SubDyn: Wrong reaction loads when SoilDyn includes nonlinear loads

[RBergua]

Bug description
When modeling soil-structure interaction (SSI) through a linear stiffness matrix in SubDyn or SoilDyn (i.e., CalcOption = 1), the results are the expected ones and SoilDyn and SubDyn return consistent results. See for reference:

However, once the user includes a nonlinear SSI (i.e., SoilDyn CalcOption = 3), the reaction loads provided by SubDyn (e.g., ReactFXss) are not consistent. See for reference:

I was wondering what these reaction loads from SubDyn were actually representing. After doing some numbers, I can confirm that the reaction loads from SubDyn account for the linear part of the SSI and disregards the nonlinear part. Since the (linear) stiffness matrix is an input that I provide and SubDyn returns the displacements and rotations at the seabed, I'm able to compute analytically the loading that comes from the SSI linear part. For example:
Stiffness matrix:
6.336198E9 0.0000e+00 0.0000e+00 0.0000e+00 -5.015421E10 0.0000e+00
0.0000e+00 6.336198E9 0.0000e+00 5.015421E10 0.0000e+00 0.0000e+00
0.0000e+00 0.0000e+00 1.119691E10 0.0000e+00 0.0000e+00 0.0000e+00
0.0000e+00 5.015421E10 0.0000e+00 8.111942E11 0.0000e+00 0.0000e+00
-5.015421E10 0.0000e+00 0.0000e+00 0.0000e+00 8.111942E11 0.0000e+00
0.0000e+00 0.0000e+00 0.0000e+00 0.0000e+00 0.0000e+00 2.552673E11

Linear force at the seabed computed analytically:
{M1N1TDxss}·6.336198E9 + {M1N1RDye}·(-5.015421E10)

By doing this, I can see an excellent agreement between the linear part of the SSI and the reaction load from SubDyn:

Summarizing: The reaction loads provided by SubDyn only account for the linear stiffness part from SoilDyn. Note that this is just a sensor output issue. The physics are properly implemented. Also, the proper (total) loading is available fom SoilDyn output sensor.

I'm looking at the source code and thinking about disabling the SubDyn reaction loads when SoilDyn involves nonlinear SSI. From FAST_Subs.f90 I can see:
Init%InData_SlD%SlDNonLinearForcePortionOnly = .true.

So, it should be a matter of having a flag that disables these SubDyn reaction loads accordingly. I'm working on that...

[jjonkman]

Dear @RBergua,

I haven't fully thought through this, but according to the documentation of the SubDyn reaction calculation (https://openfast.readthedocs.io/en/main/source/user/subdyn/theory.html#outputs-and-time-integration), the reaction load is calculated by summing the nodal load contribution from each element at each reaction node, subtracting out the external applied loads at those nodes, then summing across all reaction nodes. My understanding is the "external applied load" in your case is the nonlinear part of the load from SoilDyn. So, is the behavior you are describing the expected one, given the description? In that case, the result is simply an artifact of splitting the nonlinear soil reaction into a linear part handled within SubDyn and a nonlinear part handled externally.

Best regards,

[RBergua]

@jjonkman I would say that the behavior is aligned with the description. The documentation reads:
For a given reaction node, the 6-vector of loads is obtained by summing the nodal load contributions from all the elements connected to that node expressed in the global frame (no account of the sign is done here), and subtracting the external loads (F_HDR) applied on this node.

However, when including SoilDyn with nonlinear forces (e.g., REDWIN DLL), the nonlinear loads are also subtracted. So, in reality, the current behavior would be: and subtracting the external loads (F_HDR+F_SLD) applied on this node.

This is also observed in the source code with the -F_ext:

openfast/modules/subdyn/src/SubDyn_Output.f90

Line 474 in 2895884

ReactNs((I-1)*6+1:6*I) = FK_elm2 - Fext !Accumulate reactions from all nodes in GLOBAL COORDINATES

And, yes, this is an artifact of splitting the nonlinear soil reaction into linear (SubDyn) and nonlinear (SoilDyn) parts.