Empirical Bayes Method for Highway-Rail Grade Crossing Accident Prediction
Jacob Mathew, University of Illinois, Urbana ChampaignShow Abstract
Rahim Benekohal, University of Illinois, Urbana Champaign
Accident prediction at railroad grade crossings is often performed with the U.S. Department of Transportation (USDOT) accident prediction formula, as described in the FHWA Railroad–Highway Grade Crossing Handbook. This method was developed in the 1980s and remain unchanged except for the normalizing constants which are updated every few years. This paper discusses the Empirical Bayes (EB) method developed to estimate accident prediction values at grade crossings and results are compared with the USDOT accident prediction model. The Empirical Bayes approach allows us to include “variables” related to both the physical characteristics and the accident history at that location to enhance the prediction accuracy of the model as well as to address the regression to mean bias in the model.
Effectiveness of Treatments to Reduce Congestion at Level Crossings: A Traffic Simulation Evaluation
Gregoire Larue, Queensland University of TechnologyShow Abstract
Marc Miska, Queensland University of Technology
Gongbin Qian, Queensland University of Technology
Christian Wullems, Queensland University of Technology
David Rodwell, Queensland University of Technology
Edward Chung, Hong Kong Polytechnic University
Andry Rakotonirainy, Queensland University of Technology
Railway level crossing closures can disrupt traffic flow significantly, especially in peak hours. The current increases in road and rail traffic worsen the situation and can result in congestion known to increase risky behavior from road users due to impatience and poor decision making. In this study, seven of the most problematic level crossings around Brisbane, Australia were surveyed. The effectiveness of a set of treatments was tested and discussed using computer simulation models. The study found that variability of warning times is the major cause of unnecessary boom gate down time. Our study shows that reducing the variability in warning times could reduce congestion significantly by 10-40s on average at the investigated level crossings. Major cause for the variability in warning times are trains stopping at stations that are not equipped with express train identification, and the difference between posted line speeds and actual train speeds. Various treatments were found to be effective at reducing level crossing closure duration by reducing the variability in warning times. This study shows the potential for short to medium term treatment of congestion issues at active level crossings, which are necessary with the current increased rail and road traffic flows. Such improvements would reduce errors and violations from level crossing users, and as a result improve level crossing safety.
A Resource Optimization Model for Improving Railway-Highway Grade Crossing Safety in Canada
Lalita Thakali, University of WaterlooShow Abstract
Liping Fu, University of Waterloo
Jay Rieger, Transport Canada
Shuangshuo Wang, University of Waterloo
This paper presents a new approach to the problem of allocating federal resources and identifying upgrading projects for improving safety at grade crossings in Canada. The proposed approach is unique in two key aspects. First, a risk-based network screening process is adopted to identify the priority sites for providing a justifiable basis for distributing the total budget at a regional level as well as narrowing the search space in the subject optimization step. Secondly, a mathematical programming approach is applied to formalize the resource allocation process with an explicit consideration of the expected benefits – risk reduction and the costs of implementing the projects. This approach is expected to improve the process of identifying the optimal set of upgrading projects within each region, thus maximizing the return of investment. A full-scale case study using real data from the Canadian crossing network is conducted to demonstrate the applications of the proposed model and its potential in generating solutions that balance the need for achieving fair allocation of funds among the regions and retaining the optimality of the identified improvement projects.
Analysis of Data Retrieved from a Staged Train Collision with a Motor Vehicle
James Poslusny, North Carolina State UniversityShow Abstract
Sarah Searcy, North Carolina State University
Christopher Cunningham, North Carolina State University
Little empirical data exists on the effects of a train collision with a motor vehicle, especially for an event that occurs when both the train and the motor vehicle are moving. The North Carolina Department of Transportation (NCDOT) Rail Division’s BeRailSafe program, in partnership with Operation Lifesaver and Aberdeen, Carolina & Western Railway, coordinated the first staged collision between a moving motor vehicle and train ever conducted in North Carolina. For this unique experiment, a retired 2009 Dodge Charger police cruiser was impacted while in motion by a moving EMD GP40-2LW locomotive on an at-grade crossing. The car was dragged for five seconds by the train before it was deposited on the railroad right-of-way. Analysis of data retrieved from the Event Data Recorder that was installed in the motor vehicle shows a crash pulse with extreme peaks in both the lateral and longitudinal acceleration and a statistically significant spike in mean acceleration during the crash.
Pedestrian and Bicyclist Behavior at Highway-Rail Grade Crossings: An Observational Study of Factors Associated with Violations, Distraction, and Crossing Speeds During Train Crossing Events
Brendan Russo, Northern Arizona UniversityShow Abstract
Emmanuel James, Northern Arizona University
Taylor Erdmann, Northern Arizona University
Edward Smaglik, Northern Arizona University
Pedestrian-train collisions at highway-railroad grade crossings (HRGCs), though relatively rare events, often result in severe injuries or fatalities. In the ten year period from 2008-2017, there were 1,470 pedestrian-train collisions reported in the United States resulting in 908 fatalities and 492 injuries according to the Federal Railroad Administration. Furthermore, pedestrians in over 27% of these crashes were coded as disregarding gates at HRGCs. In order to examine both pedestrian and bicyclist behavior which may lead to such unsafe behavior and potential crashes and resultant injuries and fatalities, this observational study utilized pedestrian and bicyclist behavior data extracted from high-definition video recordings of highway-rail grade crossings in downtown Flagstaff, Arizona. Several pedestrian and bicyclist behavior characteristics were extracted from the videos including the prevalence of pedestrian and bicyclist violations (e.g. disregarding flashing lights and/or gates), pedestrian and bicyclist distraction (e.g. cell phone use while crossing), and pedestrian and bicyclist crossing speeds. Using these data, along with rail-grade crossing features and train crossing parameters, and pedestrian and bicyclist demographic data, statistical analyses are performed to examine factors which are significantly associated with pedestrian and bicyclist violations, distraction, and crossing speeds at highway-rail grade crossings. Ultimately, the results of this study provide important information which may be useful in planning engineering, educational and/or enforcement strategies to reduce or mitigate the impacts of unsafe pedestrian or bicyclist behavior at HRGCs.
A Human Behavior Analysis of Highway-Railroad Grade Crossings Based on Environmental Conditions and Driver Demographics
Alawudin Salim, Michigan Technological UniversityShow Abstract
Pasi Lautala, Michigan Tech Transportation Institute
Although the number of highway-railroad grade crossing (HRGC) accidents has significantly decreased over the recent decades, they remain one of the greatest sources of injuries and fatalities in railroad industry. From 2010 to 2014, an average of nearly 2,100 accidents per year have taken place at HRGCs in the United States, leading into more than 250 fatalities each year. Previous studies have identified several factors that increase accident risk at HRGCs and one of the leading causes has been identified as driver behavior. However, there is less understanding on what conditions/causes encourage driver behavior that results in increasing accident risks at HRGCs. Several past studies have used accident events to investigate the HRGC safety, but this study attempts to understand drivers’ actions at HRGCs by investigating successful driving events obtained from naturalistic driving study (NDS). The study uses a three-points quantitative evaluation criterion evaluate 9,128 NDS traversals data across 286 grade crossings. It calculates the average behavior scores and performs statistical tests to reveal differences under environmental (weather and time of day) conditions and for different driver demographics (gender and age). The study results found that drivers received higher behavior scores in snow and lower scores in rain conditions, but only limited pairings were statistically significant. Drivers received significantly lower behavior scores during the night time than day time traversals. On the other hand, there were limited statistical difference between males and females across different age groups. Keywords: Highway-Railroad Grade Crossing, Grade Crossing Accidents, Driver Behavior, Naturalistic Driving Data
An Evaluation of Traffic Control Devices and Driver Distraction at Grade Railroad Crossings
Radhameris Gómez, University of Massachusetts, AmherstShow Abstract
Nicholas Campbell, University of Massachusetts, Amherst
Francis Tainter, University of Massachusetts, Amherst
Michael Knodler, University of Massachusetts, Amherst
Donald Fisher, OST-R/Volpe Center
Driver behavior and poor judgement account for 94 percent of automobile and train collisions at railroad-highway crossings (1, 2). This research sought to understand and mitigate the impacts of driver distraction driver performance in level at-grade railroad crossings. To accomplish this, the effectiveness of warning devices for at-grade railroad crossings was studied in a driving simulator environment. Fifty-three participants were recruited to drive 2 simulated worlds, each containing 6 different scenarios of railroad grade crossings. Twelve experimental scenarios of railroad grade crossings were developed to evaluate driver performance as participants traversed these crossings. The application of driving simulator technology is advantageous in that allows us to model and manipulate real word scenarios in a controlled experimental setting. Participants were either distracted, or not distracted, and the drivers’ glance and driving behavior were observed. The scenarios centered on varying four key safety factors at each crossing: 1) The presence or absence of a vehicle ahead of the driver, 2) The visibility of the advance warning sign, 3) The visibility of the crossbuck and the flashing lights; And lastly, 4) the state of the flashing lights The greatest difference in the proportion of glances would be seen between participants who were distracted and those who were not. Distracted drivers, particularly those performing the in-vehicle task, had difficulty in detecting grade crossings accordingly as compared to non-distracted drivers. In all groups, participants had difficulty in navigating the grade crossing environment, which provides reason to investigate supplementary treatments in future research endeavors.
Market Basket Analysis of Safety at Active Highway-Railroad Grade Crossings
William Saunders, Louisiana Department of Transportation and DevelopmentShow Abstract
Julius Codjoe, Louisiana Department of Transportation and Development
Saleh Mousa, Louisiana State University
Grace Ashley, University of Louisiana, Lafayette
The number of crashes at United States’ approximately 211,893 highway-railroad grade has been steadily rising over the years. Part of the cause is drivers stopping within the dynamic envelope zone (DEZ) of the train when in queue. By evaluating the effectiveness of regulatory signage, this study analyzes the efficacy and potential of the market basket analysis (MBA) as a valid modeling technique in transportation engineering research. First, a comparative safety analysis is undertaken with percentage change calculations and a chi-squared test in order to determine the effectiveness of regulatory signage. Then a comprehensive multivariate MBA using parameters influenced by visibility, perception, and maneuverability is performed to further analyze driver awareness at the grade crossings. The safety analysis concluded that the regulatory signage had an overall positive effect on DEZ stopping behavior at 4 out of 5 of the sites investigated with two anomalies at Sites B and D. The resulting p-value from the chi-squared test for all sites indicated a significant association between these variables level of compliance and signage installation. The MBA demonstrated its value by confirming the results of the safety analysis and increasing the number of parameters which can be analyzed simultaneously. This study offers the scientific community a new approach to understanding driver behavior and the results provide valuable insight for developing preventive measures, which will ultimately help reduce accidents at grade crossings. MBA can provide information for transportation engineers and railroad administrators in identifying important parameters in determining intersections that require effective signage installation.