Phased Transit Service Design Based on Mean-Variance Theory

Transit travel demand has an important characteristic of being stochastic. The decision-makers with different risk attitudes (i.e., risk-neutral, risk-averse, and risk-seeking) will choose various transit service designs in response to the stochastic demand, which seems to be especially critical in the phased transit design. In order to study the impacts of stochastic travel demand on the transit service design, we firstly propose a mean-variance based decision-making approach for decision-makers with different risk attitudes. Correspondingly, the study proposes a model for phased transit corridor design, which can concurrently optimize the line length, transit station location, and headway in each phase. Then, the local decomposition method and optimization theory are adopted to find the analytical solution. Finally, the study compares the transit corridor designs with different design strategies (i.e., full-covered, once-and-done, and phased designs of the transit line) and risk decision-making attitudes. The results indicate: 1) the phased design has a lower expected system cost than the full-covered and the once-and-done designs, and a more robust scheme than the once-and-done design; and 2) risk decision- making attitudes substantially affect the parameters (i.e., the transit length, station location, and headway) of transit system under different design strategies; for example, the risk-averse decision-makers tend to seek denser station locations in the full-covered design, but relatively lower station density in the once-and-done design in phased design. The issue becomes more complex in the phased transit design. This study serves to provide a risk-based decision-making model of phased transit service under the increasing stochastic demand for the government manager and transit operators.