TL;DR
This paper introduces a dual-toll pricing strategy to mitigate the risks associated with transporting hazardous materials (hazmat) on road networks. By modeling the problem as a bilevel optimization, the authors develop methods to set tolls for both regular vehicles and hazmat carriers, aiming to reduce accident risk while considering user costs and network congestion.
What Problem Does This Paper Address?
The transportation of hazardous materials (hazmat) poses significant risks to public safety and the environment. While authorities aim to mitigate these risks, existing policies often focus on network design (e.g., curfews) or single toll pricing, which doesn't fully account for the interaction between hazmat and regular traffic. This paper addresses the challenge of effectively regulating hazmat transportation by proposing a "dual-toll" system that simultaneously charges tolls to both regular vehicles and hazmat carriers. The goal is to incentivize drivers to choose routes that minimize overall risk, considering that regular traffic congestion can influence hazmat risk.
Methods
The authors formulate the problem as a bilevel optimization model, where the upper level represents the government's decision on toll setting, and the lower level represents the users' (regular vehicles and hazmat carriers) route choices.
- First-Best Dual-Toll Pricing: For scenarios where all links are tollable, they propose a two-stage inverse optimization approach. This involves first finding a minimum-risk flow pattern and then using inverse optimization to derive the corresponding tolls. A key challenge addressed is the non-convexity of the problem, for which they introduce a post-iteration step to guarantee the existence of valid tolls.
- Second-Best Dual-Toll Pricing: When some links are untollable, the problem becomes more complex. They formulate this as a mathematical program with equilibrium constraints (MPEC) and propose a "2-Step EDO heuristic" (Equilibrium-Decomposed Optimization) to solve it. This heuristic decomposes the problem into manageable subproblems and uses iterative line searches.
- Risk Measurement: A duration-population-frequency risk measure is used, which accounts for congestion induced by regular traffic.
- Travel Delay Function: The nonlinear Bureau of Public Roads (BPR) function is used to model travel time, capturing congestion effects.
Datasets
The authors use the Sioux Falls network for their case study. This is a well-known benchmark network in transportation research, with parameters adapted from Suwansirikul et al. (1987). They also consider various numbers of regular and hazmat OD pairs for testing.
Key Insights and Ideas
- Dual-Toll Effectiveness: The dual-toll policy can significantly reduce hazmat transportation risk by encouraging divergent routing for both regular and hazmat vehicles.
- Interplay of Traffic: Regular traffic congestion plays a crucial role in hazmat risk. The dual-toll system leverages this by influencing regular traffic to indirectly mitigate hazmat risk.
- Trade-offs: There's an inherent trade-off between risk reduction, user costs (travel delay and tolls), and revenue. The dual-toll policy aims to find a balance that is acceptable to both the regulator and users.
- Post-Iteration for Validity: The non-convexity of the minimum-risk problem necessitates a post-iteration step to ensure that valid, non-negative tolls can always be found, even with approximate solutions.
- Heuristic Approach for Complexity: For more complex scenarios (second-best pricing or multi-objective optimization), heuristic methods like the 2-Step EDO are essential for finding practical solutions within reasonable computational time.
Key Results
- Risk Reduction: The dual-toll policy consistently reduced hazmat transportation risk, with reductions as high as 63.36% in some scenarios (Table 4).
- User Costs: While risk is reduced, regular vehicles often experience increased travel delay and pay tolls. However, the authors explore variations of the objective function to balance these costs.
- Computational Efficiency: The proposed heuristic algorithms (Modified Frank-Wolfe and 2-Step EDO) were significantly more computationally efficient than exact methods (MILP solvers, LINGO) for solving the complex bilevel problems.
- Comparison to Single-Toll: The dual-toll policy generally outperformed single-toll policies in terms of risk reduction and hazmat travel delay, though it might impose higher costs on regular vehicles.
Limitations and Future Work
- Incompatible Units: The current risk measurement model cannot directly incorporate regular traffic flow rate into the number of people exposed to a hazmat accident due to incompatible units.
- Equity Concerns: The benefits of the dual-toll policy are not always equitably distributed among users, leading to potential user dissatisfaction.
- Future Work:
- Regular Vehicle Exposure: Incorporating regular vehicle exposure into the risk assessment model.
- Equity in Risk Distribution: Developing policies that address equity concerns in distributing risk among different population centers.
- Cost Equity: Integrating cost equity among network users into the decision-making process.
- Nonlinear Travel Delay: While the BPR function is nonlinear, further exploring the impact of more complex or stochastic travel delay functions.
Summary updated