As train speeds continue to increase, the dynamics of the interface between train and infrastructure is an increasingly important factor in current collection performance. European legislation prescribes that assessment of the quality of the current collection system shall be performed according to measurements and/or simulations. Due to the increase in computer power in recent years, simulations have become increasingly attractive as an addition or replacement to real line measurements.

The goal of this work is to judge current collection quality through dynamic time-history simulations of the pantograph-catenary interface. First, the current state-of-art is reviewed, subsequently a simulation approach is determined, applied and tested to the norm EN50318:2002. Based on simulation results and comparison to measured values, the simulation approach is validated according to the norm EN50318:2002.

As a consequence of the large finite element models used in the valid model, a method for enhancing the simulation speed without losing non-linearities is determined. Changes in solver and contact model are identified as possible improvements, as is a modal reduction of the model. In order to apply the proposed solver improvements, a test model is built in Matlab. The valid Ansys model is used to test simulation times as different contact models are used. A modal reduction approach is implemented in Matlab.

It is found that as a consequence of solver changes, simulation times may improve by up to 40%. Furthermore, contact model changes may result in improvements of up to 77%. Modal reduction, when applied to the current model, has been found inefficient due to high amounts of interface DOF with respect to the full size of the model. Therefore, no results are presented for the modally reduced system.

Non-optimized simulations currently require around 24 hours of simulation time. It is to be expected that multiple simulations need to be executed in order to simulate all possible pantograph combinations, thus performing all simulations may require days to weeks. This situation is deemed undesirable. Based on the findings in this work, it is concluded that simulation times may improve by a ratio [6:43] if solver and contact model are chosen wisely. Therewith simulation times per simulation may be reduced to 3.4 hours.