Field validation of a three-dimensional dynamic track-subgrade interaction model

Abstract

An efficient three-dimensional dynamic track-subgrade interaction model has been formulated and then validated by field investigations at various field and traffic conditions including the effect of different train speeds and types of trains. The model contains a two-dimensional discrete support track model and three-dimensional computation-efficient finite element soil subgrade model. In the two-dimensional track model, the rail beam is modelled as an Euler-Bernoulli beam. The two-dimensional track model discretizes the tie and ballast as rigid bodies with designated spacing. The three-dimensional finite element subgrade model is simulated by plane-stress quadrilateral finite elements. The longitudinal direction of the subgrade model is expanded in the frequency domain and is assumed to be homogeneous. Therefore, the computing time could be largely reduced. A moving dynamic loading is applied on top of the rail. The model is capable of taking train speed variations and the profile change of the cross section into consideration. Multiple field instrumentation tests covering the two train types and different train speeds at the test site were then conducted to verify the accuracy of the dynamic track-subgrade interaction model. Testing site is located on the Amtrak's highest speed line (Northeast Corridor: 250 km/h) near Kingston, Rhode Island in the United States. A method to obtain the tie deflection from accelerometer data at Kingston was proposed and then validated at another site on the Northeast Corridor. Tie deflections measured in the field were compared with those predicted by the three-dimensional dynamic track-subgrade interaction model. It is concluded that this model can predict track performance accurately for the Kingston site.