A Data-driven Approach to Enhance Worker Productivity by Optimizing Facility Layout

The facility layout problem (FLP) is the problem of determining non-overlapping positions of departments on the shop floor to minimize material handling costs. Traditional methods for solving FLPs consider pairwise (from-to) flows to optimize layouts. This paper shows that these traditional methods underestimate the total travel distance of a layout when departments have more than a single input/output point and some flows consist of visits to more than two departments. To accurately calculate the traveled distances, the actual routes of the workers and transporters (so-called connected movements) in the system need to be determined. The connected movements of the workers in a facility can now be captured using the Internet of Things network and stored in the cloud server for analysis. We propose a mixed-integer non-linear programming model for the FLP that minimizes the total travel distance using these connected movements as the input data. Because of the complexity of the problem, a biased random key genetic algorithm is used to find the layout. To ensure the validity of the method, a case study is carried out at a fertilizer production company that implemented an Internet of Things network to capture worker movement data to minimize worker productivity loss via an improved layout. By using these connected movements, the best layout for the case company is found. The results of the proposed data-driven optimization method indicate that leveraging connected movements can reduce the total travel distance by 10.6% compared to the best possible layout generated by the traditional pairwise method in the case study