Stochastic Dynamic Programming Framework for Multi-Echelon Supply Chain Resilience Optimization Under Compound Demand Uncertainty

Abstract

Contemporary multi-echelon supply chains exhibit compounding vulnerabilities to demand-side stochasticity and systemic disruption cascades, rendering conventional linear optimization paradigms structurally insufficient. This study develops a stochastic dynamic programming (SDP) framework integrating Markov decision processes with robust scenario-tree decomposition to optimize resilience allocation across distributed supply network nodes. Employing a mixed-integer nonlinear programming (MINLP) formulation calibrated against panel data from 214 manufacturing firms across six economies (2019–2023), the model yields Pareto-optimal trade-off frontiers between operational cost minimization and disruption-recovery velocity. Empirical validation demonstrates a 23.4% improvement in mean-time-to-recovery metrics relative to deterministic baseline models. The findings offer prescriptive decision support for supply chain executives navigating post-pandemic structural volatility, geopolitical fragmentation, and nearshoring reconfiguration pressures.