Poster
Client Context
Dell Logistics is a subdivision of Dell Technologies, a multinational computer technology company based out of the United States. This project is specifically focused on carrier selection for the Less-Than-Truckload (LTL) bulk, ground orders for Dell’s North America operations. The current system selects the cheapest carrier by transportation freight cost at an order by order level. However, Dell Logistics is incurring increased costs due to damaged shipments and late deliveries. Dell would like a carrier selection model that incorporates the effects of damages and late deliveries in addition to the transportation cost.
Project Objective
The primary motivation of this project is to lower the COD (Cost of Dissatisfaction), which is currently $4.3M from damages annually, while maintaining the target OTP (On-Time Performance) of 99.5% orders on-time. The objective is to create a better carrier selection process for Dell Logistics through a TCO (Total Cost of Ownership) system that incorporates both COD and OTP, metrics that Dell currently tracks but does not use. Given details for an order, this TCO system will advise on the best carrier to use for each specific shipment. The new process will decrease the frequency and magnitude of carrier-related damages, reflected in the COD, while also sustaining carrier OTP metrics, by ensuring that all carrier decisions take into account historical performance. This objective is based on Dell Logistics’ current carrier selection system, TMS (Transportation Management System), which only takes into account transportation freight costs, so there is a major opportunity in creating a new data-driven model.
Design Strategy
The Total Cost of Ownership can be separated into three components: Freight Cost, Cost of Dissatisfaction, and On-Time Performance. The freight cost is a deterministic value which can be calculated using the rate cards provided by Dell Logistics. To capture the behavior of the two non-deterministic aspects, COD and OTP, both Linear Regression and Logistic Regression Models are employed to generate predicted values for COD and likelihood of late delivery on an order by order basis. For each new order, the TCO incurred by each carrier can be evaluated using the values combined and will ultimately offer Dell Logistics data-driven recommendations per order on the carrier with the best performance.
Deliverables
The creation of this new system will serve as a proof of concept that can improve the carrier selection process for Dell Logistics by potentially reducing logistics-related costs through COD and improving or maintaining OTP performance of each carrier. Dell will be provided with an order-level script and visualization of the models, an aggregate dashboard with historical trends, and a user manual documenting methods and future steps. The TCO Model could contribute hundreds of thousands in savings through the reduction of COD while allowing Dell Logistics to keep their carriers accountable.
Project Information
Student Team
Sean Chua, Sahas Mehta, Shashvat Parikh, Geoffrey Thomas, Eric Wang, David Wu, Yichen Yuan
Faculty Advisor
Faculty Evaluator
