Statistical and simulation methods for evaluating stationary and mobile work zone impacts
In 2014, nearly 10% of overall congestion on freeways was due to the presence of work zones (WZs), equivalent to 310 million gallons of fuel loss (FHWA, 2017a). In terms of safety, in the US, every 5.4 minutes, a WZ related crash occurred in 2015 (96,626 crashes annually) (FHWA, 2017b). Maintenance work involves both Stationary Work Zones (SWZs) and Mobile Work Zones (MWZs). There are many analytical and simulation-based tools available for analyzing the traffic impacts of SWZs. However, the existing traffic analysis tools are not designed to appropriately model MWZs traffic impacts. This study seeks to address this gap in existing knowledge through the use of data from MWZs to develop and calibrate traffic impact analysis tools. This objective is accomplished through data fusion from multiple sources of MWZ, probe vehicle and traffic detector data. The simulation tool VISSIM is calibrated for MWZs using information extracted from videos of MWZ operations. This is the first study that calibrated the simulation based on real driving behavior behind a MWZ. The three recommended calibration parameters are safety reduction factor of 0.7, minimum look ahead distance of 500 feet and the use of smooth closeup option. These calibration values can be used to compare MWZ scenarios. Also, the data collection framework and calibration methodology designed in this study could be used in future research. The operational analysis concluded that a moving work activity lasting one hour or more are suggested to be done when traffic volumes are under 1400 veh/hr/ln, and preferably under 1000 veh/hr/ln, due to the drastic increase in the number of conflicts. In addition, three data driven models were developed to predict traffic speed inside a MWZ: a linear regression model and two models that used Multi-Gene Genetic Programming (MGGP). The second objective is to develop models and tools for safety assessment of stationary work zones. In the WZ safety literature, few studies have quantified the safety impact of SWZ and almost no quantitative study assessing MWZ safety impact. Using Missouri data, this study introduces 20 new crash prediction models for SWZs on freeways, expressways, rural two lane highways, urban multi-lane highways, arterials, ramps, signalized intersections, and unsignalized intersections. All the models except freeway SWZs are proposed for the first time in the literature. The mentioned SWZ models are implemented in a user-friendly spreadsheet tool which automatically selects the most appropriate model based on user input. The tool predicts crashes by severity, and computes the crash costs. For MWZs, there is no crash data available to develop crash prediction models. Thus, this dissertation analyzed conflict measures as a surrogate for safety impacts of MWZs. Conflict measures were generated from the trajectories of traffic simulation model. The safety trade-off plots between conflicts and combination of MWZ's duration and traffic volume were introduced. A transportation agency can use these plots to determine, for example, if they should conduct a MWZ for a short duration when the volume is high or for a longer duration when the volume is lower.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 License.