Urban Road Network Design with Balance between Vulnerability and Reliability

Under normal events, travelers have route choice behaviors with risk aversion while under abnormal events, their route choice behaviors have a feature of both risk aversion and regret aversion. For this reason, an urban road network design model is constructed. It is a bilevel programming model with two lower level models in which reliability measure depicts road network performance under normal events, and vulnerability measure represents road network performance under abnormal events. The objective of the upper model is to optimize road network vulnerability measure（i.e., maximize road network accessibility measure）while it is subject to a predefined reliability constraint. The lower model include two stochastic user equilibrium models based on the utility theory and on the regret theory respectively. The results of numerical examples show that, compared with the model simply for vulnerability, the proposed model can acquire higher reliability with reduced accessibility to some extent. If investment budget is 0.9 × 10⁷, the average road network accessibility and road network reliability of the proposed model are 0.138 8 and 0.969 6 respectively, whereas as to the model simply for vulnerability, they are 0.140 5 and 0.334 1 respectively. Totally, the accessibility measure decreases by 1.20% and the reliability measure increases by 190.21%. Besides, the neglect of differences in travel decision behaviors may lead to a sub-optimal even false network design.