Repository navigation
feat(openfast): warn when line nodes lie outside the SeaState wave grid - #24
Conversation
SeaState evaluates its wave kinematics on a grid (X_HalfWidth, Y_HalfWidth, Z_Depth) and clamps a point outside it to the boundary. The r-test MD_Shared SeaState inputs, which the OpenFAST tutorial and examples/openfast reuse, span 100 m and 40 m depth: they cover the platform but not the moorings, whose nodes reach 840 m and 200 m depth. Those nodes took the kinematics of the grid boundary, and their fluid loads were wrong without a message: SeaState's own clamping notice is a module-initialisation warning, which the OpenFAST glue discards. The shell now checks the positions it samples, at the static equilibrium and at every CableDyn step until it has warned once, and prints one warning with the number of nodes outside the grid, the farthest node in each direction and the SeaState inputs to enlarge. Waves are checked in x, y and depth, a SeaState current in depth only; nodes below a grid that reaches the seabed are not reported. FAST.Farm and still water sample nothing and check nothing. After the warning the check costs nothing. The tutorial and examples/openfast now enlarge the grid (half-widths 900 m, Z_Depth to the seabed) and quote their results for it: with the MD_Shared grid the line-1 fairlead tension standard deviation was 18 % too high (0.136 against 0.115 MN). The manual's OpenFAST page gains a SeaState grid section. The troubleshooting table had its header row below two rows, which now sit under it. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
The SeaState grid check counted every line node, so a rope or jumper above water reaching beyond the grid triggered the warning although it carries no wave load there. It now runs after sampling and counts only nodes at or below the sampled free surface (and at or below the still water level without wave stretching). The message states that it is issued once in its first sentence and is built without truncation, dropping the farthest-node details if they would not fit. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
Rerun the OpenFAST tutorial cases with the current solver on the enlarged SeaState grid (CableDyn and MoorDyn, supercycling, still water, power cable, line failure), on the MD_Shared grid, and on the two refined grids, and recompute the quoted figures with the same post-processing. All quoted tensions, statistics, the grid warning and the memory estimate are unchanged at their printed precision. Three figures change: the t = 0 Curv4N38 of the power-cable case (9.01821E-02 to 9.01822E-02 1/m), the grid-refinement bound in the SeaState grid section (less than 0.05 % to less than 0.1 %, since NX = 73 moves the mean surge by 0.08 %), and the wall time of the 400 s line-failure run (about 30 s to about 60 s). Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
The tutorial's run time is about a third SeaState's wave grid (6 of 16 s), not most of it; the quoted dtM console lines match the padding the shell prints; a test comment quotes the t = 0 FairTen1 of the example with its enlarged SeaState grid. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 30502d9211
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".
Under Wheeler stretching (WaveStMod = 3) SeaState interpolates its wave kinematics at the node depth mapped linearly from [-EffWtrDpth, wave elevation] to [-EffWtrDpth, 0], not at the node's physical depth. The wave-grid check compared the physical depth with -Z_Depth, so a node just above the grid bottom under a crest, whose mapped depth lies below it, was missed, and one under a trough could be reported needlessly. The check now uses the same mapping (clamped to the still-water level, as SeaState does). The other stretching models are unchanged. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: d5c9f92abe
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".
With the host waves off and the host current on, fx_wl takes the still-water level for CableDyn's wetting, while SeaState still interpolates the current at the depth Wheeler-mapped with the wave elevation it sampled. The grid check now keeps that sampled elevation per node and applies the mapping whenever SeaState stretches with WaveStMod 3, not only when the host waves are on. The manual says that a dynamic (InflowWind-driven) current field, read on its own grid, is not part of the SeaState grid check. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4a46decfff
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".
… check runs The per-node SeaState elevation that the wave-grid check reads was allocated on every mooring step, also after the one warning had been issued. It is now allocated and filled only while the check is still active; once it has warned, the sampling step allocates nothing. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2d67c3ca7f
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".
SeaState measures its wave field and kinematics grid from the still-water level, and its node queries take SWL-relative positions, while the mooring samples it at positions relative to the mean sea level. With MSL2SWL /= 0 the sampled kinematics, the waterline and the wave-grid check were all shifted by MSL2SWL. A run with host waves or current now refuses a nonzero MSL2SWL at initialisation, by name. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2a9736d805
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".
The MSL2SWL refusal ran before the deck's WaterKin policy was known, so a deck taking both its waves and current from WaterKin (which never queries SeaState) was refused too. It now runs where the samples are taken, only when host waves or current are enabled. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
|
Codex Review: Didn't find any major issues. Bravo. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
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". |
A dynamically populated current field (an InflowWind-driven MHK current) is read on its own field, at the node position raised by the water depth, not on the SeaState grid, so the SeaState grid check did not cover it. For a gridded (Grid3D) current field the shell now also checks the submerged sampling nodes against the field's vertical extent [GridBase, GridBase + 2 ZHWid] and lateral half-width (in the field's frame when the box is rotated) and warns once, with the farthest excursion. Other field types have no extent to check. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 595dd0eaa2
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".
The InflowWind current-grid warning wrote six real values with five ES descriptors, which raises a runtime I/O error when it fires. The format now has six; the message was checked with gfortran -fcheck=all. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
Summary
SeaState evaluates its wave kinematics on a grid (
X_HalfWidth,Y_HalfWidth,Z_Depth) and clamps any point outside it to the boundary. The grid in r-test's MD_Shared SeaState inputs, reused by the OpenFAST tutorial andexamples/openfast, spans 100 m and 40 m depth. That covers the platform but not the moorings, whose nodes reach 840 m out and 200 m deep. Those nodes took the boundary kinematics, and their fluid loads were wrong with no message: SeaState's own clamping notice is a module-initialisation warning, which the OpenFAST glue discards.The warning. The CableDyn shell now checks the positions it samples, at the static equilibrium and at every CableDyn step until it has warned once. It then prints one warning:
The message is built without truncation and drops the farthest-node details if they would not fit.
What is checked:
The tutorial and
examples/openfastnow enlarge the grid (half-widths 900 m,Z_Depthto the seabed) and quote their results for it. With the MD_Shared grid, the line-1 fairlead tension standard deviation was 18 % too high (0.136 against 0.115 MN).Docs: the manual's OpenFAST page gains a SeaState grid section. A troubleshooting table whose header row sat below two rows is fixed.
Tests
pre-commit run --all-filesand the Sphinx build with-Wpass.Checklist
pre-commit run --all-filespassesctest --test-dir build --output-on-failurepassespython -m pytest tests python/testspasses (if Python or validation records changed)[Unreleased](heading added if missing) for user-visible changesdoc/updated for changed inputs, options, or outputsfeat:,fix:,docs:, ...)git commit -s) under theDeveloper Certificate of Origin