Flat Surface Lapping: Process Modeling in an Intelligent Environment

The process of lapping has been long considered an art due to the tremendous amount of variability and subjectivity involved. The quality of lapping differs from operator to operator and the results are highly inconsistent. The material removal rate, surface finish, and flatness all depend on the proper control of lapping parameters such as lapping pressure, lapping speed of rotation, lap ring material, weight and size, abrasive size and type, workpiece material and hardness. To attain the desired outcomes, it is imperative to select proper values for the lapping control parameters. Moving the art of lapping into a science and quantifying the results can solve many of the above problems. In this research, a portable mechanical lapping tool was designed and tested along with manual lapping. Lapping processes were studied by conducting designed experiments, literature search, and consulting experts. The results from the experiments were explored in detail using various statistical techniques to explain the relationships among potential parameters and to see the possibility of lapping model development. A preliminary intelligent computerized lapping system (advisory system) was also developed as a framework for future work. Representative qualitative models and rules for lapping were proposed based on lapping literature and lapping experts' knowledge. However, it was found that the domain knowledge obtained from different sources was often clouded by imprecision and uncertainty, and the available data of manufacturing problems were frequently imprecise and incomplete. To overcome this problem, fuzzy logic concepts were applied in developing a protocol for the knowledge-based system. This research is an initiative of well-designed experiments and data analyses in investigating potential parameters of flat surface lapping with an application on reconditioning valve discs and nozzle seats.