work zone, simulation, logistic regression


Traffic safety and mobility of roadway work zones have been considered to be one of the major concerns in highway traffic safety and operations in Florida. In intent to expose Florida's work zones crash characteristics, the Florida Traffic Crash Records Database for years 2002, 2003 and 2004 were explored. Statistical models were estimated and Florida's work zone crash traits for single vehicle crashes and two-vehicle crashes were drawn. For the single-vehicle crashes, trucks were found more likely to be involved in single vehicle crashes in freeway work zones compared to freeways without work zones. Straight level roadways are significantly affected by the presence of work zones. The lighting condition is also one of the risk factors associated with work zone single-vehicle crashes. In fact, at work areas with poor or no lighting during dark conditions, motor vehicles are more prone for crashes compared to non-work zone locations with poor or no lighting during dark. The weather condition is positively associated with single-vehicle work zone crashes. Results showed that during rainy weather, drivers are less likely to be involved in work zone crashes compared to the same weather conditions in non-work zone locations. This fact may be due to the vigilant driving pattern during rain at work zones. For the two-vehicle work zone crashes, results showed that drivers younger than 25 years of age and drivers older than 75 years old have the highest risk to be the at-fault driver in a work zone crash. Male drivers have significantly higher risk than female drivers to be the at-fault driver. The model conspicuously shows that drivers under the influence of narcotics/alcohol are more likely to cause crashes (i.e. at-fault driver) at work zones. Road geometry and the lighting condition were significant risk factors associated with two-vehicle work zone crashes. Freeways straight segments are more susceptible to crashes in work zone areas. Poor lighting or no lighting at all during dark can lead to significantly higher crash hazard at work zones. Foggy weather causes a significant mount in work zone crash risk compared to non-work zone locations. In addition to that, work zones located in rural areas have higher crash potential than work zones located in urban areas. After examining the current Florida work zone Maintenance of Traffic (MOT) plans, known as the Motorist Awareness System (MAS), it was realized that this system is static hence does not react to changing traffic conditions. An ITS-based dynamic lane management system, known as dynamic lane merging system, was explored to supplement the existing MAS plans. Two forms of dynamic lane management were recognized as dynamic lane merging namely the early merge and the late merge. These two systems were designed to advise drivers on definite merging locations. Previously deployed dynamic lane merging systems comprise several Portable Changeable Message Signs (PCMS) and traffic sensors. The addition of multiple PCMSs to the current MAS plans may encumber the latter and usually requires relatively extensive equipment installation and relocation which could be inefficient for short term movable work zones. Therefore, two Simplified Dynamic Lane Merging Systems (SDLMS) were designed, deployed, and tested on Florida's short term movables work zones. The first SDLMS was a simplified dynamic early merge system (early SDLMS) and the second SDLMS was a simplified dynamic late merge system (late SDLMS). Both SDLMS consisted of supplementing the MAS plans used in Florida work zones with an ITS-based lane management system. From the two-to-one work zone configuration (first site), it was noted that the ratio of the work zone throughput at the onset of congestion over the demand volume was significantly the highest for the early SDLMS compared to the MAS and late SDLMS. Travel time through the work was the lowest for the early SDLMS, followed by the late SDLMS, and then MAS. However, the differences in mean travel times were not statistically significant. It was also concluded that the early SDLMS resulted in higher early merging compared to the MAS and that the late SDLMS in higher late merging compared to the MAS. The first site was used as a pilot for testing the system since data collection was limited to two days for each MOT type. Hence, operational measures of effectiveness (MOEs) could not be evaluated under different demand volumes. It should also be noted that the RTMS was not available during the MAS data collection which disabled us from collecting speed data. From the three-to-two work zone configuration site, data was collected extensively relative to the first site. The RTMS was available for all three MOT types tested which enabled the collection of the speed data that are used as a safety surrogate measure. The mean speed fluctuation in the closed lane was the highest under the MAS system for all demand volumes and in all three lanes. Comparing the dynamic early merge and the dynamic late merge mean speed fluctuations in the closed lane and the middle lane, results showed that the mean speed fluctuation for the early merge are lower than those of the late merge under all demand volumes. However, the difference in the mean speed fluctuation is only statistically significant under demand volume ranging between 1 and 500 veh/hr. As for the shoulder lane, it was noted that the speed mean speed fluctuation is significantly the lowest for demand volumes ranging between 1500 veh/hr and 2000 veh/hr under the late SDLMS compared to the early SDLMS and the MAS. The ratio of the throughput over demand volume was taken as the operational MOE. Results showed that the Dynamic early merge performs significantly better than the regular MAS under demand volume ranging between 500 veh/hr and 2000 veh/hr. Results also showed that the dynamic late merge perform better than the MAS under volumes ranging between 1500 veh/hr and 2000 veh/hr and significantly poorer than the MAS under low volumes. Therefore, the late SDLMS is not recommended for implementation under low volumes. Results also showed that the late SDLMS performs better than the early SDLMS under higher volume (ranging between 1500 veh/hr to 2000 veh/hr). A simulated work zone with a two-to-one lane closure configuration was coded in VISSIM and operational and safety MOEs under MAS, early SDLMS, and late SDLMS were compared under different drivers' adherence rate to the merging instructions, truck percentage in the traffic composition, and traffic demand volumes. Results indicated that throughputs are higher in general under the early SDLMS, travel times are lower under the early SDLMS. However, overall, the early SDLMS resulted in the highest speed variance among MOT types. The MAS resulted in the lowest speed variances overall.


If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at

Graduation Date



Radwan, Essam A.


Doctor of Philosophy (Ph.D.)


College of Engineering and Computer Science


Civil and Environmental Engineering

Degree Program

Civil Engineering








Release Date

July 2014

Length of Campus-only Access


Access Status

Doctoral Dissertation (Open Access)