We describe a five-step knowledge-based method for the analysis of reaction networks. We have implemented a prototype system that realises this method. The system contains well-known concepts from the field of chemical engineering and reactor engineering and introduces a number of new concepts. The method derives a temperature and a back-mixing strategy for the reactor to optimise conversion and selectivity. We demonstrate its applicability with four different examples. We argue that this method is better than current approaches such as the geometric or attainable region approach and optimisation of complex reactor networks, mainly because the system allows inspection of its reasoning and of the knowledge used to reach its conclusions and because the decisions are presented in a form that is suitable for reactor selection. The presented method is part of a more comprehensive knowledge-based system (KBS) that does the complete reactor selection task. The prototype system is implemented in prolog.