Supervisory control theory provides means to synthesize supervisors from a model of the uncontrolled plant and a model of the control requirements. Currently, control engineers lack experience with using automata for this purpose, which results in low adaptation of supervisory control theory in practice. This paper presents three modeling guidelines based on experience of modeling and synthesizing supervisors of large-scale infrastructural systems. Both guidelines see the model of the plant as a collection of component models. The first guideline expresses that independent components should be modeled as asynchronous models. The second guideline expresses that physical relationships between component models can be easily expressed with extended finite automata. The third guideline expresses that the input-output perspective of the control hardware should be used as the abstraction level. The importance of the guidelines is demonstrated with examples from industrial cases.