Вы находитесь на странице: 1из 6

Numerical Methods in Geotechnical Engineering Schweiger (ed.

) 2006 Taylor & Francis Group, London, ISBN 0-415-40822-9

Three-dimensional analyses of transition zones at railway bridges


M.E. Smith, P.-E. Bengtsson & G. Holm
Swedish Geotechnical Institute, Linkping, Sweden

ABSTRACT: Numerical analyses were performed to evaluate the behaviour of a railway bridge transition zone under the passing of trains. A numerical parametric study was performed to study the effects of the train velocity and the stiffness of the embankment materials. The parameter study was performed using the dynamic mode of FLAC3D. Results are expressed in terms of the maximum net horizontal stresses on the back of the abutment, and maximum vertical deflections behind the abutment for the case in which the train is directly above the back of the first bridge abutment. This paper provides a description of the methods used in the numerical parametric analyses, and results of the parameter study, including a discussion of the trends disclosed by the analyses.

1 INTRODUCTION Railway lines in Europe are 60 to 100 years old, and are not designed in accordance with modern railway traffic. Due to faster and heavier modern trains, existing railway bridges are experiencing problems, such as deterioration and/or loss of ballast material, and the increase of differential settlements within the bridge transition zone. These issues have an adverse effect on the safety, reliability, and economy of the railway line, and therefore, many existing bridge systems require upgrading. Engineers are faced with the task of assessing the performance of existing bridges, and, if necessary, designing the strengthening or repair systems. As part of the work performed for the EU project Sustainable Bridges, the authors have evaluated the stress distribution and deflections within the transition zone due to the passing of high speed trains. The analyses were performed using the finite difference computer code FLAC3D; train loads were applied using the dynamic mode. A bridge geometry representative of concrete bridges in Europe was considered. A numerical parametric study was performed to study the effects of the (1) train velocity, (2) stiffness of the ballast and subballast material, and (3) stiffness of the backfill/ embankment fill. This paper provides a description of the methods used in the numerical parametric analyses, and results of the parameter study, including a discussion of the trends disclosed by the analyses.

2 DESCRIPTION OF NUMERICAL ANALYSES 2.1 Previous studies by others

Olofsson & Hakami (2000) performed threedimensional analyses using FLAC3D to evaluate the interaction between the train track, bridge abutment, and backfill at the transition zone. These analyses included structural beam elements to represent the rail and sleepers, and structural shell elements to represent the bridge abutments. The train load was applied statically by advancing a force along the structural nodes of the beam elements. Adolfsson et al. (1999) present the results of FLAC3D analyses performed to evaluate the critical train speed vibration effects at a soft site in Ledsgrd, Sweden. These analyses considered a railway on a simplified soft soil section with no bridge. No structural elements were used in the analyses, and the train load was applied using the dynamic mode in FLAC3D. The analyses described by Olofsson & Hakami (2000) and Adolfsson et al. (1999) were made available to the authors and were a valuable source of information. 2.2 Numerical analyses

Numerical analyses of a simplified bridge geometry were performed using the finite different FLAC3D (Fast Langrangrian Analysis of Continua in 3 Dimensions) computer program (Itasca 2002). The

237
Copyright 2006 Taylor & Francis Group plc, London, UK

Figure 1. Cross-section of numerical model. Figure 2. Mesh geometry of numerical model. Table 1. Base case material property values used in the numerical analyses. Ballast Dry density (kg/m3) Elastic modulus (MPa) Poissons ratio Bulk modulus (MPa) Shear modulus (MPa) Friction angle (deg) Dilation angle (deg) 1900 193 0.30 161 74 40 4 Subballast 1900 160 0.30 133 42 37 3 Fill 1700 47.9 0.31 42 18.3 31 2

base case geometry was chosen based upon a typical bridge geometry provided by the Swedish Railroad Administration, Banverket, and is representative of many concrete bridge systems. The bridge embankment profile is shown in Figure 1, where it can be seen that the profile consists of an upper ballast layer 0.4 m thick, which is underlain by a layer of subballast 1.2 m thick, which in turn is underlain by embankment backfill. A stiff, non-yielding, bearing layer is assumed to exist beneath the embankment fill. As can be seen in Figure 1, the geometry is symmetrical about its centre-line. The material property values used in the numerical analyses are listed in Table 1. A linear-elastic perfectly plastic model with a Mohr-Coulomb failure criterion was used to represent the ballast, subballast, and embankment fill. No material damping was used. The mesh geometry used in the numerical analyses is shown in Figure 2. The total length of the model is 56 m. The length of the bridge is 10 m. Zone sizes are on average 0.36 0.31 0.28 m, and a total of 63,376 zones were used. In FLAC3D, the spatial element size should be smaller than approximately one-tenth to oneeighth of the wavelength associated with the highest frequency component of the input wave (Itasca 2002). For the analyses presented herein, the wavelength of the input wave is equal to 7.25 m, which corresponds to the length of the simplified load associated with one bogie. No structural elements were used to model the bridge abutments, or the rail and sleepers. It was desired to model the train dynamically, and since the

Figure 3. Approximation of the load from one bogie.

timestep in the dynamic mode is determined by the largest material stiffness, using structural elements would require unrealistically long model run times, and may cause numerical problems. Therefore, the bridge structure was represented by fixed gridpoints at the bridge-soil interface. For these analyses, a hypothetical train load was used. The embankment system is relatively stiff and in order to obtain observable behavourial patterns, a relatively high train load was used. The applied train load was calculated based upon an axle load of 250 kN. For the base cases analyses, the maximum applied train load, qmax, is approximately 72 kPa. The train load was applied dynamically along the top of the model over a width of 1.25 m from the centre-line. To gain a clear understanding of the behaviour of the system under a dynamic load, only one bogie of the train was considered. The approximated load representing one bogie is schematically shown in Figure 3. The shape of the load in Figure 3 is a simplified approximation of the load beneath two wheels of one bogie. This load was input as a cosine-wave, with a maximum magnitude of 72 kPa. The cosine-wave

238
Copyright 2006 Taylor & Francis Group plc, London, UK

Table 2. Variable material property values. E (MPa) Ballast 160 193 300 47.9 160 300 20 47.9 70 Poissons ratio 0.30 0.30 0.30 0.31 0.30 0.30 0.31 0.31 0.30

Subballast

Embankment fill

Bold values are base case values.

was applied successively over each zone. One load variable was created for each successive zone. Each load variable is a function of time, and has a duration that is equal to the length of one bogie (7.25 m) divided by the velocity of the train. The first step in the numerical analyses was to initialize the vertical and horizontal stresses so that they satisfy both equilibrium and the gravitational gradient. To avoid shear stresses from developing in the soil at the bridge abutments during the initialisation process: (1) the model profile was built up in incremental layers for a total of four layers, with each layer being equilibrated prior to adding the subsequent layer, and (2) the gridpoints at the bridge abutments were fixed in the horizontal directions only, and not in the vertical direction. Once the model was brought to initial equilibrium, all gridpoints at the bridge abutments were fixed all directions prior to applying the train load. 3 PARAMETER STUDY A parametric study was performed to study the effects of (1) the train velocity, (2) the stiffness of the ballast and subballast material, and (3) the stiffness of the embankment fill. The approach adopted for the numerical parameter studies relies on a base case analysis, with systematic variation of parameter values from the base case. The train velocity was varied over a range of 50 to 350 km/hr. The stiffness of the ballast, subballast, and embankment fill were varied systematically according to the values listed in Table 2. Results and conclusions presented in this section pertain to the behaviour of the system at the first bridge abutment when the train is directly above the back of the abutment. Results are expressed in terms of net horizontal stress, which represents the additional horizontal stress applied to the back of the structure due to the train load, and maximum vertical deflections behind the abutment.

Figure 4. Distribution of net horizontal stresses behind first bridge abutment (v 350 km/hr). (Contours given in kPa.)

A cross-section of the net horizontal stresses at the first bridge abutment is shown in Figure 4 for the case in which base case material properties were used and the train velocity was equal to 350 km/hr. A few general observations can be made based on the results shown in Figure 4. First, below a depth of 1.2 m, the net horizontal stresses in the fill are relatively small. This stress distribution pattern is not the same, however, for the case in which a soft embankment fill is used, as discussed below. Second, the maximum net horizontal stress for this case is on the order of 80 kPa, and occurs over a depth of 0.2 to 0.45 m. For all the analyses in this study, the net maximum horizontal stresses were observed in the ballast layer or at the ballast-subballast layer interface. The variation in the maximum net horizontal stresses, net h,max, with train velocity and material property values is given in Table 3. In all cases, the maximum net horizontal stresses were observed beneath the centre of the train load. The values presented in Table 3 are for the case in which the train is directly above the back of the first abutment. The variation of net horizontal stresses with train velocity and material property values is discussed in the following sections. A cross-section of the vertical deflections 1 m behind the first bridge abutment are shown in Figure 5 for the case in which base case material properties were used and the train velocity was equal to 350 km/hr. The deflections in Figure 5 are the vertical deflections calculated when the train is directly above the back of the first abutment. When the train is at this location, the maximum vertical deflections occur approximately 1 m behind the train (unless noted otherwise). For the base case analysis, the maximum vertical deflection behind the first bridge abutment is 2.87 mm (Figure 5).

239
Copyright 2006 Taylor & Francis Group plc, London, UK

Table 3. Case

Maximum net horizontal stresses at the first abutment. Net h,max (kPa) 36.6 38.5 52.1 79.4 87.3 79.2 107.4 42.6 78.5 49.2 Depth (m) 0.2 0.2 0.2 0.20.4 0.4 0.2 0.2 0.4 0.2 0.2 % Change from base case

v 50 km/hr v 150 km/hr v 250 km/hr v 350 km/hr Ballast (E 160 MPa) Ballast (E 300 MPa) Subballast (E 47.9 MPa) Subballast (E 300 MPa) Fill (E 20 MPa) Fill (E 70 MPa) Bold value is base case value.

10 neg. 35 46 neg. 38

Table 5. Maximum vertical deflections, v,max, behind the first abutment as a function of material stiffness (v 350 km/hr). E (MPa) Ballast 160 193 300 47.9 160 300 20 47.9 70 v,max (mm) 3.03 2.87 2.71 5.46 2.87 1.83 9.10 2.87 1.43 % Change 5.6 5.6 90.2 36.2 217* 50.1

Subballast

Embankment fill

Figure 5. Distribution of vertical deflections 1 m behind first bridge abutment (v 350 km/hr). Table 4. Maximum vertical deflections, v,max, behind the first abutment as a function of train velocity. Velocity (km/hr) 50 150 250 350 v,max (mm) 1.45 1.49 1.93 2.87

*Observed 3 m behind train.

3.1

Train velocity

A few general observations can be made based on the results shown in Figure 5. First, most of the vertical deflections occur in the ballast and subballast layers. Second, as would be expected, the maximum vertical deflections are concentrated beneath the width of the train. The variation in the maximum vertical deflections, v,max, with train velocity and material stiffness values are given in Tables 4 and 5, respectively, and are discussed in the following paragraphs.

The train velocity was varied over a range of 50 to 350 km/hr. The base case material property values, listed in Table 1, were not varied. In general, as the train velocity increases, the net horizontal stresses in the ballast and subballast layer increase, as can be seen in Table 3. However, the relationship between train velocity and maximum net horizontal stress on the back of the abutment is not linear. The largest incremental increase in the maximum net horizontal stresses at the first bridge abutment occurs when the train velocity increases from 250 to 350 km/hr. As can be seen in Table 4, as the train velocity increases, the maximum vertical deflection behind the train increases. The largest increase in maximum vertical deflections occurs for the case in which the train velocity increases from 250 to 350 km/hr. The goal of the Sustainable Bridges project is to evaluate the behaviour of bridge systems under high speed trains, which may have velocities up to 350 km/hr in the future. The remainder of the parametric analyses were performed using a train velocity of 350 km/hr.

240
Copyright 2006 Taylor & Francis Group plc, London, UK

Figure 6. Net horizontal stresses behind first bridge abutment (v 350 km/hr) for stiff subballast case. (Contours given in kPa.)

Figure 7. Net horizontal stresses behind first bridge abutment (v 350 km/hr) for soft fill case. (Contours given in kPa.)

3.2

Ballast stiffness

The elastic modulus of the ballast was varied over a range of 160 to 300 MPa. Changing the stiffness of ballast material has only a small effect on the net horizontal stresses and the calculated vertical deflections behind the first bridge abutment (see Tables 3 and 5, respectively). This is likely due to the fact that the thickness of the ballast is relatively thin compared to the thicknesses of the subballast and embankment fill materials. 3.3 Subballast stiffness

350 km/hr. As can be seen in Figure 6, the maximum net horizontal stresses are on the order of 40 kPa and occur at a depth of about 0.4 m. The net horizontal stresses below a depth of 1 m are relatively small. In general, as the stiffness of the subballast layer increases, vertical deflections behind the first abutment decrease. 3.4 Embankment fill

The elastic modulus of the subballast was varied over a range of 47.9 to 300 MPa. When the stiffness of the subballast layer is equal to 47.9 MPa, it is equal to the stiffness of the embankment fill, and for this case, it is as if there is no subballast layer and the embankment fill extends to the bottom of the ballast layer. When the stiffness of the subballast layer is equal to 300 MPa, the subballast is 1.6 times stiffer than the ballast layer. The stiffness of the subballast layer greatly affects the calculated maximum net horizontal stresses, which occur at the ballast-subballast interface. Of all the parameters evaluated, the greatest decrease in maximum net horizontal stress occurs for the case in which the stiffness of the subballast is greater than that of the ballast layer. This decrease in maximum net horizontal stresses is greater than the decrease that corresponds to the decrease in train velocity from 350 to 250 km/hr. A cross-section of the net horizontal stresses at the first bridge abutment are shown in Figure 6 for the case in which the stiffest (E 300 MPa) subballast layer was used and the train velocity was equal to

The elastic modulus of the embankment fill was varied over a range of 20 to 70 MPa. An increase in the stiffness of the fill corresponds to an decrease in the net horizontal stresses in the ballast and subballast layers. A cross-section of the net horizontal stresses at the first bridge abutment are shown in Figure 7 for the case in which the softest fill layer was used and the train velocity was equal to 350 km/hr. A decrease in stiffness of the embankment fill has little affect on the calculated magnitude of maximum net horizontal stress. As can be seen in Figure 7, the maximum net horizontal stress occurs at about a depth of 0.2 m and is approximately 79 kPa. However, net horizontal stresses on the order of 30 kPa occur at the subballastfill interface, and net horizontal stresses on the order of 20 kPa are observed to a depth of approximately 2.7 m. This behaviour is different than that indicated in Figure 4, in which the net horizontal stresses below a depth of 1.2 m are relatively small. The greatest vertical deflection behind the first bridge abutment was calculated for the case in which the soft embankment fill was used. This high value of calculated deflection occurs approximately 3 m behind the train.

241
Copyright 2006 Taylor & Francis Group plc, London, UK

4 LIMITATIONS The results discussed in this paper are only for the conditions investigated. One limitation of the current analyses is that structural elements were not used to represent the rail and sleepers. As a result, the true distribution of the train load through the rail and sleepers was not taken into account. Future analyses will include structural elements to represent the rail and sleepers and/or the bridge, with a statically applied train load. Another limitation of the current analyses is that a stiff bearing layer was assumed to exist beneath the embankment fill. It is expected that a soft bearing layer beneath the embankment fill would have an effect on the distribution of stresses and deflections in the system. The analyses presented in this paper constitute only a small portion of the analyses being performed for the Sustainable Bridges project. Analyses are currently underway to evaluate the effects of combinations of various realistic train loads and velocities, and multiple train passes. Future analyses will also be performed using a typical masonry arch bridge geometry. 5 SUMMARY AND CONCLUSIONS Numerical analyses were performed to evaluate the behaviour of a bridge system under the passing of trains. A numerical parametric study was performed to study the effects of (1) the train velocity, (2) the stiffness of the ballast and subballast material, and (3) the stiffness of the backfill/embankment fill. The parameter study was performed using threedimensional, dynamic analyses using the computer program FLAC3D. Results were expressed in terms of the maximum net horizontal stresses on the back of the abutment, and maximum vertical deflections behind the abutment for the case in which the train is directly above the back of the first bridge abutment. The train velocity was varied initially, and then, for the remainder of the parameter study analyses, a train velocity of 350 km/hr was used. For the analyses performed, it was observed that the maximum net horizontal stresses occur in the ballast or at the ballast-subballast interface. The magnitude of the maximum net horizontal stresses is affected the most by the stiffness of the subballast layer. However, the vertical deflections measured behind the train at the first abutment are affected the most by the stiffness of the embankment fill. Other key findings from the analyses include:

There is a substantial increase in the maximum net horizontal stresses when the train velocity increases from 250 to 350 km/hr. Net horizontal stresses in the embankment fill are relatively small, except for the case in which a soft fill is used. When the train is located directly above the back of the first bridge abutment, the maximum vertical deflections are observed to occur approximately 1 m behind the train, except for the case in which a soft fill is used. Changing the stiffness of ballast material has little effect on the net horizontal stresses and the calculated vertical deflections behind the first bridge abutment. As the stiffness of the subballast layer increases, vertical deflections behind the first abutment decrease. An increase in the stiffness of the fill corresponds to a decrease in the net horizontal stresses in the ballast and subballast layers. When a soft embankment fill is used, the magnitude of the maximum net horizontal stress is not affected; however, greater net horizontal stresses are observed at greater depths. When a soft embankment fill is used, vertical deflections increase substantially. These deflections are observed to occur 3 m behind the train.

ACKNOWLEDGEMENTS The authors acknowledge the important contributions of Alexander Smekal, Bo Andrasson and Olof Olofsson. The analyses described in this paper were performed within the EU Sixth Framework Programme; the project title is Sustainable Bridges Assessment for Future Traffic Demands and Longer Lives, TIP3CT-2003-001653.

REFERENCES
Adolfsson, K., Andrasson, B., Bengtsson, P.-E., Bodare, A., Madshus, C., Massarsch, R., Wallmark, G., & Zackrisson, P. 1999. Evaluation and Analyses of Measurements from the West Coast Line. Report. Itasca Consulting Group 2002. FLAC3D Fast Langrangian Analysis of Continua is 3 Dimensions, Itasca Consulting Group, Minnesota, Minn., USA. Olofsson, S.s-O. & Hakami, E. 2000. Interaction SprBro-Grund-Jord, Sttningar i vergng mellan bro och till fartsbank fr lanserade broar. Itasca Geomekanik, Borlnge.

As the train velocity increases, the net horizontal stresses in the ballast and subballast increase.

242
Copyright 2006 Taylor & Francis Group plc, London, UK

Вам также может понравиться