國際航空貨運近年來成長迅速，亞太地區成長量尤為突出。面對貨運需求的榮景，航空公司盤櫃設備需求量將隨之大增。盤櫃設備為航空公司長期投資之作業性設備，其規模必須作適當的採購規劃，但目前訪談之航空公司並無適當的決策協助工具。本研究透過對實務問題之了解，利用時空網路建構盤櫃設備規模最適化之數學規劃模式，其中亦考慮短期調度之限制，並建議利用分枝限界法進行求解。文中針對國內某航空公司既有之全貨機歐洲網路作為實證對象，並進行參數之敏感度分析。從結果中發現該實證案例之盤櫃規模，約為配置飛機之單位容量的3.61倍。而航空公司預估之貨運承載率愈低，準備之盤櫃數量也隨之小幅下降；貨載承重比例愈低，準備的盤櫃數量呈非線性的微幅增加。而貨櫃基本用量的減少與調度機場安全存量的減少，都將使購置成本呈線性遞減。

The volume of international air cargo has rapidly grown in recent years, especially obvious in the Asia Pacific area. The required quantities of unit load devices (ULDs) will also increase when airlines confront the booming trend of air cargo services. ULD is the equipment that airlines have to purchase as a long-term investment. The fleet size problem of ULDs ought to be addressed in a suitable planning of purchase. However, the studied company in this research did not have any tool to assist the decision making on ULD preparation. Based on the detailed exploration in practical operations, this study exploits the time-space network to formulate a mathematical programming model for optimizing the fleet scale of ULDs in consideration of short-term repositioning constraints. We suggest to obtain the exact solution by applying the branch-and-bound algorithm. A case extracted from the freighter network of European services of an international airline in Taiwan is tested. According to the test results and sensitivity analysis, we find that the scale of ULDs is almost 3.61 times the slot capacity of the deployed airplanes. The lower the estimated load factor, the less the decrease in the number of prepared ULDs. As the ratio of cargo density becomes lower, the number of prepared ULDs is nonlinearly increased slightly. Furthermore, the purchase cost have a linear decrease when the basic usage of containers and inventory of the main airport are decreased, respectively.