Bilevel Optimization Approach to Design of Network of Bike Lanes

Abstract

A bike lane is an effective way to improve cycling safety and to decrease greenhouse gas emissions with the promotion of cycling. Improvements include high-quality off-road facilities and on-road bike lanes. Whereas construction of off-road lanes is not always possible because of urban land constraints and construction costs, on-road lanes can be a cost-effective alternative. An optimization framework for the design of a network of bike lanes in an urban road network was proposed. This framework identified links on which a bike lane could be introduced. Allocation of a lane to cyclists would increase the use of cycling, although it could disadvantage auto traffic. The presented approach balances the effects of a bike lane for all stakeholders. A bilevel optimization was proposed to encompass the benefits of cyclists and car users at the upper level and a model for traffic and bike demand assignment at the lower level. The objective function was defined by a weighted sum of a measure for private car users (total travel time) versus a measure for bike users (total travel distance on bike lanes). A genetic algorithm was developed to solve the bilevel formulation, which included introduction of a special crossover technique and a mutation technique. The proposed optimization will help transport authorities at the planning stage to quantify the outcomes of various strategies for active transport.