Repository navigation
feat(damping): mesh-independent axial damping ratio; static balance and as-laid fixes - #112
Conversation
|
@codex review |
Codex Review Summary: ✅ CompletedThis 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. |
|
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". |
2decd36 to
6de62cd
Compare
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6de62cdee9
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".
|
@codex review |
|
Codex Review: Didn't find any major issues. 🎉 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". |
…mode A negative LINE TYPES BA is MoorDyn's damping ratio per segment, BA = zeta L0 sqrt(EA m) on each element, so the first axial mode of a line is damped at about zeta pi/(2N) on N elements and less on a finer mesh. - OPTION axialDampingRatio (axial_damping_ratio): segment (the default, unchanged) or mode1. Under mode1 a negative BA is the damping ratio of the line's first axial mode, the same on any mesh: BA = 2 zeta EA / omega_1 on every element, a damping proportional to the stiffness, with omega_1 the first axial frequency of the line with both ends held from its sections (CD_Axial_Mode1_Omega, a fixed internal mesh), so a composite line is damped at zeta too. It applies on every route that builds lines from a deck (EI = 0 and finite-EI lines); a line with a negative BA and a viscoelastic or Syrope section is refused by name. The Python deck validator accepts the same values. - An initialisation note for every line with a negative BA per segment: the damping ratio of its first axial mode on its final mesh, from the resolved BA (CD_Axial_Mode1_Damping, the Rayleigh quotient of the axial chain), and the element count. It reaches the driver report, CableDyn_GetInitNotes and the OpenFAST console. - Fix: a negative BA of an EI = 0 line was resolved with the length, EA and mass of the element at the mirrored position from the other end, so the sections of an asymmetric composite line exchanged their axial damping. Of the examples, composite_chain_syrope_chain.dat changes (the largest difference 619 N, 6.6e-4 of the value); the others are bit-identical. driver_axial_damping_ratio fits the free decay of a taut line's first axial mode: zeta pi/(2N) at N = 50, 100 and 200 within 0.1 % under the segment form, with the note; 2 % on every mesh under mode1, uniform and composite, EI = 0 and finite-EI; and the composite segment form at the ratio of its own elements' damping (27 % off before the fix). Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
…rent The end reactions of a finite-EI cable at rest in a uniform current (CD_HermiteCable_Dyn_End_Force: FairTen/AnchTen, Point<P>F) balance its submerged weight and its Morison drag to the static tolerance, about 2e-8 of the end force, with pinned (bending-free) and clamped ends, suspended and touching down, in currents of 0.5 to 2 m/s at headings 0 to 180 deg across the plane of the ends, on 25- and 100-element meshes. static_current_balance integrates the documented drag law independently on the converged cubic-Hermite centreline: with the solver's 3-point rule the balance closes to 2.3e-8; a 10-point rule differs by the drag quadrature, 7.4e-8 on 100 elements and 2.0e-6 on 25. An imbalance near 1e-4 comes from the check, not the equilibrium: integrated on straight chords between the nodes the drag misses by a share that falls with the square of the element length (2.8e-5 on 1.25 m elements in a 1 m/s current, 1.1e-4 on 2.5 m, 4.5e-4 on 5 m), and per unstretched length by the axial strain times the drag (7e-5 here). theory.rst now states the length each path takes the drag over (unstretched on EI = 0 lines, the deformed centreline on finite-EI lines) and how a check must integrate it. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
…solver applies With staticsFriction asLaid on a bathymetry, the friction reference of a node moves with its sink into the penalty seabed, which on a slope shifts it horizontally by its load over its seabed stiffness. The load was the submerged weight of the line and the net in-water weight of an attachment wherever the node lay. The static solve and the dynamics give back the buoyancy of the dry part of a finite-EI section within its radius of the still-water level, and an attachment above the surface carries its dry weight. On a seabed shallower than a cable's radius the reference therefore sank less than the node, and the run rested off its laid path across the slope, held there by a spurious friction force. CD_HermiteCable_As_Laid_Offsets now sinks each node under the normal load the solver applies at its laid position: the line laid without its sink, each node at the height its load sinks it to, and the dry part's load there (CD_HermiteCable_Dry_Buoyancy, the solver's own law) solved by Newton's method with its exact height derivative; attachments above the surface add their buoyancy back. The section diameter travels with the friction reference by reference arc, so the reference rebuilt on a refined mesh uses it too. Where the line lies below the surface by more than its radius nothing changes, bit for bit. The seabed lies below the still-water level everywhere (WtrDpth and bathymetry depths must be positive), so a grounded node with an attachment stays submerged and an EI = 0 line, whose submergence follows its node heights, is wholly submerged where it is grounded: the attachment case cannot arise and is handled by the same rule. static_friction_as_laid gains shallow_cross: a 0.5 m cable laid on a seabed 0.1 m deep with a 5 degree cross slope. Its run lay 7.75e-4 m off its laid path (7.72e-4 m predicted from the two sinks); it now rests 4.4e-6 m off it, and a march from the static state stays at rest. Every example is bit-identical. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
The [Unreleased] entries of the open changes are added together in one change, so that they do not conflict with each other. Signed-off-by: Jae Hoon Seo <jaehoon.seo@inha.ac.kr>
|
@codex review |
3d99c95 to
1c6bf8e
Compare
|
Codex Review: Didn't find any major issues. 🎉 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". |
…residual The as-laid reference's dry-part load is solved by Newton's method. On a net-buoyant cable over a seabed shallower than its radius, a node held off the floor is wholly dry and one sunk under that load wholly wet, so the full step cycled between the two (about -208 and 728 N/m for a 0.5 m cable of 180 kg/m on a bed 0.01 m deep with kBot = 1000) and initialisation failed with "the dry-part load of the laid line did not converge". A step that does not reduce the residual's 2-norm (Armijo, slope 1e-4) is now halved from the last accepted loads, down to 2^-30, and the iteration budget is 200. The case converges at the 70.3 N/m root, and a march from its static state stays at rest (drift 1.5e-8). static_friction_as_laid gains buoyant_shallow, which failed before. Full CTest 370/370 on both BLAS kernels. 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". |
This PR has three commits, one per item.
1. Negative BA (
feat(damping), plus one fix)A negative LINE TYPES BA is MoorDyn's damping ratio per segment, BA = ζ L0 √(EA m). On N elements it damps a line's first axial mode at about ζπ/(2N), so the result depends on the mesh. The free axial decay of a taut line confirms this: 3.140 / 1.570 / 0.785 % at N = 50 / 100 / 200, against 3.142 / 1.571 / 0.785 % predicted.
axialDampingRatio(axial_damping_ratio) option.segmentis the default and is unchanged.mode1makes a negative BA the damping ratio of the line's first axial mode on any mesh: BA = 2ζ EA/ω₁, which is stiffness-proportional.CableDyn_GetInitNotesand the OpenFAST console.composite_chain_syrope_chain.datchanges: 619 N at most, 6.6e-4 of the value.2. Static force balance of a finite-EI cable in current (
test(statics))Across 88 cases (pinned or clamped ends, suspended or touchdown, 0.5–2 m/s, four headings, 100 or 25 elements), end reactions + weight + drag balance to 2.3e-8 of the end force with the solver's own quadrature. That is the static tolerance.
An independent check that integrates the drag on straight chords sees an O(h²) difference: 2.8e-5 / 1.1e-4 / 4.5e-4 at 1.25 / 2.5 / 5 m elements. That is the size reported earlier as "1 N in 11.8 kN", so it is not a solver defect. The new test
static_current_balancegates the balance, andtheory.rststates the length measure the drag is integrated over on each route. No code changes.3. As-laid seabed friction reference (
fix(friction))A finite-EI section on a seabed shallower than its radius is partly dry. The solver gives back the dry part's buoyancy, but the as-laid reference still sank each node under the full submerged weight.
The reference now sinks each node under the load the solver applies, including the dry part, by Newton's method with its exact derivative.
Tests
-Ware clean.