Case Study: Optimizing the Distribution Network for a Frozen Food Company

Background and Challenges

The company operates sustainable fish farms in Asia and Central America and sells the fish in the US and Canada. Most of the fish is frozen to withstand the long transportation and to level out the variations in demand.

The company needed to improve the cost structure to improve the margins. A new CEO was recruited and he identified the logistics cost as a key opportunity based on their high percentage of the total cost and the complexity of the distribution network paired with anecdotical indications of less than logical solutions.

The distribution structure was very complicated with 15 temporary storage locations and cross dockings, 7 main DCs, and a shipment pattern where all DCs and storage locations shipped everywhere.

Some DC/storage locations were “mandatory” to facilitate large customers pick-up structures. They could be moved but it would require negotiations with important customers, which is not a preferred activity.

Some complicating factors were the limited number of goods entry points and the limited capacity of frozen food 3PL operators.

The management of the distribution network was very manual without a TMS system or a supply chain planning and execution system. The second task was to develop efficient processes for managing the distribution network.

Structuring the project

The company lacked the tools and the experience to optimize the network. Establish, with a well-respected network optimization practice, was hired to recommend the best distribution network for both service and costs.

The study took 10 weeks and was very collaborative. By nature, a network optimization study is very data driven but it is very important to never lose focus on the implementation and ongoing operation that will follow. To achieve this it was a key success factor to have the operations staff actively involved.

Daily informal contacts to clean and validate data and requirements, discuss assumptions combined with weekly structured meetings to keep the project organized was a good formula.

The Analysis

All goods flows were mapped out and the cost and cost drivers identified.

Data was imported from ERP system and from carriers. The cleansing and validation being a major task.

The center of gravity was identified for 2 DCs and up to 9 DCs, with and without the “mandatory” locations. A complicating factor here was that the “mandatory” locations in themselves impacted the center of gravity. However, when looking at the next level locations (the large customers’ store locations) it was not changing the center of gravity much.

At the time of the study the capacity of frozen warehouses was very limited. This was included as a modeling factor through restricting eligible solutions to locations with theoretical capacity to handle the required volumes.

The existing rates for both transportation and warehousing were benchmarked to ensure that they wouldn’t skew the results when used in the model and to identify potential opportunities in negotiations. The rates held up pretty competitive due to the very specialized crew that was only focusing on reefer trucks transportation.

With all restrictions and rates in the model, the optimization was run, and the evaluation could start. The optimization was complicated due to the numerous restrictions and there had to be a lot of “what if” analysis to identify what restrictions were impacting the solution and what is practically implementable.

Based on experience from many network optimization projects, it is important to start with the mathematical optimization, but the qualitative aspect of setting the framework for the model and transforming a theoretical result into a feasible solution that will work in the real world is what matters.

The Solution

It was determined that the network could be simplified, reducing the logistics costs by an estimated 10-18%. The range is due to the fluctuations in the rates for warehousing and transportation at the time. This is done by reducing to 4 main warehouses and 2 cross docking locations so that the service levels can be maintained or improved. In the case where extra capacity would be needed, it could be obtained around the logistics hub, eliminating an overcomplex network with too many lanes.

With the new network, the administrative burden eased due to the reduced complexity. A TMS was not possible in the short-term due to an ongoing ERP implementation, so a freight management company was recommended to manage the transportation.