A central component of the design of complex artifacts is the representation and use of the physical principles behind each artifact’s behavior. Armed with a vast body of knowledge ranging from abstract rules of thumb to precise numerical models, design engineers guide and simplify their analyses through selective attention, approximation, and abstraction. During the course of analysis, they often shift between different perspectives and simplifying assumptions, seeking those views which make needed distinctions most apparent. Selecting the perspective and level of detail appropriate for each task is crucial. Analyzing every aspect of an artifact using the most accurate models available is generally prohibitive, even for simple artifacts. More importantly, parsimonious descriptions of structure and behavior enhance the designer’s abihty to identify the most important and relevant elements of behavior and to determine the most useful parts of the design to modify. Today, most of the knowledge of physical principles and how to use it resides with the engineer. Our goal is to capture most of this information in a computable form and automate as much of the modehng process as possible.