The Home Depot (THD) is a $261B publicly traded company founded in 1978 and is the largest home improvement retailer in the United States, supplying tools, construction products, and services. The Home Depot Supply Chain Delivery team is a team within the Logistics department whose focus is customer experience and determining the best ways for products to be delivered to customers. The system we are concerned with is the network of buildings being built to enable next-day deliveries to customers, relieving pressure from THD stores.
With the introduction of these distribution centers, The Home Depot needs to know how to best incorporate them into a transportation plan based on key factors such as handling and transportation costs. This problem is more challenging than conventional vehicle routing problems. Not every facility contains every SKU and a significant number of orders cannot be fulfilled entirely by one distribution center. The nature of products vary widely, from large items such as drywall to small items such as screws, and different products require different modes of transportation. There are multiple methods to get product from facilities to customers, some of which involve transfers: items may be sent from a facility to a customer, backhauled from one facility to another before being sent to the customer on a route. Our objective was to create a transportation plan minimizing cost.
We created a software to solve an optimization problem that produces a daily transportation plan. Our objective was to minimize distribution cost, accounting for transportation and handling costs. This plan indicates when it is efficient to backhaul products from a store to a distribution center when the distribution center cannot fulfill the complete order. We wrote a Mixed Integer Program which incorporates several inputs regarding order information and product/vehicle availability as well as constraints for capacity, time, and other delivery rules. The output indicates the cost of the transportation plan, which facility location to source product from, when to backhaul a product from a store, what type of transportation is needed, etc.
We delivered a daily transportation plan to the client as well as our customizable mixed integer program. We visualized our output in Tableau so that the client can easily visualize different routes and transportation alternatives in order to draw insights. This program can be run daily to output the optimal transportation plan for the following day, or it can be used periodically for analysis purposes. We also provided a tool that takes into account demographical data as well as THD's historical order profile data and predicts future order density to be used for planning purposes.
Value and Impact
By using our software or by adapting the logic from the MIP, THD will have the ability to optimize routing in their new delivery network. Our solution will give THD logic to make decisions about how their new distribution centers should be included in their delivery network to minimize cost. We found that backhauling becomes cost effective when economies of scale are fully utilized. There is no immediate gain with small orders, but as order size increases and systems become more complex, the benefit is added. Strategically, this tool allows them to plan and judge how much it may cost for them to provide different shipping time capabilities in the long run and help them to evaluate different transportation alternatives. With the option to generate random orders which are pulled from mapped distributions of their past delivery data, an advanced heuristic model which projects orders based on data not directly related to the order history, and a filter view to see orders and item demand specific to past days, The runs provide insights about how their current delivery network can improve: where specific items should be stocked (seeing if an item that is consistently backhauled may be better of being stocked at an FDC/MDC), creating a heatmap of routes frequency, how many vehicles they should anticipate to be going in and out of a facility, etc.