RTR Eurailpress

MODERN TRACK TECHNOLOGY Frequencies of ballasted tracks Hard steel for curves and points Concrete support slabs RAIL VEHICLES AND COMPONENTS Performance of RU 800 S Electro-hydraulic brakes Ethernet on board November 2010 | Volume 50 Euro 20,– | 13914 www.eurailpress.de/rtr ISSN 0079-9548 4|2010 EUROPEAN RAIL TECHNOLOGY REVIEW RTR INNOTRANS AND INNOVATIONS City Tunnel Leipzig CIS for Melbourne InnoTrans review RTR’S 50th ANNIVERSARY YEAR

It’s an excellent train of thought that leads to a decision in favour of the reliable diesel engines from MAN. With performance ranging from 110 kW (150 hp) to 662 kW (900 hp), extremely economical and environmentally friendly, these really powerful engines are always on track for success. A decision that sets the right signals for the future. MAN Nutzfahrzeuge AG, Engines and Components, Vogelweiherstr. 33, 90441 Nuremberg, Germany man-engines@man.eu MAN engines. For powerful trains. Engineering the Future – since 1758. MAN Nutzfahrzeuge How much hp gets you on track? www.man-engines.com P U B L I C I S

RTR 4/2010 3 Editorial „ The final contribution dealing with “infrastructure” takes the form of a report on the new railway tunnel in the city centre of Leipzig. This is the first S-Bahn tunnel constructed in the centre of a German city using the twin-bore principle. One of its interesting details is its system of electrification, using roofmounted power rails. The second main topic in the volume deals with new components for railway wagons, passenger stock and information systems. The electro-hydraulic brake presented here is admittedly still only at its experimental and testing stage. It looks as if it is going to be able to brake more effectively, while also consuming less energy. Another report deals with the installation of Ethernet systems in passenger coaches for the provision of information for passengers. A purpose-made broadband wireless system is capable of replacing elements that up until now have proved critical, such as the couplings for sensitive electrical lines carrying messages between two vehicles. The precondition is naturally the use of the correct, robust technology, and here it helps to resort to components that have already been tried and tested in industrial environments. Information technology is represented with a contribution from Australia. Richard Hammerton presents the system architecture of the new control and information system (CIS), installed in the greater-Melbourne region for the metropolitan and regional train network. Finally, as readers would expect, this issue of RTR concludes with a review of the 2010 edition of the world’s largest specialist trade fair for the railway sector, InnoTrans in Berlin. We trust, ladies and gentlemen, that you will find many useful inputs for your own professional work as you read through the reports in this magazine. Kindest regards, We are pleased to publish this edition of European Rail Technology Review (RTR) once again with news of the latest progress from the world of railways. This time, we concentrate first and foremost on further advances in railway technology. Dissemination of knowledge in this field and the spread of information about new components and processes is the aim of this magazine, and that is what it has been doing for fifty years now. The first main topic in this issue is railway infrastructure. We begin with a scientific treatise on the vibration behaviour of concrete sleepers under the load of trains running at high speeds. After only a few years of operating its first two high-speed lines, which were opened in in 1991 (i.e. Hannover –Würzburg and Mannheim–Stuttgart), the then Deutsche Bundesbahn was forced to recognise that the ballast laid on them was showing signs of premature abrasive wear. Some of it even needed to be replaced after only five years in service, namely where was a hard material underneath it, such as bridges or the floors of tunnels. Simultaneously with the replacement of the worn-out ballast, softer rail pads were incorporated too and, on a number of particularly conspicuous bridges, matting was inserted under the ballast. Dr. Pahnke, the responsible expert at the time in the Bundesbahn’s centre office for technical matters in Munich, took charge of tackling this whole problem, along with Dr. Müller-Boruttau and Dr. Breitsamter. The outcome of their painstaking, time-consuming investigation is presented here. It was recognised that the trains (ICE trains sets, to be precise) induced vibrations in the concrete sleepers lying on the ballast at frequencies close to those of the sleepers’ natural frequency and with virtually no damping. The recommended remedy is to fix vibration-damping pads to the sleepers, since just elastic pads with no damping effect are inadequate for this purpose. A further topic covered in this issue is the properties of rails. Under this heading we report on the behaviour of rails made of harder materials than usual in curves and points. For special types of applications, rails embedded elastically in prefabricated slabs offer a valid alternative. Our final report in this block deals with experience with the RU 800 S track-reconstruction machine, which has been designed to handle several of the work processes involved in the replacement of both ballast and rails. It is able to complete its work with very much shorter periods of track closure, and the newly laid track is of a better quality. Dr.-Ing. Eberhard Jänsch Editor-in-chief Dear Readers, (Eberhard Jänsch)

RTR 4/2010 4 Contents RTR 4/2010  3 Eberhard Jänsch Dear Readers  6 Ulf Pahnke Frank H. Müller-Boruttau Norbert Breitsamter Frequencies of the ballasted track  16 Bernhard Knoll Hard rails in tight curves  20 Albert Gelz Hard steel for every type of point  22 Heinrich Gall Concrete support slabs for tracks in a Deutsche Bahn depot  24 Fred Beilhack RU 800 S – performance comparison after three years  27 Dirk Stecher Michael Menschner The Leipzig City Tunnel  31 Michael Kühnlein Julian Ewald Matthias Liermann Hubertus Murrenhoff Self-energising electro-hydraulic brake (SEHB) 20 24 Include RTR in your mediaplan! Mark this date in your diary! Next issue is RTR No. 1/2011 with extra print run to Rail-Tech and Sifer! MODERN TRACK TECHNOLOGY Frequencies of ballasted tracks Hard steel for curves and points Concrete support slabs RAIL VEHICLES AND COMPONENTS Performance of RU 800 S Electro-hydraulic brakes Ethernet on board November 2010 | Volume 50 Euro 20,– | 13914 www.eurailpress.de/rtr ISSN 0079-9548 4|2010 EUROPEAN RAIL TECHNOLOGY REVIEW RTR INNOTRANS AND INNOVATIONS City Tunnel Leipzig CIS for Melbourne InnoTrans review RTR’S 50th ANNIVERSARY YEAR Ad-deadline is on 16th February 2011. Book your advert right now! Email: Silvia.Sander@dvvmedia.com  35 Olaf Schilperoort Fast, wireless communication along the whole length of trains  38 Richard Hammerton Upgrading Information and Communication Technology systems in Victoria (Australia)  43 Eberhard Jänsch Christoph Müller InnoTrans – Even more railway than ever before! Briefly from Around the World: Unimog U 400 shunting vehicle for DB Regio p. 48 Solution for the application of mobile railhead lubrication p. 48 Needle bearings for the Shinkansen p. 48 MFS material transport and silo wagons – an optimum solution for the transport of material p. 49 Largest-ever Freudenberg seal helps drive the world’s longest tunnel p. 49

27 43 Front cover: Complete mobility. Ensuring mobility is the number-one challenge in our society. We need networked traffic and information systems to remain mobile in future – for safe, costeffective and environmentally friendly passenger and cargo traffic. That is why, with “Complete mobility”, Siemens creates integrated efficient transport and logistics solutions, from infrastructure equipment for rail and road traffic, rail vehicles through to airport logistics and postal automation. (Photo: Siemens) MODERN TRACK TECHNOLOGY Frequencies of ballasted tracks Hard steel for curves and points Concrete support slabs RAIL VEHICLES AND COMPONENTS Performance of RU 800 S Electro-hydraulic brakes Ethernet on board November 2010 | Volume 50 Euro 20,– | 13914 www.eurailpress.de/rtr ISSN 0079-9548 4|2010 EUROPEAN RAIL TECHNOLOGY REVIEW RTR INNOTRANS AND INNOVATIONS City Tunnel Leipzig CIS for Melbourne InnoTrans review RTR’S 50th ANNIVERSARY YEAR International Trade Fair 22 – 24 June 2011 Berlin Exhibition Grounds www.publictransport-interiors.com Messe Berlin GmbH · Messedamm 22 · 14055 Berlin · Germany Tel. +49(0)30 / 3038-2212 · Fax +49(0)30 / 3038-2190 www.publictransport-interiors.de · pti@messe-berlin.de

RTR 4/2010 6 1 Measured frequency spectra In order to answer the question, what kind of dynamic movements occur in the ballasted tracks, numerous vibration velocity measurements under traffic load were taken two years after the Hanover-Würzburg high speed line was put into operation in 1991. Frequency spectra of rail and sleeper were recorded in vertical direction during the passage of six ICE 1 trains at about 250 km/h under the coaches. The 6 measured spectra as well as the energetic mean value can be seen in Fig. 1. The measured graphs show several peak values and much lower values in the frequency range between 0 and 1000 Hz. In the peaks most energy is transmitted and the greatest wear is initiThe wear is mainly caused by movement of the sleepers in the ballast bed. Each ballast stone has initially a sharp-edged shape with rough surface. With time the sleeper movement leads to a smooth and rounded stone, which no longer can stabilise the track system. How to explain the strong wear of the track? This question is answered by the analysis of measurement results on a ballasted high speed track in comparison with calculated results. The measurement results are given by vibration velocity spectra of rail and sleeper during the passage of a high speed train. In the following, we take an ICE 1 train on the Hanover –Würzburg 250 km/h high speed line in Germany as an example. This line is fitted with ballasted tracks with prestressed concrete monobloc sleepers (B 70, 2.60 m long, mass 300 kg). The calculated results are developed from two mathematical models, one for the unloaded track und a spatial one with a wheel set on the two rails of the ballasted track. By applying the dynamic system analysis, a simplified but accurate mathematical model has been developed, which describes the most important dynamic characteristics of ballasted tracks. Thereby it is possible to design permanent ways in future, which are dynamically less sensitive for excitations by traffic. And so it is possible at last to optimise the dynamic properties of ballasted tracks, considering the design speed of the track. Frequencies of the ballasted track Realizing the fact that the ballasted track in high speed lines has half the life time of standard tracks we should ask the questions if on the one hand everything is done correctly and on the other hand which dynamic load is transmitted to the track. Scientific researcher, Former member of BZA DB AG, Munich ulf-pahnke@starnberg-mail.de Dr.-Ing. Ulf Pahnke Expert imb-dynamik, Inning imb-dynamik@t-online.de Dr.-Ing. Norbert Breitsamter Managing director, licensed expert for construction dynamics imb-dynamik, Inning imb-dynamik@t-online.de Dr.-Ing. Frank H. Müller-Boruttau Fig. 1: 1/3rd octave ver tical vibration velocity spectra of rail and sleeper on ground during 6 passages of ICE 1 passenger wagons with 250 km/h and the energetic mean value (black). Velocity in mm/s eff., SLOW.

RTR 4/2010 7 Frequencies of the ballasted track n To produce the spatial mathematical model, we take two parallel rails of equal length, in the middle of which the wheel set is placed and rigidly connected with its wheel discs. For each of these two connections there is one equation of vertical displacement only for the wheel disc and the contact point of the rail. We are interested in the eigenvalues of the vertical vibration (natural frequencies) of the wheel set on the ballasted track as well as in the eigenfunction (mode) of the spatial structure. Therefore we build up the following mechanical system including its masses. The wheel set is connected to the vehicle body by the bogie and unsprung with the rails. The eigenfrequency of the vehicle body is tuned to about 1 Hz and that of the wheel set on the ballasted track to about 50 Hz and more. So the wheel set has a much softer connection to the vehicle body than to the permanent way. In order to simplify the vibration analysis of the wheel set we will neglect the connection to the vehicle body, allowing an error of about 1 Hz in the result. For the desired analysis of the wheel set vibrations an error of this magnitude is negligible, since the wear of the wheels causes much bigger deviations. In the mathematical model the axle of the wheelset and the rails are continuous beams with given bending stiffness but ne2 Mathematical models of the frequency analysis The mathematical model of the ballasted track is described by a continuous rail on discrete supports. The supports consist of the sleepers with elastic rail pads on top. The sleepers are supported by single elastic springs on top of an inflexible bottom plate (Fig. 3). The stiffness of these springs has to be determined from the stiffness of the ground. It must be taken into account that the 30 cm thick ballast bed distributes the sleeper’s load at an angle of 75°. Loads may be placed anywhere on the rail and the reaction in the rail can be calculated at arbitrary places. Since the mathematical model uses slab elements, the results are exact and no approximate solutions. The accuracy of the solutions can be proved by the equations of chapter 6. In order to simplify the plane model (in longitudinal direction) is calculated with half the ballasted track with one rail only (Fig. 4). So we use the wheel load only instead of the wheel set load. The ends are free to rotate and to deflect. The plane mathematical model may have arbitrary length. Here we use 15 to 20 supports, spaced at 65 cm each (9.75 to 13 m length). A change to the 60 cm spacing of the real track would raise the eigenfrequencies of the loaded track (chapter 4) by less than 2.3% only. For the unloaded track (chapter 5) they would be changed by less than 1‰ only. ated. Therefore it is important to find out the cause of each peak individually. The phenomena in the lower frequencies between 0 and 40 Hz – the so called quasistatic area – with the unavoidable, periodically repeated single loads are well known. p The first peak at 8 Hz is caused by wheelsets located about 9 m apart of each other. This are the wheelsets between the bogie pivots of each of two coupled coaches. p The second peak at 20 Hz is caused by wheelsets being about 4 m apart. This are the wheelsets beyond the bogie pivots of each of two coupled coaches, p whereas the distance of 2,50 m between the axles of each bogie is of minor importance. Above the 40 Hz we find true vibrations in the track, which partially change their amplitude with velocity. It can be seen later, that they are caused by resonances initiated by dynamic excitations of eigenfrequencies (natural frequencies) of the vehicle as well as ballasted track components. At 250 km/h dynamic excitation frequencies between 0 and 130 Hz arise from the out-of-roundness of the wheels. The 1st wheel harmonic initiates about 25 Hz, the 2nd 49 Hz, the 3rd 74 Hz, the 4th 98 Hz and the 5th 123 Hz (Fig. 2). When the train’s speed exceeds 200 km/h the out-of-roundness of the 3rd to 5th wheel harmonic are decisive. But all attempts to explain the high velocities between 400 and 1000 Hz in Fig. 1 by excitations of out of round wheels with frequencies between 50 and 123 Hz failed. A pure reflecting upon the out-of-roundness cannot explain the relation between speed and excitations that we have introduced up to now. Therefore the following calculations are applied to a wheel-set-track-system. A similar system was used in G.B. Morys’s dissertation which succeeded in simulating the formation of out-of-roundness of the wheels by calculation. Fig. 3: Longitudinal mathematical model of the ballasted track Fig. 4: Cross section of the mathematical model of the ballasted track Fig. 2: Frequencies of excitation at 250 km/h

RTR 4/2010 8 n Frequencies of the ballasted track glecting the shear stiffness. The continuous mass is replaced by discrete masses. On the rail (UIC 60) 3 mass points are placed at equal distances between the sleepers. The sleepers get one mass for each support, and the mass of the wheel set is represented by 6 masses. The four brake discs on the ICE 1 axle are represented by 4 masses. The wheel discs at the ends of the axle are concentrated in 2 masses. Furthermore the moments of inertia of the 2 wheel discs about axles parallel to the rail are to be considered because of their mayor influence on the bending of the axle. So for the whole wheelset only 8 equations are needed, much less than in other papers. The wheel discs are supposed to be rigid and inflexible. Because of the flat conical shape of the wheel discs their large shape stiffness may be assumed. For the n mass points of our mathematical model we determine n displacement equations with n components in an (n∙n) matrix. For this matrix the characteristic polynomial of nth degree can be calculated, the zeros of which define the n eigenvalues λ1, λ2, λ3 … λn. The n eigenvalues λ are calculated from the matrix by known subroutines as roots of the characteristic polynomial. From the eigenvalues λ we get the eigenfrequencies in Hz by taking the reciprocal square root and divide by 2π. The n independent eigenfunctions (modes) belonging to the n eigenvalues are calculated by inserting the n eigenvalues into the original equation system and setting it equal to zero. No commercially available program system was applied. 3 Choice of the calculation parameters The fundamental postulate of this paper is to develop a mathematical model the results of which should not be contradictionary to the measured results. Therefore as few assumptions as possible had to be made. For example no multilayer ground was presumed, a possible damping was neglected, the influence of Hertzian stress between wheel and rail was neglected and the simpler Euler beam was chosen instead of the Timoshenko beam like in other papers. The reason for such simplifications is that we cannot guarantee the exactness and homogeneity of the ground parameters. Particularly for those a locally unavoidable error of 20% cannot be excluded, while the total error of the excluded assumptions will hardly be greater. Also the variation of vehicle properties and track due to wear is not understood partially. We are aware of the fact, that the interaction of the ballast layer with the other components has not yet been defined in this dynamic investigation. From the static point of view its distributing property is well known. Therefore this is considered in the calculation of the ground spring. Initially the dynamic interaction of the ballast mass was not defined. This had to be postponed to later stage at which a comparison between various calculated results and measured values were possible. For the wheel set a half worn wheel set was assumed. The parameters chosen as best fitting after multiple repeated calculations and comparisons with the measured results are shown in Tab. 1. The real ground stiffness proved to be about 20% higher than the minimum value required at the time of construction. For the three rail fastenings before and behind the load the parameters of a loaded rail pad (RP) were taken. 4 Calculated results of the spatial wheel-set-ballastedtrack-system In order to assess the calculated spatial results it was decisive to have one diagram for each frequency for to get the full survey at a glance. Additional the most important parameters and calculation results had to be readable numerically. As examples the displacement modes of the first two eigenvalues have been drawn in Figs. 5 and 6. The longitudinal curve of the mode is placed in the left part of the picture and the transverse mode in the right. In the right part we find the system above, below it is the displacement mode of the wheel set in a different scale and below we see the bisecting lines of the 3 sleeper modes before and behind the wheel set. The left side of wheel set and track in the transverse mode is shown in black and the right one in red. The longitudinal modes are seen from the right, so they are red for symmetric modes. On the left side important system values are printed. If the wheel set is not put in the middle between the sleepers but above the sleeper the eigenfrequency is raised by 2.6%. As for other modes a similar frequency deviation can be found, only this position is shown here. In Fig. 7 all eight modes of the wheel-setballasted-track-system are shown, in which elements of the wheel set have the greatest displacement. There are 4 symmetric and 4 anti symmetric modes. Their eigenfrequencies are situated between 47.5 and 754.3 Hz. If we are looking for modes in which there are strong interactions between wheel set Parameter Calculated value Bedding modulus of the ground 120 MN/m3 Stiffness of the unloaded RP 60 N/mm Stiffness of the loaded RP 160 N/mm Tab. 1: Calculation parameters Fig. 6: Spatial modes of wheel set and ballasted tracks at the second eigenfrequency of 58.7 Hz. Track section 975 cm long. Fig. 5: Spatial modes of wheel set and ballasted track at the smallest eigenfrequency of 47.5 Hz. Track section 975 cm long.

RTR 4/2010 9 Frequencies of the ballasted track n in the front wave of the rail we look at this unloaded area of the ballasted track in the following chapter. 5 Calculated results of the unloaded track-system Figs. 8 and 9 are showing the long-wavy eigenfunctions (modes) of the unloaded ballasted track with the week rail pads (RP). The dense sequence of the two eigenvalues – at 81.5 and 81.7 Hz – points out that we have an accumulation area. Rails and of rail but interacting with the brake discs. Between 160 and 210 Hz there are large sleeper displacements in the direct surrounding of the wheel set, where the stiffer rail pads are placed. Comparing these calculated results with the measured ones of Fig. 1 we find large amplitudes of the sleepers between 50 and 125 Hz as well. The excitations of Fig. 2 are acting at these frequencies as well, such that resonance results. But the considerable lowering above the frequency of 125 Hz can not yet be explained. As we find high frequency movement of the sleepers and track elements with large displacements, the mode in 7C with roughly 115 Hz is striking. There the sleepers are interacting with the brake discs even beyond the direct influence of the wheel set where the soft rail pads of the unloaded track are located. The maximum displacement of the sleepers in the unloaded area is about 60% of the brake disc displacement. Comprehensive research shows that wheel set, rails and sleepers are swinging together in same direction at frequencies between 50 and 115 Hz. Above 115 Hz the sleepers are moving increasingly independent Fig. 7: 8 spatial modes of the ICE 1 wheel set on the ballasted track, in which elements of the wheel set have the greatest displacement. Track section 975 cm long. Schwellenabstände = sleeper distances [cm]

RTR 4/2010 10 n Frequencies of the ballasted track Fig. 9: 2nd plane mode of the unloaded track, belonging to the second eigenfrequency of 81.7 Hz. Track section 1300 cm long. Fig. 8: 1st plane mode of the unloaded track, belonging to the smallest eigenfrequency of 81.5 Hz. Track section 1300 cm long. Fig. 10-1: Essential modes of the unloaded, ballasted track-system. Track section 1300 cm long. Stützpunktabstände = Suppor t distances [cm]

RTR 4/2010 11 Frequencies of the ballasted track n of the sleepers. As this mass seams to be small, the moved ballast mass (in the equations) seams to be hidden somewhere behind the relation between ground spring and sleeper mass. For frequency calculations a small deviation of this relation is of little influence, as it is standing under the square root. Consequently the mass of the ballast was not considered for the loaded and the unloaded track up to now. The smallest eigenfrequency of the ballasted track of about 80 Hz is mainly determined by the bedding stiffness of the ground. In order to have a better survey over the characteristic of the modes between 80 can be obtained by including the full ballast mass. This is true for all calculated eigenfrequencies with more than 80 Hz and of less than 135 Hz in chapter 4. A homogeneous distribution of the sleeper load by the ballast is possible only if the whole volume in the frustum of a pyramid below the sleeper is filled with ballast. In practice the lower side of the sleepers show that they have been supported by a few ballast stones only, where the discrete formation of small but deep craters can be observed. So just the small moved ballast mass in the few pyramids below the big craters should be considered dynamically together with mass sleepers are vibrating in the same sense and with almost the same amplitude. The length of the model is that of the unloaded track between the two bogies of an ICE 1 passenger vehicle. The ends of the rails are free to rotate and deflect, though there is no displacement at the ends of the rail section. On the right side of the Fig. 8 and 9 the important construction and calculation parameters are printed. The results of chapter 4 were obtained without considering the ballast mass. That is caused by investigations in this chapter. Calculating with the parameters of table 1 no coincidence with the measured results Fig. 10-2: Essential modes of the unloaded, ballasted track-system. Track section 1300 cm long. Stützpunktabstände = Suppor t distances [cm]

RTR 4/2010 12 n Frequencies of the ballasted track 80 kN/mm, we would not get these horizontal stages and a plateau, but oblique lines. With a value of 110 kN/mm we find horizontal stages and a plateau again, but they are displaced to different frequencies from those of Fig. 1 then. So the conformity of the shoulders in measured and calculated diagrams verify that in this case a stiffness of 60 kN/mm for the unloaded track (as listed in Tab. 1) is valid. At last we have found the explanation for the fact that in Fig. 1 the greatest amplitudes of the rail velocity – in the plateau between 550 and 850 Hz – are created by resonance on the excitations of short-pitch corrugation with about 1160 Hz. The length of the undulation is about 6 cm. This phenomenon is a good example of the fact that the excitation spectrum is widened by the impulse kind of corrugation. Remarkable in Fig. 11 is that there are 3 areas of accumulation at 81, 132 and 222 Hz. A fourth one comes together at 7596 Hz in Fig. 10P. This is the eigenfrequency of a rail beam with the length of a third of the sleeper distance, laying on its ends, whose mass is concentrated in the middle. With continuous rail mass the 4th accumulation point (AP) is shifted to infinity. Between 0 and 81 Hz as well as between 132 and 222 Hz we find two “MODE FREE” areas. The existence of two accumulation points (AP) in the eigenvalues of simply connected plane elastic areas is known from [1]. There Heise has shown the existence of 2 AP for the example of plane circular discs by 16 Integral Equations of geometric and static boundary value problems. Corresponding for tracks with the composite of the two elements rail and sleeper 4 AP have to exist. The AP at 81 and 132 Hz may be attached to the sleepers, while the AP at 222 Hz belongs to the rail as well as the AP in infinity. Considering the extreme amplitudes in the frequency spectrum of Fig. 1, we find very small values in the area between 125 and 220 Hz. This effect can be explained now by the fact that no resonance can happen in a mode free area. In Fig. 12 it is remarkable that the excitation frequencies have their direct response in the bands of eigenfrequencies of the systems of wheel-set-track and unloaded track. The self excitation of the wheel sets by the 2nd wheel harmonic at about 50 Hz is unfortunately close to the smallest eigenfrequency of the wheel sets on the track of about 48 Hz. Ballast is best consolidated by frequencies between 35 and 50 Hz. This is shown experimentally in [2]. The excitation by the 3rd wheel harmonic at 74 Hz is close to the smallest eigenfrequency of the unloaded ballasted track at Under “Rail” and “Sle” the biggest amplitudes of the rail and the sleepers are put together. The composition of these values in the range between 0 and 1000 Hz is shown in Fig. 11. In these graphs of the mode attributes we recognize in which part of the system the biggest dynamic displacements and the strongest interactions between rail and sleepers are happening at which frequency. With the graphs for rail and sleeper in Fig. 11 only we cannot yet find a connection to the measured graphs of Fig. 1. After several idle attempts with derived values a successful possibility was opened by using the biggest difference of displacements between rail and sleeper in the rail fastenings (named “DRS” in Figs. 8 and 9). In Fig. 11 these differences “DRS” are shown for all frequencies. In the graph of DRS for values higher than 222 Hz we find a parallel effect to Fig. 1 in the three “stages” or “shoulders” where the 3rd stage leads to a plateau between 650 and 850 Hz. If we would not choose a rail pad stiffness of 60 but and 800 Hz, 16 of the essential modes are presented in Fig. 10. Within the range of 80 to 100 Hz rails and sleepers are vibrating together in the same direction and with almost the same amplitude. Between 100 and 132 Hz the sleepers develop their own behaviour more and more, until in the mode 10K at 132.2 Hz nearly the sleepers only are vibrating. With these frequencies the rail is not yet vibrating as a beam, it is just bent by the opposed vibration of neighbouring sleepers (analogy to the pinned-pinned mode at about 1200 Hz). In the modes A to F the rail is quasi shifted, from G to K it is bent by the sleepers. Discontinuously the behaviour of the rail changes in the next mode L of Fig. 10 at 222 Hz. It vibrates in oppositely oriented phase to the sleepers and with increasing amplitudes compared with the sleepers. A full view over the properties of the modes in the range of frequencies seams to be desirable. Therefore in the lower left part of Figs. 8 and 9 mode attributes are shown. Fig. 12: Excitation and eigenfrequencies of the wheel-set-track-system at 250 km/h Fig. 11: Mode attributes of the unloaded track-system

RTR 4/2010 13 Frequencies of the ballasted track n values in the range between 0 to 1000 Hz. Now we are happy to see that in Fig. 12 a possibility is hidden to simplify this operation, as it is very helpful for the design of ballasted tracks if we know the mode free limits. So the coincidence of excitation and eigenfrequencies may be avoided. At the upper limit of the 1st mode free area (Nr. 1 in Fig. 12) there is the mode A in Fig. 10. At the lower limit of the 2nd mode free area (Nr. 2 in Fig. 12) there is the mode K in Fig. 10. And at the upper limit of the 2nd mode free area (Nr. 3 in Fig. 12) there is the mode L in Fig. 10. Answering the question that we had in the beginning, if everything is done correctly, we have to say: The ballasted track that we have on high speed tracks at present guaranties strong wear but does not prevent it. 6 Simplified calculation of the mode free limits In order to find the 3 limits of the mode free areas we had to calculate all the eigenabout 80 Hz. This frequency is loosening the density of ballast [2]. The strong 3rd wheel harmonic of out of round wheels is uncontrolled occurring in service also. Both effects together are probably the cause of the frequent, unintentional irregular changes of the track position. The 98 Hz of the 4th wheel harmonic are close to the 98.7 Hz where the greatest amplitudes arise in the front wave of the unloaded track in Fig. 10E. Similarly unwanted are the 123 Hz of the 5th wheel harmonic, which are close to the 125.4 Hz of the Fig. 10H where the sleepers have the greatest amplitudes. There they also have a strong interaction with the brake discs of the wheel set. The practical experience with permanent ways shows that already existing irregularities of the track position are contributing to more intensive deterioration of the track position. This experience is confirmed by the results here because of the resonances. But it could be worse, if we did not have the cost free transfer of energy by the half space damping at small frequencies, as the damping of the rail fastenings is poor. In Fig. 12 in combination with Fig. 1 we can see, that unfortunately for a velocity of 250 km/h the eigenfrequencies of the unloaded ballasted track with concrete sleepers are placed in the limits of the excitation frequencies. Fig. 13: Mechanical models for the calculation of the mode free limits.

RTR 4/2010 14 n Frequencies of the ballasted track How close the results of this simplified calculation are to those of a track with 20 sleepers (less than 1‰) is shown in Tab. 2. In Fig. 14 we see the simplified calculated limits of a track with wooden sleepers and their low weight by the equations 1 to 5. In this calculation the weaker rail pad of the ZW 900 was replaced by the stiffer one of the ZW 687a. Additional the ground stiffness was weakened, as the springy bottom of the wooden sleeper with its supposed stiffness of 80 kN/mm had to be put in series with the ground stiffness. All the other parameters were kept. From the result in Fig. 14 we can see, that the lower weight of the wooden sleeper compared with the heavy weight of the B 70 sleeper has positive consequences in the range between 50 and 130 Hz, in spite of the fact that an additional spring under the sleeper had to be introduced. The limit 1 is about 12 Hz higher compared with Fig. 12. By this the distance to the frequency of the 3rd wheel harmonic is 18 Hz instead of 6 Hz with the B 70. So the amplitudes with the wooden sleeper caused by the 3rd wheel harmonic should be smaller than half the value of the concrete sleeper B 70. Besides this in Fig. 14 the limits 2 and 3 are placed much higher than for the B 70. That means, that the corrugation excitation in the lower frequencies is reduced by the 2nd mode free area between 400 and 700 Hz. Additional to the half space damping of the ground we get more damping by the wooden sleeper. This investigation allows some conclusions concerning elastically soled sleepers [3,4]. By the additional spring at the bottom of the B 70 the limit 1 is shifted to lower frequencies. By this the 1st eigenfrequencies of the wheel set (at about 50 and 60 Hz) could find resonance with the 1st eigenfrequency of the unloaded track, as the upper limit of the 1st mode free area is lowered. More promising it might be to choose damping soles [5] instead of elastic soles, if a sufficiently durable, non elastic material with pure damping could be found. In this case the limit 1 could not be shifted below the 80 Hz. References: [1] Heise U.: The spectra of some integral operators for plane elastostatical boundary value problems, Journal of Elasticity (8) 1978, H. 1, p. 47-79 [2] Fischer J.: Einfluss von Frequenz und Amplitude auf die Stabilisierung von Oberbauschotter, Dissertation Graz 1983 [3] Müller-Boruttau F.H., Kleinert U.: Besohlte Schwellen, ETR 50 (2001), H. 3, p. 90-98, published under D:\ Transferdateien\HWS\Büro\01-01-26ETR-Bericht. DOC, 23.01.01 14:12 (Vikipedia) [4] Trevin J.-M.: Gleis, Deutsche Patentanmeldung DE 691 13 884 T2 [5] Cronau H.: Querschwelle für Eisenbahngleise, Deutsche Patentanmeldung DE 297 21 118.8 including the mass of the rail we end up in differences of large numbers and consequently in large errors. But reliable values are available with the model B by neglecting the mass of the rail, but including the stiffness of the rail in the middle. So the model B contains one mass only, the mass of half the sleeper. In order to calculate the stiffness ctot we put the stiffness of the rail section in its middle in series with the stiffness of the rail pad. With “a” being the distance between two sleepers and with the bending stiffness of the rail we get c EI a 3 3 48 = (2) The total stiffness ctot of all springs acting at the sleeper is found as c c c c tot = + + 1 2 3 1 1 1 /( / / ) (3) With half the sleeper mass m1 we get the rotational frequency ω2 in eq. (4). ω2 1 = c m tot (4) By this the lower limit 2 of the second mode free area is known now. The conversion of the rotational frequencies ω1,2,3 into the frequencies f1,2,3 in Hz may be done by eq. (5). f1 2 3 1 2 3 2 , , , , / = ω π (5) With eq. 1 to 5 the exact calculation of the accumulation points (AP) of the infinite rail is possible now (Fig. 14), as the models in Fig. 13 belong to the infinite rail. Common in the three modes A, K and L is that with the mass of one sleeper and a rail section with the length of the sleeper distance a vibrator with two masses on a rigid base can be designed which has the same eigenfrequencies. In the models A and B of Fig. 13 there are the following parameters: c1 = stiffness of the ground spring c2 = s tiffness of the rail pad in the rail fastening m1 = mass of half the sleeper m2 = m ass of the rail between two sleepers, distance a With these symbols in the following eq. (1) we get the squared rotational frequency ω1 at the limit 1 (mode A) with the negative sign in front of the root and the squared rotational frequency ω3 at the limit 3 (mode L) with the positive sign before the root. Equation 1 is valid for the model A in Fig 13. ω2 1 3 1 2 2 2 1 1 2 1 2 2 2 1 1 2 2 2 , ( ) ( ) = + + ± + + c c m c m m m c c m c m m m − 2 1 2 1 2 c c m m (1) ω2 1 3 1 2 2 2 1 1 2 1 2 2 2 1 1 2 , ( ) ( ) = + + ± + + c c m c m c c c − 2 1 2 1 2 c c [ ] The eigenfrequency of Model B cannot be calculated from eq. (1). If we try to use the equations belonging to model B in Fig. 13 by calculating the rotational eigenfrequency at the limit 2 by Limit with 20 sleepers by eq. 1 to 5 1 81,469 81,464 2 132,249 132,313 3 222,270 222,657 Tab. 2: Frequencies [Hz] Fig. 14: Excitation frequencies of a wheel-set-track-system with wooden sleepers at 250 km/h and limits of the mode free areas.

3rd edition ofWorld Rail Market Study Now available! World Rail Market Study 2010 Based on a survey conducted in the 50 largest rail markets worldwide, the study provides an update on the installed base and market volumes. Unique market growth predictions are displayed for the short-term and 2020 time horizon per product segment and regions. On top of that, the UNIFE Rail Market Study elaborates strategic conclusions for the industry and reflects on market developments taking into account the impact of the economic downturn and several possible political and economic scenarios. A study conducted by UNIFE, the European rail industry in co-operation with Boston Consulting Group (BCG) published by Eurailpress It is the largest study of its kind and a major reference for the rail community. www.eurailpress.de l www.railwaygazette.com Contact: DVV Media Group GmbH l Eurailpress &NBJM TFSWJDF!FVSBJMQSFTT EF t 1IPOF t 'BY More information (incl. executive summary) at www.eurailpress.de/wrms

RTR 4/2010 16 (track 1), had been jointed sections with a length of 60 m, a “54 E2” rail profile and an “R 320 Cr” grade of steel on wooden sleepers. The wear had been so intense that the rails had needed grinding every 2–3 years and replacing every 4-5 years. In the course of the replacement work, the 60-metre rail sections were continuously welded and safety caps were placed on all the sleepers. The rails laid on track 2 had the “54 E2” profile and were made of the “R 350 HT” grade of steel, whereas those laid on the experimental track (track 1) were in 60-metre lengths and made of various different grades of steel from the manufacturers Voestalpine and Corus (recently renamed Tata Steel Europe). The “Westbahn” railway line between Vienna and Salzburg is more than 150 years old and is the most important eastwest link in Austria. It is a constituent part of the Trans-European Network (TEN) and at the same time the most heavily-trafficked line in the whole network of the Austrian Federal Railways (ÖBB). With nearly 360 trains a day (counting both directions) and a load of around 30 million gross tonnes passing over each track each year, this line is extremely demanding in terms of both operations and maintenance of the infrastructure. It is, in particular, on the first forty kilometres of the line at the Vienna end, known as the “Wienerwald section” that there are tight curves in the track with radii as small as 270 m, permitting maximum speeds of only 70 km/h. The outlay on maintaining the permanent way over this section of the line is very considerably higher than on the other parts of it. 1 Test section on a mainline track In June 2003, the ÖBB laid tracks from various suppliers with high mechanical properties, including a Brinell hardness of 350 HB and more, in a tight curve as an experiment and then observed the development of rail wear and the formation of rail corrugations over a period of five years. The Railway Institute at the University of Innsbruck was appointed to accompany the experiment by making measurements throughout, by presenting the results in graphic form and by interpreting them. The halt of Eichgraben/Altlengbach lies on a right-hand curve extending over about one kilometre on the double-track railway line from Vienna to St. Pölten and Salzburg (km 28.921 – 29.843) with a radius of 280 m, a cant of 160 mm and a maximum permitted speed of 75 km/h (Fig. 1). It was the eastbound Salzburg–Vienna track (track 1) that was chosen for the experiment. At this location, the line climbs with a gradient of approximately 10‰. The daily load on the track is around 80000 gross tonnes, comprised 55% of freight trains and 45% of passenger trains. The older tracks in use before the relaying operations in 2002 (track 2) and 2003 Hard rails in tight curves The combination of increasing loads on tracks, high tractive forces and limited maintenance budgets is forcing railway infrastructure managers to optimise the wheel/rail inter face still fur ther. The use of hard materials for rails might be one way of contributing to extending the service life of rails laid in tight curves. Head of permanent-way technical division ÖBB Infrastruktur Bau AG, Vienna bernhard.knoll@oebb.at Dipl.-Ing. Dr. techn. Bernhard Knoll Professor University of Innsbruck, Railway and public-transport section guenter.prager@uibk.ac.at Ao. Univ.-Prof. Dipl.-Ing. Dr. techn. Günter Prager Professor emeritus University of Innsbruck, Railway and public-transport section erich.kopp@uibk.ac.at Em. Univ.-Prof. Dr.-Ing. Erich Kopp Fig. 1: Experimental curve on the Salzburg-St. Pölten-Vienna line. Radius 280 m, cant 160 mm, maximum speed 75 km/h. This track is used predominantly by eastbound traffic (Source: the authors)

RTR 4/2010 17 Hard rails in tight curves „ ones, and several short-pitch ones merge into long-pitch ones again too. This has consequences for the mean depths of corrugations, which may fall after initially rising on account of differing numbers of corrugations. It is only once a high corrugation intensity has been reached that the corrugation lengths remain uniform. The characteristic measurement logs are produced from this time onwards, and the mean corrugation depth no longer declines. This point in time has not yet been reached for any of the thirty measurement sections in this particular experimental curve. For that reason, it is difficult to issue a judgment sufficiently supported by facts on the basis of the available mean values alone. The decision on whether or not the time has come for grinding the inner rail is usually taken considering the mean corrugation depth as a comparative parameter. After a period of five years and a load on the track of approximately 150 million gross tonnes, the mean corrugation depth is generally less than 0.10 mm. It is only at two measuring points at the end of the curve that higher values of up to 0.18 mm were measured, affecting the steel grades “370 LHT” and “R 350 HT”. ning metre (-/m) and the corrugation intensity (as a percentage). The parameters of vertical wear (W1, mm), lateral wear (W3 at less than 45°, mm) and the worn area (A, mm2) were recorded throughout the whole of the observation period over the entire length of the curve, as were the highest and lowest values. Diagrams were produced to assist in the interpretation of these wear parameters for each measurement point of the inner and outer rail for each of the individual rail profiles. 3 Results 3.1 Rail corrugations caused by wheel skid At the beginning of the investigation, both short-pitched and long-pitched rail corrugations developed and were irregularly distributed. This confirms a phenomenon that had been observed on earlier occasions. It is only once the corrugations cover the whole of the sections being measured that the corrugation intensity tends towards 100%. In the course of time, the long-pitch corrugations evolve into several short-pitch Corus provided rails of the types “350 SHH” and “380 MHH”, while Voestalpine provided the types “R 350 HT”, “370 LHT” and “400 UHC”. The sequence of the rails (Fig. 2) was computed using a “fairness factor”, which was intended to consider the dependency of wear on the actual position within the curve as accurately as possible and to ensure a fair comparison of the various grades of steel. New welding instructions for these rails needed to be developed at short notice. Looking back with the benefit of five years’ experience of the experimental track, it is now possible to ascertain that the method chosen was indeed a successful one. Even at the time of writing, there is no such thing as a standard solution for the welding of rails made of hard materials. 2 Arrangement and execution of measurements After the test rails had been laid and ground on 27 June 2003, the zero measurements of the whole curve were made by the University of Innsbruck on 2 July 2003. These have since been followed by ten further measurements of the rail corrugations and wear at a rate of two measurements per year (in the winter and summer months), which it was hoped would reveal any influence of climatic environmental conditions. Two measuring points (on the inside and outside) were set up on each of the 60-metre test-rail sections approximately one third and two thirds along each of them. At each of the measurement cross-sections, the top surface of the inner rail was measured over a length of four metres with a Cemafer mechanical longitudinal profile measuring device. The recorded data was digitised and interpreted according to the length of the rail corrugations caused by wheel skid, their depth, their number and their intensity. On each occasion, a measurement was made of track 2 (i.e. at 33 mm from the bolt contact point on the running edge in the direction of the middle of the rail head. The corrugation intensity (expressed as a percentage) is defined as the product of the number of medium-pitch corrugations and the mean corrugation length relative to the measured length and is used as a comparative parameter in the interpretation of the results. The measurement of the wear parameters (vertical and lateral wear) was performed at each measurement cross-section on the inner and outer rail using an electronic transverse-profile measuring device of the “MiniProf” type. The presentation of the values measured on the inner rail is done by indicating the mean corrugation depth (in mm), the mean medium-pitch corrugation length (in mm), the mean number of corrugations per runFig. 2: Arrangement of the sixty-metre experimental rails and numbering of the measurement cross-sections (Diagram: DVV) Fig. 3: Lateral wear on the outer rails of the curve [mm] at the measurement cross-sections along the length of the curve (Source of Figs. 3–5: the authors)

RTR 4/2010 18 „ Hard rails in tight curves For the sake of completeness, it ought to be mentioned that the inner rail was ground for the first time on 31 November 2007 for the purpose of reducing noise, although there was not yet any compelling reason to do so from the permanent-way engineering point of view. That alone, however, means a lengthening of the grinding cycle of at least 1–2 years compared with the previous situation with less wear-resistant rails. The length of the corrugations showed very marked scatter between measurements. There is, however, a detectable tendency for corrugations to form with wave lengths of around 0.18–0.25 m. The corrugation intensity in the experimental curve that is the subject of this report was scattered between some 30% at the beginning of the curve and approximately 60% at the end of it. The mean number of corrugations per running metre in the zones around the measuring points is also very unevenly scattered, ranging from 1 to 5. 3.2 Wear parameters The lateral wear on the outer rail of the curve displays a very considerable increase from the beginning of the curve to its end. This is a common occurrence on track curves over which trains run predominantly in one direction. Figure 3 shows this development over the whole length of the experimental curve. This phenomenon was also taken into consideration in deciding on the sequence in which to install the various grades of steel. The chosen trend line was a second-order polynominal. According to the final measurement made on 6 May 2008 (grey line), the lateral wear values on the outer rail range from less than 2 mm to 3.9 mm at the end of the curve. No lateral wear occurs on the rail on the inside of the curve, but instead of that there are rolling laps over the whole curve, scattered with an order of magnitude of 0.1–0.7 mm. The vertical wear on the outer rail (Fig. 4) is in the approximate range of 0.7 to 1.5 mm. The vertical wear on the inner rail, on the other hand, is within the range of 2.9 to 3.9 mm (Fig. 5). One conclusion that is applicable equally to both rail suppliers (Corus and Voestalpine) is that both the vertical and lateral wear is clearly dependent on the grade of steel: the harder the rail material, the less the wear. (Editor’s remark: It is the 400 UHC grade, which was laid in the measurement crosssections 5, 8 and 11, that displays the lowest amount of wear). 3.3 Development of the rail surface In the course of the five-year observation period, the following changes were ascertained in the surface of the rails: Several wheel-slip marks. These were caused by fully-laden freight trains that became “stuck” on the upgrade, Minor head checks on the running edge of the outer rail over the whole length of the experimental track, and Slight depressions in a few of the welded joints. These show up the urgent need for the development of suitable weld portions for welding hard rails to one another. 4 Concluding remarks The Austrian Federal Railways have now completed a five-year test of rails from Corus and Voestalpine with different degrees of hardness, laid in a tight curve on a heavily trafficked railway line. The evaluation of the measurements performed by the University of Innsbruck showed clearly the advantages of rails with strong mechanical properties, including a Brinell hardness of greater than 350 HB – namely a slowing down in the formation of corrugations caused by wheel skid on the inner rail at the same time as much reduced vertical and lateral wear on both the inner and outer rail. The use of harder rails made it possible to extend the tamping and grinding intervals and also the service life of rails in the track by a very considerable margin (likely to be a factor of 2-3). It would, however, be wrong to ignore the fact that harder rails might also cause undesired side effects, which have not been adequately analysed to date. These might include, inter alia, developments affecting the surface of rails (such as head checks and spalling), impacts on the wearing properties of wheels and stresses acting on other components of the permanent way, especially the rail fastenings, the fatigue strength of the rail and the propensity for factures to occur in rails and/or welds. Given the positive experience with the experimental section, hard rails (> 350 HB) are now being laid in greater numbers in tight curves on heavily trafficked lines within the ÖBB network, and developments are continuing to be observed. Fig. 4: Ver tical wear on the rail on the outside of the curve [mm] at the measurement crosssections along the length of the curve Fig. 5: Ver tical wear on the rail on the inside of the curve [mm] at the measurement crosssections along the length of the curve