Kamla, Jwan Jameel Shekh Mohammed
(2016)
Analysing truck position data to study roundabout accident risk.
PhD thesis, University of Nottingham.
Abstract
In order to reduce accident risk, highway authorities prioritise maintenance budgets partly based upon previous accident history. However, as accident rates have continued to fall in most contexts, this approach has become problematic as accident ‘black spots’ have been treated and the number of accidents at any individual site has fallen. Another way of identifying sites of higher accident risk might be to identify near-miss accidents (where an accident nearly happened, but was avoided), which are likely to be much more prolific than actual accidents, therefore they are useful in identifying high-risk sites. The principal aim of this research is to analyse potentially unsafe truck driving conditions that involving harsh braking incidents (HBIs) that may indicate accident risk.
Most modern truck fleets now record position as part of fleet management. This research used position data collected by a truck fleet management company for 8000 trucks in the United Kingdom (UK) over a 2-year period (2011-2012) to identify incidents of harsh braking. This data was compared with STATS19 accident data events (specifically truck accidents) occurring in 70 selected roundabouts (284 approaches) over an 11-year period (2002-2012), to test the hypothesis that the HBIs could represent accident near-misses and therefore increased accident risk. The data used for model prediction comprised all vehicle accidents, truck accidents, HBIs, geometric properties, and traffic characteristics for whole roundabouts, within the circulatory lanes, and at approaches to the selected roundabouts. Random-parameters negative binomial (NB) count data models were used to estimate model parameters and the models were compared with fixed-parameters NB count data models.
It was found that random-parameters count data models provide better goodness of fit and more variables were found to be significant, giving a better prediction of events. It is concluded that HBIs are influenced by traffic and geometric variables in a similar way to total and truck accidents, therefore they may be useful in considering accident risk at roundabouts. They are a source of higher volumes of data than accidents, which is important in considering changes or trends in accident risk over a much shorter time. The most important variables were Average Annual Daily Traffic (AADT) and percentage of truck traffic, which were found to have a positive influence on accidents and HBIs. Regarding the geometric variables, signalisation, circulatory roadway width, number of arms and two-lane indicator were the most important factors influencing accidents and HBIs. In addition to these models, numbers of HBIs was used as an independent variable in the models of total and truck accidents, along with traffic and geometric variables. From the results it can be concluded that at all approaches, HBIs are related to total accidents along with traffic and geometric variables, which can be used to study safety measures. A good predictive model for truck accidents at M-class approaches based on HBI, traffic and geometric parameters was identified that can be used for prioritising safety at these approaches in order to make roundabouts safer. For A- and B-class approaches a better fit model were identified when HBI were used as input variable along with traffic and geometric variables compared to the model without using HBI as input variable, but the influence of HBIs was negative (high HBIs with low numbers of accidents) which is probably an indicator of future accident risk in these locations. For at-grade roundabouts, a better fit model was obtained for total and truck accidents when it is compared to the model without HBIs, but the influence of HBIs was negative; this is probably an indicator of high accident risks in these at-grade roundabouts, however further investigation is required with more observations. These results for truck HBIs could help highway authorities to identify sites of increased accident risk more rapidly and without waiting for an accident history to develop.
| Item Type: |
Thesis (University of Nottingham only)
(PhD)
|
| Supervisors: |
Parry, Tony
Dawson, Andrew R. |
| Keywords: |
Road accidents, near-miss accidents, position data, truck accidents, harsh braking, roundabouts, random- and fixed-parameters count data model, traffic and geometric variables |
| Subjects: |
H Social sciences > HE Transportation and communications |
| Faculties/Schools: |
UK Campuses > Faculty of Engineering |
| Item ID: |
37386 |
| Depositing User: |
Kamla, Jwan
|
| Date Deposited: |
13 Dec 2016 06:40 |
| Last Modified: |
12 Oct 2017 21:33 |
| URI: |
https://eprints.nottingham.ac.uk/id/eprint/37386 |
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
