Traffic Engineering

Marguerite Zarrillo

Dr. Marguerite Zarrillo performs research in the area of Surface Transportation Engineering with special expertise in Intelligent Transportation Systems, ITS. She designs computer simulation models of traffic flow, including interruptive flow models and queuing models at toll collection facilities and intersections.   One project developed models to compute the maximum number of vehicles possibly processed in one hour at a toll facility given plaza characteristics and traffic characteristics. The models were applied to the OOCEA's (Orlando Orange County Expressway Authority's) network of toll roads in Orange County, Florida. The project was funded by a grant from CATSS (Center for Advanced Transportation Systems Simulation), a center affiliated with the University of Central Florida and supported through the USDOT. Results of her work were presented at TRB 2002 and were published in the TRB RECORDS #1781. Additional papers concerning this work were presented at ASCE 7th AATTE Conference in Boston on August 4, 2002.

Another CATSS funded project developed a Decision Support System, DSS, for Choosing the Best Toll Plaza Lane Configuration. TNCC, Toll Network Capacity Calculator, was developed by Ali Jaffrey, a mechanical engineering graduate student. SHAKER was developed by Daniel Schmitt, a physics graduate student. Each of these models predict the maximum number of vehicles possibly processed through a toll plaza of a specific lane configuration pattern. They both were calibrated using real processing rate data taken at plazas on the OOCEA's Toll Network of Highways in Orange County, Florida. SHAKER is a model based on the physics motion equations and the probability of finding ETC (electronic toll collection) vehicles in varying queuing situations, and although TNCC is a model based on optimization techniques, a completely different methodology, it predicts toll plaza capacity within 3% of SHAKER's predictions.

The DSS is an interactive webpage displaying the OOCEA's network as maps using the ArcGIS/ArcIMS software. The network was divided into approximately 300 highway segments and the capacity for each segment was computed using the HCM (Highway Capacity Manual) and using SHAKER or TNCC for the 20 segments that contained on-road toll plazas. Traffic hourly volumes on all segments were taken from data collected by PB&J on a typical workday, November 5, 2002, and compared to the computed capacity. Segments that were bottlenecked, in other words, segments that had smaller capacities than their approaching traffic volumes, were displayed as red segments on the web maps. Segments that were near bottlenecks, or segments that had approaching traffic volumes that were 90% of the capacity, became orange. Segments that were potential bottlenecks, in which its capacity was exceeded by the capacity of the upstream adjacent segment, became yellow. All other segments were displayed as green segments. This was done for 16 different hours of the day, so that there were 16 maps in total.

In addition, DSS users were allowed to play with the stored data in the Oracle database. By varying traffic approach volumes to the highway segments, for example, users could determine which segments and which portion of the network were vunerable to bottlenecks. Users may also vary parameters such as highway segment characteristics, the number of lanes, or the traffic characteristics such as the percent-trucks, or the toll plazas' characteristics such as the lane configuration pattern. By playing with on the DSS, users could predict trouble spots on the network, and they could decide how best to alleviate bottleneck situations on the computer before implementation. In this way, the DSS is a support system. As an example, when toll lanes are configured such that there are more lanes available to ETC users than necessary and thus not enough lanes devoted to manual toll collection, the plaza can become a bottleneck. DSS users can reconfigure toll plazas to determine performance and bottleneck status under different lane configurations.

Two papers were published as a result of the research on the DSS. One was entitled "Providing Bottleneck Status on a Network of Highways via a Decision Support System, DSS, Website". It was presented at TRB's Annual Meeting 2006. Another one was entitled "DSS and GUI, Decision Support System and Graphical User Interface to Assist in Choosing Appropriate Lane Configurations at Toll Facilities". It is presented at TRB's Annual Meeting 2005 and published in the TRB RECORDS 1935, 2006.

Some of Dr. Zarrillo's newest research efforts are in the area of simulation. By simulating the task of driving on a laptop, subjects can be tested for driving performance under various additional activities such as using a cell phone, reading a changeable message sign or eating a donut. Another simulation task is done by using the sophisticated software, PARAMICS. Construction is underway using PARAMICS of the roadways in and around UMass Dartmouth include Ring Road and its parking facilities, Old Westport Road and Route 6. Traffic volume data is patterned around the class schedules and is input to the PARAMICS software. Evacuation strategies are also going to be studied.