Solving the one-dimensional cutting stock problem under discrete, uncertain, time-varying demands using a hybrid of special-purpose Benders' decomposition and column generation

This study shows the integration of Benders' decomposition, column generation, and a special-purpose algorithm for solving the one-dimensional cutting stock problem under discrete, uncertain, time-varying demands. The results show that there is a linear relationship between the processing time and the number of scenarios involved in the raw material cutting patterns. Moreover, the algorithm results are not different from using column generation. During the search for production patterns, the quality of the solutions relies on the convergence tolerance. As the tolerance approaches zero, the solutions become near-optimal, but the computational time increases substantially. In addition, the experimental results indicate that the proposed method is suitable for large-scale problems.