A Monte Carlo-based approach to demand forecasting and stochastic optimization in supply chains

Economic shipment planning under demand uncertainty is still the major problem in today’s supply chains, in whichtransportation cost, inventory holding cost, and stockout cost are considered simultaneously. This work contributes by comparing demand forecasting-based planning with stochastic optimization, where the latter is enhanced with Monte Carlo simulation, and closes withthe introduction of a Monte Carlo-Enhanced Route and Inventory Optimization (MC-RIO) approach as a suitable planning option in uncertain logistics environments. The probabilistic MC-RIO framework superimposes by routing and inventory decisions the probabilistic generation ofdemand, allowing for considering concurrently shipment routing and inventory decisions in the face of a finite set of demand possibilities. Numerical tests are based on 1000 Monte Carlo scenarios and consider perishable products, capacity-limitedvehicles and penalty-oriented stockouts. Results show that while MC-RIO achieves a total expected logistics cost of $350 in the baselinecase, demand forecasting with deterministic approximation and classical stochastic programming delivers $330 and $340, respectively. We also observe that the performance of forecasting-based planning degrades dramatically as demand becomesmore variable, resulting in penalties 36% higher for stockouts. Instead, in high-variability scenarios, the MC-RIO approach is more robust, achieving a reduction in stockout costs on the order of 40–55%, at the cost of moderateincrement in holding stocks. Sensitivity analysis also shows that variability in demand and the cost of holding inventory are the most significant factors, and the total expected cost rises from $350 to $450 forhigh demand volatility and from $350 to $470 for doubled holding cost scenarios. The results corroborate that our proposed MC-RIO framework can achieve a balanced and risk-aware shipment policy that is better than the conventionalones in the presence of volatility. In summary, this work suggests Monte Carlo-based optimization as a robust and scalable decision support tool for cost-effectiveshipment planning in the face of uncertainty.