SupplyChainDigest Editorial Staff

Paper in the most recent Journal of Business Logistics by Ohio State research Keely Croxton and Walter Zinn argues that the process and tools for optimize logistics network design have not really accounted for the costs of inventory in the solution – focusing instead on warehousing and transportation costs. Inventory requirements were then typically overlaid on that optimal network design.

The paper is full of the usual academic muckety-muck (not as bad as many), but there is some useful thinking here – SCDigest will take the suffering for our readers and relay the key insights between the endless citations to other research and impossible to understand tables.

First, the authors argue that both mathematical models and software tools to determine optimal network designs have not well accounted for both traditional logistics costs and inventory. The authors notes that: “Software companies have also recognized the limitation. Insight, Inc., specializing in network design software, and Optiant, specializing in inventory deployment software, have recently formed an alliance to link their systems so that the two problems can be solved in an iterative, but still serial, fashion.”

The researcher propose a new set of algorithms for including inventory in network design, and test their theory on the network of a mid-size retailer (150 stores). There’s a lot of math, but here’s the summary:

• Including inventory does reduce total system/network costs, and will generally do so while either keeping the total number of warehouses in the network the same, or reducing the number. In the retailer analysis, the number of warehouses in the design would be reduced from four to two, and total network costs reduced by $4.5 million dollars, or about 10%.
• Obviously, the impact of the savings is greater depending on the value of the inventory.
• There is an especially strong impact from including inventory for C items, or items with highly variable demand.
• With growing use of third party providers, reducing or eliminating many of the fixed costs of physical networks, it becomes even more important to include inventory in the optimization. In these cases, “If inventory cost is included in network design, the network will increasingly be determined by the trade-off between transportation and inventory cost. Consequently, managers who reduce the fixed warehousing cost and fail to include inventory cost in network design will adopt a solution that has too many warehouses and is unnecessarily costly.”
  
  

There’s a lot more math for our APICS readers to wade through. Obviously, many companies do consider inventory in network design, which is one of the drivers behind the trend in some companies for larger, fewer DCs. Nevertheless, it seems Croxton and Zimm have applied some useful mathematical rigor to optimizing the decision.

We can’t provide a link, but send an email at the Feedback button and we will send you a copy of the paper.

Is the cost of inventory well considered in optimal network design? Have we lacked the tools to do so effectively? Let us know your thoughts.