Optimization of the production process in an automotive spring industry

Abstract

In this thesis, the Cutting Stock Problem is studied, considering three approaches set in an automotive spring industry. The first two approaches deal with the optimization of the process of one-dimensional bar cutting, minimizing material losses and inventory costs in Integrated Lot Sizing and Cutting Stock Problems. Two mathematical models and their respective solution methods based on column generation were proposed. The first refers to short term decisions, considering parallel machines and matters relative to the items and final products. The second model focuses on the medium term, considering the purchase of objects as one of the decision variables, besides demand, inventory costs and limits for objects, items, and final products. The third approach considers assigning items to the hardening furnace as a Cutting Stock Problem. The proposed mathematical model is based on an arc flow formulation. Results with real data show that the three mathematical models obtained significantly superior solutions in viable computational time, in comparison to the company's practice. Tests with random instances were performed allowing for an analysis of the influence of several parameters of these problems. Generally, small instances with large items present better results, with reduced gaps and computational times.