clay model comparison with LPILE

[emabcede30]

I am obtaining different shear and moment profiles versus results with LPILE using clay model. The loading is 20kN in axial and shear at the top of the pile.

-Soil Pile Profile and water table

-Deflection, Moment, Shear and Mobilized Soil Reax.

-py curves

The py curves are not similar with small deformations. does this have to do with nonlinear bending stiffness of the material properties? or just how the py curves are constructed at each incremental deformation? I am unsure how these are implemented in openpile. I used solid circle concrete.

i added these in construct.py for a concrete circular section

        elif self.material == "Concrete":
            # unit weight
            self._uw = 22.78  # kN/m3 ~145lb/ft3
            # young modulus
            self._young_modulus = 24_870_062.324  # kPa ~4700*sqrt(f'c = 28 MPa)
            # Poisson's ratio
            self._nu = 0.15

#####################################################

            elif self.kind == "Solid-Circle":
                A = m.pi / 4 * (d**2)
                I = m.pi / 64 * (d**4)
                area.append(A)
                area.append(area[-1])
                second_moment_of_area.append(I)
                second_moment_of_area.append(second_moment_of_area[-1])

and added shear coefficient for solid-circle shapes in kernel.py

        elif model.pile.kind == "Solid-Circle":
            nom = 6 * (1 + nu)
            denom =  7 + 12*nu + 4*(nu**2)
            kappa = nom / denom

to replicate reaction profile results, use the code:

from openpile.construct import Pile, SoilProfile, Layer, Model
from openpile.soilmodels import API_clay
from openpile.analyze import winkler

sp = SoilProfile(
    name="pier3-Static",
    top_elevation=0.0,
    water_line=13.10,
    layers=[
        Layer(
            name="Soft Clay",
            top=0,
            bottom=-12.0,
            weight=14.0,
            lateral_model=API_clay(Su=15, eps50=1.98/100, kind="static"),
        ),
        Layer(
            name="Medium Stiff to Stiff Clay",
            top=-12.0,
            bottom=-60.0,
            weight=17.0,
            lateral_model=API_clay(Su=59, eps50=0.84/100, kind="static")
        )
    ],
)

p = Pile(name = "pier3",
    kind='Solid-Circle',
    material='Concrete',
    top_elevation = 13.10,
    pile_sections={
        'length': [50.0 + 13.10],
        'diameter': [2.0],
        'wall thickness': [1e-10],       
    }
)

# Create Model
M = Model(name="pier3", 
        pile=p, 
        soil=sp, 
        element_type="Timoshenko", #Timoshenko
        coarseness=0.5)

# Apply bottom fixity along x-axis

M.set_support(elevation= (-1 * 50), Tx=True)
M.set_support(elevation=13.10, Rz=True)

M.set_pointload(elevation=13.10, Px=20.0, Py=20.0)

Result = winkler(M)

#copy results to clipboard, paste to excel and graph
Result.deflection.to_clipboard(index=True)
Result.forces.to_clipboard(index=True)
Result.py_mobilization.to_clipboard(index=True)
M.get_py_springs(kind='element').to_clipboard(index=True)

your thoughts/help to identify the cause of the differences is much appreciated. this branch can be used to replicate the results in openpile:

https://github.com/emabcede30/openpile/tree/frankeRollins2013

as an anecdote, i am getting exactly same shear and moment profiles using api sand

[emabcede30]

The issue is now resolved. Just need to expand the points of the clay p-y curve.

in py_curves.py, under def api_clay(...) expand the number of points between 0 and 0.1*y50

ylist_in = [0.0, 0.001 * y50, 0.005 * y50, 0.008 * y50, 0.01 * y50, 0.02 * y50, 0.05 * y50, 0.08 * y50, 0.09 * y50, 0.1 * y50, 0.21 * y50, 1 * y50, 3 * y50, 8 * y50, 15 * y50, ymax]

and increase the output length in the parameter list from 20 to 1000

output_length: int = 1000

likewise, in construct.py, increase the spring dimension in def create_springs(...) to 1000 as well

spring_dim = 1000

The results are now similar

-Deflection, Moment, Shear and Mobilized Soil Reax.

-py curves

There are still other nuances like layering correction and nonlinear bending stiffness. thoughts on these issues are helpful.

[TchilDill]

Hi @emabcede30,

Thanks for raising this, and sorry for the late reply here.

I note you somewhat fixed your issue.

The issue comes from the different versions of the clay curve that was documented in API and other standards (all with relatively minor tweaks to one another). And also comes from the fact you are looking at really small displacements.

The one used in OpenPile is API 2014 with a conservative initial stiffness as defined in DNV RP-C212. However I can see it's too rigid of an implementation. I suggest to add all the different versions possible of this curve via an input argument to API_clay soil model.

[TchilDill]

@emabcede30 for your LPILE model, did you use:

• API Clay RP2A (2000)
• API Clay RP2GEO (2014)
• Soft clay (Matlock) model

from looking at the LPILE manual, the current implementation in OpenPile aligns with API Clay RP2GEO (2014). so I am a bit surprised to see that we would need to increase the number of points to match LPILE, that would make sense though if you were using the Matlock model.

see below from (https://www.pilegroups.com/single-post/api-p-y-curves)