Optimization of industrial cutting stock problems - Production scheduling in a dynamic stochastic environment

The Cutting Stock Problem (CSP) arises when large items have to be cut into smaller items to satisfy demand. CSPs occur in a wide range of industries and an extensive body of literature exists on solving these problems. Still, methods discussed in the literature do not incorporate common industrial considerations such as due dates of orders and dynamic, stochastic production environments. Thus, existing methods are of limited use for industrial applications.;We first consider a static CSP where we integrate due dates of orders, and propose a new optimization model which integrates the generation of cutting patterns with due date restrictions on the orders. We then extend this approach to a multi-period, dynamic environment. We show the importance of considering future demand and how it can lead to considerable cost savings. This offers interesting insights into the structure of the problem, and explains why new and counterintuitive postponement strategies are effective.;Traditionally, CSPs have been considered as a pure make-to-order (MTO) or make-to-stock (MTS) production process. We propose a combined and dynamic approach. We discuss the structure of the MTS vs. MTO decision, illustrate when holding inventory is beneficial, and evaluate the cost reductions of a combined MTS and MTO production strategy. To solve the resulting decision problems, we introduce novel methods where we link stochastic programming to approximate dynamic programming, thus, improving the estimates of future costs.