Filtering field and history data

You can create filters in Abaqus/CAE and apply them to field output requests or history output requests for Abaqus/Explicit analyses. Abaqus filters data while the analysis is running; filtering during the analysis (real time filtering) can reduce the size of the output database by excluding high-frequency data before it is saved. Real time filtering also avoids potential aliasing problems in the resulting data. Aliasing is a loss of valid results data; aliasing will occur if the sampling frequency (the frequency at which the data are saved) is less than twice the highest frequency expected in the results. For example, if a sine wave was sampled at only two points, the aliased result would appear to be a straight line, whereas sampling at least four points would reproduce the shape of the curve.

See Also
In Other Guides
Filtering Output and Operating on Output in Abaqus/Explicit

The Filter toolset allows you to create the following filters for use with Abaqus/Explicit field output requests and history output requests:

  • Butterworth

  • Type I Chebyshev

  • Type II Chebyshev

For more specific information about these filters, see Filtering Output and Operating on Output in Abaqus/Explicit. The different filter types are distinguished by their capabilities to transition from the acceptance of low-frequency data to the rejection of data above the filter's Cutoff frequency. An ideal filter would stop all data above the cutoff frequency and have a flat response (no affect on accepted data); real filters include a transition band around the cutoff where some data are accepted and they usually have some effect on the accepted data. The Butterworth filter provides a maximally flat response magnitude with a wider transition band (slower transition) than the Chebyshev filters. The Chebyshev filters introduce an oscillation—a ripple—in the response magnitude, but they have a narrower transition band than the Butterworth filter. The two types of Chebyshev filters differ in where in their response ripples occur; the Ripple factor indicates how much oscillation you will allow in exchange for an improved filter response. You can also specify the Order of the filter, which determines the size of the filter's transition band: the higher the order, the narrower the transition band, although the computational cost increases as the order increases. The filter order must be a positive, even integer no greater than 20.

In addition to the filters that you can define using the Filter toolset, Abaqus also includes a default Antialiasing filter. Abaqus automatically sets the cutoff frequency of the Antialiasing filter based on the time interval at which the field output or history output is saved during the analysis. When you define a filter, you must specify the cutoff frequency based on your knowledge of the frequencies expected in the solution. Whether you set the cutoff or Abaqus calculates it, no checks are performed to ensure that the cutoff frequency is appropriate. If the cutoff is set too low, valid data will be filtered from the results; if it is set too high (above half the sampling frequency), no filtering will be performed.

Select ToolsFilterCreate from the main menu to create a new filter definition; select Edit from the same menu to make changes to an existing definition. Either command opens the filter editor, which allows you to select the options and provide the data needed to define your filter.

To apply a filter to an analysis, include it in a field output request or history output request for an Abaqus/Explicit analysis procedure (for more information, see Defining output requests).