Demand-oriented public transportation is a new way to provide mobility in both urban area and rural area. This paper aims to simulate an on-demand transportation system which will be implemented in Elde region, Germany. The work flows of sending a trip request and driving are first simulated using Java. The involved data models are explained and implemented. An ant colony algorithm is developed for the routing optimization. By applying the measures, request acceptance rate, average waiting time, vehicle occupancy and vehicle capacity occupancy, in different request arrival rate, the number of vehicles, vehicle capacities and the maximum acceptable riding time, it is found that the vehicles with four seats are most suitable for this system. Also in order to maintain a high acceptance rate more than 80%, the vehicle occupancy rate should be more than 70%. The analysis and evaluation results provide suggestions for vehicle resource assignment and system management.