ENH: Updated noise utility to use new noise models

This commit is contained in:
andy 2016-02-25 12:56:25 +00:00
parent 7c66e69136
commit c520c3b62b
2 changed files with 116 additions and 111 deletions

View File

@ -1,6 +1,7 @@
EXE_INC = \
-I$(LIB_SRC)/randomProcesses/lnInclude \
-I$(LIB_SRC)/sampling/lnInclude
-I$(LIB_SRC)/sampling/lnInclude \
-I$(LIB_SRC)/surfMesh/lnInclude
EXE_LIBS = \
-lrandomProcesses \

View File

@ -3,7 +3,7 @@
\\ / F ield | OpenFOAM: The Open Source CFD Toolbox
\\ / O peration |
\\ / A nd | Copyright (C) 2011-2015 OpenFOAM Foundation
\\/ M anipulation |
\\/ M anipulation | Copyright (C) 2016 OpenCFD Ltd.
-------------------------------------------------------------------------------
License
This file is part of OpenFOAM.
@ -25,157 +25,161 @@ Application
noise
Description
Utility to perform noise analysis of pressure data using the noiseFFT
library.
Utility to perform noise analysis of pressure data.
Control settings are read from the $FOAM_CASE/system/noiseDict dictionary,
or user-specified dictionary using the -dict option. Pressure data is
read using a CSV reader:
\heading Usage
\verbatim
pRef 101325;
N 65536;
nw 100;
f1 25;
fU 10000;
graphFormat raw;
Control settings are read from the $FOAM_CASE/system/noiseDict dictionary,
or user-specified dictionary using the -dict option. Presure data can be
supplied as a single time history at a given point location, or a surface
of time histories, based on the run-time selectable \c noiseModel, e.g.
pressureData
\vebatim
noiseModel <geometryType>Noise;
<geometryType>Coeffs
{
fileName "pressureData"
nHeaderLine 1; // number of header lines
refColumn 0; // reference column index
componentColumns (1); // component column indices
separator " "; // optional (defaults to ",")
mergeSeparators no; // merge multiple separators
outOfBounds clamp; // optional out-of-bounds handling
interpolationScheme linear; // optional interpolation scheme
...
}
\endverbatim
where \<geometryMode\> is either \c point or \surface
Both operation types require:
\verbatim
pRef 101325;
N 65536;
fl 25;
fu 10000;
fftWriteInterval 1;
dataMode point;
\endverbatim
where
\table
Property | Description | Required | Default value
pRef | Reference pressure | no | 0
N | Number of samples in sampling window | no | 65536
nw | Number of sampling windows | no | 100
N | Number of samples in sampling window | no | 65536 (2^16)
fl | Lower frequency band | no | 25
fU | Upper frequency band | no | 10000
graphFormat | Output graph format | no | raw
fu | Upper frequency band | no | 10000
fftWriteInterval | Output interval for FFT data | no | 1
dataMode | Input data mode, 'point' or 'surface' | yes |
\endtable
Current graph outputs include:
- FFT of the pressure data
- narrow-band PFL (pressure-fluctuation level) spectrum
- one-third-octave-band PFL spectrum
- one-third-octave-band pressure spectrum
Point-based pressure data is read using a CSV reader, and output in a
graph format:
\verbatim
pointData
{
csvFileData
{
fileName "pressureData";
nHeaderLine 1;
refColumn 0;
componentColumns (1);
separator " ";
mergeSeparators yes;
}
graphFormat raw;
windowOverlapPercent 0;
nWindow 100; // fixed number of windows
}
\endverbatim
where
\table
Property | Description | Required | Default value
csvFileData | CSV file data dictionary - see CSV.H | yes |
graphFormat | Output graph format | no | raw
windowOverlapPercent | Window overla percent| yes |
nWindow | Number of sampling windows | no | 1, calculated
\endtable
Surface-based pressure data is specified using a surface reader, and output
using a surface writer. The reader type must support multiple time
handling, e.g. the ensight format. Output surfaces are written on a
per-frequency basis:
\verbatim
surfaceData
{
reader ensight;
writer ensight;
inputFile "postProcessing/faceSource1/surface/patch1/patch1.case";
pName p;
windowOverlapPercent 50;
// nWindow 0; // let code decide if not present
}
\endverbatim
where
\table
Property | Description | Required | Default value
reader | Surface reader type | yes |
writer | Surface writer type | yes |
inputFile | Pressure data surface file | yes |
pName | Name of pressure field | no | p
windowOverlapPercent | Window overla percent| yes |
nWindow | Number of sampling windows | no | 1, calculated
\endtable
Current outputs include:
- FFT of the pressure data (Pf)
- narrow-band PFL (pressure-fluctuation level) spectrum (Lf)
- one-third-octave-band PFL spectrum (Ldelta)
- one-third-octave-band pressure spectrum (Pdelta)
Note:
- If using the windowOverlapPercent entry, the code will automatically
determine number of windows
- However, if supplied, the nWindow entry will take precedence
SeeAlso
CSV.H
surfaceReader.H
surfaceWriter.H
noiseFFT.H
noiseModel.H
windowModel.H
\*---------------------------------------------------------------------------*/
#include "noiseFFT.H"
#include "argList.H"
#include "Time.H"
#include "functionObjectFile.H"
#include "CSV.H"
#include "noiseModel.H"
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
using namespace Foam;
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
Foam::scalar checkUniformTimeStep(const scalarField& t)
{
// check that a uniform time step has been applied
scalar deltaT = -1.0;
if (t.size() > 1)
{
for (label i = 1; i < t.size(); i++)
{
scalar dT = t[i] - t[i-1];
if (deltaT < 0)
{
deltaT = dT;
}
if (mag(deltaT - dT) > SMALL)
{
FatalErrorInFunction
<< "Unable to process data with a variable time step"
<< exit(FatalError);
}
}
}
else
{
FatalErrorInFunction
<< "Unable to create FFT with a single value"
<< exit(FatalError);
}
return deltaT;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
int main(int argc, char *argv[])
{
argList::noParallel();
#include "addDictOption.H"
#include "setRootCase.H"
#include "createTime.H"
#include "createFields.H"
Info<< "Reading data file" << endl;
CSV<scalar> pData("pressure", dict, "Data");
// time history data
const scalarField t(pData.x());
// pressure data
const scalarField p(pData.y());
if (t.size() < N)
word dictName("noiseDict");
if (args.optionFound("dict"))
{
FatalErrorInFunction
<< "Block size N = " << N
<< " is larger than number of data = " << t.size()
<< exit(FatalError);
dictName = args["dict"];
}
Info<< " read " << t.size() << " values" << nl << endl;
IOdictionary dict
(
IOobject
(
dictName,
runTime.system(),
runTime,
IOobject::MUST_READ
)
);
Info<< "Creating noise FFT" << endl;
noiseFFT nfft(checkUniformTimeStep(t), p);
nfft -= pRef;
fileName baseFileName(pData.fName().lessExt());
graph Pf(nfft.RMSmeanPf(N, min(nfft.size()/N, nw)));
Info<< " Creating graph for " << Pf.title() << endl;
Pf.write(baseFileName + graph::wordify(Pf.title()), graphFormat);
graph Lf(nfft.Lf(Pf));
Info<< " Creating graph for " << Lf.title() << endl;
Lf.write(baseFileName + graph::wordify(Lf.title()), graphFormat);
graph Ldelta(nfft.Ldelta(Lf, f1, fU));
Info<< " Creating graph for " << Ldelta.title() << endl;
Ldelta.write(baseFileName + graph::wordify(Ldelta.title()), graphFormat);
graph Pdelta(nfft.Pdelta(Pf, f1, fU));
Info<< " Creating graph for " << Pdelta.title() << endl;
Pdelta.write(baseFileName + graph::wordify(Pdelta.title()), graphFormat);
autoPtr<noiseModel> model(noiseModel::New(dict));
model->calculate();
Info<< nl << "End\n" << endl;