This paper presents a novel method based on mixed integer linear programming (MILP) that addresses the optimization of water network in batch chemical processes with single contaminant to minimize wastewater discharge. In this method, a batch cycle is divided into several time intervals and a unit would be divided into several sub-units, if necessary. It ensures that each unit/sub-unit operates in only one time interval. Each unit/sub-unit is attached with a buffer tank to relax the time constraint. The sequence of units is pre-determined in ascending order of their outlet concentration, so that units and their buffer tanks only provide water to afterward units. A superstructure of units, buffer tanks, and water-using connections is developed. The corresponding mathematical model is a MILP problem that guarantees the global optimum and obtains a solution quickly. Heuristic rules are used to analyze and omit the redundant buffer tanks. The proposed method is applied to a case study in this paper. The results show that the method is effective.