FFT Analysis

This example provides a detailed guide to using the FFT analysis capabilities of Remote OpenFAST Plotter.

FFT Analysis Basics

Fast Fourier Transform (FFT) analysis converts time domain signals into the frequency domain, allowing you to:

  • Identify dominant frequencies in your data

  • Detect resonances in the structure

  • Observe harmonic relationships

  • Compare frequency content across different simulations

Configuring FFT Parameters

The FFT tab provides several configuration options that affect your analysis:

  1. Averaging Method:

    • None: Simple FFT without averaging (higher variance)

    • Welch: Welch’s method - divides the signal into overlapping segments, computes FFT for each, then averages (reduced variance, better statistical stability)

    • Bartlett: Similar to Welch but uses non-overlapping segments

  2. Window Function:

    • Hanning: General-purpose window with good frequency resolution

    • Hamming: Modified Hanning window with different coefficients

    • None: No windowing (may introduce spectral leakage)

  3. Segment Size:

    • Controls the frequency resolution

    • Larger segments give finer frequency resolution but less averaging

    • Expressed as 2^N samples per segment

  4. Overlap:

    • Percentage of overlap between segments (used with Welch’s method)

    • Higher overlap increases the amount of averaging

    • Typical values range from 50% to 75%

  5. Plot Appearance:

    • X-scale: Linear or logarithmic frequency axis

    • Y-scale: Linear or logarithmic amplitude axis

    • Plot style: Overlay or separate plots for multiple signals

FFT Analysis Step-by-Step

Here’s a detailed walkthrough for performing FFT analysis:

  1. Load Files:

    • Start by loading one or more OpenFAST output files containing the signals you want to analyze

  2. Access FFT Tab:

    • Click on the FFT tab in the main navigation

  3. Select Signals:

    • Choose the signals you want to analyze

    • For wind turbine analysis, common signals include: * Blade root moments (e.g., “RootMycX1”) * Tower base moments (e.g., “TwrBsMxt”) * Generator or shaft torque

  4. Configure FFT Parameters:

    • For most wind turbine analysis, recommended settings are: * Averaging Method: Welch * Window Function: Hanning * Segment Size: 2^11 to 2^13 (adjust based on signal length) * Overlap: 50% to 75%

  5. Set Axis Scales:

    • Select appropriate X and Y scales: * Logarithmic X-scale is useful for viewing a wide frequency range * Logarithmic Y-scale helps visualize peaks of different magnitudes

  6. Calculate FFT:

    • Click the “Calculate FFT” button to generate the frequency domain plot

  7. Analyze Results:

    • Identify dominant peaks in the frequency spectrum

    • Look for expected frequencies (e.g., 1P, 3P, tower frequencies)

  8. Add Annotations:

    • Mark important frequencies using the annotation system

    • Common annotations for wind turbines: * 1P: Once-per-revolution frequency * 3P: Three times per revolution (for three-bladed turbines) * Tower natural frequencies * Blade mode frequencies

  9. Save or Export Results:

    • Export the annotated FFT plot as HTML for documentation

    • Save annotation sets for future use

Example: Comparing Natural Frequencies

Here’s a specific example for identifying and comparing natural frequencies:

  1. Load baseline model and modified model files:

    test_files/5MW_Land_DLL_WTurb.outb
test_files/5MW_Land_BD_DLL_WTurb.outb

  2. Configure optimal FFT parameters:

    • Averaging: Welch

    • Window: Hanning

    • Segment Size: 2^12

    • Overlap: 50%

    • X-scale: Logarithmic (to emphasize lower frequencies)

  3. Select tower base moments:

    • Choose “TwrBsMxt” from both files

    • Click “Calculate FFT”

  4. Add annotations for known frequencies:

    • Add 0.32 Hz with label “Tower FA” (fore-aft mode)

    • Add 0.31 Hz with label “Tower SS” (side-side mode)

    • Add 0.2 Hz with label “1P” (assuming 12 RPM rotor speed)

    • Add 0.6 Hz with label “3P”

  5. Analyze differences:

    • Compare peak locations between baseline and modified model

    • Look for frequency shifts indicating structural changes

    • Identify if any peaks align with forcing frequencies (1P, 3P)

  6. Export the comparison:

    • Click “Export FFT as HTML” to create a shareable document

    • Include annotations in the export

Advanced FFT Techniques

For more advanced analysis:

  1. Segment Size Optimization:

    • Smaller segments (e.g., 2^8): Better for statistical stability, less frequency resolution

    • Larger segments (e.g., 2^14): Higher frequency resolution, less averaging

    • Choose based on your signal length and analysis goals

  2. Window Function Selection:

    • Hanning: General purpose, good compromise between leakage and resolution

    • Hamming: Slightly different mainlobe/sidelobe characteristics

    • No window: Maximum frequency resolution but may have spectral leakage

  3. Frequency Range Focus:

    • Use X-axis limits to focus on specific frequency ranges

    • For wind turbines, often 0-2 Hz contains most relevant dynamics

  4. Multiple Signal Comparison:

    • Compare the same signal across different files to identify changes

    • Compare different signals from the same file to identify relationships

Troubleshooting

Common issues with FFT analysis:

  • Low Resolution: Increase segment size for better frequency resolution

  • Noisy Spectrum: Use Welch’s method with more averaging (smaller segments)

  • Missing Peaks: Ensure your simulation time is long enough to capture low frequencies

  • Unexpected Harmonics: Check for physical phenomena or numerical issues in the simulation