Poster
Client Context
The team partnered with Saia, the 9th largest less-than-truckload carrier in the United States. Less-than-truckload companies allow for customers to purchase a fraction of space on a truck rather than a full truck, and use terminals as intermediary points for shipments rather than shipping directly from origin to destination. Terminals have two primary types of operations: linehaul, or the shipping of freight to and from other terminals, and city driver, or the pickup and delivery of freight to customers. Saia operates on a network consisting of 175 terminals across the United States, along with parts of Canada and Mexico, and consequently handles roughly 30,000 shipments per day across its network. Saia grows its market share by adding terminals to its network, a decision that involves a multi-million dollar capital expansion. Historically, Saia has added 3-5 terminals a year; however, due to increased profitability in recent years, Saia is now targeting 10-15 terminal additions this year.
Project Objective
The addition of even one terminal can cause unprecedented wide scale changes in the network, and it is difficult to predict cost and revenue implications on the network as a result. This project focuses on unintended cost and freight flow consequences. For example, a new terminal that is too large can lead to underutilization of this terminal and a terminal that is placed in an inopportune location can also lead to unanticipated transportation costs. Other times, unpredictability in freight flow changes resulting from terminal expansion can lead to unintended consequences. Currently, it is difficult to assess these potential impacts of a new terminal on freight flows and costs within Saia's network without first building this terminal. Then, to provide Saia with visibility into a new terminal's impact on cost and freight flow, the team has developed a system model. The system model will aid Saia in estimating city driver costs and linehaul driver costs as well as routing freight for a new network. Respectively, these are the costs associated with pickups and drop-offs at end customers, costs associated with transportation between terminals, and optimal freight routes from origin terminal to destination terminal. The system model will provide Saia with data-driven insights regarding terminal expansion prior to building a terminal.
Design Strategy
City driver costs, namely labor, fuel, and maintenance, have been estimated using a Traveling Salesperson Problem approximation that calculates the lengths of city driver routes for all terminals in a network. Load planning, the decision of which terminals to send shipments between and how many shipments and trucks to send between terminals, for a network with the addition of a new terminal has been approached via a mixed-integer program that optimizes routes to minimize network-wide costs, comprised of driver wages, fuel costs, maintenance costs, and terminal handling costs. The optimization problem outputs which terminals to send freight and trucks between, along with how much freight and trucks to send between these terminals. These outputs can be studied to view the effects of a new terminal on Saia's network prior to the construction of this terminal. The Traveling Salesperson Problem approximation was validated by looking at zip-code service assignments and number of drivers required at each terminal. The load plan mixed-integer program was validated by looking at which terminals are connected in the model versus in Saia's network, and differences in terminal capacity driven by the model's chosen routes and Saia's operated routes. Through study of these validation metrics, the team has been able to confirm that the model handles this network optimization problem in a reasonable manner.
Deliverables
The deliverable is a system model that provides the cost-related outputs of city driver and linehaul operations with the addition of a new terminal to Saia's existing network, along with a visualization for linehaul operations. Using a Graphical User Interface (GUI), Saia provides details about the new terminal and can provide cost-related inputs if needed - this is the only touchpoint with the model that Saia needs to make. After these inputs are passed in, the TSP approximation is used to provide daily city driver costs at each terminal and the mixed-integer program provides freight and truck flows between terminals. This allows Saia to understand how their network will be impacted by terminal expansion on the cost-side and make proactive decisions based on the results. The outputs of the system model are visible in Excel files. For the load plan problem, the outputs have also been visualized through the matplotlib package in Python. Terminals, the routes between them, and density of freight movement on these routes are the primary components of the visualization. This model is intended to use each time Saia is considering a new terminal - the model takes roughly twenty-four hours to run, which is reasonable given the multimillion dollar decision of building a new terminal.
Project Information
Student Team
Joey Abi-Sarkis, Abhishek Mattipalli, Maya Menon, Jay Patel, Santhosh Saravanan, Abhinav Sehgal, Pooja Sharma, Yashovarman Singh
Faculty Advisor
Faculty Evaluator
