Dear scientists and researchers,
Finally, we were able to obtain distributions for individual soil mechanical parameters in order to be able to answer geotechnical questions not only with deterministic but also with probabilistic approaches. Now the validation of the observed distributions is more difficult than initially expected.In the literature there are only isolated references to observed distributions [e.g. Lumb (1974), Lee et al. (1984), Benson (1993), Lacasse and Nadim (1996), Baecher & Christian (2003)] and the normal distributions described therein, among others, provide physically illogical results in further simulations due to their value range below zero.
Which distributions do you use to model soil mechanical parameters, such as the void ratio, the dry and saturation weights, the hydraulic conductivity or the shear strength?
Dear Niklas,
The soil parameters such as dry and saturated unit weights, Young's modulus and shear strength are used to describe the mechanical properties of soils. The hydraulic conductivity is used to describe the liquid flow characteristics of soils..ok all the best in your research..
I have found lognormal distributions to be better suited for a lot of Geotech parameters, it helps to avoid those negative values if you are doing some reliability modelling. See a relevant publication of mine below. That been said normal distributions work ok, see the distribution for hydraulic conductivity for a project I am working at the moment in the figure attached. Have you checked Phoon Kok Kwang' s book?
Article An assessment of statistically based relationships between c...
All the best,
Marcelo
Dear Niklas, as you already mentioned various authors [Lacasse & Nadim, Wolff et al., Lumb, JCSS] suggest to use a Gaussian distribution to describe, for instance, the effective friction angle. In principle, however, Lognormal distributions are also a reasonable choice and recommended by other authors [Schultze]. Schneider & Schneider recommend the use of Lognormal distributions in particular for larger coefficients of variation. Additionally, if you favour a Gaussian distribution, you could also model it as a truncated distribution. The hydraulic conductivity k is commonly considered as lognormally distributed.
- Badges
- Science topic
- Similar topics
- Geoscience
- Cartography
- Mapping