Mathematics Applied To Engineering Modelling and Social Issues Frank T. Smith Download PDF
Mathematics Applied To Engineering Modelling and Social Issues Frank T. Smith Download PDF
Mathematics Applied To Engineering Modelling and Social Issues Frank T. Smith Download PDF
com
https://textbookfull.com/product/mathematics-
applied-to-engineering-modelling-and-social-
issues-frank-t-smith/
textbookfull
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
https://textbookfull.com/product/applied-stochastic-modelling-
second-edition-byron-j-t-morgan/
https://textbookfull.com/product/mathematics-applied-to-
engineering-and-management-1st-edition-mangey-ram-editor/
https://textbookfull.com/product/applied-engineering-
mathematics-1st-edition-brian-vick/
https://textbookfull.com/product/functional-analysis-in-applied-
mathematics-and-engineering-first-edition-pedersen/
Mathematics of the Financial Markets Financial
Instruments and Derivatives Modelling Valuation and
Risk Issues 1st Edition Alain Ruttiens
https://textbookfull.com/product/mathematics-of-the-financial-
markets-financial-instruments-and-derivatives-modelling-
valuation-and-risk-issues-1st-edition-alain-ruttiens/
https://textbookfull.com/product/mathematical-methods-in-
engineering-and-applied-sciences-mathematics-and-its-
applications-1st-edition-hemen-dutta-editor/
https://textbookfull.com/product/applied-mathematics-in-
hydrogeology-lee/
https://textbookfull.com/product/information-technology-and-
applied-mathematics-peeyush-chandra/
https://textbookfull.com/product/graduate-medical-education-
issues-and-options-1st-edition-frank-c-wilson/
Studies in Systems, Decision and Control 200
2 0 0 t h Vo l u m e o f S S D C · 2 0 0 t h Vo l u m e o f S S D C · 2 0 0 t h Vo l u m e o f S S D C
Frank T. Smith
Hemen Dutta
John N. Mordeson Editors
Mathematics
Applied to
Engineering,
Modelling, and
Social Issues
Studies in Systems, Decision and Control
Volume 200
Series editor
Janusz Kacprzyk, Systems Research Institute, Polish Academy of Sciences,
Warsaw, Poland
The series “Studies in Systems, Decision and Control” (SSDC) covers both new
developments and advances, as well as the state of the art, in the various areas of
broadly perceived systems, decision making and control–quickly, up to date and
with a high quality. The intent is to cover the theory, applications, and perspectives
on the state of the art and future developments relevant to systems, decision
making, control, complex processes and related areas, as embedded in the fields of
engineering, computer science, physics, economics, social and life sciences, as well
as the paradigms and methodologies behind them. The series contains monographs,
textbooks, lecture notes and edited volumes in systems, decision making and
control spanning the areas of Cyber-Physical Systems, Autonomous Systems,
Sensor Networks, Control Systems, Energy Systems, Automotive Systems,
Biological Systems, Vehicular Networking and Connected Vehicles, Aerospace
Systems, Automation, Manufacturing, Smart Grids, Nonlinear Systems, Power
Systems, Robotics, Social Systems, Economic Systems and other. Of particular
value to both the contributors and the readership are the short publication timeframe
and the world-wide distribution and exposure which enable both a wide and rapid
dissemination of research output.
** Indexing: The books of this series are submitted to ISI, SCOPUS, DBLP,
Ulrichs, MathSciNet, Current Mathematical Publications, Mathematical Reviews,
Zentralblatt Math: MetaPress and Springerlink.
John N. Mordeson
Editors
Mathematics Applied
to Engineering, Modelling,
and Social Issues
123
Editors
Frank T. Smith Hemen Dutta
Department of Mathematics Department of Mathematics
University College London Gauhati University
London, UK Guwahati, Assam, India
John N. Mordeson
Department of Mathematics
Center for Mathematics of Uncertainty
Creighton University
Omaha, NE, USA
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
v
vi Preface
The Chap. “A Survey on the Melnikov Theory for Implicit Ordinary Differential
Equations with Applications to RLC Circuits” deals with the development of the
Melnikov theory in studying implicit ordinary differential equations with small
amplitude perturbations, and in particular, the persistence of orbits connecting sin-
gularities in finite time provided that certain Melnikov like conditions hold. Further, it
considered the achievements on reversible implicit ordinary differential equations.
Some applications to nonlinear systems of RLC circuits are also presented.
The Chap. “Numerical Solution of Space-Time-Fractional Reaction-Diffusion
Equations via the Caputo and Riesz Derivatives” considered the numerical solution
of space-time-fractional reaction-diffusion problems used to model complex phe-
nomena that are governed by dynamic of anomalous diffusion. The time- and
space-fractional reaction-diffusion equation is modelled by replacing the first-order
derivative in time and the second-order derivative in space, respectively, with the
Caputo and Riesz operators. It proposed some numerical approximation schemes
such as the matrix method, average central difference operator and L2 method. It
applied the Laplace transform technique in time and the Fourier transform method
in space to give a general two-dimensional representation of the analytical solution
in terms of the Mittag-Leffler function. The proposed methods are tested for
applicability on a range of practical problems.
In the Chap. “An Extended Langhaar’s Solution for Two-Dimensional Entry
Microchannel Flows with High-Order Slip”, Langhaar’s assumptions for the
entrance region of two-dimensional micro-channels (micro-tube, slit channel and
concentric annular micro-channel) have been implemented using high-order slip
models. Different slip models have been used, and velocity profile, entrance length
and apparent friction factor have been obtained in an integral form. The advances in
micro-fabrication technology have brought numerous applications to the field of
micro-scale science, and engineering and micro-channels are inseparable part of
microfluidic technology.
The Chap. “Dynamics of Solitons in High-Order Nonlinear Schrödinger Equations
in Fiber Optics” aims to construct kink, bright and dark solitons of a generalized
higher-order nonlinear Schrödinger equation in a cubic-quintic non-Kerr medium by
applying a modified extended mapping method. It also presented the formation
conditions on solitary wave parameters in which kink, dark and bright solitons can
exist, and graphically illustrated the collision of the constructed soliton solutions that
help realizing the physical phenomena of nonlinear Schrödinger equation. Further, it
outlined descriptions of various issues on integrability. The stability of the model in
normal dispersion and anomalous regime are discussed by using the modulation
instability analysis.
The Chap. “MHD Mass Transfer Flow Past an Impulsively Started Semi-Infinite
Vertical Plate with Soret Effect and Ramped Wall Temperature” presents an exact
solution to the problem of a hydromagnetic natural convective mass transfer flow of
an incompressible viscous electrically conducting non-Gray optically thin fluid past
an impulsively started semi-infinite vertical plate with ramped wall temperature in
Preface vii
towards the successful completion of this book project. The editors are indebted to
several well-wishers, colleagues, editors and supporting staff at Springer for timely
and efficient cooperation which helped in executing this project smoothly.
xi
xii Contents
xiii
xiv Contributors
H. Lombera Rodríguez
Centro de Informática Industrial, Universidad de las Ciencias Informáticas,
19370 La Habana, Cuba
e-mail: [email protected]
J. I. Tello (B)
Depto Matemática Aplicada a las TIC. ETSIS Sistemas Informáticos, UPM,
28037 Madrid, Spain
e-mail: [email protected]
the relationship between friction, wear and lubrication, especially applied to journal
bearings. Gustave Adolph Hirn (1815–1890), accomplished in 1847 the first exper-
iments on hydrodynamic lubrication and rediscovered the laws of Amontons and of
Coulomb. In 1879, Robert Henry Thurston (1839–1903), published the results of his
study on friction and lubrication. He showed that, with increasing speed, the friction
coefficient of a lubricated bearing diminishes below its static value, passes through
a minimum and then increases. He also specified that the speed corresponding to the
minimum of friction depends on the load applied to the bearing [48].
In 1883, Nikolai Pavlovich Petrov (1836–1920) introduced the results of his stud-
ies and tests on lubricated bearings. He proved that, among the physical characteris-
tics of an oil, the viscosity has a preponderant role in bearing friction. He stipulated
that a fluid film totally separates the surfaces of both shaft and bearing, and that a
constant pressure should be produced in this film. Petrov, also looked through the
work of Hirn and reused the term of mediate friction to characterize hydrodynamic
lubrication [48]. In 1885, a remarkable discovery was the existence of hydrodynamic
pressure in the lubricant film of a bearing by Beauchamp Tower (1845–1904), which
served as a basis for accomplishing the theory of lubrication. Fortunately, Tower’s
discovery results provided experimental confirmation to Reynolds, who was working
on a hydrodynamic theory of lubrication at that time. The result of this was a theory
of hydrodynamic lubrication published in the Proceedings of the Royal Society by
Reynolds; see [82]. In that early work, Reynolds proposed the equation that at present
is named after him and provided the first analytical proof that a viscous liquid can
physically separate two sliding surfaces by hydrodynamic pressure, resulting in low
friction and theoretically zero wear [90]. That work represents the seminal paper
on Lubrication Theory and in fact, most of mathematical models of hydrodynamic
lubrication processes between solid surfaces have the Reynolds Equation (RE) as
their key point. A rigorous approach for the deduction of the classical linear RE from
Navier–Stokes may be found in [4].
It was not until the beginning of the 20th century that Reynolds theory on hydro-
dynamic lubrication was used for calculating thrust and journal bearings. In 1902,
Richard Stribeck confirmed the hydrodynamic effects and performed the original
research into the limits of hydrodynamic lubrication. He proposed the relationship
between friction, load, speed and viscosity that is still used today to present the var-
ious types of lubrication. In most cases, the friction and lubrication relationship is
characterized with basis on μv F
(oil viscosity × sliding velocity/normal load) factor,
in a diagram called Stribeck curve. This diagram summarizes the limits of hydrody-
namic lubrication; see Fig. 1.
Three zones can be identified, each one corresponding to a type of lubrication
depending on the level of pressure established in the contact. For low pressure (0.1
to 50 MPa), zone 1 corresponds to boundary lubrication; surface separation is ensured
by lubricant molecules attached to the surfaces; see Fig. 2a. This type of lubrication is
related to the physico-chemistry of surfaces and of lubricants, for low and moderate
speeds and for relatively low loads. In zone 2, the hydrodynamic effect described by
RE takes some importance and tends to separate the areas still in contact over a part
of their asperities; this type of lubrication is the mixed lubrication; see Fig. 2b. Zone
On the Reynolds Equation and the Load Problem in Lubrication … 3
1 2 3
friction
0 v
F
coefficient is linearly ascending due to fluid film lubrication. When load increases or
oil viscosity and/or velocity decreases, the μv F
factor falls. It means that the sliding
velocity v and the oil viscosity μ are unable to generate sufficient oil-film pressure p
to support the entire load F [58]. Then, the fluid film becomes thinner and, therefore,
friction coefficient decreases up to a minimum value. Note that for even smaller val-
ues of μvF
, fluid film thickness is further reduced, and contact appears. Then, friction
coefficient increases as the μv F
factor decreases [69]. Such rise in friction coefficient
is also related to oil viscosity increase in some regions at contact area under high
contact pressure. These phenomena characterize the mixed lubrication regime. Addi-
tional reduction in μv F
factor makes contact stronger. Film thickness becomes smaller
than the height of surface asperities and then boundary lubrication regime will occur.
Due to this behaviour, the Stribeck curve is also represented with the film thickness
along the horizontal axis [48].
In 1904, Arnold Johannes Wilhelm Sommerfeld used a change of variables and
succeeded in obtaining an analytical solution of the RE for infinitely long journal
bearings ( ∂∂xp = 0), where p stands for the pressure and x, the geometrical coordinate,
is in the direction of the journal axis [48]. Nevertheless, the used boundary conditions
did not correspond to the physical reality, as they did not take into account the film
rupture in the bearing. In addition, the pressure distribution obtained was negative
in the divergent zone of the film. In 1914, Ludwig Karl Friedrich Gümbel suggested
that only the positive part of the pressure distribution should be included for the
calculation of the bearing load, omitting the negative part. The load calculated was not
exact either. Herbert Walker Swift in 1932, and later W. Stieber in 1933 independently
presented boundary conditions for film exit, representing the reality in an improved
way [91, 95]. With their proposals, the fluid film breaks in its divergent zone along
the boundary, where the pressure has the value of the saturated pressure and its
gradient is zero. Such conditions agree with the continuity of flow at the film exit
and are named the boundary conditions of Reynolds. They are universally used for
calculating bearings with constant loads [48]. This model considers as interface
conditions:
∂p
pc = = 0, (1)
∂n
where pc stands for the cavitation pressure and n stands for the unitary normal
vector to the free boundary. In 1941, a numerical method for solving the RE with
such boundary conditions was proposed in Christopherson [25].
In 1953, Fred William Ocvirk proposed to neglect circumferential pressure gradi-
ent compared to the axial one in the RE. This way he was presenting his approximate
method for short bearing [73]. The solution is analytical and uses Gümbel’s boundary
conditions. The results are almost exact for journal bearings having an L/D ratio
(bearing length over diameter) smaller than 0.25. Calculations are in consequence
considerably simplified [48].
A type of slider, with steps, consisting of two parallel parts, but shifted, was
described by John William Strutt, Lord Rayleigh. He showed that this type of slider
exhibits a load carrying capacity for a given minimum film thickness greater than for
On the Reynolds Equation and the Load Problem in Lubrication … 5
any known slider type [48]. In 1917, Lord Rayleigh was the first one in calculating
the load and the friction torque of a hydrostatic thrust bearing.
The review of all existing papers since the beginning of last century represents
a considerable effort that goes beyond the aims of this chapter. We presented here
only a brief review of major achieved progresses in the beginning of hydrodynamic
lubrication, that will allow us to succeed “standing on the shoulders of giants”.
Misalignment
In the case where misalignment is not allowed, the mathematical model assumes
that the clearance normalized film thickness, h̄, only depends on the circumferential
coordinate and is expressed as:
where ρ̄ ∈ [0, 1) stands for the normalized shaft eccentricity, α is the shaft angular
coordinate and θ represents a point on the external circumference. The assumption
posed by Eq. (2) is equivalent to the supposition that both shaft and bearing axes
are perfectly parallel to each other, and that the eccentricity ρ̄ does not depend on
the axial coordinate. Thus, this expression restricts the physical simulating capacity
of the study since in reality it becomes impossible to fully avoid radial and angular
misalignments [51]. The most common causes of misalignment are elastic and ther-
mal deflections of the shaft and bearing misalignment, as a result of assembly errors.
Large misalignment can decrease the bearing clearance and its load capacity. It can
increase the temperature and has the potential to reduce the operating velocity thresh-
old [51]. In addition, misalignment and residual unbalance are the typical causes for
rotor vibration. Both excitations are responsible for most common machine dynamic
problems happening in the field.
One of the first documented researches on journal bearing misalignment is re-
ported by McKee and McKee [70], who experimentally observed that measured
peak pressures move from the bearing mid-plane towards the bearing ends when the
journal is subjected to misalignment. Same result was found by Bouyer and Fillon
[18] in an experimental analysis of misalignment effects on hydrodynamic plain
journal bearing performance. They experimentally studied the hydrodynamic plain
journal bearing submitted to a misalignment torque. The misalignment caused more
significant changes in bearing performance when the rotational speed or load was
low [18]. Piggott showed that a 40% reduction in bearing load capacity was induced
by a 0.0002 rad misalignment. These observations clearly revealed the importance of
misalignment in bearing performance [78]. Subsequently, Dubois et al. [41] showed
that the pressure distribution of a misaligned bearing was not symmetric, and re-
ported that the maximum pressure was located as well at the bearing ends. They
6 H. Lombera Rodríguez and J. I. Tello
while in the latter the pressure is taken to be a constant [12]. The main difference
between models comes from the way to obtain the free boundary that separates
lubricated and cavitated areas. In Sects. 7 and 10 we give details on the derivation of
both cavitation models.
Several papers have used the theory of variational inequalities taking advantage
that the pressure in the full filled area is greater than the saturation pressure. In
fact, the idea was reinforced when Cryer justified the work of Christopherson [25],
associating that study to an obstacle problem [33]. This is known as the Reynolds
cavitation model [12]. In 1975, Rohde and McAllister presented a variational for-
mulation for hydrodynamic lubrication, from which the associated free boundary
problem arose naturally. The Finite Difference Method (FDM) and the FEM were
discussed as strategies for obtaining approximate solutions [84]. In fact, due to the
nature of the Reynolds cavitation model and easy computational approach, it has
been used in a large list of mathematical works; see [22, 26, 27, 34, 66, 83] for
instance. In general, for the numerical resolution of this model, techniques based
on the FEM have been widely used. The discrete problem has been solved by the
classical Gauss-Seidel method or a point-overrelaxation method, including both a
projection technique to consider cavitation; see [21, 34] for instance. In this work,
we also use the Reynolds cavitation model including a FEM discretization. Never-
theless, we propose to solve the system of linear equations by minimizing a convex
functional, using a Preconditioned Conjugate Gradient Method (PCG) with both pro-
jection and restarting strategies. The choice obeys two major reasons: the fact that
we solve a convex functional and that matrices resulting from the discretization of
Partial Differential Equations (PDE) [e.g. FEM or Finite Volume Method (FVM)], in
addition to be sparse are usually ill-conditioned, for which preconditioning is widely
recommended.
Another model to describe cavitation is the Elrod–Adams model [47]. In that
work, the authors introduce the hypothesis that the cavitation region is a fluid-air
mixture and an additional unknown ϑ appears (the saturation of fluid in the mixture1 ).
This model, which still relies on the RE has been widely used in Tribology [68].
Unlike some other models, such as the Reynolds cavitation model, it does allow the
starvation phenomena to take place. Its interest also relies on the evidence that it
is a mass-preserving model. In [5, 45] comparisons for journal bearings are made,
between their operating parameters computed by the Reynolds and the Elrod–Adams
cavitation models.
Vijayaraghavan and Keith [97] analysed the effect of cavitation on the performance
of a line-grooved misaligned bearing for both flooded and starved inlet conditions.
They used the mass-conserving cavitation algorithm in their analysis. They took into
account the lubricant rupture and the reformation phenomena. One year later, they
showed that at the higher degrees of misalignment, the performance characteristics
of the bearing are significantly different from those for an aligned journal bearing
[98].
Numerical methods for solving the Elrod–Adams model for cavitation in different
devices and conditions were presented in [6, 13, 45, 46], among others. Similarly,
numerical experimentations of various schemes based both on stationary upwind
methods and pseudo-stationary techniques were conducted in [21]. These methods
are mainly based on the characteristics discretization for the non-linear convection
term and a duality method for the multivalued non-linear saturation-pressure relation,
posed by the Heaviside operator. Namely, they use an approach based on the Method
of Characteristics (MC) to discretize a total derivative in the final formulation. This
technique was also used in [42–44, 67] among others, and it is the strategy proposed
to solve our problem as well. Additionally, the first three used a Yosida regularization
for the Heaviside operator as in [14]. In contrast, in [67] it is used a regularization
of this function by a cubic interpolating Hermite polynomial that allowed to express
the solution of the direct problem as a minimum of a convex functional.
Most of the papers previously mentioned deal with imposed geometry in the asso-
ciated RE, i.e. the gap function h for the journal bearing is a given datum and the
unknown is the pressure p. In real engineering applications the position of the shaft,
that defines the gap function h, is unknown. So, Newton’s second law is introduced
to obtain that position. The problem consists in finding the pressure of the lubricant,
its concentration ϑ in the cavitation area and the shaft position. If misalignment is
considered, two more variables need to be found, which stand for the angular mis-
alignment projections. The problem is considered as an inverse problem where the
coefficient h depends on the unknown p.
Díaz and Tello [36] addressed such a problem, considering the simple case in
which the surfaces are two parallel planes, and assuming prescribed the total force
applied upon one of the surfaces. They provided some sufficient conditions on the
total force in order to solve the inverse problem. Ciuperca et al. [28] also studied
analytically the inverse problem for a more general geometry. Specifically, they
studied the asymptotic behaviour of the position in the evolution problem.
Furthermore, Ciuperca et al. [29] studied the inverse problem for journal bearings
using the Reynolds cavitation model. In that work the inner cylinder is parallel to
the exterior one and misalignment is not allowed. They proved the existence of shaft
equilibrium positions when the hydrodynamic force created by the pressure film
balances an external radial force. The authors proved the non-existence of contact
for any force, even for the case where the shape of the external surface presents
some rugosity. Additionally, Ciuperca and Tello considered the problem for both
cases, a rigid surface moving over a flat plane and the elastohydrodynamic problem;
see [31, 32] for instance. Similarly, Ciuperca, Jai and Tello studied the existence
of equilibrium positions for the load problem in Lubrication Theory. In their work,
considering the Elrod–Adams model, the balance of forces allows to obtain the
unknown position of the surfaces, defined with one degree of freedom [30].
On the Reynolds Equation and the Load Problem in Lubrication … 9
As for the numerical resolution of the inverse problem which entails the balance
between an imposed load on the device and the hydrodynamic load we can mention
the work of [42]. They developed a numerical scheme which combines fixed point
algorithms, the MC, duality techniques and finite element approximations. In [44]
the authors used an implicit Euler method to deal with the dynamical shaft problem
coupled with the fluid hydrodynamic problem. At each time step the resulting non-
linear system is solved by the Broyden method combined with the Armijo–Goldstein
criterion to choose a proper step length in the descent direction. Conversely, in [67]
the authors proposed a different approach to deal with the shaft model. It was based
on first solving the Elrod–Adams equation for a known position by minimizing
a convex and lower semi continuous (l.s.c) functional and then using an iterative
method to reach the equilibrium, namely a trust-region strategy. In general, there
are a lot of gradient based algorithms for continuous optimization that can be used
for solving problems like the one addressed in this work. They allow to find a local
minimum, but the optimized function needs to be continuous and differentiable.
Thus, their usefulness is limited due to such prerequisites. Line search and trust-
region approaches are two of the fundamental strategies in optimization algorithms
that must be mentioned; see [72] for a wide explanation on these approaches.
On the other hand, metaheuristics are a family of optimization techniques, which
have seen increasingly rapid development and application to numerous problems
in computer science and other related fields. Normally, they require the problem to
be partitioned into a set of components to look for the solution in an optimal com-
bination or permutation of them. One of the more recent, prominent and actively
developed metaheuristic is Ant Colony Optimization (ACO) which was inspired by
the ants’ foraging behaviour. It was originally introduced by Dorigo [38], to solve
discrete optimization problems where each decision variable is characterized by a set
of components; see [39, 92] for instance. Many successful implementations of the
ACO metaheuristic have been applied to a number of different discrete optimization
problems [63]. These applications mainly concern NP-hard combinatorial optimiza-
tion problems including problems in routing [49], assembly sequence planning [37],
bioinformatics [15] and many other areas.
ACO was initially designed to solve the Traveling Salesman Problem (TSP),
where a salesman must visit a list of cities exactly once, using the shortest possible
route. The cities and paths between them can be represented as a connected graph,
and the ants move from one city to another following the pheromone trails on the
edges. Let Ti j (t) be the trail intensity on edge (i, j) at time t. Then, each ant chooses
the next city to visit depending on the intensity of the associated trail. When the ants
have completed their city tours, the trail intensity is updated according to:
m
Ti j = Ti kj , (4)
k=1
where Ti kj is the pheromone quantity laid by the kth ant on edge (i, j), defined as:
1
, if edge (i, j) is in the trajectory of the kth ant,
Ti kj = Wk (5)
0, otherwise,
with Wk the tour length of the kth ant [38]. The transition probability Pikj from city
i to city j for the kth ant is defined as:
[Ti j ]a [ηi j ]b
Pikj = , (6)
l∈allowedk [Til ] [ηil ]
a b
where ηi j = 1/di j is called visibility and di j is the associated cost to travel from
city i to city j; a and b are parameters that control the relative importance of trail
versus cost, and allowedk is the set of allowed cities the kth ant can move to from city
i [50]. Genetic Algorithm, Simulated Annealing, Tabu Search and Particle Swarm
Optimization are other approaches we find in the literature to deal with combinatorial
optimization problems; see for instance [75, 79, 89, 100] for a detailed explanation
of them.
Since the emergence of these approaches as combinatorial optimization tools, at-
tempts have been made to use them for addressing continuous problems [87]. Now,
these metaheuristics that were originally developed for combinatorial optimization
are adapted to the continuous case. Examples include the Continuous Genetic Al-
gorithm [24], Enhanced Simulated Annealing [86], or Enhanced Continuous Tabu
Search [23]. There are also included some ant related methods. In this sense, Socha
and Dorigo [88], proposed one of the most popular and easy to implement ACO
algorithms for continuous domains, called Ant Colony Optimization for continuous
domain (ACOR ). It uses a solution archive as a form of pheromone model for the
derivation of a probability distribution over the search space. However, its use in
problems with many decision variables have some limitations, reported in Leguiza-
món and Coello [60]. Thus, Leguizamón and Coello [61] proposed an Alternative
Ant Colony Optimization for continuous domain (DACOR )2 which could be more
appropriate for large scale unconstrained continuous optimization problems. Later
on, Liao et al. [62] proposed an Incremental Ant Colony Optimization with Local
Search for continuous domain (IACOR -LS). This algorithm uses a growing solution
archive as an extra search diversification mechanism and a local search to intensify
the search. Subsequently, Liao et al. [64] proposed an ACO algorithm for continu-
ous optimization that combines algorithmic components from ACOR , DACOR and
IACOR -LS. They called it Unified Ant Colony Optimization for continuous domain
(UACOR ). It is unified, because from UACOR , we can instantiate the original ACOR ,
DACOR and IACOR -LS algorithms by using specific combinations of the available
algorithmic components and parameter settings. Since in our inverse problem, we
only deal with four decision variables and considering that ACOR has proven to be
an efficient, versatile and easy to implement tool for continuous optimization, we
propose its use in our work. However, we do deal with a large scale direct problem
and motivated by the inherent parallelism of the ACOR and possible computation
speed up we suggest an implementation of the algorithm with parallel regions for
time-consuming tasks, using Open Multi-Processing (OpenMP).
4 Other Topics
Moreover, we mention other topics which have also received attention on the subject
of misaligned journal bearings. Literature concerning the topics of thermohydrody-
namic and elastohydrodynamic lubrication can be found in [1, 19, 54, 59, 76, 77, 93,
96, 99]. Besides, lubrication is not the only way to decrease the effect of friction; the
materials used and the quality of polished surfaces are also of major concern. How-
ever, if surfaces are extremely polished, it is probable a contrary trend to decrease
load capacity. It has often been observed in engineering practice that there is a risk of
sudden seizure if the surface is too smooth. In this sense, it is commonly believed that
small asperities play a useful role as a reservoir for the lubricant between asperities
[90]. The effect of surface roughness on the performance characteristics of bearings
can be found in [53, 85]. In general, roughness is one of the challenges of the field;
see [7–11, 68] and references there in, to study its effects in journal bearings for
different scenarios. More recent results on this topic can be found in [93, 94].
The current research on journal bearings also includes the applications of non-
Newtonian fluids to improve performance of modern machines [56]. Such non-
Newtonian fluids have shown that the stress is not directly proportional to the shear
strain, and the formulation of the governing equations needs to be changed. Litera-
ture on non-Newtonian fluids includes the works of [1, 17, 35, 57, 74]. The general
characterization of the behaviour of a misaligned journal bearing considering all
mentioned factors and including the prediction of its final position is very complex.
For that reason most researches focus on specific topics.
In this section we depict the formulations of the fluid film thickness of a journal
bearing, for the parallel and the misaligned case. Actually, there are similar results
that can be found in the literature; see [48, 51] for instance. However, for making
this chapter self-contained we present their derivations in this section.
12 H. Lombera Rodríguez and J. I. Tello
Figure 3 depicts the cross section of a journal bearing. The inner cylinder, the shaft of
radius R, rotates in counter-clockwise direction at a constant velocity ω, about the X
axis. The film pressure generated by the moving surfaces, forces the lubricant through
a wedge shaped zone of thickness h, which varies according to the angle α. It is
assumed a coordinate system in which “y” represents the circumferential coordinate,
“z” is the coordinate across the fluid film and “x” depicts the journal bearing axial
dimension, orthogonal to the zy-plane. Let Ob and O j be the centres of the bearing
and shaft, respectively. The origin of coordinate “y” is located over the line segment
Ob O j , to place the minimum gap of the device at an angle = π. Moreover, the
reference z = 0 is taken on the bearing surface. The model characterizes the parallel
misalignment, where the shaft is allowed to move with two degrees of freedom.
Let M be a point over the bearing surface, represented by the angular coordinate
= (Ob A, Ob M), where Ob A and Ob M are line segments. The fluid film thickness
is given by:
h = Ob M − Ob M = Rb − Ob M . (7)
Applying the sine rule for the triangle O j M Ob we have:
O j M R R R ρ
= = = = , (8)
sin(O j Ob M ) sin(O j Ob M ) sin(π − ) sin() sin(α)
where
ρ R
sin(α) = sin() and Ob M = sin(Ob O j M ). (9)
R sin()
Φ
R
O
M’ ˜ R Z
M O
h
On the Reynolds Equation and the Load Problem in Lubrication … 13
Notice that:
ρ
(Ob O j M ) = − α̃ = − arcsin sin() , (10)
R
and therefore:
R ρ
Ob M = sin − arcsin sin() . (11)
sin() R
we can calculate the sine of the sum indicated in Eq. (11), from which we obtain:
R ρ 2 1/2
Ob M = sin() cos arccos 1 − sin()
sin() R
ρ
− cos() sin arcsin sin() ,
R
ρ 2 1/2
R ρ
Ob M = sin() 1 − sin() − cos() sin() ,
sin() R R
ρ 2 1/2
Ob M = R 1 − sin() − ρ cos(). (13)
R
ρ C
< 1. (15)
R Rb
h ≈ Rb − R + ρ cos(),
h ≈ C + ρ cos(). (16)
In Fig. 3 we can notice the relation = θ − α. Thus, Eq. (16) can be rewritten as:
Another random document with
no related content on Scribd:
Fam. 2. Oculinidae.—Colonial forms, dendritic or encrusting, with
relatively large and rather prominent calices separated by
considerable stretches of compact coenosteum. The zooids bear a
crown of ten to forty-eight or more capitate tentacles.
The genera of Astraeidae that form colonies may be divided into two
groups: the Gemmantes and the Fissiparantes. In the group
Gemmantes asexual reproduction is effected by gemmation, and
each zooid of a colony is a distinct individual with two pairs of
directive mesenteries. Among the best known of recent corals
included in this group may be mentioned Galaxea. In this genus
there is a good deal more coenosteum between the calices than
there is in most of the Astraeidae. The calices are long and project
some distance above the coenosteum. The septa are exsert. In
Galaxea esperi examined by Fowler[416] there are twelve septa,
twelve pairs of mesenteries, and twenty-four tentacles, of which
twelve are very small and twelve rather larger. The colour is green or
brown. The genus is found in shallow water in the tropics of the old
world.
Fig. 176.—A single calyx of Pocillopora septata, showing Co, the columella; S, S,
the septa; Th, the theca wall. (After Gardiner.)
The free adult Fungias are derived from a fixed stock called the
trophozooid, from which the young Fungias are detached by
transverse fission (see p. 388). The thecal wall of the young Fungia
when detached from the trophozooid is perforated, but the pores
become largely filled up during the later growth of the coral.
The only genera occurring within the British area are Epizoanthus
(with six species), Parazoanthus (with four species), and Zoanthus
sulcatus.
There are six complete mesenteries in each zooid, but as they bear
no retractor muscles it is not certain that they represent the first six
protocnemes of other Zoantharia. In a great many species the zooids
are oval in shape, the longer diameter being parallel with the axis of
the branch. The mouth and stomodaeum are compressed and at
right angles to this diameter. It is usually assumed that the
mesenteries attached to the angles of the stomodaeum are the
directives, and that the remaining pair, which is axial in direction,
corresponds with the first pair of protocnemes. The axial pair of
mesenteries is frequently very well developed and alone bears the
gonads. When other mesenteries are formed they always arise in
bilateral pairs between the axial mesenteries and the directives. The
tentacles correspond with the intermesenteric chambers. In some
genera there is a constriction of the zooid between the pairs of the
tentacles on each side of the axial mesenteries and the directive
tentacles. This gives them the appearance of a division into three
zooids with two tentacles apiece, one with a mouth and two without a
mouth; and as the mouthless parts alone bear the gonads on the
single axial mesentery, they have been called the "gastrozooids" and
"gonozooids" respectively. This must not be regarded, however, as a
case of true dimorphism, as the cavities of the so-called gastrozooid
and gonozooids are continuous.
The Antipatharia are widely distributed in nearly all the great seas of
the world. Some species are found in shallow water in the tropics,
but most of them occur in depths of fifty to five hundred fathoms. The
genus Bathypathes is only found at enormous depths ranging from
1070 to 2900 fathoms. Specimens of Cirripathes spiralis,
Antipathella gracilis, and another species have recently been
obtained in deep water off the west coast of Ireland,[424] but these
are the only Antipatharia known to occur within the British area.
CHAPTER XV
CTENOPHORA
The Ctenophora are spherical, lobed, thimble-shaped, or band-like
animals, usually very transparent and gelatinous in structure. They
are exclusively marine, and are found floating at or near the surface
of the sea.
Although they are generally classified with the Coelenterata, they are
regarded by some authors as having closer affinities with the
Polyclad Turbellaria (cf. Vol. II. p. 7). They agree, however, with
neither of these divisions in their essential characters, and the only
way to indicate and emphasise their unique position is to place them
in a separate Phylum.
Fig. 180.—Hormiphora plumosa. Ab, position of the aboral sense-organ; Ct, rib
of ctenophoral plates; M, mouth; t, tentacle, with two kinds of pinnae. (After
Chun.)
The ova are very small when discharged and undergo complete
segmentation in the sea water. The development of the Cydippidea
is really direct, but there is a stage passed through in which the
tentacles are relatively very prominent and situated close to the
aboral pole, and this stage is very different in appearance from the