foamConductivity.py

Go to the documentation of this file.

import os
from modena import *
import modena.Strategy as Strategy
from fireworks.utilities.fw_utilities import explicit_serialize
from jinja2 import Template
import json
import polymerConductivity
import gasConductivity
import gasMixtureConductivity
import json

json.encoder.FLOAT_REPR = lambda o: format(o, '.12g')

@explicit_serialize

class FoamConductivityExactTask(ModenaFireTask):
    
    def task(self, fw_spec):
        eps = self['point']['eps']
        dcell = self['point']['dcell']
        fstrut = self['point']['fstrut']
        temp = self['point']['T']
        xCO2 = self['point']['x[CO2]']
        xCyP = self['point']['x[CyP]']
        xO2 = self['point']['x[O2]']
        xN2 = self['point']['x[N2]']
        # Write input
        inputs={"upperBoundary": {"temperature": temp+1,"emittance": 0.9}}
        inputs["lowerBoundary"]={"temperature": temp-1,"emittance": 0.9}
        inputs["gasComposition"]={
            "O2": xO2,
            "N2": xN2,
            "CO2": xCO2,
            "Cyclopentane": xCyP
        }
        inputs["gasDensity"]=1.2
        inputs["solidDensity"]=1.1e3
        inputs["sourceOfProperty"]={
            "porosity": "DirectInput",
            "cellSize": "DirectInput",
            "gasComposition": "DirectInput",
            "strutContent": "DirectInput",
            "wallThickness": "DirectInput"
        }
        inputs["porosity"]=eps
        inputs["cellSize"]=dcell
        inputs["morphologyInput"]="strutContent"
        inputs["wallThickness"]=0.5e-6
        inputs["strutContent"]=fstrut
        inputs["strutSize"]=1e-6
        inputs["foamThickness"]=3e-2
        inputs["spatialDiscretization"]=200
        inputs["useWallThicknessDistribution"]=True
        inputs["wallThicknessStandardDeviation"]=0.2
        inputs["numberOfGrayBoxes"]=10
        inputs["numericalEffectiveConductivity"]=False
        # inputs["structureName"]=
        inputs["testMode"]=False
        with open('foamConductivity.json','w') as f:
            json.dump(inputs, f, indent=4)
        os.mkdir('inputs')
        os.rename('foamConductivity.json','inputs/foamConductivity.json')
        # Execute the detailed model
        # path to **this** file + /src/...
        # will break if distributed computing
        ret = os.system(os.path.dirname(os.path.abspath(__file__))+'/src/kfoam')
        # This call enables backward mapping capabilities
        self.handleReturnCode(ret)
        # Analyse output
        # os.getcwd() returns the path to the "launcher" directory
        try:
            FILE = open(os.getcwd()+'/foamConductivity.out','r')        except IOError:
            raise IOError("File not found")

        self['point']['kfoam'] = float(FILE.readline())

        # os.remove('foamConductivity.json')
        # os.remove('foamConductivity.out')


f_foamConductivity = CFunction(
    Ccode='''
#include "modena.h"
#include "math.h"

void tcfoam_SM
(
    const modena_model_t* model,
    const double* inputs,
    double *outputs
)
{
    {% block variables %}{% endblock %}

    const double alpha = parameters[1];
    const double beta = parameters[0];

    const double sigma=5.67e-8;

    double fs,Xs,Xw,X,kappa,kr;
    double kfoam;
    double kgas=gasMixtureConductivity;
    double kpol=polymer_thermal_conductivity;

    fs=alpha*fstrut;
    Xs=(1+4*kgas/(kgas+kpol))/3.0;
    Xw=2*(1+kgas/(2*kpol))/3.0;
    X=(1-fs)*Xw+fs*Xs;
    kappa=4.09*sqrt(1-eps)/dcell;
    kr=16*sigma*pow(T,3)/(3*kappa);
    kfoam = (kgas*eps+kpol*X*(1-eps))/(eps+(1-eps)*X)+beta*kr;

    outputs[0] = kfoam;
}
''',
    # These are global bounds for the function
    inputs={
        'eps': {'min': 0, 'max': 0.995},
        'dcell': {'min': 0, 'max': 1e-1},
        'fstrut': {'min': 0, 'max': 1},
        'gasMixtureConductivity': {'min': 0, 'max': 1e-1},
        'polymer_thermal_conductivity': {'min': 0, 'max': 1e0},
        'T': {'min': 273, 'max': 550},
        'x': {'index': gasConductivity.species, 'min': 0, 'max': 1},
    },
    outputs={
        'kfoam': {'min': 0, 'max': 1e0, 'argPos': 0},
    },
    parameters={
        'param1': {'min': -1e9, 'max': 1e9 + 2, 'argPos': 0},
        'param2': {'min': -1e9, 'max': 1e9 + 2, 'argPos': 1},
    },
)

try: #initialization by initModels
    with open("./inputs/init_foamConductivity.json") as fl:
        foaming_ini=json.load(fl)
except IOError: #automatic initialization
    with open("../inputs/init_foamConductivity.json") as fl:
        foaming_ini=json.load(fl)


m_foamConductivity = BackwardMappingModel(
    _id='foamConductivity',
    surrogateFunction=f_foamConductivity,
    exactTask=FoamConductivityExactTask(),
    substituteModels=[
        gasMixtureConductivity.m_gasMixtureConductivity,\
        polymerConductivity.m_polymer_thermal_conductivity\
    ],
    initialisationStrategy=Strategy.InitialPoints(
        initialPoints=foaming_ini,
    ),
    outOfBoundsStrategy=Strategy.ExtendSpaceStochasticSampling(
        nNewPoints=4
    ),
    parameterFittingStrategy=Strategy.NonLinFitWithErrorContol(
        testDataPercentage=0.2,
        maxError=0.01,
        improveErrorStrategy=Strategy.StochasticSampling(
            nNewPoints=2
        ),
        maxIterations=5  # Currently not used
    ),
)


m_simulation = Strategy.BackwardMappingScriptTask(
    script=os.path.dirname(os.path.abspath(__file__))+'/src/kfoam' +
        ' && cp foamConductivity.out ../results/foamConductivity' +
        ' && cp hahtf.out ../results/foamConductivity'
)