Multi-Echelon Planning is Not Multi-Echelon Optimization

Inventory optimization is probably the most loosely applied term in service logistics. The lack of clarity about just what optimization is allows every software solution to claim to have it, when in reality they actually have some sort of multi-echelon planning capability.

Multi-echelon optimization (MEO) establishes inventory levels based on the relationships between all the locations across your network and all of the parts at the lowest cost. It considers:

  • The effective lead times between stocking locations.
  • The overall service level you’re trying to achieve.
  • Demand for all parts across your global supply chain.
  • Budget constraints.

MEO manages a network holistically, considering the millions of tradeoff decisions between which parts to stock and where to minimize overall cost. 

In contrast, multi-echelon planning creates stocking levels based upon individual locations and individual part decisions within a larger supply chain. It doesn’t consider the relationships between those locations, or how stocking parts in one location versus another will impact the cost of achieving your desired service levels.

Multi-echelon planning in practice

Let’s use a simple example. Say you have 40 locations across your service parts supply chain consisting of:

  • One global warehouse.
  • 11 regional warehouses.
  • 28 forward-stocking locations.

All a multi-echelon planning algorithm would do is consider the individual fill rates you set for each of those locations, and then create stocking plans based on those targets, ignoring the fact that demand and supply can be buffered by multiple locations when the network is considered in its entirety.

These models require that each location fills the demand that arises at the point in the supply chain, rather than relying on the broader network as a source of supply. This significantly increases the amount of inventory required for the same level of service results.

Multi-echelon optimization in practice

Given that same network, MEO considers the matrix of availability targets for part groups and location types. Instead of evaluating each location in isolation of the network that supports it, the algorithm computes the marginal contribution to level of service for each dollar of inventory “spent”, so investment is allocated to where it creates the most value. 

Where possible, inventory can be held in higher echelons where it has higher utilization, and rebalancing and pooling logic manages a dynamic network. Critical slow moving parts are assigned high levels of service to ensure that they are stocked in sufficient quantities. The net result is a set of inventory targets that achieve the service level targets at absolute minimum inventory investment.

Since the network is considered as a whole, strategic modeling of changes to the network such as opening or closing locations, rerouting physical flows, collapsing echelons, imposing budget constraints etc. can all be evaluated quickly.

Here’s the bottom line: Your supply chain isn’t a collection of isolated locations, so it shouldn’t be treated like one. Instead, it is a network of highly related locations. Multi-echelon optimization allows you to capture the value of those relationships. Multi-echelon planning does not. Which would you rather use?

If you want a more in-depth explanation of what mutli-echelon optimization does, watch the on-demand webinar hosted by my colleagues, Vinod Arekar and Steven Caldwell:

MEO Service Parts Planning Webinar replay