In order to analyze situations of delay propagation in urban railways where trains are running densely, it is required to simulate the detailed movement of trains between stations considering the interaction between trains. At present, two simulation models are known - the macroscopic simulation model and the microscopic simulation model. The macroscopic simulation is not appropriate for this purpose because they only give the arrival and departure times and do not show the movements between stations and the interaction between signaling systems and trains. Thus, the microscopic simulation has to be used. The microscopic simulation, however, has a drawback that the simulation speed is very low. In this paper, we introduce a train traffic simulation algorithm that can simulate not only arrival and departure times of trains but their detailed movement between stations. The algorithm is designed as an extension of the longest path based macroscopic simulation algorithm so that it can deal with running times for each block section. In order to get realistic simulation results, we have introduced an idea to dynamically change the weights of arcs based on an estimation of the signal aspects. Thus, we can simulate trains' movement between stations reflecting the interaction between consecutive trains.