Constructive heuristics for the Mixed Capacity Arc Routing Problem under Time Restrictions with Intermediate Facilities

• A Capacitated Arc Routing Problem variant with application to waste collection is investigated.
• The problem models a mixed network, intermediate facilities, and route time restrictions.
• Constructive heuristics are generalised and evaluated on new realistic benchmark sets.
• Performance of constructive heuristics differ between existing and new benchmarks.
• Constructive heuristics struggle to minimise vehicle fleet size.

The Mixed Capacity Arc Routing Problem under Time Restrictions with Intermediate Facilities (MCARPTIF) is an extension of the Arc Routing Problem under Capacity and Length Restrictions with Intermediate Facilities (CLARPIF) with application in municipal waste collection. This paper evaluates four constructive heuristics capable of computing feasible solutions for the MCARPTIF with a primary objective to either minimise total cost or to minimise the fleet size. The heuristics were adapted from Path-Scanning and Improved-Merge for the Mixed Capacitated Arc Routing Problem, and compared against two Route-First-Cluster-Second heuristics for the MCARPTIF. The objective was to identify the best performing heuristic for application purposes. In practice, the CARP is often solved for real-time or near real-time decision support. Computational time required by the heuristics was thus also evaluated. Identifying the best heuristic proved difficult due to a lack of realistic MCARPTIF benchmark sets, with the two CLARPIF sets predominantly solved in the literature not resembling actual waste collection instances. Route-First-Cluster-Second heuristics, linked with a new vehicle reduction heuristic performed the worst on the two CLARPIF sets, yet performed the best on new waste collection sets taken from the literature and introduced in this paper. Improved-Merge performed the best on two existing CLARPIF sets and on a realistic set with Intermediate-Facilities incident with the vehicle depot, but struggled on all other sets and in minimising fleet size. Path-Scanning was the most robust heuristic, performing reasonably well on all benchmark sets and both objectives. Results further show that due to the high computational time of one of the Route-First-Cluster-Second heuristics, which was only exposed on realistically sized sets, the slightly worse version is the best alternative when real-time support is required for waste collection applications.

Figure optionsDownload as PowerPoint slide