Fixture Configuration Design Using a Computer Experiment
Abstract A new approach to fixture configuration design is developed based on the design and analysis of computer experiments (DACE). In the past, fixture configuration design undergoes three steps in an iterative manner: locator position design, clamp position design, and clamping force design. The fixture and workpiece were generally assumed to be rigid bodies. In this study, a sampled computer experiment is conducted with a finite element analysis (FEA). Surrogate models are developed based on the responses of this experiment. The purpose of building these models is to represent complex physical relations among the design variables in a simple functional form such that design analysis can be conducted efficiently. The surrogate response models can be used to provide a directional guidance in optimization, to identify feasible fixture configuration designs for flexible workpieces, and more importantly to enable a simultaneous design of locator/clamp positions and clamping forces. In the surrogate models workpiece deflections and locator reaction forces are used as variables. Distance-based Latin Hypercube Sampling Design (LHD) is used to increase the number of the sample points and to evenly spread them in the design space. Multivariate Adaptive Regression Splines (MARS) and Gaussian Stochastic Kriging (GSK) models are employed to capture nonlinear relationship among the design variables. The proposed approach is illustrated with examples.