Optimal distribution (re)planning in a centralized multi-stage supply network under conditions of the ripple effect and structure dynamics

The first contribution of this study is multi-objective problem formulation that opens additional perspectives for decision-making beyond cost-oriented optimization in the supply chain as a non-stationary dynamic system along with a linear programming (LP) model.

Abstract

In this paper, an original approach to formulate and solve a multi-period and multi-commodity distribution (re)planning problem for a multi-stage centralized upstream network with structure dynamics considerations is proposed. First original idea of this study is description of the supply chain as a non-stationary dynamic system along with a linear programming (LP) model. This allows distribute design and control variables between dynamic and static models. Second original idea is to transit from the classical LP model to maximal flow problem by excluding demand constraint from the LP model. The first contribution of this study is multi-objective problem formulation that opens additional perspectives for decision-making beyond cost-oriented optimization. Second, the maximal flow LP model allows the finding of a feasible solution even for unbalanced supply and demand cases without relaxing hard capacity constraints. Third, this allows improve service level at the strategic inventory holding point. Fourth, structure dynamics and ripple effect can be taken into account. Structure dynamics allows considering different execution scenarios and developing suggestions on replanning in the case of disturbances. The graph of structural reliability allows identify the optimistic and pessimistic scenarios. These scenarios are used for computational experiments with the developed model and the industrial models. With the developed model, the practical issues of scenario-based risk identification strategy and operational distribution planning can be interlinked.