Iterative beam search for car sequencing

The car sequencing problem seeks a production sequence of different car models launched down a mixed-model assembly line. The models can be distinguished by selected options, e.g., sun roof yes/no. For every option, car sequencing applies a so-called sequencing rule to avoid that consecutive models requiring this option lead to a work overload of the respective assembly operators. The aim is to find a sequence with minimum number of sequencing rule violations. This paper presents a graph representation of the problem and develops an exact solution approach based on iterative beam search. Furthermore, existing lower bounds are improved and applied. The experimental results reveal, that our solution approach is superior compared to the currently best known exact solution procedure. Our algorithm can even be applied as an efficient heuristic on problems of real-world size with up to 400 cars, where it shows competitive results compared to the current best known solutions.