Process optimization is often performed using process simulators with limited optimization functionality or algebraic modelling languages, which require time consuming model formulation and implementation. Here a hybrid approach is chosen as an alternative, which uses advantages of both methods: MATLAB as an optimization platform is coupled with the process simulator UniSim Design. The approach is evaluated using the example of the industrial methanol synthesis. For the optimization, gradient-based and derivative-free algorithms are applied and thereby existing work in this field is extended reasonably.