AI-Based Tolerance Optimization for Cost Reduction

Abstract

Manufacturing industries are under constant pressure to deliver high-quality products while minimizing production costs. Among the many influencing factors, the selection of dimensional tolerances plays a crucial yet often underestimated role. Tolerances define allowable variations in part geometry and directly impact functionality, assembly fit, and manufacturability. While tighter tolerances may improve precision and performance, they typically require advanced machining, rigorous quality control, and higher manufacturing costs. The optimization challenge lies in selecting tolerances that are tight enough to ensure assembly performance but not overly restrictive to cause unnecessary cost inflation. This is especially relevant in components like shafts and hubs that demand precise clearance or interference fits based on their functional requirements. To address this, the present study explores the use of Artificial Intelligence (AI), particularly Genetic Algorithms (GAs), as a decision-support tool for optimal tolerance selection. GAs are evolutionary algorithms inspired by natural selection and are effective in solving complex, multi-objective optimization problems. Here, the GA evaluates various combinations of International Tolerance (IT) grades for shaft and hub components, considering both cost and functional fit constraints. The methodology follows ISO 286 standards for tolerance grades and fit classifications. A Python-based simulation environment was developed to implement the GA, enabling automated generation of cost-efficient tolerance combinations that satisfy minimum clearance or interference requirements. This AI-driven approach marks a shift from experience-based to data-driven tolerance design. It provides a scalable framework that supports engineers in achieving cost-effective, functionally robust designs, ultimately enhancing manufacturing efficiency and competitiveness.