Traffic Models – Their Value to the Development of Cycle Networks

Traffic Models – Their Value to the Development of Cycle Networks


Eoin O'Mahony, AECOM, Dan Brennan, AECOM, Joe Seymour, AECOM


This paper summarises the application of traditional traffic modelling techniques to the development of cycle networks.A case study of the Greater Dublin Area Cycle Network will be presented.


This submission has been prepared by the National Transport Authority and AECOM Roughan & O’Donovan in relation to the Greater Dublin Area Cycle Network Plan. The project was delivered in 2013 and set the challenging task of developing a strategic cycle network for the Dublin City, Fingal, South Dublin, Dun Laoghaire Rathdown and Wicklow areas.

The proposed network will treble the existing cycle lanes in urban areas from 500km to 1,485km in length, and will provide over 1,300km of new connections between towns in the rural areas of the Greater Dublin Area.

The planned network includes safe, accessible and direct routes along primary and secondary roads to meet the demands of work and school commuters and greenway routes - off road facilities through parks, and along waterways – which will be more generally used for leisure and tourism.

Drawing on a rich dataset a bespoke cycling demand model was created to assess the future demand in the busiest areas.
The Census POWSCAR database was released in August 2012, and reports all journeys to work and education by District Electoral Division (DED) for 2011. This information was extracted for input to a cycling traffic models, thereby giving good Origin-Destination information without the necessity for widespread Roadside Interview Surveys which can result in unsatisfactory disruption of road users. The POWSCAR information also provided travel mode and time of departure, thereby allowing journeys by bicycle during the AM period to be isolated.

The cycling demand matrices were assigned onto a transport network using VISUM v12 strategic modelling software. The main source of network information for the model was taken from NAVTEQ data. Supplementary cycling network links that are not part of the road network were coded manually based on information from other vector sources.

An assignment of the cycle matrix to the base year model enabled a comprehensive check of the network to be undertaken. In addition, cycle count data on the ‘canal cordon’ (i.e. entry points into Dublin City) was used to compare the modelled flows to observed data.

A number of network checks were undertaken such as the locations of zone centroid connectors, zone to zone movements and route choice. As a result of this network calibration process, a closer match between modelled and observed data was achieved. No matrix estimation or other calibration procedures were undertaken on the base year cycling matrices.

A target of increasing cycling numbers to 75,000 each morning by 2021 has been set. This would represent a three-fold increase in cycling over 2011 levels and would mean the cycle network could carry as many commuters in the morning in 2021 as is currently carried by bus.


Association for European Transport