This paper proposes an algorithm which reduces transportation cost while goods are delivered in time in a logistics system. The logistics system assumed in this paper is the system in which multiple cars moves various goods from spatially distributed warehouses to stores. For reducing transportation cost, the car assignment algorithm which allocates goods to minimal cars employs the BF method; routing of each car is modelled as the TSP and is solved by using the genetic algorithm. For evaluating the proposed algorithm, the logistics system is modelled and simulated by using the DEVS formalism. The DEVS formalism specifies discrete event systems in a hierarchical, modular manner. During simulation, each car is modelled as a message and traverses warehouse and stores. When a car arrives at a warehouse or a store, predetermined amount of goods are loaded or unloaded. The arrival time and departure time of cars are analyzed and eventually whether goods are delivered in the desired time bound is verified.