This paper tackles a railcar fleet sizing optimization problem in the chemical industry. In some sectors of the chemical industry, railcars are used not only as a means of transportation but also as storage space; thus, it is necessary to consider inventory in the determination of the optimal fleet size. Because of the long lifespans of railcars, this problem is formulated as a long-term capacity expansion problem considering various economic factors. A mixed-integer programming approach to modeling and solving this problem is proposed and a substantial improvement is demonstrated, compared to existing rule-based methods.