An alternative method for the assessment of railhead traction

Wear 271 (2011) 62–70 Contents lists available at ScienceDirect Wear journal homepage: www.elsevier.com/locate/wear An alternative method for the assessment of railhead traction S.R. Lewis a , R. Lewis a,∗ , U. Olofsson b a b Department of Mechanical Engineering, the University of Sheffield, Mappin Street, Sheffield S1 3JD, UK Department of Machine Design, Royal Institute of Technology, Stockholm, Sweden a r t i c l e i n f o Article history: Received 3 September 2010 Accepted 3 October 2010 Available online 12 October 2010 Keywords: Friction Rail Pendulum tester Contaminant Friction modifier a b s t r a c t Work has been carried out to develop a fast test method for the determining of railhead traction levels. Current methods used in the field are time consuming and offer relatively little control of external or test parameters. A pendulum rig has been used for this investigation and adapted to measure railhead friction levels under various states of contamination. The rig consists of an aluminium tubular pendulum; on the end of which is a spring mounted, rubber pad (slider pad). The rig functions on the same principles as used in a Charpy impact test, i.e. energy is lost as the slider pad comes into contact with a surface (in this case the rail head). This loss in kinetic energy is measured and can be translated into a friction coefficient. Tests have been carried out to validate the placing of the contaminants on the rail prior to testing and also to determine the setup of the rig. High speed video has also been used to determine the speed of the slider. The pendulum was also tested in the field and showed good correlation in comparison with a hand pushed Tribometer. Pendulum results have been compared to those from twin disk simulations of the wheel/rail contact and good correlation can also be found. © 2010 Elsevier B.V. All rights reserved. 1. Introduction Maintaining the correct levels of adhesion in railway transportation is essential. High levels of adhesion are required when a train is approaching (for braking) or leaving a station (in traction), for example. On the other hand low friction levels are desired in curves where flange contact with the rail leads to high wear rates, noise and vibrations being generated. A number of methods have been used to assess traction coefficients in the wheel/rail contact. A number of laboratory bench techniques have been used including pin-on-disc [1], disc-on-flat [2] and twin disc testing (with a line contact) [3,4]. Twin disc testing has also been carried out using scaled wheel and rail profiles [5]. These types of rig allow test parameters to be very closely controlled and are ideal for ranking type tests. However, the test contact is a simplification of the real wheel/rail contact and it is difficult to incorporate changes in environmental conditions (such as temperature and humidity fluctuations) that affect adhesion so greatly. In order to obtain greater accuracy in the representation of the contact geometry, loading and environmental conditions, more complex tests are required. Examples of these include full scale laboratory testing [6] as well as field measurements which ∗ Corresponding author. Tel.: +44 (0)114 2227838; fax: +44 (0)114 2227890. E-mail addresses: [email protected] (S.R. Lewis), [email protected], [email protected] (R. Lewis), [email protected] (U. Olofsson). 0043-1648/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2010.10.035 have been taken using track mounted tribometers [7,8] and instrumented trains [9]. However, at the same time as the accuracy of the contact conditions increase using full scale/field methods, the level of control of operating parameters decreases. These tests are also very time consuming and expensive to run. Track access, particularly in the UK is also difficult to arrange. Gallardo-Hernandez and Lewis [10] compiled data from a variety of tests that indicated that laboratory and field data matched up quite well, indicating that laboratory tests were appropriate for assessing traction coefficients. There is a need, however, for a quick and relatively simple railhead adhesion test method, which is portable and can be used either in the laboratory or out in the field. This would reduce test time and cost and allow a speedy response when adhesion problems are identified on a particular stretch of track for example caused by “wet rail syndrome”, which cannot currently be achieved. In this work a pendulum friction tester (see Fig. 1) was assessed to determine its potential for measuring railhead adhesion. The test method was originally developed for the assessment of skid resistance on road surfaces and has recently been used for testing floor materials. This is not the first time that such an instrument has been used to assess levels of railhead adhesion. A rig of this type was used on actual track during work to develop a stimulant leaf layer to use in field testing of methods to overcome leaf problems, such as laser treatment, traction gels, etc. [11]. Limited data was collected, however, and the technique was not taken any further. In this work a comprehensive set of measurements were taken using the pendulum in the laboratory using a range of test condi- S.R. Lewis et al. / Wear 271 (2011) 62–70 63 manufacturer [14]. This was derived for a contact length, between the rubber pad and the test surface, of 127 mm (5 in.): = Fig. 1. Schematic representation of the pendulum rig showing: (a) energy loss scale; (b) height-adjustable pivot; (c) pendulum swing arm; (d) spring mounted slider (rubber pad attached here); (f) height adjustable feet. tions. Initial tests were used to determine the best method to use. Subsequent tests used this method in studying the effect of different substances in the contact (both naturally occurring and actually applied to the railhead). The pendulum was also field tested on a section of Stockholm underground track which was been treated with Keltrack© friction modifier. These results were compared to those of a Salient Systems Tribometer, which was being used as part of scheduled track maintenance. An extracted piece of underground track was also lab tested. All laboratory and field test data was then compared with a range of data taken from the literature for other measurement techniques. 2. Pendulum test apparatus The pendulum tester was originally designed in the 1940s and is widely used today as a road surface friction assessment tool in the cases of accidents or experimental road surfaces [12]. The test is also used as a standard method of assessing slip resistance of flooring [13]. The pendulum rig works on the same energy loss principle as used in the Charpy impact test. The rig’s mode of operation is as follows: 1. The pendulum swing arm is released from its starting position (as shown in Fig. 1). 2. It then swings through a radius of arc equal to length (b)–(d). 3. As it approaches 90◦ of swing, rubber pad (d) comes into contact with the test specimen (floor, road or in this case rail). 4. The slider is pivoted about point (g) and mounted on a constant force spring which runs up the swing arm (c), this ensures that there is a constant normal force between the pad and the specimen throughout the contact. 5. Friction between the rubber pad and the specimen will produce an energy loss which is then measured on scale (a) as the arm swings on its upstroke. A formula for converting “energy loss”, as read from the scale on the rig, into friction coefficient has been specified by the pendulum  110 SRV − 1 3 −1 (1) where  is friction coefficient and SRV (Slip Resistance Value) is the “energy loss” reading. There are two main adjustments on the pendulum rig. The first is at point (e). This is used to alter the height of the pivot above the specimen. This is mainly used as a fine adjustment for the strike length (distance which pad is in contact with the specimen). The other point of adjustment is at the feet (f). These are primarily used to level the rig but can be also used to adjust the height of slider to reach a particular test surface. Various types of slider can be used with the rig representing different situations. The main two used are the Four-s and TRL rubber. The, harder, Four-s type rubber represents the heel of the average shoe and the TRL (softer pad) with properties similar to the average road tyre and also the heel of a foot. Rubber hardness is measured using the International Rubber Hardness Degrees, IRHD system. The harder rubber has an IRHD value of 96, and the softer, 55 [14]. The hardness test used in the IRHD differs from hardness tests on other materials like metals and ceramics which measure the ability of the material to withstand plastic deformation, i.e. indentation. Instead it measures the modulus of the rubber using a spherical indenter and observing the depth of indentation with a given force [15]. The IRHD scale ranges from 0 (no modulus) to 100 (infinite modulus). The Four-s slider was chosen as results were closer to those seen in other forms of testing. However, tests were also carried out with the TRL slider and are presented as a comparison. 3. Initial tests The aim during initial laboratory testing was to trial the rig and to establish a specification for the main testing to be carried out to assess a number of different railhead conditions. Part of the main testing was to involve using contaminants and friction controlling products, such as oil, water, leaves, traction enhancers and friction modifiers, so it was essential to see how the rig worked with these on the railhead. Nominally dry tests were initially carried out using the method prescribed for road surface friction testing using a new Fours rubber pad. This involved (as detailed in Section 2) using a contact length of 127 mm. Friction coefficients were compared with those from twin disc tests conditions as a reference. Further testing was carried out to optimise contaminant application; establish the effect of using worn pads and pads made from different rubbers. 3.1. Determining dry friction levels The contact length between the rubber pad and the railhead was set at 127 mm by adjusting the height of the pendulum pivot above the rail. The height of the pendulum can be adjusted to alter the strike length of the pendulum. The strike length was altered during the trials and it was shown that a small change can have a significant impact on the indicated friction level. Thus it is critical to check during testing that the strike length is 127 mm. The friction coefficients were then calculated using Eq. (1) from the energy loss readings. Tests were carried out over a number of days during which the environmental conditions in the laboratory varied slightly, as shown in Fig. 2, where the friction coefficients are also plotted. Small changes in humidity throughout the testing period are shown to have a noticeable effect on the friction reading. Temperature measurements taken during the tests did not vary significantly. This effect 64 S.R. Lewis et al. / Wear 271 (2011) 62–70 evenly dispersing the liquid contaminants within it, as shown in Fig. 3(c) and (d). The trial tests were also used to determine the quantity of liquid contaminant to be used. It was found that 4 ml of water and 3.5 ml oil filled the marked contact area. The friction modifiers were too viscous to syringe onto the rail and hence had to be applied directly and then spread with a spatula to give a consistent film thickness. This technique offered little controllability in terms of mass or volume of product applied. However, there was a fixed contact length and width meaning variations in how much product the slider was in contact with were small. Leaves also showed great inconsistencies in their results depending on where they were placed on the rail. It was decided that the leaves should be placed at the beginning of the contact. This ensured that the leaf was in contact with the rubber pad throughout the whole contact stroke. This method greatly increased consistency of the results. Fig. 2. Variation of dry pendulum test results with relative humidity. has been seen before in friction assessment with varying humidity [1]. The humidity in the laboratory is typically around 40% which yields a friction coefficient of 0.5–0.6 which compares favourably with that seen in the saturated slip region of dry creep curves developed during twin disc testing [10]. This appeared promising so a contact length of 127 mm was maintained for all subsequent tests. 3.2. Application of liquid contaminants and products Testing was carried out to determine a controllable technique of applying contaminants to the rail. At first liquid contaminants (water and oil) were spread across the rail at a constant volume. This provided inconsistent results, however, as some water ran off the rail (see Fig. 3(a) and (b)). This problem was solved by marking the area for which the pendulum was in contact with the rail and 3.3. High speed video A high speed camera was employed to firstly determine the amount of time in which the pad was in contact with the rail, and also to observe what exactly was happening to the contaminants when struck by the pendulum. The camera set-up is shown in Fig. 4. The camera trigger was controlled via a laptop computer. Halogen lighting was also used to increase the contrast and clarity of the video capture. The camera was set to focus solely on the contact patch for better clarity and hence did not capture the full swing of the pendulum. Some screenshots taken from a film of a wet contact are shown in Fig. 5. The time in contact was determined with various entities placed on the track and are summarised in Table 1. These were to be used to determine sliding speeds for film thickness calculations. Fig. 3. Contamination application experiments: (a) water spread along the rail; (b) oil dispersed along the rail; (c) water dispersed within marked contact area; (d) oil dispersed within contact area. S.R. Lewis et al. / Wear 271 (2011) 62–70 65 Table 1 Contact time analysis. Condition Contact time (s) Average sliding speed (m/s) Friction coefficient Wet Trackside transit Leaf Keltrack 0.05000 0.03417 0.04666 0.03500 2.540 3.717 2.722 3.629 0.3 0.2 0.2 0.1 a bow wave. Towards the end of each stroke the water would flow out from the open ends of the pad. 3.4. Comparison of hard and soft pads Fig. 4. Camera set-up for high-speed video analysis. Trackside Transit and Keltrack© are both types of friction modifier manufactured by Kelsan Corp. The high-speed footage was also used to observe the contaminants as they were struck by the slider. Footage of water and leaves were the most interesting while videos of oil and friction modifiers did not show much of interest. Leaf footage confirmed what was originally thought; in that the leaf was swept along the rail by the pendulum. It also showed that water is also extracted from the leaf towards the end of the stroke. It was noted during tests how more water would be left on the track after a test with a leaf. Footage also revealed that water would gather up in front of the pad much like The pendulum is designed to work with two types of slider. Just by touch, it is possible to establish that the softer of the two pads (TRL) gave greater adhesion compared with the Four-s type. This was also shown to be the case in dry tests with the TRL pad which gave an average friction coefficient of 1.2, which is twice the value yielded by the Four-s slider. Tests were also carried out using a TRL pad with a wet rail. This yielded an average friction coefficient of 0.14. This was somewhat lower than results from the Four-s pad. The higher reading for a dry test with the TRL slider was due to the softness of the rubber. For the tests with water, the low result may be due to the softer pad deflecting more easily and hence assisting in the formation of a hydrodynamic lubrication film. As the results using a Four-s pad had already been shown to give results in line with those from twin disc testing it was decided to use this type of pad for subsequent tests. Fig. 5. Sequence of high-speed video with leaf and contaminant. 66 S.R. Lewis et al. / Wear 271 (2011) 62–70 Table 3 Oil–water mixture results. Mixture (water:oil) Energy loss C.O.F. Fig. 6. Comparison of old and new pad data (error bars indicate one standard deviation). Water 25.8 0.25 Oil 15.3 0.15 50:50 12.8 0.12 25:75 11.8 0.11 75:25 12.8 0.12 reading was taken. Table 2 gives details of all the tests carried out including information on the contaminants. Note that the friction modifiers tested including Sandite were not able to be applied by syringe and were spread on the rail manually. Even though this gave little control over the volume applied or film thickness, results gave an acceptable level of repeatability. 5. Laboratory test results 3.5. Worn versus new pads 5.1. Water Initial tests were performed using a previously used slider. This slider had worn due to multiple tests and had developed a chamfer on its trailing edge. Tests from the worn pad were then repeated with a brand new unworn (with no chamfer) to observe what effect this had on performance. Fig. 6 shows the differences in readings between old and new pad. The chart illustrates that in all but one of the contamination cases the new pad gives a lower reading than the old one. Only for oil is this trend reversed. It also shows that dry data is closer to levels seen under twin disk testing with the new pad with the standard devaition almost halved. This may be of significance as when the rig is used in the field it will be used predominantly on dry rail where any contamination has been previously rolled over by passing traffic. Hence, any contamination on the rails will be in the form of a very thin dry film. Further comparisons between old and new pad performance are made during later testing (details in Section 5). 4 ml of water was dispersed within the contact area during wet tests. The dry control value for this test was averaged 0.64. Wet results varied between 0.25 and 0.28 showing good consistency and giving a mean coefficient of friction of 0.25. This is in good agreement with results from field and small scale testing (see Section 7). 5.2. Oil Oil used in these tests was SAE multi-grade 10W40 Diesel typical of oils which could potentially leak onto the rail from locomotives. Due to the higher density of the oil only 3.5 ml was required to fill the contact area. The dry control reading for these tests was 0.62. The readings for oil showed a friction coefficient of 0.15. 4. Laboratory test methodology 5.3. Oil–water mixtures The Four-s (harder) pad was used for the main series of testing. The contact surfaces were both cleaned prior to each test. In the case of the rail this was cleaned with acetone to remove any oil and rubber left from previous tests. The pad was conditioned using P400 (abrasive) paper to maintain a consistent profile and remove any contamination. Test contaminants were then dispersed onto the rail as uniformly as possible. The pendulum was then allowed to swing and energy loss was recorded from the scale on the upward part of the swing. Before each test a dry control (no contaminants) To simulate rain falling on an already, oil, contaminated rail, various oil and water mixtures were tested in the following (water:oil) ratios: 50:50, 25:75 and 75:25 with 1 representing 0.01 ml, e.g. 25:75 represents 0.25 ml of water to 0.75 ml oil. The dry control friction coefficient for this test was 0.68. Results showed little variation in energy loss measured for the amount of oil in the mixture. Average results are shown in Table 3. It can be seen how oil has the overriding influence in the mixture; even when water is the predominant part of the mix, friction stays low. Table 2 Summary of laboratory tests carried out. Test no. Contaminant Type Amount (ml) Proportion (ml) No. of repeats 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Dry Water Oil Oil–water mix Oil–water mix Oil–water mix Trackside transit dry Trackside transit wet HiRail dry HiRail wet VHPF dry VHPF wet Sandite dry Leaf Leaf Leaf residue – Tap 10W40 motor oil 10W40 + Tap water 10W40 + tap water 10W40 + tap water Friction modifier Friction modifier Friction modifier Friction modifier Friction modifier Friction modifier Friction modifier Sycamore (veins down) Sycamore (veins up) Sycamore (soaking water) – 4 3.5 1 1 1 – – – – – – – – – 4 – – – 0.5:0.5 0.25:0.75 0.75:0.25 – – – – – – – – – – 6 6 3 4 4 4 6 3 6 3 6 3 3 6 6 6 S.R. Lewis et al. / Wear 271 (2011) 62–70 67 Table 4 Tests with friction modifiers. Friction modifier E.L. wet C.O.F. wet E.L. dry C.O.F. dry Dry control T.T. HiRail VHPF Sandite – – 60.8 0.65 9.3 0.09 40.0 0.41 9.7 0.091 46.2 0.49 13.2 0.12 64.2 0.73 – – 49 0.52 5.4. Leaves The leaf tested was a Sycamore, common in Britain and certain to be found trackside. The dry control test prior to the leaf contaminated tests gave an average value of 0.74. The average frictional value was 0.23 with the leaf veins on the trackside and 0.17 with the veins on the pad side of the contact. The Sycamore leaves seem to reduce friction to similar levels as water (Table 4). However, in field and small scale testing which more accurately represents the true wheel rail contact, leaves will actually reduce friction below that seen under oil contaminated conditions. 5.5. Leaf residue The water which the leaves were soaked in was also tested. This soaking water was syringed onto the rail head at 4 ml in an identical method to the pure water tested. Soaking water gave lower results than pure water With an average value of 0.13 (water was 0.25), showing that fluid components of leaves can contribute to lowering friction. 5.6. Friction modifiers Four different types of friction modifier were tested in all including three Kelasn® products (using old and new pads). The fourth FM tested was Sandite; a mixture of water-based gel and sand. The dry control value averaged out at 0.65. All three of the Kelsan products were tested both wet and dry. For practicality reasons Sandite was only tested dry. Results for each friction modifier are shown in Table 4. The results for all tests are summarised in Figs. 6 and 7. 6. Pendulum field tests 6.1. Field test method The pendulum (with Four-s slider) was taken to a section the Stockholm underground rail network for field trials (Fig. 8). The section of track chosen was treated with Keltrack© friction modifier. This was the same product as has been tested in the Sheffield university laboratory. Tests were also carried out with a Salient Fig. 8. Pendulum setup on Stockholm underground track. Systems Tribometer for comparison purposes. The track was first measured with the Tribometer at 100 m intervals. As the pendulum took a long time to set-up and time on the track was limited, sections where the largest difference in Tribometer readings were chosen to do the pendulum measurements. The largest differences in friction reading were seen at the 200 and 700 m sections on the outer rail (on the outside of the curve), with average friction readings of 0.48 and 0.36 respectively. The inner rail also displayed the largest differences at these distances and with average readings of 0.71 and 0.54. A stand was constructed for the pendulum so that it was supported above the ballast level and could clear the height of the rail. Once the area of track for measurement was chosen the stand was securely laid on the ballast and then the rig placed on top. The height of the rig was also adjusted to level the device and give the appropriate strike length. Once in position 6 consecutive swings of the rig were done to get an average friction coefficient. Due to time constraints during field testing the pad was unable to be conditioned. The rail was unable to be cleaned also as this would have interfered with the operation of the rail. The results achieved, however, suggested that neither issue caused a problem. The pendulum performed well on three out of four rail sections chosen. The section where the rig did not perform was at 200 m on the inside line. During these tests the rig would judder as the pendulum struck the rail and large variations in reading were shown. This phenomenon has also been witnessed in the laboratory, usually after tests with oil. However, during these field tests the rail was dry with no obvious residue or contamination. Also the rig behaved as normal when moved to the outer rail at the same location. What should be noted here though is that this is the same section of rail where the highest Tribometer reading of 0.71 was recorded. 6.2. Pendulum and Tribometer results Fig. 7. Pendulum test results with friction modifiers. Pendulum readings were compared with Tribometer readings for the corresponding section of rail tested and to laboratory readings on a section of inner rail which was cut from the Stockholm underground measurement section. This 40 cm long piece of rail was cut from the inner rail at approximately 700m. After the section was removed from the track it was wrapped in aluminium foil to protect from chemical interference of the third body layer during transportation. It should be noted here that this section of rail was not cleaned according to the methodology in chapter 4, as this would have contaminated the third body layer on the rail. The laboratory reading is provided in Fig. 9 just for comparison purposes. 68 S.R. Lewis et al. / Wear 271 (2011) 62–70 lubrication is through a mixture of boundary and hydrodynamic action. Conversely, in the case of the field, all moisture has been extracted from the film and friction reduction is therefore via a mechanism of a low shear solid film. This does show, however, that the pendulum is suitable for taking friction measurements in the field as well as the lab. 7. Comparisons with other methods Even though the pendulum test does not fully represent the actual wheel/rail contact it can be seen that results show similar patterns as seen in twin disk testing [10,16]. Results from twindisk testing are shown in Figs. 13 and 14 for dry and contaminated contacts. Fig. 9. Comparison of Tribometer and pendulum readings at 700 m on inner rail. Fig. 11. Comparison of Tribometer and pendulum at 200 m on the outer rail. Fig. 10. Comparison of Tribometer and pendulum readings at 700 m on the outer rail. The above charts show that with both of the field measurement techniques there is variation in readings depending on location. Average reading of the field testing was 0.45 and 0.46 for the Tribometer and pendulum respectively. This is very good correlation between the two different measurement methods. In all but one case (Fig. 9) the standard deviation is higher from the pendulum readings. However this does not seem to have much effect on the averages of each method. The ways in which both of these methods work must be taken into account though. The pendulum is a spot tester, i.e. it measures the average friction within a relatively small area (127 mm) compared to the Tribometer. The Tribometer on the other hand calculates an average friction over a larger areas; typically 3 m long (Figs. 10 and 11). Fig. 12. Comparison of Keltrack© measurement in Sheffield lab and field with average Tribometer reading for Stockholm underground. 6.3. Comparison of pendulum lab results with field Pendulum laboratory tests at Sheffield University with Keltrack© , described earlier (see Section 5.6), were also compared with the results from the Stockholm underground tests (see Fig. 12). It should be noted that the pendulum field bar in Fig. 12 shows the mean of the three tests on the underground. The results show that there is a good correlation between all three tests. While the readings from the pendulum in the field and Tribometer in the field are close (0.46 and 0.45 respectively) the laboratory reading is lower at 0.41. However this is probably due to the fact that the FM film in this case is thicker than seen at the underground and therefore partially liquid (the film in the field was thoroughly dry and very thin). Thus in the laboratory the rail Fig. 13. Creep curve from twin disk testing for a variety of contaminants [10]. S.R. Lewis et al. / Wear 271 (2011) 62–70 Fig. 14. Creep curve from twin-disk testing for oil/water mixtures [16]. When analysing the results from the above charts it needs to be noted that the pendulum test is a pure slip condition test and therefore should be directly compared to the saturation value from the twin disk curves (peak value in Figs. 13 and 14). Fig. 15 gives a direct comparison between pendulum results and saturated values from twin disk testing (from [10,16,17]). It is shown above that even though absolute data for both of the test methods is not the same, there is a correlation between the data particularly for dry, water, oil and oil/water mixes. However, results for leaves are an order of magnitude out. It was clear during testing that the contact conditions are very much different from the actual wheel/rail contact. During the pendulum testing with leaves, the pendulum would strike the leaf first and then sweep the leaf along the rail. The contact pressure was not as high as seen in twin-disk testing, and hence the leaf was not crushed. 8. Discussion The pendulum test rig was initially used for road surface friction and was later adopted as a method for assessing floor surface slipperiness. This work has been focused on re-applying this rig for the assessment of rail-head friction under dry or contaminated conditions and with friction modifiers. Initial trials were run to asses the optimum setup of the rig and how to apply various contaminants evenly. Work was also done to find the speed of the pendulum as it contacted the rail using high-speed footage. From testing with the two different types of pad it was quickly found that the harder Four-s type rubber offered more accurate results when comparing with twin disk testing. Initial trials did show quite high variance in the results however this was rectified by evenly dispersing the contaminants in the contact area. Thus results are critical on the initial set-up. It should be noted that in this case there is a rubber on steel contact with pure sliding occurring. This may account for the higher friction results seen in dry tests however, the relative ranking of Fig. 15. Direct comparison between pendulum and twin disk data (from [10,16,17]). 69 results for contaminated conditions are in line those seen in twindisk testing. It can therefore be concluded that this method has potential as a quick check for adhesion in the field. Sliding distance for the pad will significantly alter the results yielded by the rig. This was investigated during the initial trials. It was found that the manufacturers setting of 12.7 cm gave the best results with data resembling that of other test methods. It was noted during lab testing, that a chamfer would form at the trailing edge of the slider as the pad wore. Original tests were performed with a slider which had a large chamfer. Results yielded from this slider were higher than those seen from a new slider which had no chamfer. However, humidity changes cannot be factored out of these differences in readings due to slider chamfer. It is suggested that a series of tests are required to focus on the size of chamfer and correlate this to any readings from the rig. This could also be used to determine the wear rate of the slider and hence growth rate of the chamfer. Results of this testing should enable a slider life to be determined helping to define standard replacement rates for pads. Variations in readings given by the rig were investigated further. It was suspected that these variations may be due to changes in temperature and humidity. It has been shown that humidity is a contributing factor for this phenomenon (see Fig. 4). However, temperature did not very enough during testing to noticeably affect readings. Humidity and temperature variation in combination with pad wear are external variables beyond the user’s control. This means that it is difficult to isolate the contributions of each individual factor to the eventual reading. As such it is suggested that further investigation is needed to clarify this under controlled temperature and humidity conditions. The pendulum rig is designed to work with two different slider types; one hard and one softer one measured using the IRHD system [15]. Tests were carried out with both types, and results show that the harder, Four-s, Slider yields results which are closer to those seen in twin-disk testing and the field. It is therefore recommended that the Four-s slider be used for further tests. Most conventional methods of measuring railhead adhesion are time/resource consuming. Methods such as field testing require a huge amount of time to plan and organise and working with railway vehicles is cumbersome, with little control. There are also hand-pushed Tribometers which are designed to be used in the field, however, track access is still required. Although these do not entirely simulate the wheel/rail contact, they do offer a more accurate method of measuring railhead friction. There are a number of techniques used in the laboratory also. These include pin-on-disk and twin disk testing; both of which are time consuming in their own right and offer varying levels of controllability and representation of the real wheel/rail contact. The pendulum rig may not offer such a close representation of the wheel rail contact, however, results of this investigation show that it yields results similar to both field and laboratory data. The pendulum is also a quick test method, easy to use and gives a greater cost advantage over the other methods mentioned. The other unique feature of the pendulum is that it can be used both in the lab and the field [11]. When the rig is used in the field this will be mostly on a dry rail where any contaminant has been crushed and dried by traffic to form a thin dry film. This is also be the case for friction modifiers. This means that some of the solid contaminants used in these tests were not accurate representations of the field. This is because of the difficulty of replicating field conditions within the lab. This explains the differences seen between the pendulum and the twin-disk data for some of the solid contaminants. To investigate if there were any hydrodynamic lubrication effects occurring in the slider/rail contact a 2D form of Reynolds equation was used. Hydrodynamic film thicknesses for the Four-s slider were calculated for wet application cases, including; water, 70 S.R. Lewis et al. / Wear 271 (2011) 62–70 oil and the wet friction modifiers. Pad sliding speed was calculated from the high-speed video footage (see Table 1). These theoretical films were compared to the surface roughness (Ra ) of the rail. It was assumed that if the calculated film was smaller than the Ra value of the rail then there would be no hydrodynamic action in the real application. By this assumption calculations suggested that there was no film in the case of oil or water contamination; the calculated films being in the order of nanometers. However, the friction modifiers which are of greater viscosity gave an average film thickness of 39.08 ␮m for Trackside Transit and HiRail (both of similar viscosity), and 360.47 ␮m for VHPF. Calculations were also made for the TRL pad to investigate whether larger deformations of the softer material would aid in generation of a HDL film. Results show that this case gave larger film thicknesses than for the Four-s pad. However, as with the Four-s, they indicated oil and water films were still in the order of nanometers, hence no film was formed. The film thickness for the Trackside Transit and VHPF was 108 ␮m and 994 ␮m respectively. It must be noted that films generated here were for wet friction modifiers. However, friction modifiers are designed to work dry. In this case it is clear that there would be no film generated for a dry contact. This also means that there is no hydrodynamic action seen when using the pendulum and only boundary lubrication is seen. The rig was taken to a section of the Stockholm underground for validation of field use. Readings were compared with those of a Salient Systems Tribometer on the same section of track (currently the standard measurement means of railhead friction in Sweden and the UK). There was good correlation between the two methods proving the pendulum’s suitability for field measurement. It was interesting to note that results for each method were so similar even though they have differing ways of operation. That is the pendulum is a very localised test method which measures average friction within a relatively small 127 mm contact. This makes it ideal for investigating local phenomenon such as leaf fall and “wet rail syndrome”. (This is where a patch of rail has very low friction after heavy rain fall for no observable reason.) The Tribometer on the other hand takes measurements over 3 m and thus would not be suitable for investigation of such conditions. There were however some notable flaws in the design of the pendulum for use in the field. Firstly it is very heavy and can be quite difficult to carry for long distances, i.e. when walking from measurement to measurement point. This is also an issue when testing on live sections of track where the rig needs to be removed from the rail quickly. Setup of the rig in its current form is also very time consuming. Changes in height between the ballast and the rail head between measurement points meant time was taken re-adjusting the level of the rig so that a 127 mm contact could be achieved. Two design changes to the pendulum are proposed in order to counter these problems. Firstly the material of which the rig is constructed should be changed from Steel to Aluminium or a lightweight composite. The rig could also be made more compact so that it is easier to transport. To counter the issue of changes in rail height; it is proposed that a new design should incorporate a clamping device so that it can be securely mounted to the rail. It should be noted however, that the comparison between field and lab data has only been done for one case, i.e. Keltrack© Friction Modifier. For a more comprehensive study field measurements should be taken on dry uncontaminated rails, wet rails, etc. 9. Conclusions The tests described within this work have shown that the pendulum tester has potential as a method for assessing railhead adhesion. The results have also identified suitable set-up parameters which are detailed below: • The pad to be used for testing is the Four-s type rubber. • Pad strike length needs to be 12.7 cm, as specified by the manufacturer [14]. • It is suggested that an area of 12.7 cm length be marked out on the rail head in order to disperse liquid contaminants evenly. • Difference between Four-s and TRL readings – softer pad able to deflect more hence hydrodynamic lubrication. • TRL pad may be more suitable for liquid contaminants results closer to twin disk. • TRL pad shows friction coefficient of 1.2 for dry tests. This is an order of magnitude higher than values form twin disk and field testing – not suitable for dry tests. • As pad wears it conforms to the rail section and forms a chamfer at the trailing edge. This wear can effect results and more testing is needed to determine useful pad life. • Field tests show good agreement with those in the lab and also with the Salient Systems Tribometer meaning this method shows good potential as an alternative railhead friction measurement technique. • The current rigs design would need to be changed for it to be a serious competitor with the Salient Systems Tribometer. References [1] U. Olofsson, K. Sundvall, Influence of leaf, humidity and applied lubrication on friction in the wheel-rail contact pin-on-disc experiments, Proceedings of the IMechE Part F, Journal of Rail and Rapid Transit 218 (2004) 235– 242. [2] T.M. Beagley, I.J. McEwen, C. Pritchard, Wheel/rail adhesion – the influence of rail head debris, Wear 33 (1975) 141–152. [3] T.M. Beagley, I.J. McEwen, C. Pritchard, Wheel/rail adhesion – boundary lubrication by oily fluids, Wear 31 (1975) 77–88. [4] H. Chen, T. Ban, M. Ishida, Adhesion between railwheel under water lubricated contact, in: Proceedings of the 2nd World Tribology Congress, Vienna, Austria, September 3–7, 2001. [5] S. Kumar, P.K. Krishnamoorthy, D.L. Prasanna Rao, Wheel–rail wear and adhesion with and without sand, Transactions of the ASME: Journal of Engineering for Industry 108 (1986) 141–147. [6] X.S. Jin, W.H. Zhang, J. Zeng, Z.R. Zhou, Q.Y. Liu, Z.F. Wen, Adhesion experiment on a wheel rail system and its numerical analysis, Proceedings of the IMechE Part J: Journal of Engineering Tribology 218 (1) (2004) 293–303. [7] M. Broster, C. Pritchard, D.A. Smith, Wheel/rail adhesion – its relation to contamination on British railways, Wear 29 (1974) 309–321. [8] H. Harrison, T. McCanney, J. Cotter, Recent developments in coefficient of friction measurements at the wheel rail interface, Wear 253 (2002) 114– 123. [9] N. Nagase, A study of adhesion between the rails and running wheels on main lines: results of investigations by slipping adhesion test bogie, Proceedings of the IMechE Part F: Journal of Rail and Rapid Transit 203 (1989) 33– 43. [10] E.A. Gallardo-Hernandez, R. Lewis, Twin disk assessment of wheel/rail adhesion, Wear 265 (9–10) (2008) 1309–1316. [11] R. Brookes, Wheel rail interface technology management; autumn performance trials, Network Rail (2003). [12] BS EN 13036-4:2003, Road and Airfield Surface Characteristics. Test Methods. Method for Measurement of Slip/Skid Resistance of a Surface. The Pendulum Test. [13] AS/NZS 4586:2004, Slip Resistance Classification of New Pedestrian Surface Materials. [14] BS 7976-1:2002, Pendulum Testers: Part 1 – Specification. [15] BS ISO 48:2007, Rubber, Vulcanized or Thermoplastic. Determination of Hardness (Hardness between 10 and 100 IRHD). [16] R. Lewis, E.A. Gallardo-Hernandez, T. Hilton, T. Armitage, Effect of oil and water mixtures on adhesion in the wheel/rail contact, Proceedings of the IMechE Part F: Journal of Rail and Rapid Transit 223 (3) (2009) 275–283. [17] Z. Li, O. Arias-Cuevas, R. Lewis, E.A. Gallardo-Hernàndez, Rolling–sliding laboratory tests of friction modifiers in leaf contaminated wheel–rail contacts, Tribology Letters 33 (2009) 97–109.