The Capacitated Electric Vehicle Routing Problem (CEVRP) is one of the most important tasks in modern logistics areas. The aim of the CEVRP is to optimize the routing of an electric vehicle which contains limited electric energy and customer service. This paper presents the modification of the simulated annealing (SA) algorithm for solving the electric vehicle routing problem. In this paper, adaptive clustering is applied to generate the high quality of the initial population. Afterward, the simulated annealing algorithm is separated into two halves according to the SA accepted criterion, the first half called the positive filter technique, and the second half called the negative filter technique. The proposed idea is presented to enhance the exploration and exploitation search capabilities of the present algorithm. The performance of the proposed algorithm is tested with the small and medium size of the standard benchmark dataset. The results on seven sets of benchmarks test set show that the proposed algorithm is very effective in solving the electric vehicle routing problem.