考量預售折扣之鐵路客運平行班次座位配置模式

Abstract

鐵路成為運輸業中的重要運具已有許多年，而其永續性更受到許多政府與組織的青睞。也因此透過技術的發展與資源的投入，許多鐵路系統在速率以及容量方面陸續獲得提升。然而，鐵路系統在營運面上的管理技術卻相對較少受到重視，尤其不像航空公司擁有精緻的營收管理系統來大幅增加營收，鐵路公司對於透過使用營收管理來進行市場區隔以及差別訂價時的態度則較為保守。另外，目前也沒有太多的研究去評估營收管理在鐵路的效能，並進一步發展更合適的決策模式。 本研究聚焦於發展一個具有多重起訖區間之鐵路系統的座位控管模式，而其中一個重要的複雜度在於多重起訖區間必須共用系統的容量。另外，由於鐵路預購折扣機制的實施，以及出現平行班次的情況，使需求在選擇行為上產生異質性，使得本研究的營收管理問題比起傳統的座位配置模式更加具備挑戰性。 本研究將發展數個靜態的線性規劃模式以產生座位控管決策，來因應具有異質特性的鐵路需求。並設計各種不同的情境，模擬不同模式之控管決策在不同情境下的收益結果，藉以驗證所發展模式和求解方法的可用性。 最後，本研究以台灣高鐵之資料做數值實驗，根據數值測試的結果，同時考慮需求在預售折扣、平行班次之異質性的平行模式，是最具有優勢的模式，藉以證明平行模式有助於改善鐵路系統的收益情形。

Railways have been an important mode of transportation for years, and it is mostly preferred by many governments and organizations from the aspect of sustainability. Thus, the speed and capacity of many railway systems has been upgraded, due to the development of technology and the investment of massive resources. However, relatively less attention has been paid to the management skills and techniques of operating railway systems. In particular, unlike the airline industry with sophisticated revenue management (RM) systems to generate considerable extra revenue, the railway industry tends to conservative about implementing RM to exercises market segmentation and price differentiation. Meanwhile, not much research work has been done to evaluate the effectiveness of RM in the railway industry and develop suitable models to provide better decision support. This study focuses on the seat inventory control for a railways system with multiple original-destination (OD) pairs sharing the train system capacity. Due to the instinct stochastic feature of railway demand, this decision model is more complicated when compared with the traditional simple seat allocation model. In particular, the behavior of demand may be more than one choice due to the discount mechanism and parallel trains, makes this RM problem much more challenging than traditional seat allocation models. This study plans to develop several static linear programming (LP) models that can generate the control decision capable of dealing with heterogeneous demand. Moreover, different scenarios are designed in the simulation experiment to validate the applicability of the developed models and solution approaches, based on these simulated revenues. Finally, based on the numerical experiment from Taiwan High Speed Rail Company (THSRC), the parallel model, which considers heterogeneous demand, has the best revenue outcome. This result proves that the parallel model is helpful for improving the revenue in a railway system.