Study of GVNS-RVND hybrid algorithm for the split load vehicle routing problem and its implementation

The focus of this article is the study of the GVNS-RVND Hybrid algorithm to solve the Split Load Vehicle Routing Problem with Simultaneous Delivery and Pickup (SLVRPSDP) problem. The GVNS-RVND Hybrid algorithm consists of four main stages, namely the grouping of customers using the split load method, the initial formation using the Clarke and Wright Algorithm, and the perturbation stage. The random solution obtained from the perturbation stage is corrected using the Randomized Variable Neighborhood Descent (RVND) Algorithm. The implementation of the GVNS-RVND Hybrid algorithm is designed using the Borland Delphi 7.0 programming language. The output of the program is the route formed along with the distance traveled and the visualization of the route formed. The results obtained on the SLVRPSDP problem in the real case at the branch of the China Express company, the GVNS-RVND Hybrid algorithm produces a solution of 309.53 km in the real case of 24 customers with a gap value of 3.3% from the Tabu Search solution algorithm, and produces a solution of 431.79 km in the case of real 30 customers with a gap of 1.14%. Based on the data obtained from this study, it can be said that the solution produced by the GVNS-RVND Hybrid algorithm in both cases has a better solution when compared to the Tabu Search algorithm.