A Tree-Search Heuristic for Stochastic Production-Distribution Planning with Transportation Mode-Dependent Lead Times

Explore a tree-search heuristic for stochastic production-distribution planning with transportation mode-dependent lead times in this 48-minute seminar from GERAD Research Center. Delve into the complex problem of simultaneous production and transportation decisions, including selecting optimal transportation modes for shipping goods to customers. Learn about the trade-offs between shorter lead times, higher costs, and increased flexibility in reacting to demand changes. Examine the multi-stage problem with a discrete finite time horizon and stochastic customer demand, solved using a rolling horizon framework and static-dynamic problem representation. Discover the tree-search heuristic based on Anytime Column Search and Limited Discrepancy Search, featuring a node selection strategy that aggregates scenarios for efficient solution improvement. Gain insights into decomposition of time periods, dynamic programming, branch and cut techniques, flow-based guiding heuristics, and the impact of scenario trees on stochastic solutions.

Intro
Summary of the SPDPL
One-Warehouse Multi-Retailer Problems
Production-Routing Problems
Dual Sourcing
Summary of the literature
Decomposition of the time periods.
Rolling horizon framework to take the decisions
SPDPL multi-stage static-dynamic model
Dynamic programming
Branch and cut
Flow-based guiding heuristic
Flow improvement sub-problem
Instances
Comparisons with CPLEX
Value of stochastic solutions
Scenario trees to sample randomness
Impact of the scenario tree
Generalizing the transportation modes
Preliminary experiments
Conclusion