White box / Analysis methods / Modal analysis

Sparse generalized eigenproblem

Analysis method · assumptions, essential equations, full derivation, algorithm, code and verification references.

Browse another sheet Analysis methods / Modal analysis
Analysis methodsModal analysisSheet 1 / 1
Analysis methodROUTE-DEPENDENT · registry controls each family

The default sparse path uses eigsh/ARPACK; dense conversion is explicit and bounded by an active-DOF cap.

01 Model and scope

Constrained stiffness and mass define a generalized eigenproblem. Positive eigenvalues map to natural frequencies; eigenvectors are arbitrary up to scale and sign, so residual, mass orthogonality and subspace comparisons matter alongside frequency values.

Route assumptions and calculation checks

  1. Mass positive-definiteness, residual, mass orthogonality, and effective modal mass are recorded diagnostics.
  2. Eigenvector sign and scale are arbitrary; repeated modes should be compared as a subspace.
  3. Shell drilling DOFs without mass may be condensed and reconstructed under a documented route.

Useful foundations and prerequisites

02 Essential equations and symbols

Kffφi=λiMffφiK_{\mathrm{ff}} \varphi{}_{i} = \lambda{}_{i} M_{\mathrm{ff}} \varphi{}_{i}ωi=λi;fi=ωi/(2π)\omega{}_{i} = \sqrt{\lambda_i}; f_{i} = \omega{}_{i}/(2π)ri=∣∣Kffφi−λiMffφi∣∣2/max(∣∣Kffφi∣∣2,∣λi∣∣∣Mffφi∣∣2,1)r_{i} = ||K_{\mathrm{ff}}\varphi{}_{i}-\lambda{}_{i}M_{\mathrm{ff}}\varphi{}_{i}||_{2} / max(||K_{\mathrm{ff}}\varphi{}_{i}||_{2}, |\lambda{}_{i}| ||M_{\mathrm{ff}}\varphi{}_{i}||_{2}, 1)ΦTMffΦ≈IΦ^T M_{\mathrm{ff}} Φ \approx{} I

Definitions and conventions

Kff,MffK_{\mathrm{ff}}, M_{\mathrm{ff}}
Stiffness and mass matrices after fixed DOFs and documented massless coordinates are eliminated.
λi=ωi2\lambda{}_{i} = \omega{}_{i}^2
Eigenvalue equals squared circular natural frequency; λ must be positive for a real oscillation frequency.
φi\varphi{}_{i}
Mode shape: relative displacement pattern, arbitrary in sign and scale until a normalization convention is applied.
rᵢ
Dimensionless eigenpair residual; a small value indicates the computed vector satisfies the discrete equation.
ΦTMffΦΦ^T M_{\mathrm{ff}} Φ
Mass orthogonality for distinct modes under mass normalization; degenerate/repeated eigenspaces need subspace comparisons.
fᵢ
Natural frequency in hertz; distinct from angular frequency ω in radians per second.
Sparse generalized eigenproblem · method mapThe constrained mass and stiffness define eigenpairs; frequency, shape, residual, and mass orthogonality are checked together.
4 detailed steps

03 Step by step derivation

Follow the assumptions, equations and discrete operations in order. The full explanation stays visible.

With homogeneous constraints eliminated, free coordinates satisfy M_ff q̈ + K_ff q = 0. Assume harmonic motion q(t)=φe^{iωt}: then (K_ff−ω²M_ff)φ=0. Set λ=ω²; only positive λ gives a real natural frequency f=√λ/(2π). A computed pair is checked with its residual, while mass-normalized distinct modes satisfy ΦᵀM_ffΦ≈I. These checks diagnose numerical quality; they do not prove every material/element/constraint route is qualified.

1 · Reduce to the free dynamic coordinates

After supports and documented massless coordinates are handled, the semi-discrete undamped system has free-coordinate mass and stiffness matrices. Positive-definite mass on the retained subspace is a precondition for interpreting the eigenvalues as ordinary structural frequencies.

Mffq¨+Kffq=0\mathbf M_{ff}\ddot{\mathbf q}+\mathbf K_{ff}\mathbf q=\mathbf0

2 · Assume harmonic motion

Substitute a sinusoidal mode q(t)=φ exp(iωt). The acceleration contributes −ω²φ, giving the generalized eigenproblem. The solver sets λ=ω²; only positive real λ yields a real undamped natural frequency.

q(t)=ϕeiωt,(Kff−ω2Mff)ϕ=0\mathbf q(t)=\boldsymbol\phi e^{i\omega t},\qquad (\mathbf K_{ff}-\omega^2\mathbf M_{ff})\boldsymbol\phi=\mathbf0Kffϕi=λiMffϕi,fi=λi2π\mathbf K_{ff}\boldsymbol\phi_i=\lambda_i\mathbf M_{ff}\boldsymbol\phi_i,\qquad f_i=\frac{\sqrt{\lambda_i}}{2\pi}

3 · Normalize and compare modes correctly

Mass normalization makes distinct modes orthonormal under M when the matrices are symmetric and the eigenvalues are distinct. The sign and scale of each eigenvector are arbitrary. Nearly repeated modes may rotate within the same eigenspace, so compare subspaces or mass-weighted correlation rather than component signs alone.

ϕiTMffϕj=δij\boldsymbol\phi_i^T\mathbf M_{ff}\boldsymbol\phi_j=\delta_{ij}R=ΦTMffΦ≈I\mathbf R=\boldsymbol\Phi^T\mathbf M_{ff}\boldsymbol\Phi\approx\mathbf I

4 · Check solver quality and report the route

Record residuals, orthogonality, effective modal mass, mass definiteness, requested mode count, and the selected sparse or bounded dense route. Shell drilling coordinates without physical mass may need documented condensation and reconstruction; that transformation is part of the result's provenance.

ri=∥Kffϕi−λiMffϕi∥2max⁡(∥Kffϕi∥2,∣λi∣∥Mffϕi∥2,1)r_i=\frac{\|\mathbf K_{ff}\boldsymbol\phi_i-\lambda_i\mathbf M_{ff}\boldsymbol\phi_i\|_2}{\max(\|\mathbf K_{ff}\boldsymbol\phi_i\|_2,|\lambda_i|\|\mathbf M_{ff}\boldsymbol\phi_i\|_2,1)}

04 Algorithm and implementation

Read the linked versioned source for the actual assembly and solver behavior. The derivation above explains the mathematics and is not a verbatim source excerpt.

05 Verification and limitations

ROUTE-DEPENDENT · registry controls each family

TET4 and MITC4 cases expose frequencies, residuals, and mode-shape comparisons.

Each reference supports its own model, observable and analysis route. Read the original contract and limitations before transferring a conclusion.

Recorded repository calculations

See this method in a calculation.

Each case keeps its mesh, measured quantities, source record and limitations attached to the result.

Browse every calculation case →

06 Sources and related sheets

Continue the reasoning

Back to sheet contents ↑