Intersections are the most complex locations in a traffic system and are likely to have a higher crash count than any other location in the system. Intersection safety is related to traffic operations, such as traffic signal and approaching volume. The objective of this study is to determine the contributing factor for left-turn crashes at signalized intersections by a statistical modeling process and to develop crash prediction models. Potential contributing factors representing the characteristic of a left-turn operation were identified and considered for inclusion in crash prediction models. HCS (Highway Capacity Software) 2000 was utilized for computing some traffic indicators such as volume to capacity ratio for potential inclusion in the models. Three years of crash data were collected in the College Station area. The Signal timing and Volume data were obtained from public works in College Station. The volume data was sorted into three time periods and signal timing data were obtained for three different time periods: AM, noon, and PM. The division of time periods results from timing plans being changed for different periods. Relationship between crash count and each factor was explored to identify whether the factor has the potential to influence the crash count. Afterwards, the prediction models were developed using the negative binomial structure because of many zero samples. Akaike Information Criteria was used for selecting the model having the best fit. Wald tables provided that variables have significance in affecting the left-turn crash count. Left-turn type, sequence, volume, control delay, and post speed limit were identified as significant factors impacting left-turn crash count at a signalized intersection.
| Identifer | oai:union.ndltd.org:TEXASAandM/oai:repository.tamu.edu:1969.1/ETD-TAMU-1911 |
| Date | 02 June 2009 |
| Creators | Lee, Sunghoon |
| Contributors | Zhang, Yunlong, Lord, Dominique, Wehrly, Tom |
| Source Sets | Texas A and M University |
| Language | en_US |
| Detected Language | English |
| Type | thesis, text |
| Format | electronic, application/pdf, born digital |
Page generated in 0.0023 seconds