Journal of Theoretical and Applied Information Technology 10 th July 2015. Vol.77. No.1 © 2005 - 2015 JATIT & LLS. All rights reserved . ISSN: 1992-8645 www.jatit.org E-ISSN: 1817-3195 125 A WATER FLOW-LIKE ALGORITHM FOR CAPACITATED VEHICLE ROUTING PROBLEM .1 SUHAILA ZAINUDIN, 2 MOKHTAR MASSOUD KERWAD, 3 ZULAIHA ALI OTHMAN 1,2,3 Centre for Artificial Intelligence Technology, Faculty of Information Science and Technology, University Kebangsaan Malaysia, 43600 Bangi, Selangor Darul Ehsan, Malaysia. Email: 1 suhaila.zainudin@ukm.edu.my, 2 kerwadd@yahoo.com, 3 zao@ukm.edu.my. ABSTRACT The goal of the capacitated vehicle routing problem (CVRP) is finding a useful vehicle route which is a fundamental issue of logistic management. CVRP consists of finding the optimal routes for distributing various items between customers and depot by a fleet of vehicles at a minimum traveling distance without exceeding the capacity of each vehicle. Among many proposed schemes, meta-heuristic algorithm is a well-known optimization method for solving non-deterministic polynomial-time problems. The population- based meta-heuristics has shown the ability to obtain an excellent solution in many domains but consumes time. This is due to the nature of the algorithms that have fixed a number of the solutions, so they suffer from high computation times to reach the solution. In this paper a new approximation algorithm named Water Flow-like Algorithm (WFA) is proposed to tackle CVRP. It is inspired by the natural behavior of water flowing from a higher to a lower level, which is self-adaptive and dynamic based on population sizes and parameter settings. The performance of the proposed algorithm is evaluated using 14 benchmark datasets. The result shown performance of the proposed algorithm is comparable with other recent enhanced algorithms. Therefore, it indicates that WFA is a potential alternative to solve the CVRP using other enhancement on the algorithm. Keywords: Combinatorial Optimization, Vehicle Routing Problem, Meta-heuristics, Dynamic solution, Water Flow-like Algorithm. 1. INTRODUCTION Capacitated Vehicle Routing Problem (CVRP) is a combinatorial optimization problem that has received considerable attention recently because finding an efficient vehicle is an important issue of logistic management [1], [2]. Hence, effective transportation management can increase competitiveness and reduce the environmental impact of organizations due to more optimal routes and shorter distances. CVRP was first designed by [3] and concerns the design of a set of least cost routes for a set of vehicles to service a number of customers with identified demands [4]. The problem combined two Non-deterministic Polynomial-time (NP) hard problems, the traveling salesman problem and the bin-packing problem (BPP), hence CVRP is also NP-hard [5]. As an NP-hard problem, a number of approaches were proposed to solve it, and they vary in terms of complexity, efficiency and their ability to solve the problem. In general, these techniques can be categorized into exact and approximate algorithms. Although exact algorithms such as dynamic programming [6], branch and bound [7], branch and cut [8], and branch and price [9] are suitable for small size instances, as problems become large and heavily constrained exact methods are no longer suitable to solve the problem and often fail to get an optimal solution owing to the computational time required [10]. On the other hand, approximate algorithms can obtain satisfactory solutions in competition time, but there is no guarantee to find global optimal solutions [11]. These algorithms can be classified as classical heuristics and meta-heuristics [12]. Some classical heuristics are saving heuristics of Clarke and Wright [13], and sweep algorithm has been proposed by [14]. Meta-heuristic combines basic heuristic strategies in higher level frameworks in order to explore search space more efficiently. Since these algorithms have good abilities to explore search space and are unquestionably efficient to get from a local optimum, they are