Order release planning with predictive lead times: a machine learning approach

An essential task in manufacturing planning and control is to determine when to release orders to the shop floor. One key parameter is the lead time which is the planned time that elapses between the release of an order and its completion. Lead times are normally determined based on the observed time orders previously took to traverse the production system (flow times). Traditional order release models assume static lead times, although it has been shown that they should be set dynamically to reflect the dynamics of the system. Therefore, we present a flow time estimation procedure to set lead times dynamically using an artificial neural network. Additionally, we implement a safety lead time to incorporate the underlying cost ratio between finished inventory holding and backorder costs in the order release model. We test our proposed approach using a simulation model of a three-stage make-to-order flow-shop and compare the forecast accuracy and the cost performance to other forecast-based order release models from the literature. We show that our proposed model using artificial neural networks outperforms the other tested approaches, especially for scenarios with high utilisation and high variability in processing times.