Tailored Priority Allocation in the Bottleneck Model with General User Heterogeneity

We propose to enhance the efficiency of road bottlenecks by strategically implementing metering-based priority (MBP) schemes. Under MBP, a portion of the bottleneck capacity is reserved for priority users but made available for nonpriority users when no priority users are queuing. Previous studies have found that MBP is Pareto-improving regarding individual trip costs with homogeneous users, but its effectiveness becomes ambiguous when users have heterogeneous scheduling preferences. To address this, we consider a finite number of user groups with group-specified scheduling preferences. The design of optimal MBP schemes to minimize the total trip cost is formulated as a bilevel problem, allowing for varying fractions of priority users across groups. Under the identified conditions, convex optimization algorithms can be used to solve optimal MBP schemes. When these conditions are not met, we propose a general solution framework to find solutions with satisfactory accuracy. We study the theoretically optimal system efficiency achievable by MBP through numerical simulations. We also explore the benefits of integrating MBP with other travel demand management policies, such as high-occupancy vehicle lanes. Importantly, the implementation challenges of MBP schemes are also discussed, particularly the difficulty of distinguishing users based on their preferences. We investigate the efficiency of implementing optimal MBP schemes in an aggregated manner, emphasizing the significance of selecting appropriate aggregating patterns. We also propose a type of heuristic MBP scheme that ensures that nonpriority users' departures can be unaffected by MBP. Such schemes are Pareto-improving and remarkably do not require observing individual preferences. These heuristic schemes can be decentralized by assigning priority status through pricing in certain cases. Numerical results demonstrate that the heuristic approach achieves comparable efficiency levels to optimal MBP schemes in considered scenarios.