The Highway Safety Manual (HSM) (AASHTO 2010) introduced quantitative methods to be used by transportation engineers and practitioners for safety and capacity assessment. Although the HSM includes methods to predict crashes for many different road facility types, it only gives two Crash Modification Factors (CMFs) to calculate the effect of increase or decrease of freeway work zone length and duration on the crash count. The HSM methodology for work zones is based on 36 freeway work zones with high traffic volumes in California. HSM models were calibrated in a recent research by Rahmani et al. (2016) using data from Missouri and the study determined a calibration factor of 3.78 which creates concerns since it is significantly larger than 1. This report describes the research conducted to make 15 different models to predict crashes for work zones on three facility types (freeway, expressway and rural two-lane highways) using Missouri data. For work zone safety studies, different databases such as work zone characteristics, crash database and road network information need to be linked together. The tremendous amount of effort required for data collection and checking process makes work zone safety studies challenging. Of the 20,837 Missouri freeway, 8,993 expressway and 64,467 rural two-lane work zones that were analysed in this report, samples of 1,546 freeway, 1,189 expressway, and 6,095 rural two-lane work zones were used to make eight, four and three models respectively. The samples were extracted using work zones longer than 0.1 mile and with a duration of greater than 10 days. The thresholds for minimum work zone length and duration were developed using a theoretical method devised by the authors. Most work zones in database were small work zones with short durations and no crashes. Using all of these work zones in the sample is possible but increases the uncertainty of the resulting model’s predictions. However, by increasing the minimum length and duration threshold the sample size decreases. Thus, there is a trade-off between dropping more small work zones and the sample size. This study tested different length and duration thresholds to extract the sample, and made work zone crash prediction models. By comparing the accuracy of the developed models, the optimum thresholds for minimum length and duration were found. Table ES-1 presents the characteristics of the work zones such as length, duration, AADT and number of crashes for all three facility types. The table shows that the work zone data represented a wide variety of work zones. In work zone databases, the footprint of a work zone is typically recorded as the beginning and end of the work area. To account for the crashes that occur in the advance warning area, transition area, buffer area and termination area of work zones, most studies in the literature considered a constant threshold before the start and after the end of each work zone. The model used by the HSM (similar to most studies in the literature) classified all crashes within 0.5 mile (0.8 km) of the beginning and 0.5 mile (0.8 km) after the end of the work zone as work zone crashes. In contrast, as a new contribution this study used more accurate variable MUTCD (FHWA 2009a) recommended temporary traffic control plans’ thresholds for freeway, expressway and rural two-lane work zones. All 15 of the models developed in this study were programmed in a user-friendly spreadsheet tool for practitioners. An illustrative example is presented to show how this software can be used for assessing the safety of different work zone plans. Figure ES-1 and ES-2 show the software graphical user interface and an example of output respectively. This study also included a survey of DOTs, FHWA representatives, and contractors to assess the current state of the practice regarding work zone safety. Two separate online surveys were developed. One survey was for contractors and the other survey was for both DOT and FHWA representatives. There were seven respondents to the contractor online survey and 29 respondents (27 DOT respondents and 2 FHWA respondents) to the DOT and FHWA online survey. In addition, follow-up phone interviews were conducted with one contractor, eight DOT representatives, and one FHWA representative. Speed reduction was the most important factor identified by the contractors for freeway work zone safety while the number of intersections was the most important factor identified by the contractor respondents for facilities with at-grade intersections. The factors that more than half of the contractors took into account for work zone safety evaluations included traffic volumes, crash history, site characteristics and experience. Work zone traffic volume was the most significant factor affecting freeway work zone safety identified by the DOT and FHWA respondents, and the number of intersections in the work area was the most significant factor identified by these respondents for facilities with at-grade intersections. The factors that more than half of DOT and FHWA respondents considered for work zone safety evaluation include traffic volume, crash history, site characteristics and experience. The survey results indicate that many agencies look at work zone safety informally using engineering judgment. Respondents indicated that they would like a tool that could help them to balance work zone safety across projects by looking at crashes and user costs. (Author/publisher)
Samenvatting