This paper presents a mixed integer linear programming (MILP) mathematic model for short-term scheduling of multipurpose batch plants based on the representation of Maximum State Task Network (mSTN) for process recipe. When the model is built, the assignment of tasks and units to event points is handled through one set of binary variables and the continuous-time representation is used. This model can accommodate dedicated, multipurpose, and shared storage modes and can tackle all storage policies. In order to improve the computational efficiency, some effective techniques are developed to cut down the size of the proposed model based on the characteristics of batch plants and the model. Examples reveal the effectiveness and applicability of the model and improvement techniques.