Robust car sequencing for automotive assembly

Abstract Just-in-sequence material supply is the status quo in the automotive industry. In this process, the assembly sequence of vehicles is set several days prior to production, and communicated to the suppliers. The committed sequence is essential for efficient operations both at the original equipment manufacturer and its suppliers. In practice, however, sequence stability is insufficient. Short-term disruptions, such as quality problems and missing parts, put the sequence at risk. If a disruption occurs, the affected vehicle is removed from the sequence. The resulting gap is closed by bringing the succeeding vehicles forward. Such sequence alterations, however, cause workload changes and potentially work overloads at the assembly stations. As a remedial measure, additional sequence alterations are necessary, which further disturb material supply. Robustness against short-term sequence alterations is currently a key objective of automotive manufacturers. In this paper, we propose a sequencing approach that includes the vehicles’ failure probabilities in order to generate robust sequences. Robust sequences are sequences that can be operated without modifications, even when vehicles fail. We develop a branch-and-bound algorithm that optimally solves small-sized instances. For large-sized instances, we design a sampling-based adaptive large neighborhood search heuristic. The superiority of our approach is validated in a simulation study using real-world data from a major European manufacturer. We find reductions in the expected work overloads of 72% and 80%, compared to the industry solution, and compared to an approach taken from literature which does not take failures into account.