Constraints in airport space and a limited number of Transportation Security Administration (TSA) officers combined with an increase in air travelers and a rising amount of security threats can put more pressure on TSA staff and resources. Because of this, developing new operational and workforce management tools is critical for TSA to improve their efforts of airport screening under challenging conditions.
Jorge A. Sefair, Ph.D., associate professor with the University of Florida Department of Industrial & Systems Engineering (ISE), is currently the principal investigator on a project that works closely with the TSA to improve its decision-making processes in many aspects, such as passenger volume prediction, staffing and capacity planning. The project is funded through the Center for Accelerating Operational Efficiency, a Department of Homeland Security (DHS) center of excellence.
“Our models are expected to facilitate the decision-making process for TSA, ensuring the efficient use of the resources without compromising the security level of the screening operation,” Sefair said. “The goal of TSA is to maintain (and increase) the standard of security in all travel operations while at the same time offering a positive experience to passengers reflected in short wait times. Our tools are intended to facilitate this mission.”
Sefair said working closely with TSA decision-makers to develop models that fit their needs requires the group of researchers to understand the complex operations of security checkpoints and baggage screening. This is an essential element since the models need to reflect their operation to be adopted.
“Some of the model features include movement of officers across checkpoints, design of work breaks and overtime schedules, ramp-up and ramp-down rules when opening or closing screening lanes, estimation of queue lengths at every 15-minute interval of the day, among others,” Sefair said. “The resulting challenge is the computational time required to solve some of our models. To overcome this challenge, we implemented various techniques to reduce the complexity of the mathematical model, decompose the problem into a series of subproblems that are easier to solve, and exploit the properties of an optimal operational plan.”
Sefair said an example of the research group’s work was seen at Phoenix Sky Harbor International Airport, where they developed a model for TSA to estimate the passenger arrival preferences during the COVID-19 pandemic. Since it was not possible to collect field data in person at the time, the group relied on CCTV footage and computer vision algorithms to count the number of passengers arriving at the checkpoint. The group then used those numbers to feed a statistical learning model to determine the updated arrival curve, which found that passengers preferred to arrive closer to their departure time in comparison to travel prior to the COVID-19 pandemic. Sefair said this was helpful for the TSA to adjust the passenger arrival predictions and their corresponding operational decisions (e.g., checkpoint staffing).
“Given the requirements expressed by our partners at TSA’s ITF [Innovation Task Force], we recently extended our models to capture the baggage screening operations, and we are hoping to see our tools being used nationwide” Sefair said. “There are opportunities to extend our work to other components of DHS such as the Customs and Border Protection (CBP) agency, as they face similar challenging problems at the US borders and ports of entry.”
Sefair continues to collaborate on this project with co-PI’s Ron Askin, Ph.D., from Arizona State and Eduardo Perez, Ph.D., from Texas State University. Furthermore, UF Ph.D. student Jiao Jiang is helping on the project by working on tractable approximations of queue performance metrics using data science and optimization.
Brady Budke
Marketing and Communications Specialist
Herbert Wertheim College of Engineering