Strengthening your supply chain one link at a time.
The power of network modeling rests with the ability to answer increasingly complex supply chain questions. Some of these questions create inherent challenges with respect to network modeling and optimization. I’ve been a network modeler for over 15 years and one practice I’ve found useful is to bifurcate the model to enhance network strategy optimization. Supply chain topics which promote multiple model techniques are discussed below.
Peak vs Non-Peak Periods
Many supply chains have facilities constrained during peak periods which can skew optimization results. Typically, a network model represents one year of customer demand and the related logistics costs. Therefore, plant and warehouse capacities are portrayed as an annual throughput constraint. In constrained networks, it is useful to look at those peak vs non-peak periods to better utilize all available capacity and determine meaningful customer assignments. The non-peak model would therefore represent an optimal network solution with respect to cost and customer service areas. By default, the peak model (due to capacity constraints) is less optimal, but helps plan for ramp ups in capacity at lesser utilized facilities and potential customer assignment changes during those periods.
Plant Direct vs Distribution Center Shipping
In some supply chain networks, the use of plant-based shipping is desirable if logistically possible. Optimally, you would ship products made at a given plant and ordered in sufficient quantities (enough volume to fill a truck) by a customer from that plant. Splitting plant direct shipments from the overall network allows for cleaner customer assignments for the volumes shipping through distribution centers. In a combined model customer single sourcing becomes more difficult since a customer may receive plant direct shipments from multiple plants as well as shipments from a distribution center. Cleaner customer service areas and isolation of plant direct cost vs the cost to serve through a distribution center drive the desirability of using multiple models. A variation of this analysis would separate TL vs LTL customer volumes whereby TL volumes typically ship from locations closer to source points while LTL volumes typically originate closer to customer.
Fast vs Slow Moving SKUs (Velocity)
Companies with larger SKU counts may explore separate networks for fast moving SKUs vs slow moving SKUs. The motivation to do this may be to reduce inventory, alleviate demand planning complexity or improve customer service. I’ve seen the desire to create a network with fast moving SKUs in regional distribution centers (improve customer service) while slow moving SKUs are more centralized (reduce inventory and demand planning complexity). As in the case of plant direct shipping, customer single sourcing becomes complicated since a customer may get shipments from a fast mover distribution center and a slow mover distribution center.
The use of multiple models to enhance supply chain optimization is a practice that I’ve grown accustomed to utilizing. Great care must be taken to separate customer demand, plant/vendor capacity, distribution center capacity and any pre-flows that are required. The benefits of using two models (flexibility to address relevant questions and cleaner results) outweigh the additional build time.
—Dan Gunter, St. Onge Company
