A Hybrid Stochastic Optimization and Simulation Approach for Multi-Stage Production–Procurement Planning under Uncertainty

Paulina Kus Ariningsih (Supply Chain Center, Fakultas Teknologi Rekayasa, Universitas Katolik Parahyangan, Jawa Barat, 40141, Indonesia)
Carles Sitompul (Supply Chain Center, Fakultas Teknologi Rekayasa, Universitas Katolik Parahyangan, Jawa Barat, 40141, Indonesia)
Titi Iswari (Supply Chain Center, Fakultas Teknologi Rekayasa, Universitas Katolik Parahyangan, Jawa Barat, 40141, Indonesia)
Giovano Alberto (Supply Chain Center, Fakultas Teknologi Rekayasa, Universitas Katolik Parahyangan, Jawa Barat, 40141, Indonesia)

Spektrum Industri
Vol. 24 No. 2 (2026): Spektrum Industri - October 2026

Multi-stage manufacturing firms that source raw materials through both long-lead-time import contracts and volatile spot markets must coordinate procurement and production under supply and demand uncertainty. Managing these decisions separately leads to excessive reliance on spot markets, unstable inventories, and unbuffered downside cost risk. This study aims to develop a novel integrated stochastic optimization framework that jointly plans procurement and multi-stage production for a real-world textile supply chain. The framework's novelty lies in embedding a Conditional Value-at-Risk (CVaR) measure directly into a multi-stage stochastic input–output production–procurement model, so that sourcing-channel selection and production scheduling are optimized jointly rather than sequentially. The approach combines a risk-aware CVaR procurement model with a stochastic input–output production scheduling model that captures material flow across processing stages. The two are accompanied by a Monte Carlo scenario-generation architecture that represents demand and spot-price uncertainty. System performance is evaluated across 100 stochastic scenarios over a 52-week horizon, using a dataset calibrated from an industrial textile system. Results show the framework maintains a 100% service level (zero expected backlog) across all scenarios. As the initial-inventory target rises from 1× to 2× the target Days of Inventory on Hand (DOH), total operational cost increases from about USD 2.15 million to USD 2.80 million (a 30% rise), while the 0.95-level CVaR increases from about USD 1.66 million to USD 2.17 million. This reveals an explicit, quantifiable trade-off between inventory buffering and downside cost risk. The model also shifts sourcing away from expensive spot purchasing as buffering increases. This study contributes to developing a framework that offers a practical, risk-aware decision-support tool for firms balancing import contracts against spot volatility, and demonstrates the value of synchronizing procurement and production rather than planning them in isolation.