An Optimization-Based Qualitative/Algorithmic Approach to Transit Service Planning: Addressing the MBTA Green Line Extension

Abstract

When changes to transit operations are necessary to accommodate changes in the network, demand levels, or agency resources, there is a risk that more obvious solutions (e.g., adjusting headways without changing service patterns) may be unnecessarily detrimental to the quality of the service provided. Complex trunk-with-branches transit networks present both opportunities and challenges for service planning in this context. There may be a large number of potentially feasible operating schemes that could address the problem, with some presenting worthwhile trade-offs that result in much better outcomes for passengers. However, identifying the most promising alternatives from such a large set is a difficult task. While human judgment is a critical part of the process, particularly in the analysis of the most promising solutions, subjectivity from human judgment introduced too early on in the alternative identification process can lead to a suboptimal selection of alternatives.

This research proposes and demonstrates the benefits of a combined qualitative/algorithmic approach to service planning. The proposed approach combines scenario planning, optimization, and qualitative analysis to generate solutions that are robust against uncertainty while providing consistently high passenger level-of-service. An integer optimization program is used to model complex trunk-with-branches transit networks, which outputs a set of service patterns that satisfy various constraints (e.g., passenger capacity, agency resources, fleet composition, infrastructure limitations) while minimizing detriments to passenger level-of-service, namely wait time and transfers. The value of the subsequent qualitative assessment is increased by the use of optimization, as comparisons are being made between high-performance operating schemes.

This approach is applied to the MBTA Green Line to propose service plans after the construction of the Green Line Extension (GLX), which adds an additional two branches to the current four. This extension is occurring during the COVID-19 pandemic, which has resulted in a significant reduction in demand and tightening of agency resources. Both events warrant and facilitate a shift in service patterns. Four phases of post-GLX evolution of demand and resources were considered to illustrate short- and long-term operating conditions. In most cases, plans generated by the qualitative/algorithmic approach included single-car train operations during the peak period to reduce expected wait time relative to the current plans. The alternatives identified may allow post-GLX operations to achieve a pre-pandemic level of service even before agency resources have fully recovered. The research suggests that the qualitative/algorithmic approach can allow service planners to maximize the potential benefits of paradigm shifts such as the GLX.