Simple soil models

Good afternoon, i am a PhD student studying the modeling of rail tracks
for high speed trains and i have been using Diana 9.1 and 9.2.
Regarding Diana 9.2 i have some questions regarding the simple soil models.

- For example in the Hardin Drnevich model, the strain that is used to
update the shear modulus is the maximum shear strain in each point
(regardless of the direction) or is the shear strain in each of the
element local axes?

- When the shear strain is updated, the correlation between the shear
strain and stress is altered. And the relation between axial stress and
strain remains the initial or is also updated accordingly (in short: G
is updated, but does E change or is always constant?)

- Regarding these models, in the theoretical presentation of them
(chapter 23) the following is said "The models should behave according
to the so-called extended Masing rules: ". Does it mean that these
models do follow the Masing rules, or they should, but they don't?

While trying to answer these questions by myself I tried to use a simple
model which i am sending pasting in attachment:
rapidshare.com/files/373078823/Triaxial.rar.html

I found some of the results strange for me so maybe i am doing something wrong:

- I computed the linear response (diana -m hardin.dat linsta.com) and I
computed the non-linear response with a very large convergence factor
(diana -m hardin.dat nonlinear.com) where the model converged for 0
iterations. I was expecting that the results would be the same but in
fact, the strains are the same in the local axes, the normal stresses
are also the same, but the shear stresses are exactly the double of the
linear response. Also, the principal stresses are not the same in both
cases.

- I also computed the nonlinear response with a tighter convergence
criterion (diana -m hardin.dat nonlinear2.com), but since the shear
strains are much lower than the characteristic shear strain of the soil,
the shear modulus should remain basically the same as the Gmax, as
presented in formula of chapter 23.1 of the Diana Manual. However, the
stresses and strains obtained are considerably different than those of
the previous two cases.

In short, I am asking for a few clarifications of these
models and to try to understand why these seemingly strange results
occur. I should stress that I also used the other soil model
(Ramert-Osgood) and obtained similar discrepancies.

Hope to ear from you soon.
Best regards.