Optimizing Capacitated Vehicle Scheduling with Time Windows: A Case Study of RMC Delivery

TL;DR

This paper addresses the complex problem of optimizing ready mixed concrete delivery schedules considering multiple constraints and objectives, introducing novel algorithms with proven efficiency and effectiveness.

Contribution

It provides a theoretical foundation, proves NP-Completeness, and develops two new algorithms for the RMCDP, improving solution feasibility and computational efficiency.

Findings

Graph-based greedy algorithm with polynomial time complexity

Heuristic Priority Algorithm effectively handles dynamic and multi-objective aspects

Proposed algorithms outperform state-of-the-art methods in feasibility and speed

Abstract

Ready Mixed Concrete Delivery Problem (RMCDP) is a multi-objective multi-constraint dynamic combinatorial optimization problem. From the operational research prospective, it is a real life logistic problem that is hard to be solved with large instances. In RMCDP, there is a need to optimize the Ready Mixed Concrete ( RMC) delivery by predetermining an optimal schedule for the sites-trips assignments that adheres to strict time, distance, and capacity constraints. This optimization process is subjected to a domain of objectives ranging from achieving maximum revenue to minimizing the operational cost. In this paper, we analyze the problem based on realistic assumptions and introduce its theoretical foundation. We derive a complete projection of the problem in graph theory, and prove its NP-Completeness in the complexity theory, which constitutes the base of the proposed approaches. The first approach is a graph-based greedy algorithm that deploys dynamic graph weights and has polynomial time complexity. The second approach is a heuristic-based algorithm coupled with the dynamic programming and is referred to as Priority Algorithm. This algorithm is carefully designed to address the RMCDP dynamic characteristic, and satisfies its multi-objectivity. In comparison with the state-of-arts approaches, our algorithm achieves high feasibility rate, lower design complexity, and significantly lower computational time to find optimal or very slightly suboptimal solutions.