A sensitivity analysis of a performance model is conducted. The performance model evaluated was developed from AASHO road test data, and it uses pavement surface roughness as the distress criterion. In order to evaluate the sensitivity of predicted pavement performance to various design factors, a factorial experiment was established assuming a three-layer model of the pavement structure. Eight different factors were considered in the development of the factorial experiment: (1) initial present serviceability index (PSI), (2) asphalt concrete modulus, (3) asphalt concrete thickness, (4) granular base thickness, (5) coefficient k sub 1 of the base resilient modulus-bulk stress relationship, (6) exponent k sub 2 of the base resilient modulus-bulk stress relationship, (7) coefficient m sub 1 of the subgrade resilient modulus-deviatoric stress relationship, and (8) exponent m sub 2 of the subgrade resilient modulus-deviatoric stress relationship. Predictions of service life from the model evaluated were found to be sensitive to asphalt concrete thickness, initial PSI, asphalt concrete modulus, and the constants m sub 1 and m sub 2 defining the stress dependency of the resilient modulus of the subgrade soil. In addition, because of the influence of the stress dependency of unbound pavement materials, there is strong indication that optimum values for base-related variables exist for different pavement conditions. The results obtained showed the importance of a sensitivity analysis for evaluating the behaviour of a performance model over a range of conditions considered to be of practical interest. The information generated from a sensitivity analysis is of value in evaluating the most effective pavement design for a given set of conditions and in developing guidelines for the proper application of a performance model.
Abstract