Study On The Traffic Characteristics Of The Quadruple Left-turn Lanes At Signalized Intersection

The traffic operation of multiple left-turn lanes can improve the left-turn capacity at signalized intersections. However, as the number of left-turn lanes increases, the more serious interference between vehicles occurs at the intersection which reduces the traffic efficiency and safety. At present, the studies of this field mainly focus on interactions with double or triple left-turn lanes. By contrast, the research on the operational characteristics of vehicles under more complex traffic organization forms, e.g. Quadruple Left-turn Lanes(QLL) at signalized intersection is very rare.In this study, the operating characteristics of QLL vehicles during left turn process were analyzed comprehensively. Specifically, subjective initiative of drivers and randomness of lane selection behaviors were considered. Various statistical methods and regression analysis were adapted to probe the main factors influencing vehicles’ lane-changing rate during left turn and lane-changing interval selection. As a result, the lane-changing rate model and the lane-changing interval selection model were built. Based on these two models, the capacity calculation model was constructed, which could successfully apply to the quadruple left-turn lanes group. The specific content of this paper is structured as follows:(1) The influencing factors of traffic flow operation in quadruple left-turn lanes of signalized intersection were analyzed. The characteristics of QLL’s traffic flow were examined and summarized preliminarily. Based on comparison analysis of traffic survey data, the differences of saturation flow rate among QLL, single left-turn lane, double left-turn lanes and triple left-turn lanes were revealed. Through the number of vehicles statistics of the QLL entrance and exit at signalized intersection as well as the research of distribution characteristics of lane changing position, the pattern of vehicles Lane Changing During Left-turn(LCDL) was inspected. Lastly, the fluctuation of the left-turn vehicles’ headway which was under the effect of lane-changing behavior was studied. And the saturation flow rate of QLL with and without the influence of lane-changing behavior was compared. Therefore, the lane-changing behavior could influence the performance of QLL, which was analyzed and demonstrated from different angles.(2) The lane-changing rate model of QLL group was established. By considering the differences between vehicles’ lane-changing intention, the characteristics of vehicles’ lane-changing behavior performance were also analyzed. Then, the various factors which may affect vehicles’ lane-changing behavior during left turn were examined. For the release process of queue and random arrival, the multiple linear regression method was chosen to build the lane-changing rate models respectively. It was found that intersection exit weaving distance, traffic flow rate and percent heavy vehicles had significant impacts on LCDL rate. At last, the results of goodness of fit and instance validation showed that the model could not only fit the statistical data but also have a good performance in the prediction of new data.(3) The lane-changing interval selection model of QLL vehicles was established. From the microscopic angle, this chapter studied the lane-changing interval selection of individual vehicles. Firstly, the lane changing behaviors at the intersection were divided into two categories: one was lane changing from Inside to Outside(I20) and the other was lane changing from Outside to Inside(O2I). Then according to the lane-changing behavior in different intervals, the vehicles of each category were divided into two groups: LCDL vehicles and Lane Changing After Left-turn(LCAL) vehicles. Accordingly, the research of microscopic model was converted into a binary problem of lane changing interval selection. Next, binary logistic regression method was adopted to build I2 O and O2 I lane changing interval selection models, respectively. The methods of marginal effects and adjusted predictions were used to interpret the established model, and quantitatively analyze each independent variable. Finally, the model was verified by goodness of fit and tested with field data. The results showed the constructed model was robust.(4) According to the lane-changing rate model, a calculation formula of adjustment factor for lane-changing rate was deduced. Hence, the traditional capacity calculation model in HCM 2010 was revised, which is suitable for QLL group at the signalized intersection.Finally, the main research contents, research results and deficiencies of the whole dissertation were presented. Future research directions were also discussed.