Press Release

Profit-driven network redesign through value-creation services

A model that shifts the focus onto revenue generation over cost reduction opens new opportunities in decision-making

Editor’s Note: The SCM thesis Profit-Driven Network Redesign Through Value-Creation Services was authored by Morgan DeHaan and Yujia Ke and supervised by Dr. Milena Janjevic (mjanjevi@mit.edu). For more information on the research, please contact the thesis supervisor.

Network design is a key strategic decision in supply chain management. Traditional network design concentrates on cost savings. However, can we utilize these facilities to create profit, given that there are value-creation services that facilities can undertake? In our capstone project, we worked with a logistics service provider to restaurant chains across the United States. The company envisions a redesign of its service network based on two components: (1) the addition of new flexible distribution centers (referred to as iDCs) located closer to high-volume demand areas and (2) the value-creating services offered from the iDCs to restaurants beyond deliveries (such as inventory reserves, food preparation, and reverse materials handling).

With the context of network redesign with iDCs and focus on profits, we aimed to give a comprehensive technical approach to answer three questions: (1) how to identify urban clusters in the current demand areas, (2) how to infer/impute missing transportation cost data, and (3) how to determine the best locations of future iDCs.

An overall methodology to redesign networks

Our problems of identifying demand-dense urban clusters and allocating iDCs to high impact locations was addressed through a hybrid approach.

To identify urban areas, we analyzed demand distribution at the ZIP code level and refined three different clustering algorithms to accommodate our case, ultimately proceeding with a revised k-means solution. The result of clustering was a set of potential iDC locations serving as the discrete candidates in the facility location models in the final step.

We then explored our cost data sets and proposed an imputation method combining k nearest neighbors (KNN) and linear regression to fill missing transportation cost data. KNN was primarily used in supplier-to-facility legs where we did not know exact locations of suppliers, while linear regression was used for the facility-to-customer legs to find correlation between existing costs and distances.

Lastly, we formulated the facility location models. We developed a multi-commodity cost minimization model as our baseline, which combined p-median and set covering building blocks. The objective comprised of transportation costs of both middle and last mile. Building on that model, developed a profit maximization model. The distinction from the former model is the addition of potential revenues of value-creation services, so total profit (revenues less transportation costs) became the new objective. To capture uncertainties, we proposed a stochastic optimization model incorporating best, average, and worst scenarios of revenues.

Results and conclusions

In identifying urban areas, we decided on the result of refined k-means algorithm due to its good interpretability and categorization in the dataset.

By running the two facility location models, the cost-based model yields an output reflective of cost to deliver, while the profit-based model produces an interesting output reflecting high revenue-generating regions. We observe shifts of iDC locations from the southeastern United States to the northeastern and southwestern parts of the country. The iDC number in the Southwest decreases from 59 to 55 out of 100 possible outcomes in the latter model, and the number in the Northeast increases accordingly. The finding is justified by the profitability data (input) of value-creation service, where the Northeast, Southwest, and Northwest are projected to be the most profitable regions.

However, the numbers did not change as much as we expected. A major factor is the relatively few traditional distribution centers in the current network, located in the Southeast. Thus, the profits realized by adding one iDC to the Northeast are not as significant as the costs saved by adding it to the Southeast. That said, it may be compelling to explore how opening additional traditional distribution centers in the Southeast would impact the iDC location results.

Contributing to the industry, our primary finding relates to a profit-driven network design model incorporating both costs and revenues. It creatively uses revenues to dictate location choices. This is a key opportunity for supply chain strategy because it shifts the focus from cost reduction to prioritizing revenue generating factors in decision making. Furthermore, the proposed holistic approach is also generalizable to network designs in other industries.

Supply Chain Management Review

Every year, approximately 80 students in the MIT Center for Transportation & Logistics’s (MIT CTL) Master of Supply Chain Management (SCM) program complete approximately 45 one-year research projects.

These students are early-career business professionals from multiple countries, with two to 10 years of experience in the industry. Most of the research projects are chosen, sponsored by, and carried out in collaboration with multinational corporations. Joint teams that include MIT SCM students and MIT CTL faculty work on real-world problems. In this series, they summarize a selection of the latest SCM research.