Best Practices For Crash Modeling And Simulation

Author : Victor Muchuruza
Publisher :
ISBN 13 :
Total Pages : pages
Book Rating : 4.:/5 (796 download)

DOWNLOAD NOW!

Download or read book Simulation of Traffic Crashes Using Cell Based Micro-simulation written by Victor Muchuruza and published by . This book was released on 2006 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: ABSTRACT: The deterioration of the safety of operation, coupled with the persistent increase in rearend crashes, is of great concern in finding accurate and realistic methods of modeling traffic flow and preventing traffic crashes. For some decades safety evaluation methods have relied on analysis of historical crash data. Since crashes are random and rare events and, in most cases, are independent events, it is difficult to find a sufficient number of crashes on a road section in a relatively short time period (e.g., a month or even a year). Thus, multi-year collection of crash data is used in safety analysis. Another safety evaluation method that has been practiced though in small scale is traffic conflict techniques (TCT). The advantage of using TCT in safety evaluations is the ability to test or study a safety strategy or improvement applied on the roadway facility in a relatively short period of time compared with traditional methods, which are dependent on crash data. However, use of TCT is not popular; perhaps because it needs extensive resources to collect, extract, and analyze conflict information. Moreover, like crash data analysis, use of TCT also makes concerned authorities reactive to the problem by responding to the crashes that have already occurred. Therefore, alternative proactive safety evaluation techniques that can improve the quality of traffic safety evaluation are needed at this time. One way of using proactive safety evaluation techniques and thus become more preventive than reactive towards dealing with the overall safety problem is to utilize the capability of traffic micro-simulation to assess safety on highways through examination of hazardous vehicle movements in the traffic stream. Using micro-simulation predictive methods, it may be possible to diagnose safety problems and apply appropriate remedial measures, rather than waiting until a crash occurs to remedy the problem. This means, a hazard can be early identified and possibly corrected before implementation of highway projects. In addition, the use of simulation tools to evaluate the safety of a traffic system can be advantageous because such tools provide extensive results for any study area within a relatively short time along with other traffic operational measures like level of service, delays, travel times, and capacities. Therefore, the objective of this dissertation was to analyze numerically the likelihood of the traffic crashes that might occur on the highway using cellular based micro-simulations. The modeling considered occurrence of rear-end crashes on high-speed highways with two lanes of traffic in each direction. Narrowing the safety evaluation to rear-end crashes, this study sought to analyze these crashes by providing simulation evidence of association between time-based traffic safety indicators and driver attributes with the likelihood of conflict or collision. To meet the study objectives, a stochastic cellular automata traffic model have been extended to use field-derived vehicle and driver characteristics. The vehicles' acceleration submodel in the simulation is categorized into different regimes depending on the prevailing traffic conditions. The vehicles' evolutions in the proposed micro-simulation model are based on kinematic equations to enhance the realism of their advancements. Behavioral variance in the model is introduced by taking in consideration both driver aggression and responsiveness to the traffic conditions. The model is calibrated using field data. Comparison of simulated spacings and speeds obtained from the simulation output with vehicle trajectories data obtained from the field return a Mean Absolute Percentage Error (MAPE) of less than 10% and a Theil's coefficient of inequality (U) of about 0.002. These statistics inferred that the proposed model worked well in replicating traffic on the field. In addition, correlation results showed that simulation results not only agree to the theoretical results but also to the detector data collected from the field. The driver behavior was found to contribute more in the likelihood of crashes which was determined by amount of great deceleration that driver apply to maintain safety during movement. The likelihood of vehicles to crash in the model was formulated from the Gamma distribution functions. Closer examination of the probability of a vehicle to crash in the model indicated that the likelihood of crashing is high when the traffic is flowing close to the maximum flow.