This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: Before a stick of deodorant reaches a consumer's bathroom counter, it moves through a complex supply chain. Raw materials and packaging components are sourced, assembled into a finished product and shipped through warehouses and distribution centers before finally reaching a store shelf.
At each step, companies must decide how much extra inventory to keep. A product can be available, in stock and ready to ship, yet still be out of reach when customers need it most. In today's interconnected supply chains, even a small disruption can cascade across networks, leading to empty shelves or production delays.
"Companies can have plenty of inventory and still run out of products because they are in the wrong place," explained Andre Calmon, associate professor of operations management at Georgia Tech's Scheller College of Business. "Our research helps companies decide where to keep extra inventory so they can meet demand without tying up unnecessary money in products sitting on shelves." Calmon and his co-authors, Philippe Blaettchen, Georgina Hall and Mohit Tawarmalani, received the 2026 INFORMS Computing Society Harvey J. Greenberg Research Award for their paper, "From Trees to Treewidth: Inventory Management in Complex Supply Chain Networks." The paper is published in the journal Management Science.
For decades, major companies like Procter & Gamble, Intel, HP and Boston Scientific have used a planning framework called the Guaranteed Service Model (GSM) to decide where inventory should be stored across a supply chain. The goal is simple: keep products available when customers need them while minimizing inventory costs. Calmon and his co-authors developed a new mathematical method that makes those decisions much faster.
"Our main contribution is a fundamental theoretical advance," Calmon said. "We developed the first exact method for the Guaranteed Service Model whose computing time is tied to a measure of network structure called treewidth." While the underlying mathematics are complex, the core idea is surprisingly intuitive. Rather than analyzing an entire supply chain at once, the method breaks a large network into smaller, overlapping sections.
Researchers can solve those sections more efficiently and then combine them into a single optimal plan. As a result, the structure of a supply chain can matter more than its overall size when determining how quickly the problem can be solved. The result is a faster way to identify the best inventory strategy.
Extract — continue reading at the source.
