Reasonable transportation risk models are conducive to achieving the green reform of hazardous material logistics industry. However, existing multi-depot vehicle routing programming models for hazardous material transportation may result in overemphasis on either global risk or local risk. To overcome such shortcomings, we develop two novel two-stage programming models that consider weight variety in risk measures. The ordered weighted averaging risk-based model effectively reduces both the global risk and the maximum local risk with respect to the weight distribution in the aggregation process of local risks, and the state variable weight risk-based model helps reduce the global risk and the maximum local risk based on the variable weights associated with local risk values. Furthermore, we design a constraint reduction mechanism and a variable neighbourhood search-based hybrid parallel genetic algorithm to handle the proposed models, such that they could rapidly reach the near-optimal solution using multiplication processors. Experimental investigations demonstrate that the proposed models achieve a good balance between overall risk and local risk, and proposed algorithm can obtain a satisfactory approximate solution within an acceptable time frame.
- hazardous materials transportation
- hybrid parallel genetic algorithm
- Multi-depot vehicle routing
- ordered weighted averaging
- state variable weight