Developing of Estimating Multilane Roundabouts Control Delay Models (Case Study: Kermanshah Roundabouts)

There have been various methods exhibited for roundabout control delay which mostly, predicated on NCHRP572 report(Roundabouts in the United States)and limited to bilinear roundabouts .These methods have been exhibited for unsignalized intersections and predicated on traffic behavior in these intersections. . In this research, the roundabouts with several input and circulating lanes have been simulated and respect to simulating outcomes, the control delay models developed. The model developing and sensitivity analysis done by SPSS22 software with applying different modeling methods as linear regression, non linear regression, neural networks, weight estimation, Optimal scaling and Least Squares Analysis and the best models selected as respect to maximum R-Square of developed models. Verification of the models is performed in comparison with results of models and measured amounts of  field study in 3 roundabouts with different  geometric and traffic conditions. According to this verification, in general, the difference between results of simulating models  and field study measured amounts, was  in range of 0 to 14 percents (in average less than 6 percents). In 3 selected case studies, the average  differences between simulating models  and field study measured amounts  was 3.3 , 4.8 and 6 percentages ,which suggesting the validity of models, proposed for calculating the control delay of multilane roundabouts. In this paper, the  roundabout  control delay models has been acquired  based on behaviour of roundabouts traffic, so there are  more accommodation with real operation of roundabouts. Also these models have been proposed for multilane roundabouts initialy and overcome the limitations of existing models.