|dc.description.abstract||This thesis reports on a research project aimed at developing comprehensive methods for quantifying the benefits and risks of urban cycling, investigating the environmental exposures of cyclists and developing a tool for optimising the design of cycling infrastructure. These goals have been achieved through a combination of statistical modelling, mathematical programming and data collection and analysis. There are many good reasons to encourage urban cycling such as the reductions in the social costs of air and noise pollution and the promotion of active and healthy lifestyles. However, there are also risks associated with urban cycling such as traffic collisions and increased inhalation of air pollutants. Promotion of cycling also requires investment of resources such as capital and road space. For these reasons it is essential that all of these benefits and risks can be quantified in common units to enable evidence-based policy formulation. In order to mitigate the risks of cycling, the factors affecting the variability in these risks also need to be understood so that they can be mitigated. Finally, in order to design measures to promote cycling in such a way as to optimise for these impacts, it must be possible to model the change in travel behaviour and associated impacts resulting from such measures.
The first step taken in this thesis was to conduct a thorough review of the literature relating to quantification of health and environmental impacts of transportation, in-travel environmental exposures and modelling of travel behaviour. The literature review revealed that there was significant heterogeneity in the modelling techniques and a lack of studies which considered the variation in benefits and risks experienced by individual cyclists. A comprehensive framework was therefore developed to quantify the health, environmental and travel time impacts of cycling. Different mathematical models from the literature were used to quantify impacts related to physical activity, pollution inhalation, traffic collisions, noise, vehicle emissions and travel time. Using census data, this framework was applied to a case study of an increase in cycling in Dublin. It was shown that, at a societal level, the total net health and environmental impacts of increased cycling in Dublin would be strongly positive. When travel costs are also considered, the uncertainty becomes greater but the best estimate of the net impact is still positive. A framework was also developed for quantification of the expected benefits and risks experienced by an individual cyclist switching from driving to cycling. The application of this framework to a case study of Dublin was the first study to investigate the variation in health impacts of cycling experienced by individuals of different ages, genders and travel distances. A stochastic simulation approach was employed in order to calculate distributions of each health impact reflecting both the uncertainty in the model parameters and the variation among individual characteristics. It was shown that for some groups, in particular males aged 20-30, cycling can have negative expected net health impacts.
The literature review also indicated that one particular health impact of cycling?in-travel pollutant inhalation?was relatively poorly understood in terms of the factors affecting its variation. In order to investigate this further using an exposure study, a new environmental sensing node?the BEE node?was first designed and built using relatively low-cost pollution sensors and electronics. The BEE node?s gaseous pollutant measurements were calibrated and validated to a degree of accuracy comparable with much bulkier and expensive equipment. Using the BEE node, two studies of the environmental exposures of cyclists in Dublin were designed and carried out. Volunteer cyclists cycled through Dublin while collecting data about the pollution and noise they were exposed to. These data were analysed along with time-resolved information about the cycling facilities they used, the vehicle traffic volumes they interacted with and the weather conditions. The analysis produced new insights such as the observation that while segregated cycle lanes decreased pollution exposures, roadside cycle lanes and bus lanes actually increased exposures.
Having developed models for quantifying the benefits and risks of cycling, the next step was to develop a model for predicting the change in levels of cycling and driving which would result from a given intervention. This required a disutility function for cycling and while a small number of studies have previously suggested functional forms, no study had attempted to calibrate or validate such a function. A new method for calibrating a cycling disutility function was therefore developed. The method involves formulating the calibration problem as a Mathematical Programme with Equilibrium Constraints (MPEC) and solving the MPEC using a descent-based method. A new cycling disutility function?the DOC function?was also proposed based on previous research into the factors affecting cycling disutility. Using the newly developed calibration method, this function was calibrated and its accuracy was validated based on data from the Dublin network.
The final contribution of this thesis was a tool for systematically designing a cycle network in order to optimise the resulting net impacts to the network users and society. This Network Design Problem (NDP) is formulated as an MPEC and a solution approach is presented which is uses a genetic algorithm (GA) to find the optimal solution. The problem formulation and solution algorithm are tested using a numerical example and the GA algorithm was shown to efficiently converge to a near-optimal solution for the cycle network design.||en