Scribd
Upload
4 views6 pages

Power Spectral Density in Fatigue Analysis

The document discusses the importance of Power Spectral Density (PSD) in fatigue analysis, explaining how it is used to transform acceleration or load data into stress data for predicting fatigue life. It outlines the steps to calculate PSD from time series data, including sampling, applying Fourier Transform, and normalizing the results. Additionally, it emphasizes the significance of ensuring signal stationarity and applying windowing techniques to minimize spectral leakage.

Uploaded by

Sudhanshu Pandey
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
4 views6 pages

Power Spectral Density in Fatigue Analysis

The document discusses the importance of Power Spectral Density (PSD) in fatigue analysis, explaining how it is used to transform acceleration or load data into stress data for predicting fatigue life. It outlines the steps to calculate PSD from time series data, including sampling, applying Fourier Transform, and normalizing the results. Additionally, it emphasizes the significance of ensuring signal stationarity and applying windowing techniques to minimize spectral leakage.

Uploaded by

Sudhanshu Pandey
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

The Role of Power Spectral Density in

Fatigue Analysis
• For those just starting in fatigue analysis, we begin with
the PSD of accelerations or loads, which we transform
into the PSD of stresses using system dynamics and ma-
terial properties.
• From the stress PSD, we compute spectral moments.
These moments are key inputs for predicting fatigue life
by estimating stress cycles and their contributions to cu-
mulative damage.

LinkedIn � - Dr. Lonny Thompson, Clemson University, Jan. 20, 2025. 1 of 6


Let’s Define Power Spectral Density (PSD)
• A Power Spectral Density (PSD) describes how the power
of a signal is distributed over frequencies, expressed as
power per unit frequency (e.g., Acceleration Power per
unit frequency, G2 / Hz), where G is the acceleration of
gravity.
• PSD is invaluable when analyzing random or time-varying
signals, like vibrations, noise, or earthquake data.
• If you have a signal that fluctuates unpredictably, the
PSD helps you break it down into its frequency compo-
nents to identify which frequencies contribute most to
the signal’s behavior.
• Instead of looking at the raw signal in the time domain,
PSD provides a frequency-domain view.
• The GRMS (G Root Mean Square) value is a single scalar
metric used to represent the overall energy or intensity
of a vibration signal across a specified frequency range.
It is derived by integrating the PSD, which quantifies the
distribution of power per unit frequency over the fre-
quency range of interest and then takes the square root
of the result.

LinkedIn � - Dr. Lonny Thompson, Clemson University, Jan. 20, 2025. 2 of 6


Steps to Calculate a PSD from Time Series Data
1. Sample the Time Signal Data: Sample your signal as a
time series: a set of values recorded at regular intervals
(e.g., acceleration or displacement over time).
2. Apply the Fourier Transform: Apply a Fourier Trans-
form (usually a Fast Fourier Transform, FFT) to convert
the signal from the time domain to the frequency do-
main. This process decomposes the signal into its fre-
quency components.
3. Calculate the Power: Calculate the power at each fre-
quency by squaring the magnitude of the Fourier Trans-
form coefficients. Proper scaling based on the sampling
rate and time window is needed to ensure the power is
accurate.
4. Normalize with Frequency: Divide the power by the
frequency resolution to obtain a true PSD, reflecting power
per unit frequency. This step ensures consistency across
signals with different durations or sampling rates.

LinkedIn � - Dr. Lonny Thompson, Clemson University, Jan. 20, 2025. 3 of 6


5. Smooth the PSD (Optional):
• If your signal is noisy, the PSD can be improved by
splitting it into overlapping regions, calculating the
PSD for each region, and then averaging them. This
technique is called the Welch method and produces
a cleaner, less noisy PSD.
• It works by applying a window function such as a Ham-
ming or Hann window to each region to minimize
spectral leakage and averaging the results to reduce
random fluctuations.
• The Hamming window balances frequency resolution
and leakage suppression, while the Hann window pri-
oritizes leakage suppression at the cost of slightly broader
main lobes.
• The Welch method improves on the basic periodogram
and is often the preferred approach for PSD estima-
tion.
• While it smooths the PSD, segmenting the signal may
slightly reduce frequency resolution.
• This method is usually recommended for obtaining a
more reliable PSD, especially for random or noisy sig-
nals. However, for transient or short-duration signals,
raw PSD estimation may be more appropriate.

LinkedIn � - Dr. Lonny Thompson, Clemson University, Jan. 20, 2025. 4 of 6


Key Ideas to Keep in Mind
• Ensure the signal is stationary, as PSD assumes the sta-
tistical properties of the signal do not change over time.
• Apply windowing to minimize spectral leakage, which
can distort frequency measurements.

LinkedIn � - Dr. Lonny Thompson, Clemson University, Jan. 20, 2025. 5 of 6


Graphs to Guide You:
• The first graph shows a synthetic signal (e.g., acceleration) as
a function of time. To illustrate the process, the acceleration
signal is a combination of sine waves with different frequen-
cies and amplitudes representing a noisy time-domain signal.
• The second graph, the PSD, shows how the power of the sig-
nal is distributed across frequencies, with power (G2/ Hz) plot-
ted against frequency (Hz). Peaks are observed at the domi-
nant frequencies (2 Hz, 5 Hz, 8 Hz, and 11 Hz), reflecting the
signals’ primary components.
Raw Signal (Acceleration vs. Time)
1.5

1
Acceleration (G)

0.5

-0.5

-1

-1.5
0 1 2 3 4 5 6 7 8
Time (s)
0
Power Spectral Density (PSD)
10

-2
10
Power (G2/Hz)

-4
10

-6
10

10-8
0 5 10 15 20 25 30 35 40 45 50
Frequency (Hz)

LinkedIn � - Dr. Lonny Thompson, Clemson University, Jan. 20, 2025. 6 of 6

You might also like