Testing criteria for non-ballasted track and embedded track systems

Transcription

1 Testing criteria for non-ballasted track and embedded track systems ABSTRACT André Van Leuven Dynamic Engineering St Louis, MO The EC co funded research project Urban Track aims at reducing the total life cycle cost of urban rail systems by 25%. This goal will be achieved through the development of low cost track designs that take into account the maintenance aspects in order to reduce maintenance costs and increase track availability. The new track systems will have a low track modulus to minimize ground borne noise and vibrations and minimize rail wear. Acceptance of track systems depends on proving they are fit for service by amongst others successfully completing a repeated load test. However, experience has shown that under site specific conditions and especially in small radius curves, very resilient rail fasteners have failed in service despite the fact that the design passed the repeated load test in the lab according to the existing norms. In addition, there is no clear normative frame for repeated load testing of embedded track. Current practice is based on the standard for testing of direct fixation fasteners. This paper discusses the development of a method to define the repeated load test conditions for resilient rail fasteners and embedded continuously supported fastenerless rail systems and situations. For existing track systems, the proposed method is based on the use of experimental on site data; for new systems the proposed method is based on the results of multi body simulations with for track and vehicle. The repeated load tests are executed by applying an artificial force in one point of the railhead of a sample assembly of a limited length. The artificial force is applied in such a way that the same lateral and vertical rail displacements are obtained that were derived from the measurements or the simulations. The method for existing rail systems and situations starting from rail displacement measurements is illustrated in detail as it was applied to the testing of very resilient rail fasteners. INTRODUCTION Tom Vanhonacker, D2S International Heverlee, Belgium Performance requirements and test methods for direct fixation fasteners are described in following documents: European norm EN 13481: Railway applications Track Performance requirements for fastening systems (Part 1 to 7); European norm EN 13146: Railway applications Track Test methods for fastening systems (part 1 to 8). It has been found that these test methods and performance requirements do not always correspond to the needs and in service conditions experienced by very resilient direct fixation fasteners. Due to their specific design and characteristics, their increased resiliency and height (due to the presence of a more resilient base plate pad), it is necessary to verify the test conditions described in the above-mentioned documents. More specific, this is done for the load conditions during repeated load testing. It is found that the loads applied during repeated load testing should be re-evaluated for very resilient rail fasteners. For embedded tram tracks, despite their increasing popularity, there is at present neither a clear normative frame nor a common set of functional requirements under which Embedded Rail Systems may be approved and certified; no specific EN or ISO standards are available. Up to now, Embedded Fastenerless Rail Systems have been tested using test procedures derived from the specifications for discrete and continuous supported rail on slab track (EN &6), the recommendations and guidelines compiled by CROW in the Netherlands (RAGERS 2001) and experience. 1

2 VERY RESILIENT DIRECT FIXATION TRACK For conventional direct fixation track, performance requirements and test methods are described in the norms and documents cited earlier. Although the norms make a distinction in test conditions for Main Line and Light Rail, it is also noted that in order to perform a representative fatigue test, there are many other contributing factors, which should determine the exact test conditions. It is however difficult to list test conditions for each possible situation, especially for very resilient rail fasteners as they are often only used in very particular and short sections of the network. One could argue that an analytical approach can be used to determine the forces acting on the fastener. In reality, loads are distributed unevenly over the two rails and unevenly down to the rail fasteners. They are often very difficult to quantify as they not only consist of static or quasi-static loads but also of dynamic loads caused by track irregularities, discontinuities at welds and joints, switches, corrugations, vehicles defects such as wheel flats, hunting etc. In general, the forces, which act on a railway track as a result of vehicle passage, can be split up in three components: - vertical; - lateral; - longitudinal. The vertical and lateral loads generated by moving railway equipment are applied by wheel treads and flanges to the rails, which in turn must be held in place by fastenings and sleepers and ballast or slab. Lateral, vertical and torsional stiffnesses of the rail distribute the lateral loads to the fasteners and consequently to the sleepers and ballast or slab. The amplitude of the loads, which must be restrained, depends on not only the dimensions, configuration, weight, speed, and tracking characteristics of the rolling stock but also upon the geometric characteristics of the track structure. The latter includes not only the track alignment but also the local track geometry such as track irregularities and deviations from the original design. One could assume that lateral load distribution mimics the vertical load distribution to the discrete supports. However, the AREMA manual (American Railway Engineering and Maintenance of- Way Association) warns this assumption may be nonconservative since field observations suggest that such a distribution underestimates the actual load environment. The in-service situation is considerably complex due to the presence of several coupled wheel sets, various positions the vehicle can assume in a curve and the adhesion forces between the wheel and the rail. The total forces (especially lateral forces) cannot be predicted with great accuracy. Longitudinal loads are caused by accelerating and braking, temperature forces, track creep and shrinkage stresses caused by rail welding. Resiliency related issues Besides the difficulties to quantify, the exact forces acting on rail fasteners, very resilient rail fasteners are characterized by a specific design that influences the inservice loads to a greater extend than what is the case for conventional direct fixation fasteners. Most very resilient fixation fasteners have two elastic layers: a rail pad and a base plate pad. The base plate pad is responsible for a significant increase in height of the fastener. Due to this increase, more pronounced rotation moments are exercised on the fastener, causing additional displacements. The high vertical resiliency will also decrease the vertical reaction forces on the fastener whilst this is not necessarily the case for the lateral reaction forces. For example, the fastener loads can be calculated using the Zimmermann approach considering the rail as an elastically supported beam. When varying the vertical dynamic stiffness of the support points from 120 kn/mm to 5 kn/mm and considering a 24 ton axle load (120 kn wheel load), a UIC 54 rail, a fastener spacing of 600 mm, the wheels on the same bogie spaced at 2400 mm and the wheels of the nearest bogie of the adjacent car at 4800 mm and 7200 mm, than the maximum load on a single fastener decreases according to the figure below. Fastener Load Fastener load vs. fastener stiffness Fastener Stiffness Figure 1 2/7

3 It can be seen that for dynamic fastener stiffness above 70 kn/mm the wheel load of 120 kn results in a fastener load of 56 kn and higher. However, for the fastener stiffness of 20 and 30 kn/mm, the fastener load is reduced to respectively 35 and 40 kn. While the fastener loads are decreased, vertical deflections have increased from below 0.8 mm to respectively about 2 and 1.5 mm. A similar calculation is done using Finite Elements, showing the load distribution of a bogie (20 ton axle load) over the adjacent fasteners for four different vertical stiffnesses. It can be seen that the resiliency not only influences the maximum reaction forces but also the distribution of the loads over the different fasteners. The difference in load distribution and resiliency will result in different rail displacement patterns. Current testing loads and angle The current testing methods are developed for standard rail fasteners with a relatively high dynamic stiffness of around 45 kn/mm. Distinction is only made between mainline track and light rail track. For light rail track, the test conditions are fixed at a load of 50 kn that is applied under an angle of 45. From what we have previously seen, it can be concluded that: We have previously seen that the vertical fastener load decreases with decreasing fastener stiffness, but that the lateral load is not influenced. Hence, the ratio between lateral loads and vertical loads has changed dramatically. Maintaining the fixed angle approach may result in testing conditions that have a far too high vertical load and a far too low lateral load. Adapting the vertical load to better match the expected fastener load, whilst maintaining the same fixed angle may result in the application of a lateral load that is also far below the reality. FATIGUE TESTING BASED ON DISPLACEMENT CONTROL Figure 2 The current approach of fastener load estimation seems valid and safe for stiff track. However, for very resilient rail fasteners, the load distribution pattern is different. The current approach does not account for the effect of the highly increased resiliency. In addition, the approach does not to cover variations in sleeper spacing and wheel spacing. Considering the complexity of the forces acting on the fasteners and the increased displacements caused by an increased resiliency and specific fastener design, it can be concluded that fatigue testing imposing a certain test load, trying to represent the most adverse in-service load, is too safe or even not safe when it concerns very resilient direct fixation track. For very resilient rail fasteners, an approach simulating the most adverse rail displacements is more accurate, as these displacements are directly responsible for fatigue failure. The prediction and/or measurement of rail displacements are practically more feasible. Taking into consideration the existing normative reference and the above-explained discrepancy between operating conditions and test conditions that must guarantee safety, a new approach is required to define the test conditions during the repeated load testing of Very Resilient Rail Fasteners and Embedded Continuously Supported Fastenerless Rail Systems. The proposed method moves away from defining the test conditions by the fixed load and angle to be applied, and replaces these with test conditions that must result in the same displacements as those predicted from simulations or measured in the field. Thus, the repeated testing of a single fastener or a sample system with limited length (fastenerless systems) in the laboratory is done by simulating the most adverse in-service rail movements, simultaneously in vertical and lateral directions by excitation of the test system with a repeated artificial force applied in a single point on the railhead under the appropriate angle. The most adverse rail movements are determined: By measuring the rail displacements on site in the most adverse situation (high speed, small curve ) for existing systems and situations. This method was performed at STIB s metro network in Brussels and is further illustrated. Or 3/7

4 By computer simulations with multibody simulation, software where the exact site conditions can be simulated. The output of these relative fast software calculations are rail displacements. Testing is than done on a rail assembly with limited length. An artificial force is applied in a single point of the railhead in such a way that the measured or calculated rail displacements are replicated. The angle under which the force is to be applied has to be derived from initial tests on the lab assembly (in order to obtain the correct ratio between lateral and vertical rail displacement). The applied force to obtain the required rail displacements can be increased with coefficients to cope with dynamic effects and bad rail or wheel quality, as done in the existing norms. This design results in a high but acceptable rail deflection during train passage and in a low wheel-rail resonance frequency. We will discuss here the very resilient rail fasteners were installed in in 2005in the STIB metro network in Brussels, between station Bizet and La Roue. CASE STUDY A very resilient direct fixation fastener with a vertical dynamic stiffness below 10 kn/mm was designed and installed at several locations in the course of the European Funded research project CORRUGATION. As most very resilient fixation fasteners, it has two elastic layers: a rail pad and a base plate pad. The base plate pad is precompressed by specially designed springs, which pre-load the base plate with a load, which is about 80 % of the normal static load on the fastener during vehicle passage. The vertical rail deflection is controlled and limited by the pre-compression of the base plate pad. In comparison with existing fasteners, the selected material for the base plate pad has an extremely low vertical static stiffness, which marginally increases as a function of the pre-load. On the other hand, the horizontal stiffness of the material is high. The ratio of dynamic to static stiffness does not exceed 1.5. Figure 2 Very resilient rail fasteners installed between Bizet and La Roue at STIB s metro network in Brussels Rail displacement measurements on site for existing systems and situations In order to verify the in-service dynamic behavior of the installed rail fastening system, displacement measurements were performed during vehicle pass by to determine the maximal displacements of the rail relative to the sleeper. Figure 1 Detailed view on the preloaded direct fixation fastener Figure 6 Vertical and lateral displacement sensors at the outer rail 4/7

5 The passage of each wheel over the fastener is responsible for a lateral and vertical displacement of the rail/base plate. Lateral displacements of the higher rail (6Y) are generally situated between 6 mm to 8 mm. Maximum values of 8.5 mm are attained. Vertical displacements of the base plate (4Z) of the higher rail are situated between 3.5 mm and 4.5 mm. Although very resilient rail fasteners are distributing the vertical loads more effectively over several fasteners, the measurements clearly illustrate that they can give rise to increased rail displacements, not only caused by the higher resiliency of the base plate pad but also due to the increased height of the fastener itself. Displacement of rail, relative to sleeper (La Roue - Bizet) PK Z (Vertical) 2Z (Vertical) 3Z (Vertical) 4Z (Vertical) 5Y (Lateral) 6Y (Lateral) Displacement (mm) 1 Figure 8 Test setup in the laboratory 0-1 Test procedures -2-3 The angle under which the force is to be applied has to be derived from initial tests on the lab assembly, in order to obtain the correct ratio between lateral and vertical rail displacement. Once the angle is determined, the applied force can be increased to obtain the required rail displacements as measured in-service. Two types of load tests are performed on the assembly: static and dynamic tests. The static tests serve to give a first indication of the displacements obtained by applying a certain load under a certain angle. Dynamic test are later performed to produce detailed information on the actual displacements Time (seconds) Figure 7 Simulation of the displacements measured in exploitation on a single system in the laboratory The fastener assembly is mounted on a shortened wooden sleeper as used on site. The wooden sleeper is fixed horizontally and vertically and serves as a reference for the horizontal and vertical displacement measurements. The objective of this experiment is to determine an artificial equivalent test load (and its angle of application) resulting in the same vertical base plate displacements and lateral rail displacements as measured in exploitation. A loading mechanism between the rail and the actuator allows free rotation of the rail under load. Static tests The actuator is positioned at an angle α of 30, 32.5, 35, 45 and 50. At each position of the actuator, a static load of consecutively 10 kn, 20 kn and 25 kn is applied. Each load is applied successively two or three times. The displacement measurements are only registered during the the third loading cycle. Dynamic tests The actuator is positioned at an angle α of 30, 32.5, 45 and 50. At each position of the actuator and before starting the dynamic tests, the test assembly is slowly loaded to the maximum load (depending on the 5/7

6 experiment this is 18 kn, 20 kn or 25 kn), at a rate not exceeding 100 kn/min. Afterwards and at each position of the actuator, an alternating load is applied from a minimum load of 1 kn to the maximum load at a frequency of 3 Hz. Each test is executed for 60 seconds. For each experiment, maximum displacements are registered. Test results Following conclusions can be drawn: With the load under an angle of 30, vertical movement of the base plate (V1) is larger than the lateral movement of the rail (H1) for loads of 20 and 25 kn. With the load under an angle of 32.5 and more, this effect is inversed: lateral movement become larger than vertical movements. Vertical displacements at the inside (V2) are negative in most cases. Physically, this means there is an asymmetric movement of the fastener assembly. This effect is also observed for the measurements on site in exploitation, although it is not as pronounced. With the load under an angle of 50, the ratio of vertical/lateral movement equals the ratio as measured in exploitation. A dynamic load excitation up to 18 kn also corresponds with the absolute amplitudes of the displacements as measured on site. These test conditions will be used to perform final fatigue testing. Fatigue test on the very resilient rail fastener Test conditions The test conditions for the fatigue test are identical as for the simulation test. Test procedures Before the start of the dynamic fatigue tests, a preliminary static test is performed to verify the stability of the assembly. Preliminary static test The actuator is positioned at an angle α of 50 (compared to the vertical line). The static test consists of applying three loading/unloading cycles (0.5 kn/s) up to 32.5 kn. Dynamic fatigue test The actuator is positioned at an angle α of 50. A sinusoidal oscillating load between 1 kn and 18 kn is applied to the rail at a frequency of 3 Hz. The test is programmed for 4x10 6 cycles. During the test, the minimal and maximal loads are registered together with the minimal and maximum displacements corresponding with these forces. The fatigue tests were conducted with the same sleeper and fastener as for the first tests to simulate the displacements measured in exploitation. As indicated above, the holes for the anchor bolts in the wooden sleeper show some ovalization. After the loading cycles, the ovalization is measured again and compared to the ovalization before the start of the fatigue tests, however, no significant differences are measured. There is no significant evolution in the maximum and minimum displacements, there is even a tendency of stagnation after cycles. The maximum observed displacements at the end of the test correspond with the displacements measured onsite. The tests according to the updated test specifications did not result in bolt failure with bolts of an 8.8 steel quality. Rail Displacement Calculation with multi-body model software for new systems and situations With a multibody model software, future in-service rail displacements can be predicted accurately. Input values required for this calculation are related to the type of vehicle (sprung and un-sprung masses, axle spacing), rail type, vehicle speed, curve radius, superelevation, cant, spacing between the fasteners (in the case of discrete support conditions) and rail system stiffnesses. The determination of the fatigue test conditions based upon the multibody model calculation method requires also the identification of the most adverse site conditions. With the calculated rail displacements, a first lab test can be set-up simulating these displacements as was illustrated above for a very resilient direct fixation fastener. Actual fatigue testing should then be performed with the artificial force and angle acquired from this first test but should be increased with coefficients to cope with dynamic effects and bad rail or wheel quality, as done in the existing norms. Prof. Eisenmann determined this so 6/7

7 called Dynamic Amplification Factor in the seventies, based on a large number of field measurements. Local Rail Fixation Stiffness Calculation using a Finite Element Software A specific difficulty for the calculations with the multibody model is one of the input values: the determination of the local rail fixation stiffness values at the wheel/rail contact (lateral, vertical and rotational stiffness) are required input values. These lumped stiffness values can however be derived from calculations using finite element models of the relevant track systems. CONCLUSION It has been illustrated that the current normative reference with respect to fatigue testing is not applicable and may even be unsafe when it concerns Embedded Continuously Supported Fastenerless Rail Systems or very resilient rail fasteners. Very resilient discrete rail fasteners are subject to increased displacements caused by their very high resiliency and specific design. The load distribution over the adjacent fasteners changes drastically compared to conventional direct fixation fastener track. For Embedded Continuously Supported Fastenerless Rail Systems any normative reference is missing. The proposed method is based upon the use of experimental on site data, for existing systems, or upon the use of multibody model calculation results of track and vehicle. An artificial force is applied in a single point of the railhead of a sample assembly with limited length,. This is done in such a way that the rail displacements during testing match those measured or calculated. The angle under which the force is to be applied has to be derived from initial tests on the assembly in order to obtain the correct ratio between lateral and vertical rail displacement. During the simulations, the forces can be increased with coefficients representing dynamic effects, track irregularities or poor wheel quality as is currently done in the existing norms. The method for existing rail systems and situations, starting from the rail measurements is illustrated in detail in this document for a very resilient rail fastener installed in Brussels. The second method, starting from multibody model calculations is only outlined in general terms. 7/7