Improvement of axle load spectra characterization by a mixture of three distributions

Traffic is one of the most important variables used in pavement design methods. There are currently two methods for characterizing wheel load repetitions: equivalent single axle load (ESAL) and load spectra, which is a more precise characterization of traffic but relies on the same input data used to calculate ESALs. Although the load spectra are described using a bimodal mixture distribution, samples taken in Colombia indicate that the load spectra exhibit an intermediate region that can be modelled with an additional distribution. This research proposes the use of a mixture of three distributions: two log-normal distributions for the description of the peaks and a normal distribution for the intermediate region, finding a closed statistical solution for the formal evaluation of its properties, and deduces the expression of the nth moment that facilitates obtaining important features of the proposed distribution. In a case of application in two geographic regions of Colombia, with more than 53,000 tandem and tridem observed axles, it was found that the proposed model effectively improves the characterization of the load spectra.