Perform General Signal Processing for Computational Aeroacoustics

Process a surface output or point probe file created from data as a part of the General Signal Processing (GSP) workflow that is utilized for Computational Aeroacoustics (CAA).

Examples of this workflow include the post-processing of a rotating fan and greenhouse noise simulations. You may process time-dependent point or surface data to interpret the noise source from aerodynamic and aeroacoustic simulations.

Hydrodynamic dB Map Creation

The purpose of the GSP workflow is to generate hydrodynamic surface dB maps from a time-dependent surface pressure, written in HyperView (.h3d) format.

The source .h3d file must be created directly from the solver or converted from merged ENSIGHT data (.case files) via the HVTrans application. The source file should contain the time-dependent pressure for one or more surfaces at a high enough frequency and long enough duration to support the requested frequency range of interest.

Using the source time-dependent pressure map, you can obtain the Sound Pressure Level (SPL) in decibels (dB) or weighted levels to represent human hearing more closely in dB(A) or dB(C). The SPL can be obtained for a single range of frequencies [fmin, fmax] represented at the center of the broadband, or optionally, weighted for each fractional octave band across the requested frequency spectrum. The SPL is obtained by computing the Fast Fourier Transform (FFT) of the time dependent pressure signal, modifying the resultant power spectral density and logarithmically rescaling, accounting for the reference pressure of air for each element present on the surface of interest. The value for SPL is computing by using the relationship shown in Equation 1. Figure 1 provides an example of the data that is utilized and created during the GSP workflow.

S P L i = 10.0 log 10 ( p 2 ( f i ) p r e f 2 ) MathType@MTEF@5@5@+= feaagKart1ev2aaatCvAUfeBSjuyZL2yd9gzLbvyNv2CaerbuLwBLn hiov2DGi1BTfMBaeXatLxBI9gBaerbd9wDYLwzYbItLDharqqtubsr 4rNCHbGeaGqiVu0Je9sqqrpepC0xbbL8F4rqqrFfpeea0xe9Lq=Jc9 vqaqpepm0xbba9pwe9Q8fs0=yqaqpepae9pg0FirpepeKkFr0xfr=x fr=xb9adbaqaaeGaciGaaiaabeqaamaabaabaaGcbaGaam4uaiaadc facaWGmbWaaSbaaSqaaiaadMgaaeqaaOGaeyypa0JaaGymaiaaicda caGGUaGaaGimaiGacYgacaGGVbGaai4zamaaBaaaleaacaaIXaGaaG imaaqabaGcdaqadaqaamaalaaabaGaamiCamaaCaaaleqabaGaaGOm aaaakiaacIcacaWGMbWaaSbaaSqaaiaadMgaaeqaaOGaaiykaaqaai aadchadaqhaaWcbaWaaWbaaWqabeaacaWGYbGaamyzaiaadAgaaaaa leaacaaIYaaaaaaaaOGaayjkaiaawMcaaaaa@4DD0@

The above equation represents the sound pressure level of ith element on a given surface, using Pref=2.0e-5Pa. Where SPL has units of decibels (dB).

Figure 1. Diagram showing the process for creating and post-processing boundary surface data


  1. From the Post ribbon, click the Signal Processing > Surface tool.
    Figure 2.


  2. Define the surface-based signal processing attributes in the dialog.
    Figure 3. Input dialog for GSP (Hydrodynamic dB Surface Map focus)


  3. Load the data file of interest into the tool using the file browser icon, .
    Fields are populated with the data that is contained in the input source (.h3d).
  4. The following attributes are populated once the data is initially loaded. Assign each of the attributes according to best-practice guidelines.
    File name
    Source data as provided from the solver (.h3d file) or the pre-processed compact form of the pressure vs. time signal (.gsph5 file).
    Surface
    Surface components that are sent to the GSP algorithm. Multiple components contained within the input source may be selected by entering the advanced selection context.
    Time duration
    Beginning and ending time values in seconds. The fields may be modified based to control the number of records considered during the calculation. Auto-calculate can be utilized to redefine the original time-series bounds.
    Band type
    Output band summation option. Single band provides a single dB map comprised of SPL from the full frequency range. Octave (fractional octave) provides a frequency dependent dB map for each of the Nth octave bands, with N=1, 3, 12 that fall within the frequency range of interest. The octave band center frequency is reported within the output file.
    Weight
    Decibel weighting for dB (default), dBA or dBC. All data in the dB map will be weighted according to your selection of weight. Provides dBA weight for interpreting dB maps associated with the range of human hearing.
    Note: dBA ~= dB at a frequency of 1kHz.
    Frequency range
    Minimum and maximum cutoff frequency to be displayed in the output results. Auto-calculate can be utilized to redefine the original supported cutoff frequency bounds.
    Note: The minimum value is set to resolve the minimum frequency that is supported by ~4 cycles of the transient period, the maximum is set to the Nyquist Frequency.
  5. Define any additional options that are needed to compute the hydrodynamic surface dB map and band filtered animation through the two hamburger menu icons in the main menu dialog. Select the FFT parameters through the first menu, as shown in Figure 4.
    Figure 4. FFT parameters for the GSP calculation


    Type
    Modifies the subsequent selection of the FFT bandwidth for windowing.

    When Type=Block size, the Block size enumerated list is accessible, defaulting to 4096 records. The associated Bandwidth for that Block size is automatically calculated and presented to you.

    When Type=Bandwidth size, the Bandwidth real number is accessible, defaulting to the bandwidth associated with a Block size of 4096 records. When the Bandwidth is modified, the associated Block size is automatically calculated and presented to you.

    Window Type
    Windowing algorithm used to reduce spectral leakage that may be present within the input pressure signal. Hanning, Hamming and Blackman are available.
    Window Overlap (%)
    Define the percentage overlap for each window used during the FFT of the input pressure signal.
  6. Review the data contained within the input source through the second hamburger menu.
    The menu shows the Time range (seconds), Record length (samples), Data Frequency (Hz) and Time step (seconds).
    Figure 5. Source data parameters, non-modifiable


  7. Once the parameters have been defined, click Calculate.

    This performs the calculation in a sub-process, allowing you to return to the UI and prepare any additional post-processing content. During the calculation, a status indicator is updated with the progress of the sub-process. To query the status of the calculation, press the icon.

    Once the calculation is complete, you are presented with the option to load the data into the post-processing session, as shown in Figure 6. Multiple datasets may be loaded into the session at once if required.

    Note: Once the calculation has been completed, the definition for the GSP is stored in session, accessible in the Post Browser for future inspection.
    Figure 6. GSP progress bar and results complete, load data into the post-processing session


Band Filtered Transient Animation Summary

Provides a workflow to generate a Band Filtered Transient Surface Animation (BAFA) from a time-dependent surface pressure source file, written in HyperView (.h3d) format.

Utilizing the same source input file as described in the hydrodynamic dB map summary, you can generate a BAFA according to the frequency range of interest. As in the hydrodynamic map process, the source file should contain the time-dependent pressure for one or more surfaces at a high enough frequency and long enough duration to support the requested frequency range of interest. If previously computed, or with the Surface Output Type option: Both, the BAFA will utilize the FFT and filter the time domain data for output.

Figure 7 shows dialog that assigns the input attributes needed to process a BAFA map.
Note: The input file name is populated with a .gsph5 instead of the original .h3d file. This allows for a significantly reduced time to compute the filtered pressure compared to reading the original source data.
Figure 7. Input dialog for GSP (BAFA focus)


Point Probe Summary

Provides a workflow to compute noise levels from point probe measurement files using .csv files created directly from ultraFluidX.

The point probe, or computational measurement files, contains individual point data for the time history of pressure in Pascals (Pa). The measurement file is provided by the solver during the simulation and can be utilized to calculate the noise levels on the surface or in the volume near the region of interest. The probe positions are prepared prior to running the solver and are created during the output data stream while the simulation progresses. Computing the noise levels in frequency space is necessary to evaluate the signal validity, period of record and to approximate the sound pressure level at individual locations after post-processing the raw data. Figure 8 provides a workflow diagram outlining the procedure to compute the SPL or PSD for a given probe.
Figure 8. Diagram showing the process for creating and post-processing Point Probe data


  1. From the Post ribbon, click the Signal Processing > Probe Point tool.
    Figure 9.


  2. Define the point probe-based signal processing attributes in the dialog.
    Figure 10. Input dialog for GSP (Point Probe focus)


  3. Load the data file of interest into the tool using the file browser icon, .
    Fields are populated with the data that is contained in the input source (.csv).
  4. The following attributes are populated once the data is initially loaded (Figure 11). Assign each of the attributes according to best-practice guidelines.
    File name
    Source data as provided from the solver (.csv file).
    Probe selection

    All, individual, or a subset of probes (read from the header names) for 1D processing. The advanced selection context may be used to chose specific probes.

    Time duration
    Beginning and ending time values in seconds. The fields may be modified based to control the number of records considered during the calculation. Auto-calculate can be utilized to redefine the original time-series bounds.
    Plot type
    Define output plot type. Sound Pressure Level (SPL) or Power Spectral Density (PSD).
    Band type
    Output band summation option. Narrowband provides the full spectrum of requested frequency range output at the data source sampling frequency. Octave (fractional octave) provides a frequency dependent dB map for each of the Nth octave bands, with N=1, 3, 8, 12 that fall within the frequency range of interest. The octave band center frequency is reported for the computed bands.
    Weight
    Decibel weighting for dB (default), dBA or dBC. All data in the dB map will be weighted according to your selection of weight. Provides dBA weight for interpreting dB maps associated with the range of human hearing.
    Note: dBA ~= dB at a frequency of 1kHz.
    Frequency range
    Minimum and maximum cutoff frequency to be displayed in the output results. Auto-calculate can be utilized to redefine the original supported cutoff frequency bounds.
    Note: The minimum value is set to resolve the minimum frequency that is supported by ~4 cycles of the transient period, the maximum is set to the Nyquist Frequency.
    Figure 11. Populated input dialog for GSP (Point Probe focus)


Batch Automation of GSP Surface Workflow

The GSP workflow can be utilized by running an automation script that generates the hydrodynamic dB map or BAFA without the user interacting with the GUI.

The batch automation can be accessed within the same distribution as the GUI (HWDesktop – HMCFD) from the command line. For example:

"<install_directory> \hwdesktop\hwx\plugins\hwd\profiles\HyperworksCfd\hwcfdGsp.bat" gspParams.json

Where the .json file contains the attributes that were provided in the GUI workflow, referenced above, and stored as follows (by default, in the application launch directory).

{"gspArgs": {
        “resultFilePath": "01_windshield_pass.gsph5",
        "outGspFname":"01_windshield_pass_octave.gsph5",
        "partName": "Glasses_windshield_pass", 
        "resultsName": "Glasses_windshield_pass", 
        "useCellData": true,
        "dataArrayName": "pressure", 
        "timeRange": [0.29910001, 1.1], 
        "bandType": "octave", 
        "scale": "dB", 
        "frequencyRange": [20.0, 11251.486], 
        "surfaceMapType": 2, 
        "exportData": 2, 
        "exportStride": 100, 
        "exportDataFilename": "windshield_pass_pressure.mp4", 
        "exportResultType": 1}
"fftArgs": {
        "nperseg": 4096, 
        "noverlap": 2048, 
        "window": "hann"}}