Investigations and field results for high speed moving crossings

Ph.Pouligny1, S.Nierengarten1, J. Plu1, J.Y. Grunchec1, JM.Pauchet1; M.Sebes2, D.Boulanger3;

Ph. Mugg4, L.Winiar5, W.Schöch6

1SNCF, Paris, France; 2INRETS, Paris, France; 3CORUS RAIL, Hayange, France;4VOSSLOH COGIFER, Reichshoffen, France;

5RAILTECH, Raismes, France; 6SPENO, Genève, Suisse.

Abstract

Experience from various high speed lines operated by SNCF has shown a frequent development of head-checking cracks on moving frogs, exhibiting a similar orientation to train direction than for plain track. After some months, these superficial cracks are growing down to a threshold depth which can be detected by ultrasonic testing and which makes it necessary to organize a costly replacement of the frog. SNCF decided to experiment two different technical solutions :

- Designing new moving frog with bainitic blades to improve future moving frogs behavior, - Grinding off damaged steel in the frog point area so as to displace contact path out of

the head checking zone. This study presents the theoretical simulation and field tests realized to design and assess these points.

Introduction

Wheel running over switches, especially over the frog, is not so smooth than in plain track because of a number of specific irregularities in the wheel/rail contact tread. The most visible irregularity is caused by the existing gap between the wing rail and the frog point. For this reason, high speed lines are equipped with steel moving frogs which eliminate the existing gap of rigid frogs. Although this strongly alleviates dynamical and shocks caused by passing wheels, there still remains running irregularities which cannot be totally suppressed, because passing wheels experience changes in tread diameter over the different successive transversal profiles resulting from crossing construction. According to SNCF experience, this situation creates rapidly Head checks on moving frog’s blades and leads to a important increase of maintenance costs. At this time, we needed to develop a technical solution to improve the design of future moving frogs and a second solution to repair the moving crossings already in track, and because of fundamental safety, experiments and field tests on lines circulated by 300 km/h operated trains need important previous studies. Thus, wheels passing over moving crossings generate nevertheless dynamic forces directly on milled and tooled steel blades. Then complete dynamic simulations have to be performed to :

- evaluate shocks and contact forces levels undergone by the blade, - estimate the stability of the train passing over the grinded moving crossings - assess chosen grinding profile for the moving frogs

In the other hand, bainitic steel must be evaluated in order to be sure that beyond its well-known “head-checks proof” qualities :

- it keeps its stiffness and flexibility qualities, - it can be manufactured in SNCF or VOSSLOH factories, - its weld ability and durability in high speed line.

In 2005, INRETS realized a previous simulation based on theoretical crossings profiles before and after moving frog grinding. The results were not completely satisfying because the systematic plastic deformation of the surface (especially the wing rail) could not be taken into account as required in contact stresses calculation. Nevertheless, this first work permitted INRETS to develop a new approach numerical model of railway dynamics with varying rail profiles. The computed values of Y/Q ratio also allowed us to undertake such a high speed line test with grinded moving crossings. At the beginning of 2006, two moving crossings where chosen on Paris-Lyon high speed line. They both presented early RCF defects.

Figure 1: View of crossing # 2703 Figure 2: Chronology of simulations and grindings After an early visit before the grinding (with rail profile measurements), the two crossings have been visited every two months in the year 2006 and then every six months in 2007. Crossing # 2513 :

- Paris to Lyon HSL, run at 270 km/h from blade to tail. End of a curve; - Grinded in the middle of may 2006 with curative anti head checks profile(AHCC)

Crossing # 2703 :

- Lyon to Paris HSL, run at 300 km/h from tail to blade. Alignment; - Grinded in the middle of may 2006 with preventive anti head checks profile (AHCP).

The study has been undertaken in two steps :

- Assessment of the model with comparison between calculated and measured rolling contact zones,

- Calculation of the effect of the grinding on the dynamic behavior of the vehicle rolling over the movable frog.

Data for theoretical simulation

Beyond classical data for multi-body vehicle model (in this case TGV Duplex for an entire train simulation), this project needed a very complete description of the track at each part of the crossing and of the moving frog itself.

Figure 3: Extract of track geometry around the crossing # 2703

Cv=vertical curvature

Transversal profiles were precisely measured in each movable frog, before and after grinding according to the drawing below :

Figure 4: Measurement points for the moveable frog

Figure 5: Complete profiles for moveable frog and TVG wheels

Slop

e (‰

) C

v (k

m-1

)

Beginning of the frog’s blade

The new wheel profiles shown Figure 5 come from standard NF F 01-115 used by SNCF on TGV. For the realism of the contact stresses results, INRETS did the simulation with 28 different wear levels for TGV wheel, including new wheel profile. From these wheel and rail profiles, the contact zones table was calculated at each moveable frog measurement points. As sketched Figure 6 below, the rolling contact zones are divided with narrow strips, and elsewhere with large strips.

Figure 6: Mesh of a moveable frog with rolling contact zones.

Perspective effects removed The semi Hertzian method [1] has been implemented in the VOCOLIN multibody research software which was used for every simulation of dynamic behaviour, including the contact zones determination.

Validation of the simulation

In July 2007, fourteen months after grinding, rolling contact zones on the rail surface have been measured and compared with calculated contact zones. This comparison showed that a correct simulation needed not only to consider the first leading passing wheel, but the superposition of many. Taking into account the dynamic behaviour of the TGV train set and most of all the variations of the wheel profiles due to wear, simulation get much closer to the reality.

Figure 7: Rolling contact zones (A) – On track measurements (B) – One leading wheel; (C) – 28 different worn wheel profiles

In the following stages of the study, especially to evaluate the grinding influence on dynamic behaviour and W/R contact forces, we shall consider the average value by merging the wheel/rail contact forces calculated for 28 wear wheel profiles.

Influence of the grinding (VOCOLIN simulations)

Eight cases have been simulated : 1. Crossing #2513 with its own profiles and track geometry. Tip to heel circulation.

a) before grinding b) after AHCC grinding

2. Crossing #2513 with #2703 profiles and #2513 track geometry. Tip to heel circulation. a) before grinding b) after AHCP grinding (#2703 profiles)

3. Crossing #2703 with its own profiles and track geometry. Heel to tip circulation. a) before grinding b) after AHCP grinding

4. Crossing #2703 with #2513 profiles and #2703 track geometry. Heel to tip circulation. a) before grinding b) after AHCC grinding (#2513 profiles)

Cases #1 and #3 are representative of the field tests situation at present time. We calculated the cases #2 and #4 to clearly mark the difference between the influences of the grinding and those coming from track geometry and defects. The chosen comparison criterion is the Y/Q ratio. The Figure 8 below sketches the maximum values of Y/Q calculated on moveable frog area in cases #1 to #4.

Figure 8: Comparison between Y/Q ratio maximum values

Y/Q ratios increase in a important way after AHCC grinding. It is not significant with AHCP grinding. The case #4 increases over 0.8 but we have to consider that :

- case #4 doesn’t represent a “on track” case, - the calculated Y/Q signal is not 20Hz low-pass filtered as it should be in reel line tests

measurements This study aimed to analyse the consequences of moveable frog grinding on the TGV train set dynamic behaviour and the W/R contact forces generated by the passing wheels. The

simulations sketch a Y/Q peak value during the wheel transfer inside the moveable frog, between blade and wing rail. After the validation of the model and the results offered by VOCOLIN research software, we concluded that AHCP grinding is safe in relation to dynamic behaviour and passing wheel shocks. Although the AHCC grinding gave higher Y/Q increase, we stood confident about dynamic behaviour because of the shape of the signal. Indeed, the peak generated by the wheel transfer is to fast to have any influence on the axle stability. This dynamical point of view of the moveable frogs grinding must be confirmed by the metallurgical analysis offered by observations on track.

Influence of the grinding (Track observations)

The track observations of the crossing and moveable frogs are based on the following parameters :

- visual observations, - transversal and longitudinal profiles - surface hardness - Mauzin recordings of track geometry

Visual observations on moveable frog tip

Crossing #2513 measurement point P9 (3,4 m from the tip)

One year after grinding, wheels still don’t roll over the grinded zone

Before grinding Severe head checking

29/08/2006 3 months After AHCC grinding

7.5 MGT

Deep HC is not completely removed by AHCC

26/10/2006 6 months After AHCC grinding

11.2 MGT

Rolling contact zone is changing but doesn’t reach the remaining

HC area

03/07/2007 One year After AHCC grinding

26.8 MGT

Rolling contact zone reached the middle of the rail head and is

close to the remaining HC area

Crossing #2703 measurement point P8 (2,8 m from the tip)

After one year in track, it seems that AHCP grinding tested on the crossing #2703 didn’t move the rolling contact zone durably out of the HC area. As early as the second visit, the contact

take place on the HC cracks again. 18/07/2006

2 months After AHCC grinding 4.4 MGT

Deep HC is not completely

removed by AHCC

29/08/2006 3 months After AHCC grinding

7.5 MGT

The main HC crack keeps growing (1). Two new cracks

have appeared.

26/10/2006 6 months After AHCC grinding

11.2 MGT

Visual stability for the three cracks

Transverse profiles

No abnormal wear revealed by the frog blade transverse profile measurements.

Figure 9: Crossings #2513 and #2703 transverse profiles

Conclusion about the moveable frog grinding

This study enables the SNCF to assess the possibility of moveable frog maintenance by AHCC or AHCP grinding. INRETS has adapted its multibody research software VOCOLIN to semi-hertzian method applied with varying rail profiles. Thus it was possible to calculate and evaluate in a dynamic point of view, the grinding profiles in moveable crossings. Visual observations in track have shown that an AHCC grinding can be used on crossings with head checks. Nevertheless, this type of grinding is not applicable on new crossings because blade head geometry is excessively changed with creation of heavy load areas and stresses concentration. In this case, AHCP grindings are sufficient to avoid or delay head checks.

Bainitic moving frog field tests

For many years, SNCF and CORUS have been performing tests by installing in the same site new bainitic steel rails adjoining usual grade 900A. After more than 100 MGT in France and well performances on various sites worldwide, we knew the qualities of bainitic steel B320 against head checks. Wear and hardness behaviour are well known too.

Figure 10: HC field tests B320 vs 900A

Moreover, mechanical characteristics of the bainitic steel enable VOSSLOH COGIFER and SNCF to make two B360 bainitic moving frogs intended to field tests.

Figure 11: Vossloh bainitic moving frog setting up

After More than 2 years field tests, the B360 bainitic moving frog is about to be validate. . Behaviour in HSL track : No rolling contact fatigue. . Weldability for aluminothermic welding : Ok, but few studies are in progress to solve

little martensite problem. . Weldability for restoration of rails by electric arc welding SNCF laboratory examinations are satisfactory . Manufacturability SNCF and VOSSLOH factories tests are

satisfactory. . Compatibility with track circuits SNCF examinations during field tests are

satisfactory.

Conclusion

This project has performed few experiments in a large range of techniques from numerical dynamic simulations to hardness and rail profiles measurements in field tests. Thanks to competence of our suppliers and partner’s:

- INRETS for theoretical approach and numerical simulation, - CORUS for bainitic steel rail, - VOSSLOH COGIFER for switch and crossing manufacturing, - SPENO for grinding technology, - RAILTECH for aluminothermic welding experience,

SNCF has been able to undertake this study which will lead soon to validate two solutions against advanced HSL crossings removals due to moveable frogs RCF. By increasing the global lifetime of these essential components of RFF high speed infrastructure, we expect to reduce the global maintenance costs of existing and future HSL. SNCF is having now a group of validated grinding profiles to improve maintenance on specific HSL crossing. On a theoretical part of this project, the RCF issue in moveable frog gave the opportunity to develop the simulation of vehicle dynamic behaviour and the associated wheel/rail contact forces, in the case of varying profile as it occurs in crossings.

It is a real benefit for railways dynamic software and especially infrastructure components stresses simulation to be able to take into account, beyond usual track parameters, every scale of geometry variations :

- Alignments, curves and transitions - Track geometries defects - Varying transverse rail profiles or rail surface small defects

This opens a wide field of investigations especially forces and stresses submitted by rails or track components in presence of surface punctual irregularities or defects.

Perspectives

With a similar methodology, the next step will be to experiment bainitic switch blades on high traffic classic lines. A site near PARIS is about to be chosen and the field tests with such a switch and crossing will soon start. The new capability of VOCOLIN to match in the same railways model every scale components from an entire TGV to a small rail surface defect, has still to be developed and improved but it will be useful in future studies like rail damage analysis or derailments investigations.

Acknowledgements

This study was mainly launched few years ago by Louis Girardi who retired last year. As the project is close to the end, we would like to thank him warmly. This kind of subject, which needs large participation and competences, couldn’t be achieved without the contribution of many people in SNCF and partner enterprises. We thank all of them.

References

[1] M. Sebes, J.-B. Ayasse, H. Chollet, P. Pouligny, B. Pirat. Application of a semi Hertzian method to the simulation of vehicles in high speed switches – INRETS (2007).

[2] M. Sebes, J.-B. Ayasse. Meulage des pointes mobiles. Simulation de circulations. – INRETS (2007).

[3] J. PLU, J.-M. Balabaud. Approche théorique de la maintenance des profils de pointe mobiles – SNCF IGEV (2007). Technical report.

[4] D. Boulanger. Zones d’essai de nuance de rail, CORUS (2007).