1. ABSTRACT

One of the most daunting challenges in Computer Integrated Manufacturing (CIM) is bridging the gap between Computer Aided Design (CAD) and Computer Aided Manufacturing (CAM). Many specific approaches have been developed for limited manufacturing domains. For example, a solid model may be directly converted into Numerical Control (NC) code for a computerized milling machine. But, when we try to extend the planning to cover a number of domains, there are no satisfactory techniques. This still leaves many unresolved problems when converting a product from design features to manufacturing features.

Computer Aided Process Planning (CAPP) is the umbrella term for automated approaches to determining production process parameters from a design. This thesis has a description of the various approaches and methods used when constructing these systems. For the most part, the existing research has focussed on limited production domains, while there is a definite need for a system that can combine a large number of different manufacturing processes. Such an approach is presented in this thesis.

This approach begins with a design stored in sets that are linked together with Boolean equations. The information in this representation can be collected using modern solid modeling systems. The primary difference is that current CAD systems perform operations and then discard the information provided by the user. When this information is kept, it provides a powerful tool for reasoning about manufacturing a product.

The design is loaded into a hierarchical, non-linear planning system. The planner first looks for assemblies and reused features. Each of the features and assemblies are examined separately. In a first pass, the equation for a part is examined for known forms in the equation using templates. If a template is matched to part of an equation, then corresponding production rules are examined. These rules are of the form ‘if <conditions> then <actions>’. The planner will find a number of alternatives for each operation, and try to plan from beginning to end. If for some reason the planner is not able to complete the plan, it will backtrack and try another alternative. If the planner is unable to find a satisfactory plan, then it will simplify the Boolean equation and begin planning again.

The planner’s capabilities are demonstrated through the use of examples. The results show that the planner is able to plan with mixed production technologies, select plans on the basis of cost, generate alternative operations, deal with new technologies, and recover from failures. The thesis is concluded with a discussion of the method in general, including some of the future developments, some of the shortcomings, and some of the outstanding research questions for the method.