The dynamic interaction of pantograph and catenary in electrified railway systems has considerable research attention, both in terms of simulation and of measurements. The simulation benchmark of 2015 proved a good base of agreement between established models, but also showed that deviations between the different models increase when the simulated system approaches critical speeds. From these deviations it can be concluded that some of the models in the benchmark better depict the system behaviour around critical speed than others. Therefore, this paper analyses and compares measurement data from the Green Train project in Sweden, which includes speed at and over critical level, to simulations performed with one of the participating models, the finite element (FE) based CaPaSIM model. Based on this comparison, modelling alternatives are implemented and tested in CaPaSIM, to examine if they improve the accuracy of the model in the speed range around critical speeds. These alternative modelling approaches are to a large extent based on previous studies in the field, such as other models that participated in the benchmark. The varied aspects include choice of element type for single components of the catenary and element mesh resolution. On the pantograph side, the effect of separately modelling independent flexible collector strips is studied. The modelling choices that prove to be most promising during the initial variation tests are presented and recommendations are given for further investigations of possible improvements.