Safety Impacts of Variable Speed Limit Strategy Were Evaluated Using a Before-and-after Analysis and a Machine Learning-based Crash Prediction Model in Utah.
Utah, United States
I-80 Hybrid Regulatory Speed Limit Signing Design and VSL System Evaluation
Summary Information
The Utah Department of Transportation (UDOT) established a Variable Speed Limit (VSL) zone to improve traffic safety along a section of U.S. Interstate 80 (I-80). Beginning Fall 2019, UDOT’s original changeable message signs (CMS) with white digits and black backgrounds were replaced with CMS with amber digits to improve visibility during inclement weather. This study evaluated the effectiveness of the new CMS in improving speed compliance and performed a comprehensive before-and-after analysis of VSL system performance. A before-and-after analysis was conducted using data obtained from roadside detectors, crash records, and video recordings. Data were collected in Fall 2018 and 2019 and Winter 2019 and 2020 from eight signs in the eastbound and seven signs in the westbound direction. A road safety assessment model using Machine Learning (ML) was also created to predict road safety and speed compliance rates using crash records and road geometric features.
METHODOLOGY
Researchers collected data on a biweekly basis during October 2018, October 2019, January 2019, and January 2020. Vehicle travel speeds and traffic flow data were collected from stationary roadside detectors. Weather data were also collected during these periods to quantify its impact on VSL compliance rate and road safety. Driver compliance rates were calculated by comparing the average travel speed to the displayed speed limit, and by measuring the ratio of drivers traveling less than 10 miles per hour (mi/h) above the speed limit. Video snippets were recorded multiple times during bother summer and winter seasons by vehicle onboard cameras during field trips to compare the visibility of original and new signs. Crash data were obtained through state records; crash frequency and severity were calculated as direct measurements of effective (MOE) of the VSL. In addition, safety models were developed using a ML technique to evaluate crash frequency and severity on the I-80 corridor using the occurrence data, weather index, and crash records. The impact of improved visibility on crash severity was assessed by adding the variable of VSL legend color to the model.
FINDINGS
- Compared with the white legend VSLs, when visibility was affected most (e.g., in wintertime) the compliance rate with the amber signs improved by nine percent on average.
- Based on the crash records, crash frequency decreased by about 50 percent (compared to the white digit message signs) after installing amber digit signs. Crash severity also decreased slightly.
- Based on the data analysis, after the installation of the new CMS, average vehicle travel speeds in the VSL zone decreased and variations from the displayed speed limit lowered.
- Based on the comparison of the video snippets recorded during the field trips, the original white legend VSLs have reduced visibility in direct sunlight during the summer as compared to the new amber legend VSLs. The visibility of amber legend was also not significantly affected by inclement winter weather, such as fog and snow.
- Based on the developed safety models using ML techniques, crash frequency was estimated to decrease by up to 80 percent and severity by 8.26 percent with the widespread installation of amber CMSs.
