Skip to main navigation Skip to search Skip to main content

Probabilistically calibrated robust optimization for multi-period inventory control under demand uncertainty

Research output: Contribution to journal › Article › peer-review

Abstract

Uncertainty is a fundamental challenge in multi-period inventory management, particularly when reliable probability distribution information is unavailable. Robust optimization provides a tractable alternative to stochastic programming but often results in overly conservative solutions due to heuristic calibration of uncertainty sets. This paper develops a probabilistically calibrated robust optimization framework for a multi-period inventory control problem with uncertain demand by integrating a priori probabilistic bounds into robust counterpart formulations based on box and ellipsoidal uncertainty sets. The proposed approach explicitly links uncertainty set size to allowable levels of constraint violation, providing a clear probabilistic interpretation of robustness without requiring full knowledge of the demand distribution. Several probability bounds from the literature are incorporated, covering bounded and unbounded uncertainty as well as different levels of distributional information. Computational experiments examine the effects of uncertainty set geometry, probability bounds, and violation levels on solution conservatism and performance, showing that probabilistically calibrated robust counterparts can substantially reduce conservatism while preserving tractability, and that ellipsoidal uncertainty sets generally yield less conservative solutions than box sets under comparable feasibility guarantees. The results offer practical guidance for selecting uncertainty set parameters and probability bounds in robust multi-period inventory planning.

Original languageEnglish
Article number100849
JournalResults in Control and Optimization
Volume25
DOIs
StatePublished - Dec 2026

Bibliographical note

Publisher Copyright:
© 2026 The Author(s)

Keywords

  • Conservatism control
  • Demand uncertainty
  • Multi-period inventory control
  • Probabilistic guarantees
  • Robust optimization
  • Uncertainty sets

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Modeling and Simulation
  • Control and Optimization
  • Artificial Intelligence
  • Applied Mathematics

Fingerprint

Dive into the research topics of 'Probabilistically calibrated robust optimization for multi-period inventory control under demand uncertainty'. Together they form a unique fingerprint.

Cite this