A multistage stochastic programming model for the strategic supply chain design

Publication TypeConference Paper
Year of Publication2018
AuthorsDaniel Ramón-Lumbierres; F.-Javier Heredia; Robert Gimeno Feu; Julio Consola; Román Buil Giné
Conference Name23th International Symposium on Mathematical Programming
Conference Date01-06/07/2018
Conference LocationBordeaux
EditorMathematical Optimization Society (MOS)
Type of Workcontributed presentation
Key Wordsresearch; supply chain; postponment; multistage stochastic programming
AbstractSupply chain management has been widely developed through the evolution of manufacturing, distribution, forecasting and customer behavior, encouraging the introduction of postponement strategies in its various forms. At these strategies, semi-finished goods are stored in certain operations of the chain, called decoupling points, waiting for the placement of demand orders, which trigger production flows from decoupling points to the remainder operations. Such a design problem facing the speculation/postponement paradigm must intrinsically include elements that "unveil" demand orders when they are placed, that is, the modelling approach should keep demand orders as random variables until their placement, when they are disclosed. This work proposes a multi-stage stochastic programming model that decides the optimal allocation of decoupling points, as well as a process selection among alternative designs for any general supply chain case, where the stochastic parameters, demands by period and product, are represented through a scenario tree, which is in turn generated using the forecasting. Both a risk-neutral model and a risk-aversion approach with stochastic dominance constraints are presented and solved with multi-stage instances of test cases based on real manufacturing problems defined in collaboration with the Accenture consultancy company.
URLClick Here
ExportTagged XML BibTex