The objective of this research is to gain insight into automating engineering reasoning about physical systems through the construction and utilization of a qualitative model addressing the domain of fatigue and fracture in metals. Engineering problem solving requires an understanding of how physical systems behave even when only incomplete and nonnumeric information is available on the parameters which influence that behavior. For this reason, qualitative physics is one of the critical areas requiring key advances to support intelligent computer aided engineering. Engineers make extensive use of abstraction and simplifying assumptions in construction and manipulation of mental and analytic models. Formalization of modeling assumptions and construction of a multilayered qualitative domain model capable of supporting reasoning at various levels of abstraction and under various sets of assumptions are goals of this project. The domain of application is that of fatigue and fracture in metals. This aspect of material behavior has wide engineering applicability. The envisioned qualitative model encompasses a range of physical phenomena within the domain such as high cycle fatigue and low cycle fatigue. More specific models can then be automatically generated by setting initial assumptions. After the qualitative model has been built, its capabilities for qualitative simulation and comparative analysis will be studied. The end product will be an intensive qualitative domain model and a systematic understanding of its predictive behavior.
