Preface to special issue on parallel computing and co-simulation in railway research

Due to the high cost of physical testing and the increasing acceptance of computer simulation modelling, railway research has become more and more reliant on computer simulations to advance the state-of-the-art. With the collective efforts of researchers from all around the world, computer models and simulation techniques in railway research have been developed to advanced levels incorporating unprecedented amounts of detail. For example, three-dimensional wheel-rail contact models, traction models with consideration of gear transmission and mechatronic control, discrete element ballast models and multibody draft gear models are now available. However, these advanced models are often developed from the perspective of different research interests and by various researchers using different programming tools. In real-world railway engineering, computing models that require multiple detailed model components are sometimes needed. There are also motivations from a theoretical research perspective to integrate multiple detailed model components as such a combination does provide more accurate simulation results. In these cases, there is a research question about how to connect legacy models that were developed by different researchers and often using different programming tools so as to avoid developing new models from scratch. The reason is obvious as new developments are time consuming. Under these circumstances, co-simulation is an effective approach to resolve the question. Co-simulations can be achieved in various ways, such as shared memory and communication protocols. Also, the co-simulations can be achieved among different software packages no matter whether the packages are in-house or commercial.