Train timetable is the key document to regulate railway traffic through sequencing train movements to keep the appropriate order. Timetable stability and on-schedule rate are closely related. Delays caused by disturbances in train operations can be absorbed by a high quality timetable with high stability, and the on-schedule rate then can be assured. This paper improves the stability of timetables of several connected railway sections to assure the on-schedule rate with a simulation method. Firstly, we build a macroscopic network model of train operation in a railway network using the Petri net theory. Then we design the train tracking subnet model, the station subnet model and arrival-departure track subnet model. At last we propose a computing case, simulating the train operation process based on the presented models, and the simulation results prove the feasibility and availability of the models. The approach presented in this paper can offer valuable decision-support information for railway operators preparing train timetables.