Methods to improve the response time of a CAPP system that generates non-linear process plans

Feature-based Computer Automated Process Planning applications often suffer from exponential growth of the search space and thus CPU time, with increasing complexity of a part. This paper will explain some methods that were used to improve the response time of a newly developed CAPP system that is capable of generating net-shaped process plans, called non-linear process plans, containing several alternatives. One method to improve the performance is called Opportunistic process planning. The approach consists of dividing the features into two sets: one set, called the important features, is used to generate a process plan; the other features, called non-important features, are added to the process plan afterwards. This generation method highly resembles the reasoning pattern of a human process planner. Another innovative method for performance improvement is called Feature grouping. This method combines features that have strong resemblance, and considers them as only one feature during the process plan generation. Other reasoning methods, described in this paper, are: combined variant/generative planning and constraint-based search. All of the above-mentioned methods allow us to reduce the search space significantly. The developments have been carried out in the framework of the Esprit project 6805, COMPLAN (Concurrent Manufacturing Planning and Shop Control for small batch production). The COMPLAN CAPP system is able to cover a wide workpiece spectrum (customizable towards specific needs). It has been intensively tested for three workpiece families (i.e. hydraulic valves, bending tool-knives and prismatic parts) of a pilot end-user, a producer of sheet metal machinery. The system can be easily extended towards other part spectra. Also the embedded process-planning knowledge is easy to configure or alter.