Development of an integrated design methodology for a new generation of high performance rail wheelset

Transcription

1 Development of an integrated design methodology for a new generation of high performance rail wheelset K. Bel Knani 1, S. Bruni 2, S. Cervello 3, G. Ferrarotti 4 Abstract An integrated design methodology, based on the use of numerical simulation tools, is established and applied for the development of innovative railway wheels and axles having outstanding reliability performances. The advanced modelling of the dynamic vehicle mission, the determination of service loads and operational stresses and strains taking into account rail-wheel contact, together with appropriate predictive criteria related to in-service damage phenomena (e.g. fatigue, wear), offer a comprehensive framework for the assessment of wheelsets durability performance at the early design stages. 1. Introduction Wheelset engineering is facing new severe specifications on wheel wear (to decrease needs for profile turning) and weight (to decrease aggressiveness on the track), without penalising reliability, running safety and total life cycle costs. At the same time, the mission is changing due to: _ the need to operate the rolling stock both on tracks with low radius curves and on high speed tracks; _ the increase of commercial speeds on conventional tracks (allowed by tilting technology as well as by increased bogie performances); _ the decrease in the quality of rail tracks and rolling stock due to maintenance reduction. Therefore, new design methodologies are needed to meet the above mentioned requirements, and in particular it is necessary to merge into the design of the wheelset the more updated numerical tools for the mathematical modelling of railway vehicle dynamic behaviour, suitable models for the damage mechanisms acting on the wheelset, and appropriate experimental test rigs for the final validation of the project. This paper illustrates some of the interim results of the European Community funded Hiperwheel project, which features the participation of major European research institutes and railway vehicle component suppliers. 1 Fiat Research Centre, Strada Torino 50, Orbassano (TO) Italy 2 Department of Mechanical Engineering Politecnico di Milano, Via La Masa Milano (MI) Italy 3 Lucchini C.R.S., Via G. Paglia 45, Lovere (BG) Italy 4 Mechanical Dynamics Italy s.r.l., Via Vigliani 25/4, Torino (TO) Italy

2 Aim of the project is to develop an integrated procedure for wheelset design, based on the combined application of refined multi-body vehicle models to predict the loads acting on the wheelset, and appropriate damage criteria to assess wheelset durability with respect to the different damage phenomena like metal fatigue, rolling contact fatigue, wear, fretting in the shrink fits. It is expected that this procedure will lead to the design of a wheelset for high speed trains with reduced vibro-acoustic impact and optimised total life cycle costs, and also to the design of an innovative, low weight hybrid wheel for mass transit vehicles, made of an aluminium hub and steel axle and rims. Moreover, an experimental validation of the results will be carried out, by means of suitable tests to validate component durability. These experimental activities will also include the full scale testing of the whole wheelset, using the roller rig owned by Lucchini C.R.S. Using this test bench, it will be possible in particular to simulate the wheelset running behaviour under realistic running conditions, thereby allowing to obtain an overall validation of the results of the project. Among the different objectives of the project, the present paper deals in particular with the definition and validation of the multi-body model of the rail vehicle, with the numerical techniques to generate representative load spectra for the wheelset, and with the experimental procedures to verify the innovative wheelset design. Some other important achievements of the project, focussing on the subject of rolling contact fatigue models, are reported in [1], while other results concerning metal fatigue, wear and fretting will be published soon. 2. An integrated approach towards the design of an innovative mounted wheelset The main feature of the project is to propose a multi-disciplinary approach to the design of the wheelset. To this end, different competencies are merged to achieve a real optimisation of wheelset characteristics and performances. The main modules composing the integrated design procedure can be resumed as follows: 2.1 Identification of mission profile Based on available measurements, the mission loads for the wheels and axle are identified, as function of rail vehicle service (running speed, cant deficiency in curve), vehicle parameters, track conditions and environmental factors. The results of these measurements are elaborated in terms of load spectra, for the different load components (especially wheel / rail contact forces), this experimental data base is then used as reference for the following steps of the project.

3 2.2 Multi-body modelling Multi-body models are defined to simulate the dynamic behaviour of the railway vehicle (see section 3 for a detailed description). Specific analyses are carried out to assess the necessity of modelling rail and wheelset flexibility. Additional analyses are performed to improve the numerical efficiency of models, so as to develop a suitable predictive tool for the subsequent development phase. Different missions are simulated, including irregular straight tracks and curve, different flexibility of track, to evaluate the effects of changed vehicle and track parameters on dynamic contact loads and, consequently, on design loads spectra. 2.3 Procedures for fatigue life prediction and material characterisation From the theoretical point of view, fatigue life prediction (FLP) criteria are developed to estimate the life of the wheelset assembly under the dynamic loads occurring during real vehicle mission. All possible damage phenomena, including metal fatigue, rolling contact fatigue (RCF) in its various forms [1] and fretting are considered. Moreover, the effect of wear and its interconnections with RCF is studied. From the experimental point of view, a thorough characterisation is performed for axle and wheel materials, under static, cyclic and random loads allowing to build a complete material data base. Moreover, specific tests for wear studies are performed on a rolling disk machine. 2.4 Integration of FLP techniques into FEM/FEA models Considering different loading conditions associated with the vehicle mission, the stress distribution induced in the wheelset structure is determined using FEM techniques. Particular attention is addressed to: _ simulate the wheel-axle shrink-fit coupling; _ determine the wheel-rail contact patch; _ evaluate the sub-superficial stresses beneath the rolling tread. Then the integrated use of FEM results, FLP (Fatigue Life Prediction) criteria and of the material database described in point 2.3 will allow to estimate the life of the wheelset assembly under the dynamic loads occurring during real vehicle mission. The influence of wear affecting both rails and wheels rolling tread will be evaluated, from the structural reliability point of view, through a prediction of the changing wheel-rail interaction as service wear proceeds.

4 2.5 Numerical techniques to predict noise emission The vibro acoustic impact of the wheelset is of course of great importance within the project. Therefore, a specific workpackage of the project deals with the development of numerical models for the prediction of noise emission. The availability of these numerical tools will make possible to compare different design solutions, in terms of wheel web shape, and also to quantify the effectiveness of specific noise absorber devices. The part of the project dealing with noise is not covered by the present paper, and the related results will be soon published. 3. Multi-body model of the rail vehicle A mathematical model of a railway vehicle running in tangent track and in curve was developed and is being used in the project to estimate the values of wheel rail contact forces under realistic operating conditions, that are the input for all durability analysis to be performed during the design of the innovative wheelset. In order to allow an efficient use of the vehicle model throughout the whole project, the choice has been made to build it using a commercial multi-body code, instead than specific home made software. To this end, the ADAMS/Rail package was chosen. The main advantage of this software is represented by the availability of refined models of wheel-rail contact forces [2] and by the possibility of including into the analysis the effect of body flexibility (e.g. for the wheelsets). The mathematical model has been set up referring to the ETR470 Pendolino rail vehicle. This kind of vehicle is very well suited for the purposes of the project, as it is used for both high speed service (e.g. in Italy between Florence and Rome) and for service on standard lines at high values of cant deficiency [3]. Data for this type of vehicle were kindly made available by Fiat Ferroviaria S.p.A. (now Alstom Ferroviaria). The complete model of the vehicle is composed by three different sub-assemblies: the carbody, the front bogie and the rear bogie. The carbody and bogies are treated as rigid bodies and defined giving its mass characteristics (mass, moments of inertia and the position of the center of gravity) and specifying the position of the bogie with respect to the carbody itself. For the wheelsets, two alternative schematizations have been defined: one as a rigid body, and the other as a flexible body. Figure 1 reports a view of the bogie model as represented in the ADAMS/rail environment. The front and rear bogies are equal except for the position of the yaw dampers, which is symmetrical with respect to the middle of the carbody. The single bogie is basically composed by the bogie frame, two wheelsets, suspensions and dampers connecting the bogie frame to the wheelsets and to the carbody. The masses of other components in the bogie, like auxiliary elements, springs and dampers, are reduced to the bogie frame.

5 Primary and secondary suspensions are represented with linear and non-linear elastic elements, while the corresponding dampers are treated as viscous elements. The connections between the bogie frame and the wheelsets are represented by the primary suspension vertical springs and by concentrated elastic bushing elements representing the axle-boxwheelset connection. The yaw damper is represented by a viscous damper in series to a spring. Multi-body model of the vehicle, detail of one bogie 3.1 Model of the active lateral suspension Based on the results of numerical simulations performed on the model and comparisons with measured contact forces available from line tests, it was observed that the active lateral suspension system has a large influence on the load transfer between the inner and outer wheels when curve negotiation is simulated. Therefore, a description of this active device was included into the multibody model of the vehicle. To this end, at each time step of the simulation the non compensated lateral acceleration on the front bogie is computed; this quantity is then low-pass filtered to depurate the effects of high frequency lateral motions of the bogie, and is fed into a proportional controller defining the reference value of pressure in the pneumatic circuit connected to the lateral actuator. The actual value of air pressure in the circuit is obtained feeding this reference value into an ARMA filter, whose values have been calibrated to reproduce the time delay due to air compressibility, and finally the value of the lateral force applied by the actuator is computed as the product of the actual pressure times the section of the actuator.

6 3.2 Model of wheelset deformability A finite element model of the wheelset was developed in order to investigate wheelset deformability in the low-medium frequency range and its influence on the dynamic behavior of the complete system, with particular reference to contact forces. For wheel modeling, brick elements were used, while the axle was schematized using Timoshenko beam elements. Concentrated masses were used to represent additional parts mounted on the axle, such as brake disks and axle boxes. The deformable wheelset model was then introduced into the complete multi-body model of the rail vehicle, using a mode superposition approach. This allows to include in the model the main effects of wheel deformability and, at the same time, to keep the complexity of the model within reasonable limits. 3.3 Model validation Once defined the mathematical model of the railway vehicle, extensive validation activities were performed in order to assess to which extent the model can be assumed representative of the actual behavior of the vehicle in tangent track and in curve, with particular reference to the vertical and lateral components of wheel-rail contact forces, as these quantities are the most important output of the model towards the assessment of wheelset durability. These validation activities allowed to verify that the model is able to reproduce with good accuracy: _ the natural frequencies of the vehicle; _ the critical speed of the vehicle as function of the actual shape of wheel and rail profiles; _ the steady-state and dynamic components of wheel-rail contact forces during curve negotiation, as function of curve radius and of the non compensated lateral acceleration Due to space limitations in this document, it is not possible to report in detail these activities. Nevertheless, a global validation of the model in terms of predicted/measured load spectra will be reported in the next section. 4. Simulation of reference mission and prediction of load spectra Besides the definition of the rail vehicle multi-body model, a procedure to compute the load spectra for the different components of wheel-rail contact forces has been set-up. The procedure is schematically resumed by Figure 2 and can be described as follows. As a first step, based on the available service measurements, a number of representative running conditions are identified for the vehicle. These are selected in order to be sufficiently representative of the different loading conditions encountered by the wheelset during standard service.

7 outline of the procedure for the numerical estimation of the contact force load spectra Table I reports the different running condition considered: as can be observed, for the vehicle running in tangent track different speeds have been considered in the range km/h, while lower speeds are not considered as they have minor relevance with respect to the definition of the load spectra. As to the curved track considered, the running conditions have been selected in order to be representative of both ordinary lines, where relatively small curve radii can be present, and high speed lines. Therefore, different curve radii in the range m have been selected and for each of them, two different running speeds have been considered, one producing a noncompensated lateral acceleration slightly greater than 1 m/s 2, and the other corresponding to a lateral acceleration above 2 m/s 2 (except for the curves with very large radius), which can be considered as a present realistic value for tilting trains [3].Two different levels of irregularity have been considered, and assumed as representative respectively of ordinary and high speed lines. The high level irregularity has been used for all simulations at speeds below 180 km/h, while the low level irregularity has been used at higher speeds. A total number of sixteen running conditions has been therefore defined as representative of the whole vehicle mission profile; for the conditions with curved track, one curve to the left and one curve to the right were simulated, including reasonable lengths of entry and exit spirals. As represented in figure 2, for each condition a simulation was performed using the mathematical model described in section 3, and a cycle count has been performed on the different force components obtained from the simulation.

8 n. track irregularity speeds [km/h] curve radius [m] non compensated lateral acc. [m/s 2 ] 1 tangent track high tangent track low 190, 220, curve high 100, , curve high 120, , curve high 140, , curve high 160, , curve low 180, , curve low 200, , 1.3 Table I: different running conditions of the vehicle considered for the prediction of load spectra In this way, sixteen separate load spectra are computed, one for each running condition, and the cumulative load spectrum is then obtained as a weighted sum of these elementary spectra. Different weights can be used in this sum, in order reproduce different kinds of mission profile, in particular, the following two conditions were considered: _ an ordinary line mission profile, where no speed higher than 190 km/h was considered, and a high percentage of sharp curves was assumed; _ a mixed track, composed partially by an ordinary line and partially by a high speed line (as is at present in Italy for the line between Rome and Milan). In this case speeds up to 250 km/h were considered, and a higher number of curves with large radius was considered. The total load spectra obtained by this analysis for the right wheel in the front bogie leading wheelset in the case of the ordinary line mission profile are reported in figures 3 and 4 respectively for the vertical and lateral force components.

9 numerical load spectra, vertical contact force numerical load spectra, lateral contact force For this mission profile, a comparison is available from line measurements carried out on the low speed line between Chiusi and Orte in Italy. The experimental values of contact forces recorded

10 during these line tests were therefore processed in order to obtain the load spectra of the vertical and lateral force components. The results of this procedure are represented in figures 5 and 6 for the same wheel for which the numerical results are shown. measured load spectra, vertical contact force measured load spectra, lateral contact force

11 Instead than comparing the absolute values of cycle counts, which are of course function of the mileage, it is more appropriate to compare the shape of the diagrams, which is fairly similar for the experimental and numerical results. In particular, the diagram for the vertical force components shows in both the numerical and experimental result a symmetric shape with respect to the value of the static load per wheel. On the contrary, the spectrum of the lateral force component is not symmetrical, as wheel flanging on the considered wheel only occurs when the vehicle is curving to the left. These features of the line tests measurements, as well as the maximum and minimum load values, are reproduced with good accuracy by the numerical results, allowing to conclude that the numerical procedure outlined in this section can be used to estimate numerically the spectra of the force components acting at wheel-rail interface. A final note in this section concerns the counting method used to define the load spectra. The results presented in this section have been obtained using the level crossing method, which is a widely adopted counting method. Nevertheless, in the case of wheel-rail contact forces, a particular loading condition occurs in which one loading cycle takes place at each wheel revolution, the peak value of the stress occurring when the considered material point on the wheel surface flows into the contact area. Therefore, the counting method to be adopted should be carefully considered. 5. Experimental methodologies to design and verify the innovative wheelset With the need to optimise wheelset geometries and develop innovative materials to reduce the nonsuspended masses and improve life cycle costs, the use of full scale test rigs has become necessary for final experimental design validation and for the product reliability guarantee. In recent years, a number of special test rigs were developed at Lucchini CRS laboratories. These test rigs provide the opportunity for rapidly and safely test innovative concepts and components; they are suitable either for standard prototype design omologation than for research purposes, generally in the field of wheel-rail contact forces measurements, rolling contact fatigue, wear mechanisms and stress analysis under realistic conditions. Experimental wheel omologation is generally obtained by performing a static test on test rig BS500 where the same loads used in the wheel FEM calculation can be applied to half wheelset and strains can be measured by strain gauges placed on the web, from here the most damaging fatigue stress cycle can be calculated. An equivalent fatigue stress cycle can be dynamically reproduced on the test rig BDR, where a rotating radial force, applied to the axle end, generates a rotating bending in the wheel web. The same type of dynamic test is used to define a Woler curve and find the fatigue stress limit for a new material in the full scale geometry conditions.

12 Axle design is simpler because it follows the classic beam calculation theory, but more experimental information about full-scale fatigue limits should be provided to improve geometry optimisation when using non conventional materials. Full-scale nominal fatigue limits can sensibly change in different parts of the axle, where different parameters should be taken into account such as: surface radius at section transitions, surface roughness, press-fit pressures on the wheel seats and journals; for this purpose, a new test rig BDA was developed to characterise innovative materials. More complex dynamics are simulated by the roller rig BU300 in which a complete wheelset together with its primary suspensions can be mounted. This testing facility has been specifically designed to reproduce as close as possible the real behaviour of the wheelset in a wide variety of operating conditions. To this end, as the result of a co-operation between Lucchini C.R.S. and Politecnico di Milano, the test rig control system has been interfaced with a mathematical model for the simulation of railway vehicle dynamics in straight track and curve, so that the reference signals to the different actuators can be directly derived from the results of the simulation of the dynamic behaviour of a specific railway vehicle, taking into account the effect of wheel and rail profiles, track layout and irregularities, train speed and several other operating conditions. As this latter test stand will be hugely employed in the Hiperwheel project, a more detailed description of the test rig itself and of its use within the project is provided below. 5.1 Description of the BU300 roller rig The BU300 roller rig, shown in figure 7, is composed by two wheels driven by a DC motor, bearing two profiled rail rings. The wheelset is mounted on the roller and is connected to a transverse beam representing the half-bogie through a primary suspension composed by helicoidal springs and viscous dampers. Moreover, two yaw dampers are placed between the wheelset and the beam, in order to control the amplitude of possible hunting motions of the wheelset.

13 the BU300 dynamic test rig for 1:1 scale tests on a mounted wheelset (courtesy of Lucchini C.R.S.) Two hydraulic actuators are placed vertically over the transverse beam, and allow to separately impose the value of the vertical force acting on each wheel, while another hydraulic actuator applies a lateral force on the beam, with a maximum of 150 kn in each direction. Moreover at each side of the test rig, a couple of electric servomotors are longitudinally placed at two different heights and connected between the transverse beam and a fixed frame. These units are used to control the transverse beam yaw movement. The three hydraulic actuators are force controlled, while the electric servomotors are operated in displacement stroke control. As mentioned above, the references for the actuators can be derived from the numerical simulation of rail vehicle running behaviour. Full details about the procedure to derive the reference signals for the control system are reported in [4]. 5.2 Use of the test rig within the project The availability and the flexibility of the roller test rig makes possible the final experimental validation of the design procedure developed within the Hiperwheel project, and to quantify the overall improvement of the innovative wheelset designs with respect to existing ones.

14 In fact, the roller rig will allow to thoroughly test the demonstrators (prototypes) obtained as the result of the design activities performed within the project. As described above, testing conditions very close to real service ones can be obtained by feeding appropriate references into the test rig control system and, at the same time, a big number of mechanical quantities can be easily measured, and some environmental parameters (temperature, state of wheel / rail surfaces) can be kept under strict control, thereby allowing a direct and quantitative comparison of alternative wheelset designs. Among the measurements which are carried out during the tests, the most important can be resumed as follows: strain measures are performed on the wheel web and on the axle, by means of a telemetry system; these data will form a base for fatigue stress analysis. Wheel rail contact forces, also measured through strain measurements in the axle and wheels, are important for both durability and safety assessment. For this latter purpose, also the measurements of wheelset dynamic behaviour, including lateral wheel-rail relative displacement and wheelset yaw are of great importance. At the same time, vibration measurements on the axle-boxes (and if needed also on the wheel web) will provide important information about the noise generated by the wheel in different testing condition. It is also possible to apply acoustic holography techniques to fully characterise the noise emitted by the wheel. Finally, the measure of wheel profiles is performed at regular time intervals, allowing to characterise the behaviour of the wheels with respect to wear rate of growth and to the possible occurrence of wheel polygonalisation. Conclusions In this paper, some of the interim results of the Hiperwheel project, presently in progress, have been outlined. In particular, a description of an integrated procedure for the design of an innovative, high performance wheelset has been provided. This procedure is based on the joint use of refined numerical tools to perform the design stage, and top level testing facilities to verify and quantify the achievements of the project. Significant improvements in the design of railway wheelsets in terms of reduced noise and vibration impact, improved durability and reduction of life cycle costs are expected as the final result of the project. Acknowledgements The work reported in this paper has been carried out within the European project HIPERWHEEL, funded by the European Commission under Contract G3RD-CT

15 References [1] Ekberg, A., Kabo, E., Andersson, H., Predicting Rolling Contact Fatigue of Railway Wheels, 13 th International Wheelset Congress, Rome, September 2001 [2] Kik W., Piotrowsky J., A Fast, Approximate Method to Calculate Normal Load at Contact between Wheel and Rail and Creep Forces During Rolling Proc. 2 nd Mini Conference on Contact Mechanics and Wear of Rail/Wheel Systems, Budapest, July, [3] Roberti R., Bruni S., Development of Operations of Tilting Train on Italian Network WCRR 2001 Congress, Köln [4] Bruni S., Cheli F., Resta F., A Model Of An Actively Controlled Roller Rig For Tests On Full Size Railway Wheelsets, Appearing in the Journal of Rail and Rapid Transit