佈設左轉專用車道之對於號誌化路口安全之影響

Abstract

設置左轉專用道之目的在於安全地將直行車流與左轉車流分離，並透過偏移設計使得左轉駕駛可獲得更良好的視距，避免左轉車與對向直行車發生衝突，然國內目前缺少左轉專用道之設計規範，導致國內各地出現不同設計方式之左轉專用道，本研究試圖量化不同設計方式之左轉專用道之安全效益，並使用三個左轉專用道設計變數，其分別為主線連續性、主線順暢性、順暢性主導之左轉車偏移量，來區分研究對象中的五個不同設計方式之左轉專用道，將此三項變數與其他路口資料及事故數設為自變數與應變數，分別針對左轉穿越側撞、左轉側撞、左轉擦撞、直行擦撞及追撞，五種碰撞型態建立傳統迴歸模型以進行事前事後分析。模型輸出結果顯示，三種變數對於左轉設計之安全效益評估具有一定可信度，表示當左轉設計為附加式，且偏移程度越高時，可減少左轉穿越側撞、左轉側撞、左轉擦撞與直行擦撞，而追撞則會因爲擁有主線連續性而上升，但亦會因為其餘兩變數而下降。由於地方主管機關在沒有充分證據下無法評估效益，影響其設置意願，故本研究後續透過適合度指標選擇較有解釋力的模型，並使用該模型之三個左轉專用道設計變數之估計值估算事前事後可減少之事故數，並將其轉換為事故成本，加入工程成本做十年經濟效益分析，顯示將任一左轉專用道進行改善，皆符合B/C > 1，且多數案例僅不到一年即可達到收支平衡，表示安全所帶來之效益遠大於工程成本。

The aim of installing left-turn lanes is to safely segregate straight and left-turning traffic by enhancing visibility of left-turn road users to prevent conflicts with opposing through traffic. In Taiwan, a lack of design standards results in varied left-turn lane designs. This study attempts to quantify the safety benefits of different design for left-turn lanes, using three design variables: mainline continuity, mainline smoothness, and smoothness-dominant left-turn offset. These three variables, along with other intersection data and accident counts, are set as independent and dependent variables, respectively, to establish classical regression models for five collision types: left-turn crossing-side impact, left-turn side impact, left-turn sideswipe, through sideswipe, and rear-end collision, for pre post analysis. The output results of the models show that the three variables have a certain level of credibility in assessing the safety benefits of left-turn designs. When the left-turn design is additional and the degree of eccentricity is higher, it can reduce opposite left-turn collision, left-turn side collision, left-turn sideswipes, and through sideswipes. However, rear-end collisions may increase due to having mainline continuity but decrease due to the other two variables. Given the authorities'' challenge in assessing benefits, this study chooses an interpretable model using fitness indicators. It estimates the pre- and post-accident reduction by converting the three left-turn lane design variables into accident costs and adding them to engineering costs for a ten-year economic analysis. The findings reveal that improving any left-turn lane achieves B/C larger than 1, underscoring the safety benefits outweighing engineering costs.