Total-cost Frameworks for Multimodal and Environmentally Preferable Streets

Efforts to reduce the environmental impacts of transportation have often overlooked efficiencies obtained by considering the relevant engineering and economic aspects of roadway infrastructure as a system. Concerns over greenhouse gas emissions (GHGs), degrading infrastructure in the face of limited maintenance budgets, and the declining use of private automobiles all motivate this more comprehensive approach investment prioritization. Here, a three-part framework is presented consisting of a multi-segment discrete pavement management system (PMS), a roadway use-phase microsimulation that evaluates lane allocation between modes, and a spatial-Bayesian bicycle volume model to synthesize necessary but generally unknown street-specific bicycle usage from available data. A more complete assessment of the costs and benefits of various road-use scenarios is quantified with respect to user costs, agency costs, and GHG emissions.

The PMS presented here incorporates GHG emissions using a multi-objective evolutionary algorithm to produce a Pareto-set of discrete long-term maintenance plans. By using deterioration models and cost estimates from the Virginia Department of Transportation (VDOT), comparisons with historical practice are also possible. VDOT has historically relied on corrective maintenance in validation area, but an optimized management plan could achieve the same average pavement condition with 60% of the cost and 50% of the GHG emissions. Solutions from a network-wide optimization also dominated aggregated single-segment solutions, justifying the computational cost of the method. The use phase still accounts for the majority of roadway impacts, however, so the second component of this work employs microsimulation and a vehicle emissions model to consider the coupled effects of pavement condition and vehicle fuel consumption, as well as travel time costs, by mode.

Both microsimulation and a probabilistic analysis of all feasible combinations of travel lanes, bicycle lanes, and curb parking show mobility reductions on road segments of insufficient width for heavy vehicles to pass bicycles without encroaching on oncoming traffic. This delay is positively correlated with uphill grades and increasing traffic volumes and is inversely proportional to total pavement width. A high bicycle mode share is therefore negatively correlated with total costs and emissions for lane configurations allowing motor-vehicles to safely pass bicycles, while the opposite is true for configurations that inhibit passing. As a result, curb parking exhibits spatial opportunity costs well in excess of feasible hourly use fees when the parking lane could have been devoted to bicycle mobility, even before considering safety benefits often used to justify such conversions. The results are sensitive to street-specific bicycle mode share, however, and these data are not commonly known without dedicated field observation, thus precluding network-wide analysis.

The final component of this work employs Markov-chain Monte Carlo sampling to address the temporal factoring of bicycle count observations into annually representative posterior distributions of critical parameters on direction roadway links that have been sampled. A novel spatial factoring method employs Bayesian updating to combine uncertain volume estimates from a regional travel demand model with the temporally factored intermediate distributions by applying a stochastic edge correlation matrix. For a small city in the United States with some volunteer bicycle counts and no permanent counting infrastructure, the model is able to estimate edge-specific bicycle usage network-wide with large but well-characterized uncertainty.

Overall, the results provide quantitative evidence that efforts to reallocate limited pavement space to bicycles, like those being adopted in several US cities, could appreciably reduce costs for all users and illustrate the value of a total cost approach to investment optimization in evaluating these decisions. Future work will continue to address the data gap between bicycle and motorized transportation with respect to travel demand and safety.