An Investigation into Dynamic Modulus of Hot-Mix Asphalt and Its Contributing Factors

Abstract

One of the key elements of mechanistic-empirical (M-E) flexible pavement design is the characterization of material properties. One material property in particular, the dynamic modulus of HMA, E*, influences tensile strain levels, therefore it is necessary to investigate this property to successfully predict fatigue cracking. E* can be determined directly by laboratory testing or it can be estimated using predictive equations as a function of mixture properties. The more recently developed M-E design program, the Mechanistic-Empirical Pavement Design Guide (MEPDG), offers both methods to characterize E*.

An investigation into the Witczak 1-37A and the Witczak 1-40D E* predictive equations, both utilized by the MEPDG and another recently developed predictive equation, the Hirsch model, was completed. Comparisons were drawn with E* laboratory results for mixtures constructed as part of the National Center for Asphalt Technology’s (NCAT) 2006 Test Track structural study. The investigation revealed that the Hirsch E* model most accurately predicted measured E* values, while the Witczak 1-40D overpredicted values and the Witczak 1-37A varied inconsistently. To validate and optimize M-E designs it is necessary to link pavement performance to material properties. The field parameters that influence E*, load duration and temperature, and the field parameter that is most affected by E*, tensile strain, were measured under varying speeds and temperatures at the NCAT Test Track. These measurements enabled comparisons with the MEPDG analysis procedure. In comparing load durations, it was found that those determined by the MEPDG were nearly 70% greater than those measured in the field. These load durations enabled the computation of E* of each HMA layer, and the prediction of strain in a layered elastic program. Strains estimated from both load duration methods (MEPDG and measured) closely replicated each other; however they poorly replicated those strains found in the field, indicating the inaccuracy of the time-frequency relationship currently used in the MEPDG. It is suggested that State DOT’s wishing to supplement laboratory E* testing while utilizing the MEPDG for design, substitute laboratory results required for a level one design with predictions from the Hirsch model. It is not recommended that the MEPDG be used as a primary design method until further refinement of the time-frequency relationship, and further investigation into the accuracy of the pavement distresses can be completed.