[Back]

R. Lackner:
"Multi-Scale Modeling of Asphalt as the Basis for Reliable Predictions of the Performance of Flexible Pavements";
Talk: 1st International Conference on Finite Element for Process (LUXFEM 03), Luxembourg (invited); 2003-11-13 - 2003-11-14.

A reliable assessment of the performance of road asphalts requires
suitable procedures and models for the evaluation of key material properties. In case of asphalt used for trafficked pavements, these key properties are its resistance to rutting, to cracking at low
temperatures, and to fatigue failure under repeated load cycles.
Asphalt is composed of bitumen, aggregate, and air voids, showing a complex thermo-rheological behavior. E.g., the low viscosity
of asphalt at high temperatures (T > 135 C) is a necessary prerequisite during the construction and compaction process of high-quality asphalt layers. When the surface temperature reaches
70 C during hot summer periods, however, this viscosity should be significantly higher in order to minimize the development of permanent deformations (rutting). The desirable decrease of viscosity and, hence, increase of stiffness with decreasing temperature at hot and medium temperatures (0 C < T < 70 C) are, on the other hand,
disadvantageous at low temperatures (T < 0 C), causing low-temperature cracking in asphalt pavements. This optimization problem concerning the behavior of asphalt at different temperature regimes is a main objective of the Christian Doppler Laboratory ``Performance-Based Optimization of Flexible Road Pavements" (TU Wien), headed by Ronald Blab. For the optimization process of a multi-composed material such as asphalt, three different modes can be distinguished:

(a) variation of mixture characteristics (e.g. binder/aggregate-ratio),
(b) change of constituents used (e.g. different bitumen or filler type), and
(c) allowance of additives (e.g., polymers to modify the bitumen).

In order to account for the wide range of asphalt mixtures, resulting from the large number of different bitumen, fillers, and aggregates used in engineering practice, the multiscale concept is chosen for the determination of rheological properties of asphalt. Within the multiscale concept, four additional observation scales are introduced below the macroscale, namely (i) the bitumen-scale (asphaltene and maltene morphology), (ii) the mastic-scale (bitumen+filler), (iii) the mortar-scale (mastic + aggregate with d < 2 mm), and (iv) the
asphalt-scale (mortar + aggregate with d >2 mm). Within each observation scale, the characteristics (such as structure and material properties) of the constituents present at this scale are considered. In the first year of research within the Christian
Doppler Laboratory, the modeling efforts concentrated on the material identification at the bitumen- and mastic-scale. Experimental studies on bitumen, employing differential scanning calorimetry (DSC), thermo gravimetry (TG), bending beam rheometer (BBR), and atomic force microscopy (AFM) were performed. The obtained experimental results allow to assess the glassy/crystal configuration of the bitumen microstructure at the bitumen-scale and its relation to rheological properties. Moreover, physical hardening effects resulting from the orientation of molecules within the glassy regime and an ongoing oxidation process (aging), both affecting the glassy/crystal state of bitumen, are considered at the bitumen-scale.
The characteristics of fillers, such as the specific surface and the
particle-size distribution, enter the multiscale model at the mastic-scale. Based on the effective volume of the filler, homogenization at scale (ii) provides the material properties of the mastic. The so-obtained mastic properties show good agreement with the results obtained from laboratory tests.

The material properties obtained from homogenization will serve as input for macroscopic material models used in the numerical simulation of flexible pavements. Changes of asphalt properties in consequence of loading and/or environmental conditions can be considered at the respective scale of observation and, by the use of the multiscale model, be incorporated into analyses performed at the macroscale.