The effect of approach gradient, weather conditions and queue position on discharge headways at signalized intersections in the city of St. John's

Abstract

This experiment was designed to determine the effect, if any, of approach gradient, weather conditions and queue position on discharge headways at signalized intersections in the City of St. John's. -- Many of the factors that affect vehicle operation at signalized intersections were examined through an extensive search of the existing literature. These factors were considered during the selection process when the intersection approaches on which data was collected were selected. -- A suitable mathematical model was selected and the appropriate hypotheses were generated. -- Headway data was collected during the Spring of 1986 on five intersection approaches having a range in gradients from -7.2% to +7.2%. -- This data was analyzed using the SPSS/X statistical software package. It was determined that the headway data that was originally collected did not meet the assumption of Normality. Accordingly, the data was transformed using a logarithmic transformation to produce a more Normal distribution, and was analyzed again. -- The factorial experiment that was performed indicated a statistically significant interaction between weather conditions and approach gradient. However, subsequent attempts to formulate equations quantifying the relationship between approach gradient and vehicle headways at each level of the weather factor were not successful. The regression procedure produced equations with very low values of the coefficients of determination. At the same time, the significance level of the F-test procedure results gave indications that a strong linear relationship existed between approach gradient and discharge headways. -- It was concluded that the significant result of the F-test procedure was a result of the large number of residual degrees of freedom resulting from the very large database. -- Consequent examination of the data revealed an error in the database that had been used during those regression procedures. However, when the regression procedures were performed on the corrected database, the results were largely unchanged. -- It was, therefore, concluded that there is no practical quantifiable relationship between approach gradient and discharge headways on the approaches to signalized intersections in the City of St. John's. -- However, a practical and statistically significant relationship was developed between the elapsed time from the start of the green phase, a quantity which is related to discharge headway, and vehicle position in the queue, and an appropriate prediction equation was developed that accounts for 93.5% of the variability in the data.