Göran Johansson

Simulation of train-turnout interaction and plastic deformation of rail profiles

Railway turnouts (switches and crossings) require more maintenance than other parts of the railway network. Multiple wheel–rail contacts are common, and impact loads with large magnitudes are generated when the conventional wheel–rail contact conditions are disturbed at various locations along the turnout. The dynamic interaction between train and turnout is simulated in order to predict the forces and creepages in the wheel–rail contacts, and the sizes and locations of the contact patches. Furthermore, the change in rail profile because of plastic deformation is calculated by finite element analysis at a selected position along the switch rail. Contact loads and contact locations, taken from the vehicle dynamics simulation, are then used as input data in the finite element analysis. The objective of the study is to gain knowledge about the influence of different damage mechanisms on the life of a turnout. This is useful in an optimization of turnout geometry with the purpose to improve vehicle ride dynamics and to decrease maintenance costs.

On the modeling of large ratcheting strains with large time increments

Purpose - This paper aims to speed up finite element analyses of structures with a highly nonlinear material response subjected to many loading cycles. Design/methodology/approach - An approach where large time increments are taken in an adaptive fashion is presented. The size of the large time increments typically spans several loading cycles and is based on Taylor series expansions of the response combined with error control. Findings - The suggested adaptive algorithm is simple compared with some well-known alternatives in the literature. It also has the inherent convergence property that it reduces to the classical time incrementation in the case where the estimated error is too large. Research limitations/implications - The algorithm is suitable for (restricted to) a special class of problems where the material response versus a representative time sequence are smooth curves. The simplicity of the method results in a robust algorithm. Originality/value - Similar algorithms have been presented earlier in the literature but the present work introduces some enhancements, e.g. accounting for general internal variables also in the error estimate. In addition, the present work considers a more complex constitutive model compared with earlier work within the research field.