keijiro et al., 1991 manufactured high strength ceramic tiles from slurry
containing cement and mineral materials, e.g., feldspar, siliceous stone and
SiO2 by moulding the slurry followed by drying and firing. The cement is
selected from Portland and/or white Portland cements. Abdel-Fattah and
Nour, 1981; Abdel-Fattah et al., 1982; Abdel-Fattah and Eh-Didamony,
1981) studied the CKD supplied by the factory. The dust was fired
between 1000-1250ºC, then two calcines (70:30) and (50:50) were
prepared by blending the raw dust and kaolin. It was stated that it consisted
mainly of dolomitic limestone, minor amount of Na2O, K2O, SiO2, 2
C2S.CaCO3 and 2 C2S.CaSO4. The formed phases in the 1st calcined mix,
was found to be mainly of gehlenite (C2AS), but in the 2nd calcined mix, it
was mainly of β-C2S.
Using of CaO sources in ceramics fields and tile production is
recommended as a whiting and a flux (Darweesh, 2001). The cement kiln
dust (CKD) is a mixture of finely divided particles and partially calcined
raw materials with condensed volatile salts (Anlagenbau et al., 1995;
Saltyvskaya, 1985). So, the use of dust along with kaolin in the form of
calcined mix proved also to be used successfully in replacing feldspar in
porcelain bodies. Darweesh, 2010 also studied and utilized the alkaline
wastes obtained from cement kilns to produce porcelaineous bodies.
Darweesh, 2010 exploited the CKD waste from cement factories as a
source of CaO in ceramic industry by 25% to produce wall and floor tiles.
W.A, B.D. and A.P. as well as B.S. and T.F. were researched (Figures 29-
31). The addition of the waste improved and enhanced all properties of the
prepared units. Darweesh, 2001; Darweesh and khalil, 2001 demonstrated
that adding 5 up to 10% waste to the aumina cement improved the specific
Nanomaterials, Ceramic Bulk and Bioceramics 237
Figure 29. Water absorption, bulk density and apparent porosity of the ceramic bodied
of the base batch and the batches containing CKD waste.
Figure 30. Compressive and bending strengths as well as abrasion resistance of the
base batch and the batches containing CKD waste.
238 H. H. M. Darweesh
Figure 31. Firing shrinkage and linear thermak expansion of the base batch and the
batches containing CKD waste.
Figure 32. Chemically-combined water content of the various alumina cement pastes
containing cement kiln dust waste hydrated up to 28 days.
Nanomaterials, Ceramic Bulk and Bioceramics 239
Figure 33. Free lime content of the various alumina cement pastes containing cement
kiln dust waste hydrated up to 28 days.
Figure 34. The compressive strength of the various alumina cement pastes containing
cement kiln dust waste hydrated up to 28 days.
3. BIOCERAMICS
3.1. Introduction
The sol-gel produces porous Al2O3 and ZrO2 toughened alumina (ZTA)
substrates that can be prepared and impregnated with a synthetic body fluid
(SBF) and dicalcium phosphate (DCP) solutions for different periods to
prepare the bioactive composites. The resulting porous ceramic substrates
are then characterized by measuring their physical properties in terms of
B.D and A.P. The phase composition and the microstructure of the
prepared scaffolds or porous ceramic substrates can be determined by
using XRD and SEM techniques (Mohamed, 2012; El-Hady et al., 2015;
Hench and West, 1990).
El-Hady et al., 2015; Hench and West, 1990 used a certain polymer
based on the gelatin from bovine skin (Figure 37) and agarose (Figure 38)
to prepare composite scaffolds using the sol-gel technique. They also used
ciprofloxacin (Figure 39) to avoid bacterial infections (Daniel and Vallet-
Regi, 2010; Balamurugan et al., 2008; Barba-Izquierdo et al., 2000).
Figure 40. The Sol-gel processing could be used to produce nanostructured layers
and/or coatings as well as nanoporous membranes. 1- Hydrolysis. 2: Condensation. 3:
Gelation, 4: Evaporation and Drying. 5: Nanoporous membranes.
powder be based on the kind of the catalyst and the porosity of the gel
(Saravanapavanand Hench, 2003).On the other hand, Brinker and Scherer,
1990 considered that the gelation is essentially depends on the ratio of
H2O/TEOS. At lower ratios, the gelation time became quicker.
Water is added as an initiator and it can be generated by condensation
reactions, e.g., esters by a condensation of carboxylic acids and alcohols
(Iarry, 1991). In some researches, ethyl alcohol could be used as a solvent,
where hydrolysis takes place by water into a non-aqueous solution as
alcohols, whilist Saravanapavan and Hench, 2003 reported that CaO ratio
is < the theoretical value. This is attributed to the variation in the leaching
of cations during aging and drying stages when the process was carried out
in presence of the pore liquor (ethyl alcohol). In the present study ethyl
alcohol did not used.
After a transparent green sol was obtained, The sol was held at 60°C
(aged) to reach to a viscosity close the gel point, where transparent gel was
formed (in the present procedure two hours at 60°C, the gelation began.
This temperature was in agreement with other research (Saravanapavan
and Hench, 2003; Balamurugan et al., 2006; Crayston, 2003; Lao et al.,
2006; Balamurugan et al., 2005; Liu and Miao2004; Wang et al., 2008;
Lao et al., 2007), and it was chosen according to observation of
temperature at which the solution become completely miscible or in other
word when the solution become completely miscible the temperature
already reach to 60°C. So, we kept the solution at this temperature to
obtain transparent gel system. Hydrolysis and condensation continue
giving an increase in viscosity (Chen et al., 1999; Saravanapavan and
Hench, 2003). Minimizing of gelling time and rising of Ca2+ ions are due
to a lack of ionic charges on sol-gel by the salt (Saravanapavanand Hench,
2003; Wang et al., 2003). Gupta et al., 2005 reported that gelation
increases the viscosity, shrinkage, weight loss, but decreases the pore size
distribution. Scherer, 1999 also reported that after the gelation, the
chemical reactions still proceeded, increased the rigidity and shrinkage.
Gel was dried slowly up to 110°C to remove water and also adsorbed
moisture from pores. Heat will densify the gel by removing of water.Then,
discs transformed from transparent to translucent opaque. The drying
246 H. H. M. Darweesh
chosen due to its high wear resistance, their excellent scratch resistance,
good frictional properties and fracture toughness (Nalwa, 2000; Tseng and
Nalwa, 2009; Ashby, 2009).
The high stability of oxide ceramics as low corrosion and low ion
release is due to the high heat of formation of their molecules, in addition
to its biocompatibility with the physiological environment, that make these
materials reliable for a variety of applications as biomedical applications
including knee-prostheses and dental implants. Hence, ceramic oxides of
Al2O3, TiO2 and ZrO2 have been introduced as reinforcing agents (Kokubo
et al., 2003; Wan et al., 2008; Arstila et al., 2007; Liu, and Ding, 2002;
Wei-hong et al., 2006).
Sol-gel helps the formation of nano-particles, nano-films, and nano-
porous membranes (Figure 40). A precursor solution in a solvent, e.g.,
alkoxides and cations can form a colloidal suspension (SOL-GEL) due to
some polymerization reactions, which by adding a surfactant the dispersed
particles could be kept in a suspension used to extract the particles for
more processings to be a substrate. Solvent evaporation creates dense or
nano-porous films. Sol-gel techniques can develop many new materials as
paints, nano and/or bioceramics (Wang et al., 2008; Lao et al., 2007;
Binnaz and Hazar, 2007; Padilla et al, 2005; Long et al., 2006; Long et al.,
2006; Meseguer-Olmo et al., 2008; Wu et al., 2007; Wan et al., 2008;
Arstila et al., 2007; Peng et al., 2004).
CONCLUSION
REFERENCES
Darweesh, H. H. M., Building materials from siliceous clay and low grade
dolomite rocks, Ceramics International, 27, 2001, 45-50.
Darweesh, H. H. M., “Utilization of Cement Kiln Dust Waste as a source
of CaO in Ceramic Industry,” Sil. Industrials (Ceramic Science and
Technology), 66 (3-4) 2001, 47-52.
Darweesh, H. H. M. & Khalil, N., Effect of cement bypass dust waste on
the physico-mechanical properties of alumina cement, Ind. J. Chem.
Techn. India, 7, 2001, 287-291.
Danto Co. Ltd. “Decorative tiles,” Jpn. Kokai, 79/70, 250, Jan. 14, 1979.
Drews, M., “Wall and floor tiles,” Ceramic Monographs 2.4.1. Verlag
Schmid GmbH, 1983.
El-Alfi, E. A.; Radwan, A. M. & Darweesh, H. H. M., Effect of sand as
non-plastic material on ceramic properties of clay bricks, Inter Ceram
(Intern. Cer. Review), 53 (5) 2004, 330-333.
El-Hady, B. I.; El-Kady, A. M. & Hassan, M. M. A., Preparation and
characterization of biocomposites for localized bone treatment based
on Agarose and Gelatin, Ph. D. Thesis (Biophysics), University
College of Women, Ain Shams University, 2015.
Fiori, C. & Fabbri, B., Granite containing bodies for the production of
Stoneware tiles, Res. Inst. Ceram. Techn. National Res. Council,
Faenza, Italy, Inter Ceram 32 (1) 1980, 21-22.
Genin, G., Influence of Quartz on the strength of porcelain, Stekelo.
Keram., 15, 1958, 35-41.
Giomlem, K. & Lyng, S., Anorthosite as a ceramic raw material-Part II;
Anorthosite for tile manufacture, Trans. Brit. Ceram. Soc. 73, 1974,
139146.
Grim, R. E., Clay Mineralogy, McGraw Hill, NewYork, 1962.
Gupta, R.; Mozumdar, S. & Chaudhury, N. K., “Fluorescence
spectroscopic studies to characterize the internal environment of
tetraethyl-orthosilicate derived sol–gel bulk and thin films with aging,”
Biosensors and Bioelectronics 20, Issue 7, 2005, 1358-1365.
Haiyan, L. & Chang, J., Fabrication and characterization of bioactive
wollastonite /PHBV composite scaffolds, Biomaterials, 25, 2004,
5473-5480.
Nanomaterials, Ceramic Bulk and Bioceramics 255
Hayashi, S.; Nakagawa, Z.; Yasumori, A. & Okada, K., Effects of H2O in
EtOH-H2O disperse medium on the electrophoretic deposition of
CaSiO3 Fine Powder,” Journal of the European Ceramic Society, 19,
1999, 75-79.
Hench, L. L. & West, J. K., The sol-gel process, Chemical Review, 1990,
90-133.
Iarry, H. L., Bioceramics from concept to clinical, Amer. Cer. Soc., 74,
1991, 1487-1491.
Ibrahim, D. M. M; Abdel-Fattah, W. I. & Nour, F., Air-cooled blast
furnace slag as a fluxing material in the Ceramic Industry, Ber. Dt.
Keram. Ges. 57, 9-10, 1980, 187-194.
Jackson, G., “Introduction to white wares,” Elsevier Publishing Co., 1st
Edn., 1969.
Joachim, S.; Manfred, S. & Peter, T., Ceramic composition for light-
colored gehlinite-free glazed earthenware tiles, VEB kombinat Fliesen
and Sanitaerkeramic “Kurt Buerger,” Ger. (East), DD 205, 883, CI
CO 4B33 /00, Jpn. Jan, 11, 1984.
Julian, R. J.; Lisa, M. E. & Larry, L. H., Optimising bioactive glass
scaffolds for bone tissue engineering, Biomaterials, 27, 7, 2006, 964-
973.
Julie, C. D., “R & D 2002: Nano Ceramics.” MIT Technology Review.
Retrieved Dec. 1, 2002.
Kalnin, I. L., Strength & elasticity of Whitewares-Part I; Relation between
Flexural strength and elasticity, Ceram. Bull., 46, 1174, 1967,
Kalnin, I. L.; Cucka, P.; Warshow, S. I. & Seider, R., “Strength and
elasticity of Whitewares,” Ceram. Bull., 47, 1968, 498-503,
Kashlyak, L. L. & Kareev, Yu. B., Production of tiles for floor and Façade
tiles in the Leningrad Ceramic Products Plant, Keram from St. 8, 10,
1978.
Ke, S.; Wang, Y.; Pan, Z.; Ning, C. & Zheng, S., Recycling of polished tile
waste as a main raw material in porcelain tiles, J. Cleaner Prod. 115,
2016, 238–244.
Keijiro, S.; Takashi, o.; Yoshifumi, K.; Yoshiyuki, O; Masaru, T. & Kozo,
M., Manufacture of ceramics containing Portland cement, Sumitomo
256 H. H. M. Darweesh
Cement Co. Lmt., Jpn. Kokai, Tokyo Koho JP, o3, 215, 350, 1991,
[91, 215, 350] (C1. CO 4B35/16). Chem. Abst. 115: 285800p.
Khalil, A., “Advanced Sintering of Nano-Ceramic Materials.” Ceramic
Materials - Progress in Modern Ceramics, 2012.
Khrundzhe, A. V. & Babushkin, V. I., State Reginal electric power plant
wastes used for the manufacture of ceramic tiles, USSR, Steklo,
Keram. 3, 1983, 6-8.
Kim, K, and Hwang, J., Characterization of ceramic tiles containing LCD
waste glass, Ceram. Int. 42, 2016, 7626–7631.
Kimm, F., “Ceramics Don't Have to Be Brittle.” Cal Tech. Retrieved 11,
2014.
Kimm, F., “Made-to-Order Materials.” Cal Tech. Retrieved 5, 2013.
Kingery, W. D.; Bowen, H. K. H. K. & Uhlmann, D. R, Introduction to
Ceramics, 2nd edn., John Wiley and Sons, New York, London, Sydney,
Toronto, 361, 594, 1975.
Kisel, I. I.; Strodyreva, G. V. & Demidovisch, V. B., Possibilities of using
local clay in manufacturing floor tiles, Steklo, Sitally Silik, 1980.
Klein, I. Sol-gel optics: Processing and applications, Springer Verlag,
ISBN 0792394240, 1994.
Koch, C. C., Nano structured materials: Processing properties and
applications, Noyes Publications, 2002, ISBN 0-7155-1451-4.
Kokubo, T., Bioactive glass ceramics, Properties and applications,”
Biomaterials, 12, 1991, 155-163.
Kokubo, T.; Kim, H. & Kawashita, M., Handbook of Advanced Ceramics,
2003, 385-416. Chapter 14 - 14.1 Ceramics for Biomedical
Applications.
Kong, Y. M.; Kim, S. & Kim, H. E., Reinforcement of hydroxyapatite
bioceramics by addition of ZrO2 coated with Al2O3, Amer. Cer. Soc.,
82, 1999, 2963-2968.
Konta, J., Properties of ceramic raw materials, Ceramic Monograph 1.1.4.
Verlag Schmid G.m.b.H, 1981.
Lao, J. Nedelec, J. M.; Moretto, Ph. & Jallot, E., Biological activity of a
SiO2–CaO–P2O5 sol–gel glass highlighted by PIXE–RBS methods,
Nanomaterials, Ceramic Bulk and Bioceramics 257
Chapter 8
ABSTRACT
The need for simulation of low cost automation and life cycle cost
analysis for a special kind of machine is a promising research area in
industries. This machine is capable of machining (undercut, head
diameter, fillet, chamfering) operation of engine valve, used in
automobile industry. The objective of this work is to reduce rejection,
lower the cost, increase productivity and quality of performance. In this
chapter, a methodology to design, develop and simulate a twin spindle
turning Special Purpose Machine is presented based on the data collected
from hydraulic, pneumatic, electro pneumatic which will serve as low
Corresponding Author Email: lalhmingsanga.mech@nitmz.ac.in
264 Pattanayak Satyajit and Hauchhum Lalhmingsanga
NOMENCLATURE
ABBREVIATIONS
1. INTRODUCTION
2. METHODOLOGY
1) Chute on/off
2) Loader on/off
3) Pusher on/off
4) Ejector on/off
5) Collet on/off
Simulation of Low Cost Automation and Life Cycle Cost … 267
Figure 3. Pusher.
Figure 4. Collet.
The Special Purpose Turning Machine, having two spindle head unit,
is used for carrying out the turning head diameter, undercut, chamfering
head and neck operations during manufacturing of engine valves.
Assembly of Special Purpose Turning Machine is shown in Figure 7.
The engine valve manufacturing involves a sequence of operations
starting from casting to final finishing, out of which turning head diameter,
undercut, chamfering and fillet form four intermittent operations. These
operations are required for turning head diameter as per dimension,
undercut operation for holding the spring, chamfering operation for good
fittings.
The Special Purpose Turning Machine will have the combined
advantages of a CNC Machine and Special Purpose Machine. This
machine is built only for the purpose of turning head diameter, undercut
270 Pattanayak Satyajit and Hauchhum Lalhmingsanga
A-A+B+C+D-C-B-D+E+E-
• Very complicated
• Large space is required
• Difficult in assembling and dissembling
• Difficulty in maintenance.
• One labour is always required when process is going on, for
operating the start button at every cycle.
(V4) and to blank end area of gripper cylinder. therefore, gripper cylinder
moves forward, (A+ B- C- D+ E-).
Step 10: Ejector cylinder moves forward as shown in Figure 21. When
gripper cylinder is at D+ position and Y9 solenoid is actuated at the same
time. Air goes to compressor from port 1 to port 4 (V5) and to blank end
area of ejector cylinder. Therefore, ejector cylinder is ejecting the part from
collet, (A+ B- C- D+ E+).
6) When the pusher is retracted and at the same time, loader is also
retracted. After 01 s, CR4 is activated which means Y4 solenoid in
5/2 DCV2 is actuated (B-).
7) When loader is retracted, the required machining work begin. After
completing machining work, ejector cylinder is activated. Then,
CR7 is activated which means Y7 solenoid in 5/2 DCV4 is actuated
(D+). Before ejecting the part, the gripper cylinder should unclamp
the part from collet.
8) After unclamping the part, the ejector cylinder is extended and
eject the part from collet. Then, the ejector cylinder is retracting
and one cycle is completed. Then, CR9 is activated which means
Y9 solenoid in 5/2 DCV5 is actuated (E+). The next round has one
timer set at 01 s. After 01 s, CR10 is activated which means Y10
solenoid in 5/2 DCV 5 is actuated (E-). This complete one cycle of
automation by using PLC programming.
1
𝑃ℎ𝑐𝑜𝑛𝑣 = 𝑠 (1)
𝐿𝑆
𝐿𝐶𝑜𝑛𝑣 = 𝑃ℎ × 𝑤 (2)
𝑇𝐶+𝐿
𝐶𝑝 (𝐶𝑜𝑛𝑣) = 𝐿𝑆
(3)
b) Chamfering
c) Fillet
d) Undercut/Grooving
1
𝑃ℎ𝑁𝐶 = 𝑠 (4)
𝐿𝑆
𝐿𝑁𝐶 = ×𝑤 (5)
𝑃ℎ
𝑇𝐶+𝐿
𝐶𝑝 (𝑁𝐶) = 𝐿𝑆
(6)
𝑇𝐶
𝐵𝑃 = (𝐶 (7)
𝑝(𝐶𝑜𝑛𝑣) −𝐶𝑝(𝑁𝐶)
1
𝑃ℎ𝑆𝑃𝑀 = 𝑠 (9)
𝐿𝑆
𝐿𝑆𝑃𝑀 = 𝑃ℎ × 𝑤 (10)
Simulation of Low Cost Automation and Life Cycle Cost … 285
𝑇𝐶+𝐿
𝐶𝑝 (𝑆𝑃𝑀) = (11)
𝐿𝑆
𝑇𝐶
𝐵𝑃 = (12)
(𝐶𝑝(𝑁𝐶) −𝐶𝑝(𝑆𝑃𝑀)
From Table 1, it can be seen that the first machine is not suitable, when
cost and quality are the only factor. The second machine also has slow
production rate and low labour cost. But number of quality of the product
Simulation of Low Cost Automation and Life Cycle Cost … 287
is less and electricity charge is higher than other machines. This machine’s
total cost of savings is $13.6. When quality is the only factor, the second
machine is not suitable.
The third machine produces the parts at a higher production rate and a
lower labour cost. The number of quality product is high and electric
charge is minimum as compared to second machine. The savings are again
offset, this time by the tool cost. If the production run were greater, this
method would have been the least expensive. Our requirements are high
volume production rate, the number of quality product, less rejection and
reduction of cost of production rate. In this machine skilled labour is not
required. So, the third machine is selected for production (Graphically
shown in Figure 24).
CONCLUSION
REFERENCES
Chapter 9
ABSTRACT
In sheet metal incremental forming, the sheet blank edge rigidly
clamped deforms using a spherical tool, assisted by a CNC controlled
incremental tool movement. Limitation in Single Point Incremental
Forming (SPIF) is that it is limited to the production of parts with wall
angle between 70-80 degree. The formability of a metal sheet differs
according to the direction of the tool movement because of the plane
anisotropy. The formability of a sheet metal is better accounted for using
Forming Limit Diagrams. A considerable amount of knowledge about the
forming forces is required for the preservation of tooling and machinery.
A dynamical model is to be developed and using numerical techniques
and FEM simulations, an approximated value of axial force will be
computed. Based on these results a decision is to be made regarding a
servo-controlled motor for the robot used for base rotation.
Corresponding Author Email: murtazabohra17@gmail.com.
292 Bohra Murtaza and Y. V. D. Rao
1. INTRODUCTION
In SPIF, the sheet blank’s edges are usually rigidly clamped and the
sheet is then deformed usually by a hemispherical tool which moves as per
the numerical computer program feed, following specific contours which
result in the required shape of the sheet blank. Incremental sheet forming
can be performed on any universal milling machine which is having at
least 3-axis CNC control system. Thus, the basic elements of incremental
forming processes are the material to be formed, a sheet blank holder
clamping the blank, universal forming tool and the forming machine with
the CNC control.
3.1 Clamping
Clamping is done to fix the metal sheet onto the blank holder. There
are two types of clamping methods used. In the first one, Solid or Rigid
clamping the sheet is fixed on the blank holder using nuts and bolts. And in
the second case, Ball clamping the sheet is clamped or fixed by forcing
free rotating balls along its perimeter. The ball clamping method is more
effective than solid clamping as it provides a relatively freer flow of the
sheet during the ISF process (Khare and Pandagale, 2014). There is a lot of
scope for research in ball clamping method.
Experiments show that spherical tool gives more depth and provides
better surface roughness when compared to a hemispherical tool, this is
because of its smooth rolling action over the blank surface without
rubbing. Kim and Park, 2002 found that the formability of SPIF process is
improved when a ball tool of a particular size is used with a small feed rate
and a little friction. The feed rate is the speed the forming tool with which
moves around the mill bed. Decreasing the feed rate also increases the time
of the process, therefore it should be wisely decided according to the shape
of the part and surface finish required. The formability also differs
according to the direction of the tool movement because of the plane
anisotropy.
Effect of Lubrication
Lubrication is very important in incremental sheet forming. It is used
to decrease the wear of tool and sheet. Also, lubrication helps in getting a
Estimation of Axial Force in Incremental Sheet Metal Forming 297
better surface finish and helps in dissipating heat generated at the contact
area between tool and metal sheet.
It is observed that spherical tool without lubrication gives better
formability because the friction at the tool and sheet interface increases the
tool pressure lowering state of stress in the sheet (Khare and Pandagale,
2014).
FLD remains same. Figure 5 shows that the FLD for SPIF lies above the
FLD obtained for conventional forming. This shows the higher formability
of SPIF compare to conventional forming. The main reason for the higher
formability of SPIF is because the strain here is localized near the forming
tool.
CONCLUSION
REFERENCES
Aerens R, Eyckens P, Bael A Van and Duflou JR. Force prediction for
single point incremental forming deduced from experimental and FEM
observations. 2009, Int J Adv Manuf Technol doi: 10.1007/s00170-
009-2160-2.
Bambach M, Hirt G, and Ames J, Modelling of optimization strategies in
the incremental CNC sheet metal forming process. In Materials
processing and design: modeling, simulation and applications, 2004.
Proceedings of the Eighth International Conference on Numerical
methods in industrial forming processes (Numiform 2004), AIP
Conference Proceedings, Vol. 712, Columbus, Ohio, USA, 13–17 June
2004, pp. 1969–1974 (American Institute of Physics, Melville, New
York).
Crina, Radu, Determination of the Maximum Forming Angle of some
Carbon Steel Metal Sheets, 2011, Journal of Engineering Studies and
Research – Volume 17 No. 3.
Ghosh AK and Hecker SS, Stretching limits in sheet metals: in-plane
versus out-of-plane deformation. 1974, Metall. Trans. A 5A (1974)
1607– 1616.
Goodwin GM, Application of strain analysis to sheet metal forming
problems. 1968, Metall. Ital. 60 (1968) 767–771.
302 Bohra Murtaza and Y. V. D. Rao
C
B
carbon nanotube, 177, 179, 183
barriers, 58, 92, 102, 103, 108, 109
carboxylic acids, 245
beam-column joint, vi, viii, 165, 166, 167,
casting, 198, 200, 202, 206, 269
168, 169, 171, 173, 174
ceramic, viii, 175, 186, 187, 188, 190, 191,
beams, 1, 166, 174, 226
192, 193, 198, 199, 205, 206, 207, 208,
bedding, 209
306 Index
211, 212, 214, 216, 219, 220, 221, 223, crystals, 17, 19, 20, 22, 23, 29, 30, 33, 34,
224, 225, 227, 229, 231, 232, 233, 234, 35, 40, 41, 42, 43, 45, 46, 47, 51, 52, 53,
235, 236, 237, 239, 240, 242, 248, 249, 54, 55, 115, 211, 215, 219, 247, 250
250, 251, 252, 254, 255, 256, 257, 258,
259, 260, 261, 262
D
ceramic bulk, vi, 175
ceramic materials, 186, 220, 240, 250, 251,
decomposition, 28, 30, 187, 205, 215, 216,
252
243
Chebyshev approximation theory, viii, 136
defect engineering, 18, 51, 52, 55
chute, 266, 267, 270, 274, 275, 279
defect formation, 20, 21, 40, 41, 42, 43, 44,
clay, 186, 187, 190, 191, 192, 193, 195,
46, 50, 51, 52, 55
198, 199, 200, 201, 202, 203, 204, 205,
defective structure, v, vii, 17, 18, 20, 21, 34,
206, 207, 214, 215, 216, 218, 224, 229,
35, 42
250, 253, 254, 256, 259, 261
defects, 19, 23, 35, 43, 50, 51, 52, 56, 219
CNC technology, 292
deformation, 4, 36, 37, 38, 39, 171, 194,
coal, 191, 220, 260
196, 197, 297, 301
coatings, 243, 252, 257
differential equations, 23, 35, 63
collet, 266, 267, 268, 269, 270, 276, 277,
diffusion, 17, 19, 22, 23, 26, 27, 36, 39, 40,
281
41, 42, 48, 49, 53, 56, 100, 115
computational algorithm, 17, 18, 23, 47
directional control valve, 264, 271
computer simulations, 69, 104
dislocation, 19, 20, 21, 22, 26, 31, 34, 35,
computer technology, 18, 47
36, 37, 38, 39, 40, 41, 43, 44, 46, 47, 48,
configuration, 6, 48, 50, 59, 64, 70, 80, 82,
51, 52, 54, 55
84, 86, 91, 93, 95, 105, 109, 110, 111,
displacement, viii, 2, 36, 37, 153, 159, 166,
128, 179
171, 210, 222
convergence, 73, 74, 98, 101, 102, 108
displacement isomorphism, viii, 153
cooling, 19, 20, 21, 22, 23, 30, 33, 34, 35,
distributed memory computer, 125
39, 40, 45, 46, 50, 54, 212, 223, 226, 233
distribution, 11, 20, 21, 22, 26, 30, 31, 32,
correlation function, 99, 105
33, 34, 35, 43, 44, 45, 64, 70, 72, 73, 76,
cost, ix, 263, 264, 265, 266, 281, 282, 283,
77, 81, 82, 84, 87, 89, 90, 94, 95, 100,
284, 285, 286, 287, 289, 292
104, 105, 180, 181, 198, 199, 204, 208,
cost reduction, 264, 265, 266
244, 245
Coulomb interaction, 47
distribution function, 20, 26, 30, 31
crystal growth, 19, 20, 32, 34, 39, 42, 43,
DNA, 57, 58
45, 52, 55
crystal structure, 42, 43, 210
crystalline, 33, 51, 53, 54, 209, 215, 228, E
248
crystallites, 198 ejector, 266, 268, 269, 270, 276, 277, 279,
crystallization, 21, 22, 23, 26, 33, 34, 35, 281
220, 243, 247, 251, 260 elastic deformation, 36, 40
Index 307
energy, 20, 38, 53, 58, 65, 69, 70, 72, 73, forming forces, 291, 300
74, 76, 77, 80, 81, 82, 83, 86, 87, 88, 89, forming limit diagrams, 291, 292, 297, 302
90, 91, 92, 93, 95, 96, 97, 99, 100, 101, four precision point synthesis, 138
102, 103, 104, 107, 108, 109, 113, 115, fracture toughness, 240, 249
118, 120, 122, 124, 125, 127, 128, 130, free energy, 48, 58, 96, 99, 100, 101, 102,
132, 133, 167, 178, 179, 189, 190, 220, 108, 109, 114, 115, 128, 130, 248
228 freedom, 1, 59, 60, 61, 63, 65, 66, 67, 68,
energy consumption, 220 71, 80, 81, 83, 92, 109, 131, 154, 156
energy density, 87, 90, 93 Freudenstein equation, 136, 138, 141
energy supply, 190 Freudenstein-Chebyshev approximation
engineering, vii, ix, 18, 41, 51, 52, 55, 165, method, 135
182, 190, 191, 250, 251, 252, 260, 289 friction, 168, 296, 297, 298
entropy, 47, 48, 49, 50, 53, 71, 76, 91, 94, fuel cell, 185
128 function generation, 135, 136, 137, 148,
epicyclic gear train, v, viii, 153, 154, 155, 149, 150, 151
156, 163, 164 functional representation, 154
expansion, 4, 93, 176, 208, 209, 210, 215,
216, 219, 229, 233, 234, 235, 238
G
I M
incremental sheet forming, 292, 296, 302 acromolecular systems, 57, 58, 59, 109, 113
ions, 204, 243, 244, 245, 246 macromolecules, viii, 117, 120
iron, 207, 209 magnetic characteristics, 180
ISF using hydraulic fluid, 294, 295 magnitude, 1, 4, 13, 22, 36, 38, 63, 73, 77,
isomerization, 117 92, 107, 112
isomers, 159 manufacturing, vii, ix, 14, 178, 185, 190,
isomorphism, 153, 159, 163 191, 192, 220, 256, 264, 266, 269, 270,
isotope, 53 282, 283, 284, 288, 292
Markov chain, 116
materials, vii, viii, 18, 19, 20, 42, 51, 52, 53,
J
167, 175, 177, 180, 181, 182, 183, 184,
185, 186, 187, 189, 190, 192, 206, 208,
joints, 154, 155, 165, 166, 167, 174
212, 216, 218, 220, 229, 231, 236, 240,
247, 249, 250, 251, 252, 254, 255, 260,
K 261, 281
materials science, viii, 18, 51, 175, 249, 260
kinematics synthesis, 136 matrix, 3, 26, 36, 37, 110, 111, 140, 153,
kinetic equations, 23 157, 158, 159, 161, 162, 219, 221, 290
kinetic model, 31, 39, 46, 54 MD simulation, 57, 58, 59, 63, 64, 70, 71,
kinetics, 23, 26, 28, 32, 36, 42, 52, 108, 118, 72, 73, 75, 76, 77, 88, 89, 90, 91, 95, 99,
133 102, 103, 104, 106, 108, 110, 112, 113
MD technique, 58
mechanical properties, 176, 235, 247, 248,
L
254, 261, 302
landscape, 58, 100, 104, 108, 109, 118, 127, melting, 21, 48, 205, 208, 210, 216, 217,
128, 132, 133 219, 221, 227, 229
lattice parameters, 37 microdefects, 18, 19, 20, 22, 23, 26, 30, 35,
least-square method, 135 41, 43, 44, 46, 54, 55
life cycle cost, vi, ix, 263, 264, 265, 281, model, 1, 3, 6, 8, 17, 18, 19, 22, 23, 26, 31,
287, 288 32, 33, 34, 35, 36, 38, 39, 40, 41, 42, 46,
linkage adjacency matrix, 157, 161 47, 48, 49, 50, 51, 54, 55, 84, 108, 111,
liquids, 31, 117, 126, 128, 204 112, 113, 115, 117, 122, 123, 125, 128,
loader, 266, 267, 270, 275, 276, 281 129, 131, 132, 133, 135, 166, 167, 168,
low cost automation, vi, ix, 263, 264, 265, 169, 170, 171, 172, 173, 174, 264, 291,
266, 288, 289 300, 301
low temperatures, 50, 53, 70, 213 modulus, 5, 26, 38, 194, 201, 219, 240, 246
moisture content, 196, 198, 199, 200, 201,
203
Index 309
molecular dynamics (MD), v, viii, 57, 58, physical and mechanical properties, 261
59, 61, 63, 64, 65, 67, 68, 69, 70, 71, 72, physical properties, 31, 51, 204, 228, 242
73, 75, 76, 77, 79, 81, 83, 85, 87, 88, 89, physicochemical properties, 243
90, 91, 93, 95, 96, 97, 99, 101, 102, 103, plastic deformation, 292
104, 105, 106, 107, 108, 109, 110, 111, plasticity, 167, 187, 194, 195, 196, 197,
112, 113, 114, 115, 116, 117, 118, 119, 198, 203, 204, 205, 208, 215
120, 121, 122, 123, 124, 125, 126, 127, point defects, 19, 21, 47, 48, 50, 51, 53, 54
128, 129, 130, 131, 132, 133 polymerization, 242, 243, 249
molecular nanotechnology, 178 polymers, 113, 185, 302
Monte Carlo method, 120, 124 polypeptide, 125, 131
porosity, 176, 213, 221, 223, 225, 235, 237,
245, 250
N
precipitation, 19, 20, 22, 26, 31, 33, 34, 35,
36, 40, 42, 46, 47, 50, 51, 54, 209, 236
nano-Al2O3, 175, 183
precision points, viii, 135, 136, 145, 146,
nanoceramics, viii, 175, 184, 185, 187, 188,
148, 150
189, 190, 259, 261
probability, 70, 72, 73, 74, 77, 83, 84, 86,
nanomaterials, vi, viii, 175, 176, 177, 181,
87, 89, 90, 92, 93, 94, 95, 96, 102, 103,
182, 183, 184, 187, 249, 251
104, 105, 106, 117
nanometers, 177, 180, 184, 188
probability density function, 94, 95
nanoparticles, 175, 176, 180, 182, 184, 187,
probability distribution, 70, 72, 77, 84, 89,
188, 189, 250
93, 96, 102, 103, 104
nano-SiO2, 175, 183
production costs, 282, 283, 284
nanostructured materials, 180, 181
productivity, 263, 264, 265, 266, 281, 285,
nanostructures, 179, 182, 184, 188
286
nanotechnology, 175, 177, 178, 249, 251,
programmable logic controller, ix, 264, 279
253, 259, 260
programming, viii, 47, 51, 279, 281, 287
programming languages, 47
O properties of ceramics, 189, 261
protein(s), 57, 58, 68, 76, 78, 84, 85, 103,
optimization, 3, 15, 23, 92, 114, 120, 136, 106, 109, 112, 113, 114, 115, 119, 122,
137, 142, 143, 144, 145, 146, 148, 150, 123, 124, 126, 127, 128, 129, 130, 131,
301 132, 133
oxygen, 22, 23, 24, 26, 30, 31, 33, 34, 40, protein folding, 58, 76, 78, 85, 103, 122,
54, 216, 228 128, 130, 131, 133
pusher, 266, 267, 268, 270, 275, 276, 279,
281
P
quartz, 192, 193, 207, 208, 209, 210, 211, 83, 87, 88, 89, 90, 91, 95, 96, 99, 100,
212, 214, 215, 216, 219, 248, 250, 254 101, 102, 103, 104, 105, 106, 108, 112,
115, 116, 118, 119, 121, 122, 123, 126,
127, 128, 129, 132, 133, 165, 167, 263,
R
264, 273, 274, 275, 276, 277, 278, 287,
288, 289, 301
random walk, 70, 73, 74, 75, 119, 132
single crystals, 20, 21, 22, 29, 31, 34, 39,
raw materials, 175, 190, 191, 193, 207, 218,
41, 43, 44, 46, 47, 48, 54, 55
220, 223, 224, 236, 249, 256, 260
single point incremental forming, 291, 292,
recombination, 21, 35, 36, 47, 48, 49, 50,
293, 294, 301, 302
53, 54
sintering, 176, 187, 188, 189, 220, 231, 247,
reduction of time, 266
248, 255, 258, 262
rheological and electromagnetic properties,
SiO2, 34, 175, 176, 182, 183, 185, 187, 209,
177
210, 213, 215, 218, 219, 220, 224, 230,
rotation graph, 155, 156, 157, 160, 163
236, 244, 247, 248, 252, 257, 258
rotational analysis, 159
sol-gel, 187, 188, 189, 242, 243, 244, 245,
rotational isomorphism, viii, 153, 163
248, 252, 255, 259, 260
SOL-GEL techniques and applications, 242
S solid state, viii, 18, 31, 34, 51, 55
special purpose machine, vi, ix, 263, 264,
seismic analysis, 166 266, 269, 284
semiconductor(s), 17, 19, 35, 40, 42, 51, 52, special purpose turning machine, 266, 269,
54, 55, 182, 185 286, 287
servo controlled, 292, 301 spindle collet, 267, 268
shear, 26, 37, 38, 166, 167, 169, 172, 174, strain gauges, vii, 1, 2, 3, 4, 5
194 strengthened beam-column joint, 166
sheet metal forming, vi, ix, 291, 295, 297,
299, 301
shrinkage, 176, 190, 198, 199, 200, 201, T
202, 205, 206, 208, 215, 217, 220, 238,
techniques, vii, 64, 69, 90, 125, 126, 137,
245, 246, 250, 253
187, 191, 242, 249, 287, 291, 295, 300,
sigmoidal function generation, v, 135, 137,
301
148
technology(ies), vii, 2, 18, 20, 41, 42, 51,
silica, 182, 208, 209, 210, 220, 233, 235,
52, 55, 178, 190, 252, 259, 265, 292
253, 259, 261
thermal expansion, 176, 208, 209, 210, 215,
silicon, vii, 17, 19, 20, 21, 22, 23, 24, 26,
219, 229, 233, 234
29, 30, 31, 33, 34, 37, 39, 40, 41, 42, 43,
transducer, vii, 1, 2, 4, 5, 7, 9, 10, 12, 14
44, 45, 46, 47, 50, 51, 52, 53, 54, 55, 56,
transfer vertex, v, viii, 153, 156, 160, 161,
228
163
simulation(s), v, vi, viii, ix, 9, 18, 57, 58,
tri-axial loadings, 1
59, 63, 69, 70, 71, 72, 73, 74, 75, 76, 78,
Index 311
W Yang-Mills, 32