We present a mixed-integer linear programming model for minimizing the makespan of batch processes occurring in the chemical industry. The formulation of the model is based on the starting and processing times of the batches without using an external uniform time grid. A special form for the objective function is chosen and additional redundant constraints are added in order to speed up the solution process. A comparison with the classical uniform time discretization model shows that in many cases the size of the new model as well as its solution time is smaller. Computational results on several instances are reported. The modular structure of the model allows a combination with other continuous-time models for batch processing problems.