31.11.4. Using the FFT Utility

The ANSYS Fluent FFT utility is available through the Fourier Transform Dialog Box (Figure 31.89: The Fourier Transform Dialog Box).

 Results →   Plots →   FFT → Set Up...

Figure 31.89:  The Fourier Transform Dialog Box

The Fourier Transform Dialog Box

31.11.4.1. Loading Data for Spectral Analysis

FFT analysis requires an input signal data file consisting of time-sequence data. To load an input signal data file into the Fourier Transform Dialog Box, click Load Input File.... This displays The Select File Dialog Box where you can browse through your file directories and locate your data file containing your time-sequence data. To remove a file from the Files list, select it and then click Free File Data.

If you computed acoustic signals “on the fly”, you have the option of processing signal data from a file or processing receiver data stored in memory. To analyze signal data from an existing input file, select Process File Data under Process Options and proceed as described above. To analyze receiver data stored in memory, select Process Receiver under Process Options and select the appropriate receiver in the Receiver list.

Click Plot FFT to display the spectral analysis data and, if you have enabled Acoustics Analysis, to calculate the overall sound pressure level in dB based on the Reference Acoustic Pressure and display it in the console.


Note:  You can specify the axis and axis variables if your input report file contains multiple report definitions.


31.11.4.2. Customizing the Input and Defining the Spectrum Smoothing

With the input signal data file loaded into the Fourier Transform Dialog Box, you may want to view a plot of the input signal and/or customize the data set in preparation for applying the FFT algorithm. To do so, click the Plot/Modify Input Signal button to open the Plot/Modify Input Signal Dialog Box (Figure 31.90: The Plot/Modify Input Signal Dialog Box).

Figure 31.90:  The Plot/Modify Input Signal Dialog Box

The Plot/Modify Input Signal Dialog Box

The Plot/Modify Input Signal Dialog Box allows you to analyze a portion of the input signal, view input Signal Statistics—Min, Max, Mean, Variance, total Number of Samples, and Min Frequency (that is, the finest possible frequency resolution)—and set title and label information for the input signal plot. Additionally, this dialog box allows you to enable and control the smoothing of the resulting spectrum by subdividing the input signal into multiple segments and averaging the segment-based spectra.


Note:  You can specify the axis and axis variables if your input report file contains multiple report definitions.


31.11.4.2.1. Customizing the Input Signal Data Set

By default, the entire data set is analyzed. To analyze a portion of the input signal, enable the Clip to Range option and specify the data range by entering Min and Max values under X Axis Range. To have the -axis quantities reduced by the Mean value of the relevant signal property, enable the Subtract Mean Value option.

The Set Defaults button will reset the original values for the Min and Max fields under X Axis Range and turn off the Clip to Range option.

31.11.4.2.2. Spectrum Smoothing Through Signal Segmentation

For long broadband noise signals, the computed Fourier spectrum may display spurious fluctuations of amplitudes between the neighboring modes. In order to obtain a smooth spectrum, you can split the signal into multiple overlapping segments so that ANSYS Fluent can apply the FFT algorithm on each segment and then average the resulting spectra. This procedure allows you to significantly suppress the spurious fluctuations at the cost of coarsening the frequency resolution of the spectrum. Note that the arithmetic averaging of spectra is performed for the squares of Fourier amplitudes, thus conserving the signal energy.

To use the smoothing procedure, enable the Subdivide into Segments option and define the segment size and the overlap of the subsequent segments in the Segment Control group box. Make a selection from the Control Method list to specify how you want to define the segment size: select Samples if you want to specify the number of Samples per Segment; select Frequency if you want to specify the desired Frequency Resolution in Hertz units (the segment length in seconds is then equal to ). To help determine a suitable segment size, you can use the Number of Samples and the Min Frequency information in the Signal Statistics group box. Note that the final segment length is selected by ANSYS Fluent to adhere to the largest supported number of data samples (see Table 31.2: Numbers of Data Points Supported by the Prime-Factor FFT Algorithm). The actual signal length and the number of segments will be displayed in the console when you plot or write the Fourier spectrum. The number of segments depends on the Overlap of the subsequent segments, which is specified regardless of which Control Method you select. If you select a Window function (see Windowing for details), it is recommended that you use an Overlap value of at least 0.125, in order to cover the signal portion affected by the window function. Generally, an Overlap value of 0.5 is recommended. The exact number of samples that are in the overlapping region of the segment can be determined from the segment hop size, which is displayed in the console. The segment hop size is the distance (in terms of samples) between the starting samples of any two adjacent segments (that is, the shift from one segment to the next), and has a minimum value of 1.

The Set Defaults button will reset the segment control values and disable the Subdivide into Segments option.

31.11.4.2.3. Viewing Data Statistics

To aid in the signal analysis, whether for the entire input signal or for a certain range of data, the Signal Statistics group box in the Plot/Modify Input Signal Dialog Box displays signal information such as the minimum, maximum, and average signal values, as well as the signal variance, total number of samples, and finest possible frequency resolution.

31.11.4.2.4. Customizing Titles and Labels

You can create a new title or edit the original title for the input signal plot by entering a text string in Signal Plot Title. Likewise, you can create a new axis label or edit the original axis label by entering a text string into either Y Axis Label or X Axis Label.

31.11.4.2.5. Applying the Changes in the Input Signal Data

To apply any changes you have made in the Plot/Modify Input Signal Dialog Box and view a plot of the input signal, click Apply/Plot.

31.11.4.3. Customizing the Output

In most practical applications with CFD data, you may want to find out how much power or energy is contained in a certain frequency range, but do not want to distinguish positive and negative frequency. In recognition of this, all the outputs from the FFT module in ANSYS Fluent pertain to one-sided spectra for the range of positive frequency.

The Fourier Transform Dialog Box (Figure 31.89: The Fourier Transform Dialog Box) and Plot/Modify Input Signal Dialog Box (Figure 31.90: The Plot/Modify Input Signal Dialog Box) enable you to set several different functions for the and axes, apply different FFT windowing techniques, and set various output options.

31.11.4.3.1. Specifying a Function for the Axis

You can choose the -axis function using the Y Axis Function drop-down list. Available options for the -axis functions are as follows. Note that the functions related to acoustics (all of which are measured in dB) are only available when the Acoustics Analysis option is enabled.

The definitions are provided for , and include contributions from both elements of a complex conjugate pair and -. Note that in ANSYS Fluent, the value of is always an even number.

Power Spectral Density

is the distribution of signal power in the frequency domain. Its value and units depend on the X Axis Function choice. For the detailed spectral representation with all resolved harmonics (that is, when X Axis Function is either Frequency, Strouhal Number, or Fourier Mode), the Power Spectral Density () has units of the signal magnitude squared over the frequency (for example, Pa2/Hz) and is defined for the frequency as

(31–36)

where is the frequency step in the discrete spectrum, and the Fourier mode power is computed as

(31–37)

For the octave analysis (that is, when the X Axis Function is either Octave Band or 1/3-Octave Band), the Power Spectral Density has units of the signal magnitude squared (for example, Pa2), and is defined for the frequency band as

(31–38)

where includes all of the Fourier modes belonging to the band.

Magnitude

is the amplitude. For the detailed spectral representation with all resolved harmonics (that is, when X Axis Function is either Frequency, Strouhal Number, or Fourier Mode), the Magnitude () is defined for the frequency as

(31–39)

where is the mean signal value.

For the octave analysis (that is, when the X Axis Function is either Octave Band or 1/3-Octave Band), the Magnitude is defined for the frequency band as

(31–40)

where is calculated according to Equation 31–38.

Sound Pressure Level (dB)

is the decibel level. For either general or acoustic data, when the sampled data is pressure (for example, static pressure or sound pressure), the Sound Pressure Level (dB) () is calculated in decibel units using

(31–41)

where is the Power Spectral Density for either a particular Fourier mode or a particular frequency band (see Equation 31–36 and Equation 31–38). is the reference acoustic pressure, with a default value of 2 10-5 Pa; you can revise this value, as described in Enabling the FW-H Acoustics Model.

Sound Amplitude (dB)

is similar to the Sound Pressure Level (dB), and is a logarithmic conversion of the pressure signal Magnitude into decibel units. The Sound Amplitude (dB) () is calculated for either a Fourier mode or a frequency band using

(31–42)

A-Weighted, Sound Pressure Level (dB A)

is the calculated sound pressure level weighted by the A-scale function to more closely approximate the frequency response of the human ear. A-Weighting is applied for loudness levels below 55 phons (55 dB at 1 kHz) and is the most commonly used weighting function. See Figure 31.91: A-, B-, and C-Weighting Functions for a graphical representation. This function is only available when the X Axis Function is either Octave Band or 1/3-Octave Band.

B-Weighted, Sound Pressure Level (dB B)

is the calculated sound pressure level weighted by the B-scale function. B-Weighting is applied to loudness levels between 55 and 85 phons, though it is rarely used. See Figure 31.91: A-, B-, and C-Weighting Functions for a graphical representation. This function is only available when the X Axis Function is either Octave Band or 1/3-Octave Band.

C-Weighted, Sound Pressure Level (dB C)

is the calculated sound pressure level weighted by the C-scale function. C-Weighting is applied for loudness levels above 85 phons and is commonly used for high-intensity sound such as traffic studies. See Figure 31.91: A-, B-, and C-Weighting Functions for a graphical representation. This function is only available when the X Axis Function is either Octave Band or 1/3-Octave Band.

Figure 31.91:  A-, B-, and C-Weighting Functions

A-, B-, and C-Weighting Functions

Further graphical customizations for the -axis are available by clicking the Axes... button. For more information, see Modifying Axis Attributes.

31.11.4.3.2. Specifying a Function for the X Axis

There are three options for the -axis function you can choose from in order to plot or write the detailed spectrum with all resolved Fourier modes; these three options are related to the discrete frequencies at which the Fourier coefficients are computed. There are also two additional options available when the Acoustics Analysis option is enabled, which allow the octave and 1/3 octave band analysis. You can apply specific analytic functions for the -axis using the X Axis Function drop-down list.

Available options for the -axis functions are as follows. The definitions are provided for , because the corresponding definitions for the -axis functions include contributions from both elements of a complex conjugate pair and -. Note that in ANSYS Fluent, the value of is always an even number.

Frequency (Hz)

is defined as:

(31–43)

where is the number of data points used in the FFT.

Strouhal Number

is the nondimensionalized version of the frequency defined in Equation 31–43 :

(31–44)

where and are the reference length and velocity scales specified in the Reference Values Task Page.

Fourier Mode

is the index in Equation 31–34 and/or Equation 31–35, which represents the th term in the Fourier transform of the signal.

Octave Band (Hz)

is a range of discrete frequency bands for different octaves within the threshold of hearing. The range of each octave band is double to that of the previous band (see Table 31.3: Octave Band Frequencies and Weightings).

1/3-Octave Band (Hz)

is a range of discrete frequency bands for different 1/3 octaves within the threshold of hearing.

Table 31.3:  Octave Band Frequencies and Weightings

Lower Freq. (Hz)

Center Freq. (Hz)

Upper Freq. (Hz)

dB A

dB B

dB C

11

16

22

-56.7

-28.5

-8.5

22

31.5

45

-39.4

-17.1

-3.0

45

63

90

-26.2

-9.3

-0.8

90

125

180

-16.1

-4.2

-0.2

180

250

355

-8.6

-1.3

0.0

355

500

710

-3.2

-0.3

0.0

710

1000

1400

0.0

0.0

0.0

1400

2000

2800

1.2

-0.1

-0.2

2800

4000

5600

1.0

-0.7

-0.8

5600

8000

11200

-1.1

-2.9

-3.0

11200

16000

22400

-6.6

-8.4

-8.5


Further graphical customizations for the -axis are available by clicking Axes.... For more information, see Modifying Axis Attributes.

31.11.4.3.3. Specifying Output Options

You can write out the FFT data directly to a file by choosing the Write FFT to File option under Options in the Fourier Transform Dialog Box. Once the Write FFT to File option is selected, click Write FFT to display a file selection dialog box where you can choose a file and/or a location to hold the FFT data. If Acoustics Analysis is selected, the overall sound pressure level will be calculated in dB (based on the Reference Acoustic Pressure) and displayed in the console at this time.

Further customizations for how the FFT data is displayed are available by clicking Curves.... For more information, see Modifying Curve Attributes.

31.11.4.3.4. Specifying a Windowing Technique

You can use the various windowing techniques described in Windowing by selecting any of the Window options in the Plot/Modify Input Signal Dialog Box. By default, None is selected so that no windowing technique is applied.

31.11.4.3.5. Specifying Labels and Titles

You can assign a title for your FFT plot using the Plot Title text field. You can also assign -axis and -axis labels for your FFT plot using the Y Axis Label and X Axis Label text fields, respectively. By default,ANSYS Fluent assigns the Y Axis Label and the X Axis Label to the particular selection of Y Axis Function and X Axis Function.


Release 18.2 - © ANSYS, Inc. All rights reserved.