Modeling of Asphalt Rutting in Airfield Pavements Using Advanced Mechanistic Models

Abstract

The objective of this research is to develop an accurate and realistic model that predicts the viscoplastic response, i.e. permanent deformation, in flexible airfield pavements. The proposed approach builds on the mechanistic shift model which has proven to yield accurate predictions of viscoplastic strains in asphalt pavements. However, the shift model was developed and calibrated for highway pavement conditions and data. This research investigates and addresses the needed modifications to the shift model in order to accommodate the different loading conditions found at airfield pavements. The major difference between highway and airfield loading conditions is the shape and amplitude of the applied loading cycle. While highway pavements are mainly subjected to typical haversine loading due to moving vehicles, airfield pavements experience static loading at locations where airplanes stop frequently during stacking at taxiways, during preparation for take-off or proceeding to their final parking position after touchdown.

The shift model, developed for continuous sets of haversine loading, may loose prediction accuracy for such loading conditions. Thus, this research develops a load conversion module that utilizes a load shape shift factor to convert airplane static load cycles into an equivalent number of haversine load cycles, based on the concept that equal viscoplastic strains can be induced by different loading patterns at different combinations of stress amplitudes, loading times, and temperatures.

As a first step in developing the load shape shift factor, the test conditions of the stress sweep rutting (SSR) test, which is typically used to calibrate the shift model, were modified to reflect realistic airfield loading conditions. This included the increase of both the deviatoric stress levels as well as the confining pressure in the pavement under wheel loading. SSR tests were then conducted using the modified conditions in order to calibrate the reference shift model prior to load conversion. Then, in order to develop and calibrate the proposed load shape conversion model, a comprehensive testing program consisting of confined load and recovery tests at multiple conditions was carried out. The experimental plan included nine combinations of temperatures and deviatoric stress levels, with both haversine and creep load and recovery tests conducted at each combination. The testing program was designed to isolate the effect of loading shape on the viscoplastic response of asphalt such that a load shape shift factor can be determined from creep and haversine tests at each loading condition. After quantifying the load shape shift factor for a variety of test conditions, the data was used to perform multiple linear regression analysis in order to calibrate the relationship between the shape factor and both stress and temperature. This allowed for the prediction of the shape factor at a wide range of conditions other than the tested load combinations.

Finally, and in order to validate the model at a wide range of loading conditions, other than those used in the calibration process, creep and recovery tests were conducted at additional temperatures and stress levels. The calibrated conversion model was then used alongside the reference shift model to predict the viscoplastic response at those additional conditions. Validation results proved that the conversion model, integrated with the reference shift model, is able to accurately predict the viscoplastic strain in asphalt concrete for repetitive creep and recovery loading.

As a last step, and after the concept of load shape shifting had been validated, a thorough sensitivity analysis was conducted in order to reduce the testing effort required to calibrate the conversion model to the extent possible. The aim was to reduce the required number of samples, test temperatures, and deviatoric stress levels while maintaining accurate viscoplastic (VP) strain predictions. As a result, the testing effort was significantly reduced by excluding the intermediate test temperature and stress level. Furthermore, the required number of samples was cut by half by excluding the need to conduct haversine load and recovery tests. Instead, the haversine VP strain data, which is required to calculate the load shape shift factor at each condition, was predicted using the reference shift model that was calibrated for airfield conditions.