Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Civil, Construction, and Environmental Engineering


Civil Engineering

First Advisor

Shauna Hallmark


Work zones provide challenging and hazardous conditions not only for vehicle drivers, but also for highway workers who are injured or killed by errant vehicles. Over, 96,000 work zone crashes occurred in 2015 which equates to a work zone crash every 5.4 minutes. Several factors have been noted as contributing to work zone crashes. Driver factors have not been as well studied as other factors. The main objective of this study was to evaluate safety in work zones utilizing the SHRP2, Roadway Information Database (511 data) to identify potential work-zones. The study looked into the effectiveness of each temporary traffic control device on drivers’ change in speed on four-lane and multi-lane divided highways. The study also evaluated drivers’ lane change behavior on freeways with lane closure. The research team manually coded the locations of work zone features starting from first work-zone sign to the end of work-zone. The change in speed from a point upstream of the legibility distance of each work zone feature was compared to the speed just past the feature. Driver distraction and eye glance were also included. A linear mixed effects model was used to predict drivers’ change in speed in the work zone. For work zones on four-lane divided highways, speed feedback signs, lane end sign, and changeable message signs were found to be effective in reducing driver speed before the merge point. Non-forward related glance was seen to increase driver speed inside the work zone. Work zone speed limit signs were seen to be more effective within half mile inside a work zone. Presence of static work zone signs were more effective when the cones were placed as channelizing device inside the work zone. Vertical panels as channelizing device were used to decrease driver speed more effectively compared to concrete and cones. The change in speed model for multilane work zones showed static work zone signs to be effective in the upstream portion of start of taper of a work zone. Work zone speed limit signs are effective when placed within half mile upstream. Lane end signs are effective in all the sections of locations in the downstream model. Drivers reduced speed due to presence of any worker or equipment inside the work zone. Driver’s lane change behavior in work zones with lane closure on four-lane divided work zones were analyzed. It was seen that with presence of rear accommodating vehicle in the open lane, the drivers tend to merge early in a lane closure scenario in a four-lane divided (farther from work zone activity area). Similarly, presence of enforcement sign before merging, tends to increase distance of lane merge from the end of taper showing that the drivers merge early in a lane closure scenario in a four-lane divided. Non-forward related glance was associated with drivers merging early in a lane closure scenario in a four-lane divided. The study also showed that driver moving over to left from right lane closure were choosing to merge early than when they were moving from right lane closure to left lane. This phenomenon cannot be fully justified as the sample size of this study was small. Head to head configuration was associated with drivers merging late. Influence of distraction and cell phone use was seen on drivers’ lane change behavior. When drivers were distracted, the arrowhead CMS sign was not seen to be effective, meaning that the drivers did not choose to merge early in work zones lane closure scenario. Similarly, when they were distracted by cell phone, the normal speed limit signs were not effective to influence the drivers to merge early for a lane closure ahead scenario. Several different analyses were conducted in order to evaluate the data from different perspectives. The different models had different response variables (i.e. change in speed, lane merge distance). The change in speed model assume that a driver decelerating or decreasing speed when they encounter a work zone feature were interpreted as positive behaviors. However, they do not capture drivers who may have slowed their speed entering the work zone and then maintained their speed. As a result, they would not have needed to slow when encountering additional features. For the lane merge analysis, it was assumed that drivers merging earlier that is farther ahead of the activity area were showing safe driving behavior than the drivers merging at the vicinity of the activity area. Overall, this study was successful in identifying active work zones from the RID data and reduce valuable information from the forward videos to evaluate driver behavior in work zones.

Copyright Owner

Amrita Goswamy



File Format


File Size

139 pages