Calibration of Mechanistic-Empirical Models for Flexible Pavements Using California Heavy Vehicle Simulators

Abstract

Calibration of mechanistic-empirical models for pavement design is a very complex process. The heavy vehicle simulator (HVS) is ideal for the first step in this calibration process. The short test section can be carefully constructed with well-characterized materials and instrumented to measure the pavement response. The climatic conditions may be controlled or monitored closely, all load applications are known exactly, and perhaps most important, the pavement may be tested until it fails. This overcomes the problems of real pavements, which have uncertainties with regard to materials, loads, and climatic conditions and which are normally designed with a high reliability leading to very few failures. The mechanistic-empirical models of an incremental-recursive computer program, known as CalME, have been initially calibrated using data from 27 flexible pavement test sections tested with the two HVSs owned by the California Department of Transportation. Most sections were instrumented with multidepth deflectometers to compare the measured pavement deflections (at several depths) to the deflections predicted by the mechanistic model. Resilient deflections were compared for the complete time history of each test, and each test was carried to failure in regard to rutting (12.5 mm), cracking (2 m/m2), or both. This involved the calibration of models for changes in layer moduli, including the effects of asphalt fatigue. The comparison of measured and predicted response is essential to ensure that the pavement response is predicted reasonably well by the mechanistic model. Once this was achieved, models for permanent deformation of the individual pavement layers were calibrated against the measured permanent deformation of the layers, again using the complete time history of each test.