Centralized and decentralized supply chains: Performance maps for comparing the cost-effectiveness of distribution network configurations

Optimizing Distribution Networks (DNs) is crucial for retailers, impacting service levels and logistics costs. A key DN configuration decision is the stock deployment policy, which entails choosing between centralized, decentralized, and hybrid DNs for each Stock Keeping Unit (SKU). Choosing the stock deployment policy is complex due to many variables influencing the decision (e.g., number of customers served, SKU purchasing costs, customer demand, etc.). Moreover, this decision must be revisited whenever customer demands changes, which can be time-consuming when DN resilience is challenged by geopolitical changes, market trends, and disruptive events. Dimensional Analysis (DA), and particularly the Buckingham Theorem (BT), shows capabilities to support retailers in guiding and streamlining stock deployment decisions. After modeling the stock deployment problem in a mathematical form, BT can identify its influential variables, extract knowledge on how variables mutually interact when affecting the stock deployment performance, and aid informed decision-making on the most cost-effective policy. Accordingly, BT enables creating performance maps which compare the characteristics of different DNs and SKUs, then suggesting similar stock deployment decisions for similar (scaled) DNs and SKUs. Despite the potential utility of these performance maps, no prior study has explored BT's capabilities for stock deployment decisions. This paper bridges this gap by proposing BT to create supportive maps for multidimensional scaling, similarity analysis, and economic performance prediction across centralized, decentralized, and hybrid DNs. The resultant maps provide retailers with visual decision support tools for associating similar DNs and SKUs with optimal stock deployment policies, ultimately improving DN performance and resilience.