Precise mathematical models are not always available for some process units which are too complicated to be understood clearly in mechanism and variable relationship. These process units are difficult to be analyzed and simulated in conventional approaches. A rule based approach is proposed in this paper to model the fuzzy units. Conversion between fuzzy and crisp numbers at the input and output of fuzzy units is discussed in details and a slicing approach for conversion is proposed A case study shows that this approach provides a way to incorporate fuzzy models into conventional process simulators.