Accidents and Air Quality Models for EMME/2

1. Accidents and Air Quality Models for EMME/2 Marwan AL-Azzawi

2. Project Goals • Develop series of mathematical models to describe relationship between flows, link attributes, accidents & pollution • Allow the user to take account physical characteristics of highways • Models can be used in any EMME/2 traffic study.

3. Background • Accurate estimation of impacts of traffic on accidents and air quality are growing concern to local and central governments. • Many modern traffic studies also examine impacts of traffic on accidents and pollution, in addition to usual design issues. • Accident and pollution models are normally estimated as a function of highway type and traffic volumes. • But in many instances the road geometric layout is omitted. • This raises a problem with regards to taking into account the different designs and characteristics of different roads.

4. Data Collection • Accident Data • Large data sets input into sophisticated databases • Measured against different road sections and flow levels • Multiple regression analysis • Accident data between 1993 to 1999, for 32 local authorities in Scotland • Road data has over 30,000 miles of road sections in urban, rural and semi-rural areas • Total of over 10,000 accident records and 1 billion vehicle miles • Air Quality Data • From various test vehicles which measure exhaust emissions from vehicles • Different travelling speeds to show effects of congestion • Automatic pollution measuring equipment set up at various locations throughout Scotland (from 1995 to 1999). • 5 different towns and cities

5. Types of Accident Models • Separate models for Fatal, Serious and Slight casualties • Total is the addition of the model results • Variables include • annual traffic flow (million of vehicles) • width of the road section (metres) • Length of the road section (metres) • visibility along the road section (metres) • surface quality (0 for ‘poor’, 1 for ‘good’) • if road section is in an urban area (1 for ‘urban’, 0 if ‘non-urban’) • if road section is in a rural area (1 for ‘rural’, 0 if ‘non-rural’) • if road section is in a semi-rual area (1 for ‘semi-rural’, 0 if not)

6. Types of Air Quality Models • Separate models for the 3 main traffic pollutants Carbon Monoxide (CO), Hydrocarbons (HC) and Nitrogen Dioxide (NO2) • E = k + aV + bV2 + cV3 • E = emission rate per vehicle (g/km) • V = vehicle speed (km/h) • a, b, c and k are co-efficients • Different co-efficients for different road types • Relates speed of travel across the link, thereby taking into account the effects of rising and falling congestion levels on the rate at which traffic emissions vary.

7. Results of Tests • The models tested against actual measured accident and air pollution data • Results show new models are statistically satisfactory, with the accident models a little better than the air quality • This is arguably due to the vast amount of data used to develop the accident database

8. Future Developments • Accident Models • Further disaggregate the models • Allow the effects of built-up development activities (e.g. retail frontage parking, loading and unloading services, etc) • Would closer represent situations in busy town or city centres • Pedestrian and cycling accident models • Include vehicle/pedestrian and vehicle/cycle accidents • Air Pollution Models • Develop models of other air pollutants (e.g. Carbon Dioxide, Particulate Matter, Lead, etc) • Allow examination of pollutants agreed at environmental health summits, including Rio Summit (1992) and Kyoto Agreement (1999)

9. Conclusions • New accident and air quality models should improve traffic studies • Help simulate affects of congestion • Further work on-going to develop model parameters for other road types, and other accidents and pollutants