台北都會區進城費之最佳收費周界與費率水準之研究

On the Optimal Toll Cordon and Toll Level of In-Town Congestion Charge of Taipei Metropolitan

鄭雅丹(Ya-Dan Cheng)

指導教授 ： 邱裕鈞

國立交通大學/管理學院/交通運輸研究所/碩士(2010年)

https://doi.org/10.6842/NCTU.2010.00459

摘要

本研究目的為發展一個雙層數學規劃模式，來決定台北都會區進城最佳化擁塞收費方案，上層決定最佳化收費方案來最大化社會福利，下層決定最佳化運具選擇(透過羅吉特模式)來最大化道路使用者的效用，以及決定最佳化的路線選擇(透過使用者均衡模式)來最小化旅行成本。針對汽機車持有者，進行大規模的問卷調查，以校估台北都會區的羅吉特模式。共有三個共生變數：旅行成本、旅行時間、收費費率為顯著的變數。

提出兩種最常用的擁塞收費的方案進行最佳化後加以比較：(1)以路線為基礎的方案，(2)以周界為基礎的方案，以路線為基礎的方案於本研究路網中的每個路線上決定最佳化收費費率，以周界為基礎的方案為決定最佳化收費周界(收費中心與收費半徑)與收費費率。使用基因演算法的軟體-Evolver來解決這兩個方案。

根據文獻，設計一個小路網，共有兩個起迄點需求與七條路線，來進行簡例驗證，用來檢驗本研究提出的模式之合理性與適用性。針對某些重要的參數進行敏感度分析，也同時描述系統之間的關係。最後進行台北都會區的實例測試。結果顯示，徵收進城擁塞費後，以線段為基礎的方案可以增加17.76%的社會福利，減少7.55%的總旅行時間。值得提到的是，收費路線不一定根據高交通流量進行收費。最佳化收費水準從7元到100元(收費上限)。同時，以周界為基準的方案可以增加47.09%的社會福利，減少13.03%的總旅行時間，最佳化收費周界的中心點接近南京建國路口，收費半徑為4.5km，收費費率為每次100元，於上午尖峰時間，只有進城要進行收費。顯然地，以線段為基準的方案的績效較以周界為基準的方案來得好。

關鍵字

擁塞定價 ； 雙層數學規劃法 ； 多項羅吉特 ； 基因演算法

並列摘要

This study aims to develop a bi-level mathematical programming model to determine the optimal in-town congestion charge scheme in Taipei metropolitan. The upper level is to determine the optimal congestion charge scheme for maximizing the social welfare (SW); while the lower level aims to determine the optimal mode choice (by a Logit model) for maximizing road users’ utility and the optimal route choice (by an user equilibrium model) for minimizing their travel cost. A large-scale questionnaire survey is conducted on car or motorcycle owners in Taipei metropolitan for the estimation of the Logit model. Three generic variables: travel cost, travel time and toll level are significantly estimated.

Two types of congestion charge schemes are respectively proposed and compared: the Link-based scheme and the Cordon-based scheme, where the Link-based scheme determines the optimal toll level on each link in the study network and the Cordon-based scheme determines the optimal toll cordon (center and radius) and toll level. An genetic algorithm software – Evolver is used to solve the both schemes.

An exemplified example with a small network of 2 OD pairs and 7 links is designed to examine the reasonability and applicability of the proposed models. Sensitivity analysis on some important parameters is also performed to depict the relationships among the system. At last, a case study on Taipei metropolitan is conducted. The results show that the Link-based scheme can increase the social welfare by 17.76% and decrease total travel time by 7.55% by levying in-town congestion toll. It is worth noting that the tolled link may not be those which originally bear high traffic flows. The optimal toll levels range from NT$7 to NT$100 (upper toll limit). Meanwhile, the cordon-based scheme can increase social welfare by 47.09% and decrease total travel time by 13.03%. The optimal center and radius of toll cordon is located near the intersection of 10th Ave. and 7th Blvd and 4.5 km with a toll level of NT$50 per entrance during the morning peak hour. Obviously, the Cordon-based scheme performs better than the Link-based scheme.

並列關鍵字

congestion pricing ； bi-level mathematical programming ； Logit model ； genetic algorithm

參考文獻

5. 陳佛鐘(民98)「徵收進城擁擠費對私人機動車輛使用行為之影響分析」，國立交通大學交通運輸研究所碩士論文。

10. Brownstone, D., Ghosh, A., Kazimi C. , Golob T.F., and van Amelsfort D. (2003). "Drivers’ willingness-to-pay to reduce travel time: evidence from the San Diego I-15 congestion pricing project." Transportation Research Part A 37(4): 373-387.

11. Chu, C. and Tsai, J. (2008). "The optimal location and road pricing for an elevated road in a corridor." Transportation Research Part A 42(5): 842-856.

12. de Groot, J. and Steg, L. (2006). "The role of value orientations in evaluating quality of life consequences of a transport pricing policy." Transportation Research Part D 11(2): 160-165.

13. Larsen, O. (1995). "The toll cordons in Norway: an overview." Journal of transport geography 3(3): 187-197.