CalculiX Modal Analysis
SkillWeb & browsingThis skill gives your AI the workflow for CalculiX modal analysis, so it can find a model's natural frequencies and mode shapes — the ways a structure naturally vibrates. Once added, your AI can compute those results, read the frequency table, and display any mode shape in the browser viewer, which helps when checking a design for resonance.
Available today. Use it from your connected AI after setup.
No other account needed.
Add the skill, then ask your AI to run a modal analysis on your CalculiX model. It will report the natural frequencies and can render individual mode shapes in the browser for you to inspect.
Then ask your AI: use the CalculiX Modal Analysis skill
What your AI can do with it
- Calculate the natural frequencies of a CalculiX model
- Determine the matching mode shapes
- Read back the results as a frequency table
- Render any mode shape in the browser viewer
- Check a design for resonance
What this skill tells your AI
The instructions your AI receives, as published by cai-aa/cae-agent-hub in Skill/calculix/calculix-modal-analysis/SKILL.md and read by ahel’s review.
Free-vibration eigenanalysis on a CalculiX deck: a *FREQUENCY step requests N
eigenpairs; ccx prints the eigenvalue table and per-mode eigenvectors to the
.dat, so both frequencies and mode shapes parse from text (no .frd needed).
When to Use
Use when an agent must extract natural frequencies, check resonance/vibration
margins, or visualize mode shapes on a CalculiX model. Driven by the same MCP
tools as static runs (run_solver → read_results → export_results).
Not for: static stress/deflection (use calculix-fem), transient/dynamic
response (not yet supported), or sizing optimization (calculix-sizing-optimization).
Workflow
- Confirm the deck has
*DENSITYunder its*MATERIAL— frequencies need mass; without density ccx fails the eigenvalue solve. - Confirm the step is
*FREQUENCYwith the wanted mode count on its data line, and*NODE PRINT, NSET=<all>/Uso eigenvectors reach the.dat. No load is applied (free vibration); keep the*BOUNDARYclamp set. run_solver_tool— submit the deck. Note: a*FREQUENCYrun writes a header-only.sta(no increments in an eigenvalue solve); success accepts a non-empty.datinstead, so this is normal, not a failure.read_results_tool— returnsfrequenciesas[{mode, eigenvalue, freq_rad_s, freq_hz}]andn_modes(a doubly symmetric section gives degenerate pairs — f1 = f2 — which is expected).export_results_toolwithmode=N— writesresult_mesh.jsonholding that mode's eigenvector as a stress-free displacement field; the viewer renders the mode shape with its usual auto-scaled deformation.
Rules
- Units follow the
.inp(commonly mm-t-s-MPa → frequencies in Hz). - Eigenvectors are mass-normalized; their magnitude carries no physical displacement meaning — only the shape does. The viewer auto-scales.
- Fully-integrated C3D8 hexes shear-lock in bending: ccx frequencies run ~5-10% ABOVE the Euler-Bernoulli hand calc. For tight margins, refine the mesh through the thickness (or switch element type) and report the gap.
- Hand-calc check for a clamped-free bar: f_n = (beta_n^2 / 2pi) * sqrt(E I / (rho A L^4)), beta_1 = 1.8751.
Example
MCP/CalculiX/examples/cantilever_modal.inp is the public cantilever benchmark
with a 5-mode *FREQUENCY step. ccx gives f1 = f2 ~ 502 Hz (degenerate bending
pair on the square section) vs the Euler-Bernoulli hand calc ~ 464 Hz (+8%,
C3D8 shear locking). Mode 1 exports straight into the viewer as the classic
half-sine bend.
Signals
- GitHub stars
- 880
- Forks
- 115
- Last commit
- Sep 2026
Advanced
- Catalog kind
- skill
- Gateway key
calculix-modal-analysis- Source
- github.com/cai-aa/cae-agent-hub