International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

Design Optimization of SGCI Insert Embedded in

Pre Stressed Concrete Sleeper
Srihari Palli, and Azad Duppala

Abstract---One of the most spectacular developments railway The tie plates are through bolted to the ties/sleepers. The
tracks went through since the beginning of the railways can be seen “direct fixation” fastening eliminates the tie plate Concrete
in the improvement of their fastening systems. The pre-stressed shoulders cast into the sleeper providing lateral support to the
concrete sleepers in a railway track use Spheroidal Graphite Cast Iron
clips.
(SGCI) inserts together with Elastic Rail Clip and rubber/elastomeric
pad to hold the rails. Fastener suppliers are rising to meet the Bolts, threaded into stainless steel inserts cast into the
challenges in a variety of ways. One such a way is to economize the sleepers are used to secure the clips to the sleeper/tie. The
design of the SGCI insert by reducing its weight. As there are four direct fixation clips are less rigid than those used in the
SGCI inserts embedded in a normal single/mono-block pre-stressed indirect fixation.
concrete sleeper, reducing the weight of the insert is necessary both The Inserts/Iron Shoulders are manufactured from pearlitic
from the economical point of view and the design point of view.
malleable iron (M.C.I. Inserts) and spheroidal graphite iron
This work is concentrated on optimizing the design of the insert,
considering various geometric parameters used in its design by (S.G.C.I. Inserts). Pre-stressed concrete sleepers use SGCI
reducing its volume which is in direct proportion to its weight. The inserts together with Elastic Rail Clip and rubber/elastomeric
best feasible/optimum solutions are obtained for reduced weight of pad to hold the rails [3], [7]. An illustration of a SGCI insert in
the insert. Stress analysis of the insert under its operating conditions a direct fixation system is shown in Fig. 1. The shank/shaft
for different load cases has been done. Parametric model of the insert part of the insert is embedded into the sleeper before the
has been built to carry on the design optimization of the SGCI insert
sleeper is casted with concrete. The top part of the insert
model using ANSYS, considering the geometric parameters as design
variables and volume as the objective. On the basis of the algorithmic resting on the sleeper surface is a combination of the tapered
optimization carried out using ANSYS , it has been found that the part called the shoulder and semi cylindrical part with a
weight of the insert can be reduced from 1.51 Kg to 1.38 Kg for the cylindrical hole in it called the centre-leg [4].
worst state of stresses occurring in the most effective load condition Two elastic rail clips are provided on both sides of the rail
keeping the maximum induced stresses within the permissible limit. and hold the flange side of the rail with the SGCI insert. The
The optimum design set obtained has been compared with that of the
elastic liners are used between the rail flange and SGCI insert,
neural network optimization that has been carried out.
and together with ERC they reduce the dynamic loads on the
Keywords--Design Optimization, Fastening system, Rail Insert, insert. A rubber/elastomeric pad is placed between the rail and
Parametric Model concrete sleeper, and reduce the dynamic loads acting on the
surface of the rail. As there are four SGCI inserts embedded in
I. INTRODUCTION a normal single/mono-block pre-stressed concrete sleeper,
reducing the weight of the insert is necessary both from the
T HE combination of rails, fitted on sleepers with a suitable
fastening system and resting on ballast and subgrade is
called the railway track or permanent way [2]. The name
economical point of view and the design point of view. The
present work is concentrated on optimizing the design of this
insert, considering various geometric parameters used in its
permanent way is given to distinguish the final layout of track
design by reducing its volume which is in direct proportion to
from the temporary tracks that are laid for conveyance of earth
its weight.
and materials during construction works.
The most important function of fastening systems is to
provide strong and flexible connection between rail and its
supporting structure that can be sleeper or slab. Two types of
fastening systems are used to secure the rails to the sleepers,
indirect fixation and direct fixation [1]. The “indirect
fixation” fastening uses a steel tie plate beneath the rail and a
tie pad between the plate and the tie.

Srihari Palli is with Aditya Institute of Technology and Management,

Tekkali, India (corresponding author’s phone: +91-9985660927; e-mail:
[email protected]). Fig.1 Illustration of a SGCI insert in a direct fixation fastening [4]
Azad Duppala is with Aditya Institute of Technology and Management,
Tekkali, India (e-mail:[email protected]).

http://dx.doi.org/10.15242/IIE.E0314116 457
 International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

II. MODELING OF THE RAIL INSERT and volumes are created through the splines and arcs using
The insert considered is a SGCI insert which is used in the modeling commands used in the MACRO written for
concrete sleepers in Indian Permanent way. The insert is modeling of the insert.
modeled parametrically in ANSYS to run the optimization B. Modeling the shoulder and centre leg of the insert:
with ease and efficiency [6]. Key dimensions which are used
The shoulder of the rail insert lies on the surface of the
in designing the insert are taken as parameters. These
sleeper on which the elastic rail clip rests. The shoulder is
geometric parameters are the independent quantities that are
modeled by tapered lines and its length is taken a geometric
varied in order to achieve the optimum design.
parameter. The centre leg of the insert is a half cylinder and is
The shaft cross section of the insert comprises of a set of
modeled as a semicircle extruded throughout the length of the
ellipses and these ellipses are guided by arcs. Hence major and
shoulder. The thickness of the centre leg is taken as a
minor axes of the base ellipse and the guiding smaller ellipse
geometric parameter. The diameter of the hole (22mm)
are taken as the parameters (‘a’ and ‘b’, ‘c’ and ‘d’
depends on the cross-section of the elastic rail clip and is a
respectively), the vertical distance between the base ellipse
fixed parameter.
and the smaller ellipse is taken as another parameter (‘h’). The
Boolean Operations were used to combine various parts i.e.
top most part of the shaft has another smaller ellipse with a
shaft, shoulder, and centre leg into a single volume. As, the
variable minor axis and is considered as another parameter
behavior of the SGCI rail insert on a whole has to be studied
(‘e’). The length (‘l’) of the shoulder of the insert which is
for optimization, free meshing with tetrahedral elements has
projected outside the sleeper and the thickness (‘t’) of the
been used for the FE model of the rail insert. Solid45, 3D
centre leg are also taken as geometric parameters and shown
structural solid element type was used for FE modeling. The
in Fig.3.
total number of elements were 13895 and nodes were 3704.
A. Modeling the shaft of the insert The FE model of the insert is shown in Fig.3.
The shaft of the insert is embedded in the concrete sleeper,
its cross section is of an ellipse and its profile is designed as a
curved surface to increase its contact area. As there is no
direct command to create an ellipse in ANSYS, the ellipse is
created through a set of key points satisfying the basic ellipse
equation [5].
x2 y2
+ =1
a2 b2
The key points are input in polar co-ordinate form i.e.,
(a cosθ, b sinθ). These key points are joined through a smooth
curve (spline). Using the symmetry these smooth curves are
reflected about Y-Z and X-Y planes to complete the ellipse.

Fig.3 FE Model of the Rail Insert

III. STRESS ANALYSIS OF RAIL INSERT UNDER OPERATING

CONDITIONS
The elastic rail clip contacts the insert at locations I and C
as shown in Fig.4. The forces FI and FC will be acting at these
locations, in addition to these forces a horizontal force Fh
( Fh = Fy − µ × Fz ) also acts on the insert coefficient of
friction μ between rail and rubber pad or between rubber pad
and concrete may be assumed as 0.4 (a conservative
estimate). The force Fh can be safely assumed to be equally
distributed along the length of the insert as shown in the Fig.4.
For analysis of stresses on the insert, six different load cases
Fig. 2 SGCI Rail Insert with geometrical parameters Modeled have been considered as per the RDSO (Research Division &
using Parametric Modeling Standards Organization for Indian Railways) specifications
and are explained below. The six load cases are further
Similarly the guiding smaller ellipse is also created and line simplified into three effective load cases to arrive at the worst
arc guiding the profile of the shaft through the specific key state of stress conditions. The insert in different stages of the
points of the larger and smaller ellipses is created. Now, areas

http://dx.doi.org/10.15242/IIE.E0314116 458
 International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

optimization procedure is subjected to forces from these three The axial force accounting for the hammer force (Fa) =
load cases. 14700 N is assumed to be acting at the top most node of the
centre-leg part of the insert as shown in Fig. 4.
A. Load case 1:
The principal stresses in different parts of the insert vary
Only vertical load of 15000 kg acts on the rail in this case. from 26.1E 7 N/m2 (compressive) to 195.0E 7 N/m2 (tensile).
According to RDSO specifications it gives the force analysis The Vonmises (S-Equivalent) stresses in the insert vary from
on the insert in this case gives the loads as 15.638E 3 N/m2 (tensile) to 183.0 E 7 N/m2 (tensile). When
FI = 28760 N, FC = 17980 N, Fh = 0. embedded in the sleeper block the principal stresses vary from
The force FI acts at the inside upper surface in the insert 4E 7 N/m2 (compressive) to 20E 7 N/m2 (tensile). The
hole and has been taken as uniformly distributed. The force FC principal stresses in the concrete block vary from 45.8620E 4
acts on a small elliptical area made due to contact between N/m2 (compressive) to 4.92E 7 N/m2 (tensile).
ERC (Elastic Rail Clip) and face of the insert shoulder at point
C. The principal stresses in different parts of the insert vary IV. DESIGN OPTIMIZATION OF RAIL INSERT
from 2.09E 7 N/m2 (compressive) to 11.3E 7 N/m2 (tensile).
Design optimization is a technique that seeks to determine
The Vonmises (S-Equivalent) stresses in the insert vary from
an optimum design. By "optimum design," means the one
11.133E 3 N/m2 (tensile) to 17.3 E 7 N/m2 (tensile). When
design that meets all specified requirements but with a
embedded in the sleeper block the principal stresses vary from minimum expense of certain factors such as weight, surface
1.93E 7 N/m2 (compressive) to 9.09E 7 N/m2 (tensile). The area, volume, stress, cost, etc. In other words, the optimum
principal stresses in the concrete block vary from 18.5576E 4 design is usually one that is as effective as possible. In
N/m2 (tensile) to 2.17E 7 N/m2 (tensile). ANSYS, Design optimization is a technique available to
B. Load case 2: obtain the optimum design. Actually, any ANSYS item that
can be expressed in terms of parameters can be subjected to
The toe load has been increased to 1500 kg in addition the design optimization.
elastomeric pad is missing (frictional coefficient μ = 0), force The ANSYS program offers two optimization methods to
analysis on the insert gives accommodate a wide range of optimization problems. The
FI = 39200 N, FC = 24500 N, Fh = 68600 N. subproblem approximation method is an advanced zero-order
The lateral load Fh acts just above the projected part of the method that can be efficiently applied to most engineering
insert on which the rail rests through the elastic liner. The problems. The first order method is based on design
principal stresses in different parts of the insert vary from sensitivities and is more suitable for problems that require
58.6E 7 N/m2 (compressive) to 609.8E 7 N/m2 (tensile). The high accuracy.
Vonmises (S-Equivalent) stresses in the insert vary from A single criterion optimization of the SGCI insert has
42.091E 3 N/m2 (tensile) to 604.0 E 7 N/m2 (tensile). When been carried out for individual load cases using the Design
embedded in the sleeper block the principal stresses vary from Optimization tool in ANSYS. The model is built
8.63E 7 N/m2 (compressive) to 48.9E 7 N/m2 (tensile). The parametrically and explained in the macro file. Assumptions
principal stresses in the concrete block vary from 0.124E 7 considered for this optimization process are:
N/m2 (compressive) to 10.5E 7 N/m2 (tensile). • The insert alone is considered and its base is fixed to
account for the maximum stress condition.
• To make the maximum utility of insert material, the state
variable is taken as a fraction which is defined as ratio of
minimum tensile strength of the insert material to factor of
safety times the Vonmises stress (S-Eq). Optimization has
been carried out for maximum Vonmises stress and Total
Vonmises stress condition occurring in the insert.
• The limits of the design parameters considered as design
variables for optimization process are taken to be within
the limits of the topology of the insert.
A. Design Variables
Geometrical parameters considered as design variables for
optimization are explained below. The original values of each
of these geometrical parameters are presented in Table I. For
all the optimization processes carried out for all the three load
Fig. 4 Different Forces acting on the Rail Insert cases considering both the maximum Vonmises stress as well
C. Load case 3: as the total Vonmises stress the limits for the design variables
are taken as per the topology requirement of the insert.
The force acting in the axial direction of the 22 mm hole in
a -----Major axis of the base ellipse
the insert at the centre leg part to account for the hammer
b ------Minor axis of the base ellipse
striking force, along with forces c ------Major axis of the small guiding ellipse
FI = 39200 N, FC = 24500 N, Fh = 0. d ------Minor axis of the small guiding ellipse

http://dx.doi.org/10.15242/IIE.E0314116 459
 International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

e ------Minor axis of the small guiding ellipse at the top Total Vonmises stress (S -Eq) total is extracted by summing
part of the shaft up each element’s S-Eq and accepting it as a parameter from
h ------Vertical distance between base and small ellipses the results data. For the State variable i.e. for the fraction, a
f-------Length of the shoulder part maximum limit has been used which is approximately taken as
t ------- Thickness at the centre-leg above the hole. 60% of the original value.

TABLE I C. Objective Function

LIMITS AND VALUES OF DESIGN VARIABLES The objective of the optimization process is to minimize
Design Variable Minimum Maximum Original Value the weight of the insert, since there is no direct option to
Name Limit (m) Limit (m) (m)
a 0.015 0.02 0.0175 extract weight from the post processing results data volume of
b 0.01 0.015 0.0125 the insert is extracted from the results data as the volume is in
c 0.01 0.015 0.012 direct proportion to the weight. Hence volume is taken as the
d 0.005 0.009 0.007 objective parameter for the design optimization processes. The
e 0.007 0.012 0.012
h 0.015 0.023 0.02 density of the Spheroidal Graphite Cast Iron (SGCI) material
f 0.055 0.062 0.05868 is taken as 7800 kg/m3.
t 0.009 0.012 0.0105 For the Objective the convergence tolerance for all the
load cases is taken as 1E-09.
B. State Variables
State Variable taken here is a fraction defined as a V. RESULTS OBTAINED IN ANSYS DESIGN OPTIMIZATION
parameter for different types of stresses as:
The design sets have been obtained in the optimization for
Fraction (Fr) =
Minimum Tensile strength of SGCI different load cases considering maximum Vonmises stress as
(1.2 * Maximum Vonmises stress) well as total Vonmises stress. It has been found that for load
500E 10 case 2 considering total von-Mises Stress in which the worst
→ Fraction (Fr) = state of stresses occur the weight of the insert has been
[1.2 * ( S − Eq) max ] reduced from 1.511 kg to 1.378 kg as shown in Table II. From
Maximum Vonmises stress (S -Eq) max is extracted from the Fig. 5 to Fig.9 the variations of design variables, state
post processing results data. Minimum tensile strength of variables and objective with each design set are shown.
SGCI is taken as 500E 10 N/m2 (IS code 1865:1991) [8].
Fraction (Fr1) = Minimum Tensile strength of SGCI
(1.2 * Total Vonmises stress)
→ Fraction (Fr1) = 500E 10
[1.2 * ( S − Eq) total ]
TABLE II
OPTIMUM VALUES OF GEOMETRIC PARAMETERS FOR REDUCED WEIGHT OF THE INSERT
DESIGN SETS OBTAINED BY OPTIMIZATION USING ANSYS
PARAMETER PARAMETER
NAME TYPE SET 1 SET 2 SET 3 SET 4 *SET 5* SET 6

Fr1 (SV) > 1.99 > 1.567 > 1.754 1.54 1.436 1.482

a (DV) 1.75E-02 1.7214E-02 1.6830E-02 1.5832E-02 1.5412E-02 1.5386E-02

b (DV) (DV) 1.25E-02 1.2345E-02 1.1461E-02 1.1236E-02 1.1137E-02 1.11092E-02

c (DV) (DV) 1.20E-02 1.2318E-02 1.0952E-02 1.0834E-02 1.0624E-02 1.0724E-02

d (DV) (DV) 7.00E-03 7.2400E-03 6.6270E-03 6.3820E-03 6.2510E-03 6.2538E-03

e (DV) (DV) 1.00E-02 9.8720E-03 9.7401E-03 9.7540E-03 9.3262E-03 9.1672E-03

h (DV) (DV) 2.00E-02 2.1630E-02 1.8630E-02 1.8261E-02 1.7683E-02 1.8164E-02

f (DV) (DV) 5.868E-02 5.9027E-02 5.7162E-02 5.7346E-02 5.7164E-02 5.7189E-02

t (DV) (DV) 1.05E-02 1.1600E-02 9.8351E-03 9.6411E-03 9.8341E-03 9.5436E-03

Volume (m3) (Objective) 1.9367E-04 2.0104E-04 1.8234E-04 1.7918E-04 1.7667 E-04 1.7715E-04

Weight
1.5106 1.56811 1.4225 1.3976 1.37802 1.38177
(Kg)

**SEq (N/m2) 2.09224E+12 2.6586E+12 2.3753E+12 2.70562E+12 2.90157E+12 2.81151E+12

* ---------- Optimum Design Set

**SEq ------------ Vonmises(Equivalent) Stress
> ---------- In Feasible Set

http://dx.doi.org/10.15242/IIE.E0314116 460
 International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

Fig. 8 State Variable (Fraction-Fr1) (Vs) Design set number for Load
Fig. 5 Design variables of ellipses (Vs) Design set number for Load case 2 considering (S-Eq)total
case 2 considering (S-Eq)total

Fig. 9 Objective-Volume (Vs) Design set number for Load case 2

Fig. 6 Design variables of other parameters (Vs) Design set number considering (S-Eq) total
for Load case 2 considering (S-Eq)total
By varying each individual design variable (geometric
parameter) and keeping all other design variables fixed,
volume of the insert has been found. Using these values of
parameters and variation of each individual parameter with the
volume has been plotted in Fig.10.

Fig. 7 Total Vonmises stress(S-Eq)total (Vs) Design set number for

Load case 2

Fig. 10 Volume Vs Parameters

http://dx.doi.org/10.15242/IIE.E0314116 461
 International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

von-Mises stress condition the weight of the insert has

A. Results Obtained in stress Analysis of Insert before and
been reduced from 1.5106 Kg to 1.36102 Kg.
after optimization for the three Load cases. 2. When considering maximum von-Mises stress condition
1. For the first load case the maximum von-Mises (S- the weight of the insert has been reduced from 1.5106 Kg
Equivalent) stress in the insert has been found as 17.3E 7 to 1.3863 Kg for load case 2, whereas considering total
N/m2 (tensile). After optimization the maximum von- von-Mises stress condition the weight of the insert has
Mises (S-Equivalent) stress in the insert has been found as been reduced from 1.5106 Kg to 1.37802 Kg.
81.821E 10 N/m2 (tensile) and is far lesser than the tensile 3. When considering maximum vonMises stress condition
strength of the insert material (500E 10 N/m2 ). the weight of the insert has been reduced from 1.5106 Kg
2. For the first load case the maximum von-Mises (S- to 1.3602 Kg for load case 3, whereas considering total
Equivalent) stress in the insert has been found as 604.0E 7 von-Mises stress condition the weight of the insert has
N/m2 (tensile). After optimization the maximum von- been reduced from 1.5106 Kg to 1.3503 Kg.
Mises (S-Equivalent) stress in the insert has been found as
281.151E 10 N/m2 (tensile) and is far lesser than the VI. NEURAL NETWORK MODEL FOR OPTIMIZATION
tensile strength of the insert material (500E 10 N/m2 ).
Using the data from the above plots from Fig.5 to Fig.9 in
3. For the first load case the maximum von-Mises (S-
a neural network model, the values of each individual
Equivalent) stress in the insert has been found as 183.0E 7
parameter by reducing the volume of the insert i.e. by
N/m2 (tensile). After optimization the maximum von-
reducing its weight at an interval of 100 gm have been
Mises (S-Equivalent) stress in the insert has been found as
obtained and are shown in Table III. Neural network model is
223.15E 7 N/m2 (tensile) and is far lesser than the tensile
developed using the neural planner program for estimating the
strength of the insert material (500E 10 N/m2 ).
parameters [9]. The program implements the back propagation
algorithm. The configuration of each model with the obtained
B. Results obtained in Single criterion Design Optimization
individual geometric parameter has been subjected to all the
of insert three load cases for stress analysis. Comparison of target
1. When considering maximum von-Mises stress condition volume and weight achieved with neural network has been
the weight of the insert has been reduced from 1.5106 Kg shown in Table IV.
to 1.35108 Kg for load case 1, whereas considering total

Fig. 11 Schematic diagram of Neural Network Model

TABLE III
VALUES OF VARIOUS PARAMETERS OBTAINED THROUGH NEURAL NETWORK
Weight of the Design Variables (m)
Volume(E-03)
SGCI Insert
(m3) a b c d e h f t
(Kg)
1.5106 0.19367 0.0175 0.0125 0.012 0.007 0.01 0.02 0.05868 0.0105
1.4106 0.18089 0.0169682 0.012281 0.011576 0.006818 0.01001 0.019024 0.058511 0.010464
1.3106 0.16807 0.0168559 0.012204 0.011487 0.006769 0.010014 0.018785 0.058238 0.010403
1.2106 0.15525 0.0168416 0.012133 0.011505 0.006744 0.010025 0.01876 0.058056 0.010392
1.1106 0.14243 0.0169144 0.012011 0.011611 0.006724 0.010051 0.018852 0.057696 0.010415

TABLE IV
COMPARISON TABLE OF THE VOLUME OBTAINED THROUGH ANSYS AND THE VOLUME GIVEN BY NEURAL NETWORK.
Target Volume of the
Volume of the insert Target Weight of the insert Weight of the insert Percentage reduction in
insert given by Neural
3 calculated in ANSYS (m3) given by Neural Network (Kg) calculated in ANSYS (Kg) Weight/Volume for Neural Network
Network (m )
0.18089E -03 0.18945E -03 1.4106 1.4777 4.518
0.16807E -03 0.18775E -03 1.3106 1.4644 10.45
0.15525E -03 0.18752E -03 1.2106 1.4630 17.22
0.14243E -03 0.187497E -03 1.1106 1.4624 24.05
0.12961E -03 0.19186E -03 1.0106 1.4965 32.44

http://dx.doi.org/10.15242/IIE.E0314116 462
 International Conference on Advances in Engineering and Technology (ICAET'2014) March 29-30, 2014 Singapore

VII. CONCLUSIONS Srihari Palli, received M. Tech. from Indian Institute of Technology
Roorkee, India in 2005. He is an Associate Professor in the Department of
Design evaluation of SGCI insert used on concrete sleepers Mechanical Engineering, Aditya Institute of Technology and Management
by the Indian Railways has been carried out. The stress Tekkali, India. His research interests include Vehicle Dynamics, Design
analysis of the insert under its operating conditions for Optimization, Computer Integrated Manufacturing, Product Life Cycle
Management, Digital Manufacturing, Manufacturing Process Management,
different effective load cases has been carried out. The best Finite Element Analysis, Machine & Structural Dynamics. He has a
feasible/optimum solutions have been obtained for reducing cumulative 9.5 years of academic and industrial experience. He is an associate
the weight of the SGCI insert. On the basis of this study it can member of Institute of Engineers India (IE(I)).
be concluded that
Azad Duppala, is an Associate Professor in the Department of Mechanical
Engineering, Aditya Institute of Technology and Management Tekkali, India.
1. The influence of each individual design variable on the He received his M.Tech from Andhra University Visakhapatnam, India. He
weight of the insert has been found and it can be has recently submitted his Ph.D in Andhra University. His research interests
concluded that the design variable ‘h’ defining the height include Condition Monitoring, Optimization, Quality Management, Fault
Diagnosis, Alternate Sources of Energy and Machine Dynamics. He has over
of the shaft/shank of the insert is most critical to the a 20 years of rich academic and industrial experience. He has various
weight of the insert, i.e. slight reduction in the value of ‘h’ publications in National & International journals and has successfully
results in a significant reduction the weight of the insert. organized & attended many technical conferences and workshops. He is a
Subsequently, design variables ‘a’ and ‘b’, the major and member of Institute of Engineers India (IE(I)) and Life Member of Indian
Society for Technical Education (ISTE).
minor axes of the base ellipse are more critical to the
weight of the insert as compared to other design variables.
2. On the basis of the algorithmic optimization carried out
using ANSYS, it has been found that the weight of the
insert can be reduced from 1.51 Kg to 1.38 Kg for the
worst state of stresses occurring in the most effective load
condition i.e. load case 2, keeping the maximum induced
stresses within the permissible limit. Hence the optimized
weight of the SGCI insert has been found to be 1.38 Kg.
3. The limited parametric optimization carried out using the
neural network indicates that the target weight could not
necessarily be achieved by using the obtained design
variable values from the neural network. Hence
parametric optimization by neural network as formulated
cannot be recommended for achieving the desired results.
4. The studies conducted clearly indicate that reduction in
weight of the SGCI insert is a distinct possibility, without
compromising on the enhanced factors of safety
associated with its functioning under normal operating
conditions. This aspect needs further studies collaborated
by experimental work before a configuration can be
adopted for mass production, leading to enormous savings
for the Indian Railways.

REFERENCES
[1] Eszter Ludvigh, "Elastic Behaviour of Continuously Embedded Rail
Systems", Department of Highway and Railway Engineering Budapest
University of Technology and Economics, H–1521 Budapest, Hungary
March 31, 2001.
[2] S.C.Saxena and S.P.Arora, "A Text Book of Railway Engineering",
Sixth Edition, 2001, pp. 9.1- 9.21.
[3] Website of Pama International Ltd., "Other Rail Track
Fastenings",<www.pamainternational.com>, February,2005.
[4] Website of Pandrol Clips, <www.pandrol.com>, January,2005.
[5] "ANSYS Commands Manual", Ansys 7.
[6] "ANSYS Modeling and Meshing Guide", Section 5.10.6.2,
Ansys7.
[7] Mindess S., Yan C., Venuti W. J, "Impact resistance of prestressed
concrete railroad sleepers", International Symposium on precast concrete
railway sleepers, Madrid, 1991, pp. 487-504.
[8] IS code 1865:1991, "Mechanical Properties measured on test pieces
from separately cast Test Samples", Section 12, pp. 5-6.
[9] Venkatesh K., Pandey A. D. and Mukherjee S. (2002), “Application of
Artificial Neural Networks in Estimating SPT N-Value.” Proceedings of
Indian Geotechnical Conference 2002, Vol. 1, Indian Geotechnical
Society, Dec. 20-22, 2002, Allahabad, India, pp. 505-508.

http://dx.doi.org/10.15242/IIE.E0314116 463