Many asset-intensive organizations stock Maintenance, Repair and Operations (MRO) inventory at multiple locations to cater to geographically-distributed demand. Where there are common items used at the different locations, it is advantageous for them to share this common inventory for greater efficiency. In doing so, they set up a network of interacting nodes that can be optimized to obtain additional inventory savings. Achieving these savings, however, requires a network-centric analysis to optimize the stocking decisions at each of the nodes, such that the *global* inventory in the *network as a whole* is optimal with respect to defined management objectives. Such an analysis is significantly more complex than that for a stand-alone location because the analysis must consider the interactions between the nodes, the stochastic nature of MRO inventory demand and supply as well as practical operational constraints.

An inventory network can assume different topologies as defined by the relationships between its nodes. A simple topology might consist of a set of independently-supplied stores that are in the same region, i.e. geographically close enough, so that they can share common stock items (Fig 1). A more complex topology is the so-called multi-echelon arrangement consisting of multiple *echelons* of nodes with a supplier-consumer relationship between them, i.e. the nodes at any particular echelon supply the nodes at the next lower echelon. Fig 2 shows an example of a two-echelon network where the Regional Distribution Centers (RDCs) in echelon 1 supply the stores in echelon 2, which then supply the end users. While the stores are primarily supplied by the RDCs, they also share inventory between them as required.

Fig 1: Multiple Stores in a Single Region with Shared Inventory

Fig 2: Two-Echelon Inventory Network

For any given network topology, typical inventory optimization objectives are either to minimize **Total Cost = (Stock-Out + Holding + Purchasing)** or to minimize **Total Inventory** subject to a minimum end-user service-level constraint. This is achieved by calculating and implementing the optimal Min/Max levels at each node.

One simple approach to this problem is to ignore the network interactions between the nodes and optimize each one as a stand-alone location. Unfortunately, while it is simple, this approach produces a sub-optimal solution with respect to the putative management objectives for the very reason that it ignores the nodal interactions.

An intermediate approach is to aggregate the demand of nearby nodes, compute optimal Min/Max levels for this aggregate demand and then allocate the resultant inventory among the aggregated nodes in some appropriate fashion.

A comprehensive approach would be to employ discrete-event simulation to model the dynamics of the inventory network and combine the simulation model with an appropriate optimization algorithm to determine the optimal nodal Min/Max levels with respect to the defined management objectives. The advantage of employing a simulation model is that it has the flexibility to cater to the myriad practical complexities and constraints of an inventory network, and thus eliminate the need to make restrictive simplifying assumptions. On the other hand, the simulation-model approach is also computationally-intensive, and therefore requires greater computing power to execute within a reasonable time frame.

One way to mitigate the computational demands of the simulation-model approach is to restrict the number of items that are optimized using the network-centric approach. Not all items are suitable candidates for network-centric optimization. Criticality, unit price, usage volume, lead time and usage-location count are some of the considerations in determining an item’s suitability for network-centric optimization.

Additional practical considerations include the transportation costs in moving the shared inventory from one location to another and the administrative overhead of the network-centric management approach.

Distributed, multi-location MRO inventory optimization is significantly more complex and challenging than single-location inventory management. However, with the application of the appropriate expertise and analytical tools, it also offers asset-intensive organizations an opportunity to gain additional savings on their MRO inventory.

Check out our webinar on the 12 Best Practices of Inventory Optimization to learn more about the 12 critical elements that must absolutely be a part of any successful MRO inventory optimization initiative.