Verification of mechanistic prediction models for permanent deformation in asphalt mixes using accelerated pavement testing

Abstract

Permanent deformation (rutting) is the most critical load-associated distress that develops on asphalt pavements significantly affecting their performance. Past research work focused on estimating permanent deformation of asphalt mixes using empirical prediction models or prediction models based on linear elastic material models. In recent years, mechanistic and mechanistic-empirical prediction models have been developed to take into account the behavior of asphalt material (viscoelastic, viscoplastic or elasto-visco-plastic). This research project aims to evaluate existing mechanistic models that predict permanent deformation (rutting) in asphalt mixes by comparing computed permanent deformation to that measured in a full-scale accelerated pavement test. Six pavement sections were constructed in the Civil Infrastructure Systems Laboratory (CISL) of Kansas State University with six different asphalt mixes. The sections were loaded with up to 700,000 load repetitions of a 22,000lb single axle. The transverse profiles at the pavement surface were measured periodically. For material characterization, asphalt mix samples fabricated in the laboratory, were subjected to dynamic modulus (|E*|), static creep - flow time (Ft), dynamic creep - flow number (Fn), triaxial and uniaxial strength tests, repetitive shear at constant height (RSCH) and frequency sweep at constant height (FSCH). The finite element software, Abaqus, was used to simulate and evaluate four permanent deformation prediction models, which are: creep model, elasto-visco-plastic model, viscoelastic model and Drucker-Prager model. The predicted permanent deformation was then compared to permanent deformation measured in CISL for the six of asphalt pavement sections. It was found that, with some improvements, creep and elasto-visco-plastic models could be used to predict permanent deformation in asphalt mixes. The viscoelastic model greatly under-predict permanent deformation, and the Drucker-Prager model with hardening criteria over predicts permanent deformation as compared to values measured in CISL.