Stochastic Optimization for Tactical Demand Allocation in Large-Scale Supply Chains

Selene Silvestri
Massachussets Institute of Technology, USA

This work introduces a novel stochastic modeling approach to solve large-scale network design problems.

We propose a multi-phase approach based on Sample Average Approximation that includes a pre-processing phase to reduce the feasible solution space and increase tractability of the problem.
We apply the methodology to a real-world case study with a world-leading pharmaceutical company to improve tactical and operational demand allocation decisions.

By explicitly accounting for complexities including multi-commodity flow and customer demand uncertainties, we present a solution approach that accurately represents real-world supply chain networks and constraints.
We demonstrate that the proposed stochastic solution strategy outperforms both the current state network and the expected value optimization.

We also explore the impact of using intra-network redeployments and on-demand transportation mode alternatives on demand allocation decisions and cost metrics. Through the analysis, we highlight the resilience of the stochastic solution to changes in network settings, as well as the benefit of incorporating flexibility in the network.

Register

Dr. Silvestri is a Research Scientist at the MIT CTL Intelligent Logistics Systems Lab. Her current research focuses on supply chain network design and inventory management. Her work is conducted in collaboration with global organizations and aims to leverage optimization methods to enhance their decision-making in supply chain management.
Dr. Silvestri received her Ph.D. in Computer Science from the University of Salerno, in Italy. During her Ph.D. she was a Visiting Student at the Interuniversity Research Center on Enterprise Networks, Logistics and Transportation (CIRRELT) in Montréal, Canada. Following her doctoral studies, Dr. Silvestri was a Postdoctoral Fellow at HEC in Montréal, Canada, affiliated with the Institute for Data Valorisation (IVADO) and CIRRELT. Before joining MIT, Dr. Silvestri worked as a Pre-Sales Technical Consultant in Optimization at FICO.