The ability to answer prediction questions is crucial to reasoning about physical systems. A prediction question poses a hypothetical scenario and asks for the resulting behavior of variables of interest. Prediction questions can be answered by simulating a model of the scenario. An appropriate system boundary, which separates aspects of the scenario that must be modeled from those that can be ignored, is critical to achieving a simple yet adequate model. This paper presents an efficient algorithm for system boundary selection, it shows the important role played by the model’s time scale, and it provides a separate algorithm for selecting this time scale. Both algorithms have been implemented in a compositional modeling program called TRIPEL and evaluated in the plant physiology domain.