CPL - Chalmers Publication Library
| Utbildning | Forskning | Styrkeområden | Om Chalmers | In English In English Ej inloggad.

INNOTRACK Deliverable 4.2.5 -- Improved model for the influence of vehicle conditions (wheel flats, speed, axle load) on the loading and subsequent deterioration of rails

Francis Franklin ; Jens C. O. Nielsen (Institutionen för tillämpad mekanik, Dynamik) ; Elena Kabo (Institutionen för tillämpad mekanik, Material- och beräkningsmekanik) ; Anders Ekberg (Institutionen för tillämpad mekanik, Material- och beräkningsmekanik)
Paris : UIC, 2009. - 47 (and 6 appendices, 47+15+9+22+35+53 pp) s.

The aim of this deliverable is an improved model for the influence of vehicle conditions on the loading and subsequent deterioration of rails. In particular, this report focuses on: 1. Wear data from SUROS twin-disc tests with CORUS 260 and premium grade pearlitic rail steels (CORUS 400, VA 350 and VA 400), and metallurgical analysis (micro-hardness and shear strain measurement) of section disc samples. 2. The effect of wheel hardness on rail wear rate, and vice versa: findings from the academic literature, and observations from InnoTrack SP4 SUROS twin-disc tests. 3. Development, calibration and validation of Newcastle University’s wear and crack initiation model (dynarat, a.k.a. the ‘brick’ model): a. development of material models for each of the tested rail steels; b. calibration of the wear model against the results of the CORUS 260 dry tests; and c. validation of the wear model against the results for the premium grade steels. 4. Wear modelling (using the calibrated wear model) of wheel-rail contact on top of the rail head, focussing on the effects of load and traction; develop a method for quickly estimating wear. 5. The effect of out-of-round (OOR) wheels on wear and crack initiation. Wear-Hardness Correlation Four pearlitic rail steels were tested against VAS R7 wheel steel. Three twin-disc tests were performed for each rail steel: 5000 cycles dry; 5000 cycles dry followed by 5000 cycles wet (i.e., water-lubricated); and 15000 cycles dry. In general, the harder the rail disc material becomes at the surface, the harder the wheel disc material becomes at the surface. Rail disc wear decreases when rail steel hardness increases. In wet tests, wheel disc wear rate drops as rail disc hardness increases. In the system as a whole (i.e., considering both wheel and rail discs), using harder CORUS 400 and VA 400 rail steels lowers the total wear rate. From a review of the academic literature, there is no conclusive finding that harder rails wear wheels more, or vice versa. In general, harder materials wear less, but material hardness is not the only determining factor of wear performance; microstructure and strain-hardening behaviour are critical factors, and rolling contact fatigue performance is equally important. However, as a fairly general rule: → To reduce system wear, harder steel grades should be used for both wheel and rail. Wear Model Development, Calibration and Validation The mixed (dry-wet) SUROS tests caused severe rolling contact fatigue, affecting hardness readings and wear rates, but the dry tests have provided an excellent resource for wear model development and calibration. Test specimens have been sectioned and analysed. Microhardness measurements have suggested a possible softening effect at very small strains, and a new strain-hardening equation is used to fit to the strain-hardness data for each of the four steels. Wear model calibration has led to a number of core developments to the model itself. The wear rate predictions for the premium grade rail steels match measured values (approximately) for the 5000 cycles dry tests, but over-predict the wear rate for the 15000 cycles dry tests by a factor of 2-3. (The dry-wet test predictions do not match, but are not expected to, since the ratcheting wear model does not account for major surface deterioration caused by significant surface cracking.) → Following major development, the wear model has been calibrated successfully for CORUS 260 rail steel under dry contact conditions. → The wear model has been partly validated for the premium grades. Additional test work and metallurgical analysis should lead to improved material hardening models. Effect of Vehicle Characteristics: Rail Wear Predictions The wear model, calibrated for CORUS 260 and dry contact, was used to study the effect on rail wear of vehicle characteristics through their effect on the wheel-rail contact. The patch was assumed to be elliptical and the pressure distribution to be Hertzian; in addition, the contact was assumed to be on the top the rail, suitable for straight track, not curves, and the traction to be longitudinal only. Traction coefficient has a significant effect on the wear rate. For distributed traction systems the traction coefficient may often be about 0.1, i.e., an average wear rate of about 0.75nm/cycle. For locomotives the traction coefficient may be 0.3 or even higher, i.e., an average wear rate of 1.5nm/cycle or more. → There was a very clear linear trend of wear rate against peak contact pressure (for the range of pressures studied). → Wear equations, giving wear rate for a given pressure and traction coefficient, have been extrapolated which can be used for quick estimation of rail wear. Out-of-Round Wheels The effect of pressure variation (with wavelengths above about 20mm) on rail wear rate was studied by considering each wheel pass as an independent event. Wear simulations were thus performed by varying the normal load with each passing wheel, and the predictions compared with the constant average-load case. No significant difference was observed. → Out-of-round pressure variations do not affect rail wear significantly. The effect of periodic variation of wheel-rail contact patch pressure on propagation of semi-circular cracks up to 12mm radius (i.e., penetrating to a depth of about 6mm at 30°angle to the surface) was studied using the ‘2.5D’ Green’s-function-based model. The following conclusions were reached: → For pressure variations with a wavelength less than about 2mm, the maximum pressure should be used to calculate crack growth rate. → For pressure variations with a wavelength greater than about 20mm, there is no advantage to modelling pressure variation within a single load pass, and that modelling successive wheel passes with different static pressures would be sufficient. → Out-of-round wheels with roughness features between these two wavelengths would accelerate crack propagation, but would require more detailed modelling. Rail crack growth and rail breaks For an analysis of long crack growth and rail breaks numerical simulations validated and calibrated towards full-scale field tests featuring flatted wheels have been employed. The main conclusions from the simulations were the detrimental influence of rail temperature and low ballast stiffness (where hanging sleeper(s) is an extreme case. The influences of the most important parameters have been quantified. The results are presented in Annexes 4, 5 and 6 and summarized in Section 6.2. The main conclusions and operational recommendations are: → Critical crack sizes (i.e., crack sizes for which fracture is likely) for rail head and foot cracks depend significantly on the temperature (or rather the temperature below the stress free temperature). For cold conditions, critical crack sizes of roughly 1 and 3 cm are found for rail foot and rail head cracks respectively. → Crack growth rates are significantly increased close to fracture. Consequently, operationally allowed crack sizes need to be much smaller. Exactly how much smaller depends on the accuracy of inspections (i.e., how small cracks can with certainty be detected) and inspection intervals. To guide in this decision, crack growths evaluated for different operational conditions and presented in Annex 6 can be employed. → A decrease in the assured largest crack size after an inspection (i.e., the size of a crack that with full certainty can be found at an inspection) will have a major influence on the needed length of the inspection intervals. → Low ballast stiffness will normally lead to higher rail bending moments. To avoid this influence the ballast stiffness per half sleeper should be kept above some 30 MN/m. Hanging sleepers will remove the beneficial effect of high ballast stiffness and should be avoided. In particular this seems to be the case for high-speed operations (200 km/h in the current study). → It is recommended to combine the mitigation of a hanging sleeper with an inspection for rail head and rail foot cracks. → The temperature will have a very significant effect on both crack growth rates and risk of final fracture. To this end it is recommended that the magnitude of allowed wheel–rail impact forces be related to the temperature. Further, inspection intervals need to be significantly reduced during cold periods. Guidance in defining alarm limits and inspection intervals can be obtained from the results presented in Annex 6. → Due to the significant increase in crack growth rates in cold climate, it is recommended that there is an inspection before a cold period to minimize the occurrence of larger cracks that may propagate to fracture. → The wheel–rail impact force will have an effect on the risk of fracture. To establish alarm limits is a balance between allowed forces and allowed crack sizes. Due to this, a higher alarm limit can be allowed if shorter crack sizes are assured (e.g., by more frequent inspections). → It is wise to introduce multiple alarm levels for several reasons: o A wheel that induces a high impact load is likely to cause damage on the vehicle (in the wheel, in the roller bearing, etc.). This may lead to increased costs and operational disturbances. o Also, wheels that induce impact loads below an alarm limit corresponding to rail breaks may cause smaller, arrested cracks to start growing. In particular, this is likely to be an issue for rail head cracks where a higher load may cause a crack to deviate transversally, which eventually may lead to a rail break. o The introduction of low-level alarms is likely to give the maintenance organisation improved possibilities of planning and optimising maintenance procedures. If only a one-level alarm exists there is an obvious risk that a vehicle that just passes the limit may fail in a subsequent control where operational conditions are slightly different. This will, obviously, result in unnecessary costs and operational disturbances. o For the same reason as outlined above, it is recommended that the low-level alarm limit is gradually decreased over a period before the introduction of “cold climate” alarm limits.

Denna post skapades 2010-01-14. Senast ändrad 2014-09-17.
CPL Pubid: 107380