Sensitivity analyses

Parameters listing

To proceed to the Sensitivity analyse, we should define the initial list of parameters

Wood stove parameters
Parameters name	Nominal Value	Standart deviation
Nominal Heating Power	6000 W	5%
Nominal Temperature	80°C	5°C
Nominal efficiency	90%	5%
Nominal pressure drop	6000	10%
Water volume	0.009 m3	3%
Mass of boiler	9kg	3%
LHV of the wood	5 kWh/kg	5%
Density of th fuel	1000 kg	10%
CO2 emisson at combustion	0.007 kgCO2/kgFuel	15%

Boiler pump and mixing circuit parameters
Parameters name	Nominal Value	Standart deviation
Thermal conductivity ( lambda/thickness)	0.04 / 4 cm	7%
Pipe length	5 m	3%
Nominal pressure drop	3000 Pa	10%
KV of the 3-Way valve	50 (m3/h.Bar)^(1/2)	20%

Hot water Tank parameters
Parameters name	Nominal Value	Standart deviation
Volume of the Tank	3 m3	5%
height of the Tank	2 m	3%
Insulation thickness	25 cm	7%
Insulation Heat conductivity	0.04 W/(m.K)	10%

Control and Monitoring control
Parameters name	Significant	Standart deviation
Minimal temperature at the bottom of the tank	50°C	2°C
Threshold to switch boiler off	1°C	1°C
Pump delay	60s	5%
Proportionnal gain of PID	0.1	10%
Integral gain of PID	120	10%
Minimal return temperature to the wood stove	60°C	2°C
Test to block boiler if room air temperature is sufficiently high	16°C	1°C

Only sensitivity analyzes are performed in this part of the tutorial. The results of the various sensitivity analyzes (component and system) will therefore be presented. In practice, the process can start with a component and continue until it is calibrated and then start again with another component. We will try here to be methodical and therefore not to use this way of doing.

Sensitivity of the boiler

The following script was applied to obtain the weight of the table "Wood stove parameters

# -*- coding: utf-8 -*-
"""
@author: wthomare
"""
import buildingspy.simulate.Simulator as si
from SALib.sample import saltelli
from SALib.analyze import sobol
import os
from buildingspy.io.outputfile import Reader
import numpy as np
import shutil
from multiprocessing import Pool
import copy

TSimulation = 604800

model = u"FBM.Tutorial.FirstTuto"
s = si.Simulator(model,u"dymola",u"case", packagePath="//media//wthomare//DONNEES//comis")
s.setStopTime(TSimulation)
s.setSolver('dassl')
s.translate()
s1=copy.deepcopy(s)


problem = {
    'num_vars': 9,
    'names': [
    'boi.Q_flow_nominal',
    "boiler.T_nominal",
    "boiler.a",
    "boiler.dp_nominal",
    "boiler.VWat",
    "boiler.mDry",
    "datFue.h",
    "datFue.d",
    "datFue.mCO2"],
    'bounds': [[5700,6300],
               [75,85],
               [85,95],
               [5000,7000],
               [0.0873,0.0927],
               [8.73,9.27],
               [17100,18900],
               [900,1100],
               [0.00608,0.00805]]}

W = saltelli.sample(problem,100, calc_second_order=True)
po = Pool(processes = 4)

def SaveResult(path,texte):
        fichier = open("outputsMorris.txt","a")
        fichier.write(texte + "\n")
        fichier.close()

# Function to set common parameters and to run the simulation
def simulateCase(s):

    s.setStopTime(TSimulation)
    # Kill the process if it does not finish in 10 minute
    s.setTimeOut(600)
    s.showProgressBar(False)
    s.printModelAndTime()
    s.simulate_translated()


def evaluate(i):
    s1.setOutputDirectory('//media//wthomare//DONNEES//FyPy//case %d' % (i))

    s1.addParameters({"boiler.Q_flow_nominal' :W[i,0]})
    s1.addParameters({"boiler.T_nominal" :W[i,1]})
    s1.addParameters({"boiler.a" :W[i,2]})
    s1.addParameters({"boiler.dp_nominal" :W[i,3]})
    s1.addParameters({"boiler.VWat" :W[i,4]})
    s1.addParameters({"boiler.mDry" :W[i,5]})
    s1.addParameters({"datFue.h" :W[i,6]})
    s1.addParameters({"datFue.d" :W[i,7]})
    s1.addParameters({"datFue.mCO2" :W[i,8]})

    simulateCase(s1)

# Main function
if __name__ == '__main__':
   po.map(evaluate,range(len(W)) )
   for i,X in enumerate(W):
        resultFile = os.path.join("//media","wthomare","DONNEES","FyPy",'case %d' % (i),u"ValidationPan.mat")
        outTemp=Reader(resultFile,"dymola")
        SaveResult("//home//wthomare//Bureau//PyWorDir//SaveOutputFile//Boiler",str(outTemp.integral('Solaire.y')))
        shutil.rmtree('//media//wthomare//DONNEES//FyPy//case %d' % (i))

   Y = np.loadtxt("outputsMorris.txt", float)
   Si = sobol.analyze(problem, Y, print_to_console=False)
   shutil.rmtree('case')
   print Si['S1']

Results :

Tables resume wich parameters was significant on our output for each component then for the system

Wood stove parameters
Parameters name	Significant
Nominal Heating Power	Yes
Nominal Temperature	No
Nominal efficiency	Yes
Nominal pressure drop	No
Water volume	No
Mass of boiler	No
LHV of the wood	Yes
Density of th fuel	No
CO2 emisson at combustion	No

Boiler pump and mixing circuit parameters
Parameters name	Significant
Thermal conductivity ( lambda/thickness)	Yes
Pipe length	No
Nominal pressure drop	No
KV of the 3-Way valve	Yes

Hot water Tank parameters
Parameters name	Significant
Volume of the Tank	No
height of the Tank	No
Thermal conductivity ( lambda/thickness)	Yes

Control and Monitoring control
Parameters name	Significant
Minimal temperature at the bottom of the tank	Yes
Threshold to switch boiler off	No
Pump delay	No
Proportionnal gain of PID	No
Integral gain of PID	No
Minimal return temperature to the wood stove	Yes
Test to block boiler if room air temperature is sufficiently high	Yes

Comis Project Website

First Tutorial