Source code for test_tank_models

''' '''
'''
 ISC License

 Copyright (c) 2016, Autonomous Vehicle Systems Lab, University of Colorado at Boulder

 Permission to use, copy, modify, and/or distribute this software for any
 purpose with or without fee is hereby granted, provided that the above
 copyright notice and this permission notice appear in all copies.

 THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

'''
import sys, os, inspect
import numpy
import pytest
import math

from Basilisk.utilities import SimulationBaseClass
from Basilisk.utilities import unitTestSupport  # general support file with common unit test functions
import matplotlib.pyplot as plt
from Basilisk.utilities import macros
from Basilisk.simulation import spacecraftPlus
from Basilisk.simulation import sim_model
import ctypes
from Basilisk.simulation import gravityEffector
from Basilisk.simulation import spice_interface
from Basilisk.utilities import simIncludeThruster
from Basilisk.simulation import thrusterDynamicEffector
from Basilisk.fswAlgorithms import vehicleConfigData
from Basilisk.simulation import fuelTank

filename = inspect.getframeinfo(inspect.currentframe()).filename
path = os.path.dirname(os.path.abspath(filename))

# uncomment this line is this test is to be skipped in the global unit test run, adjust message as needed
# @pytest.mark.skipif(conditionstring)
# uncomment this line if this test has an expected failure, adjust message as needed
# @pytest.mark.xfail() # need to update how the RW states are defined
# provide a unique test method name, starting with test_
[docs]def test_tankModelTest(show_plots): """Module Unit Test""" [testResults, testMessage] = tankModelConstantVolume(show_plots) assert testResults < 1, testMessage [testResults, testMessage] = tankModelConstantDensity(show_plots) assert testResults < 1, testMessage [testResults, testMessage] = tankModelEmptying(show_plots) assert testResults < 1, testMessage [testResults, testMessage] = tankModelUniformBurn(show_plots) assert testResults < 1, testMessage [testResults, testMessage] = tankModelCentrifugalBurn(show_plots) assert testResults < 1, testMessage
def tankModelConstantVolume(show_plots): # The __tracebackhide__ setting influences pytest showing of tracebacks: # the mrp_steering_tracking() function will not be shown unless the # --fulltrace command line option is specified. __tracebackhide__ = True testFailCount = 0 # zero unit test result counter testMessages = [] # create empty list to store test log messages model = fuelTank.cvar.FuelTankModelConstantVolume model.propMassInit = 10 model.r_TcT_TInit = [[1],[1],[1]] model.radiusTankInit = 5 trials = [(0, 0), (10, -1), (5, -1)] #mFuel, mDotFuel true_ITankPntT_T = [ [0,0,0,0,0,0,0,0,0], [100,0,0,0,100,0,0,0,100], [50,0,0,0,50,0,0,0,50] ] true_IPrimeTankPntT_T = [ [0,0,0,0,0,0,0,0,0], [-10,0,0,0,-10,0,0,0,-10], [-10,0,0,0,-10,0,0,0,-10] ] true_r_TcT_T = [ [1,1,1], [1,1,1], [1,1,1] ] true_rPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] true_rPPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] accuracy = 1e-8 for idx, trial in enumerate(trials): model.computeTankProps(trial[0]) model.computeTankPropDerivs(*trial) dataITank = model.ITankPntT_T dataITank = [0] + [dataITank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataITank, true_ITankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed ITankPntT_T test") dataIPrimeTank = model.IPrimeTankPntT_T dataIPrimeTank = [0] + [dataIPrimeTank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataIPrimeTank, true_IPrimeTankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed IPrimeTankPntT_T test") dataR = model.r_TcT_T dataR = [0] + [dataR[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataR, true_r_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed r_TcT_T test") dataRPrime = model.rPrime_TcT_T dataRPrime = [0] + [dataRPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPrime, true_rPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed rPrime_TcT_T test") dataRPPrime = model.rPPrime_TcT_T dataRPPrime = [0] + [dataRPPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPPrime, true_rPPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed rPPrime_TcT_T test") if testFailCount == 0: print("PASSED: " + " Fuel Tank constant volume unit test") snippetName = 'ConstVolPassFail' passFail(testFailCount, snippetName) assert testFailCount < 1, testMessages # return fail count and join into a single string all messages in the list # testMessage return [testFailCount, ''.join(testMessages)] def tankModelConstantDensity(show_plots): # The __tracebackhide__ setting influences pytest showing of tracebacks: # the mrp_steering_tracking() function will not be shown unless the # --fulltrace command line option is specified. __tracebackhide__ = True testFailCount = 0 # zero unit test result counter testMessages = [] # create empty list to store test log messages model = fuelTank.cvar.FuelTankModelConstantDensity model.propMassInit = 10; model.r_TcT_TInit = [[1],[1],[1]] model.radiusTankInit = 5 trials = [(0, 0), (10, -1), (5, -1)] #mFuel, mDotFuel true_ITankPntT_T = [ [0,0,0,0,0,0,0,0,0], [100,0,0,0,100,0,0,0,100], [31.498026247371826,0,0,0,31.498026247371826,0,0,0,31.498026247371826] ] true_IPrimeTankPntT_T = [ [0,0,0,0,0,0,0,0,0], [-16.666666666666668,0,0,0,-16.666666666666668,0,0,0,-16.666666666666668], [-10.499342082457275,0,0,0,-10.499342082457275,0,0,0,-10.499342082457275] ] true_r_TcT_T = [ [1,1,1], [1,1,1], [1,1,1] ] true_rPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] true_rPPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] accuracy = 1e-8 for idx, trial in enumerate(trials): model.computeTankProps(trial[0]) model.computeTankPropDerivs(*trial) dataITank = model.ITankPntT_T dataITank = [0] + [dataITank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataITank, true_ITankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed ITankPntT_T test") dataIPrimeTank = model.IPrimeTankPntT_T dataIPrimeTank = [0] + [dataIPrimeTank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataIPrimeTank, true_IPrimeTankPntT_T[idx],9,accuracy): print(dataIPrimeTank, idx) testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed IPrimeTankPntT_T test") dataR = model.r_TcT_T dataR = [0] + [dataR[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataR, true_r_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed r_TcT_T test") dataRPrime = model.rPrime_TcT_T dataRPrime = [0] + [dataRPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPrime, true_rPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed rPrime_TcT_T test") dataRPPrime = model.rPPrime_TcT_T dataRPPrime = [0] + [dataRPPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPPrime, true_rPPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank constant volume unit test failed rPPrime_TcT_T test") if testFailCount == 0: print("PASSED: " + " Fuel Tank constant volume unit test") snippetName = 'ConstDensPassFail' passFail(testFailCount, snippetName) assert testFailCount < 1, testMessages # return fail count and join into a single string all messages in the list # testMessage return [testFailCount, ''.join(testMessages)] def tankModelEmptying(show_plots): # The __tracebackhide__ setting influences pytest showing of tracebacks: # the mrp_steering_tracking() function will not be shown unless the # --fulltrace command line option is specified. __tracebackhide__ = True testFailCount = 0 # zero unit test result counter testMessages = [] # create empty list to store test log messages model = fuelTank.cvar.FuelTankModelEmptying model.propMassInit = 10 model.r_TcT_TInit = [[1],[1],[1]] model.radiusTankInit = 5 trials = [(0, 0), (10, -1), (5, -1)] #mFuel, mDotFuel true_ITankPntT_T = [ [0,0,0,0,0,0,0,0,0], [100,0,0,0,100,0,0,0,100], [50.0,0,0,0,50.0,0,0,0,50] ] true_IPrimeTankPntT_T = [ [0,0,0,0,0,0,0,0,0], [0,0,0,0,0,0,0,0,0], [-8.75,0,0,0,-8.75,0,0,0,-12.5] ] true_r_TcT_T = [ [1,1,1-5.0], [1,1,1], [1,1,1.0-15.0/8.0] ] true_rPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,-3.0/8.0] ] true_rPPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,-17.0/30.0] ] accuracy = 1e-8 for idx, trial in enumerate(trials): model.computeTankProps(trial[0]) model.computeTankPropDerivs(*trial) dataITank = model.ITankPntT_T dataITank = [0] + [dataITank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqual(dataITank, true_ITankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank emptying unit test failed ITankPntT_T test") dataIPrimeTank = model.IPrimeTankPntT_T dataIPrimeTank = [0] + [dataIPrimeTank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataIPrimeTank, true_IPrimeTankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank emptying unit test failed IPrimeTankPntT_T test") dataR = model.r_TcT_T dataR = [0] + [dataR[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataR, true_r_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank emptying unit test failed r_TcT_T test") dataRPrime = model.rPrime_TcT_T dataRPrime = [0] + [dataRPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPrime, true_rPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank emptying unit test failed rPrime_TcT_T test") dataRPPrime = model.rPPrime_TcT_T dataRPPrime = [0] + [dataRPPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPPrime, true_rPPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank emptying unit test failed rPPrime_TcT_T test") if testFailCount == 0: print("PASSED: " + " Fuel Tank constant volume unit test") snippetName = 'EmptyingPassFail' passFail(testFailCount, snippetName) assert testFailCount < 1, testMessages # return fail count and join into a single string all messages in the list # testMessage return [testFailCount, ''.join(testMessages)] def tankModelUniformBurn(show_plots): # The __tracebackhide__ setting influences pytest showing of tracebacks: # the mrp_steering_tracking() function will not be shown unless the # --fulltrace command line option is specified. __tracebackhide__ = True testFailCount = 0 # zero unit test result counter testMessages = [] # create empty list to store test log messages model = fuelTank.cvar.FuelTankModelUniformBurn model.propMassInit = 10 model.r_TcT_TInit = [[1],[1],[1]] model.radiusTankInit = 5 model.lengthTank = 5; trials = [(0, 0), (10, -1), (5, -1)] #mFuel, mDotFuel true_ITankPntT_T = [ [0,0,0,0,0,0,0,0,0], [83.33333333333334,0,0,0,83.33333333333334,0,0,0,125], [41.66666666666667,0,0,0,41.66666666666667,0,0,0,62.5] ] true_IPrimeTankPntT_T = [ [0,0,0,0,0,0,0,0,0], [-8.3333333333334,0,0,0,-8.3333333333334,0,0,0,-12.5], [-8.3333333333334,0,0,0,-8.3333333333334,0,0,0,-12.5] ] true_r_TcT_T = [ [1,1,1], [1,1,1], [1,1,1] ] true_rPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] true_rPPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] accuracy = 1e-8 for idx, trial in enumerate(trials): model.computeTankProps(trial[0]) model.computeTankPropDerivs(*trial) dataITank = model.ITankPntT_T dataITank = [0] + [dataITank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataITank, true_ITankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank centrifugal burn unit test failed ITankPntT_T test") dataIPrimeTank = model.IPrimeTankPntT_T dataIPrimeTank = [0] + [dataIPrimeTank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataIPrimeTank, true_IPrimeTankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank centrifugal burn unit test failed IPrimeTankPntT_T test") dataR = model.r_TcT_T dataR = [0] + [dataR[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataR, true_r_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank centrifugal burn unit test failed r_TcT_T test") dataRPrime = model.rPrime_TcT_T dataRPrime = [0] + [dataRPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPrime, true_rPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank centrifugal burn unit test failed rPrime_TcT_T test") dataRPPrime = model.rPPrime_TcT_T dataRPPrime = [0] + [dataRPPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPPrime, true_rPPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank centrifugal burn unit test failed rPPrime_TcT_T test") if testFailCount == 0: print("PASSED: " + " Fuel Tank constant volume unit test") snippetName = 'UniformBurnPassFail' passFail(testFailCount, snippetName) assert testFailCount < 1, testMessages # return fail count and join into a single string all messages in the list # testMessage return [testFailCount, ''.join(testMessages)] def tankModelCentrifugalBurn(show_plots): # The __tracebackhide__ setting influences pytest showing of tracebacks: # the mrp_steering_tracking() function will not be shown unless the # --fulltrace command line option is specified. __tracebackhide__ = True testFailCount = 0 # zero unit test result counter testMessages = [] # create empty list to store test log messages model = fuelTank.cvar.FuelTankModelCentrifugalBurn model.propMassInit = 10 model.r_TcT_TInit = [[1],[1],[1]] model.radiusTankInit = 5 model.lengthTank = 5 trials = [(0, 0), (10, -1), (5, -1)] #mFuel, mDotFuel true_ITankPntT_T = [ [0,0,0,0,0,0,0,0,0], [83.33333333333334,0,0,0,83.33333333333334,0,0,0,125], [57.291666666666671,0,0,0,57.291666666666671,0,0,0,93.75] ] true_IPrimeTankPntT_T = [ [0,0,0,0,0,0,0,0,0], [-2.0833333333333335,0,0,0,-2.0833333333333335,0,0,0,0.0], [-8.3333333333333339,0,0,0,-8.3333333333333339,0,0,0,-12.500000000000002] ] true_r_TcT_T = [ [1,1,1], [1,1,1], [1,1,1] ] true_rPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] true_rPPrime_TcT_T = [ [0,0,0], [0,0,0], [0,0,0] ] accuracy = 1e-8 for idx, trial in enumerate(trials): model.computeTankProps(trial[0]) model.computeTankPropDerivs(*trial) dataITank = model.ITankPntT_T dataITank = [0] + [dataITank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataITank, true_ITankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank uniform burn unit test failed ITankPntT_T test") dataIPrimeTank = model.IPrimeTankPntT_T dataIPrimeTank = [0] + [dataIPrimeTank[i][j] for i in range(3) for j in range(3)] if not unitTestSupport.isArrayEqualRelative(dataIPrimeTank, true_IPrimeTankPntT_T[idx],9,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank uniform burn unit test failed IPrimeTankPntT_T test") dataR = model.r_TcT_T dataR = [0] + [dataR[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataR, true_r_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank uniform burn unit test failed r_TcT_T test") dataRPrime = model.rPrime_TcT_T dataRPrime = [0] + [dataRPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPrime, true_rPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank uniform burn unit test failed rPrime_TcT_T test") dataRPPrime = model.rPPrime_TcT_T dataRPPrime = [0] + [dataRPPrime[i][0] for i in range(3)] if not unitTestSupport.isArrayEqualRelative(dataRPPrime, true_rPPrime_TcT_T[idx],3,accuracy): testFailCount += 1 testMessages.append("FAILED: Fuel Tank uniform burn unit test failed rPPrime_TcT_T test") if testFailCount == 0: print("PASSED: " + " Fuel Tank constant volume unit test") snippetName = 'CentrifugalPassFail' passFail(testFailCount, snippetName) assert testFailCount < 1, testMessages # return fail count and join into a single string all messages in the list # testMessage return [testFailCount, ''.join(testMessages)] def passFail(testFailCountInput, snippetName): if testFailCountInput < 1: textMsg = 'PASSED' textColor = 'ForestGreen' else: textMsg = 'FAILED' textColor = 'Red' texSnippet = r'\textcolor{' + textColor + '}{'+ textMsg + '}' unitTestSupport.writeTeXSnippet(snippetName, texSnippet, path) if __name__ == "__main__": tankModelEmptying(False)