Mimicking road surfaces in FD and FE models.

[soilsteelguy]

I am trying to mimic a granular road surface, and apply a surface pressure representative of a live load vehicle (using FLAC and PLAXIS). The problem is, when the pressure is applied, the modeled surface experiences very large deformations. The Mohr Coulomb model I am using suggests that a regular highway vehicle would sink into the ground on any gravel road without a pavement layer. We know that in reality this is not the case.

The soil I am trying to model is a well-graded (3" minus), compacted (95% SPD), granular material. Following the rule for cohesionless soils, I set cohesion equal to zero. Without the cohesion, the soil surface in the model fails under any significant pressure. I have tried increasing the soil stiffness, but even making it another order of magnitude stiffer doesn't help.

The properties I am using are: Es=25MPa(3.6ksi), Poisson's ratio=0.3, friction angle=40, dilation angle=0, cohesion=0, tension=0.

I do not want to significantly affect the stiffness of the soil (especially the surface layer) because the live load pressure distribution through the depth of the soil is of key concern.

Anyone have any ideas?

 

[MCC1966]

I am wondering how your simulation runs with C=0 . Usually this leads to numerical instability. please try to give c law value ;e.g, c=2 and let me know how things change
Thanks

 

[soilsteelguy]

Using c=2kPa(0.3psi) helps, a tiny bit. I still get more than 50mm deflection at the "road" surface.

If I bump it up to c=50kPa(7psi), I get more reasonable results, but still more deformation than you would expect in the field (approx. 20mm).

The problem with adjusting the cohesion of the soil, is that it alters the live load pressure distribution through the soil. A higher cohesion spreads the load more, giving a wider pressure distribution and resulting in a smaller vertical pressure at equivalent points of depth.

Thanks for your input thus far.

 

[Ron]

Your elastic modulus is quite low for the soil you describe. I would use something on the order of 8 to 10 ksi for that material.

Remember you are trying to force a non-homogeneous material to act as an isotropic homogeneous mass in your model.

You might want to check your parameters using an elastic layer analysis program such as ELSYM5 or Everstress 5.0.

While you don't have cohesion, you do have confinement when acts similarly since it resists lateral shear stresses.

Your vertical deformation in a pavement structure should be in values of 10^-4 to 10^-6 range.