Return to search

Establish skid resistance thresholds for local authority roads in the UK using statistical models

Skid resistance is considered as one of the contributing factors that affect traffic accidents; it is considered as an important property of a road surface that is required to maintain a safe road network. The main aim of this thesis is to establish new skid resistance thresholds for local authority roads. This will be accomplished through the following objectives: 1. to define new site categories based on accident layouts at different network features (e.g. junctions, roundabouts) and the relationships between accidents and geometric characteristics (such as radius of curvature and gradient), 2. to estimate the impact of traffic characteristics (natural logarithm of annual average daily flow, percentage of heavy vehicles, and speed limit); geometric characteristics (radius of curvature, gradient, number of lanes and number of minor accesses); and pavement characteristics (skid resistance, rut depth, and texture depth) on both accident frequency and rate, and 3. To analyse the relation of the expected accident frequencies and rates as a function of skid resistance. This study has included A-road networks for Norfolk, Oxfordshire and Nottinghamshire counties in England, UK. These networks are divided into different site categories these site categories are: 1. non-event; 2. bends (0 - 250) m; 3. bends (250 - 500) m; 4. roundabouts; 5. junctions; 6. gradients. Four different datasets for the period 2005-2010 have been merged to construct the final unique dataset for this study. They are: 1. accident data, 2. traffic data; 3. geometric data; and 4. pavement characteristics data. A series of fixed and random parameters Negative Binomial models have been employed to investigate the effect of skid resistance on accident frequency at different site categories for the three counties. The datasets were modelled by total accidents, by road surface condition (i.e. dry and wet), and by severity level (i.e. fatal and serious or slight). In the same way fixed and random Tobit models have been employed to investigate the effect of skid resistance on accident rate. The model estimation results suggest that skid resistance is negatively associated with the frequency and rate for all accident types at all site categories. A 10% increase in skid resistance leads to a decrease in total, dry, wet, slight, and serious and fatal accident frequencies at network level by 12.24, 10.21, 16.34, 10.68 and 4.92%, respectively. A 10% increase in skid resistance (SCRIM value) leads to a decrease in total, dry, wet, slight, and serious and fatal accident rates at network level by 6.32%, 10.62%, 12.52%, 3.31%, and 4.87%, respectively. This thesis contributes to knowledge in terms of accident prediction approach, showing that application of random parameters modelling as a new approach applied in the UK to estimate accident frequency and rate on A - road networks. This method is introduced as a sufficient approach for the researcher due to the ability to account and correct for heterogeneity, which can arise as a result of several factors relating to the characteristics of traffic, geometric and pavement characteristics. In addition, the random parameters model approach provides a reasonable understanding of the main factors that affect accident frequency and rate. Therefore, this approach allows the researcher to identify and control for confounding factors that may bias estimation. In addition, new skid resistance thresholds for different site categories are established based on the analysis of expected accident frequencies (outputs of random parameters Negative Binomial models); and the analysis of expected accident rates (outputs of random parameters Tobit models) as function of skid resistance.

Identiferoai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:765541
Date January 2018
CreatorsAlacash, Hamid Ahmed Awad
PublisherUniversity of Nottingham
Source SetsEthos UK
Detected LanguageEnglish
TypeElectronic Thesis or Dissertation
Sourcehttp://eprints.nottingham.ac.uk/55550/

Page generated in 0.0017 seconds