Evaluation of supply chain or workshop management is often based on simulation. This simulation task needs models which are difficult to design. The aim of this work is to reduce the complexity of simulation model design and to partially automate this task by combining discrete and continuous approaches in order to construct more efficient and reduced model. Model design focuses on bottlenecks with a discrete approach according to the theory of constraints. The remaining of the workshop is modeled in a less precise way by using continuous model in order to describe only how the bottlenecks are fed. This used continuous model is a regression tree algorithm. For validation, this approach is applied to the modeling of a sawmill workshop and the results are compared with results obtained previously by using a neural network model.