This paper describes a modelling method that reduces the effort required to develop a decision support system for scheduling in food industry, based on constraint satisfaction techniques. Most food manufacturers schedule manually. Part of the information used by human planners is hard to formalise. In the literature on detailed scheduling and in software libraries for building scheduling systems the recipe concept is lacking, while in semi-process industry and in data-models of Enterprise Resource Planning modules for process industry the recipe concept plays a central role. We argue that the addition of a recipe class reduces the formalisation or modelling effort. The recipe class proposed not only stores information closely related to the information captured in a general recipe in food industry but also generates activities and constraints necessary to produce a schedule. We implemented and tested a recipe-based model in a prototype for scheduling support in a salad factory.