Cranfield University
Cranfield University
Cranfield University
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Soil Compaction Models<br />
depends on the natural logarithm of the mean normal stress. This approach is described in<br />
detail by Ansorge (2005, a). Thus in SOIL FLEX the user can choose the theoretical back-<br />
ground he wants to utilize. Stress distribution is based on Söhne (1953) with an added<br />
shear component. For stress distribution in the contact area a variety of approaches can be<br />
selected. In section 6.2 two of these predictions will be compared to results of both An-<br />
sorge (2005, a) and this study.<br />
Within SOIL FLEX this work will only consider the model COMPSOIL as this requires<br />
the least amount of parameters to be estimated while maintaining all output parameters.<br />
COMPSOIL only needs an estimation of the slope and intercept of a virgin compression<br />
line. For subsequent passes the slope of the recompression line has to be calculated. Thus<br />
overall only three parameters are needed. Gupta and Larson (1982) need similar parame-<br />
ters, a reference bulk density and a reference stress state on the VCL, i.e. an intercept of<br />
the VCL; and the slope of the VCL. Additionally the model of Gupta and Larson (1982)<br />
requires the slope of the bulk density vs. degree of water saturation curve and the desired<br />
degree of saturation minus the actual saturation at a reference bulk density. Thus this<br />
model requires an additional three parameters and moreover, the recompression index is<br />
equal to zero, assuming no elastic rebound, which can lead into difficulties predicting sub-<br />
sequent passes. The model of Bailey and Johnson (1989) requires four compactability co-<br />
efficients of which two are dimensionless and two have dimensions of [kPa -1 ]. The empiri-<br />
cal determination of this number of coefficients requires a large data set to adapt the model<br />
to a certain soil condition. Keller et al. (2007) give for each of the three models pedo-<br />
transfer functions from literature to be able to adapt each individual model to given soil<br />
conditions. However, determination of variables from pedo-transfer functions is time con-<br />
suming as it involves extended laboratory testing.<br />
Another approach concerning soil compaction models and the prediction of soil compac-<br />
tion was developed by Diserens and Spiess (2004) (TASC for “Tyre/Tracks and Soil Com-<br />
paction”) and is based on the principles of Etienne and Steinmann (2002) who developed a<br />
stress distribution model utilizing the approach of Newmark (1940) for stress propagation<br />
in the soil. This model is in theory and approach similar to the model from van den Akker<br />
(2004) (SOCOMO; described in detail by Ansorge 2005, a), yet it takes a different ap-<br />
proach in both calculated pressure distribution and the occurrence of soil compaction. It<br />
merely looks at the soil pressure rather than a combined failure criteria as in SOCOMO.<br />
Ph.D. Thesis Dirk Ansorge (2007)<br />
93