Sampling-based Tolerance-Cost Optimization: The Key to Optimal Tolerance Allocation

Abstract

Limiting manufacturing-caused part variations by size, location, orientation, and form tolerances primarily aims to assure the total assembly quality. At the same time, however, the manufacturing conditions and, thus, the manufacturing costs are already predefined in the product development phase. The method of sampling-based tolerance-cost optimization, a combination of statistical tolerance analysis based on sampling techniques and metaheuristic optimization algorithms, enables an automated and optimal allocation of tolerance values and, thus, solves the conflict of objectives between costs and quality. However, limitations in effectiveness and efficiency still prevent its profitable application for solving complex, industry-relevant problems and exploiting hidden cost potentials. To close the current research gaps, the individual methods involved, in particular the sampling, non-conformance rate estimation and surrogate model-based optimization, are (further) developed and harmonized in one common approach, ensuring that reliable optimization results can be obtained in adequate computing times. Its extension to simultaneous machine selection and allocation with different batch sizes and selective assembly, considering machine-specific part tolerance distributions and geometrical, mutually dependent tolerances, significantly expands the context of use to practical aspects. A final evaluation of the developed framework proves its potential for a profitable application to practical problems and serves to identify further research potentials.