Skip to content

fix: carry the end weight moment at an anchor resting on the seabed - #14

Merged
SMI-Lab-Inha merged 4 commits into
devfrom
fix/seabed-friction-anchor
Oct 15, 2026
Merged

SMI-Lab-Inha merged 4 commits into
devfrom
fix/seabed-friction-anchor

Conversation

@SMI-Lab-Inha

@SMI-Lab-Inha SMI-Lab-Inha commented Oct 14, 2026 •

Copy link
Copy Markdown
Owner

Summary

Fixes a spurious, permanent rise of the anchor tension of a finite-EI cable laid on the seabed, found in a one-hour full-turbine OpenFAST run, and records that run.

Symptom. In the IEA-15 MW mixed model (examples/iea15mw_umaine_mixed_cabledyn.dat, a lazy-wave power cable with about 38 grounded nodes, frictionMu 0.35), the cable's AnchTen4 jumped from its static 1.22 kN to about 3.1 kN some 25 s into the run and stayed there for the whole hour, while the platform drifted 22 m toward the anchor and the line slackened. Without friction it declined smoothly to 0.6 kN.

Cause (not the friction law). The seabed contact acts at the nodes, which carry no moment. At a held anchor end resting on the floor, the consistent end moment of the end element's weight (w L² / 12) had nothing to react it:

  • the free end tangent turned 1.9° into the floor, and the element dipped 27 mm below the seabed between its nodes;
  • membrane locking put −47 kN to +127 kN of axial force in that element, against a 1.2 kN line tension;
  • the compressed part buckled sideways as soon as the hang-off swayed;
  • the bowing element pulled its neighbours toward the anchor, and friction locked them there (friction forces of 155–349 N on nodes 2–6, all toward the fairlead).

The friction model itself is correct: its force opposes slip, its return mapping only dissipates, and its anchors are saved and restored correctly.

Fix. The seabed now carries the end element's weight moment at a held line end resting on the floor, along the floor normal on a slope, in the statics and the dynamics. After review, the support also needs the end element itself to lie on the floor. It is scaled by a C1 smoothstep of the inner node's elastic normal force over its half share of the element weight. A laid element keeps the full support, and an element rising from an anchor on the floor gets none, which leaves the old result bit for bit. The static near-anchor tension is uniform (1218.204 N), and the anchor curvature drops from 1.5e-2 to 9e-7 1/m.

Results

Coupled IEA-15 MW, normal turbulence, Hs 8 m Before After
AnchTen4 median (static 1.22 kN) 2.75 kN, above 2 kN for 3592 of 3600 s 1.39 kN
AnchTen4 excursions locked wave-frequency stick-slip peaks up to 4.1 kN, each under 1 s (5 % of the time)
FairTen4 11.5 ± 1.1 kN unchanged

Example decks.

  • Byte-identical: every deck whose held end is off the floor (37 EI = 0, rod and body decks), and laid_cable_cross_current.
  • Changed: the 13 finite-EI decks with an anchor on the seabed. Their end tensions change by at most 0.34 N (0.004 %, mixed_body_rods_points FairTen8).

Statics robustness matrix. Two entries change outcome; the evidence is recorded in the matrix:

  • LW_1015__cur_fr (1 m/s current, friction) now fails closed. Its old "admissible" state was a hairpin past the anchor that the old code reached only through the mesh-sequenced fallback at its exact seabed stiffness (±0.001 % of kBot failed). A slow dynamic current ramp reaches the hairpin only by passing through compression, which this tension-only deck rules out. With tensile_safety it stops at 0.37 m/s, as the static solve does.
  • CUR_1013 (1.9 m/s current, no friction) used to fail closed and now finds an equilibrium with the grounded run turned sideways on the seabed. When the refinement of the catenary continuation stops on compression in a current, its failure now names the current limit, so the driver retries out of the plane. Patched with this one change, the old code finds the identical state.

Full-turbine touchdown record (VALIDATION.md, coupled finite-EI touchdown row): the cable gap on the coupled route was "never a full-turbine touchdown DLC".

  • Set-up: one hour after a 200 s transient. TurbSim normal turbulence at 10.6 m/s, 17 % TI; JONSWAP Hs 8 m, Tp 12 s, aligned with the wind and the cable; AeroDyn with the rotor held at 7.55 rpm (no controller), rigid blades and tower; a SeaState grid covering the lines.
  • Robustness: completes without recoveries or non-finite values, at 3.4× real time.
  • Cable: the same 38 nodes stay grounded. The hog-bend curvature (0.160 1/m) is mesh-converged; the deck's 10 m laid elements under-resolve the touchdown curvature by about 30 %.
  • Chains: the chain-only twin matches stock MoorDyn to 0.3 % over 1200 s.

Tests

  • New case_grounded_anchor_end (deck_hermite_cable): the power cable with 2 m of hang-off sway for 30 s. Before the fix the anchor tension goes 1218 → 3775 N and the anchor tangent dips (share 3.2e-2), so the test fails; after the fix it passes.
  • New check_taut_line_anchored_on_floor: a taut line anchored on the floor whose first element rises is bit-identical to the same line without a seabed, in statics and held dynamic steps. Against the first version of this PR it fails three assertions (a 7 mm move, anchor curvature 2.9e-3 to 4.3e-3 1/m).
  • hermite_cable_static slope polish: the old layout (fairlead at x = 150 m) touched down at a node, with neighbouring equilibria 3 cm apart, and failed on Linux. The fairlead now sits at x = 200 m. Every BLAS kernel reaches the same stable reference, and both backtracking polishes return to it from four times the perturbation the test uses (default, Prescott, Haswell and SkylakeX kernels; Release and Debug).
  • cable_statics_matrix: updated expectations as above, each with a why field.
  • Full suite, Windows, gfortran 16.2, Release, slow tier included: 251/251 pass (rebased on dev).
  • pre-commit run --all-files passes.

Checklist

  • pre-commit run --all-files passes
  • ctest --test-dir build --output-on-failure passes
  • python -m pytest tests python/tests passes (if Python or validation records changed) — the statics matrix harness ran in the suite
  • VALIDATION.md updated if a validation result changed
  • CHANGELOG.md updated under [Unreleased] (heading added if missing) for user-visible changes
  • Documentation in doc/ updated for changed inputs, options, or outputs (theory: the end support)
  • Conventional Commit title (feat:, fix:, docs:, ...)
  • Commits are signed off (git commit -s) under the
    Developer Certificate of Origin

A one-hour coupled OpenFAST run of the IEA-15 MW mixed deck showed the
power cable's anchor tension jump from 1.22 kN to 3.1 kN about 25 s in
and stay there while the platform drifted toward the anchor; without
seabed friction it declined to 0.6 kN.

The friction law is not at fault. The seabed contact acts at the nodes,
which carry no moment, so at a held anchor end resting on the floor the
consistent end moment of the end element's weight (w L^2 / 12) had
nothing to react it: the free end tangent turned 1.9 deg into the
floor, the element dipped 27 mm below the seabed between its nodes, and
membrane locking put -47 kN to +127 kN of axial force in it about a
1.2 kN line tension. The compressed part buckles sideways once the
hang-off sways; the bowing element pulls its neighbours toward the
anchor and friction locks them there. Without friction the end nodes
wandered by up to 0.18 m.

The seabed now carries the end element's weight moment at a held line
end resting on the floor, along the floor normal on a slope, in the
statics and the dynamics (a constant load; the tangent is unchanged).
The static near-anchor tension is uniform (1218.204 N) and the anchor
curvature drops from 1.5e-2 to 9e-7 1/m; the coupled anchor tension
stays at 1.22-1.38 kN. Lines whose held end is off the floor are
byte-identical; the shipped finite-EI examples with a grounded anchor
change by at most 0.7 N.

A regression test sways the hang-off of that cable for 30 s. Two decks
of the statics matrix change outcome, with the evidence in the matrix:
LW_1015__cur_fr fails closed (its old hairpin is reachable only through
compression, which its tension-only deck rules out), and CUR_1013 now
finds the equilibrium with the grounded run turned sideways, because a
refinement stopped on compression in a current now names the current
limit and the driver retries out of the plane.

Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
The coupled finite-EI touchdown record rested on a 20 s aggregate
trajectory and said it was not a full-turbine DLC. A one-hour run of the
IEA-15 MW mixed model (normal turbulence at 10.6 m/s, JONSWAP Hs 8 m,
AeroDyn with the rotor held, a SeaState grid covering the lines)
completes without recoveries; the hang-off tension, grounded length and
mesh-converged hog curvature are recorded, with the touchdown curvature
that the deck's 10 m laid elements under-resolve, and the chain-only
twin matches stock MoorDyn to 0.3 %. With the anchor-end correction the
laid cable's anchor tension stays at a median of 1.39 kN with sub-second
stick-slip peaks; before it, it locked at about 2.75 kN.

Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
@SMI-Lab-Inha

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector

chatgpt-codex-connector Bot commented Oct 14, 2026 •

Copy link
Copy Markdown

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review Status Commit Review trigger
📝 Code Review ✅ Completed 2026-10-14T23:55:14.749032Z 16e3a27 Manual request
ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

@read-the-docs-community

read-the-docs-community Bot commented Oct 14, 2026 •

Copy link
Copy Markdown

Documentation build overview

📚 CableDyn | 🛠️ Build #35154893 | 📁 Comparing 16e3a27 against latest (f27c5e3)

  🔍 Preview build  

11 files changed · ± 11 modified

± Modified

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 63c130f129

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread src/CableDyn_HermiteCableDynamic.f90 Resolved
Review found that the floor carried the end element's weight moment
whenever the held end node touched the seabed, so a taut line anchored
on the floor whose first element rises into the water lost the end
moment of that element's weight: the floor does not support it.

The support now also needs the element to lie on the floor: it is
scaled by a C1 smoothstep S(r) of r = N_in / (w l0 / 2), the elastic
normal force of the element's inner node from the contact law the
assembly uses over its half share of the element weight. A laid element
(r about 2) keeps the full support; an element whose inner node is
clear of the floor gets none, and its line is bit-identical to the same
line without a seabed; in between the term is smooth and its tangent is
assembled in the statics and the dynamics. A test checks the taut line
anchored on the floor against the line without a seabed, statics and
held dynamic steps, bit for bit.

The slope-polish test of hermite_cable_static failed on Linux: its
layout touched down at a node with neighbouring equilibria 3 cm apart,
and under some BLAS kernels even its reference state was unstable. The
fairlead now sits at x = 200 m, where every kernel reaches the same
stable reference, every clear node stands 0.3 m above the floor, and
both polishes return to it from four times the perturbation the test
uses (default, Prescott, Haswell and SkylakeX kernels; Release and
Debug).

The shipped finite-EI examples with a grounded anchor now change by at
most 0.34 N from dev.

Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
@SMI-Lab-Inha

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector chatgpt-codex-connector Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: c747e7e357

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

Comment thread src/CableDyn_HermiteCableStatic.f90 Resolved
Review found that inside the transition of the end support (0 < S < 1)
the energy-minimisation potential varied with the inner node through S,
while the residual carries the support load only on the end tangent, so
the potential was not a potential of the residual and the Armijo test
compared R.dq with the derivative of a different function.

Each energy line search now holds the share S0 of its starting point;
with S fixed, the gradient of the support energy is exactly the
residual's support load at every trial point. The hold is released when
a step is accepted or the search stalls, and the torsion descent, which
compares energies the same way, holds it too. The alternative, a
conjugate force on the inner node (about 0.9 MN here), is a contact
force the floor does not exert and would have changed every equilibrium
and the dynamics; the support load stays non-conservative like current
drag. A new check compares central differences of the held-share
potential with R.d along four directions with the inner node mid
touchdown (agreement 1e-5; with the share free the inner-node term is
off by 8.9e5). Every example output is unchanged.

Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
@SMI-Lab-Inha

Copy link
Copy Markdown
Owner Author

@codex review

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. Hooray!

Reviewed commit: 16e3a27d35

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

  • Open a pull request for review
  • Mark a draft as ready
  • Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

@SMI-Lab-Inha
SMI-Lab-Inha merged commit 8878f99 into dev Oct 15, 2026
15 checks passed
@SMI-Lab-Inha
SMI-Lab-Inha deleted the fix/seabed-friction-anchor branch October 15, 2026 01:15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant