Civil Engineering Reference
In-Depth Information
paths close to those used to establish the model; extrapolation to different paths
would be erroneous and invalid. Furthermore, quite often, the modeling obtained in
this way only partly represents the physical phenomena; thus, the equivalent linear
visco-elastic modeling (which will be described later) does not describe or
incorporate volume variations (subsidence) under purely deviatoric loadings.
Moreover, the stress paths represented in tests are only rudimentary idealizations of
real stresses. This type of approach represents a compromise between the
phenomenon to be modeled and ease of implementation. If it is used with good
judgment, however, it is quite a powerful tool.
Before attempting an experimental description of the phenomena to be modeled
by mathematical representation, it is important to realize that on the time scales
involved in seismic stresses, most soils exhibit non-drained behavior during the
stress. Soil permeability is not sufficient (compared to the loading rate) to permit
drainage. As a consequence, in the approach described above, we reason in terms of
total stresses. Once again, this is a simplification, as real soil behavior is controlled
by the actual stresses.
The rest of this text will be limited to examining soil behavior before failure. The
study of soil behavior on failure gives rise to different approaches made necessary
by the adopted schematization. If we had a real behavior law at our disposal, such a
distinction would not be necessary; in fact, the behavior law would allow us to
reproduce soil behavior from the smallest strains (quasi-elastic strains) to the very
high strains associated with failure.
For further descriptions of soil behavior, see [HAR 78], [PEC 8] and [PRE 78].
What emerges from the experimental statements in section 4.1.2 is that the soil
cannot be represented by an elastic model, at least as soon as the strains become
significant.
The non-linearity appearance thresholds generally correspond to low strains (in
the range 10 -4 to 10 -6 ). However, we must distinguish between reversible (or quasi-
reversible) and irreversible non-linearities, as the appearance thresholds of the latter
are higher (10 -4 to 10 -3 ).
Search WWH ::




Custom Search