[FEATURE] distributed Winkler distribution, or an integrated soil_eq=True flag on from_windio_with_monopile

[TheMercer]

What problem does this solve?

So stiffness is distributed and not lumped stiffness

Proposed API or behaviour

As in Issue 97 follow on

from pybmodes import MudlineFoundation
from pybmodes.models import Tower

tower = Tower.from_windio_with_monopile("ontology.yaml")
foundation = MudlineFoundation.from_soil_properties(
    pile_diameter=10.0, pile_length_embedded=45.0,
    pile_EI=tower._bmi.support.sec_props.flp_stff[-1],  # or pass directly
    soil_E=140e6, soil_nu=0.4,
    soil_profile="homogeneous", pile_behaviour="auto",
)
tower._bmi.support.mooring_K = foundation.as_mooring_K()
tower._bmi.hub_conn = 3

Can make it instead of a single diameter and EI. It can take the Tower definition to get the properties (if you know pile_length_embedded) then you know how much is in the mudline

Citable references

Issue 97

Alternatives you've considered

No response

Scope estimate

None

Anything else?

No response

[SMI-Lab-Inha]

The ergonomic half of this issue (one-call attach onto from_windio_with_monopile without hand-mutating BMI fields) shipped in 1.15.1 via #119, Release notes. The canonical pattern is now:

modal = (
    Tower.from_windio_with_monopile("ontology.yaml", tip_mass=rna_mass)
    .attach_mudline_foundation(foundation)
    .run(n_modes=4)
)

The distributed Winkler distribution along the embedded length (mapping per-station soil rates into distr_k rather than lumping the coupled stiffness at the mudline) remains in scope for this issue. Keeping #118 open for that.

I plan to scope the distributed-Winkler work for a 1.16 minor release. The per-station rate computation needs a clear physics gate (constant E_SO over depth has a closed form; inhomogeneous profiles need either a Yu and Amdahl Table 1 reduction at multiple depths or a different formulation altogether), and a comparison anchor against the lumped-spring 1st mode on the DTU 10 MW reference is the natural validation. Will open a focused PR with the proposed API and reduction once the design is settled.

If you have a strong preference between (a) a fluent foundation = MudlineFoundation.distributed_from_soil_properties(..., n_stations=8) builder that returns the per-station distr_k_z / distr_k arrays, or (b) an integrated Tower.attach_mudline_foundation(foundation, distributed=True, n_stations=8) flag, leave a note here and I will fold it into the design.

[TheMercer]

I think I prefer the Tower.attach_mudline_foundation(foundation, distributed=True, n_stations=8)
As then foundation can stay the soil properties input only, similar to platform and tipmass

Then attaching has the extra functionality for single 6 by 6 or distribution etc.

I think that flows better + more inline with current setups in pybmodes

[TheMercer]

I think really depends what's easiest when comes to current work follow. I can see why the first works as it just extended current input section