This paper proposes Dynamic Multiple Depots Vehicle Routing (DMDVR) to explore the feasible solution of routing transportation between PI-hubs and retailers in the same cluster. The routing solution in this research is constructed by the daily forecasting demand of a commodity crop, Pineapple, from Thailand's northern region. Each route is composed of a starting hub, the number of retailers, and an ending hub. The authors propose Mixed Integer Linear Programming (MILP) to construct the routes considering inventory and truck capacity constraints. Besides, another solution method is proposed by using a heuristic method named “Iterated Random Heuristic”. The empirical results are evaluated by using the total distribution cost and computational time. The routing transportation of this research is based on the daily delivery transportation from PI-hubs to retailers. The results show that Iterated Random Heuristic with Nearest Neighbor Search generates near-optimal solutions within a short computational time.