Skip to content

fix(openfast): restart a finite-EI cable from its committed boundary - #54

Merged
SMI-Lab-Inha merged 1 commit into
devfrom
fix/restart-checkpoint-row
Oct 26, 2026
Merged

SMI-Lab-Inha merged 1 commit into
devfrom
fix/restart-checkpoint-row

Conversation

@SMI-Lab-Inha

Copy link
Copy Markdown
Owner

Problem

openfast -restart did not continue a run with a finite-EI cable bit for bit.

The checkpoint rebuild moves each coupled end to the restored input mesh, which holds the glue's
input at the restart time. The cable's mirror held only its free DOFs, so its fairlead kept the
restored mesh's bits instead of the boundary the uninterrupted run committed. (With
NumCrctn = 0, that run stepped with the extrapolated input.) The EI = 0 lines already reload
their committed coupled motion.

Reproduction: the IEA-15 mixed deck (chains plus the lazy-wave cable), restarted at 6 s.

  • The checkpoint row itself matched.
  • From 8.8 s on, the cable's .CD.out channels drifted by up to 1.0e-7 relative (AnchTen4).
  • With supercycling (dtM = 2 DT), the drift reached 9.4e-7.
  • The chain-only deck stayed exact.

A per-step bit-checksum dump of the cable state pinned the cause. At the first step after the
restart, the inputs, accelerations, end-connection directions and parent frame matched the
continuous run, but q, v and the cached committed force differed in their last bits.

Change

The cable mirror gains a boundary block (CD_HFMF_BoundaryMirrorSize), reloaded after the move,
holding:

  • [q; v; a] of the held DOFs. The held accelerations move here from the force-blend block, so the
    configuration blend restores them too.
  • The end-connection directions.
  • The parent frame: orientation, angular velocity and angular acceleration.

Unpacking checks the orientation before it writes anything. A torsion cable's parent frame now
reloads as committed, instead of being re-derived from its end frames. The OpenFAST shell
accounts for the block in its free-DOF count, and its comments on the coupled-motion mirror now
describe what the code does.

Evidence

OpenFAST v5 built with gfortran on this branch; IEA-15 mixed deck, checkpoint at 6 s, run to 10 s.
Every .CD.out and OpenFAST output row from 6 to 10 s is compared:

case before after
dtM = DT, NumCrctn 0 6 rows differ (1.0e-7) identical
NumCrctn 1 6 rows differ (1.0e-7) identical
dtM = 2 DT 3 rows differ (9.4e-7) identical
rigid fairlead END CONNECTIONS 5 rows differ (1.0e-7) identical

The mirror test covers the new layout, its round trip and the orientation check. ctest 302/302,
pytest 2069 passed. CHANGELOG, VALIDATION.md (checkpoint-restart row) and ARCHITECTURE.md
updated.

openfast -restart did not continue a run with a finite-EI cable bit for
bit. The checkpoint rebuild moves each coupled end to the restored input
mesh, the glue's input at the restart time, and the cable's mirror held
only its free DOFs: its fairlead kept the restored mesh's bits instead of
the boundary the uninterrupted run committed (with NumCrctn = 0 that run
stepped with the extrapolated input). The EI=0 lines already reload
their committed coupled motion. On the IEA-15 mixed deck restarted at
6 s, the lazy-wave cable's .CD.out channels drifted by up to 9.4e-7
(relative) within 4 s, while the chain-only deck stayed exact.

The cable mirror gains a boundary block, reloaded after the move: [q; v;
a] of the held DOFs (the held accelerations move here from the
force-blend block, so the configuration blend restores them too), the
end-connection directions and the parent frame (orientation, angular
velocity and acceleration). Unpacking checks the orientation before it
writes anything. The parent frame of a torsion cable now reloads as
committed instead of being re-derived from its end frames.

With the fix every .CD.out and OpenFAST row from 6 s to 10 s matches the
uninterrupted run bit for bit at dtM = DT and 2 DT, NumCrctn 0 and 1,
and with a rigid fairlead END CONNECTIONS row (before: 3 to 6 rows of
the cable's channels differed in each). The mirror test covers the new
layout, its round trip and the orientation check.

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 26, 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 2026-10-26T14:14:46.127790Z 6f38248 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

Copy link
Copy Markdown

Documentation build overview

📚 CableDyn | 🛠️ Build #35414955 | 📁 Comparing 6f38248 against latest (fd1936d)

  🔍 Preview build  

14 files changed · ± 14 modified

± Modified

@chatgpt-codex-connector

Copy link
Copy Markdown

Codex Review: Didn't find any major issues. 🚀

Reviewed commit: 6f38248596

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 f267527 into dev Oct 26, 2026
15 checks passed
@SMI-Lab-Inha
SMI-Lab-Inha deleted the fix/restart-checkpoint-row branch October 26, 2026 15:38
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