Assessing Speed of Passenger Cars at Urban Channelized Right-Turn Roadways of Signalized Intersections

Abstract

This paper aims to assess passenger cars’ speed on urban channelized right-turn roadways at signalized intersections under the free-flow traffic conditions. Stepwise regression and artificial neural network (ANN) were used to develop speed models. Geometrics data for twenty-one channelized right-turns at seventeen signalized intersections in urban areas were collected. Analysis results showed a linear relationship between right-turn roadway radius and speed. Other geometric parameters were found statistically significant at the 95% confidence level. The coefficient of determinations ( $${R}^{2})$$ was found to be 87.9%, 94.2% and 97% for the average speed, 85th percentile speed regression models and ANN model, respectively. Results obtained from the comparison analysis showed that equations developed for horizontal curves are recommended with caution for predicting speed on right-turns due to differences in the range of radius values. The regression equation $$(\hbox {Eq.}~8: V_\mathrm{avg} =10.904+0.314\,R+2.286\, \mathrm{Ds}+0.046\,\mathrm{SD}+0.288\,W)$$ developed in this study offers relatively better prediction accuracy compared to all other previous models. Besides, the ANN model outperformed all other models including the developed regression equation of this study. Care should be taken when using the findings of this study as they are exclusive to the conditions and locality under which this study was performed. To generalize the results, further validation is needed using data from different localities and drivers with diverse behaviors. In addition, models of this research are for speed predictions if curves’ radii are within the range of data used (10–135 m) in order to avoid extrapolation.