Optimisation of factory floor layout in a complex manufacturing process

This contribution describes the methods used in a factory floor optimization process. The manufacturing process involves approximately 140 machines and is concentrated around a group of multipurpose CNC machines. We have analysed the current state of the manufacturing system by constructing a discrete event simulation (DES) model. The factory manufactures approximately 30,000 different products and half-finished products using a corresponding number of different technical processes. Due to this complexity it is not feasible to manually construct a simulation model that would include every process, and a method for automated model construction was developed. The resulting DES model reads the data on open work orders, technical procedures and bills of materials from the company database. The model is used to develop optimisation methods and verify modifications needed for process optimisation. Final part of the contribution presents the optimisation problem handled in the project: minimisation of the total distance the products need to travel between the machines. By introducing some simplifications, the problem can be described as a quadratic assignment problem. We present a heuristic method that is based on force-directed graph drawing algorithms. Key-Words: Optimisation of processes, heuristic optimisation, discrete event simulation, factory layout, automated model construction, force-directed graphs