R3.100 Final Report

A Guide to
Insulated
Rail Joints
A Guide to Insulated Rail Joints
DOCUMENT CONTROL SHEET
CRC for Rail Innovation Document:

Old Central Station, 290 Ann St.
Title: A Guide to Insulated Rail Joints
Brisbane Qld 4000
Project Leader: Manicka Dhanasekar
GPO Box 1422 Author: Manicka Dhanasekar
Brisbane Qld 4001 Project No.: R3.100
Tel: +61 7 3221 2536 Project Name: Insulated Rail Joints
Fax: +61 7 3235 2987

www.railcrc.net.au
Synopsis:
Insulated rail joints (IRJs) are a crucial part of the Australian rail network used for identifying trains within
a track circuit and the presence of broken rails. Improvement in the design, installation and maintenance
of IRJs is important for the capacity and safety of the network. Research was undertaken, culminating in
this guide, which developed a major field test, full-scale wheel-rail test rig, a digital image correlation
technique, and a neutron diffraction method. Significant computational optimisation and ratchetting
modelling were also developed. The research developed a ratchetting failure life prediction model and
identified failures in IRJs and suggested ways of addressing failures. It showed that the ratchetting metal
flow in the insulating gaps of the IRJs can only be addressed through modifications to the wheel contact
near the railhead edge; reducing impact load will not be an effective solution for this purpose. All other
modes of failure can only be addressed through appropriate conditioning of the track in the vicinity of
the IRJs.
REVISION/CHECKING HISTORY
REVISION DATE ACADEMIC REVIEW INDUSTRY REVIEW APPROVAL
NUMBER
(PROGRAM LEADER) (PROJECT CHAIR) (RESEARCH DIRECTOR)
0 [insert date] Colin Cole Ian Marks Chris Gourlay
DISTRIBUTION
REVISION
DESTINATION 0 1 2 3 4 5 6 7 8 9 10
Industry Participant x
for Review
Established and supported under the Australian Government’s Cooperative Research Centres Programme
CRC for Rail Innovation October 2013 Page 1

Copyright © 2013
This work is copyright. Apart from any use permitted under the Copyright Act 1968, no part may be reproduced
by any process, nor may any other exclusive right be exercised, without the permission of Queensland University of
Technology.

TABLE OF CONTENTS
1. INTRODUCTION ................................................................................................................................ 1
1.1 INSULATED RAIL JOINTS (IRJS) ................................................................................................................... 1

1.2 OUTLINE OF THE GUIDE ............................................................................................................................ 6
2. CURRENT STATE OF IRJS ................................................................................................................... 8
2.1 IRJS IN COAL CORRIDORS .......................................................................................................................... 8

2.2 IRJS IN PASSENGER-FREIGHT CORRIDORS ..................................................................................................... 9
2.3 IRJS IN IRON ORE CORRIDORS .................................................................................................................... 9
2.4 QUALITY ASSURANCE IN THE MANUFACTURE OF IRJS ................................................................................... 10
2.5 HANDLING AND PLACEMENT OF IRJS ON RAIL TRACK .................................................................................... 11
3. TOWARDS IMPROVING THE PERFORMANCE OF THE IRJS ................................................................ 14
3.1 MONITORING PROGRESSIVE DEGRADATION OF THE IN-SERVICE IRJS: .............................................................. 14

3.1.1 Progressive narrowing of the gaps of the IRJs: ........................................................................ 15
3.1.2 Progressive settlement of sleepers adjacent to the IRJs .......................................................... 17
3.2 PERFORMANCE OF IN-SERVICE IRJS UNDER THE PASSAGE OF WHEELS.............................................................. 20
3.2.1 Track input to wheel-rail contact impact forces at the IRJs ..................................................... 21
3.2.2 Track input to joint bar bending strains: .................................................................................. 25
3.2.3 Track input to response of sleepers adjacent to IRJs................................................................ 26
3.2.4 Track input to ballast pressure beneath IRJs:........................................................................... 31
3.2.5 Effect of the Angle of Cut in the Rails of the IRJs ................................................................... 32
3.3 PLASTIC STRAINS AND RESIDUAL STRESSES IN IRJS ....................................................................................... 35
3.3.1 Accumulation of Plastic Strains in Current Designs of IRJs ..................................................... 36
3.3.2 Residual stresses in the IRJs ................................................................................................... 41
3.4 RATCHETTING FAILURE: LIFE PREDICTION MODEL FOR THE IRJS ...................................................................... 46
3.4.1 Geometry Idealisation of the IRJs for Finite Element Modelling .............................................. 46
3.4.2 Material properties of the IRJs for finite element modelling: monotonic .............................. 48
3.4.3 Material properties of the IRJs for finite element modelling: cyclic ....................................... 54
3.4.4 Plastic strains: comparison of FE and experimental predictions ............................................ 57
3.4.5 Load History for Prediction of Ratchetting ............................................................................. 63
3.4.6 Ratchetting failure life prediction .......................................................................................... 64
3.5 SUMMARY............................................................................................................................................ 69
4. LONGER LIFE NEW GENERATION IRJS ............................................................................................. 71
4.1 LOW-IMPACT IRJ................................................................................................................................... 71

4.1.1 Parameters of IRJ design considered in sensitivity studies ..................................................... 73
4.1.2 Rolling vs. Sliding ................................................................................................................... 74
4.1.3 Inserted vs. glued end posts ................................................................................................... 74
4.1.4 10mm vs. 5mm gap................................................................................................................ 75
4.1.5 4-Bolted vs 6-Bolted Joint Bars ................................................................................................ 76
4.1.6 Suspended vs. Supported IRJ .................................................................................................. 77

4.1.7 Discussion ............................................................................................................................... 78

4.2 NON-RATCHETTING IRJ .......................................................................................................................... 79
5. CONCLUSIONS ................................................................................................................................ 83
REFERENCES ............................................................................................................................................. 84
APPENDIX – DIGITAL IMAGE CORRELATION FOR STRAIN MEASUREMENT ............................................... 86

LIST OF FIGURES AND TABLES

Figure 1.1 - Four IRJs form a Track Circuit
Figure 1.2 - Components of the Insulated Rail Joints (IRJs)
Figure 1.3 – A Typical Track Geometry car Data from an Australian Track
Figure 1.4 - Ratchetting Failure
Figure 1.5 - Contact Pressure Singularity
Figure 1.6 - A 12 Bolt IRJ (Akhtar and Davis, 2011)
Figure 1.7 - Top Views of High Yield Strength Steel Welded IRJs
Figure 2.1 - Ballast Pockets
Figure 2.2 - A Four Bolt IRJ positioned on the Sleeper
Figure 2.3 - IRJ made in the USA with thick joint bars
Figure 2.4 - Rail Wear Limit Parameters (CoP, 2013)
Figure 3.1 - IRJs for Monitoring of Progressive Deterioration (Zong et al. 2013)
Figure 3.2 - Digital Imaging of IRJs (Zong et al. 2013))
Figure 3.3 - Progressive Closure of the Gap of the IRJs (Zong et al. 2013)
Figure 3.4 - Location of Measuring Levels of Sleeper Tops (Zong et al. 2013)
Figure 3.5 - Field Test Setup of Instrumented IRJs
Figure 3.6 - IRJs Tested for Track Input Effects (Askarinejad et al. 2013)
Figure 3.7 - Sensors Used in the Field Experiment on IRJs
Figure 3.8 - Calibration of Wheel-Rail Contact Impact Force
Figure 3.9 - Enlarged views of Wheel-Rail Contact Force Signatures
Figure 3.10 - Typical Wheel-Rail Contact Impact Force Signatures
Figure 3.11 - Typical Joint Bar / Rail Bottom Longitudinal Strain Signatures
Figure 3.12 - Joint Bar Top Longitudinal Strain Signatures at two IRJ
Figure 3.13 - Typical Sleeper Acceleration Signatures Adjacent to IRJs & Rail
Figure 3.14 - Vertical deflection profile of a sleeper under Reference Rail
Figure 3.15 - Vertical deflection of sleepers in Hexham and Whittingham Sites
Figure 3.16 - Vertical deflection profile of sleepers under the four IRJs
Figure 3.17 - Ballast Pressure Profile due to Passage of Loaded Wheels
Figure 3.18 - Peak Ballast Pressure due to Passage of Wheels over Sleepers
Figure 3.19 - Peak Ballast Pressure due to Passage of Wheels over Joint Gaps
Figure 3.20 - Square Cut and Angle Cut IRJs
Figure 3.21 - Conceptual Diagram of the Hypothesis on Two IRJs
Figure 3.22 - Strain Gauged IRJ under Wheel Passage
Figure 3.23 - Vertical Strains in the IRJs under Wheel Passage
Figure 3.24 - Shear Strains in the IRJs under Wheel Passage
Figure 3.25 - Longitudinal Strains in the IRJs under Wheel Passage
Figure 3.26 - Lab Set-up for Plastic Strain Accumulation Measurement
Figure 3.27 - Directions of Wheel Motion and Limit Positions
Figure 3.28 - Shape of Contact Patches Adjacent to Railhead Edge
Figure 3.29 - Vertical Strain (E22) Distribution along Rail End Symmetric Axis for Varying Positions (Z)
of a Wheel Loaded with 130.7kN
Figure 3.30 - Accumulated Vertical Plastic Strain (PE22) Distribution

Figure 3.31 - Plastic Zones of Rail End Face

Figure 3.32 - Accumulated Vertical Plastic Strain (PE22) Distribution along Rail End Symmetric Axis
after the 1st and 100th Passage of a Wheel Loaded with 130.7kN
Figure 3.33 - Accumulated Vertical Plastic Strain (PE22) Distribution along Rail End Symmetric Axis
after the 1st and 100th Passage of a Wheel Loaded with 50.0kN
Figure 3.34 - Kowari Nuclear diffraction Facility for Residual Stress Measurements
Figure 3.35 - Slicing of Railheads in IRJs for Transverse (T-Slices) and Longitudinal (L-Slices) Residual
Stress Measurements
Figure 3.36 - T-Slices Measurements of Residual Stresses in the In-Field IRJs
Figure 3.37 - L-Slices Measurements of Residual Stresses in the In-Field IRJs
Figure 3.38 - L-Slices Measurements of Residual Stresses in the Lab Tested IRJs
Figure 3.39 - Simplification of the Geometry of the IRJ Assembly for FE Modelling
Figure 3.40 - Simplification of the Geometry of the Wheel for FE Modelling
Figure 3.41 - Assembled IRJ Wheel – Railhead Model for FE Meshing
Figure 3.42 - Head Hardened AS60kg/m Rail Section
Figure 3.43 - Typical Monotonic Uniaxial Tension Test Coupon
Figure 3.44 - Strategies for Extraction of Test Coupon
Figure 3.45 - Cutting of Test Coupon
Figure 3.46 - Strain Gauged Face of Test Coupon
Figure 3.47 - Colour Dotted Face of Test Coupon
Figure 3.48 - Testing of Coupon in Instron
Figure 3.49 - Stress – Strain Curves of Coupon
Figure 3.50 - Stress – Strain Curves of Head hardened and Non-hardened Zones
Figure 3.51 - Brittle Fracture of Coupon
Figure 3.52 -Variation of 0.2% Proof Strength of Steel as a function of Depth
Figure 3.53 - Stress at Different Depths for   0.03
Figure 3.54 - Average and Extrapolated Top Stress-Strain Curves for Heat Treated, Hardened Railhead
Figure 3.55 - Geometry of the Cyclic Load Test Specimen
Figure 3.56 - Half test Specimen with Symmetry Boundary condition
Figure 3.57 - FE Model of the Cyclic Load Test Specimen
Figure 3.58 - Half Cycle Stress – Plastic Strain Data
Figure 3.59 - Applied Cyclic Tension
Figure 3.60 - Output: Axial Plastic Strain Time History
Figure 3.61 - Output: Axial Stress – Axial Plastic Strain Hysteresis Loop
Figure 3.62 - Railhead Top: Bi-Material – Top and average Zones
Figure 3.63 - Vertical Strain at Unsupported Free End of Railhead
Figure 3.64 - Lateral Strain at Unsupported Free End of Railhead
Figure 3.65 - Shear Strain at Unsupported Free End of Railhead
Figure 3.66 - Strains in the Longitudinal Vertical Plane of the Loaded Rail in IRJ
Figure 3.67 - Equivalent Uniaxial Plastic Strain (PEEQ) Distribution on Unsupported Railhead at the
Gap
Figure 3.68 - Applied Load History
Figure 3.69 - Progressive accumulation of Plastic Strain
Figure 3.70 - Schematic Diagram of Limit Dimension of Metal Flow
Figure 4.1 - Rail/wheel contact force history

Figure 4.2 -Contact force history of IRJ and Continuous Rail

Figure 4.3 - IRJ Design Parameters Examined
Figure 4.4 - Contact force history of wheel pure rolling and pure sliding
Figure 4.5 Contact force history of glued and inserted joint
Figure 4.6 Contact force history of 10mm and 5mm gap size
Figure 4.7 - Contact force history of 4-bolt and 6-bolt joint bar
Figure 4.8 - Illustration of sleeper spacing and joint bar length
Figure 4.9 - Contact force history of IRJs suspended or supported
Figure 4.10 - Contact force history for Low Impact IRJ
Figure 4.11 - Stress Concentration in the Current and Non-Ratchetting IRJs
Figure 4.12 - Stresses in the Current and Non-Ratchetting IRJs
Figure 4.13 – Impact factor in the Current and Non-Ratchetting IRJs
Figure A1 - Principles of digital image analysis for strains
Figure A2 - Successive images of two adjacent ‘patches’
Figure A3 - Successive images of four adjacent ‘patches’
Table 3.1 - Properties of Rail Steel

Table 3.2 - Parameters for Chaboche Model
Table 3.3 - Summary of Applied Stress History for Seven Select points
Table 3.4 - Predicted Ratchetting at the Seven Select Points
Table 3.5 - Predicted Metal Flow into the Gap
Table 3.6 - Observed Life of IRJs in the Field
Table 3.7 - Service life of the three FE models for 150kN wagon wheel loads
Table A1 - Description of the symbols in the output text file

EXECUTIVE SUMMARY
It is estimated that approximately 50,000 Insulated Rail Joints (IRJs) exist in the Australian rail
network; they exhibit an average life of 341 million gross tonnes (MGT) (30% Coefficient of Variation).
Due to their low service life, peak annual replacement of 24% occurs, which reduces the network
capacity through closures/ speed restrictions. As IRJs are sources of impact, their surrounding
structures deteriorate at a faster rate than elsewhere in the track; hence any improvement to the
design, installation and maintenance of the IRJs can enhance the capacity and safety of the network.
With a view to better understanding and improving IRJs, the CRC for Rail Innovation funded a major
project with strong input from the Australian Rail Track Corporation (ARTC) and Queensland Rail (QR).
The focus is particularly on IRJs in heavy haul networks, although the findings can be generalised in
many instances. The major outcomes of the research are:
1. classification of the failure of IRJs with clear directions for addressing each type of failure –
for example:
a. a reduction in impact can be achieved through reduced gap size and not through
modifications to the track modulus
b. the evolution of the residual stresses due to accumulated plastic strains in the
railheads of the IRJs observed through neutron diffraction and a nonlinear finite
element method clearly has established that the ratchetting metal flow into the gaps
of the IRJs with electrical short-circuiting risk is due to the wheel contact mechanism.
This problem cannot be treated solely by using higher yield strength material. A more
robust solution will require modifications to contact parameters to
minimize/eliminate the risk.
2. development of a model for estimating the life of the IRJs that fail due to ratchetting metal
flow – through predicted life, maintenance strategies can be worked out in an informed
manner.

Acknowledgments:
Robert Taylor BEng(Civil & Mech), Formerly ARTC

Ian Marks, QR
Enda Crossin PhD, Formerly QR
Clinton Crump BEng(Civil), ARTC
Vladimir Luzin PhD, ANSTO
Carlos Valente BEng(Civil), QR
Ron Moller PhD, Formerly MD - Thermit Australia
Christopher West, Thermit Australia
David Wexler PhD, University of Wollongong
Paul Boyd BEng(Mech), CQ university Australia
Tao Pang MEng (Rail), CQ University Australia
Hossein Askarinejad MEng (Rail), CQ University
Thaminda Bandula Heva PhD, QUT
Paviz Barathi MEng(Rail), QUT
Chandrahaas Rathod MEng(Mech), UoW
Nannan Zong PhD, QUT
Adam Mayers BEng (Hons, Mech), QUT
Wirtu Bayissa PhD, formerly QUT
Nirmal Mandal PhD, CQ University, Australia
Sid Hayes, Rio Tinto
Paul Radman PhD, MD – Thermit Australia.

ABSTRACT
Insulated rail joints (IRJs) are widely used in the rail network for identification of trains within a track
circuit; they also help identify broken rails if there are any in the network. The current design and
installation practices, despite good efforts and available national standards, still identify IRJs as areas
of high vulnerability for traffic and high damage potential for the track. With a view to improving the
current situation, two major research projects have been sponsored by the CRC for Rail Innovation
and its predecessor, the Rail CRC. Data generated from other related track research projects have also
provided additional sources of information. As part of the research, an innovative full scale wheel-rail
test rig, state-of-the-art digital image correlation technique and neutron diffraction method have
been developed/ adapted in addition to significant computational optimisation and ratchetting
modelling. Three PhD theses and three Master of Engineering Research theses have emerged from
this research and another three PhD theses are currently nearing completion; ten journal papers and
six conference papers have also been published as a result of this research. The outcomes of the
research would not have been possible without the significant in-kind support from the Australian
Rail Track Corporation (ARTC) and Queensland Rail (QR); the Australian Nuclear Science and
Technology Organisation (ANSTO) has also contributed significantly. The major outcomes from the
research are as follows:
1. The major concern, ratchetting metal flow into the insulating gaps of the IRJs, is solely due to
the design of sharp edges of the railheads, where the wheels establish contact at the top of
the unsupported railhead end section. Track condition does not contribute to the ratchetting
metal flow at the localised railhead top edge. In the current design of IRJs, longitudinal
vertical compressive residual stresses of up to 300MPa within the top ~5mm layer and
vertical tensile residual stresses of 300MPa at the very localised corner of the railhead top
edge were measured.
2. Track condition is the major contributor to the variability in life of the IRJs where the failure
mode is either the joint bar fatigue or ultimate load collapse / bolt-hole fracture. Tracks with
higher track moduli are beneficial in reducing (i) strain ranges (and hence increasing fatigue
life) in joint bars, (ii) sleeper accelerations, (iii) sleeper displacements and (iv) ballast pressure.
All these aspects contribute to reduced rates of progressive damage to the track structure –
thereby minimising the need for frequent maintenance of IRJs and the surrounding track
structure, which can save significant costs.
3. Theories that emerged from the research provide models for conservative prediction of the
life of the IRJs. A non-ratchetting design involving minimal changes to current practices has
been developed through an extensive process of optimisation and experimental studies.

1. INTRODUCTION
1.1 Insulated Rail Joints (IRJs)

IRJs are bolted rail joints containing bonded insulation materials wrapped around each
contacting surface to electrically isolate them. IRJs are essential components in track
circuitries (a simplified diagram of a typical track circuitry is provided in Fig. 1.1) that control
signalling and broken rail identification systems. Joint-less track circuitry systems are
another form of controlling signals; unfortunately, such systems cannot work in tracks
containing switches and crossings and they will not be effective for broken rail identification
either. As crossings and switches are indispensable parts of rail track corridors, track
circuitries containing IRJs will remain the primary form of signalling control system unless a
major revamp of technology occurs (such as GPS based train identification and control);
even then, the existing IRJs will be left in service for decades as infrastructure projects
require significant investment, which will only be slowly realised.
FIGURE 1.1 - Four IRJs form a Track Circuit
In Fig. 1.1 it can be seen that when the ‘circuit’ is unoccupied, the receiver will receive all fed
information from the transmitter – thereby keeping the signal ‘green’. Where wheelsets are
present in between the transmitter and receiver, due to short-circuiting, the receiver will not
receive any information from the transmitter; a similar scenario will also exist when there
are single or multiple distinct rail cracks. The role of each track circuit containing four IRJs
can therefore be regarded as being critical to the safety of train operations.
Failure (electrical or structural) of the IRJs can lead to catastrophic consequences; therefore,
IRJs are regarded as high risk elements and are accordingly maintained through high
standards.
Current designs of IRJs contain a number of components:

 two rail sections cut normal to the axes of the rails
 one end post
 two joint bars (also known as fish plates)

 six (or four) bolts, nuts & washers

 electrical insulation material sheets.
The IRJ design is carried out as per AS1085.12 (2002); a general view of the IRJ assembly is
provided in Fig. 1.2.
IRJs are manufactured in factories in a thermal process in order to wrap the insulation
materials around the surfaces of all metallic components to ensure electrical isolation of
each component. Field assembly of IRJs is not allowed on main lines; however, such a
procedure is permitted in sidings and on temporary works to tackle emergency situations
(ESC220, 2011). Operation in such conditions is subject to severe speed restrictions. Clearly,
field-assembled IRJs should be replaced by the proper factory made IRJ as soon as
practicable to return to normal operation.
FIGURE 1.2 -Components of the Insulated Rail Joints (IRJs)
There is no mechanised method available for the inspection of the IRJs. This is due to the
complex contact conditions between the surfaces of various components both in the vertical
and horizontal planes (the joint bar–rail web contact plane is vertical whilst the bolt shank–
rail web hole contact plane is horizontal; there are many other contact surfaces between
bolt shanks and nuts and washers). Each contact surface can be construed as ‘cracks’ by a
laser or ultrasonic based inspection technique, making it difficult to differentiate between a
real crack in a component (e.g. joint bar) and various interfaces. These techniques (laser/
ultrasonic) work quite well in the identification of internal railhead flaws even under
high-speed operation, but are inadequate for IRJs. A visual inspection method is therefore
considered the most viable; a ‘mechanised form’ of visual inspection is a trolley mounted
camera system trialled by the Transport Technology Corporation Inc., TTCI, USA (Davis &
Akhtar 2006). The images taken by the cameras had to be qualitatively analysed, which
makes the method cumbersome and expensive.
Fig. 1.3 shows typical vertical geometry recorded by a track recording car; higher
irregularities around the IRJs compared to other parts of the track are evident in this figure.
The problem of excessive ballast settlement around rail joints and its subsequent effect on
joint life is discussed in Suzuki et al. (2005). Prompted by such distinctive track degradation
around the IRJs, rail engineers monitor in-service IRJs using digital photography for recording

and qualitative assessment with no follow-up of quantitative analysis. This guide reports a
similar method of digital imaging to illustrate quantification of degradation of IRJs.
FIGURE 1.3 – Typical Track Geometry car Data from an Australian Track
The IRJs exhibit many failure modes. It is believed that ratchetting (cumulative accumulation
of metal under plastic flow, shown in Fig. 1.4) of the railhead material in the vicinity of the
gaps of the rail joint is a failure mode of significant concern to the heavy haul rail industry as
it has the potential to short-circuit due to imminent metal contact culminating in signalling
complications.
FIGURE 1.4 -Ratchetting Failure
Ratchetting is affected by the wheel-rail rolling contact pressure singularity (which is a

theoretical ‘infinite pressure peak’ that can cause damage to any engineering material
available in the industry) as shown in Fig. 1.5. It can be seen that the wheel-rail contact away

from the railhead edge produces pressure defined by the Hertzian theory (with no
singularities); the same wheel when approaching the railhead edge unfortunately does not
obey the theory and causes contact pressure singularity.
Wheel load Wheel load
Contact pressure
Contact surface
Stress concentration
Sharp
Corners
Railhead Railhead
Unsupported or supported edges

(Rail ends)
FIGURE 1.5 - Contact Pressure Singularity
The pressure singularity leads to high levels of stress concentration at the corners defined by
the rail end and railhead top (adjacent to the gap) as shown in Fig. 1.5. It can also be seen
from the figure that away from the gap, the stress concentration exists below the railhead.
Migration of stress concentration from below the railhead top to the top of the gap is the
major reason for the onset of ratchetting in the current design of IRJs.
Other failure modes are related to fatigue/ structural aspects, namely: (1) bolt looseness; (2)
delamination or battering of the end post; (3) bolt hole cracking and (4) joint bar cracking.
These failure modes are predominantly affected by the magnitude and frequency of
application of the axle load, the support conditions and metal fatigue.
With a view to increasing the axle loads and the annual operational throughput, many
improved structural designs of IRJs have emerged in the market. A patent search on rail
joints found hundreds of designs. An analysis of the collection reveals that, from 1903 to
2011 (Page (1903), Click & Duffner (2007) and Akhtar & Davis (2011)), the primary focus of
the rail joint design remained unchanged with the concept being centered on
‘strengthening/ stiffening’ of the components/ assembly. Transport Technology Corporation
Inc., USA (Davis & Akhtar, 2006) have developed designs that possess either increased
tie-plate length and number of bolts or a more supporting mechanism using an additional
saddle design (a design trialled at TTCI is shown in Fig. 1.6).

FIGURE 1.6 - A 12 Bolt IRJ (Akhtar and Davis, 2011)
Australian heavy haul has resorted to solving ratchetting through improvement to materials
(by welding high yield strength materials at ratchetting prone locations) as distinct from
increasing structural strength as practised in the USA (Fig. 1.7).
Weld
Weld Weld
Weld
FIGURE 1.7 - Top Views of High Yield Strength Steel Welded IRJs
Although not specific to IRJs, the findings of Ringsberg et al. (2005) on laser cladding of the
contact zone at the railhead to improve the rail service life have had some effect on the
attempts to minimise ratchetting through high yield material cladding at ARTC (Fig. 1.7).
In some designs, more components are added to further stiffen the assembly. As each
component can fail in several complex mechanisms, adding more components increases the
number of failure modes and hence the risk of failure in addition to complicating the process
of inspection and maintenance. Furthermore, these designs have neither claimed nor are
capable of preventing railhead metal ratchetting failure; this is because they all have sharp
cornered railhead ends in the vicinity of the gaps (rail ends cut orthogonal to the rail axis) –
and hence suffer from the contact pressure singularity as shown in Fig. 1.5.
The designs developed from the Cooperative Research Centre for Rail Innovation (CRC)
sponsored project (the guide is the outcome of the sponsored project) are capable of
eliminating contact pressure singularities through innovative shaping of the railhead edges
in the vicinity of the gaps. Designs incorporating fewer components have been developed to

further minimise maintenance problems. It is shown that wear can be minimised through
suitable selection of material innovations which prolongs the period of re-profiling the shape
of the railhead so that the contact pressure singularity will not re-appear.
Apart from innovations in design, handling and placing of the IRJs in the field are also
important to maximise their performance/ life. The current industry standards (ESC220
(2011), RAP5138 (2006) & RTS3655 (2006)) recommend a ‘suspended’ arrangement in which
the gap of the IRJ is symmetrically positioned to the centerlines of the sleepers. Analysis of
results from the CRC research project also supports this arrangement. However, the TTCI
design of the 12 bolt IRJ is shown (Fig. 1.6) to have positioned the gap on the top of a
sleeper.
Since 2007, the size of the gap has been restricted to 4mm – 6mm in the manufacture of the
IRJs. Several IRJs were manufactured prior to 2007 with gap sizes of 8mm – 10mm which
might still be in service – especially on low traffic passenger lines. Therefore this guide
covers a range of IRJs (4 bolts/ 6 bolts; 4mm gap – 10mm gap; suspended between and
directly supported on the sleepers; with and without a coating of high yield strength
materials on the railhead).
1.2 Outline of the Guide
This guide is divided into five chapters.
The first chapter presents important outcomes from the CRC sponsored project on longer
life new generation IRJs. It also presents brief references to current international practices in
the design, installation and maintenance of IRJs.
The second chapter presents current practices regarding design, manufacture, quality
assurance testing procedures, field handling, placement and maintenance. Variations that
occur on different corridors (for example, passenger, freight, mixed) on each of the above
aspects are illustrated. Actual practices of how sleeper spacing and ballast depth are
ensured where IRJs are located in a new construction and where the IRJs are replaced in an
existing track are highlighted. Subgrade properties (e.g., California Bearing Ratio CBR) and
drainage are excluded from the discussion. Effects of various track maintenance practices on
the IRJs are also covered – in particular, rail replacement (for example, due to rail wear
limits), rail grinding, weld repair of damaged railheads, ballast tamping, sleeper renewals (for
example, damaged sleepers), ballast cleaning and subgrade strengthening.
The third chapter presents some new ideas for improving the performance of IRJs through
appropriate track design and maintenance practices in the vicinity of IRJs, which can be of
use to track engineers and track maintenance crews. Whilst designing tracks, it is suggested
that the following parameters should be given careful consideration for improved
performance of IRJs: (1) sleeper spacing, (2) ballast depth, (3) identification of ballast
pocket/ hanging sleepers, (4) mechanised and manual railhead grinding, and (5) tamping

frequency. For improved track maintenance, the parameters discussed include: (1)
methods of ensuring accurate positioning of IRJs, (2) methods of ensuring correct spacing of
sleepers allowing for the length of the IRJ assembly commensurate with the rail neutral
temperature, (3) methods of recording the conditions of the components of the IRJs and the
surrounding supports – sleepers and ballast, (4) strategies of grinding and profiling the
railheads closer to the gap in the IRJs and (5) weld repair of the railheads in the IRJ
assemblies.
Chapter 4 describes longer-life new generation IRJs that minimise impact. IRJs that could
eliminate ratchetting and hence the risks to electrical short-circuiting due to potential
metal-to-metal contact are also highlighted. Suggestions for manufacturing, installation and
maintenance of these new generation IRJs are provided.
Chapter 5 provides a summary and conclusions.

2. CURRENT STATE OF IRJs
There are approximately 50,000 IRJs in the Australian rail track, an average of 3.33 per route
km; by discounting half of them in sidings and yards, conservatively it can be estimated that
in approximately 15km of railroad, 25 IRJs might be present on average. To provide another
perspective, for a train travelling at a moderate speed of 90km/h (on average) each wheel
would encounter an IRJ in about 24 seconds of travel. Damage to the IRJs and the associated
railroad components, as well as train dynamic problems and safety issues, illustrate the
significance of the economic problem for the rail industry; there is thus a need to monitor
and understand the deterioration process of IRJs and surrounding railroad components for
better management of heavy haul corridors riddled with IRJs.
2.1 IRJs in Coal Corridors
In the Australian heavy haul coal corridors, IRJs containing six bolted joint bars and AS60Kg
rails are commonly used. These IRJs are supported in a suspended manner between
pre-stressed concrete sleepers spaced at 600mm resting on a ballast bed of 600mm
thickness; both standard gauge and narrow gauge tracks use the same track standards
although the axle loads can be higher in standard gauge tracks due to larger volume coal
wagons.
In formations on clay/ expansive soils, ballast tends to penetrate through the subgrade,
especially where the capping layers are weak or damaged. With a view to maintaining the
desired depth, more ballast is added (re-surfacing operation). The added ballast further
moves down enlarging the penetrated ballasts, which forms ‘ballast pockets’ as shown in
Fig. 2.1
Ballast Layer
Penetrated ballast Capping Layer
FIGURE 2.1 - Ballast Pockets
The ballast pockets usually act as a barrier to drainage leading to water-logging and further
deterioration of the track; this process culminates in pumping of mud from the ballast
pocket to the clean, re-surfaced new layer of ballast. As such, although the depth of ballast
may be detected as large through ground penetration radar (GPR) measurement, the track
modulus at sections containing the ballast pocket will be lower due to poor subgrade.
Therefore, IRJs in the vicinity of the ballast pocket would be highly responsive to loading and

can fail due to joint bar cracking prior to the onset of ratchetting. In heavy haul operations,
these IRJs are more vulnerable than the ones on firmer subgrade and cleaner ballast
surrounded by a good drainage system and hence should be maintained with higher priority.
As stated, the typical spacing of sleepers and the depth of the ballast layer are 600mm each,
which is achieved when a new track is constructed due to excellent quality assurance
systems and existing construction technology and equipment. However, since the first day
of commissioning, each track component will move vertically, laterally and longitudinally.
Therefore, when an IRJ is to be replaced, care should be taken to order a slightly longer
length IRJ (the normal length is between 3.2m and 4.3m) to ensure the welded ends are fully
cut out when the replacement IRJ is welded into position.
2.2 IRJs in Passenger-Freight Corridors
AS50kg/m or AS40kg/m rail is used in passenger, freight or mixed passenger – freight

corridors where the axle loads are in the order of 19T – 22T; some old AS53kg/m rails still
exist in these corridors. An IRJ found positioned on sleepers that violate the Railcorp
guidelines (ESC220, 2011) is shown in Fig. 2.2.
FIGURE 2.2 - A Four Bolt IRJ positioned on the Sleeper
Four or six bolt IRJs are found in both the passenger/ freight corridors as well as the coal
corridors, although the current practice is to use only the six bolt IRJs. Considering all these
variations in rail track, a common approach for installation and maintenance of IRJs would
therefore be appropriate, especially when a ‘national grid’ approach is taken with train
operators from one part of the nation operating on other parts of the rail track
infrastructure – this guide is an attempt to provide such guidelines.
2.3 IRJs in Iron Ore Corridors
Iron ore is perhaps the heaviest mineral transported through rail wagons; the axle load of
these wagons is typically 36T. As such, these corridors use the heaviest rail (AS68kg or
equivalent). The Western Australian iron ore corridors are privately managed with dedicated
lines; typically, IRJs manufactured in the USA are used in their track (Fig. 2.3).

FIGURE 2.3 - IRJ made in the USA with thick joint bars
2.4 Quality Assurance in the Manufacture of IRJs
The manufacture of IRJs is subject to rigorous quality assurance (QA) procedures as laid out
in AS1085.12 (2002), which is briefly highlighted in this section. The QA process is designed
to ensure the adequacy of the IRJs to remain electrically insulated after fatigue loads are
applied both in the vertical and longitudinal directions simulating the wheel loads and
thermal movement loads respectively. The electrical insulation is also checked after a
stationary load is applied causing deformation within the prescribed limits in Table E1,
AS1085.12 (2002).
In all tests, the IRJ is positioned on supports spaced at 600mm centre-to-centre (c/c) in the
vertical position without the in-field 1 in 20 cant. The simulated vertical wheel load is
positioned symmetric to the vertical axis of symmetry. Similarly, the simulated thermal
movement load is applied along the centroid of the rail section to ensure pure axial
elongation/ shortening without any torsional rotation.

A Pull-Apart Test is conducted by first applying the specified longitudinal load (for AS60kg
rail, 1130 kN) along the centroidal axis of the rail followed by a vertical load (for AS60kg rail,
210kN) on the top of the gap for 15 seconds, which is then released and re-applied for 15
more seconds. After all loads are removed, the IRJ is checked for evidence of delamination
or any permanent deformation followed by testing for electrical insulation.
A Load-Deflection Test is carried out on an IRJ by applying a vertical load (for AS60kg rail, the
vertical load is 1100kN) on the top of the gap whilst measuring the centre span vertical
deflection, which should lie within a prescribed limit (for AS60kg rail, 20mm). The load
should then be slowly increased until failure of the IRJ or 1900kN – whichever occurs first.
Where no failure occurs, an electrical insulation test is carried out.
A Fatigue Test is carried out on IRJs supported with a span of 1100mm and a prescribed
maximum vertical load (for AS60kg rail, 245kN) is applied smoothly from just under 10kN
(i.e., for AS60kg rail, the load range is 10kN - 245kN) at a frequency of 5Hz – 10Hz until 3
million cycles are achieved. At every 0.5 million cycles, the fatigue test is paused and the
electrical insulation test is conducted.
An electrical insulation test is conducted on an IRJ specimen (virgin or tested for pull-apart/
load-deflection/ fatigue) by applying a 500V DC and measuring the current (or resistance)
between each joint bar and rail, and bolt and rail.
From the information in this section, it can be seen that IRJs are manufactured with great
care regarding their quality, with a clear focus on their structural adequacy to withstand the
wheel and thermal loads as well as their integrity for electrical insulation. Ideally, these IRJs
should perform fairly uniformly; unfortunately their variability in life is quite excessive – as
low as 50MGT to as high as 500MGT (still well below the 2000MGT life of continuous rails
that are only removed from service if they have exceeded the wear limit). Developing
strategies to manage/ maintain these IRJs (given their condition assessment could not be
supported by technologies applied to continuous rails) is quite challenging, if not impossible
but will not be cost effective. The major contributor to the variability of their life is the
condition of the track on to which they are laid. The manner in which they are laid in the
field (proximity of the gap to the sleeper, the ballast support under the sleepers adjacent to
the gap, the thickness of the ballast and the California Bearing Ratio (CBR) of the subgrade)
varies significantly due to a lack of uniform guidelines.
2.5 Handling and Placement of IRJs on Rail Track
The IRJs are manufactured with lengths from 3.2m – 4m as per demand and transported to
the site, where they are welded into place. Whether the track is new or an existing one,
the procedure of determining the appropriate length of the IRJ assembly should account for
the ‘rail neutral temperature (RNT)’, which is the temperature at which the rail is subject to
zero longitudinal load. Depending on the expected temperature on the day of renewal, the
length of the IRJ is determined and ordered to be manufactured. During installation of the

IRJ on an existing track, the temperature must be accounted for in determining the exact
length of cut where the assembly should be welded into place. To ensure the previous weld
was fully removed, a longer length of the IRJ assembly would be required compared to the
previously installed IRJ assembly.
Adjustments to the length of the IRJ assembly allowing for the RNT quite often can inhibit
positioning of the gap of the IRJ at the most desired location of the mid-span between
adjacent sleepers; this is because the sleepers would have often moved, also the lengths of
cutout and new IRJ assemblies may have minor differences. In some tracks where sleeper
movement is excessive, the gap of the IRJ assembly could end up being positioned very close
to or on the top of the sleepers. To avoid such a situation, this guide suggests re-positioning
the sleepers whilst replacing the failed IRJs.
In very locally damaged IRJs (chipping or squashing of railhead close to gaps), field welding
to smooth the wheel passage is sometimes carried out – especially when all other
components of the IRJ assembly are in good condition. The field welding process should
ensure integrity of the electrical insulation.
In some instances, sleepers close to the gap of the IRJs might get damaged that require
replacement. Lifting the IRJ assembly to facilitate removal of the damaged sleepers and
insertion of the new sleeper must be handled carefully to ensure the gap of the IRJ assembly
is retained symmetric to the adjacent sleepers.
Ballast cleaning is required when excessive fouling is noticed due to particle breakage or
dust deposition due to spillage from coal wagons or airborne particles. The IRJ assembly and
the local sleepers are lifted up to facilitate the cleaning operation. When the ballast is
cleaned and the sleeper – rail assembly is laid back, care should be taken to ensure the gap
of the IRJ assembly is symmetric to the adjacent sleepers.
Strengthening of subgrades where the IRJs are present may not be a regular event, but a
possibility. When the subgrade is stabilised/strengthened using a lime-slurry injection or a
similar technique, the sleepers should be so adjusted to ensure the gap of the IRJs is
symmetrically positioned to the adjacent sleepers.
Some routine maintenance can also affect the requirement of appropriate handling of the
IRJs, These include:
 rail replacement due to exceeding rail wear limits
 railhead grinding
 ballast tamping.
Rail replacement close to IRJs may become necessary in some instances where the IRJ is
relatively new compared to the continuously welded rail (CWR) being quite worn. There are
limits to worn rail dimensions set by various rail owners; for example, the CoP (2013) set the
condemning limit of loss of height of AS60kg/m head hardened rail as 20mm with the
remaining height of railhead measured as shown in Fig. 2.4 as 24mm. These geometric limits

are set to ensure that the wheel flange would not get hit on fishplate tops; should fishplates
be fully eliminated (through another viable technology), it could be possible to further
increase the head height loss condemning limit, thereby saving billions of dollars. With the
loss of railhead height, the moment of inertia of the rail reduces and hence the displacement
increases; a limit of 9mm vertical displacement under the passage of loaded wheels is also
specified in the CoP (2013).
FIGURE 2.4 - Rail Wear Limit Parameters (CoP 2013)
When a worn continuous rail adjacent to a newly installed IRJ is replaced, potentially the gap
location may get shifted from the symmetry of spacing between the adjacent sleepers.
Repositioning of sleepers should be carried out to ensure proper positioning of the IRJs.
Railhead grinding is carried out at regular intervals as a preventative strategy (to ensure
removal of rolling contact fatigue surface cracks prior to attaining threshold limits of rapid
progression as happened in Hatfield, UK). IRJs are not given any special consideration during
the grinding operation; they are ground similar to the CWRs.
Ballast tamping is another maintenance activity that occurs regularly to minimise lateral
spreading of the ballast layer. Tamping forks penetrate to approximately half the depth
(300mm) of the ballast layer in between the sleepers. The tamping machine operation is
calibrated for sleeper spacing of 600mm; any other spacing (especially less than 600mm)
should be discussed with the operators. During the tamping operation the sleepers could
move longitudinally. Where IRJs are present, care should be taken to ensure appropriate
positioning.

3. TOWARDS IMPROVING THE PERFORMANCE OF THE IRJS
The current design of IRJs consists of insulation materials made of nylon or similar material
whose elastic modulus is about 10% of the elastic modulus of the steel. These materials are
inserted in the gap between the two sections of the rails and are termed end-posts. The IRJs
are manufactured in factories in accordance with the relevant clauses of the respective
national standards (such as the AS1085.12 (2002) in Australia) and carefully transported to
the site and installed in the rail track structure; in spite of this care, due to the discontinuity
of the elastic stiffness between the insulation materials and the rail steel, impacts occur due
to the passage of the loaded wheels across the end post gap, which elevates the severe
stress concentration at the top of the railhead in the vicinity of the joint gap. This chapter
reports on the examinations carried out to understand the performance of the current
designs of in-service IRJs with a focus on quantification for potential application to
engineering decision making. The strategies employed were: (1) establish a quantitative
measure of progressive degradation of the IRJs and the local zone of the track that supports
them; (2) capture the strains at critical locations of the IRJs, wheel-rail contact impact force
and deformation of the local surroundings under the passage of loads; (3) develop numerical
models to simulate the strains and deformations captured in (2) above; and use the
predictions of the cumulative strains of the numerical models to determine the life of the
IRJs affected by ratchetting.
3.1 Monitoring Progressive Degradation of the In-Service IRJs:
In-service IRJs degrade progressively due to repetitive passage of the ongoing traffic; the
throughput along a corridor is monitored in terms of cumulative traffic load expressed in
million gross tonnes (MGT). Maintenance is scheduled based on the number of MGTs the
track has serviced. In this method, the static wagon load is considered; the dynamic load
generated at the IRJ is transmitted to the sub-structure layers causing deterioration of
ballast leading to a higher rate of sleeper settlement in the vicinity of the gap of the IRJs not
being considered. This local rail track geometry deterioration adjacent to IRJs consequently
increases the dynamic impact load that leads to a further increase in instantaneous
deflection under the passage of the loaded wheels at sections where the IRJs are present.
From an operational safety perspective, a sudden failure of the joint bar due to high impact
loading can lead to disastrous consequences such as derailment. Even when there is no
failure of joint bars, higher deflection of the IRJ assembly can generate unacceptable
vibration of the rolling stock components forming a vicious circle with the one aggravating
the other, which could culminate in poor ride quality. The aspects of safety (IRJ failure) and
serviceability (poor ride quality) are related to the progressive degradation of the track
around the locations of the IRJs. Therefore, with a view to quantifying progressive
deterioration around the locations of the IRJs, four IRJs of varying levels of service life were
identified and monitored using digital photography and total station survey measurements.
The collected data were then analysed to quantify the trends in joint gap narrowing and
track geometry deterioration (progressive increase in dip, cross level and twist) around the
IRJs.

Four IRJs were selected as shown in Figure 3.1 based on the traffic condition and the initial
damage level of their railhead. Two of the IRJ’s were in low traffic lines leading to a nearby
yard and the other two IRJ’s were in a high traffic suburban passenger railroad network. One
each in the low and the high traffic line exhibited damage to the railhead and the other was
in relatively good condition. Whilst IRJ-1, IRJ-2 and IRJ-4 were suspended symmetrically to
the ties, IRJ-3 was positioned close to a sleeper. Geographically, all four IRJs were located
within one kilometre distance; the subgrade condition can, therefore, be regarded as quite
similar. The structural design of these four IRJs was similar with a four-bolt connection and
epoxy-fibreglass insulation end post. The yard line railroad structure is tamped infrequently
compared to the suburban line railroad.
FIGURE 3.1 - IRJs for Monitoring of Progressive Deterioration (Zong et al. 2013)
At the start of the monitoring, IRJ-1 and IRJ-2 exhibited damage; the railhead metal plastic
flow on both sides of the joint gap of IRJ-1was noticed, while in IRJ-2 the railhead cracking
and insulation material crushing were obvious. The other two IRJs (IRJ-3 and IRJ-4) exhibited
relatively better condition in terms of the railhead running surface in the vicinity of the end
post gap.
One each of the IRJs in the low and high traffic tracks exhibiting damage and two IRJs in each
of the low and high traffic tracks were selected, making comparative study meaningful for
practical interpretations. Data were collected for two purposes: (1) to monitor progressive
narrowing of the end post gap and (2) to monitor the extent of the progressive deterioration
of track geometry around the IRJs. A total of six site visits were carried out over 500 days.
3.1.1 Progressive narrowing of the gaps of the IRJs:
A scaled stencil (250mm×80mm) was positioned symmetrically over the end post gap of the
IRJ, providing a reference scale, and a digital camera was used to take multiple photos for
further graphical analysis. The camera was held at the same height at all times. Fig. 3.2
shows a sample photo of IRJ-1.

FIGURE 3.2 - Digital Imaging of IRJs (Zong et al. 2013)

Analysis was undertaken from the successive measurement of the end post gap, progressive
narrowing, and the trend of joint gap narrowing due to accumulated traffic over the period
of observation. AutoCAD was adopted for measuring the local geometry of the gap of the
IRJs. From every site visit, three stenciled photographs for each IRJ were selected based on
the best photo clarity and light exposure. The photos were imported into AutoCAD and
scaled relative to a 1:1 ratio to the scale of the stencil as shown in Fig 3.2 to ensure
consistency of the dimensions throughout the entire photo analysis. AutoCAD was used for
outlining the joint gap area (shaded in Fig 3.2). The area Arail  gap was divided by the length
d gap length to determine the average width of the joint gap wave . The change in wave as a
function of time is plotted in Fig. 3.3 for the four IRJs; these figures illustrate the progressive
changes to the width of the joint gap over the six site visits spanning 500 days. Graphical
analysis was completed on three photos of each of the four joints over the six site visits
totaling 72 sets of graphical analyses. As IRJ-2 was fully replaced between the fifth and sixth
visits (as the damage to IRJ-2 caused a brief signalling failure), data for the sixth visit of this
IRJ were omitted from the analysis.
It can be seen from the slope of the trend line in Fig. 3.3 that the average widths of the end
post gaps of these four joints have progressively reduced. Over the 500 day period of
observation, IRJ-1 and IRJ-2 that had visible initial damage exhibited gap width reduction of
1.001mm and 2.406mm respectively, while the two IRJ’s with a good appearance (IRJ-3 and
IRJ-4) exhibited smaller gap width reduction of 0.927mm and 0.394mm respectively.
Although the gap width reduction data showed variability, the trend line clearly indicates
progressive narrowing of the gap; further, the slope of the line illustrates the initial damage
and the adjacent environment.

FIGURE 3.3 -Progressive Closure of the Gap of the IRJs (Zong et al. 2013)
Although IRJ-3 and IRJ-4 exhibited similar physical conditions (no damage) in the initial
inspection, it is interesting to note that IRJ-3 (in a low traffic yard line) exhibited a higher
rate of gap closure than IRJ-4 (in a high traffic line); further investigation revealed that whilst
IRJ-3 was located closer to a sleeper, IRJ-4 was suspended symmetrically to two sleepers.
Locating the IRJ close to a sleeper induces higher impacts (Pang 2007). This observation
supports the Railcorp recommendation (ESC220, 2011) of suspending the IRJs symmetrically
to sleepers and contrasts the US recommendation of supporting onto a sleeper (Akhtar and
Davis 2011).
3.1.2 Progressive settlement of sleepers adjacent to the IRJs

To examine the progressive settlement of sleepers on either side of the end post gap, the
vertical levels of six sleepers (three on either side) were measured from a datum on each of
the site visits using total station survey equipment (See Fig. 3.4).

FIGURE 3.4 - Location of Measuring Levels of Sleeper Tops (Zong et al. 2013)
A typical visit involved positioning the total station on a pre-identified location; as the height
of the total station would not be the same on each visit, a datum (concrete plinth of a
signalling post) was selected. A reflector was held at the datum to measure the x, y, z
coordinates with reference to the height of the total station. The reflector was then moved
to each target of the ties to measure their respective coordinates. Both ends of the six
sleepers were monitored for each IRJ; thereby 36 points were measured on each visit. Firstly,
the coordinates (x, y, z) of the 12 selected measuring locations were marked as A1 – A6 and
B1 – B6 for each IRJ as shown in Figure 3.4. The six measuring points adjacent to the rail
containing the IRJ were denoted as A1 to A6 from left to right for those viewing from the
field side, and the other six closer to the rail without the IRJ were denoted as B1 to B6. As
noted previously, IRJ-2 and IRJ-4 were positioned opposite each other on the same section
of the railroad line; therefore, the measuring points adjacent to the rail containing IRJ-2
were denoted from A1 to A6; and B1 to B6 for the other six points on the side of IRJ-4.
To ensure accuracy of the measured data, the configuration of the total station relative to
the benchmark point was maintained as constant on every site visit. In the Cartesian
coordinate system, the origin of the xoy plane was directly located at the centre of the
calibrated total station, denoted as (0, 0, z) as shown in Fig. 3.4. At the benchmark point (a
concrete pedestal of an electrical signalling post), the z value (elevation) was set as
100.000m. The x axis was directed away from the origin towards the benchmark, whilst the
perpendicular direction was the y axis.
The total station survey data collected from the sleeper top locations were analysed to
evaluate the change in geometry conditions around the section containing the IRJs through
measuring the vertical position of the sleeper tops. Based on the survey data obtained from
the marked points (A1 – A6 and B1 – B6) of the sleepers, the trend in degradation of track
geometry (support settlement caused dip, cross level and twist) close to the end posts of the
IRJs over time was monitored.
The location of the benchmark and the total station in the xoz plane were kept constant and
only the elevation of the total station changed between visits; as the total station elevation

was considered the origin height and the total station reading was ‘zeroed’ (z=0) at different
heights for each visit, the absolute height difference between each measuring point along
the sleeper tops and the benchmark was obtained by subtracting the elevation data of the
measured point from that of the benchmark. Finally, a plot comparing the change in the
absolute height difference between the various visits was created to reveal the condition of
supporting ties close to the end post gap in comparison with the ties under the continuous
welded rail sections.
It should be noted that IRJ-1 and IRJ-3 were located in different sections of the low traffic
yard line and both were positioned in one rail adjacent to the measuring points A1 to A6.
Opposite these joints (IRJ-1 and IRJ-3), the measuring points B1 to B6 were aligned to the
Continuously Welded Rail (CWR). IRJ-2 and IRJ-4, on the other hand, were located in the
same section opposite each other on the high traffic line, with points A1 to A6 adjacent to
the IRJ-2 and the points B1 to B6 adjacent to IRJ-4.
It is striking to note that the average progressive settlements (APS) of the sleepers below the
two CWRs were noticeably smaller than that of the four IRJs. Over the period of 500 days,
the maximum of the APS of ties below the CWRs was 3.5mm, whilst the minimum of the APS
of the sleepers below the IRJs was 3.67mm.
From the APS of the sleepers below IRJ-1 and IRJ-3 (both in low traffic yard lines with
infrequent maintenance), it can be observed that the higher the initial damage of the IRJs
(such as in IRJ-1), the larger will be the APS of the sleeper top (4.3mm under IRJ-1 against
3.7mm under IRJ-3). This observation is confirmed through comparison of the APS of the
sleepers below IRJ-2 and IRJ-4 (both in high traffic lines) – 5.47mm under the damaged IRJ-2
against 4.91mm under the relatively good IRJ-4.
The sleepers in high traffic lines undergo higher APS compared to those in low traffic lines,
whether or not the IRJs are damaged (5.5mm vs. 4.3mm) or undamaged (4.9mm vs. 3.7mm).
The higher APS might be attributed to the location of the IRJs opposite each other that
potentially doubles the wheel impact. The APS of the sleepers under the CWRs is affected by
the APS of the IRJs opposite them, as can be seen from IRJ-1 and IRJ-3 and the
corresponding CWRs.
The differential settlement thus adds to the design cross levels and twists where the IRJs are
present. From the data, by assuming the rail tops and tie tops are separated by the height of
the rails, the maximum cross level deviation in IRJ-1 was determined as 2.23mm (which is 1
in 897 as the spacing between Ai and Bi was 2000mm). The maximum progressive twist in
the observed zone (B1 to A6) in IRJ-1 was 1 in 1917. These irregularities, therefore, demand
higher attention to maintenance of the rail track structure around the IRJs, especially the
ballast depth. In the absence of such careful maintenance, the railroad geometry is likely to
deteriorate faster with larger dips, cross levels and twists leading to potential derailment
and remedial work over the whole of the network.

In summary:
 For sleepers below IRJ-1 and IRJ-3 (in low traffic lines), the higher the initial damage
of the IRJs, the larger will be the APS of the sleeper top (4.3mm under IRJ-1 against
3.7mm under IRJ-3).
 For sleepers below IRJ-2 and IRJ-4 (in high traffic lines), the APS is 5.47mm under the
damaged IRJ-2 against 4.91mm under the relatively good IRJ-4.
 Sleepers in high traffic lines undergo higher APS compared to those in low traffic
lines, whether or not the IRJs are damaged (5.5mm vs. 4.3mm) or undamaged
(4.9mm vs. 3.7mm).
 Maximum cross level deviation in IRJ-1 was 2.23mm (which is 1 in 897 as the spacing
between Ai and Bi was 2000mm). The maximum progressive twist in the observed
zone (B1 to A6) in IRJ-1 was 1 in 1917.
3.2 Performance of In-Service IRJs under the Passage of Wheels
Rail engineers are well aware that the IRJs with similar structural and material properties
subject to similar traffic conditions exhibit variable performance in different locations of the
same corridor; yet no model is available in the open literature for quantification of the
structural behaviour of IRJs due to track parameter changes. Two extensive field tests were,
therefore, conducted on two different in-service Australian heavy haul corridors. A view of
these test setups is presented in Fig.3.5.
IRJ 3 IRJ 4
IRJ 1 Ref. Rail IRJ 2
(a) Test Setup for Track Input on the Behaviour of the IRJs
(b) Test Setup for Comparative Studies of Square and Inclined IRJs
FIGURE 3.5 - Field Test Setup of Instrumented IRJs

3.2.1 Track input to wheel-rail contact impact forces at the IRJs

Two IRJs from the test setup shown in Fig. 3.5(a) were chosen for understanding the effect
of changes from track parameters on the contact-impact forces under the passage of loaded
wheels across the gap of the IRJs. Fig. 3.6 shows the two selected IRJs. The two IRJs are
spaced 40m away from each other and are strain gauged appropriately to capture the
moving wheel load exactly. Such a type of strain gauging at the web of the rails is used for
wheel load detection in many systems; however, their usage to detect the impact at the gap
is the first of its kind and is published in Askarinejad et al. (2013). A reference rail (CWR) was
also strain gauged using the same technique.
FIGURE 3.6 - IRJs Tested for Track Input Effects (Askarinejad et al. (2013))
The main test parameter is the change to track modulus. As changes to ballast layer depth
and subgrade modulus are expensive, the sleeper spacing (albeit knowing the ballast
tamping could be problematic) was modified as shown in Fig. 3.6 to affect the track
modulus. IRJ-1 and the Reference Rail rested on a track support of 600 mm sleeper spacing,
whilst IRJ-2 rested on six sleepers with a spacing of 350 mm. Track modulus is defined here
as the track stiffness per sleeper spacing, the definition of which makes IRJ-2 rest on a track
of 1.71 times higher modulus compared to IRJ-1.
Traffic conditions in the test site were fairly uniform. The line was uni-directional and was
servicing only loaded coal trains with a 30 tonne axle load (30TAL). The track substructure
was uniform across the test site. The track bed was 450 mm thick including a ballast layer of
300 mm and a capping layer of 150 mm. The particle size, gradation and other properties of
the ballast material at the test site were in accordance with the technical specification TS

3402 (2001). The two IRJs were produced in a factory with special care given to embedding
strain gauges. The sensors on the IRJs and the surrounding track components are shown in
Fig. 3.7.
FIGURE 3.7 - Sensors Used in the Field Experiment on IRJs
A portable Data Acquisition system (DAQ) was connected to the plugs in the trackside
terminal box at the time of data logging. The data acquisition plan was programmed in
LABVIEW with a sampling frequency of 10,000 Samples/s. This sampling frequency ensured
that all of the high frequency dynamic impacts and the corresponding structural response
were recorded during the measurement. With this sampling frequency, a large amount of
data were generated for each channel requiring a suitable and efficient method to sort
through the data and extract the peak values and localised time histories for further
examination.
Data on the wheel-rail forces, rail/joint-bar strain, sleeper acceleration and ballast pressure
signatures in the time domain were determined from the experiment. The data were then
systematically analysed to compare the relative structural merits of two IRJs subjected to
similar real-life coal traffic loading. An experiment of this scale on IRJs has not been
previously carried out; therefore, the result provides new information and hence is
published in Askarinejad et al. (2013).
To examine the accuracy of the developed shear bridge load indicator, a train was passed
through the test site with an average speed of 7 km/h (considered static) and the wheel
loads were measured using the shear bridge in the reference rail. The wheel loads thus
measured were compared with the wheel loads reported by a commercial train weighbridge;

the two measurements are shown in Fig. 3.8, which shows a good correlation (R2=0.91) – or
dependable wheel-rail contact impact force determination.
FIGURE 3.8 - Calibration of Wheel-Rail Contact Impact Force

Data analyses were conducted using the DIADEM commercial data analysis software. The
peak signals from each sensor were extracted and analysed. Prior to extraction of the peak
signals, the recorded data were subject to a low pass filter to remove the high frequency
noise signals. An ideal low pass filter removes the signal contents at frequencies higher than
the cut-off frequency and doesn’t affect the signals at frequencies lower than the cut-off
frequency. For each sensor, the filtered and unfiltered signals were plotted and compared to
examine the accuracy and suitability of the employed filtering algorithm. Enlarged time
series signatures of wheel-rail contact force divided by the static wheel load from the
weighbridge, P0 , for the reference rail and an IRJ are shown in Fig. 3.9.

Reference Rail
I
RJ1
FIGURE 3.9 - Enlarged views of Wheel-Rail Contact Force Signatures

Additional data for approximately seven seconds of the passage of wheels across the test
site are presented in Fig.3.10. The data show that the maximum P for IRJs is 210kN, whilst
that of the reference rail is 150kN. The  P  ratio for IRJ is thus 1.50 whilst for the
 P0 
reference rail it is 1.07. Therefore, the IRJs can be said to add 43% dynamic load (or
impact). The increase in track modulus of 42% (determined from 600mm and 350mm
spacing) under IRJ-2 (relative to IRJ-1) had no distinct effect on the impact – which is an
interesting and important observation that can have a significant bearing for practical
applications. Later in this chapter it will be shown that the higher the track modulus, the
lower will be the sleeper deflection (and hence the joint bar strains), which would prove that
whilst the structural responses (flexural strains and the surrounding structural component
accelerations and deflections) are affected by the track modulus, the wheel impact load at
the IRJ gap is not noticeably affected. The reasoning for this is that the gap itself is a major
contributor to the impact and any additional ‘track input’ to the impact is insignificant.
Therefore, where impact is a serious track problem, trying to solve that through higher track
modulus will not be successful; reduced gap size is one means of reducing impact as shown
in Chapter 4.

FIGURE 3.10 - Typical Wheel-Rail Contact Impact Force Signatures
3.2.2 Track input to joint bar bending strains:

The bending strains measured from both sides of the rail web and joint-bars (gauge and field
sides) were averaged to eliminate the effect of eccentricity in wheel position; only the
averaged strain data were used in the analysis reported in this guide; signatures for
approximately seven seconds of the passage of wheels across IRJ-1 and IRJ-2 are presented
in Fig.3.11; the measured strains were non-dimensionalised using the yield strain of the
railhead steel (911MPa – as measured, see Section 3.3.2 of this Chapter). IRJ-1 has
consistently exhibited higher joint bar strains (top-compression and bottom-tension) –
consistent with the expectation of the lower track modulus effect. A 33% increase in strain in
IRJ-1 is observed (compared to IRJ-2) with the 42% change to track modulus – the lower the
modulus, the higher the IRJ component strains. The peak joint bar bottom strains for 100
wheels are shown in Fig.3.12. IRJ-1 consistently exhibits higher peak bending strains (due to
relatively lower track modulus) with higher variability than IRJ-2. One possible explanation of
the higher variability is the higher frequency of wheel passage over sleepers adjacent to
IRJ-2.

FIGURE 3.11 - Typical Joint Bar / Rail Bottom Longitudinal Strain Signatures
Wheel Count
FIGURE 3.12 - Joint Bar Top Longitudinal Strain Signatures at two IRJs
3.2.3 Track input to response of sleepers adjacent to IRJs
Fig.3.13 presents a typical time history of the vertical acceleration recorded on top of the
sleepers adjacent to IRJ-1 and IRJ-2; the reference rail signature is also shown. The peaks
correspond to the passage of the wheel over the relevant sleeper.

FIGURE 3.13 - Typical Sleeper Acceleration Signatures Adjacent to IRJs & Rail
It can be seen that the peak accelerations of sleepers adjacent to both IRJs are much higher
than that of the reference rail, which shows that the effect of the presence of gaps in the
IRJs excite the surroundings more than the CWR. The sleeper acceleration of IRJ-2 is lower
than that of IRJ-1, which can be attributed to the higher track modulus (42% increase in
track modulus due to reduced spacing of sleepers under IRJ-2). Higher track modulus is thus
shown to reduce acceleration of track components (which will subsequently improve their
life).
Acceleration is easier to measure relative to deflection; however these two measusres are
related. Through double integration Barati (2013), amongst others, has shown that the
acceleration signatures can be converted to deflection signatures.
Fig. 3.14 shows an example of the vertical deflection profile of a sleeper under the reference
rail calculated from the measured acceleration data. Loaded coal trains with consistently low
speeds of around 30 km/h were passing over the instrumented test section.
An average of 3.2 mm has been calculated for the vertical sleeper deflection in this case.

FIGURE 3.14 - Vertical deflection profile of a sleeper under Reference Rail

Wheel positions (two back wheelsets of rear bogies and the first two wheelsets of the front
bogie of the following wagon) in a block of four peaks provide better confidence in the
results. Similar deflection profiles of vertical sleeper deflections have been determined from
the measured sleeper-top accelerations in the nearby Whittingham track site (as part of a
major geogrid project – R3.117) with an average displacement of just 1.7 mm as shown in Fig.
3.15. Lower deflections observed in the Whittingham track site compared to those
calculated for the Hexham track site are directly attributed to the relatively newer and well
maintained track in Whittingham. This observation is supported by the fact that in
Whittingham the average speed of coal trains was 80km/h, whilst for the same coal trains on
the Hexham track the speed was restricted to 50km/h; therefore, Whittingham must have
had a higher track modulus compared to the Hexham track site. It is observed that the
vertical sleeper deflections in the Hexham track site are almost double those calculated for
the Whittingham track site (Fig. 3.15), which could be attributed to (i) train speed, (ii) axle
load, and (iii) track stiffness. Trains were indeed running at higher speeds in Whittingham
than in Hexham, which would have contributed to a higher dynamic load. In spite of the
higher loads, the deflection was only 53% of that calculated from the Hexham site
accelerations. Therefore, it can be ascertained that the track modulus of the Whittingham
site should have been more than twice that of the Hexham site. Higher track modulus also
contributes to less scattered peaks in the Whittingham track site.

Hexham Whittingham
4
Peak sleeper deflection (mm)

3
0
0 20 40 60 80 100
Wheel number
FIGURE 3.15 - Vertical deflection of sleepers in Hexham and Whittingham Sites

The order of the peak vertical sleeper deflections measured from these research projects
(Hexham & Whittingham) are consistent with the results reported for sleeper deflections
from a series of tests conducted by Cox (1995) in Dalby, England. The reported maximum
vertical deflection of sleepers in Dalby where 4 m deep piles were driven to support the
Linear Variable Deferential Transducers (LVDTs) was 1.2 mm. Priest and Powrie (2009)
reported 0.8 mm of vertical sleeper deflections from a Class 373 Eurostar passenger train-set
in England. Doyle (1980) presented a diagram where a vertical track deflection of 10.16 mm
is considered as the limit for a track which deteriorates quickly.
The double integration method, having been seen to provide sensible deflections of CWR
track, was employed on the four IRJs installed at the Hexham test site; the deflections
determined are shown in Fig. 3.16 for 100 wheel passage. The maximum deflections
observed from the four IRJs are in the order of ~6mm maximum. Consistently, the sleepers
resting on a higher track modulus (due to reduced sleeper spacing) provided lower
deflection; a 42% increase in track modulus reduced deflection by about 50%.

Enlarged View Enlarged View
FIGURE 3.16 - Vertical deflection profile of sleepers under the four IRJs

3.2.4 Track input to ballast pressure beneath IRJs:

Earth pressure cells of a hydraulic type, capable of measuring the dynamic pressure, were
selected for this experiment. The pressure cells were placed in the ballast directly below the
sleeper to measure the sleeper-ballast pressure adjacent to the IRJ and the reference rail. In
this type of pressure cell, two flat plates are welded together at their periphery and are
separated by a small gap filled with a hydraulic fluid. The applied load squeezes the two
plates together, thus building up pressure inside the fluid. The accuracy and measurement
range of this pressure cell model was suitable for capturing the maximum expected
sleeper-ballast contact pressure generated under a coal train. The time history of
sleeper-ballast pressure beneath the two IRJs and the reference rail for about seven seconds
of a train passage is presented in Fig.3.17.
FIGURE 3.17 - Ballast Pressure Profile due to Passage of Loaded Wheels
It can be seen that the pattern of sleeper-ballast pressure beneath the IRJs is different from
that of the reference rail. The pressure time history of the IRJs shows two peak values for a
single wheel passage. The first peak (short duration peak) occurs when the wheel is over the
joint gap and the second peak (long duration peak) occurs when the wheel is over the
sleeper. This confirms that a portion of dynamic impact generated at the rail joint transmits
into the track sub-structure consistent with popular belief. This observation explains the
high rate of track deterioration around the gap of the IRJs compared to continuous welded
rails. This ballast pressure variability beneath the sleepers around the IRJs is a key factor that
gives rise to irregular track settlement of sleepers around the IRJs over a period of time. The
peak values of sleeper pressure were extracted from the recorded data and are presented in
Figs 3.18 and 3.19 respectively for the cases of wheel on sleeper (WOS) and wheel on joint
(WOJ).

FIGURE 3.18 - Peak Ballast Pressure due to Passage of Wheels over Sleepers
FIGURE 3.19 - Peak Ballast Pressure due to Passage of Wheels over Joint Gaps
In these figures, the vertical axis presents the ratio of recorded ballast pressure (Pr) to the
average ballast pressure recorded under the reference rail (Pr0). The average peak pressure
under IRJ 2 is ~36% lower than those of IRJ-1. This reduction in ballast pressure under IRJ-2
illustrates the beneficial effects of the higher track modulus. A 42% increase in track
modulus reduces ballast pressure by 33%.
3.2.5 Effect of the Angle of Cut in the Rails of the IRJs
Until 2005, square-cut (rails cut perpendicular to their axis prior to assembly) IRJs were
predominantly manufactured, which are now replaced by 75 angled-cut IRJs. These two
types of IRJs are shown in Fig. 3.20.
Sleepers
90°
75°
(b) Inclined cut – top view

Square cut – top view (c)
FIGURE 3.20 - Square Cut and Angle Cut IRJs

It is generally claimed that the inclined cut IRJs outperform the square cut IRJs as a
low-impact joint within the Australian railway track industry; however, there is a paucity of
literature with regard to the actual structural merits of these two designs. There is a need to
quantify the structural response of these two types of IRJs to the passage of wheels based
on continuously acquired field data from joints strain-gauged closer to the source of impact.
It is shown that the inclined IRJs resist the wheel load with higher peak shear strains (which
may adversely affect their fatigue life) and lower peak vertical strains (which may help in
reducing ratchetting and surface damage) compared to that of the square IRJs.
It is hypothesised that the inclined IRJ can
be more sensitive to the contact band
Endpost A
(wheel position) than the square IRJ as
conceptually shown in Fig. 3.21. In Fig. 3.21,
a wheel is replaced by a Hertzian elliptical
pressure patch; in this figure, Rail-2 of the B d1 Rail -2 (a)
Rail -1
inclined joint receives the wheel load
eccentric to the longitudinal axis of the rail
(as shown by the shaded triangle in (b)). Endpost
A
This eccentricity will induce torsion about
the longitudinal axis of the rail; since the
vertical rail section in (b) is inclined to the Rail -1 B
Rail -2
(b)
d2
longitudinal axis, the rail will resist the
torsion as a biaxial twist along the cross FIGURE 3.21 - Conceptual Diagram of
section and the longitudinal section, which the Hypothesis on Two IRJs
can lead to increased shear. A similar
scenario would have occurred when the
wheel was leaving Rail–1 of the inclined IRJ. Such an eccentricity does not occur in the
square IRJ as can be seen in Fig. 3.21 (a).
Two IRJs, one square and the other inclined, were tested under in-service wheel passages
over nine months. These IRJs contained multiple strain gauges attached to both rails on
either side of the rail web; a total of 28 strain gauges were used but due to limitations in the
data acquisition system (DAQ), only a selected 14 channels were monitored (i.e., 7 channels
per IRJ). Fig. 3.22 shows the test setup and the location of critical strain gauge rosettes close
to the source of impact.
An ultrasonic moving object detector (Fig. 5) was used to trigger the DAQ once it sensed the
presence of a train approaching the test site. Power for the DAQ was provided through two
solar panels.

Passing Wheel
3-gauge rosettes
Detail ‘X’
Strain gauged IRJ

(a) Detail ‘X’ (b)
FIGURE 3.22 - Strain Gauged IRJ under Wheel Passage

A sampling frequency of 20 kHz was implemented to ensure accurate capturing of the high
frequency dynamic strain responses and data were recorded only for a period of 10 seconds
for every passing train. A total of 200,000 data points for each of the 14 channels was thus
recorded for each passing train. After 10 seconds of recording, the DAQ reverted to a low
frequency sensing mode until another train was detected. During the non-traffic period,
the data were recorded at an interval of 5 minutes; this data included all 14 strain channels
and a thermocouple channel. The thermal data were used to correct the measured
longitudinal strains for thermal effects in order to determine the pure bending strains. The
data on vertical, shear and longitudinal strains measured from the field tests are shown in
Figs. 3.23, 3.24 & 3.25 respectively.
Vertical Strains (Microstrains)
500 SQUARE IRJ

INCLINED IRJ
400
300
200
100
84.4 84.6 84.8 85.0 85.2 85.4 85.6 85.8

 10
3
Sampling Data Points
FIGURE 3.23 - Vertical Strains in the IRJs under Wheel Passage

It can be seen from Fig. 3.23 that the strain in the square cut IRJ is larger than that of the
angle cut IRJ; as traffic is the same, the square IRJ must have generated a higher dynamic
load.
This data re-confirms the general understanding that the inclined cut IRJ is a low impact
design. However, Fig. 3.24 shows that the same angle cut IRJ attracts a lot higher shear
strain compared to the square cut IRJ.

500
Shear Strains (Microstrains)

SQUARE IRJ
400 INCLINED IRJ
300
200
100
1.05 1.052 1.054 1.056 1.058

5
Sampling Data Points (b) x 10
FIGURE 3.24 - Shear Strains in the IRJs under Wheel Passage
As shear strains damage the material more than normal strains, potentially the angle cut IRJs
are more vulnerable to early failure. The test proved the same where the angle cut IRJ failed
two weeks ahead of the square cut IRJ.
The longitudinal strains (excluding thermal strains) measured on these two joints are shown
in Fig. 3.25. It can be seen that both IRJs have recorded similar strains although the inclined
IRJ shows marginally higher strains. As longitudinal strain is due to a more global response,
the type of cut would not affect performance. It can therefore be concluded that overall
there is very little difference between the angle cut and square cut IRJs.
FIGURE 3.25 - Longitudinal Strains in the IRJs under Wheel Passage

3.3 Plastic Strains and Residual Stresses in IRJs
From the information provided in Sections 3.1 and 3.2, the following inferences can be made:
(1) the current designs of IRJs (square cut/ angled cut/ four bolted/ six bolted/ supported on
the sleeper/ suspended between sleepers), suffer from ratchetting metal flow caused by
localised accumulated plastic strains and (hence) residual stresses; (2) higher moduli tracks
minimise joint bar strains, vertical deflection of sleepers adjacent to joint bar gaps and
pressure of ballast under those sleepers; however, the effect of track modulus on the impact
loads generated at the gap of the IRJs due to the passage of wheels is found to be
insignificant. It remains to be determined what are the major factors that affect the
ratchetting strains and the residual stresses at the top edge of the railhead in the vicinity of
the gaps of the IRJs; this section provides the results of some major experiments carried out
for this purpose. Firstly, structural testing was carried out on IRJs in a real life wheel-rail
loading rig, wherein the localised plastic strains were determined using a state-of-the-art

digital image correlation method. Secondly, the tested specimens were sliced carefully and
examined for the build-up of residual stresses due to the passage of the loaded wheels in a
neutron diffraction test facility. Thirdly, several IRJs retrieved from the field were sliced and
examined for residual stress build-up, which allowed us to conclude whether or not the track
had any effect on the ratchetting and residual stress build-up.
3.3.1 Accumulation of Plastic Strains in Current Designs of IRJs
An innovative, real world, full scale wheel-rail loading rig capable of applying up to 60TAL
was designed and fabricated as part of the project. A single wheel carrying up to 300kN
vertical load with up to 1Hz of forward and reverse rolling action was used to apply the load
onto the rail specimen. Fig. 3.26 shows the test rig, specimen and strain measuring
equipment (digital imaging camera fitted with a microscope lens).
Long Distance Microscope Full scale Wheel

Cannon EOS 7D
1000 watt narrow beam Test rail specimen

light
FIGURE 3.26 - Lab Set-up for Plastic Strain Accumulation Measurement

During the passage of loaded wheels on the railhead, a series of images of the railhead end
face for different wheel positions were taken; the images were analysed using the digital
image correlation (DIC – details of which are in the Appendix) method for the determination
of plastic strain accumulation at an extremely localised railhead edge. To properly simulate
the passage of loaded wheels, which is unidirectional, the wheel was unloaded when it
reached a limit location, lifted up and reversed back to the original position where it was
re-loaded and rolled back in the forward direction. During the reverse directional motion,
only two digital images were taken whilst during the forward motion 30 images were taken.
Fig. 3.27 illustrates these details.

FIGURE 3.27 - Directions of Wheel Motion and Limit Positions

Visual observation of the contact zone of the railhead was obvious as the shiny patch, as
shown in Fig. 3.28. In the expanded view of the contact zone shown in Fig. 3.28(b), the
approximate dimensions and the boundaries of the contact zone are marked.
FIGURE 3.28 - Shape of Contact Patches Adjacent to Railhead Edge

The width of the contact zone gradually increased towards the rail edge; Fig. 3.28 shows that
the contact patch has an enlarged ‘mouth’ at the edge of the railhead, which depicts the
effect of the unsupported vertical edge. The area of this shiny patch was 592mm2 calculated
based on a  13.45mm; b  8.75mm; b1  12.85mm , the symbols of which are shown in Fig.
3.27(b). The theoretical semi-axes values of the Hertzian contact patch are for wheels loaded
with 130.7kN, a  7.8mm; b  5.9mm . The measured dimensions are quite large as the
real world contact produces plastic response at the contact surfaces whilst the Hertz theory
assumes elastic response of both bodies in contact. It can therefore be concluded that the
Hertzian theory is not really applicable to heavy haul rail–wheel interface studies; nonlinear
elasto-plastic finite element modelling is more appropriate. Figure 3.29 shows that when the
wheel moves beyond the rail edge (wheel position: z  0 mm, the peak vertical strain on the
rail end face symmetric axis migrates towards the top edge. It can also be observed that the
vertical strain drops by approximately 80% within the top 10mm depth on the rail end face.

Therefore, it is clear that the top zone on the rail end face is subjected to significant
deformation at the edges under the wheel/rail contact loadings.
Vertical strain (E22)

-0.040 -0.035 -0.030 -0.025 -0.020 -0.015 -0.010 -0.005 0.000
0
Depth from top y(mm)

8
10
12
Z = -2 mm
Z = 0 mm 14
Z = 3 mm 16
Z = 5 mm
18
Z = 10 mm
20
FIGURE 3.29 - Vertical Strain (E22) Distribution along Rail End Symmetric Axis for Varying
Positions (Z) of a Wheel Loaded with 130.7kN
A number of analyses were carried out with varied wheel loads and wheel positions. When
the wheel loaded with 50kN was positioned at z  10 mm, E22 = 29,500 microstrains. With
an increase in wheel load to 130.7kN, the vertical strain increased to 36,000 microstrains. In
other words, (130.7  50) 100 / 50  161% increase in wheel load raises the vertical strain
only by ( (36000  29500) 100 / 29500  22% ). This result provides hope for higher axle
loads from the stress perspective.

The total accumulated plastic strain at a point (PE22) on the rail end face after a particular
load cycle was determined from the digital images taken after the full cycle and is presented
in Fig. 3.30; it can be seen that the first load cycle induces approximately 24,000 microstrains
(at the top). The railhead end face top zone subjected to plastic deformation is defined as
the plastic zone as shown in Fig. 3.31.

Vertical plastic strain (PE22)

-0.03 -0.025 -0.02 -0.015 -0.01 -0.005 0
0
Depth from rail top (mm)

Depth of the 2
plastic zone
4
6
8
10
12
14
16
FIGURE 3.30 - Accumulated Vertical Plastic Strain (PE22) Distribution
FIGURE 3.31 - Plastic Zones of Rail End Face

The depth of plastic zone for the 130.7kN wheel load is approximately 12mm while that
for the 50kN wheel load is approximately 7mm. Therefore, it is clear that the depth of the
plastic zone increases when the wheel load increases; however, the magnitude of the plastic
strain is not very sensitive to wheel load.
The accumulated vertical plastic strain along the depth of the rail end face centreline at the
end of the 1st and 100th load cycles for the 130.7kN and the 50kN wheel loads is presented in
Fig. 3.32 and Fig. 3.33 respectively.


0 -0.005 -0.01 -0.015 -0.02 -0.025 -0.03
Depth = 0 mm
Depth = 1.30 mm
Depth = 2.22 mm
Depth = 3.14 mm
Depth = 3.61 mm
Depth = 4.53 mm
Depth = 5.45 mm
Depth = 6.37 mm
Depth = 7.30 mm
Depth = 8.22 mm 1 Load Cycle 100 Load Cycle
Depth = 9.14 mm
Depth = 10.06 mm
Depth = 10.99 mm
FIGURE 3.32 - Accumulated Vertical Plastic Strain (PE22) Distribution along Rail End
Symmetric Axis after the 1st and 100th Passage of a Wheel Loaded with 130.7kN.

0 -0.005 -0.01 -0.015 -0.02 -0.025 -0.03
Depth = 0 mm
Depth = 1.22 mm
Depth = 2.15 mm
Depth = 3.07 mm
Depth = 4.00 mm 1 Load Cycle
Depth = 4.92 mm 100 Load Cycles

Depth = 5.84 mm
Depth = 6.77 mm
FIGURE 3.33 - Accumulated Vertical Plastic Strain (PE22) Distribution along Rail End
Symmetric Axis after the 1st and 100th Passage of a Wheel Loaded with 50.0kN
From Figs. 3.32 and 3.33, it can be seen that more than 60% of the accumulated plastic
deformation in the 100 load cycles is dominated by the first load cycle. The passage of the
first wheel on the virgin railhead is thus responsible for the significant plastic strain; the rate
of accumulation of the plastic strain reduces with the subsequent passage of the wheels.

3.3.2 Residual stresses in the IRJs

Residual stresses in the railheads near the gap of the current IRJ designs were measured
using the KOWARI Neutron Diffraction facility (Fig. 3.34) in the Australian Nuclear Science
and Technological Organisation (ANSTO), Lucas Heights, NSW.
FIGURE 3.34 - Kowari Nuclear diffraction Facility for Residual Stress Measurements
An electric discharge machine was used to cut 5mm thick slices of railheads along the
transverse and longitudinal directions as shown in Fig. 3.34. Measurements were carried out
in a gauge volume 3mm  3mm  3mm of the sliced rail samples. New, in-service (voluntarily
removed from track) and damaged IRJs (declared unfit for service and removed from track)
were obtained from a rail track owner. These IRJs were unbolted and cleaned from the glued
insulation layers prior to slicing the railheads. Slices were taken from the railheads in the
vicinity of the gap of the IRJs as shown in Fig. 3.35. Each slice contained several
measurement points. Each point was subject to neutron beam penetration of 30 minutes.
The residual stress components (vertical, lateral and longitudinal) and lattice space
measurements taken from the T-slices and L-Slices are shown in Figs. 3.36 and 3.37
respectively.
T-slice geometry Mesh in T-slice (regular 3 mm steps)
V=X
T=Y
L=Z
L-slice geometry Mesh in L-slice: 400 points (2 or 4 mm

steps)

V=X
V=X L=Z
T=Y
0
-5
L=Z -10
-15
-20
-25
-30
-35
-40
-45
-50
-55
-60
0 5 10 15 20 25 30 35 40 45 50 55 60
FIGURE 3.35 - Slicing of Railheads in IRJs for Transverse (T-Slices) and Longitudinal (L-Slices)
Residual Stress Measurements
Stress evolution at the rail ends was found to exhibit characteristics of a compressive zone of
5mm depth that is counterbalanced by a tension zone beneath, extending to a depth of
around 15mm. The changes in d0 suggest decarburisation.
Results on L-slices suggest that material adjacent to the joint has a different
thermo-mechanical history. Material near the gap is more deformed than the material in the
bulk of the rail (see differential stress maps). This is in agreement with larger material flow
visible under visual inspection. The feature on the very top corner in stress maps is an
artifact, because stress boundary conditions are not fulfilled. This is probably due to high
deformation, strain and texture gradients around the corner of this particular specimen.

NEW In-Service Damaged

d0 (Å)
1.1681
1.16815
1.1682
1.16825
1.1683
1.16835
1.1684
1.16845
1.1685
1.16855
1.1686
Transverse, σYY (MPa)
-300
-250
-200
-150
-100
-50
50
100
150
200
Vertical, σXX (MPa)
-300
-250
-200
-150
-100
-50
50
100
150
200
FIGURE 3.36 - T-Slices Measurements of Residual Stresses in the In-Field IRJs

0
NEW 0
In-Service 0
Damaged
-5 -5 -5
-10 -10 -10
Longitudinal, σzz (MPa)

-15 -15 -15
-20 -20 -20
-25 -25 -25
-30 -30 -30
-35 -35 -35
-40 -40 -40
-45 -45 -45
-50 -50 -50
-55 -55 -55
-60 -60 -60

0 5 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 45 50 55 60
-300
-250
-200
-150
-100
-50
50
100
150
200
250
300
0 0 0
-5 -5 -5
-10 -10 -10
-15 -15 -15

-20 -20 -20
-25 -25 -25
-30 -30 -30
-35 -35 -35
-40 -40 -40
-45 -45 -45
-50 -50 -50
-55 -55 -55
-60 -60 -60

0 5 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 45 50 55 60
-300
-250
-200
-150
-100
-50
50
100
150
200
250
300
0 0 0
-5 -5
-10 -10 -10
-15 -15
-20 -20 -20
-25 -25
-30 -30 -30
-35 -35
-40 -40 -40

d0 (Å)
-45 -45
-50 -50 -50
-55 -55
-60 -60 -60

0 5 10 15 20 25 30 35 40 45 50 55 60 0 10 20 30 40 50 60 0 5 10 15 20 25 30 35 40 45 50 55 60
1.1681
1.16815
1.1682
1.16825
1.1683
1.16835
1.1684
1.16845
1.1685
1.16855
1.1686
FIGURE 3.37 - L-Slice Measurements of Residual Stresses in the In-Field IRJs
L-slices taken from the lab tested specimens (described in Section 3.3.1) were also examined
for build-up of residual stresses using the Kowari Neutron Diffraction facility in ANSTO. The
results are presented in Fig. 3.38.

Un-Loaded End Loaded End Loaded End

600 Cycles 1000 Cycles
0 0
-5 0 -5
-10 -5 -10
-15 -10 -15
-20 -15 -20

Longitudinal, σzz (MPa)
-25 -20 -25
-30 -25 -30
-35 -30 -35
-40 -35 -40
-45 -40 -45
-50 -45 -50
-55 -50 -55
-60 -60
-55
0 5 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 45 50 55 60
-60
0 5 10 15 20 25 30 35 40 45 50 55 60
-200
-150
-100
-50
50
100
150
200
250
300
0
-5 0 0
-10 -5 -5
-15 -10 -10
-20 -15 -15
-25 -20 -20

-30 -25 -25
-35 -30 -30
-40 -35 -35
-45 -40 -40
-50 -45 -45
-55 -50 -50
-60
-55 -55
0 5 10 15 20 25 30 35 40 45 50 55 60
-60 -60
0 5 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 45 50 55 60
-200
-150
-100
-50
50
100
150
200
250
300
FIGURE 3.38 - L-Slice Measurements of Residual Stresses in the Lab Tested IRJs
Figs. 3.37 and 3.38 show the similarity between the build-up of the longitudinal compressive
residual stresses at the corner of the railhead; the (obvious) difference is their magnitude
and the extent. For example, the maximum longitudinal compressive residual stress
measured on the slices taken from the structural lab tested IRJ was 200MPa whilst the same
from the field loaded IRJ was 300MPa. The lab tested IRJ was subject to a maximum of only
1000 cycles of 130.7 kN loaded wheels; this corresponds to approximately two coal trains
(250 wagons). In comparison, the slices taken from the IRJs retrieved from the field were
there for months (the actual traffic was not known). The lower residual stress is entirely
acceptable – however, the feature that should be noted is the very high rate of build-up of
residual stress (200MPa) within 1000 load cycles. The very large accumulation of plastic
strains reported in Section 3.3.1 is consistent with this observation. The other feature to
note is the extent of residual stress build-up which is localised in the slices from the lab
tested IRJs – whilst the slices from the field loaded IRJs show the entire length of the slice
with maximum residual stress of 300MPa. Again, this is entirely logical as the load was only
applied for a very local length of 20mm on the railhead.

The same inferences can be made for the vertical tensile residual stresses shown in Figs 3.37
and 3.38. Interestingly, the vertical tensile residual stress magnitude is 300MPa for both the
slices from the lab tested IRJs and the field retrieved IRJs. This can be attributed to the
rather comparable loads and the extreme stress singularity at the corner. The slices from the
lab tested IRJs exhibited lower build-up of vertical tensile residual stresses on the body of
the slice compared to the field retrieved IRJ. Clearly, it can be seen that the build-up of the
residual stresses and the associated accumulation of plastic strains in the IRJs are due to the
manner in which the IRJs are designed; the contribution, if any, of the track moduli to these
localised phenomena is negligible. Whilst the track modulus is a key parameter for the
structural responses (deflection of sleepers, joint bar strains and ballast pressure), the
localised ratchetting and the associated residual stress build-up are entirely due to the
current design where the rail is cut normal to the axis of the rail making the cut section quite
vulnerable due to corner loading of the wheels.
3.4 Ratchetting Failure: Life Prediction Model for the IRJs
As presented in Sections 3.1 – 3.3, the ratchetting metal flow is solely affected by the wheel
contact at the top of the unsupported railhead end section, with no contribution from the
track parameters. This conclusion significantly helps to simplify the modelling of the complex
wheel-rail rolling contact and the elasto-plastic nonlinear material response. In particular,
the complex interaction of the rail– sleeper –ballast–capping layer – formation and the
associated variability can all be disregarded whilst modelling the wheel–rail rolling contact /
railhead metal ratchetting. In other words, the boundary condition can be simplified as fixed
along the rail foot. Further, the IRJ assembly itself can be simplified as described in this
section.
For ratchetting life prediction, stress concentrations must be accurately predicted; as well,
a reliable strain controlled hysteretic stress-strain relationship determined (preferably) from
experiment should be available. Accurate prediction of stress concentration requires a good
mesh and a reliable nonlinear FE solver with contact nonlinearity capability. Hysteresis
stress–strain curves are obtained from tension– torsion tests on cylindrical specimens, which
are complex. Due to limitations of equipment to perform such physical tests, monotonic
tests were performed from railhead material coupons, a FE model was developed and
calibrated with material properties and then extended to produce the hysteretic curves from
numerical simulations of tension–torsion tests.
3.4.1 Geometry Idealisation of the IRJs for Finite Element Modelling
FE models reported in the literature specifically address the contact pressure and railhead
stresses in IRJs. These models are either static (Chen and Kuang 2002; Chen 2003; Sandström
and Ekberg 2009 or dynamic (Wen et al. 2005; Cai et al. 2007). General findings reported
include significant deviation from the Hertzian contact and high levels of stress
concentration. FE models were developed by the research team – both static and dynamic
models were developed; both models simplified the IRJ assembly as a single part with
varying material properties as shown in Fig. 3.39.

FIGURE 3.39 - Simplification of the Geometry of the IRJ Assembly for FE Modelling
The wheel geometry was also simplified as shown in Fig. 3.40; simplification was possible as
only tangent track with wheel contact symmetric to the IRJ assembly alone was the focus of
the study. The wheel flange was therefore not modelled.
460mm
1
1 in 20
FIGURE 3.40 - Simplification of the Geometry of the Wheel for FE Modelling

The assembled wheel-rail contact model is shown in Fig. 3.41.

(a) Longitudinal view
(c) Isometric
view
(b) End view
FIGURE 3.41 - Assembled IRJ Wheel – Railhead Model for FE Meshing

For ratchetting modelling, the rail foot was fixed whilst for contact impact modelling the
track support was modelled using springs and dashpots.
3.4.2 Material properties of the IRJs for finite element modelling: monotonic
AS1085.1(2002) defines the enclosed area ABCDEF of Fig. 3.42 as the heat affected zone,
where the rail steel is deemed to be harder. Therefore, a single representative stress-strain
curve cannot be obtained for the whole of the railhead steel.
FIGURE 3.42 - Head Hardened AS60kg/m Rail Section

The fundamental material properties (Young’s modulus and Poison’s ratio), required for the
elastic analysis are specified in the Australian standards, AS1085.12(2002) and in the open
literature in Chen and Kuang (2002), and Wen et al. (2005). However, the nonlinear
properties of the railhead within or outside the hardened region are not known. Therefore,
the variation of material properties across the railhead was experimentally determined. Test
coupons extracted from varying depths of the railhead were subjected to uni-axial tension

using an INSTRON testing machine. The axial and transverse strains of the test coupons were
determined using a non-contact optical strain measuring technique outlined in the Appendix.
Non-contact optical strain measuring techniques have become popular in experimental
mechanics due to their wide range of applications. These techniques can measure very large
strains (80,000 microstrains are measured in this research) till the failure of the specimens,
whereas the bonded electrical strain gauge generally fails at around a threshold level of
20,000 microstrains (Bandula-Heva et al. 2013). Tensile test coupons were prepared as per
the provisions in AS1391 (2007); a typical coupon is shown in Fig. 3.43.
Thickness is 2.5mm
FIGURE 3.43 - Typical Monotonic Uniaxial Tension Test Coupon
FIGURE 3.44 - Strategies for Extraction of Test Coupon

Coupons were obtained from two virgin 60kg/m rails as shown in Figs. 3.44 & 3.45.
Rail test sample
FIGURE 3.45 - Cutting of Test Coupon

A strain gauge was attached on one flat side of each test coupon as shown in Fig. 3.46. Black
and red colour marker pens were used to dot the reverse side of the test coupons as shown
in Fig. 3.47 for facilitating the DIC analysis.

Strain
gauges
FIGURE 3.46 - Strain Gauged Face of Test Coupon
FIGURE 3.47 - Colour Dotted Face of Test Coupon
The coupons were tested in an INSTRON machine as shown in Fig. 3.48. Digital images were
taken sufficiently for DIC determination of strains.
Test INSTRON
coupon testing
machine
Camera
FIGURE 3.48 - Testing of Coupon in Instron

Since the application of DIC to steel strain determination is new, it was decided to validate
the steel strains from the DIC method using the engineering stress-strain curve for the rail
steel from the strain gauge data. The engineering stress-strain curves of a typical test
specimen obtained through the DIC method and from the strain gauge data are shown in Fig.
3.48. The true stress-strain curve is also included in the same figure. Even though the strain
gauges produced accurate strains during the initial part of the tensile test, strain gauges

de-bonded close to 0.02 strains (or 20,000 microstrains). Thus a full stress-strain curve based
on strain gauge readings cannot be seen. The DIC method, on the other hand, has provided
strain measurements in excess of 80,000 microstrains. Furthermore, a good correlation
between the engineering stress-strain curve from DIC and strain gauge readings can be seen
in Fig. 3.49.
1400
Stress (MPa)
True Stress-Strain:
1200
1000 Engineering Stress-Strain: DIC
800
Engineering Stress-Strain: Strain Gauge
600
400
200
Strain
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09
FIGURE 3.49 - Stress – Strain Curves of Coupon
The complete stress-strain curves of the test coupons from one of the rail specimens are
shown in Fig. 3.50; the distinction between the head hardened and non-hardened zones is
apparent. The final failure of the railhead exhibits brittle fracture as shown in Fig. 3.51;
improved ductility of steel would improve the performance of the IRJs.
1400
1200
1000
Stress (MPa)
800
A1
600 B1
A2
400 B2
A3
200 B3
Strain
0
-0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
FIGURE 3.50 - Stress – Strain Curves of Head hardened and Non-hardened Zones

Fracture without necking
FIGURE 3.51 - Brittle Fracture of Coupon

The measurements from Rail B coupons confirmed the stress-strain curves of Rail A coupons.
The 0.2% proof strength of the rail steel as a function of depth from the top of the railhead is
shown in Fig. 3.52.
1000
0.2% Yield strength
800
600
(MPa)
400 Rail A Rail B AS 1085.1 (2002)

200
Depth (mm)
0
0 5 10 15 20 25 30 35 40
FIGURE 3.52 -Variation of 0.2% Proof Strength of Steel as a function of Depth
The properties of rail steel obtained from the specimens are summarised in Table 3.1.
Table 3.1 - Properties of Rail Steel
Test coupon Young’s 0.2% Yield Max Eng. Stress Max True
Modulus, Strength (MPa) (MPa) Stress (MPa)
B1 (MPa)
207336 911 1287 1400
B2 206624 845 1206 1320
B3 210028 810 1202 1330
A1 201695 907 1297 1389
A2 210965 840 1215 1316
A3 196112 816 1194 1283
A4 212070 725 1194 1295

A5 209155 561 902 1015
A6 202836 538 954 1033
A7 207934 555 965 1051
The railhead material properties vary across the depth of the railhead. Due to practical
limitations, the properties of the top-most railhead surface could not be tested. From the
available data, an extrapolation was carried out for this purpose. This extrapolation is
important because the railhead top surface material properties are essential for accurate
determination of the plastic zone in the FE analysis.
In order to obtain the railhead surface material properties, the stress-strain curves obtained
from the top three rail test coupons (see Fig. 3.44) were extrapolated using a second order
polynomial. A sample stress extrapolation for strain ( )  0.03 is presented in Fig. 3.53.
An average stress-stress curve was also obtained based on the tensile test results of the top
three rail test coupons (see Fig. 3.44). The average stress-strain curve and the extrapolated
top surface stress-strain curve are provided in Fig. 3.54.
1240
y = 0.9248x2 - 23.221x + 1259.9
1220
1200
1180
Stress
1160
1140
1120
1100
0 2 4 6 8 10 12 14 16
Depth (mm)
FIGURE 3.53 - Stress at Different Depths for   0.03

1600
1400
1200
1000
Stress (MPa)
A1
800
B1
A2
600
B2
A3
400 B3
Avg curve
200 Top curve
0
-0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
Strain
FIGURE 3.54 - Average and Extrapolated Top Stress-Strain Curves for Heat Treated,
Hardened Railhead
3.4.3 Material properties of the IRJs for finite element modelling: cyclic
The slender test specimens used in the uniaxial tensile tests can buckle under compression;
therefore, test specimens with robust cylindrical cross sections are required for the cyclic
load test. Test specimens are generally subjected to axial and torsional loading to determine
the ratchetting material parameters of Jiang and Sehitoglu (1994) & Portier et al. (2000). In
many situations, circular hollow test specimens with 12.5mm gauge section diameter are
used. Tension-torsion testing machines are used to test these specimens and axial-torsional
extensometers are fixed on the gauge section of the test specimens to measure strains. Due
to limitations in budget and time in this project, the hysteresis loops of the material are
obtained using FE analysis in this research.
The geometry of the cyclic load test specimen is shown in Fig. 3.55. Considering symmetry,
one half of the test specimen was modelled (Fig. 3.56). In this FE model, the symmetric cross
section of the gauge length was assigned symmetric boundary conditions. Axial
displacement was specified on the circular surface of the half FE model. The entire
simulation of the test was performed under displacement control.

FIGURE 3.55 - Geometry of the Cyclic Load Test Specimen
FIGURE 3.56 - Half test Specimen with Symmetry Boundary condition
The developed half FE model of the test specimen is as shown in Fig. 3.57. In order to specify
the half-cycle test data for the kinematic hardening model, stress-plastic strain curves based
on an average stress-strain curve and the extrapolated railhead top surface stress-strain
curve were determined and plotted in Fig. 3.58.
FIGURE 3.57 - FE Model of the Cyclic Load Test Specimen

2000
1500
Avg curve Top curve
1000
Stress (σi)
500
Plastic strain (εip)

0
0 0.02 0.04 0.06 0.08 0.1 0.12
FIGURE 3.58 - Half Cycle Stress – Plastic Strain Data
To determine the ratchetting material parameters for the Chaboche three decomposed rule,
a stabilised hysteresis loop from strain controlled cyclic load test is required. It was decided
to apply five load cycles within 2 seconds with 1.2mm amplitude on a full test specimen to
obtain a hysteresis loop with significant plastic deformation. Since one half of the test
specimen was modelled, it needs only half of the amplitude (0.6mm) of cyclic elongation in
the FE analysis. The applied elongation is shown in Fig. 3.59.
0.8
Axial elongation (mm)
0.6
0.4
0.2
0
-0.2 0 0.5 1 1.5 2 2.5
-0.4
-0.6
-0.8 Time
FIGURE 3.59 - Applied Cyclic Tension
The FE model was analysed in the ABAQUS/standard environment. Since the FE model was
analysed under elongation control, the axial strain output for both the average and top
material properties exhibits similar plastic strain-time history and it is plotted in Fig. 3.60.

0.15
Axial plastic strain (εp)

0.1
0.05
0
0 0.5 1 1.5 2 2.5
-0.05
-0.1
-0.15
Time (s)
FIGURE 3.60 - Output: Axial Plastic Strain Time History
The axial stress-plastic strain hysteresis loops are plotted in Fig. 3.61. The results for the
average material properties and the railhead top surface material properties are included in
these figures.
2000
Avg Top
1500
1000
Axial Stress (MPa)
500
0
-0.15 -0.1 -0.05 0 0.05 0.1 0.15
-500
-1000
-1500
-2000
Axial plastic strain
FIGURE 3.61 - Output: Axial Stress – Axial Plastic Strain Hysteresis Loop
The FE results from the railhead top surface material properties show consistently higher
stress amplitude than the FE results with the railhead average material as shown in Fig. 3.61;
this reflects the fact that the railhead top surface has a higher yield stress material than the
average railhead material.
3.4.4 Plastic strains: comparison of FE and experimental predictions
Three decomposed kinematic hardening parameters of Chaboche required definitions of five
constants: C1 , C2 , C3 ,  1 and  2 . A stabilised uniaxial strain controlled hysteresis loop is
used to determine these five constants. The first backstress (which is the center of the yield
surface) component  1 of the Chaboche model has very large value representing the initial
nonlinear part of the backstress  . The third backstress component  3 represents linear

segment of the backstress hysteresis loop and its value is relatively small. The transition
between the nonlinear to linear segment of the backstress hysteresis loop is represented by
the second component  2 . The loading path of the backstress components  1 and  2 is
provided in Eq. 3.1(Bari and Hassan (2000))
i 
Ci
i
 
1  2 exp   i ( p  ( Lp ))  For i  1 and 2 (3.1)
where  Lp is the value of the upper and lower limit of the plastic strain in the hysteresis
loop.
  1   2   3 (3.2)
It is generally recognised that C1 should be relatively very large in line with the plastic
modulus at yielding and the parameter  1 should also be large enough to immediately
stabilise the hardening of  1 . The parameter C3 represents the slope of the linear segment
of the hysteresis loop at high strain range. The parameters C 2 and  2 are determined using
a predictor - corrector method to satisfy Eqs. 3.3 and 3.4.
C1 C2 C
   0    3 { p  ( Lp )} (3.3)
1  2 2
where,  and  0 are axial stress and yield stress respectively.
The parameters determined for the Chaboche model are tabulated in Table 3.2.
Table 3.2 - Parameters for Chaboche Model
C1 50000 61500
C2 21500 16300
C3 452 450
1 725 785

2 42 37
The top zone of the railhead part (see Fig. 3.62) was assigned the elastic-plastic material
properties based on the top surface stress-strain curve (Fig. 3.61). The railhead was
partitioned as shown in Fig. 3.61 to enable the application of two separate elastic-plastic
material properties. The railhead part other than the top zone was assigned the
elastic-plastic material properties based on the average stress-strain curve.
FIGURE 3.62 - Railhead Top: Bi-Material – Top and average Zones

The E22 contour maps on the rail end face obtained from FE analysis are presented in Fig.
3.63. The maximum vertical strain at each depth on the rail end face is seen on the end face
centreline. Therefore, the vertical strain (E22) results on the rail end face centreline
obtained from the DIC method and the FE analysis are plotted together in Fig. 3.62; good
correlation shows that the FE results are dependable.

-0.05 -0.04 -0.03 -0.02 -0.01 0

0
Depth from rail top (mm)

Vertical strain (E22)
2
4
6
8
Experimental 10
FE results 12
14
16
FIGURE 3.63 - Vertical Strain at Unsupported Free End of Railhead

The lateral and shear strains are shown in Figs 3.64 & 3.65 respectively.
0 0.001 0.002 0.003 0.004 0.005 0.006

0.00
Lateral strain (E11)
Depth from rail top
5.00
(mm)
10.00
experimental
15.00 FE results
20.00
FIGURE 3.64 - Lateral Strain at Unsupported Free End of Railhead

Coordinate Shear strain (E12)

x y FE DIC % Diff.
7.5 4.5 0.01385 0.01350 2.5%
7.5 5.5 0.01226 0.01194 2.6%
7.5 9.0 0.00355 0.00375 -5.6%
FIGURE 3.65 - Shear Strain at Unsupported Free End of Railhead
The FE results show a good correlation with the experimental results for the three strain
components (E11, E22 and E12) on the rail end face. Therefore, the FE model can be
considered as a validated model for further studies. The validated FE model was used to
obtain the other three key strain components (E33, E23 and E13) that could not be
measured using the DIC method. The strain components E33, E23 and E13 on the rail end
face in the first load cycle are presented in Fig. 3.66. The maximum E33 on the rail end face
is exhibited towards the railhead top and its value is 30,951 microstrains. The shear strain
E13 does not show a significant variation across the rail end face, while the shear strain E23
was concentrated to 1mm depth from the railhead top. The state of the equivalent plastic
strains (PEEQ) of the railhead in the vicinity of the top unsupported edge is important in the
discussion of railhead metal plasticity. Therefore, the PEEQ distributions on the rail end face
in the first load cycle with 130.7kN load were obtained using the FE analysis and are
presented in Fig. 3.67. The highly localised accumulated PEEQ and the high gradient along
the depth of the railhead can be seen from the figure.

FIGURE 3.66 - Strains in the Longitudinal Vertical Plane of the Loaded Rail in IRJ
FIGURE 3.67 - Equivalent Uniaxial Plastic Strain (PEEQ) Distribution on Unsupported

Railhead at the Gap

3.4.5 Load History for Prediction of Ratchetting

From the PEEQ distribution in Fig. 3.67, it was assumed that a cyclic load history with a mean
stress equivalent to the yield stress of the material (932MPa) at a point 1mm below the top
of the railhead would be the most critical. With this assumption, the proposed stress
histories of a point located at 1mm below the rail end face top for the 130.7kN wheel loads
are plotted in Fig. 3.68. These stress histories were used as the inputs for the Chaboche
model to predict the ratchetting strains. It was assumed that the maximum, minimum and
mean stresses at any point on the rail end face remain constant throughout the cyclic load
application simulating uniform loaded wheels.
FIGURE 3.68 - Applied Load History

Seven points at one millimetre intervals from the rail head top on the rail end face centreline
were marked and the Chaboche model was used to predict the ratchetting on these seven
points. The Von Mises stress, the minimum stress and the mean stress of these seven points
are presented in Table 3.3.
Table 3.3 - Summary of Applied Stress History for Seven Select points
Mean Minimum
Depth from Von Mises stress Yield stress Material
stress stress
rail top (MPa) (MPa) property
(MPa) (MPa)
1 mm 1247.78 315.78 -616.22 932 Top
2 mm 1189.00 257.00 -675.00 932 Top
3 mm 1185.97 253.97 -678.03 932 Top
4 mm 1192.72 260.72 -671.28 932 Top

5 mm 1113.60 313.60 -486.40 800 Average
6 mm 1078.14 278.14 -521.86 800 Average

7 mm 958.87 158.87 -641.13 800 Average

The plastic strain accumulation at 1mm depth on the rail end face centreline in the initial
five load cycles and the 100th load cycle are found in Table 3.4. It was observed that more
than 60% (0.02960/0.04544 = 65.1%) of the total plastic deformation occurred during the
first cycle of the 100 cycles at 1mm depth of the rail end face centreline. Similar trends can
also be seen at the other observation points as shown in Fig. 3.69.
Table 3.4 - Predicted Ratchetting at the Seven Select Points
Plastic strain at the stress Total accumulated

Number of load cycle
cycle plastic strain
1 0.02960 0.02960
2 0.00056 0.03016
3 0.00044 0.03060
4 0.00040 0.03100
5 0.00032 0.03132
- - -
100 0.00012 0.04544
0.05
Accumulated plastic strain
0.045
0.04 Depth=1
0.035 mm
0.03 Depth=2
0.025 mm
Depth=3
0.02
mm
0.015 Depth=4
0.01 mm
0.005
0
0 20 40 60 80 100 120
Number of load cycles
FIGURE 3.69 - Progressive accumulation of Plastic Strain
3.4.6 Ratchetting failure life prediction
The failure modes of the IRJs have been discussed in Chapter 2; ratchetting is one of the
modes. In order to determine the service life prior to ratchetting failure of IRJs, a failure
criterion for the IRJs is required; for this purpose, the quantity of the railhead metal flow

required to short-circuit the two rails of the IRJ was used as the limit of failure. The IRJ’s gap
is 6mm and the two rails touch each other when a railhead metal flows 3mm ( u3  3 mm)
into the gap as shown in Fig. 3.67. However, allowing for safety, 2mm ( u3  2 mm) of
railhead metal flow into the gap was considered as the limit in this research.
FIGURE 3.70 - Schematic Diagram of Limit Dimension of Metal Flow
A five step Life Prediction Model is proposed:

i. estimation of the service load (Wheel load) for FE analysis
ii. determination of the stress history of the railhead metal
iii. application of the Chaboche model to obtain a ratchetting curve
iv. determination of parameters of a fitting curve to the ratchetting curve and
extrapolation of the ratchetting for a large number of load cycles and
v. estimation of the service life of IRJs.
The PEEQ accumulation curve was appropriately extrapolated to determine the railhead
ratchetting for a large number of load cycles. For this purpose, curve fitting techniques
available in MS Excel and MATLAB software programs were used to determine an equation
of best fit. The proposed relationship between the number of load cycles and the ratchetting
strains is given in Eq. 3.4.
a2
(PEEQ) rat  a1 ln( N )   a4 (3.4)
e a3 N
where the (PEEQ) rat is the total accumulated plastic strain at the end of N number of
load cycles. The parameters a1 , a2 , a3 and a4 are constants unique for a ratchetting curve.
For the case considered, a1  0.0453; a2  0.28; a3  0.000063& a4  0.062 .

The ratchetting strain determined at a point in the railhead must be converted to a

corresponding displacement of that point to estimate the amount of the railhead metal flow
into the gap of an IRJ. Finite displacement of a point in a continuous body is given in Eq. 3.5
(Slater, 1975).
u3 1  u1   u2   u3  

2 2 2
E33          (3.5)
z 2  z   z   z  
By neglecting the higher order terms,

u3
E33  (3.6)
z
Rewriting,
 u3   (E33)z (3.7)
which gives the following relationship:
u3  (E33) z  k1 (3.8)
where k1 is the integration constant and u 3 is the deformation of the point considered in
the direction of the strain E33 . For a given point in the railhead the coordinate z
becomes a constant and the Eq. 3.8 is rewritten as follows;
u3  k2 (E33)  k1 (3.9)
where k 2 is another constant. It was assumed that eq. 3.9 is valid for the plastic strain and
the plastic deformation in the railhead. In order to estimate the constants k1 and k 2 in the
Eq. 3.9 , the plastic deformations u 3 for plastic strains PE33 at the railhead top edge were
obtained from several FE analyses and plotted in Fig. 3.71. The best fitting trend line for the
points in this plot is also included in the figure and the Eq. 3.9 is rewritten with the values of
the constant k1 and k 2 as shown in equation (3.10).
u3  5.0158 (PE33)  0.0308 (3.10)

0 0.005 0.01 0.015 0.02 0.025 0.03

0
Plastic strain (PE33)
-0.02
Plastic deformation (u3)

-0.04
-0.06
-0.08 y = -5.0158x + 0.0308

R² = 0.9997
-0.1
-0.12
FIGURE 3.71 - Relationship between Plastic Strain and Plastic deformation (Taken
from FE Modelling)
We know from basic mechanics,
  
1/ 2
2 
 eq   E112  E22 2  E332  2 E12 2  E232  E312  (3.11)
3 
Assuming all the strain components can be written as a function of  z using five constants,
c1 , c2 , c3 , c4 and c5 and ssubstituting Eq. 3.11 into Eq. 3.10, we obtain:
1/2
2
 
 eq   c12  c22  1  2  c32  c42  c52    E 33  cE 33
3 
 (3.12)
Assuming the (PEEQ) rat and the PE33 of the railhead edge obey Eq. 3.12, the ratio
 (PEEQ)rat  at the railhead edge was obtained from the FE analysis for different
 PE 33 

wheel loads and found that the ratio was indeed a constant (1.763), which simplifies Eq. 3.12
to:
( PEEQ)rat  1.763  PE33 (0-1)
The railhead metal flow ( u 3 mm) into the gap of an IRJ under the 130.7kN service wheel
loads was analytically determined for various tonnages and shown in Table 3.5.

Table 3.5 - Predicted Metal Flow into the Gap
Load cycles Accumulated Metal flow

Traffic load
PE33
N ( 10 )
6
(MGT) (PEEQ) rat ( u 3 mm)
1 13 0.5638 0.3198 1.5734
5 65 0.6368 0.3612 1.7808
10 130 0.6681 0.3790 1.8701
15 195 0.6865 0.3894 1.9224

20 260 0.6995 0.3968 1.9594
25 325 0.7097 0.4025 1.9882

27.4 356.2 0.7138 0.4049 2.0000
According to the railhead metal ratchetting results, the limiting quantity ( u3  2 mm) of the
railhead metal flow was achieved at approximate service life of 356MGT with the 130.7kN
wagon wheel loads. The field observations of IRJs’ service lives collected from 18 samples
(collected from the Australian heavy haul tracks) are presented in Table 3.6.
Table 3.6 - Observed Life of IRJs in the Field
IRJ No. Service life (MGT) IRJ No. Service life (MGT)
1 260 11 447
2 270 12 220
3 214 13 360
4 220 14 360
5 460 15 400
6 230 16 480
7 450 17 320
8 400 18 266
9 270 Mean 341.06
10 512 CoV 30%
The average service life of the 18 IRJs is 341.06MGT with a coefficient of variation (CoV)
approximately 30%. The minimum and maximum service lives in the collected data set were

214MGT and 512MGT respectively. The predicted service life in this study for 130.7kN is
356MGT – which shows a good correlation with the field data.
The effect of the wheel diameter on the IRJ’s service life is examined. The IRJ’s service life
was predicted based on the Chaboche model. It was decided to use 150kN wheel load for
comparison of the service life of IRJs with 700mm, 898mm and 1100mm diameter wheels.
Table 3.7 - Service life of the three FE models for 150kN wagon wheel loads
Load cycles Accumulated Metal flow

Traffic load
Model PE33
N ( 10 )
6
(MGT) (PEEQ) rat ( u 3 mm)
D700 3.46 51.9 0.7138 0.4049 2.000
D898 27.4 356.2 0.7138 0.4049 2.000
D1100 36.7 550.5 0.7138 0.4049 2.000
As per the results, where the wagon wheel diameter is 700mm, the IRJ shows a significantly
low service life which is approximately 52MGT (  3.46  10 wheel passes with 150kN load).
6
The service lives of the IRJs are approximately 356MGT and 550MGT for the load application
with 898 mm and 1100mm diameter wagon wheels respectively. The larger diameter
wheels appear quite ‘IRJ friendly’!
3.5 Summary
This chapter has presented the performance of current designs of IRJs resulting from the
past 10 years of studies on the Australian heavy haul, passenger and freight corridors
including references to international studies. The studies include experimental and advanced
computational analyses listed below:
(1) four un-instrumented, in-service IRJs were selected and monitored using digital
imaging and total station measurements for their progressive degradation over 500
days
(2) two major instrumented field experiments were carried out to understand the
performance of the IRJ assemblies and components under the passage of loaded
wheels
(3) several IRJ assemblages were tested in a laboratory environment using a full-scale
rail–wheel loading rig to clearly focus on the effect of the loaded wheels on the
localised contact on the unsupported railhead
(4) several coupons were cut from the new IRJs to determine the stress–strain
relationships under tension of the railhead material
(5) several cross sectional and longitudinal sectional slices were sampled from the new
and in-service/damaged IRJs to examine the residual stresses using the neutron
diffraction technique

(6) finite element modelling was undertaken with a particular focus on determining
comparable residual strains at the unsupported railhead edge to that of the
experimental predictions in (3) above
(7) to determine the service life of the IRJs, the researchers incorporated the residual
stresses predicted from the finite element models in (6) above into the three
parameter Chaboche metal fatigue algorithm and used a limit of 2mm as the
maximum length of the accumulated metal ratcheted along the longitudinal
direction into the gap of the IRJ to prevent short-circuiting. The predicted service life
prior to the risk of electrical short-circuiting ratchetting metal flow 356MGT is in very
good comparison with the field observed average life of 341MGT (with 30% CoV).

4. LONGER LIFE NEW GENERATION IRJs
Insulated rail joints are shown to exhibit low mean (341MGT), high variable (30% CoV)
service life. They fail due to several mechanisms that are classed under two groups: (1)
global impact load induced structural failure – e.g. joint bar bending failure or fatigue
cracking and (2) localised strain controlled failure. New generation IRJs that minimise/
eliminate these two modes of failure would serve longer in track and/ or carry higher loads/
faster traffic. The economic benefit is expected to be quite large as the IRJs are safety critical
devices; recent renewed interest can be gauged from the theses and patents Charlton
(2007), Akhtar and Davis (2011), Zong (2013), and Bandula-Heva (2013). Charlton (2007)
designed IRJs using ceramic metal composite insulation layers. In this chapter two simple
modifications to the current design are shown to bring significant improvement without
having to introduce new materials.
4.1 Low-Impact IRJ
As described earlier, the gap in the IRJ is the major reason for wheel impact which amplifies
the static wheel load – irrespective of the type of track (low or high modulus). Some old IRJs
that exist in the track were made with 10 mm gap (the most common ones were 8.5mm);
the most recent ones are made with a 6mm gap. A finite element model with varied gap
sizes (10mm – 5mm) was developed and it was found that from the impact perspective a
5mm gap can be ideal. Reducing the gap further can increase the risk of electrical
short-circuiting.
In the FE analysis, ABAQUS/ Explicit solver is used for predicting the wheel – rail contact
force – time history. The wheel and the rail are geometrically modeled, provided with
appropriate material properties and meshed. The bottom of the rail is considered fixed as
only the wheel – railhead contact plane is the primary consideration. The wheel load was
considered as 150kN. Dynamic FE analysis of the IRJ has provided the railhead/wheel contact
force time history, which is shown in Fig. 4.1.

FIGURE 4.1 - Rail/wheel contact force history

Fig. 4.1 shows that, at the beginning of the dynamic analysis, the contact force has increased
sharply just above 150KN and stabilised to the static wheel load value of 150kN after a short
period of approximately 1.2 milliseconds. As the wheel approached the end post, a drop in
the contact force (127kN) occurred due to the local deformation of the edge of the railhead
that is affected by the difference in the material properties between the two interacting
materials (rail steel and endpost Nylon). Within 0.54 milliseconds, the contact force
increased from 127kN to 174kN (or 37%) indicating the occurrence of the rail/wheel
contact-impact. The impact occurred at 7.1 milliseconds from the start of the wheel travel
with the corresponding impact factor of 1.16 (calculated from the quotient of impact force
on static load (1+(174-150)/150)).
It is believed that the wheel impact at the rail edge is due to the momentary ‘loss’ of contact
leading to wheel flight across the end post with the wheel landing on the edge of the other
railhead. The exact location of the wheel tending to lose contact and re-landing on the
railhead cannot be precisely estimated from the FE model. As 0.54 milliseconds of ‘flight
time’ of the wheel travelling at 120km/h corresponds to 18.0mm which is larger than the
end post (10mm gap) thickness, it is inferred that the hypothesis of wheel impact in the
vicinity of the end post is approximately validated.
Although the results presented so far illustrate the logical occurrence of impact in the
vicinity of the end post, to further prove the appropriateness of the FE model for the
contact-impact analysis, the end post material (nylon66) was replaced with the rail steel
itself. This modification effectively removed the joint (discontinuity), with the FE model of
the IRJ becoming a continuous piece of rail; as such the model was expected to predict no
impact. The contact force time history shown in Fig. 4.2 proves that the FE model works well,

as no impact was found with the contact force remaining at 150kN level (equivalent to static
wheel load) throughout the travel and with the distinct absence of impact.
FIGURE 4.2 -Contact force history of IRJ and Continuous Rail
4.1.1 Parameters of IRJ design considered in sensitivity studies

There are a range of designs of IRJ available. Sensitivity of a few major design parameters is
reported. The basic design parameters examined are illustrated in Fig. 4.3; the sensitivity of
these parameters to wheel/rail impact are reported in this section.
Insulation material
End post material
bonding type
Joint bar type
Gap size
Supporting system
Sleeper position
type
FIGURE 4.3 - IRJ Design Parameters Examined

The design parameters considered are:

 end post bonding detail: glue or inserted
 gap size: 5mm or 10mm
 length of joint bar: 4 bolts or 6 bolts long
 end post material: Nylon66, Fibreglass or Polytetrafluoroethylene (PTFE).
 joint suspended or directly supported on sleeper.
In addition to the above, an operational parameter, namely sliding of wheels across the joint
was also considered and compared to the rolling case. In all analyses, other than the sliding
analysis, the wheel was considered as undergoing pure rolling.
4.1.2 Rolling vs. Sliding
Two types of wheel motion, namely pure rolling and pure sliding, are investigated first.
Locked wheels due to heavy braking/ traction tend to slide and are known as the primary
reason for ‘wheel burn’ type damage even on rails with no joints. The FE model developed
was used to analyse the effect of sliding wheels near the IRJ on the contact force history.
The IRJ containing a glued end post was used for this purpose. Degree of freedom 5 of the
wheel was arrested to simulate dragged wheels. The contact force history shown in Fig. 4.4
illustrates the increase in impact force (194kN – 174kN = 20kN for a static wheel load of
150kN representing 13% increase) that is significant. It is, therefore, important that the
operating vehicles ensure good rolling of wheels through application of gentle braking/
traction torques.
FIGURE 4.4 - Contact force history of wheel pure rolling and pure sliding
4.1.3 Inserted vs. glued end posts
As the ‘inserted’ end post (non-glued) does not provide additional stiffness to the entire IRJ
structure unlike the glued end post, it is generally expected that the ‘inserted’ type would
generate higher impact. Modelling of the ‘inserted’ case of the end post is complex. To truly
model this case, the end post surfaces and the rail end surfaces should all be initially defined

as free. Progressively due to deformation, the model should account for the development of
contact between these surfaces. For simplicity, the end post for the ‘inserted’ case was
removed. Therefore it is expected that the model would predict high impact force as no
benefit of partial support from the end post is accounted for in the model. The glued and
inserted case modelling can therefore be regarded as lower and upper bound results.
Fig. 4.5 presents the contact-impact force histories of these two cases. The damage potential
due to the increased impact of each wheel passage (185kN – 174kN = 11kN for a static wheel
load of 150kN, or 8% increase) requires further investigation as the costs of glueing the end
post against the potential increase in railhead damage requires economic justification. For
the inserted (non-glued) case, the impact occurs 0.15ms later than for the glued case,
corresponding to 48mm travel for a speed of 120Km/h. This suggests enlargement in the
damage area of the IRJ.
FIGURE 4.5 Contact force history of glued and inserted joint

4.1.4 10mm vs. 5mm gap
The gap size of conventional IRJ designs normally ranges from 5mm to 10mm. In this
sensitivity study, two sizes of gap (thickness of end post material) of 5mm and 10mm have
been considered. The impact force time series for these two cases (cases (3) and (1)) are
compared as shown in Fig. 4.6. The numerical result indicates that the small gap size reduces
the impact force by 11KN (174KN-163KN) or 7.3% of the static load of 150KN. Further
economic and technical assessment is required as the thinner gap may increase the
possibility of early electrical isolation failure.

FIGURE 4.6 - Contact force history of 10mm and 5mm gap size
4.1.5 4-Bolted vs. 6-Bolted Joint Bars
Two types, namely 4 bolt long and 6 bolt long joint bars have been considered. The cross
sections of these two joint bars were kept the same and the lengths were 576mm and
830mm respectively. In this design case the 4-bolt joint bar is 254mm shorter than the 6-bolt
joint bar in the longitudinal direction. As the bolt pretension load is kept the same, the four
bolt joint bar has had a lower pretension force in the lateral direction relative to the six bolt
joint bar case. Fig. 4.7 illustrates that the 6-bolt joint bar IRJ generates a slightly larger
impact force of 178KN compared to the 4-bolt joint bar case of 174KN. A conclusion can be
drawn that the effect of the joint bar length and number of bolts on the impact force is not
evident for reasons as explained in this section.
FIGURE 4.7 - Contact force history of 4-bolt and 6-bolt joint bar

Considering the sleeper clear spacing is 564mm (Fig. 4.8), joint bars always span across
sleepers whether 4-Bolt (576mm) or 6-Bolt (830mm) designs are used. Therefore, their
effect on impact is not significant in the cases considered. However, with 6-bolt joint bars, a
larger sleeper spacing may be adopted. In the event of larger sleeper spacing, the 4-bolt case
might generate larger impacts due to the larger dip (deflection) under wheel passage.
FIGURE 4.8 - Illustration of sleeper spacing and joint bar length

4.1.6 Suspended vs. Supported IRJ
The sleeper position effects on the wheel/rail contact impact at the IRJ were studied by
positioning the end post either symmetric to the sleepers (suspended IRJ) or directly on the
sleeper (supported IRJ). Fig. 4.9 shows that the supported IRJ has generated an impact force
of 192KN, while the suspended IRJ has generated just 174KN impact force. The impact force
difference is almost 12% of the static load of 150KN. The supported IRJ also exhibited
significant post-impact vibration.
FIGURE 4.9 - Contact force history of IRJs suspended or supported

4.1.7 Discussion
Through the analyses of these eight design cases, the effects of the selected design
parameters on the wheel/rail contact-impact forces are investigated and a few conclusions
are drawn.
 The glued IRJ performs better than the inserted (non-glued) IRJ. The IRJ with a
smaller gap size generates less impact.
 The higher the flexibility of the supporting system, the lower the wheel/rail
contact-impact.
 The effects of joint bar length seem not to be significant to the wheel/rail impact
force based on the numerical results. Both the 4-bolt joint bar and the 6-bolt joint
bar have just enough length to span across the clear spacing of the sleepers. If the
sleeper spacing had been larger than the 4 bolt joint bar length, the result would
have been different.
 The stiffer the end post material (fibreglass in this case), the lower the impact forces.
This is because the material with mechanical properties closer to steel decreases the
discontinuity in stiffness in the vicinity of the end post.
 It seems not a good choice to place a sleeper directly underneath the IRJ end post.
The directly supported IRJ generates much larger impact forces relative to the
suspended IRJ.
In summary, to minimise the wheel/rail contact impact force at the IRJ, the best design
parameter combination is a fibreglass end post with 5mm gap size that is glued to the rail
sections suspended between the flexible supporting system. Fig. 4.10 indicates that the
impact force is largely eliminated in this case. From a practical perspective, this case may be
considered technically optimal.
155KN
FIGURE 4.10 - Contact force history for Low Impact IRJ

4.2 Non-Ratchetting IRJ

As explained in Chapter 3, ratchetting is mainly due to the stress concentration that occurs
at the top of the unsupported edge of the railhead. This stress concentration is entirely due
to the contact made by the loaded wheels that pass through the gap and land on the
railhead edge. Where the wheel contact is away from the corner, the railhead stress
concentration location migrates to the underside of the railhead similar to the stress
distribution in a continuous rail.
The current designs – a large number of patented systems are available in the literature –
always have their cut surfaces perpendicular to their respective longitudinal axes, which
allows the wheel to travel to the edge of the cut (unsupported) face of the railhead. All these
designs will therefore experience ratchetting regardless of how stiff the joint bars are/ how
many bolts are used/ the size of the railhead or the material of the railhead. The rate of
ratchetting may differ depending on the actual magnitude of the stress concentration and
the yield strength of the material, but the ratchetting will not disappear in these designs.
With the non-ratchetting IRJ developed as part of the research sponsored by the CRC, the
railhead is so shaped that the stress concentration, which is the root cause for ratcheting, is
completely eliminated. This was possible due to the shaping of the railhead that modified
the wheel contact in such a manner that the contact pressure singularity was fully
eliminated. The shape was developed through a rigorous process of optimisation embedded
with a nonlinear, parametric finite element algorithm and is published in Zong and
Dhanasekar (2012); full details are found in Zong (2013). For completeness, the stress
distribution (Von Mises) in a current design and the optimal design are presented in Fig. 4.11.
In this figure, the top of the railhead at the gap should be observed to appreciate the
significance of the changes to both the magnitude and location of the stress concentration in
the current design compared to the edge shaped non-ratchetting design.

Stress (MPa) Railhead

von Mises Surface Current Design
1852.86
1698.85
1544.83
1390.82
1236.81
1082.80
928.78
774.77
620.76
466.74
312.73
158.72
0.00
x y
Rail Edge Rail Symmetric
z Surface
Stress (MPa) Railhead
von Mises Surface Non-Ratchetting
923.32
838.02
762.81
687.54
612.23
536.96
461.66
386.32
311.02
235.77
160.45
85.13
0.00
x y Rail Edge Rail Symmetric
Surface
FIGURE 4.11 - Stress Concentration in the Current and Non-Ratchetting IRJs
It can be observed that, due to the migration of the stress concentration below the top of
the railhead away from the vulnerable unsupported sharp corner, the railhead at the gap
will have no reason to ratchet. Further, the stresses are just below the railhead yield and
hence will be safe. Further reduction in gap size reduces the stresses even more, making it
safe. For example, a 2mm gap size reduces well below yield, thus taking the joint stresses to
the levels of the stresses in the CWR. The design can also be regarded as a ‘gapped CWR’.
The stresses in the 6mm and 2mm gapped IRJs are presented in Fig. 4.12 for ease of
comparison of the effectiveness of simple shaping.

Stress in Current Design

2000 Stress in Non-Ratchetting
1800
Maximum Von mises Stress
1600
1400
1200
1000
800
600
6 2
Gap Size (mm)
FIGURE 4.12 Stresses in the Current and Non-Ratchetting IRJs
The shaping of the ends of the IRJs thus eliminates the ratchetting risk and hence a much
smaller gap (2mm) would be possible without any risk of potential electrical short-circuiting.
The current design that suffers from ratchetting cannot be brought so close; a 5mm limit
appears sensible.
The stresses are a static measure; the measure of dynamic impact of the two designs was
studied using the explicit dynamic FE modelling method. From the contact impact forces,
impact factors were determined and are presented in Fig. 4.13.

Impact factor- Current Design

1.2 Impact Factor - Non-Ratchetting IRJ
1.1
1
Impact Factor
0.9
0.8
 FOR TRACK MAINTENANCE CREW
o Installation of the new designs in the field
0.7
o Methods of checking/ recording
 Issues related to maintenance routine for IRJs: wear limit/ grinding/ wheel load…
 Weld0.6
repair
FIGURE 4.13 Impact Factor in the Current and Non-Ratchetting IRJs
0.5
6 Gap Size (mm) 2
It can be observed that the impact factors in the current design and the non-ratchetting
designs of the IRJs are very similar. With the 2mm gaped design, the non-ratchetting IRJ
FIGURE 4.13 – Impact factor in the Current and Non-Ratchetting IRJs
It can be observed that the impact factors in the current design and the non-ratchetting
designs of the IRJs are very similar. With the 2mm gapped design, the non-ratchetting
IRJ can potentially produce a lower impact compared to the current design of the IRJ,
the gap of which cannot be reduced to the 2mm level.
There are several significant benefits that can arise from the simple shaping of the
railhead of the IRJs. They are:
1. The elimination of the contact pressure singularity is not only beneficial to the
railhead – but also to the wheel.
2. The elimination of stress concentration eliminates the risk of ratchetting.
3. Gap size can be reduced significantly – for example, 2mm in the non-ratchetting
designs.
4. The non-ratchetting IRJ can function similarly to the CWR, thus minimising the
need for speed restrictions and track closure for frequent replacement/
maintenance of the IRJs as happens now.

5. CONCLUSIONS
The current designs of insulated rail joints are manufactured as per AS1085.12 (2002) in
factories with stringent quality measures and installed on-site with highly skilled technical
crews; they are maintained on a regular basis. In spite of all these safeguards, they exhibit
low mean (341MGT), high variable (30% CoV) service life. In comparison to the 2000MGT life
obtained from the CWR, this is very low and requires frequent replacement, which locks the
productive capacity of the rail network with speed restrictions and closures. With a view to
thoroughly exploring the response of the IRJs and suggest improvements, a major research
project was sponsored by the CRC for Rail Innovation. This guide is the result of the research;
it has reviewed the current IRJs and provided data from the field tests, lab tests and the
finite element modelling methods. The studies provided clear insight into the failure
mechanisms of the IRJs; although they fail due to several mechanisms, the failure can be
classified into just two categories:
1. localised strain related failure
2. global deformation related failure
Corrective measures should conform to these failure mechanisms; what works for solving
the localised failure will not solve the global failure and vice-versa.
The strain related failure is entirely attributed to the wheel – railhead contact that occurs on
top of the unsupported edge of the railhead. This contact violates Hertzian theory, and
introduces contact pressure singularity and very high levels of stress concentration at the
top corner, which is much higher than the yield strength of cost effective materials known in
the industry. With the ongoing increase in axle load, this localised strain will increase,
increasing the residual stresses and work against the usage of a high yield strength material
solution. Clever means of avoiding contact at the free-edge is the only feasible solution
which is referred to in this guide.
To minimise the risk of global failure (for example joint bar failure/ fatigue), the track
modulus around the IRJ can be enhanced, which would minimise the surrounding sleeper
deflections and hence the bending strain/ stress ranges in the joint bars (hence improved
fatigue life). The higher the track modulus, the lower will be the ballast pressure; this can
help keep the rail top without a dip for longer and can avoid the requirement for frequent
tamping.
The higher the track modulus, the lower will be the sleeper acceleration and hence there will
be better fatigue life.
The impact load that occurs in the IRJs is solely the function of the gap; only through
reduction in gap size can the impact be reduced. A 5mm gapped IRJ symmetrically supported
arrangement between sleepers is the lowest possible impact IRJ of the current designs; no
further reduction in gap size can be advised due to ratchetting risks of these designs. The
non-ratchetting IRJs offer the potential for a further reduction in gap size.

REFERENCES
Akhtar, M & Davis, D 2011, Rail Joint Assembly Using Embedded Load Transfer Keys and
Method Therefor, US Patent 7871015B2,USA.
AS 1085.1 2002, Railway Track Materials, Part 1: Steel Rails, Standards Australia, Sydney.
AS 1085.12 2002, Insulated Rail Joints, Standards Australia, Sydney.
AS1391 2007, Metallic Materials-Tensile Testing at Ambient Temperature, Standards
Australia, Sydney.
Askarinejad, H, Dhanasekar, M & Cole, C 2013, ‘Assessing Effects of Track Input to the
Response of Insulated Rail Joints Through Field Experiments’, Proceedings of the Institute of
Mechanical Engineers, Part F Rail and Rapid Transit, UK, 227(2), pp.176-187
Bandula-Heva, TM, Dhanasekar, M & Boyd, P 2013, ‘Experimental Investigation of
Wheel/Rail Rolling Contact at Railhead Edge’, International Journal of Experimental
Mechanics, 53, pp.943–957.
Bari, S. & Hassan, T. 2000. Anatomy of coupled constitutive models for ratcheting simulation.
International Journal of Plasticity, 16, pp.381-409.
Barati, P 2013, Response of Rail Track Superstructures with and without Geogrid
Reinforcement, Master of Engineering Thesis, Queensland University of Technology,
Australia.
Cai, W, Wen, Z, Jin, X & Zhai, W 2007, ‘Dynamic Stress Analysis of Rail Joint with Height
Difference Defect using Finite Element Method’, Engineering Failure Analysis, 14, pp.
1488-1499.
Charlton, ZI 2007, Innovative design Concepts for Insulated Rail Joints, Master of Science
Thesis, Virginia Polytechnic Institute & State University, Blacksbeg, USA.
Chen, YC 2003, ‘The effect of proximity of a rail end in elastic-plastic contact between a
wheel and a rail’, Proceedings of the Institute of Mechanical Engineers, Part F Rail and Rapid
Transit, UK, 217, pp. 189-201.
Chen, YC & Kuang, JH 2002, ‘Contact Stress Variations Near the Insulated Rail Joints’,
Proceedings of the Institute of Mechanical Engineers, Part F, Journal of Rail and Rapid Transit,
216, pp. 265-274.
Click, G & Duffner, B 2007, Insulated Rail Joint Assembly, US Patent 2007/0272762 A1, USA.
CoP 2013, Code of Practice for Rail, Engineering Track & Civil, ARTC.
Cox, S J 1995, ‘Deflection of Sleeper in Ballast’ Vehicle System Dynamics 24 (sup1), pp.
146-153.
Davis, DD & Akhtar, MN 2006, ‘Reduced Impact Bonded Insulated Joint Designs’, Railway
Track & Structures, 2006, pp. 17-19.
Doyle, NF 1980, Railway track design: a review of current practice, Occasional paper no. 35.
Canberra, Australia, Bureau of Transport Economics.

ESC220 2011, Rail and Rail Joints, Version 4.6, Engineering Standard, Track, Railcorp NSW.
Jiang, Y & Sehitoglu, H 1994, ‘Cyclic ratchetting of 1070 steel under multiaxial stress states’,
International Journal of Plasticity, 10, pp. 579-608
Page, DR 1903, Railway Rail Joint, US Patent 723,543A, USA.
Portier, L, Calloch, S, Marquis, D & Geyer, P 2000, ‘Ratchetting under tension-torsion
loadings: experiments and modelling’, International Journal of Plasticity, 16, pp. 303-335.
Priest, J & Powrie, W 2009, ‘Determination of dynamic track modulus from measurement of
track velocity during train passage’, Journal of geotechnical and geoenvironmental
engineering, 135 (11), pp. 1732-1740.
RAP5138 2006, ‘Inspection of Insulated Rail Joints’, Engineering Practices Manual, civil
Engineering, ARTC.
Ringsberg, JW, Franklin, FJ, Josefson, BL, Kapoor, A & Nielsen, JCO 2005, ‘Fatigue Evaluation
of Surface Coated Railway Rails using Shakedown Theory, Finite Element Calculations, and
Lab and Field Trials’’, Int J of Fatigue, 27, pp. 680–694.
RTS3655 2006, ‘Maintenance of Mechanical Joints and Examination of Rail Ends in Welded
Track’, Engineering Practices Manual, Civil Engineering, ARTC.
SandstrÖm, J & Ekberg, A 2009, ‘Numerical study of the mechanical deterioration of
insulated rail joints’, Proceedings of the Institution of Mechanical Engineers, Part F: Journal
of Rail and Rapid Transit, 223, pp. 265-273.
Suzuki, T, Ishida, M, Abe, K & Koro, K 2005, ‘Measurement of Dynamic Behaviour of Track
near Rail Joints and Prediction of Track Settlement’, QR of RTRI, 46(2), pp. 124-129.
TS 3402 2001, Specifications for Supply of Aggregates for Ballast, Rail Infrastructure
Corporation of NSW, Australia.
Wen, Z, Jin, X & Jiang, Y 2005, ‘Elastic-Plastic Finite Element Analysis of Non-steady State
Partial Slip Wheel-Rail Rolling Contact’, Journal of Tribology, 127, pp. 713-721.
Zong, N 2013, Development of Optimal Designs of Insulated Rail Joints, PhD Thesis,
Queensland University of Technology, Brisbane.
Zong, N, Askarinejad, H, Bandula-Heva, T & Dhanasekar, M 2013, ‘Service Condition of
Railroad Corridors around the Insulated Rail Joints’, Journal of Transportation Engineering,
ASCE, 139(6), pp. 643-650.
Zong, N & Dhanasekar, M 2012, ‘Minimisation of Railhead Edge Strains through Shape
Optimisation’, International Journal of Engineering Optimization,
DOI:10.1080/0305215X.2012.717075.

Manual for the design, installation and maintenance of insulated rail joints
APPENDIX – DIGITAL IMAGE CORRELATION FOR STRAIN
MEASUREMENT
The digital image correlation (DIC) referred to in this guide is based on a Particle Image
Velocimetry (PIV) method, which was originally developed in the field of experimental fluid
mechanics (Adrian 1991) to measure the particle velocities of a seeded flow. The PIV
technique was modified by White and Take (2003) for geotechnical testing. The MATLAB
module that they developed to analyse digital images was called ‘GeoPIV’, reflecting its
primary use for geotechnical applications (Thusayanthan et al. 2007). This GeoPIV module
was first used with a metal application to determine stress-strain characteristics of rail steel
in this research and it was reported in Bandula-Heva et al. (2013).
Figure A1 - Principles of digital image analysis for strains
Let’s consider two patches (patch 280 and 281 – Fig. A2) located horizontally next to each
other.
(u f 281  u f 280)  (uo 281  uo 280)

H  (A1)
(uo 281  uo 281)
(1967.19  1887.0933)  (1963  1883)

H   1.2  103
(1963  1883)
Similarly the vertical strain at a midpoint between two patches vertically next to each other
is determined as follows:
(v f 2  v f 1)  (vo 2  vo1)
V  (A2)
(vo 2  vo1)
The shear strain also can be determined using four patches in the first image located as
shown in Fig. A3.

 s  Angle A  Angle B (0-3)
Since the angle A and B are small
A  tan( A) and B  tan(B) (0-4)
Then the shear strain is given by
(v f 2  v f 1) (u f 3  u f 1)
s   (0-5)
(u f 2  u f 1) (v f 3  v f 1)
Patch 280 Patch 281 Patch 280 Patch 281

80
80
(u0 281- u0 280) (uf 281- uf 280)

(a) First image (b) Second image
Figure A2 - Successive images of two adjacent ‘patches’
Table A1 - Description of the symbols in the output text file
Symbol Description
Patch Patch ID (Patch number)
uo Horizontal coordinates (x) of the patch in the first image
vo Vertical coordinates (y) of the patch in first the image
uf Horizontal coordinates (x) of the patch in the second image
vf Vertical coordinates (y) of the patch in the second image
du Horizontal displacement vector of the patch in the horizontal (x)

direction
dv Vertical displacement vector of the patch in the horizontal (y)
direction
size The width of a side of square patch

Patch 1 Patch 2 Patch 1

Angle A Patch 2
Angle B
Patch 3 Patch 4 Patch 4
Patch 3
(a) First image (b) Second image
Figure A3 - Successive images of four adjacent ‘patches’

R3.100 Final Report

Uploaded by

Copyright:

Available Formats

You might also like

R3.100 Final Report

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

R3.100 Final Report

Uploaded by

Copyright:

Available Formats

A Guide to

DOCUMENT CONTROL SHEET

CRC for Rail Innovation Document:

GPO Box 1422 Author: Manicka Dhanasekar

Brisbane Qld 4001 Project No.: R3.100

Tel: +61 7 3221 2536 Project Name: Insulated Rail Joints

Fax: +61 7 3235 2987

0 [insert date] Colin Cole Ian Marks Chris Gourlay

CRC for Rail Innovation October 2013 Page 1

CRC for Rail Innovation October 2013 Page 2

1.1 INSULATED RAIL JOINTS (IRJS) ................................................................................................................... 1

2. CURRENT STATE OF IRJS ................................................................................................................... 8

2.1 IRJS IN COAL CORRIDORS .......................................................................................................................... 8

3. TOWARDS IMPROVING THE PERFORMANCE OF THE IRJS ................................................................ 14

3.1 MONITORING PROGRESSIVE DEGRADATION OF THE IN-SERVICE IRJS: .............................................................. 14

4. LONGER LIFE NEW GENERATION IRJS ............................................................................................. 71

4.1 LOW-IMPACT IRJ................................................................................................................................... 71

CRC for Rail Innovation October 2013 Page 3

4.1.7 Discussion ............................................................................................................................... 78

APPENDIX – DIGITAL IMAGE CORRELATION FOR STRAIN MEASUREMENT ............................................... 86

CRC for Rail Innovation October 2013 Page 4

LIST OF FIGURES AND TABLES

CRC for Rail Innovation October 2013 Page 5

Figure 3.31 - Plastic Zones of Rail End Face

CRC for Rail Innovation October 2013 Page 6

Figure 4.2 -Contact force history of IRJ and Continuous Rail

Table 3.1 - Properties of Rail Steel

CRC for Rail Innovation October 2013 Page 7

CRC for Rail Innovation October 2013 Page 8

Robert Taylor BEng(Civil & Mech), Formerly ARTC

CRC for Rail Innovation October 2013 Page 9

CRC for Rail Innovation October 2013 Page 10

1.1 Insulated Rail Joints (IRJs)

FIGURE 1.1 - Four IRJs form a Track Circuit

Current designs of IRJs contain a number of components:

CRC for Rail Innovation October 2013 Page 1

 six (or four) bolts, nuts & washers

FIGURE 1.2 -Components of the Insulated Rail Joints (IRJs)

CRC for Rail Innovation October 2013 Page 2

FIGURE 1.4 -Ratchetting Failure

Ratchetting is affected by the wheel-rail rolling contact pressure singularity (which is a

CRC for Rail Innovation October 2013 Page 3

Wheel load Wheel load

Unsupported or supported edges

FIGURE 1.5 - Contact Pressure Singularity

CRC for Rail Innovation October 2013 Page 4

FIGURE 1.6 - A 12 Bolt IRJ (Akhtar and Davis, 2011)

CRC for Rail Innovation October 2013 Page 5

1.2 Outline of the Guide

This guide is divided into five chapters.

CRC for Rail Innovation October 2013 Page 6

Chapter 5 provides a summary and conclusions.

CRC for Rail Innovation October 2013 Page 7

2. CURRENT STATE OF IRJs

2.1 IRJs in Coal Corridors

Penetrated ballast Capping Layer

FIGURE 2.1 - Ballast Pockets

CRC for Rail Innovation October 2013 Page 8

2.2 IRJs in Passenger-Freight Corridors