Modeling Performance of a Production Line and Optimizing Its Efficiency by Means of Genetic Algorithm

Abstract

A production line is a fundament of modern high scale FMCG industry. The performance of the line depends on various factors, out of which breakdowns, cleanings and changeovers play the most important role. The paper describes the idea of modeling production line performance by its decomposition into discrete subsystems. Every machine or workstation together with preceding buffer constitute a single subsystem, which is characterized by statistical distributions of time to repair, time between failures, processing speed and capacity. Time dedicated for cleaning and changing format parts between different production batches is also considered in the model. Subsystems are connected with each other by conveyors. The model was simulated by the given time step. In order to verify the simulation results, the data from the real production line were compared and used for adjusting the parameters of the model. The described specimen consisted of six workstations connected with conveyors. There was one high capacity buffer between the second and third station. The efficiency of the whole line as well breakdown time characterizing every machine was captured by data acquisition system. Based on the given data, the parameters of statistical distributions of time to repair and time between failures were estimated by approximation to known distributions. In addition, statistical distributions of cleaning and changeover time were derived in order to provide general performance of the production line. Genetic algorithm was introduced to optimize the line parameters in order to achieve higher efficiency and to identify potential bottlenecks.