Random Vibration FEA

Technical provenance

Applicable standards / specifications

• IEC 60068-2-64 (2008+AMD1:2019) — Environmental testing — Part 2-64: Tests — Test Fh: Vibration, broadband random and guidance
• GSFC-STD-7000B (B (2021)) — General Environmental Verification Standard for GSFC Flight Programs and Projects — Applicable to relevant GSFC flight programmes; not a universal random-vibration specification.

References

• Newland, D. E. — An Introduction to Random Vibrations, Spectral & Wavelet Analysis — Background reference for PSD-based random processes and structural response.
• NASA GSFC-STD-7000B — General Environmental Verification Standard (GEVS) — Space-hardware environmental verification reference; project-specific environments govern.

What Is It?

Random vibration FEA combines a modal analysis with a PSD input to predict the statistical response of a structure to broad-band random excitation. The output is not a time history or a deterministic peak — it is a statistical description of the response: RMS displacement, RMS acceleration, RMS stress or PSD curves at selected locations. This is the standard method for assessing structural response to acoustic, launch and transportation vibration environments.

Why It Matters

Random vibration is the governing environment for many aerospace, defence and automotive applications. Launch vehicle payload qualification, aircraft equipment qualification and electronics vibration testing are all based on random vibration PSD inputs. Random vibration FEA is the analytical method for predicting whether a structure will survive these environments. It connects the input environment to the structural response to the stress that determines fatigue and strength.

The quality of a random vibration result depends on the modal model underneath it. If the modal basis is inadequate, the random vibration result is unreliable — no matter how well the PSD input is defined.

Typical Workflow

Random vibration FEA follows a defined workflow. Each step builds on the previous one, and errors at any stage propagate through the entire chain.

1. Modal analysis — extract natural frequencies and mode shapes over the relevant frequency range
2. Input PSD definition — define the acceleration PSD environment (base excitation) or force PSD
3. Frequency response computation — compute transfer functions from input to response locations
4. Modal combination — combine modal responses statistically (typically SRSS or CQC)
5. Response PSD — compute the PSD of the response quantity at each location
6. RMS response — integrate the response PSD to obtain RMS values

Modal analysis → Input PSD definition → Frequency response computation → Modal combination / statistical response → Response PSD → RMS response

Assumptions

Random vibration FEA rests on several assumptions. Understanding these is essential for knowing when the method is valid and when its results may be unreliable.

Base Excitation vs Force PSD Excitation

Random vibration FEA can be driven by base excitation (acceleration PSD applied to supports) or by force PSD (force spectral density applied at nodes). Base excitation is the more common case — it represents shaker testing, mounting structure vibration and seismic excitation. Force PSD excitation represents applied random forces, such as acoustic pressure fluctuations. The analysis formulation differs between the two.

Modal Superposition

Random vibration FEA uses modal superposition — the response is computed as the combination of individual modal responses. Each mode is treated as an SDOF system excited by the PSD input filtered through the modal transfer function. The total response is the statistical combination of all modal responses. This is efficient and provides insight into which modes dominate the response.

Modal Participation and Truncation

The adequacy of the modal basis is critical. If significant modes are omitted — either because the extraction frequency range is too narrow or because insufficient modes were extracted — the response will be underestimated. The cumulative effective mass in the excitation direction should be checked. If it falls short of the total mass, more modes or residual mass correction may be needed.

DYNAMIC CHECK: Confirm that the extracted modal mass adequately represents the directions and frequency range relevant to the excitation. Missing modal mass means missing response.

Residual Mass and Missing-Mass Correction

When the modal basis is truncated — modes above a cut-off frequency are omitted — the response from those modes is lost. For low-frequency response, this is usually acceptable because high-frequency modes contribute little. However, for high-frequency input content or for response quantities sensitive to high-frequency modes (stress at stress concentrations, for example), the missing contribution can be significant. Some solvers offer residual mass or missing-mass correction to approximate the contribution of omitted modes.

Modal Combination Methods

The individual modal responses must be combined to produce the total response. Because the modes respond simultaneously to a random input, the combination is statistical — not a simple algebraic sum. The most common methods are SRSS (square root of sum of squares) and CQC (complete quadratic combination). CQC accounts for modal coupling and is preferred for closely spaced modes.

Outputs

Random vibration FEA produces several types of output. Each has a different engineering purpose and must be interpreted appropriately.

• Acceleration RMS — for equipment qualification and comparing against test levels
• Displacement RMS — for clearance and relative motion assessment
• Stress RMS — for strength screening and fatigue input (not a deterministic stress)
• Interface load RMS — for mount sizing and fastener load assessment
• Response PSD — for understanding frequency content of the response and fatigue analysis

RMS Stress Is Not Automatically a Fatigue Life

A common error is to compare RMS stress directly against a fatigue allowable. RMS stress is a statistical quantity — the standard deviation of the stress process. The actual peak stresses that drive fatigue are higher than the RMS. Vibration fatigue requires additional processing of the stress PSD — cycle counting, S-N data and damage accumulation — before a fatigue life can be estimated. RMS stress is a useful screening quantity but not a fatigue life predictor.

STATISTICAL INTERPRETATION: An RMS stress is not equivalent to a deterministic peak stress. It is a standard deviation. Comparing it directly against a static or fatigue allowable without statistical interpretation is incorrect.

Frequency Range and Modal Truncation

The frequency range for the random vibration analysis should cover the full PSD input range. If the PSD extends to 2000 Hz, modes up to 2000 Hz must be extracted. In some cases, modes above the PSD range may be needed for adequate effective mass. The cut-off frequency and the number of modes should be determined by checking effective mass, not by arbitrary defaults.

Verification

Random vibration FEA results should be verified by checking that response PSD peaks align with natural frequencies, that RMS values are consistent with the input Grms and structural amplification, and that the response is insensitive to further modal refinement.

• Check response PSD peaks align with natural frequencies from modal analysis
• Verify cumulative effective mass is adequate in excitation directions
• Confirm response is insensitive to additional modes (convergence check)
• Check that RMS values are physically plausible relative to input Grms
• Verify damping assumptions are documented and appropriate

Key Takeaways

• Random vibration FEA combines modal analysis with PSD input to predict statistical response
• The method assumes linearity, stationary excitation, modal damping and Gaussian input
• Modal truncation adequacy is critical — check effective mass in excitation directions
• RMS stress is a statistical quantity, not a deterministic peak — do not compare directly against allowables
• The quality of the result depends on the modal model underneath it