Fleet sizing is an essential element of chemical supply chain management. It is expensive to purchase and maintain tank cars. Further, the sizing decision is not an isolated one and is closely linked to fleet routing, inventory management and other aspects of logistics and supply chain management. Traditional methods to determine the size of fleets rely on coarse models of the supply chain. In this paper, we present a detailed agent-based simulation model of a multisite chemical supply chain that brings out the intricacies of the tank car fleet sizing problem. The model explicitly takes into account the independent decision making of the different entities as well as the interactions across operations such as replenishment planning and order assignment. Our simulation studies show that different fleet sizes and routing policies can have significant impact on the overall performance of the supply chain including on customer satisfaction and plant performance.