Calibrating MEPDG inputs prediction models for asphalt mixes containing reclaimed asphalt pavement

Loading...
Thumbnail Image
Date
2017
Authors
Esfandiarpour, Saman
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In this thesis, the pavement sustainability practices were implemented by using recycled asphalt shingles (RAS) and reclaimed asphalt pavement (RAP) in asphalt pavements. Laboratory performance of mixes containing RAS and RAP were evaluated and characterized for a cold climate such as Manitoba, Canada. In addition, pavement sustainability practices were implemented by generating a database of measured values from laboratory test results to develop and perform local calibration alternatives on dynamic modulus and creep compliance predictive models used in Pavement ME Design software, and to assess the impact of locally calibrated MEPDG models on long-term performance of mixes. Laboratory results showed that 15% RAP can be used in an asphalt mix without changing the virgin asphalt binder grade when the design binder is PG 58-28. It was found that the globally calibrated MEPDG creep compliance and dynamic modulus models are not able to accurately predict values, particularly for mixes used in cold climates, in part because these mixes constituted only a small fraction of the mixes used to develop these models. It was found that the nonlinear multiple regression is the preferred technique for local calibration of NCHRP 1-37A and NCHRP 1-40D E* models. It was noted that the existence of high RAP mixes in calibration of the E* predictive model causes an adverse effect on the reliability of calibrated models. In addition, it was found that nonlinear regression and Artificial Neural Network (ANN) models can be used as two alternatives to reliably predict creep compliance values. Results of the predicted distresses of mixes containing RAP using MEPDG software for Manitoba default Level 3, Manitoba calibrated Level 3, and Manitoba Level 1 demonstrated that the calibrated Level 3 Manitoba asphalt mix input data can be used for the design and analysis of the Manitoba mixes with comparable accuracy of the Manitoba Level 1 input data. As conducting laboratory tests for individual mixes is expensive and time consuming, utilizing locally calibrated reliable models to predict E* and creep compliance can tremendously reduce operating and testing expenses.
Description
Keywords
Dynamic modulus, creep compliance, calibration, MEPDG models, RAP, RAS
Citation