Soil Collapse due to Water Infiltration - Environmental Geomechanics - page 162

The Stephanoff and Kremakova [STE 60] approach is more ambitious since it
proposes a criterion allowing for a quantitative estimation of the collapse. The
criterion is based on the natural porosity n and the natural water content, together
with a coefficient K , which accounts for the soil type ( K = 0.08 for a clayey loess
and 0.05 for a loess). The collapse strain δ is obtained by the following relation:
δ = K(n − 40)(30 − w Nat ) . The calculations of δ in Table 6.3 were made with K = 0.08
for samples at 1.2 and 4.9 m and K = 0.05 for samples at 2.2 and 3.5 m. The
comparison between calculated and measured values shows that this criterion
significantly overestimates collapse strain.
Examination of the previous criteria and comparison with the oedometer tests
show that only the Gibbs and Bara criterion gives a reasonable estimation of the
collapse susceptibility of the loess tested here. In particular, it does not seem
possible to obtain a quantitative estimation of collapse based on standard
geotechnical parameters, as proposed by Stephanoff and Kremakova [STE 60].
6.3. Conclusion
An investigation of the collapse susceptibility of four intact samples extracted
from a profile located along the TGV line 140 km from Paris has been conducted by
using simple and double oedometer methods. The most sensitive sample of the
profile presented slight collapse susceptibility at the initial natural water content that
corresponded to the sampling conditions. Collapsibility appeared to be significantly
sensitive to changes in water content, with collapse volume decrease higher than 5%
at water contents below 12%. This increased collapsibility may explain the
sinkholes observed in areas where collapsible soil was exposed to the atmosphere,
with possible drying occurring before soaking by rainfalls. This confirms the
necessity of protecting the layer of sensitive loess during geotechnical works. SEM
observations showed a porous microstructure with heterogenous scattering of clay
aggregations that filled the inter-grain pores and worked as a linking agent between
the grains in some areas. In areas with no clay, sharp-edge angular-silt grains (15 to
30 µm in diameter) were observed with large inter-grain pores. These pores, located
in the areas with no clay, probably significantly contribute to collapse volume
decrease. Mercury intrusion porosimetry identified the changes in inter-grain pores
that occurred during collapse and showed that the smaller pores inside the clay
aggregations were not affected. The collapse structure is apparently more organized
with a well-graded PSD curve.
An examination of the relevance of various existing collapse criteria was carried
out. Although all criteria detected the higher susceptibility of the 2.2 m sample, the
comparison showed that the well-known Gibbs and Bara's criterion provided the
best answer and was in a good agreement with the results of oedometer tests.