Probabilistic Analysis of Highway Capacity Manual Models

Abstract

“Uncertainty analysis” is a relatively new term in traffic capacity analysis, but it is routine in other areas of civil engineering. Three methods are proposed to examine the uncertainty of Highway Capacity Manual models caused by variance in the inputs: one-at-a-time, point estimate, and Monte Carlo simulation. These methods have different levels of computational intensity and input requirements. They are recommended for three levels of detail in probabilistic analysis, depending on the application. Correspondingly, the results of probabilistic analysis vary from simple upper and lower values to a comprehensive output with mean values, standard deviations, confidence intervals, and delay distributions. The methods were applied to a signalized intersection to demonstrate the application of the three levels of analysis. Field data were from an actuated signalized intersection in Honolulu, Hawaii. Application of the three methods is illustrated in detail. Although the probabilistic method is considered to be more robust than the existing deterministic method, the former is not recommended for application to every project. Options are given for conditions under which the probabilistic method should be applied and the appropriate level of probabilistic analysis.