ProcessMultiShear.cpp

Go to the documentation of this file.

////////////////////////////////////////////////////////////////////////////////
//
//  File: ProcessMultiShear.cpp
//
//  For more information, please see: http://www.nektar.info/
//
//  The MIT License
//
//  Copyright (c) 2006 Division of Applied Mathematics, Brown University (USA),
//  Department of Aeronautics, Imperial College London (UK), and Scientific
//  Computing and Imaging Institute, University of Utah (USA).
//
//  Permission is hereby granted, free of charge, to any person obtaining a
//  copy of this software and associated documentation files (the "Software"),
//  to deal in the Software without restriction, including without limitation
//  the rights to use, copy, modify, merge, publish, distribute, sublicense,
//  and/or sell copies of the Software, and to permit persons to whom the
//  Software is furnished to do so, subject to the following conditions:
//
//  The above copyright notice and this permission notice shall be included
//  in all copies or substantial portions of the Software.
//
//  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
//  OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
//  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
//  THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
//  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
//  FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
//  DEALINGS IN THE SOFTWARE.
//
//  Description: Computes tawss, osi, transwss, afi, cfi fields.
//
////////////////////////////////////////////////////////////////////////////////
 
#include <iostream>
#include <sstream>
#include <string>
using namespace std;
 
#include <boost/core/ignore_unused.hpp>
 
#include <LibUtilities/BasicUtils/SharedArray.hpp>
#include <MultiRegions/ExpList.h>
 
#include "ProcessMultiShear.h"
 
namespace Nektar
{
namespace FieldUtils
{
 
ModuleKey ProcessMultiShear::className =
    GetModuleFactory().RegisterCreatorFunction(
        ModuleKey(eProcessModule, "shear"), ProcessMultiShear::create,
        "Computes shear stress metrics.");
 
ProcessMultiShear::ProcessMultiShear(FieldSharedPtr f) : ProcessModule(f)
{
    m_config["N"] = ConfigOption(false, "1", "Number of chk or fld files");
    m_config["fromfld"] = ConfigOption(
        false, "NotSet",
        "First fld file. First underscore flags position of id in name.");
}
 
ProcessMultiShear::~ProcessMultiShear()
{
}
 
void ProcessMultiShear::Process(po::variables_map &vm)
{
    m_f->SetUpExp(vm);
 
    // Skip in case of empty partition
    if (m_f->m_exp[0]->GetNumElmts() == 0)
    {
        return;
    }
 
    ASSERTL0(m_config["fromfld"].as<string>().compare("NotSet") != 0,
             "Need to specify fromfld=file.fld ");
 
    int nstart, i, j, nfields = 0;
    bool wssg      = false;
    NekDouble nfld = m_config["N"].as<NekDouble>();
    string fromfld, basename, endname, nstartStr;
    stringstream filename;
    vector<string> infiles(nfld);
    vector<std::shared_ptr<Field>> fromField(nfld);
 
    // Set up list of input fld files.
    fromfld  = m_config["fromfld"].as<string>();
    basename = fromfld.substr(0, fromfld.find_first_of("_") + 1);
    filename << fromfld.substr(fromfld.find_first_of("_") + 1, fromfld.size());
    filename >> nstart;
    filename.str("");
    filename << nstart;
    filename >> nstartStr;
    filename.str("");
    endname = fromfld.substr(fromfld.find(nstartStr) + nstartStr.size(),
                             fromfld.size());
 
    for (i = 0; i < nfld; ++i)
    {
        stringstream filename;
        filename << basename << i + nstart << endname;
        filename >> infiles[i];
        cout << infiles[i] << endl;
    }
 
    for (i = 0; i < nfld; ++i)
    {
        fromField[i]            = std::shared_ptr<Field>(new Field());
        fromField[i]->m_session = m_f->m_session;
        fromField[i]->m_graph   = m_f->m_graph;
    }
 
    // Import all fld files.
    for (i = 0; i < nfld; ++i)
    {
        if (m_f->m_exp.size())
        {
            // Set up ElementGIDs in case of parallel processing
            Array<OneD, int> ElementGIDs(m_f->m_exp[0]->GetExpSize());
            for (j = 0; j < m_f->m_exp[0]->GetExpSize(); ++j)
            {
                ElementGIDs[j] =
                    m_f->m_exp[0]->GetExp(j)->GetGeom()->GetGlobalID();
            }
            m_f->FieldIOForFile(infiles[i])
                ->Import(infiles[i], fromField[i]->m_fielddef,
                         fromField[i]->m_data,
                         LibUtilities::NullFieldMetaDataMap, ElementGIDs);
        }
        else
        {
            m_f->FieldIOForFile(infiles[i])
                ->Import(infiles[i], fromField[i]->m_fielddef,
                         fromField[i]->m_data,
                         LibUtilities::NullFieldMetaDataMap);
        }
 
        nfields = fromField[i]->m_fielddef[0]->m_fields.size();
        int NumHomogeneousDir =
            fromField[i]->m_fielddef[0]->m_numHomogeneousDir;
 
        if (nfields == 5 || nfields == 7)
        {
            wssg = true;
        }
 
        // Set up Expansion information to use mode order from field
        fromField[i]->m_graph->SetExpansionInfo(fromField[i]->m_fielddef);
 
        // Set up expansions, and extract data.
        fromField[i]->m_exp.resize(nfields);
        fromField[i]->m_exp[0] =
            fromField[i]->SetUpFirstExpList(NumHomogeneousDir, true);
 
        for (j = 1; j < nfields; ++j)
        {
            fromField[i]->m_exp[j] = m_f->AppendExpList(NumHomogeneousDir);
        }
 
        for (j = 0; j < nfields; ++j)
        {
            for (int k = 0; k < fromField[i]->m_data.size(); ++k)
            {
                fromField[i]->m_exp[j]->ExtractDataToCoeffs(
                    fromField[i]->m_fielddef[k], fromField[i]->m_data[k],
                    fromField[i]->m_fielddef[k]->m_fields[j],
                    fromField[i]->m_exp[j]->UpdateCoeffs());
            }
            fromField[i]->m_exp[j]->BwdTrans(
                fromField[i]->m_exp[j]->GetCoeffs(),
                fromField[i]->m_exp[j]->UpdatePhys());
        }
    }
 
    int spacedim = m_f->m_graph->GetSpaceDimension();
    if ((fromField[0]->m_fielddef[0]->m_numHomogeneousDir) == 1 ||
        (fromField[0]->m_fielddef[0]->m_numHomogeneousDir) == 2)
    {
        spacedim = 3;
    }
 
    int nout = 5; // TAWSS, OSI, transWSS, AFI, CFI (optional WSSG)
    if (wssg)
    {
        nout = 6;
    }
 
    int npoints = fromField[0]->m_exp[0]->GetNpoints();
    Array<OneD, Array<OneD, NekDouble>> normTemporalMeanVec(spacedim),
        normCrossDir(spacedim), outfield(nout), dtemp(spacedim);
    Array<OneD, NekDouble> TemporalMeanMag(npoints, 0.0),
        DotProduct(npoints, 0.0), temp(npoints, 0.0);
 
    for (i = 0; i < spacedim; ++i)
    {
        normTemporalMeanVec[i] = Array<OneD, NekDouble>(npoints);
        normCrossDir[i]        = Array<OneD, NekDouble>(npoints);
        dtemp[i]               = Array<OneD, NekDouble>(npoints);
        Vmath::Zero(npoints, normTemporalMeanVec[i], 1);
        Vmath::Zero(npoints, normCrossDir[i], 1);
    }
 
    for (i = 0; i < nout; ++i)
    {
        outfield[i] = Array<OneD, NekDouble>(npoints, 0.0);
    }
 
    // -----------------------------------------------------
    // Compute temporal average wall shear stress vector,
    // it's spatial average, and normalise it.
    for (i = 0; i < nfld; ++i)
    {
        for (j = 0; j < spacedim; ++j)
        {
            Vmath::Vadd(npoints, fromField[i]->m_exp[j]->GetPhys(), 1,
                        normTemporalMeanVec[j], 1, normTemporalMeanVec[j], 1);
        }
    }
 
    for (i = 0; i < spacedim; ++i)
    {
        Vmath::Smul(npoints, 1.0 / nfld, normTemporalMeanVec[i], 1,
                    normTemporalMeanVec[i], 1);
        Vmath::Vvtvp(npoints, normTemporalMeanVec[i], 1, normTemporalMeanVec[i],
                     1, TemporalMeanMag, 1, TemporalMeanMag, 1);
    }
 
    Vmath::Vsqrt(npoints, TemporalMeanMag, 1, TemporalMeanMag, 1);
 
    for (i = 0; i < spacedim; ++i)
    {
        Vmath::Vdiv(npoints, normTemporalMeanVec[i], 1, TemporalMeanMag, 1,
                    normTemporalMeanVec[i], 1);
    }
    //---------------------------------------------------
 
    if (wssg) // cross product with normals to obtain direction parallel to
              // temporal mean vector.
    {
        Vmath::Vmul(npoints, fromField[0]->m_exp[nfields - 1]->GetPhys(), 1,
                    normTemporalMeanVec[1], 1, normCrossDir[0], 1);
        Vmath::Vvtvm(npoints, fromField[0]->m_exp[nfields - 2]->GetPhys(), 1,
                     normTemporalMeanVec[2], 1, normCrossDir[0], 1,
                     normCrossDir[0], 1);
 
        Vmath::Vmul(npoints, fromField[0]->m_exp[nfields - 3]->GetPhys(), 1,
                    normTemporalMeanVec[2], 1, normCrossDir[1], 1);
        Vmath::Vvtvm(npoints, fromField[0]->m_exp[nfields - 1]->GetPhys(), 1,
                     normTemporalMeanVec[0], 1, normCrossDir[1], 1,
                     normCrossDir[1], 1);
 
        Vmath::Vmul(npoints, fromField[0]->m_exp[nfields - 2]->GetPhys(), 1,
                    normTemporalMeanVec[0], 1, normCrossDir[2], 1);
        Vmath::Vvtvm(npoints, fromField[0]->m_exp[nfields - 3]->GetPhys(), 1,
                     normTemporalMeanVec[1], 1, normCrossDir[2], 1,
                     normCrossDir[2], 1);
    }
 
    // Compute tawss, trs,  osi, taafi, tacfi, WSSG.
    for (i = 0; i < nfld; ++i)
    {
        for (j = 0; j < spacedim; ++j)
        {
            Vmath::Vvtvp(npoints, fromField[i]->m_exp[j]->GetPhys(), 1,
                         normTemporalMeanVec[j], 1, DotProduct, 1, DotProduct,
                         1);
        }
 
        // TAWSS
        Vmath::Vadd(npoints, fromField[i]->m_exp[spacedim]->GetPhys(), 1,
                    outfield[0], 1, outfield[0], 1);
 
        // transWSS
        Vmath::Vmul(npoints, DotProduct, 1, DotProduct, 1, temp, 1);
        Vmath::Vvtvm(npoints, fromField[i]->m_exp[spacedim]->GetPhys(), 1,
                     fromField[i]->m_exp[spacedim]->GetPhys(), 1, temp, 1, temp,
                     1);
 
        for (j = 0; j < npoints; ++j)
        {
            if (temp[j] > 0.0)
            {
                outfield[1][j] = outfield[1][j] + sqrt(temp[j]);
            }
        }
 
        // TAAFI
        Vmath::Vdiv(npoints, DotProduct, 1,
                    fromField[i]->m_exp[spacedim]->GetPhys(), 1, temp, 1);
        Vmath::Vadd(npoints, temp, 1, outfield[3], 1, outfield[3], 1);
 
        // TACFI
        for (j = 0; j < npoints; ++j)
        {
            temp[j] = 1 - temp[j] * temp[j];
            if (temp[j] > 0.0)
            {
                outfield[4][j] = outfield[4][j] + sqrt(temp[j]);
            }
        }
 
        // WSSG
        if (wssg)
        {
            // parallel component:
            Vmath::Zero(npoints, temp, 1);
 
            fromField[i]->m_exp[0]->PhysDeriv(DotProduct, dtemp[0], dtemp[1],
                                              dtemp[2]);
            for (j = 0; j < spacedim; j++)
            {
                Vmath::Vvtvp(npoints, dtemp[j], 1, normTemporalMeanVec[j], 1,
                             temp, 1, temp, 1);
            }
            Vmath::Vmul(npoints, temp, 1, temp, 1, temp, 1);
 
            // perpendicular component:
            Vmath::Zero(npoints, DotProduct, 1);
 
            for (j = 0; j < spacedim; ++j)
            {
                Vmath::Vvtvp(npoints, fromField[i]->m_exp[j]->GetPhys(), 1,
                             normCrossDir[j], 1, DotProduct, 1, DotProduct, 1);
            }
            fromField[i]->m_exp[0]->PhysDeriv(DotProduct, dtemp[0], dtemp[1],
                                              dtemp[2]);
            Vmath::Zero(npoints, DotProduct, 1);
 
            for (j = 0; j < spacedim; j++)
            {
                Vmath::Vvtvp(npoints, dtemp[j], 1, normCrossDir[j], 1,
                             DotProduct, 1, DotProduct, 1);
            }
            Vmath::Vvtvp(npoints, DotProduct, 1, DotProduct, 1, temp, 1, temp,
                         1);
 
            // Final wssg computation
            Vmath::Vsqrt(npoints, temp, 1, temp, 1);
            Vmath::Vadd(npoints, temp, 1, outfield[5], 1, outfield[5], 1);
        }
 
        Vmath::Zero(npoints, DotProduct, 1);
    }
 
    // Divide by nfld
    Vmath::Smul(npoints, 1.0 / nfld, outfield[0], 1, outfield[0], 1);
    Vmath::Smul(npoints, 1.0 / nfld, outfield[1], 1, outfield[1], 1);
    Vmath::Smul(npoints, 1.0 / nfld, outfield[3], 1, outfield[3], 1);
    Vmath::Smul(npoints, 1.0 / nfld, outfield[4], 1, outfield[4], 1);
 
    // OSI
    for (i = 0; i < npoints; ++i)
    {
        outfield[2][i] = 0.5 * (1 - TemporalMeanMag[i] / outfield[0][i]);
    }
 
    /* TAWSS = sum(wss)/nfld
     * transWSS = sum( sqrt( wss^2 - (wss . normTempMean)^2) )/nfld.
     * OSI = 0.5*(1-TemporalMeanMag/TAWSS)
     * TAAFI = sum(cos)/nfld
     * TACFI = sum(sin)/nfld = sum( sqrt(1-cos^2) )/nfld.
     */
 
    m_f->m_exp.resize(nout);
    m_f->m_fielddef = fromField[0]->m_fielddef;
    m_f->m_exp[0] =
        m_f->SetUpFirstExpList(m_f->m_fielddef[0]->m_numHomogeneousDir, true);
 
    for (i = 1; i < nout; ++i)
    {
        m_f->m_exp[i] =
            m_f->AppendExpList(m_f->m_fielddef[0]->m_numHomogeneousDir);
    }
 
    m_f->m_fielddef[0]->m_fields.resize(nout);
    m_f->m_fielddef[0]->m_fields[0] = "TAWSS";
    m_f->m_fielddef[0]->m_fields[1] = "transWSS";
    m_f->m_fielddef[0]->m_fields[2] = "OSI";
    m_f->m_fielddef[0]->m_fields[3] = "TAAFI";
    m_f->m_fielddef[0]->m_fields[4] = "TACFI";
 
    if (wssg)
    {
        m_f->m_fielddef[0]->m_fields[5] = "|WSSG|";
        Vmath::Smul(npoints, 1.0 / nfld, outfield[5], 1, outfield[5], 1);
    }
 
    m_f->m_variables = m_f->m_fielddef[0]->m_fields;
 
    for (i = 0; i < nout; ++i)
    {
        m_f->m_exp[i]->FwdTrans(outfield[i], m_f->m_exp[i]->UpdateCoeffs());
        m_f->m_exp[i]->BwdTrans(m_f->m_exp[i]->GetCoeffs(),
                                m_f->m_exp[i]->UpdatePhys());
    }
}
} // namespace FieldUtils
} // namespace Nektar