Langford Analytic · Knowledge Base

How to Model Multiple Random Inputs

A procedure for setting up multi-input random vibration analysis — determining correlation, selecting the analysis approach and interpreting results.

Random Vibration6 min read
how-tomultiple inputcorrelateduncorrelatedcross-spectralmulti-support

When to use this guide

Use this guide when the structure is excited at multiple points and you need to determine whether to treat the inputs as correlated or independent.

Step-by-step method

  1. Identify all input points and their PSDs
  2. Determine the physical relationship between inputs — same source? Same structure? Independent sources?
  3. If measured data is available: compute the coherence between each pair of inputs
  4. If coherence ≈ 0 at all frequencies: treat as uncorrelated — use SRSS combination
  5. If coherence ≈ 1 at critical frequencies: treat as correlated — use full cross-spectral analysis
  6. If partial correlation: use full cross-spectral analysis with the measured cross-spectra
  7. If no measured data: assess based on physical configuration — multi-support on same structure is usually correlated at low frequency
  8. For multi-support excitation: consider the support spacing relative to the wavelength at critical frequencies
  9. Document the correlation assessment and the analysis approach chosen

Related Knowledge