Regression-Based Models for Bus Dwell Time

Bus dwell time contributes to much of the bus travel time and headway variations, meanwhile influence bus service reliability and restrict bus stop capacity directly. But in terms of how passenger characteristics affect bus dwell time under different bus stop types, many of researches found in the literature are not enough. Firstly numbers of passenger alighting/boarding, bus arrival and depart rate were collected via videos, including linear and non-linear bus stops, also dwell time distribution was test respectively. Secondly quantitative analysis of dwell time vs various variables have been done. Based on field data, a multiple linear regression model was proposed and estimated. At last the dwell time equations were compared with equations found in the literature, meanwhile comparison between measured value and theoretical value has been done to test the accuracy of the regression model. The results show dwell time follow Erlang-distribution, and in non-linear bus stop average boarding time is 1.5~3.2s higher than linear bus stop. Among all the regression models, using the maximum of boarding and alighting passengers as independent variable show the most accuracy.