Engineering Failure Analysis 18 (2011) 2305–2315

Contents lists available at SciVerse ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier .com/locate /engfai lanal

Experiments into impact behaviour of railway prestressedconcrete sleepers

Sakdirat Kaewunruen a,⇑, Alex M. Remennikov b

a RailCorp – Track Engineering, Level 13, 477 Pitt St., Sydney, NSW 2000, Australiab School of Civil, Mining, and Environmental Engineering, Faculty of Engineering, University of Wollongong, Wollongong, NSW 2522, Australia

a r t i c l e i n f o

Article history:Received 17 May 2011Received in revised form 11 August 2011Accepted 27 August 2011Available online 10 September 2011

Keywords:Prestressed concrete sleeperImpact testingBallasted railway track

1350-6307/$ - see front matter Crown Copyright �doi:10.1016/j.engfailanal.2011.08.007

⇑ Corresponding author. Tel.: +61 02 8922 1151;E-mail address: sak.kaewunruen@railcorp.nsw.go

a b s t r a c t

On railway track structures, dynamic impact loads with very high magnitude but shortduration are often caused by wheel or rail abnormalities such as flat wheels and dippedrails. The possibility of the large impact loading to cause an extreme failure to an in situconcrete sleeper could be very low about once or twice in the design life cycle. However,to the current knowledge, the behaviour of the in situ prestressed concrete sleepers underthe ultimate impact loading has not yet been comprehended, resulting in the design defi-ciency. A high-capacity drop-weight impact testing machine was thus constructed at theUniversity of Wollongong, in order to evaluate impulsive resistance of in situ prestressedconcrete sleepers under impact loads. This paper describes the detail of the high-capacityimpact testing machine, as well as the instrumentation, the calibration, and the analysis offailure mode, crack propagation, flexural toughness, and energy absorption mechanismswith respect to railway prestressed concrete sleepers. The impact tests were carried outusing the prestressed concrete sleepers manufactured in Australia. An in situ track testbed was simulated in laboratory and calibrated against the frequency response functionsobtained from the experimental modal analysis. The experiments using the high-capacityimpact testing machine to investigate the impact energy transfer mechanism of the pre-stressed concrete sleepers are highlighted.

Crown Copyright � 2011 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Railway sleepers (or called ‘railroad tie’) are an important element of railway track structures. Their designed function isto distribute loads from the rail foot to the underlying ballast bed. Fig. 1 shows a typical ballasted railway track. Substructure(ballast, sub-ballast, subgrade, and formation) and superstructure (sleeper, rail pad, fastener, and rail) constitute a railwaytrack system. It is believed, based on the industry experience, that prestressed concrete sleepers have reserves of strengththat are untapped. It is thus important to ascertain the spectrum and amplitudes of forces applied to the railway track, tounderstand more clearly the manner in which track components respond to those forces, and to clarify the processes where-by concrete sleepers in particular carry those actions. In addition, cracks in concrete sleepers have been visually observed bymany railway organizations, including the field trials by the authors [1]. As noted in a review report [2], the principal cause ofcracking is the infrequent but high-magnitude wheel loads produced by a small percentage of abnormal wheels or railheadsurface defects. Those loads are of short duration but of very high magnitude. As an example, the typical loading durationproduced by the wheel flats could vary between 1 and 10 ms, while the force magnitude could reach up to 600 kN per railseat. Existing design philosophy [3] for prestressed concrete sleepers is based on the allowable stress principle taking into

2011 Published by Elsevier Ltd. All rights reserved.

fax: +61 02 8922 1154.v.au (S. Kaewunruen).

Ballast &Subgrade

SleeperRail

Pad

Fig. 1. Typical components of railway tracks.

2306 S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315

account only the quasi-static wheel loads, which results in overly conservative, deficient design for concrete sleepers. Limitstates design concept, which considers the probabilistic dynamic loading condition, is a more logical entity for the develop-ment of a new design approach for prestressed concrete sleepers. Current knowledge to date is insufficient with respect toconstituting the new limit states design concept. As a result, the research tasks are required to perform fundamental studiesof the loading conditions, the static behaviour, the dynamic response, and the impact resistance of the prestressed concretesleepers [3]. A major research effort at the University of Wollongong was to evaluate the ultimate capacities of concretesleepers under static and impact loads.

This paper focuses on the impact testing methodology of prestressed concrete sleepers, and the study on energy transfermechanism in a simulated track environment. The emphasis of this paper is also placed on the experimental verification ofthe ultimate impact resistance and failure modes of railway prestressed concrete sleepers. The prestressed concrete sleepersused were designed in accordance with Australian Standard: AS1085.14 [4]. The support condition provided by the trackwere simulated using a resilient material and validated against the in-field and laboratory vibration measurements [5]. Inthis paper, the experiments using the high-capacity impact testing machine to evaluate impact response of the prestressedconcrete sleeper in a rail track with soft support condition are demonstrated. The test results could be used for validating anumerical model for design purpose and for predicting a transfer mechanism of other track configurations upon which it isbeyond the scope of this paper.

2. Experimental overview

2.1. Specimens

The prestressed concrete sleepers were supplied by an Australian manufacturer, under a collaborative research project ofthe Australian Cooperative Research Centre for Railway Engineering and Technologies (Rail CRC). The typical full-scale pre-stressed concrete sleepers, which are often used in broad gauge tracks, were selected for these tests. The dimensions and shapeof the prestressed concrete sleeper specimens are shown in Table 1. The high strength concrete material was used to cast theprestressed concrete sleepers, with the characteristic compressive strength at 28 days of 55 MPa, and the prestressing steelsused were the high strength with rupture strength of 1860 MPa. However, the cored samples, drilled from the sleepers, weretaken for a confirmation test, as per the Australian Standard AS1012.14 [6], given that the average compressive strength at thetest age of about 2 years was 80 MPa. It had been believed that the high strength prestressing wires were of high quality and thestrength would not change during time. The test specimens of prestressed concrete sleepers are shown in Fig. 2.

2.2. Support condition

In a railway track, a sleeper is supported by a layer of loose, coarse, granular materials with high internal friction,so-called ‘ballast’. It is often a mix of crushed stone, gravel, and crushed gravel through a specific particle size distribution.Using a typical ballast bed leads to difficulties in controlling the boundary conditions in either the experiments or the

Table 1Dimensions and masses of the test sleepers.

Mass (kg) Gauge length (m) Total length (m) At railseat (m) At centre (m)

width depth width depth

299.5 1.60 2.85 0.22 (top) 0.21 0.22 (top) 0.180.25 (bottom) 0.25 (bottom)

Fig. 2. Test specimens.

S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315 2307

numerical simulations, as to equally distribute supporting pressures underneath the sleepers’ soffit, as to retain strengthafter multiple high-velocity impact blows, and as to repeat the test setup. An alternative approach that uses polymeric mate-rials (or ‘rubber mat’) to replace the real ballast has been adopted for a specific use in Australian Standard, AS1085.19J [5]. Thealternative method specifies that, when the support is subjected to an increase in static load from 50 kN to 60 kN at a railseat,it shall allow a vertical displacement between 0.1 mm and 0.5 mm (inclusive). The measurements of static stiffness of therubber mat were performed to select a suitable material. It was found that conveying belts used in the mining industry isa suitable one, based on both the static testing and the nondestructive dynamic testing [7]. It was also found that varyingthe number of rubber mat layers can mimic any track supporting bed, e.g. 2–3 layers of such could replicate a stiffness ofthe hard track bed. Note that the deformation of sleepers was the key measurable component to be focused in this studywhereas there are uncertainties in frictional dissipation of ballast. This study thus neglected the ballast friction and empha-sized on the lower bound of the problem, especially when the friction of pulverized ballast was compromised [7].

2.3. Experimental setup

In this study, the drop height and the drop mass were selected to simulate a typical impact load due to a wheel flat of20–25 mm, which could generate an impact of over 600 kN. The weight of the projectile was set as 5.81 kN (592 kg), andtherefore, the drop height becomes the only variable. The experimental setup was thus identified for specific energy absorp-tion capacity for particular sleepers, so that the back calculation for the optimum drop height could be attained. The dropheight was adjusted from a series of pre-test experiments to cause complete collapse under a single blow [3]. In order toeliminate surrounding noise and ground motion; the concrete sleepers were set up on the prepared support and placedon a strong isolated floor in the laboratory. The strong isolated floor is made of reinforced concrete using high strength con-crete. The large concrete mass is of 5 � 3 � 2.5 m over a compacted sand bed and surrounded by 50 mm-thick shock absor-ber material. The strong isolated floor contains the annex of steel rods. The rods with a threaded hole were installed toprovide the screw supports. The impact plate was installed at a railseat to transfer the load to the specimens and the railwas mounted to the strong floor using the thread bars at the other railseat. The roller was attached to the steel drop massthrough runners, which guide the descent of the drop weight hammer. This type of guiding arrangement provides very lowfriction during a free fall. The hammer was hoisted mechanically to the required drop height (at ultimate capacity of sleeper)and released by an electronic quick release system [3]. The impact test rig is capable of fitting the test specimen and its fix-ture with a working area of 1800 � 5000 mm.

2.4. High-capacity drop-weight impact testing machine

The schematic of the impact testing machine is illustrated in Fig. 3. The core of the test rig is the free-fall hammer that canbe dropped from a maximum height of 6 m, or equivalent to the drop velocity up to about 10 m/s. The impact load was mon-itored and recorded by the dynamic load cell connected to the computer. Efficiency of drop weight hammer was evaluatedthrough the calibration tests done using high speed camera. It was found that due to friction of guiding runner the hammer’sexperimental velocity reduces to 98% of theoretical velocity (Vt). To achieve the prescribed impact velocity, the required dropheight was adjusted by the coefficient ð1=0:982Þ ¼ 1=0:96. Based on the energy conservation theory, the relationship be-tween the drop height (h: in m) and the drop velocity (V: in m/s) reads

0:98V ¼ffiffiffiffiffiffiffiffi2gh

pð1Þ

g is the gravitational acceleration (=9.81 m/s2). A series of impact tests on concrete sleepers were conducted using the sametest setup as described above. A special attention was paid to the different drop heights on the prestressed concrete sleepersin order to classify the different damage severity levels. Note that this experimental test rig can be applied to investigate the

Fig. 3. Illustrative layout of the drop-weight impact test rig.

2308 S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315

failure and energy absorption mechanisms, impact load history (from the dynamic load cell), accelerations, and crack prop-agation of any engineering structure.

2.5. Calibration of impactor using a high-speed camera

Table 2 presents the efficiency of the impact test rig evaluated using the high speed camera. The theoretical free fallingvelocity was computed from the conservation of energy. The measured free falling velocity was obtained from the movingtime frames corresponding to the movement of the object. The distance marks on the impactor and a reference column wereidentically scaled to be 100 mm long. The moving time frames were captured when the mark on the impactor had fullypassed by an adjacent mark on the reference column. Three drop tests were done at each drop height and the average dropvelocity could be obtained. The tendency of the test rig efficiency, which is gently decreased as the drop height increases,could then be established. Averagely, at all drop heights, the loss of the drop velocity was found to be about only 2%.

2.6. Instrumentation

At each drop height, the measurements for impact testing included the impact load history, the accelerations, the crackwidth, and the damage severity. The dynamic load cell was used for measuring the impact load history. The maximum capac-ity of the dynamic load cell is as much as 2000 kN. The accelerations were captured by using the Dytran Series 3200B accel-erometers [8] with the peak amplitude of up to 10,000g (where g is the Earth’s gravitational acceleration or about 9.81 m/s2).Crack width was obtained by using the magnifier glass telescope with a resolution of 0.1 mm. The fibre strains weremeasured by using the foiled strain gauges manufactured by BCM Sensor Technologies [3]. Strain measurements were tar-geted at the top and the bottom fibres of both cross sections at railseat and at mid span, in order to back calculate the cur-vatures of the concrete sleeper and the resultant bending moments under impact loading.

Table 2Efficiency of high capacity impact testing machine.

Drop height (m) Measured free fallingvelocity (m/s)

Theoretical free fallingvelocity (m/s)

Efficiency (%)

1 4.33 4.43 97.762 6.12 6.26 97.705 9.65 9.90 97.43Average 97.62

S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315 2309

2.7. Data acquisition system

The built-in chassis with coupled SCXI and SCI channels from National Instruments was used to acquire the data signalsfrom the 24-channel strain gauge unit, the 4-channel digital voltage platform, and the 8-channel acceleration box. The dataacquisition system was triggered by the impact loading signal obtained from the drop-weight hammer itself during a test.Subsequently, the impact load history, the accelerations, and the strain signals could be recorded. 10-kHz low-pass filteringwas utilized to eliminate white noise and high-frequency signals.

3. Analysis of the results

Energy conservation principle has been employed successfully in impact mechanics of concrete flexural members. Thederivation of fracture energy can be attained by two means; load–deflection capacity; and impulse and momentum. Ithas been found that the conservation of energy can be achieved by considering the initial and final conditions of the impactevent without considering the detailed mechanism of the impact event [9]. At the end of the impact event, the energy is bal-anced between work done by the drop hammer and energy absorbed by associated components in any form of either strain,fracture, or kinetic energy.

3.1. Effect of inertia load

It should be noted that in these experiments the impact load was recorded from the tup or the dynamic load cell attachedto the impactor. In addition, it is well-known that the inertia forces play a major role on the actual bending load applied tothe test specimens. Therefore, the true bending load Pb(t) is the difference between the recorded impact load history Pt(t) andthe generalised inertia load Pi(t), as follows [9].

PbðtÞ ¼ PtðtÞ � PiðtÞ ð2Þ

For any particular setup, the inertia load Pi(t) can be computed from the virtual work expression as follows.

PiðtÞdu0 ¼Z

m � aðtÞduðxÞdx ð3Þ

where a(t) is the acceleration along the sleeper in x direction, du0 is a given virtual displacement, du(x) is the generalisedvirtual displacement, and m is the mass (kg) per unit length (m) of the test specimen [9].

3.2. Energy of the impactor

Kinetic energy of the impactor during the strike on a specimen can be evaluated based on the acceleration of the drop-weight hammer, ah(t). The energy spent by the drop-weight hammer (Eh) is given by [10]

Eh ¼Z

M � ahðtÞdu ð4Þ

where ut is the displacement at time t.On the other hand, the potential energy (Ep) of the drop-weight hammer at a certain height, which would be the same

amount of the kinetic energy in Eq. (5), can be read as [11]

Ep ¼ Mgh ð5Þ

where M is the drop mass (592 kg); g is 9.81 m/s2; and h is the drop height (m).

3.3. Energy loss in the impactor

According to Wang [12], the energy loss in impactor (DE) due to frictions in contact zones can be calculated from the inte-gration of impact force signal as follows.

DE ¼ 12

M1M

ZPtðtÞdt � 0:98

ffiffiffiffiffiffiffiffi2gh

p� �2

� 2ghð0:98Þ2 !

ð6Þ

The factor 0.98 is the efficiency of the test rig. M is the drop mass (592 kg); Pt(t) is the impact load history (kN); g is9.81 m/s2; and h is the drop height (m).

3.4. Energy absorbed by a test specimen

Virtual work done (dW) by an external force (impact load P in this case) can be written in a virtual work form as follows.

dW ¼ Pduo ð7Þ

where duo is the virtual displacement at the load point.

2310 S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315

Based on Birch et al., [13] the actual energy absorbed by the sleepers (EA) can be approximated from the multiplicationbetween the mean plastic impact load and the displacement integrated from the acceleration responses, as follows.

EA ¼ PmpuoðtÞ ð8Þ

and,

uoðtÞ ¼Z t

0

Z t

0€uoðtÞdtdt ð9Þ

where Pmp is the mean plastic impact force (recorded from dynamic load cell). It can be obtained from the average betweenthe maximum impact force and the summit of the elastic loading portion of impact load history. In addition, €uoðtÞ is theacceleration at the load point of time t, and uo is a displacement at the load point of time t.

4. Results and discussion

The experimental programme to evaluate the impact testing setup performance is given in Table 3. The test setuparrangement of the prestressed concrete sleeper in a track system is illustrated in Fig. 4. Impact loading was imparted tothe prestressed concrete sleeper by dropping the mass of 5.81 kN at varied drop heights step by step. The impact forces mea-sured by the dynamic load cell from the direct contact between rail and impactor are presented in Fig. 5. It was found thatthe first peak duration is too short and has not played an influential role on the flexural failure of concrete sleepers in thisstudy [9]. In general, the first high frequency impact peaks would later be filtered out by rail and fastening system [3]. Low-frequency pass signal threshold can be used to filter out the first peak of impact signals. The critical peak of impact force(second peak) corresponds to the flexural behaviour of the concrete sleeper, varying in a frequency range between 160and 200 Hz. Very small bending crack was firstly detected at the drop height of 600 mm but the small shear cracks were alsofound after few blows at the drop height of 800 mm. However, no major failure can be observed from the series of impactloads as shown in Fig. 6. At this point, the impact energy input was equivalent to the fracture toughness energy (obtainedfrom static test) of the concrete sleeper [3]. This implied that under the high strain rate circumstances, the associated trackcomponents tends to dissipate impact energy and the sleeper tends to gain hardening material strength and perform well.

The energy losses in the impactor at each drop height are tabulated in Table 4. The potential energy (Ep) was computedfrom Eq. (6). It was found that the higher drops loss lesser energy in the impactor due to contact frictions. The average per-cent loss in the impactor is about 30% so that only about 70% of the total potential energy transfers to the related track com-ponents. In a railway track, the loss could comparatively be regarded to the energy that would be absorbed partly by: first, atrain wheel in the form of elastic strain and rebound; and later, the sleeper in the form of vibrations. Accordingly the largerwheel irregularities such as wheel burn, radial or longitudinal wheel cracks, or any out-of-round wheel defect could then beworsened; or even the train suspension springs could deteriorate. In addition, the hair cracks in the sleeper can be openedup, especially at mid span (see Fig. 7) where the lowest bending mode plays a role [14]. The experimental results showedthat the vibration has a potential to accelerate the interfacial cracks but only when the impact reaches a certain level ofseverity. It could take thousands of million cycles to open the cracks up and expose the prestressing tendons [3].

Fig. 8 shows an example of acceleration responses at mid-span and at railseat of the in situ railway concrete sleepers.(Note that the displacement responses could also be obtained from the double integration of acceleration responses.) UsingEq. (9), the actual energy absorbed by the sleepers can be approximately computed from the multiplication between themean plastic impact load and the maximum displacement spectra integrated from the acceleration responses. Table 5 dis-plays the results of energy absorption by the sleeper specimen under different drop heights. Note that the energy input wasderived from the remaining energy applied to the track components. This energy input is the difference between the totalpotential energy and the energy loss in Table 4. Table 6 summarizes the energy distribution to each particular element inthe test bed. For this particular track (were found as a soft track in general), the impact energy would averagely be absorbedby the sleepers about 45% in terms of the flexural deformation only. It should be noted that this value is based on the sim-plification that exercises the maximum displacement response and the mean plastic impact load. This value is the upperbound of the actual dynamic performance of the sleeper and the track system. It was found that about 25% of the impactenergy is dissipated by both the ballast supporting system and especially the rail pads. It is important to note that in the

Table 3Experimental programme.

Support Drop height (m) Expected free fallingvelocity (m/s)

Total potentialenergy (J)

Measured maximpact load (kN)

6 Layers of rubber mat(17.5 kN/mm)

0.2 1.94 1161.5 4260.4 2.74 2323.0 7000.6 3.36 3484.5 8830.8 3.88 4646.0 1160

(a)

(b)

contact zone

impactor

rail

rail pad

sleeper/tie

alternative support [AS1085.19J]

accelerometer

Strain gauges

Fig. 4. A setup of the in situ prestressed concrete sleeper. (a) In situ condition setup, and (b) instrumentation.

0

200

400

600

800

1000

1200

1400

4.33 4.335 4.34 4.345 4.35

Time,s

Impa

ct fo

rce,

kN

800mm drop height

600mm drop height

400mm drop height

200mm drop height

Fig. 5. Impact force measurements.

S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315 2311

test setup, the hard rubber pad (high density polyethylene) was used. The softer rail pads such as rubber pads, EVA pads willprovide much more dissipation of the impact energy [14].

Based on the energy balance theory, it was discovered that the dynamic cracking moment of about 29 kNm is associatedwith the static energy absorbed by the sleeper of roughly 1000 J [15]. It could be seen from the impact testing that the

Front

Fig. 6. Initial cracks after the first series of experiments.

Table 4Energy losses in the impactor at different drop heights.

Drop height (m) Energy loss (J) Total potential energy (J) Percent loss (%)

0.2 421.9 1161.5 36.30.4 678.9 2323.0 29.20.6 991.5 3484.5 28.50.8 1267.1 4646.0 27.3Average 30.3

Fig. 7. Centre crack due to the effect of vibration energy.

2312 S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315

required drop height to create cracks in the sleeper for such track configuration is at 0.6 m. At this drop height, the impactenergy absorbed by sleeper is 1413 J, which is slightly higher than the static energy absorption [15]. This phenomenon isascribed to the effect of strain rate, as when the strain rate is high the concrete material becomes stronger but brittle.

The experiments were continued by a cumulative series of impact drops. The drop heights were increased step by step todetermine the ultimate impact behaviour of the concrete sleeper. Fig. 9 depicts the ultimate behaviour of the concrete slee-per under impact loads of about 1800 kN. It was clearly evident that the concrete sleeper fails in the splitting failure mode.This failure mode can also be found in the field (see Fig. 10). The concrete splits along the alignments of prestressing wires,breaking it into three portions. Although there was some concrete spalling at railseat and slight yielding of the prestressingwires, major interfacial crack or wires’ breakage could not be detected. This discovery has led to the concern regarding thedeficient design and analysis of prestressed concrete sleepers. The impact performance of the sleeper implied that thestrength evaluation using the static tests in accordance with the standard is unsuitable to represent the actual behaviourof concrete sleepers laid in ballasted track systems. From the test results, the ultimate impact capacity (of 1800 kN orequivalent to about 108 kNm) is much higher than the static capacity of about 70 kNm [15–18]. The difference between sta-tic and dynamic performance of the concrete sleeper is as much as 40%. This magnitude of capacity difference is referred tothe reserved strength of concrete sleeper that is untapped.

-1000

-500

0

500

1000

2.19 2.2 2.21 2.22 2.23 2.24 2.25

time, sac

cele

ratio

n,g at mid-span

at railseat

Fig. 8. Impact responses at railseat at 0.2 m drop height.

Table 5Energy losses to the support at different drop heights.

Drop height(m)

Max displacement(mm)

Mean plastic impact load(kN)

Energy absorbed by sleeper(J)

Energy input(J)

Energy loss to support(J)

0.2 1.3 406 528 740 2120.4 1.5 660 990 1644 6540.6 1.6 860 1376 2494 11180.8 1.9 1060 2014 3379 1365

Table 6Summary of energy losses.

Drop height(m)

Total potentialenergy (J)

Energy loss in theimpactor (J)

Energy loss to ballast and resilientsupports (J)

Total energyloss (J)

Energy absorbed bysleeper (J)

0.2 1162 422 212 634 5280.4 2323 679 654 1333 9900.6 3485 992 1118 2110 13760.8 4646 1267 1365 2632 2014

Fig. 9. Ultimate impact behaviour of a concrete sleeper.

S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315 2313

Fig. 10. Impact damage of a concrete sleeper found in the field due to trackwork machinery.

2314 S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315

5. Conclusions

Current design manner omits the actions of impact loads on the railway track structures due to the fact that the existingknowledge is insufficient. The goal of this research project was to develop the rational design method for prestressedconcrete sleepers on the basis of limit states design concept. This study included comprehensive investigations in loadingconditions, static and dynamic performance, and impact resistance of prestressed concrete sleepers. This paper, however,presents the use of a high-capacity impact testing machine developed at the University of Wollongong, in order to investi-gate the impact behaviour of an in situ railway prestressed concrete sleeper. The instrumentation was described in details toachieve the adequate experimental results. The fracture energy and energy losses under dynamic conditions of the in situprestressed concrete sleepers as a performance indicator of damage for the selected limit states, were quantitativelyevaluated.

The energy absorption mechanisms of the prestressed concrete sleeper and the railway track structures under impactloading were highlighted in this paper. It can be concluded that the energy balance theory is applicable to the railway sleep-ers and track system. This paper investigated the energy loss in either the impactor (e.g. in wheel suspension) as well as inrail pad and in rubber mat (e.g. in ballast support). It was found that this particular test setup represents a soft railway trackin general (track modulus: 10–20 MPa). The experimental results exhibited that the impact energy up to 60% would partlydamage the wheel, degrade the rail pad, break the ballast gravel, and accelerate the cracks in sleeper. The remaining portionof around 45% would be directly absorbed by the railway sleeper in the form of bending curvature (including fracture).Energy absorption capacity of the prestressed concrete sleeper or so-called the fracture energy can clearly indicate its dam-age severity. It was also discovered that, due to the effect of high strain rate, concrete material plays a dominant role in thedynamic failure mode of prestressed concrete sleepers. In general situations, the standard static testing evaluation does notprovide the realistic representation of the concrete sleeper under the real-life loading condition. The ultimate impact loadresistance is much higher than the prediction using static method of analysis while the dynamic failure mode could be dif-ferent from the anticipated bending failure mode. As a result, the current design concept could be considered as over con-servative; and this investigation led to the limit states and performance based design concept for the prestressed concretesleepers [17–20].

Acknowledgements

The authors are grateful to the Australian CRC for Railway Engineering and Technologies (Rail-CRC) for the financialsupport throughout this study. The authors would like to thank the technical officers, Alan Grant, Ian Bridge, Bob Roland,and Jason Knust, for their assistance during the course of this project. Also, the first author sincerely thanks the AustralianGovernment for the Endeavour Executive Award, which provided him the opportunity to visit Railway Technical ResearchInstitute (RTRI) in Tokyo, Japan.

S. Kaewunruen, A.M. Remennikov / Engineering Failure Analysis 18 (2011) 2305–2315 2315

References

[1] Kaewunruen S, Remennikov AM. Field trials for dynamic characteristics of railway track and its components using impact excitation technique. NDT&EInt 2007;40(7):510–9.

[2] Murray M, Cai Z. Literature review on the design of railway prestressed concrete sleeper. RSTA research report, Australia; 1998.[3] Kaewunruen S. Experimental and numerical studies to evaluate the dynamic behaviour of prestressed concrete sleepers under severe impact loading.

PhD thesis, School of Civil, Mining, and Environmental Engineering, University of Wollongong, NSW, Australia; 2007.[4] Standards Australia. Railway track material – Part 14: Prestressed concrete sleepers. Australian Standard: AS1085.14-2003; 2003.[5] Standards Australia. Railway track material – Part 19: Fastening system. Australian Standard: AS1085.14-2001; 2001.[6] Standards Australia. Method of testing concrete – method for securing and testing cores from hardened concrete for compressive strength. Australian

Standard: AS1012.14-1991; 1991.[7] Kaewunruen S, Remennikov AM. Experimental simulation of the railway ballast by resilient materials and its verification by modal testing. Experiment

Tech 2007;4:29–35.[8] Dytran Instruments Inc. Model series 3200B calibration and operation manual for the University of Wollongong, CA, USA; 2007.[9] Banthia NP, Mindess S, Bentur A, Pigeon M. Impact testing of concrete using a drop-weight impact machine. Experiment Mech 1989;29(1):63–9.

[10] Bhattacharya S, Krishnamurthy KC, Rajendran R, Prem sai R, Basu S. Impact studies on structural components using a free-flight drop tower.Experiment Tech 2006;18:52–8.

[11] Kaewunruen S, Remennikov AM. Investigation on static and dynamic performance of railway prestressed concrete sleepers. In: SEM annual conferenceand exposition on experimental and applied mechanics, June 3–6, Springfield, MA, USA; 2007 [CDRom].

[12] Wang N. Resistance of concrete railroad ties to impact loading. PhD thesis, University of British Columbia, Canada; 1996.[13] Birch RS, Jones N, Jouri WS. Performance assessment of an impact test rig. Proc Inst Mech Eng, Part C 1988;202(4):275–85.[14] Barke DW, Chiu WK. A review of the effects of out-of-round wheels on track and vehicle components. Proc Inst Mech Eng, Part F 2005;219:151–75.[15] Remennikov AM, Kaewunruen S. Experimental determination of energy absorption capacity for railway prestressed concrete sleepers under impact

loading. In: The 4th international conference on structural engineering and construction, September 22–24, Melbourne, Australia; 2007 [CDRom].[16] Wakui H, Okuda H. A study on limit-state design method for prestressed concrete sleepers. Concr Lib Jpn Soc Civil Eng 1999;33(1):1–25.[17] Kaewunruen S, Remennikov AM. Progressive failure of prestressed concrete sleepers under multiple high-intensity impact loads. Eng Struct

2009;31(10):2460–73.[18] Kaewunruen S, Remennikov AM. Dynamic crack propagations of prestressed concrete sleepers in railway track systems subjected to severe impact

loads. ASCE J Struct Eng 2010;136(6):749–54.[19] Kaewunruen S, Remennikov AM, Murray MH. Greener and leaner – unleashing capacity of railroad concrete ties via limit states concept. ASCE J Trans

Eng 2011;137(4):241–7.[20] Remennikov AM, Murray MH, Kaewunruen S. Reliability based conversion of a structural design code for prestressed concrete railway sleepers. Proc

Inst Mech Eng, Part F J Rail Rapid Transit. doi: 10.1177/0954409711418754.