LOGNORMAL DISTRIBUTION FOR HIGH TRAFFIC FLOWS

Abstract

The headway probability distributions for very high traffic flows are studied theoretically and empirically. Theoretical analyses show that the lognormal mechanism is applicable to individual headways for drivers in a car-following state. Thus the headway distribution of a traffic stream at high flows should follow the shifted lognormal distribution. Tests on a freeway data set with lane flow rates of 2,500 to 2,900 vehicles per day gave an excellent fit for the shifted lognormal distribution with a 0.3- or 0.4-sec shift. Previously the normal model had been the preferred headway distribution for high traffic flows, but in the present study it did not fit the data very well.