路外停車場之動態預約定價模式構建

Abstract

在有限之停車空間和固定停車費率下，都市中心路外停車需求大於停車供給。未來政策走向「道路僅供車輛通行使用」，汽機車停車需求由路外的停車場解決。對於停車業者來說，在有限之停車空間和固定停車費率和大量停車需求下，停車業者於尖峰停車需求時段時，往往只能拒絕後續停車需求，就停車業者而言，車位佔有率越高，即表示較佳收益，但對獲利能力未必是最好的。因為若服務消費者為固定費率消費者，當存有潛在高費率等級消費者不被接受，則會造成額外收益損失，即表示在同一車位中，付出成本不變，但收益並非為最佳，這對停車營運效率而言係一項損失。 本研究利用動態規劃及隨機過程，在需求不確定下，推導路外停車場動態預約定價模式。本模式考量各時段剩餘車位數、停車剩餘時間及駕駛人願付價格，求解各時段動態預約價格，以追求最高停車營收。與其他運輸系統（例如，航空、鐵路及公路客運）的傳統座位預約模式不同的是停車需求係由不同停車者的停車延時所決定，進而導致模式推導與求解之複雜度。為簡化問題，本研究分別研提兩個模式：單小時及兩小時。其中，單小時模式係假設所有停車者均恰好停1小時，需停多小時者，必須每小時分別訂位。因此，每一小時的車位均可視為可滿足一位停車者之需求，進而可將停車預約問題依時段分別化約成為數個各自獨立的車位預約問題。至於兩小時模式則假設停車者有兩種需求：停車1小時及停車2小時，故現時段發生的2小時停車需求將與下一時段停車需求同時競爭剩餘停車位，導致分時停車位必須整合求解，而不能分別加以切割求解。為推導模式，假設消費者到達型態為卜瓦松分配，而消費者願付價格則呈韋布分配。本研究設計一簡例，以驗證本模式之可用性。分析結果顯示最佳預約價格會隨著預約剩餘時間的減少而降低及隨著剩餘車位的減少而增加。此外，以總收入而言，本研究所提出的動態定價模式亦明顯優於固定價格模式。

With limited parking space and fixed parking fare, the city Centre off road parking demand exceeds the supply. “Roads are only for vehicle passing” is the future policy for confining vehicles to park in off-street parking lots. However, for the operation of parking lots, excess parking needs are always rejected during rush hours and among them would be highly likely existing high rates consumers. Therefore, higher utility rate does not equate to better profit because the users served might be mixed fare consumers (i.e., lower revenue with the same cost of parking space would decrease operation efficiency). This study proposes dynamic reservation pricing models for off-street parking facilities based on dynamic programming and stochastic processes modeling under demand uncertainties. The proposed model aims to maximize total parking revenue by dynamically determining the reservation price according to available parking lots, elapsed time for parking, and willingness to pay of drivers. Unlike traditional seat reservation models of other transportation systems, such as airline, train and intercity bus services, parking demand is defined by various parking durations (e.g. one hour or two hours) during the reservation period, resulting the closely interaction among customers at different time frames and the complexity in modeling. To simplify the model, two models: one-hour and two-hour models are respectively developed. The one-hour model assumes that all drivers are homogenous with the same parking duration of exactly one-hour. In this model, the parking lot at different time frames can be separately treated, which are similar to the traditional seat reservation of other transportation systems. However, the two-hour model assumes drivers may request for two parking durations: one-hour and two-hour, which the two-hour parking request occurs at previous hour may compete the remaining parking lots with the parking requests at the current hour. To derive the optimal dynamic reservation price, non-homogeneous Poisson process combining by Poisson arrival rate and Weibull distributed willing-to-pay price is also assumed. To demonstrate the applicability of the proposed models, a case study on a numerical example is conducted. The analytical results show that the optimal dynamic reservation price decreases as time elapses but increases as the number of remaining parking lots decreases. In terms of total revenue, the proposed dynamic pricing model performs much better that the fixed reservation price strategy.