Picking the Right Supply Chain Design for a Growing Food Business

Picking the Right Supply Chain Design for a Growing Food Business

SCM thesis researchers developed a model for perishable foods distribution

Designing supply routes during periods of rapid growth when market conditions can change rapidly can be a major challenge for retailers. SCM program graduates Cyril Khamsi and Veronica Stolear developed a model for evaluating the cost trade-offs between multiple supply network designs under a variety of business scenarios.

The research was carried out in collaboration with a major retailer for the SCM master’s thesis Efficient Supply Chain Design for Highly-Perishable Foods, and was supervised by Dr. Chris Caplice, Executive Director, MIT CTL.

Significant growth in the retailer’s nascent fresh food business, which includes high-perishable items such as sandwiches, salads, and parfaits shipped in totes, was expected. The existing system – receiving freshly made product from a supplier at a distribution center where the items were consolidated with non-perishable items and sent to stores – worked well with low volumes. But the retailer needed to know whether this model would perform as well when business volumes increases by threefold or more.

For example, when volumes picked up would consolidation continue to make sense? Also, would shipping directly to stores increase freshness, and at what cost, and should the supplier be co-located with the distribution center?

The researchers created four candidate supply routes to move fresh foods from a supplier in the US northeast to regional stores: (1) dedicated supplier-to-store, (2) supplier and distribution center co-located for consolidated shipping, (3) supplier to city cross dock for van routing, and (4) distribution center to supplier for pickup while on route to stores.

Utilizing approximation methods, a light-weight model was created to estimate the costs associated with each candidate design, including the current route.

The analysis validated the current network cost to be about $3.15 per tote. The model was then applied across all candidate route designs. It showed that delivery costs under existing demand would range from $1.75-$7.03 per tote, narrowing to $1.62-$2.68 as demand increases to 5x current levels. Co-location was consistently the least costly option, offering estimated savings of $120k per year under current demand, and increasing to $360k per year as demand increases. Aside from co-location, the current design remains the most cost-effective until 3.8x demand levels, when it is matched by Zone Skip and Direct-to-Store scenarios.

The model was also used to test the sensitivity of each networks’ costs to a variety of assumptions about future business conditions. The analysis generated important additional insights into route designs.

  • Demand must increase before dedicated fresh food networks are viable due to the lack of economies of scale with shipping fresh food as a stand-alone product.
  • Tote fill-rates significantly impact costs (can drive savings ranging from 17% to 43%)
  • Vehicle selection is an important driver of cost and delivery time, allowing for savings of 15%-34%, or yielding savings of six hours for most network designs.
  • Production policies influence freshness. A 12-hour shift in production policy will accelerate delivery to customers by 24 hours at an approximated cost of $0.01-$0.02 per item.
  • Delivery interval shifts from 2 days to 3 days with a maximum 12-hour shelf-life extension, offers cost savings to offset the premium of more rapid deliveries.

These insights help the retailer to understand key delivery cost drivers for its nascent fresh food business, and to refine its strategy for investing in its distribution network. Moreover, the same methodology can be used by other retailers that face similar delivery challenges.

This post is based on an article published by Supply Chain Management Review.

For more information on the research contact Dr. Chris Caplice at: caplice@mit.edu.




Sorry, comments are closed for this post.