In-situ investigation of the behaviour of a French conventional railway platform

Transcription

1 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang 0 0 In-situ investigation of the behaviour of a French conventional railway platform October 0 Word Count: Words + Figures * 0 + Table * 0 = Words Corresponding Author: Francisco Lamas-Lopez Phd Candidate SNCF- Engineering Phone: (+) (0)0 francisco.lamas-lopez@enpc.fr Sofia Costa d'aguiar Project Manager SNCF - R&D Phone: (+) (0) sofia.costadaguiar@sncf.fr Alain Robinet Research Division Manager SNCF - Engineering Phone: (+) (0)00 alain.robinet@sncf.fr Jean-Claude Dupla Professor Ecole des Ponts ParisTech Phone: (+) (0) dupla@cermes.enpc.fr 0 0 Yu-Jun Cui Professor Ecole des Ponts ParisTech Phone: (+) (0)0 cui@cermes.enpc.fr Nicolas Calon Project Manager SNCF - Engineering Phone: (+) (0)00 nicolas.calon@sncf.fr Jean Canou Professor Ecole des Ponts ParisTech Phone: (+) (0) canou@cermes.enpc.fr Anh-Minh Tang Professor Ecole des Ponts ParisTech Phone: (+) (0) tang@cermes.enpc.fr

2 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang 0 0 In-situ investigation of the behaviour of a French conventional railway platform Francisco Lamas-Lopez - Phd Candidate; SNCF Engineering Ecole des Ponts ParisTech Yu-Jun Cui - Professor; Phd Supervisor; Ecole des Ponts ParisTech Sofia Costa d'aguiar - Project; SNCF R&D Nicolas Calon - Project Manager; SNCF Engineering Alain Robinet - Research Division Manager; SNCF Engineering Jean Canou - Professor; Ecole des Ponts ParisTech Jean-Claude Dupla - Professor; Ecole des Ponts ParisTech Anh-Minh Tang - Professor; Ecole des Ponts ParisTech ABSTRACT Concerning the 0,000 km French conventional railway network ( % of the entire network), the speed is currently limited to 0 km/h as maximum, whereas for the,00 km high speed lines, the speed can go up to 0-0 km/h. Nowadays, there is a growing need to reduce the travel time and improve the service by increasing the speed limits on the conventional network. This paper aims to study the influence of train speed on the mechanical behaviour of an experimental site representative of the French conventional network. Emphasis is put on the behaviour of the interlayer soil. The selected experimentation site is located near Vierzon (Centre/FR). More than 0 sensors, including accelerometers and soil pressure gauges, have been installed. A testing campaign has been performed with a fully instrumented experimental intercity train composed of a locomotive and seven coaches running at different speeds (0 0 0 and 00 km/h). This campaign has revealed that when the train speed is increased, the loads transmitted to the platform increase. Moreover, these loads are amplified at the depth of interlayer soil. Indeed, the upgrade of limit speed of a line leads an increase of the vertical soil pressure and the vertical acceleration in the soil. The acceleration data obtained from the accelerometers will be used to determine the maximum vertical strain induced in the interlayer soil for different train speeds. KEYWORDS: train speed increase, interlayer, dynamic behaviour, in situ experimentation, maximum axial strain, reversible modulus

3 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang INTRODUCTION Nowadays, there is a growing need to reduce travel time in railway transportation. Therefore, an increase in train speed is needed to reach this goal. Most of the European railways network is composed of conventional lines, corresponding to service speeds under 0 km/h. The conventional lines and high speed lines must be complementary from each other. In France, almost % of the operational lines in use are conventional. Several projects aimed at upgrading the service speeds on conventional lines are currently being carried out. A good understanding of the mechanical behaviour of the materials composing the railway platform could help to optimize the maintenance operations when the upgrading operations are undertaken. The INVICSA project, initiated by SNCF (Société Nationale des Chemins de Fer) in 0, and presently under development, aims to study the influence of trains speed on the behaviour of conventional platforms. The main difference between a conventional track and a new high speed track is the existence of an interlayer, which has been slowly created in time between the ballast and the subgrade soil (-). This specific layer has been formed in time by interpenetration of ballast grains and fines from subgrade soil, added to ballast weathering and attrition. The nature and thickness of this interlayer will depend principally on the site and load history of the track. Several authors have studied the influence of train speed on the behaviour of the materials which compose the railway platform. Some of them have performed in-situ studies on problematic sites composed of loose clays with on-surface instrumentation, in order to monitor the sleeper s displacements, analyze the effect of the critical speed of a site () and analyze the vibrations induced by trains running at different speeds (,). Experimentation on railway platforms have also been performed on platforms presenting stability problems due to the presence of peat on different sites in Canada. These experimentations were carried out after occurrence of problems induced by train s speed upgrade on those sites (,). Other experimentation looked for measuring strains of platforms materials using Multi-Depth Deflectometers and strain gauges in order to analyze the contribution of each individual substructure layer to the differential settlement of the railway platform (). Moreover, some experiments have been performed, based on real scale physical models, in order to analyze the amplifications loads induced at different depths of the platform (,) as well as the accumulation of settlements on different materials (). On the other hand, the influence of train speed on the behaviour of railway platform materials has also been numerically modeled by differents authors, based on FEM, in order to analyse the mechanical behaviour of the track under trains running at different speeds (). The INVICSA project includes the development of an in-situ full scale experimentation on an instrumented section of a conventional line railway platform. The experimental site was chosen among the 0000 kilometers of the French conventional network (). The selection criteria were related to the limit speed allowed on the site, the main characteristics of the platform (alignment, cutting,...) and the conditions of the rails and sleepers, with no special maintenance operations needed since the last renewal works. The experimental site finally selected is near Vierzon, in the centre of France. The site is located at the P.K. of the 0000 line. The instrumented section of the track on the experimental site is m long. More than 0 sensors (accelerometers, geophones, soil stress sensors, porewater pressure gages,

4 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang 0 strain gages on rail and a meteorological station) were installed at different depths and positions along the experimental site. In the present study it will be used a soil stress sensor at the ITL depth (-0.0 m) and capacitive accelerometers: one at the ITL (-0.0 m) and one at the TL (-.0 m).. A representative cross section of the track, resulting from the in situ geotechnical investigations carried out, is presented in FIGURE. Technical specifications of the sensors are shown in TABLE. The soil stress sensor is 0 mm diameter to be adapted to the biggest grains in the ITL (about 0 mm). A finner subgrade soil was put around the sensor to ensure a good contact with the rest of soil. The installation methods have already been explained in detail in (-). Capacitive accelerometers were used in order to well measure the low frequencies, in the range - Hz, to double-integrate the signals and obtains displacements (). More information about the sensor's choice and site selection could be found in (). The platform is composed of about 0 cm of ballast, 0 cm of interlayer and 0 cm more of a transition layer soil until the subgrade is reached. The fresh ballast is composed by no fouled grains of diameters from 0 to. mm. The Interlayer (ITL) at the site is composed by ballast grains, with a maximal grain diameter of 0 mm, mixed with silty sand from the subgrade. About % of the total mass is finner than 0 µm. The D 0 of the ITL is around mm. The Transition Layer (TL) is composed by the same fines as the ITL but the presence of ballast grains (biggers than 0 mm of diameter) is limited to % of the total mass. The D 0 of the TL is mm at the selected site. The subgrade of the zone until meters depth is a silty sand with a D 0 of 0. mm and a Plasticity Index (PI) of. The water table in the area is located at a depth of.0 m. The bottom surface of the drainage system is also located at.0 m. More than 0 sensors have been installed on both tracks of the platform (Tracks & ). This paper will mostly present the results obtained on a capacitive accelerometer and on a soil pressure sensor installed on top of the ITL (- 0. m depth from the running surface) as well as on a second accelerometer installed at middle depth in the TL (-.0 m depth from the running surface). 0 FIGURE. Cross section of Vierzon experimental site

5 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang SENSOR Capacitive Accelerometer UNIT OF MEASUREMENT CAPACITY DIMENSIONS OTHER SPECS m/s² +/- g x x mm Soil Stress Gage kpa 00 kpa TEST CAMPAIGN ø : 0 mm H: 0. mm TABLE. Technical specifications of the used sensors in this study Frequency response : 0 Hz Natural Frequency : 0 Hz Rated Output Approx. mv/v (000x- strain). Nonlinearity : % RO In order to evaluate the influence of train speed on the interlayer behaviour, a fully instrumented Intercity train has been running over the instrumented site at six different speeds. This testing campaign has been performed during the first week of June 0. The Intercity train configuration used in the test is presented in FIGURE : Locomotive BB and Corail coaches. The locomotive BB length is. m with interbogies distance of. m and. m of interaxles distance. The total mass of the locomotive is 0 tons, about. tons per axle. Each Corail coach is. m long. The interbogies distance for this type of coach is. m between bogies of the same coach and m between bogies of adjacent coaches. As in the locomotive, the interaxle distance is. m. The mass of this type of coach is tons, resulting in. tons average mass per axle. 0 FIGURE. Scheme of the Intercity experimental train (locomotive + coaches) and their axles mass The testing program was composed of passages of the experimental train over the Vierzon experimental site, with different speeds (FIGURE ). It was organized on a days time period and had to fit with the regular commercial traffic still going on on the line, which is quite a busy one. Six passages have been carried out in the regular direction on the line over Track (Les Aubrais Vierzon); four passages have been performed over Track in the inverse direction (Vierzon Les Aubrais) and two passages have been performed in the regular direction over Track (Vierzon Les Aubrais). The results presented in this paper correspond to the passages on Track in the regular direction. In this configuration, six passages have been performed at six different speeds over track in its regular direction. In day two passages were performed at

6 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang 00 and km/h, other two passages in day at and 0 km/h and in day passages at 0 and 0 km/h were executed. FIGURE. Experimental site and instrumented experimental train The sensors installed in the platform were connected to a data logger (HBM-cx). Once a train running over the platform is detected (from strain gages glued to the rail used as triggers) a file in MATLAB format was created, recording all the installed sensors for seconds. The used sampling frequency is 00 Hz. The results from the six files generated (for each one of the different train passages) of this test campaign are analyzed in the next chapter. 0 RESULTS After each train passage, the test may be analysed based on the processing of the experimental data obtained, through the data acquisition system, from the different transducers located in the interlayer (pressure transducers and accelerometers). A specific file is obtained for each passage, with a sampling frequency of 00 Hz. The first interesting finding is that the amplification of the maximum stress registered by the soil stress sensor at the interlayer when speed is amplified when the speed is increased. FIGURE shows that the intercity train causes a σ z of about. kpa per locomotive s axle and. kpa for each one of the coach axles. The same train can load the soil up to an increase σ z of. kpa for the locomotive and. kpa for a coach when it is running at 00 km/h. The FIGURE shows the stress when the train runs at km/h, and we can realize that the stress induced in the soil by the train during its passage is intermediate with respect to the two other speeds. This increase results from the dynamic amplification of the train mass with increasing speed.

7 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang 0 FIGURE. Increase in vertical stress in the interlayer (-0. m) during the passage of experimental train at three different speeds Moreover, when the vertical acceleration at the top of the interlayer is analyzed, it may be seen that the maximum values of the acceleration increases also when trains run faster. The accelerometers used during this experimentation are of the capacitive type, with a natural frequency of 0 Hz and a frequency response of 0 Hz. FIGURE shows the Power Spectrum Density (PSD) of vertical accelerations for three different trains (0, and 00 km/h). Most energy is given by the train excitation to the platform in the first Hz. In the PSD, we can differentiate the peaks caused by the inter-bogies distance or half-coach distance (λ~ m) around the. Hz for the train at 0 km/h and Hz for the test train at 00 km/h. The inter-axles distance (λ~. m) induces frequencies between Hz for a train running at 0 km/h and Hz when the train runs at 00 km/h. Larger frequencies are excited, during the train passage, by lower wavelengths generated within the railway system as sleepers distance, wheel or rail defects. Since most displacements from a dynamic load are the result of low frequencies signals, we will analyze the accelerations corresponding to the first Hz. Each acceleration signal will be filtered with an elliptic filter. The filter s pass band is. Hz Hz. The high pass filter is used to avoid baseline effect if we integrate the signal to perform further analysis. Secondly, as we have analyzed the frequencies excited by the trains in order to filter the signals, we can avoid accelerations amplitudes from frequencies that do not induce displacements and deformations of the soil. FIGURE shows the vertical acceleration signal when trains are running at 0, and 00 km/h over the experimental site. There is a difference of about times on the acceleration amplitude when the speed is 0 km/h and the passage at 00 km/h for both locomotive and axle.

8 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang FIGURE. Power spectrum density of the untreated accelerometer signal in the interlayer (- 0.m) during passage of the test train at different speeds FIGURE. Vertical acceleration (fc= Hz) in the interlayer (-0.m) during passage of the test train at different speeds If the acceleration signal is double-integrated, the displacements of the embedded accelerometer in the interlayer may be obtained for each passage of train. The method used to perform these double-integrations has been described in (). There are smaller differences between displacements amplitudes than for the acceleration amplitudes when train speed varies. FIGURE shows the displacement signals corresponding to the passage of the intercity train at three different speeds. It is also interesting to note the differences observed between the locomotive displacement (four axles at the beginning of the signal) and the coach axle displacement (rest of axles). There is also an amplification of the displacements when the train speed is increased.

9 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang FIGURE. Vertical displacements (fc= Hz) in the interlayer (-0.m) during passage of the test train at different speeds As different passages over track in its regular direction has been performed, it can be analyzed the maximal amplitudes registered by the soil stress sensor and the capacitive accelerometer during each one of its passages. It is also needed to differentiate between locomotive and coach because the response of the sensors is not the same for both loads mass. FIGURE shows the evolution of the maximal σ z for the locomotive (average of first axles) and coach (average of last axles) for each passage. There is a linear trend for speeds between 0 and 00 km/h. As shown in FIGURE, there are no high variations in the soil stress sensor measurements during the different passages. The maximum value of σ z for each axle load is quite constant. FIGURE. Average increase of the vertical stress in the interlayer (-0.m) during passage of the test train at different speeds

10 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang If the stress amplitude per axle at 0 km/h is taken as a quasi-static value, it is possible to calculate the amplification for both locomotive and coach when the train speed increases. FIGURE shows the amplification in σ z caused by the locomotive, which is about % when train speed increases from 0 to 00 km/h. The amplification for the coach is about 0 %, because the quasi-static value of σ z is lower. But as already seen from FIGURE, the trends of load amplification for the locomotive and for coach are quite parallel. FIGURE. Amplification of σ z in the interlayer (-0.m) during passage of the experimental train at different speeds. As well as averages of vertical accelerations are calculated for locomotive and coach, it is possible to estimate the evolution of the vertical accelerations with train speed for the tested speeds (FIGURE ). The evolution of the maximal acceleration corresponds to a trending of the type: a z = x e x V () Being a z the maximal vertical acceleration at the interlayer for the train speed V with x and x two constants depending on the load amplitude and depth of the tested soil.

11 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang FIGURE. Evolution of the maximal vertical acceleration (fc= Hz) in the interlayer (-0.m) during passage of the test train at different speeds If accelerometers signals from FIGURE are double-integrate, it is possible to obtain the displacements signals for each train passage. Since there is another accelerometer embedded 0 cm deeper than the first one, it is possible to estimate the average strain within the interlayer; Both accelerometers are not installed in the same borehole, and it is therefore necessary to modify the time beginning to simulate how were the signal if Acc# and Acc# were installed in the same borehole. The location of the boreholes with respect to the rail and the sleepers is the same in both boreholes. The horizontal distance between both analyzed boreholes dist - is. m (distance between sleepers). To simulate that the two boreholes were at the same position, the signal needs to be moved a number of positions calculated from this formula: pos =dist f S V. () Being pos - the number of positions to be moved the signal, f s the sampling frequency in Hz and V the train speed in km/h. Once the signals are modified, the ε z can be calculated with the following relation: ε z =(u u ) 0/d () 0 In the formula (), u is the displacement signal calculated from accelerometer # and u the displacement signal from accelerometer #; d is the depth difference between both accelerometers (0. m). From the axial strain signal calculated for each passage with () we can look for the maximal values, and then separate the results in average for locomotive ( values each tested speed) and average for coach ( values each tested speed). The FIGURE shows the evolution of these maximal axial strains during passage of the test train at different speeds.

12 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang There is more variability for locomotive values but this is also the result of calculating the average with only axles per train passage and the assumptions made for the double integration process, based on the accelerometers data. These results show the order of magnitude of the strains for each loading cycle (bogies) during a train passage, for different loads (locomotive and coach). FIGURE. Evolution of the maximal vertical strain amplitude per loading cycle in the interlayer (-0.m) during passage of the test train at different speeds. Estimation calculated from the double integration of two accelerometers installed at different depths. So, we have the soils stress signal and the average axial strain signal calculated from displacements of acc# and acc# with relation (). Then, it is possible to plot the hysteresis cycles ( σ z -ε z ) during the passage of a train as shown in FIGURE for the different speeds. The largest cycles correspond to the bogies of the locomotive and the other cycles ( coach s bogies) are smaller in the hysteresis plots. The surface inside the cycles gets bigger when the train speed increases. Consequently, the damping ratio of the soil under train loading increases with train speed. In addition, the secant modulus can be calculated from these cycles linking the departure point of the soil loading with the point of biggest stress measured in each plot.

13 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang FIGURE. Hysteresis cycles ( σ z - ε z ) of the interlayer (-0.m) during passage of the test train at different speeds. As a result of the estimations of the reversible modulus for each one of the trains speeds, it is possible to plot the evolution of M r. FIGURE shows an estimation of the evolution of M r from the coach loading cycles plotted in FIGURE. The secant modulus for a given speed is the same for locomotive and coach. It may be observed that the reversible modulus M r decrease with speed at the same time that damping ratio (surface inside hysteresis cycles) increases. Between a quasi-static speed of 0 km/h and a speed of 0 km/h the reversible modulus decreases in % and between 0 and 00 km/h the estimated M r decreases in 0 %. The effect of the speed on the evolution of M r of the soil is non-linear

14 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang FIGURE. Evolution of the reversible modulus of the interlayer (-0.m) during passage of the test train at six different speeds 0 SUMMARY AND CONCLUSIONS This paper has presented preliminary results concerning the instrumentation of an experimental site located on a French conventional railway platform close to Vierzon (Centre/FR) in order to analyze the influence of the train speed on the mechanical behaviour of the interlayer. The interlayer is a heterogeneous layer located between ballast and subgrade in conventional railway platforms. In this paper, the data obtained from two capacitive accelerometers and a soil stress sensor embedded within the interlayer have been analyzed. A test campaign has been carried out on the experimental site. An intercity instrumented experimental train has been running over the site at six different speeds. The results have shown an amplification of σ z caused by the train when the speed increases. The order of magnitude of the amplification is about % for the locomotive loading between 0 and 00 km/h. The trend observed for the amplification is the same for the locomotive (. tons) and the coach (. tons). The amplification of vertical maximal accelerations (filtered at Hz) is about times larger for 00 km/h than for 0 km/h. The maximal amplitudes of vertical accelerations follow an exponential law depending on the load amplitude and depth of the accelerometer. From the double integration of two accelerometers located at different depths, it could be possible to calculate an average axial strain of the interlayer. The increase of speed leads to a two time increase in the average axial strain supported by the interlayer soil. The trend of the axial strain increase is the same for both locomotive and coach loads. The reversible modulus decreases (about 0%) especially between 0 and 00 km/h, contrary to the damping ratio which

15 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang 0 0 increases with speed. These results are being validated with laboratory experiences using Interlayer soils. The tests are performed on a large scale cyclic triaxial (00 mm diameter). Finally, it may be concluded that when the train speed increases, the stress taken by the soil is amplified within the platform as well as the axial strains induced by loading on the soil. The amplifications ratios trend is usually the same for the locomotive and the coach. The stress amplification caused by the passage of locomotives (largest mass loading the platform,. tons) is more aggressive to the soil, being responsible of an eventual increase of plastic deformations within the structure after a speed increase over the line. REFERENCES. Trinh, V-N. (0). Mechanical characterisation of the fouled ballast in ancient railway track substructure by large-scale triaxial tests, Soils and Foundations 0;():.. Duong, V. et al. (0). Investigating the mud pumping and interlayer creation phenomena in railway sub-structure. Engineering geology, Volume, pp -.. C. Madshus and M. Kaynia. (000). High-Speed railway lines on soft ground: dynamic behaviour at critical train speed, Journal of Sound and vibration (), -0.. Hall, L. (000). Simulations and analyses of train-induced ground vibrations. PhD thesis Royal Institute of Technology, Sweden.. Hall, L (00). Simulations and analyses of train-induced ground vibrations in finite element models. Soil Dynamics and Earthquake Engineering. pp. 0. Hendry, M. et al. (00). Train induced dynamic response of railway track and embankments constructed over soft peat foundations. In Proceedings of the th Canadian Geotechnical Confer- ence, Vancouver, B.C. pp... Hendry, M., Martin C., D., Barbour, L. (0). Measurement of cyclic response of railway embankments and underlying soft peat foundations to heavy axle loads. Canadian Geotechnical Journal. 0: 0. Chen, R.P. et al. (0). Experimental study on dynamic load magnification factor for ballastless track-subgrade of high-speed railway. Journal of Rock Mechanics and Geotechnical Engineering (0) 0. Mishra, D., Tutumluer, E., Boler, H., Hyslip, J., Sussmann, T. (0). Instrumentation and Performance Monitoring of Railroad Track Transitions using Multidepth Deflectometers and Strain Gauges. th Annual meeting of the Transportation Research Board; Washington DC, USA.

16 Lamas-Lopez, Cui, Costa, Calon, Robinet, Canou, Dupla and Tang. Chen, R.P. et al. (0). Dynamic soil pressure and velocity of slab track-subgrade in high-speed railways. Proceedings of the sixth Int. symposium on environmental vibration. - November 0, Shanghai, China. Xu, X.et al. (0). Accumulative settlement of saturated silt subgrade under cyclic traffic-loading. Proceedings of the sixth Int. symposium on environmental vibration. - November 0, Shanghai, China. Alves Costa, P. et al. (0). Influence of soil non-linearity on the dynamic response of high-speed railway tracks. Soil Dynamics and Earthquake Engineering 0 pp.. Lamas-Lopez, F. et al. (0). Field instrumentation to study the behaviour of a conventional railway platform. Proceedings of Georail 0 conference. - November 0, Marne-la-Vallée, France. Cui, Y.J. et al. (0), Investigation of interlayer soil behaviour by field monitoring. Transportation Geotechnics, Volume, Issue, pp -. Lamas-Lopez et al. (0). Assessment of the double integration method using accelerometers data for conventional railway platforms. The second international conference on railway technology research, Ajaccio, France