1. SUMMARY

In continuously welded rails high compressive and tensile forces can be created due to thermal expansion of the rail material Compressive force together with dynamic loads applied to the track by train can lead to dangerous track buckling.

Theoretical prediction of thermal stress, based on the measurement of rail temperature, are difficult due to numerous, usually unknown factors describing track structure and its displacements. Various nondestructive methods of stress evaluation were tested in laboratory and field conditions but none of them found practical application..

Paper describes ultrasonic monitoring of longitudinal forces in continuously welded rail. Presented are laboratory and field tests performed with shear waves propagated in the rail height direction and with longitudinal, subsurface waves propagated along the rail.

2. INTRODUCTION

Since the 1950s portable welding machines and termite welding technology have enabled rapid rail welding in field conditions.. Continuously welded rails (CWR) become popular what resulted in lengthened rail life and reduced expenditures on track maintenance.

Application of CWR made it necessary to control the longitudinal forces created in the rails due to temperature changes and external loads. Tensile forces in CWR can result in track cracking and compressive forces can lead to dangerous track buckling and train derailment.

To prevent track buckling some railroads companies lower the train speed during hot summer days in order to reduce dynamic loads applied to the track by trains. Another way to reduce buckling danger is to reduce compressive stress in the CWR by cuffing out a short pieces of the rails and than welding the track together again. Both methods are destructive and expensive. Specially if there is no information when and where dangerous compressive forces exist in the rail.

The evaluation of forces in CWR based on ambient or rail temperatures measurement only is difficult due to many factors influencing track bed resistance and uneven temperature distribution along the rails [1]. The second reason are other than thermal longitudinal stresses which develop in the CWR. Therefore numerous techniques of stress measurement, based on various physical phenomena, were tested for CWR.

Despite many years of experience problem is still not solved and there is no simple and fast, nondestructive method of longitudinal force evaluation in the CWR The aim of the paper is to present the possibilities and limitations of the simple ultrasonic technique for stress monitoring in CWR.

3. SOURCES OF STRESSES IN CWR

Fig. 1. Typical distribution of residual, longitudinal stress in the rail straightened in the roller straightener.

Stresses in CWR are of various origins and are created in various stages of the rail life. First stresses are introduced during rail shape rolling straightening in roller straightener. Fig. 1 presents typical distribution of longitudinal component of residual stress in the rail. Tensile stress is observed in the rail head and base, compressive one in the rail web [2]. Values of stresses can vary depending on straightening procedure and rail heat treatment. Important feature of residual stresses in the new rail is that their value integrated over the rail cross section is equal to zero and they do not create any longitudinal force in the CWR.

Additional residual stresses are created in the rails during welding. These stresses are localized and observed on the distance of about 500 mm from the weld. If welding operation was performed on the rail before it was fastened to the sleepers, welding also do not create any longitudinal force in the rail but significantly changes stress state in the vicinity of the weldment.

The next operation after welding is track geometry regulation during which both rails of the track, connected to the sleepers, are moved in horizontal and vertical directions. Depending on the magnitude of rail displacements, this operation can create longitudinal force in the CWR [3]

The main source of longitudional forces in the CWR is thermal stress caused by thermal expansion of the rail material. At the end of the CWR section this stress is equal to zero. In the middle of the section its value depends on the difference between actual rail temperature and the temperature at which the CWR was welded and connected to the sleepers. The temperature at which welding of the CWP was performed is called ,,rail neutral temperature". It is assumed that at this temperature, which is usually in the range of 18 - 30°C for UIC-60 rail longitudinal stress due to rail thermal expansion is equal to zero.

Any CWR temperature changes from ,,neutral temperature" win result in longitudinal stress. For the straight rail, properly connected to the sleepers, temperature change equal to 1°C results in a longitudinal stress change of 2.5 MPa It means that for UIC-60 rail the temperature change of 1°C results in logitudinal force equal to 18.6 kN. Durmg hot summer days the CWR temperature can reach about 70°C. Assuming that its neutral temperature was 18°C, the thermal force in one rail can reach about 1000 kN. In extreme climatic conditions the difference between maximum and minimum CWR temperatures were measured up to 120°C. In such a conditions, the thermal force acting in the track can be higher than 2000 kN [4].

The next important source of longitudinal force in the CWR is track longitudinal displacement. The movement of the track in respect to the ground can be caused by forces applied to the track during train braking and acceleration, repair works performed on the track or uneven thermal force distribution along the CWR In some regions the track is ,,shortened" and compressive forces are created in the CWR Forces due to rail displacement can be different in both rails what results in bending of the track. Asymmetry in the stress state in the track can be also the reason of track buckling.

Stress concentrations due to CWR displacement can be usually expected close to the bridges, joints or road crossings where the track bed resistance can rapidly change. In such a places, thermal and other stresses summed up can create high forces in the CWR and values of these forces are difficult to predict.

In the rail in the track, the pattern of residual stresses is changed by plastic deformation of the rail head running surface. Plastic deformation is caused by high contact stresses between rail and wheels of passing trains. The thickness of the cold-rolled layer depends on mechanical properties of the rail material and load applied to it. Usually this layer is few millimeters thick and longitudinal compressive stress is observed in the top layer and tensile stress in the bottom part of deformed layer, few millimeters below the running surface.

4. METHODS OF THERMAL STRESSES MEASUREMENTS IN RAILS

Various techniques were tested to evaluate thermal stresses in CWR Their features are shortly summarized in [5]. The simplest, but destructive method, is cutting the rail and measurement of rail ends displacements. Using this technique one can determine approximate value of stress in the rail before it was cut. The second technique, which is in use in many countries but is time consuming, is to monitor rail displacement in numerous points along the track. Every 200 m the mark is made on the rail and its position in respect to the ground is recorded. The measurements are periodicalb repeated and if the distance between two adjacent marks changes it means that the stress in the rail changes. Track fragments where distances between marks are shorter than before are in compression. Where distances between marks are longer than before - the rail is in tension [4].

Resistance strain gauges are in use on test tracks to monitor forces in the rails. Various magnetic techniques and Barkhausen noise method were also tested. Semi-nondestructive Rail Lift-Up technique was developed and tested in the USA. Stress meters based on rail temperature moment only were also developed in Australia In 1988 shear subsurface ultrasonic SH waves generated in the rail web by EMAT transducers were tested as a tool for thermal stress evaluation in CWR [6].

Unfortunately, after many years of experiments none of nondestructive techniques has found practical application in railroad companies. Some of them, like Barkhausen noise method, are sensitive to rail material magnetic properties, electric current in the rail and structure variations. Other, like one described in [6], can not provide sufficient accuracy. Stress evaluation based on rail temperature only neglects other than temperature sources of forces in the track. In result, the only technique which is widely used in practice today is the measurement of the rail displacements. This one however is time consuming and not accurate.

5. ULTRASONIC STRESS MEASUREMENTS

Ultrasonic stress measurements are based on elastoacoustic phenomena it is on the dependencies of ultrasonic wave velocities on stress. Velocity changes depend on material properties, ultrasonic wave type and relation between directions of wave propagation, polarization and stress.

Two techniques are the most popular and present high sensitivity to stress. The first one is the measurement of acoustic birefringence [7]. The first application of this technique for railroad industry, almost 25 years ago, was the measurement of hoop residual stress in the monoblock wheels [8]. For stress evaluation in raiL shear waves polarized along and perpendicular to the rail axis propagate in the direction perpendicular to the rail axis, between rail head and base. Schema of the measurement is presented on Fig. 2.

The difference in times of flight of pulses polarized along and perpendicular to rail axis is proportional to the longitudinal force in the rail averaged value of residual stresses and rail material texture. If the anisotropy due to texture and the influence of residual stresses on measured times of flight are known value of stress can be calculated according to formulas:

= B_o + B_s = 0,5 * (t_L - t_P ) / (t_L + t_P)

where B is total acoustic anisotropy, B_o is anisotropy due to texture and residual stresses, B_s is thermal stress induced anisotropy, t_L and t_P are time of flight for longitudinal and perpendicular polarization directions respectively,

= (B_o - B_s )/ _B where ~ is thermal stress and _B is elastoacoustic constant for birefringence method.

Fig. 2 Schema of acoustic birefringence measurement in the rail

Advantages of this technique are important prom practical point of view. Results are temperature independent, simple probehead for shear waves is used for generation and detection of ultrasonic waves. Probehead is coupled to the easily accessible rail head and the measurement can be performed on rails of various height. The waves propagate from the rail head to the rail base averaging the residual stresses which integrated over rail cross section are close to zero. If, for a given rail type, the rail texture due to preferred grain orientation is constant and its value is known, absolute value of longitudinal force in the rail can be determined.

Another simple approach is to use longitudinal wave propagating along the rail. Longitudinal wave propagated long the stress is the most sensitive one to stress and are used for residual stress evaluation in new rails [9]. Schema of the measurement with longitudinal subsurface waves is shown in Fig. 3

Fig. 3. Schema of the measurements with longitudinal subsurface wave propagated along the rail

Advantage of this technique is the highest sensitivity to stress and possibility to measure time of flight of ultrasonic pulses over a bug distances. Unfortunately the velocity of longitudinal wave propagated along the rail head side is influenced by numerous other than stress factors They are residual stresses, material chemical composition, material texture and temperature [103. It means that using longitudinal wave only one can not measure absolute value of longitudinal force in the raiL

However using this wave one can easily measure stress changes in chosen locations on the CWR For a given location on the raiL it can be assumed that the acoustic properties of the material are constant, residual stresses are also constant and the only factors influencing measured time of flight are temperature and longitudinal stress in the rail. The time of flight changes caused by temperature variations only can be measured in the temperature chamber for a given probehead. Such temperature correction measured as a sum of temperature velocity changes in the rail material and plastic wedges of the probehead, thermal elongation of the probehead elements and plastic wedges, has to be taken into account during stress change calculation.

To avoid any influence of plastic deformation due to the contact rail-wheel on results the measurements should be performed in the region where rail surface is not affected by wheels. Such a region is external side of rail head. The surface in this part is Sat and undisturbed pulses of longitudinal subsurface wave 2MHz frequency one can observe on the distance up to 500 mm [3]. For such a distance between probeheads the stress change equal to 10 MPa results in the time of flight change equal to about 10 ns.

Plastic deformation on the head running surface result in the change of residual stress distribution in the whole rail. The measurement performed on the rails subjected to rolling in the stand showed that significant plastic deformation on the rolling surface has negligible influence on time of flight of longitudinal subsurface wave propagated along the external head side [11].

If the first measurement can be performed on thermal stress free rail, for example during laying or destressing operation when the rail is cut and free of any longitudinal forces, next measurements performed in the same location will inform about absolute values of these forces.

Next chapter describes the results of experiments performed in the laboratory and in the field condition using No techniques described above - acoustic birefringence and measurement with longitudinal subsurface waves.

5.1 MEASUREMENT OF ACOUSTIC BIREFRINGENCE IN RAILS
Fig. 4 presents the dependencies between times of flight of shear waves polarized along and perpendicular to rail axis and longitudinal stress applied to the rail. Measurements were performed on a sample made of new UIC-60 rail and subjected to compression in the machine. The probehead, equipped with 4 MHz, 10*10 mm piezoelectric transducer, was positioned on the rail rolling surface.

Linear changes of times of flight for both wave polarization direction and high time of flight changes due to stress, specially for wave polarized along the stress direction can be seen. The difference between times of flight for two waves resulted form 10 MPa stress increment in the rail is equal to about 7,8 ns.

Fig. 4. The dependencies between times of flight of shear waves polarized along and perpendicular to rail axis and longitudinal stress applied to the rail.

······[]····· wave polarized parallel to stress direction ------o----- wave polarized perpendicular to stress direction

Using shear waves propagating from running surface to the rail base, it can be assumed that the influence of residual stresses, shown in Fig. I, on acoustic birefringence will be minimized. The waves cross the areas subjected to compression (head and base) and subjected to compression (web). This assumption was checked on two rail samples. Acoustic birefrineence measured on new rail, with residual stresses, and on rail after stress relieving annealing, were practically the same.

However experiments performed on various samples made of new rails of the same rail type and steel grade showed that measured acoustic birefringence is not repeatable. Experiments performed on used rails showed that the readings are strongly affected by rail head running surface geometry. Wheels of passing trains cause plastic deformation of the thin material layer and in used rails the running surface can be curved and not parallel to the rail base. This changes in the rail geometry result m changes in wave propagation direction and unpredictable redactions of the waves on thin rail web surfaces. As a result the received pulses are disturbed and measured times of eight are not precise. Experiments performed on new rails in the track also showed unacceptable scatter of readings.

Experiments performed with various transducer sizes and wave frequencies in order to produce narrow, low divergence ultrasonic beam and to eliminate the influence of web reflections were also unsuccessful.. Concluding, it was found that this promising technique can not be applied to used rails in service, in which rail head running surface is deformed. The scatter of readings caused by rail geometry variations was to high to evaluate thermal stresses with sufficient accuracy.

Better results were be obtained with transmitter positioned on the rail head running surface and the receiver coupled to the rail base underside. This approach however reduces the length of the path on which times of eight are measured by 50%. Base underside is usually covered with ballast, is rusty and is not visible what makes receiver positioning difficult and not practical

5.2. MEASUREMENTS WITH LONGITUDINAL SUBSURFACE WAVES


Fig. 5. The dependencies between times of eight of shear waves polarized along and perpendicular to rail axis and longitudinal stress applied to the rail.

Fig. 5 shows the dependence between time of eight of the wave propagated along the head side and stress applied to the rail sample. The distance on which the time of flight was measured was about 450 mm and the sample was made of UIC-60 rail and subjected to tension in the machine.

Elastoacoustic constant ßL longitude wave propagated parallel to stress can be calculated as:

= (t_s-t_o)/(t_s* ) [MPa^-1] where ts and to are times of flight in stresses and zero stress conditions respectively, is stress.

For UIC-60 rail = -1,25 [MPa^-1]. From the diagram h can be seen that stress increment equal to 10 MPa results in time of flight increment equal to about 10 ns. It means that using longitudinal subsurface wave propagated on the distance 450 mm, time of flight changes due to stress are higher than using birefringence technique.

The measurement with longitudinal subsurface waves were performed with multitransducer probeheads equipped with 2MHz piezoelectric transducers. The application of several transducers arranged along one line can significantly reduce unwanted influence of rail surface roughness on the measured times of flight [12].

In 1992 the measurement of stresses introduced into the CWR during track laying. It was shown that in the neutral temperature when thermal stress is assumed to be zero, in some locations on the track the bughudinal forces acting in each of the rails can reach more then 6 kN [3]. Figures presented below show result of other field measurements performed with portable DEBRO-30 device.

Fig. 6. The dependencies stress increment - rail temperature measured with subsurface 1ongitudinal wave on various tracks.

Fig. 6 shows thermal stress versus rail temperature measured on various CWR Line denoted as A shows the changes of stress observed during one day and measured on the straight CWR The slope of the straight line is equal to 2,5 MPa/°C what is equal to theoretical predictions.

It was assumed that rail neutral temperature was 22°C. Lines denoted as B and C show stress changes observed during much longer time period.The slope of the straight line is equal to 2,5 MPa/°C what is equal to theoretical predictions. It was assumed that rail neutral temperature was 22°C. Lines denoted as B and C show stress changes observed during much longer time period.

Stress changes presented as line B were measured close to the road crossing and as line C - in the distance from the road crossing [13]. For line B stress-temperature dependence is equal to 2,6 MPa/°C and for line C - only 1.9 MPa/°C. It means that depending on location on CWR the same temperature changes can result in various stress changes. Much higher stress scatter on line B comparing to lines A and C is probably due to other than temperature factors influencing stress state.

Fig. 7. Stress changes along the rail due to repair (destressing) operation.

Fig. 7 presents stress changes measured during track repair. The rail were cut, short piece of rail was removed and rails were welded together. The aim of the operation was to reduce compressive force in the CWR.

It can be seen that the operation reduced compressive stress along the 112 m track section and the average stress change is equal to 43 MPa. Variation in stress increment along 150 m long section was caused probably by variations in the rail temperature. Some parts of the track were in the shade, other in the sun.

6. CONCLUSIONS

Using small, lightweight portable and battery operated ultrasonic devices equipped with multitransducer probeheads for longitudinal subsurface waves it is possible to monitor stress changes in the CWR The measurements are very fast and can be performed without any traffic disturbances. Such measurement can help to control forces in track sections where longitudinal stress can occur due to rail displacement, temperature changes or other factors.

Multitransducer piezoelectric probeheads minimize the influence of rail surface roughness on readings and the measurements can be repeated in chosen CWR locations without time consuming rail preparation. It was shown that the evaluation of forces in the CWR based on rail temperature only is not precise enough and thermal stress depends not only on temperature but also on location on the track.

7. REFERENCES

1. J. Richards, Weather and site effects on rail temperatures, The Railway Gazette, June 19, 1964, pp. 505-512.
2. J. Deputat, J. Szelazek, A. Kwaszczynska-Klinek, Experiences in Ultrasonic Measurements of Rail Residual Stresses, Residual Stress in Rails, VoLI, 169-183, O.Oringer et al. (eds.), 1992 Kluwer Academic Publishers, Printed in the Netherlands
3. J. Szelazek, Ultrasonic measurement of thermal stresses in continuously welded rails, NDT&E International, Vol.25, No.2, pp.77-85 (1992)
4. K Towpik, Tracer Maintenance, Editor: WED Warszawa 1990, (in Polish)
5. A. Kish, M. Coltman, New techniques for rail longitudinal force measurement and rail restraint capacity evaluation., American Railway Engineering Association - Bulletin 727, pp.236-240.
6. G.A. Alers, A. Manzanares, Use of surface skimming SH waves to measure thermal and residual stresses in installed railroad tracks, Review of Progress in QNDE, Vol.9B, pp.l757-1764, D.O.Thornpson and D.E.Chimenti (eds.), Plenum Press, New York (1989)
7. RW. Benson, V.J.Realson, Acoustic-Elasticity, Product Engineering, Vol.30, No.56 (1959)
8. R.R King, J.A. Birdwell, D.E.Bray, W.N. Clotfelter, E.R.Rish, Improved method for nondestructively measuring residual stress in railway wheels, Proc.9th Symposium on NDE, San Antonio, April 25-27, 1973, pp.91-105
9. A. Brokowski, J. Deputat, Ultrasonic measurement of residual stress in rails, Proc. I I WCNDT, Las Vegas 1985, Vol.l, Taylor Publ.Co., Dallas, p; 592
10. R.B. Thompson, S.J. Wormley, J.C. Johnson, Elastic and Acoustoelastic Properties of Railroad Rail, Review of Progress in QNDE, Vol.14, pp.l891-1898, D.O.Thompson and D.E.Chimenti (eds.), Plenum Press, New York (1995)
11. J. Deputat, J. Szelazek, M. Adamski, Experiences in ultrasonic measurements of stresses in rails, Proc. Int.Conf. on Rail Maintenance for Modern Railway Operation, Delft June 1992, Kalker J.J., Cannona D.F., Orringer O. (eds.), Kluwer Academic Publishers, 1993, pp.109-118.
12. A. Brokowski, J. Szelazek, Set of ultrasonic probeheads for the measurement of time of flight of ultrasonic pulses, Patent USA, Nr 5,549,001, 27 July 1996
13. D. Utrata, Longitudinal Rail Stress Measurements, Association of American Railroads, Research Report R-779, Chicago, 1993.

The paper was presented at the annual DGZfP conference May '97 in Dresden