Estimation of gap acceptance parameters for HCM 2010 roundabout capacity model applications

This paper deals with the field data observation and successive application of estimation procedures in order to estimate follow-up headway and critical headway at roundabouts. Average follow-up and average critical headway are two critical parameters in the new roundabout capacity model presented in the 2010 edition of the Highway Capacity Manual (HCM). The HCM 2010 capacity model was developed as an exponential regression model with parameter estimates based on gap acceptance theory. Gap acceptance models are strongly affected by driver behavior and local habits. Follow-up headways can be field measured, while critical headway cannot be obtained directly. The paper mainly aims to check if general suggested values are well-suited also for Italy or not. There have been numerous techniques developed for estimating critical headway. In order to perform a better check of a sample of experimentally observed values, three different procedures are chosen and applied. The first is a quite popular mathematical method based on maximum likelihood technique. The second is a statistical method based on the median of the observed sample distribution. The third is a graphical method known as Raff's method. All these three methods require information about the accepted headway and the largest rejected headway for each driver. Therefore a sample of field data was recorded by digital camera and processed following the instructions suggested by NCHRP. The sample data of critical headway and follow-up headway are gathered in seven selected roundabouts located in Northern Tuscany (Italy). Our first obtained results indicate that the average critical headway is significantly lower than the values recommended by some international references. However, the average follow-up headway is only higher than that recommended for the State of California and it is lower than that recommended by all other international references. Finally, conclusions drawn along with insights for further research developments are suggested.