臺北市公共自行車系統營運特性分析

Operations Assessment of the Taipei Bike Sharing System

黃晏珊(Yen-Shan Huang)

指導教授 ： 鍾智林

淡江大學/商管學院/運輸管理學系碩士班/碩士(2015年)

https://doi.org/10.6846/TKU.2015.00456

摘要

因應能源短缺、全球暖化，各城市將自行車公共化而成為自行車共享(Bike Sharing)系統，臺灣各地也陸續推動，其中臺北YouBike以每輛周轉率12次/日居全球之冠。隨著系統服務範圍擴大及使用量增加，甲租乙還的時段性不平衡導致無車可租與無位可還，成為公共自行車營運與調度的核心要素，然而以往較缺乏以大規模缺車與缺位資訊為基礎的探討，故本研究以臺北市YouBike為對象，利用連續六個月、每隔五分鐘擷取的自行車租賃站即時可借車輛與可停車位數據(8,080,128筆)，以統計分析方法進行時間與空間特性的系統分析，除了嘗試建立多元營運評估指標以彌補現有不足外，同時依據租賃站規模與地域特性，檢視現行租賃站設置原則，提供未來設站及既有站點規模調整之參考，並提出相關營運策略建議。

本研究結果顯示：

1.既有營運評估指標僅侷限於自行車上線率，無法充分反應營運績效，本研究另行研議7項指標，包括(1)缺車風險指標，現為26%；(2)缺位風險指標，現為1%；(3)正常營運指標，又稱為可靠度，現為73%；(4)平均缺車與缺位等待時間，缺車為57分鐘、缺位為21分鐘；(5)波動率，現為28%；(6)恢復度，現有為14%；(7)使用率，現有平均為58%。

2.現行設站準則排除既有租賃站600公尺內設置新站，然而本研究顯示，部分租賃站缺車與缺位風險指標高於正常營運指標，此種車站應不受前述限制，並建立既有站點規模調整原則，來改善現有租賃站點。

3.YouBike時間特性包括：(1)平日從7點開始有使用人潮，上下班時間為使用尖峰時段，假日使用人潮則從9點開始，尖峰集中於下午時段；(2)假日使用率高於平日；(3)早上缺車與缺位問題比晚上少；(4)早上波動變化較小，YouBike主要使用時段尤其深夜波動變化大。

4.YouBike空間特性包括：(1)缺車風險熱點多位於交通設施、學校、商業區、休閒娛樂場所周邊；(2)缺位風險熱點部分，以學校周邊居多，其次為商圈、休閒娛樂場所；(3)市中心、交通設施周邊的租賃站不論規模大小都較易有缺車與缺位問題；(4)市中心、交通設施周邊的租賃站由於使用人次較多，導致波動率較大。

5.相關營運策略建議如下：(1)業者運補策略應隨著平、假日尖峰時段不同而變，(2)尖峰時段與市中心地區租賃站具高波動率，以及平均等待時間較長的站點，應列為運補重點，(3)針對缺車與缺位風險熱點，提出激勵機制與自主性調度，以彌補中央調度的不足，(4)平、假日使用率不同，可利用「差別定價」移轉尖峰時段的使用人潮。

關鍵字

公共自行車 ； 自行車共享 ； 大數據分析 ； 時空模型 ； 營運特性

並列摘要

In response to energy shortage and global warming, public bikes have recently been gaining global popularity, including Taiwan. Among the cities with public bike systems, Taipei’s YouBike tops the world in terms of an average daily turnover rate of 12 times per bike. Meanwhile, as more and more users and stations involved in the system, YouBike stations confront such issues as few bikes or few parking slots available due to imbalanced rental demand - a headache of public bike operations and dispatch. Distinct from previous studies, this research collected big data on YouBike with a time span of 6 months for every 5 minutes, or a total data size of 8,080,128. We statistically modeled the spatial and temporal characteristics with respect to various performance indicators. In addition, we examined the exiting disposition criteria for new YouBike stations, and proposed supplementary criteria on the aspects of station scales and locations.

The findings of this study include:

1.For the 7 proposed indicators, the Taipei YouBike system has the following performances: (1) risk of insufficient bikes as 26%; (2) risk of insufficient parking spaces as 1%; (3) reliability as 73%; (4) average waiting time for bikes as 57 minutes and that for parking spaces as 21 minutes; (5) fluctuation rate as 28%; (6) resiliency as 14%; and (7) utility rate as 58%.

2.The current station disposition criteria exclude building new stations within 600 meters of any existing rental station. However, this study shows that some stations suffered from greater risks of insufficient bikes and spaces; in such a case, new stations could (and should) be considered as an exemption from the 600-meter criterion.

3.The temporal characteristics of YouBike include: (1) the number of users increased from 7:00 on weekdays, of which the peak occurred during the commuting hours; for weekends, the number of users increased from 9:00 and reached its top in the afternoon. (2) There were more bike users on weekends than on weekdays. (3) The risk of insufficient bikes or parking spaces was lower in the morning than that in the evening. (4) The fluctuation was small in the morning, but became large in the late evening.

4.The spatial characteristics of YouBike include: (1) The stations near transit facilities, schools, and commercial and recreational areas had greater risk of insufficient bikes; (2) The stations near schools, and commercial and recreational areas had greater risk of insufficient parking spaces; (3) Regardless of the station size, those around downtown and transit facilities tended to have insufficient bikes and parking spaces, resulting in greater fluctuation rates.

5.It is suggested that: (1) The bike operator should adopt different dispatch strategies regarding the peak and off peak periods, and weekdays and weekends. (2) More attention should be paid on the peak periods and downtown stations with greater fluctuation and waiting time. (3) The operator should provide incentives and develop an autonomy dispatch mechanism for stations with greater risk of insufficient bikes and/or parking spaces. (4) “differential pricing” should be considered as the weekday utility rates were different from the weekends.

並列關鍵字

Public Bicycle System ； Bike Sharing ； Big Data Analysis ； Temporal-Spatial Model ； Operational Assessment

參考文獻

4.林緯帆（2013），探討激勵機制設計對多站點自行車租賃服務系統的影響，國立屏東科技大學工業管理系研究所碩士論文。

14.賴淑芳（2012），公共自行車接受度與滿意度研究-以臺北微笑單車為例，運輸學刊 24(3)，頁 1-24。

8.黃仁皇（2010），公共自行車騎乘特性、服務便利性、騎乘滿意度之相關研究─以臺北市微笑單車為例，朝陽科技大學休閒事業管理系碩士論文。

16.謝昉叡（2011），自行車租賃系統佈署調度暨選址最佳化之研究，中央大學土木工程學系碩士論文。

1.余書玫（2009），公共自行車租借系統選擇行為之研究，交通大學交通運輸研究所碩士論文。