Optimization of empty railcar distribution helps to improve the railcar service efficiency and meet the transportation demand. This paper defines empty railcar arcs and loaded railcar arcs from the perspective of real-time optimization, the revenue, cost and service capacity can be denoted on those arcs during the plan horizon. A time space service network of empty railcar distribution is present after mapping the relationship of those arcs. Based on the time space network, a multi-time point optimization model is proposed, with the objective of maximizing the total revenue of empty railcar distribution. At last, the model is verified by a real world example of east region of Guangzhou Railway (Group) Corporation. The result shows that the proposed time space service network and the model can reduce the difficult of the real world problem, and provide a new idea for empty railcar distribution.