Skip to content

fix(openfast): linearise from double-precision loads in a single-precision OpenFAST - #85

Merged
SMI-Lab-Inha merged 1 commit into
devfrom
fix/lin-single-precision
Nov 3, 2026
Merged

SMI-Lab-Inha merged 1 commit into
devfrom
fix/lin-single-precision

Conversation

@SMI-Lab-Inha

Copy link
Copy Markdown
Owner

Summary

The release openfast.exe is built as OpenFAST's Visual Studio Release configuration, which is single precision (ReKi = 4 bytes). The 0.4.0 linearisation of the line dynamics differenced the ReKi output mesh over a 1 µm position step. With fairlead loads of a few MN, the 7-digit mesh cannot resolve the change, so the release executable linearised the mooring wrongly:

  • the IEA-15 MW UMaine chains' surge stiffness read 3.2e3 N/m against MoorDyn-F's 7.7e4 N/m;
  • the CableDyn stiffness block was asymmetric by 1.18;
  • A and B and the line modes were unaffected; C and D were off by 4e-4.

The gfortran CMake builds this feature was validated with are double precision.

lin_eval now keeps its coupled loads and deck channels in double precision, and the two linearisation Jacobians pack those (pack_lin_outputs) in place of the ReKi mesh. A double-precision build packs the same values.

Checks

validation/scripts/openfast_lin_ab.py on the IEA-15 MW UMaine twin decks plus the mixed cable deck:

Build Result
Release workflow openfast.exe (Intel, single) FAIL: surge 3.20e3 N/m, symmetry 1.18
gfortran, DOUBLE_PRECISION=OFF, before FAIL, identical to the release executable
gfortran, DOUBLE_PRECISION=OFF, after PASS: surge 7.721e4 vs 7.716e4 N/m, symmetry 7e-6, fairlead blocks to 1.1e-3, cable deck to its static solves to 1e-4
gfortran, double (the release tree) PASS, unchanged
  • RELEASE.md now asks for the linearisation check on the release executable too.
  • pre-commit, Sphinx -W.
  • The shell compiles only inside OpenFAST; built here with gfortran in both precisions. The Intel build is the release workflow's, rerun on this change.

…ision OpenFAST

The release openfast.exe is OpenFAST's Visual Studio Release
configuration, single precision (ReKi = 4 bytes). The linearisation of
the line dynamics differenced the ReKi output mesh over a 1 um position
step; with fairlead loads of a few MN the 7-digit mesh cannot resolve
the change, and the reduced mooring stiffness of the IEA-15 MW UMaine
chains read 3.2e3 N/m in surge against MoorDyn-F's 7.7e4 N/m (symmetry
error 1.18), with the line modes, A and B unaffected. The gfortran CMake
builds this was validated with are double precision.

lin_eval now keeps its coupled loads and deck channels in double
precision, and the two linearisation Jacobians pack those
(pack_lin_outputs) in place of the ReKi mesh. A double-precision build
packs the same values. RELEASE.md asks for the linearisation check on
the release executable as well.

Checked with gfortran OpenFAST builds at DOUBLE_PRECISION=OFF: before,
openfast_lin_ab.py fails as the release executable does; after, it
passes (surge 7.721e4 vs 7.716e4 N/m, symmetry 7e-6, fairlead blocks to
1.1e-3, the mixed cable deck's point stiffness to its static solves to
1e-4).

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 Nov 3, 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 2026-11-03T13:50:48.441862Z ca0a1c0 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.

@chatgpt-codex-connector

Copy link
Copy Markdown

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

Reviewed commit: ca0a1c0c9c

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".

@read-the-docs-community

Copy link
Copy Markdown

Documentation build overview

📚 CableDyn | 🛠️ Build #35613944 | 📁 Comparing ca0a1c0 against latest (fd1936d)

  🔍 Preview build  

33 files changed · ± 33 modified

± Modified

@SMI-Lab-Inha
SMI-Lab-Inha merged commit f530efa into dev Nov 3, 2026
15 checks passed
@SMI-Lab-Inha SMI-Lab-Inha mentioned this pull request Nov 3, 2026
@SMI-Lab-Inha
SMI-Lab-Inha deleted the fix/lin-single-precision branch November 3, 2026 19:21
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