-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathfitData_Cantera.py
170 lines (132 loc) · 7.76 KB
/
fitData_Cantera.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import cantera as ct
import numpy as np
import math
# reference about fitting thermo data: https://github.com/Upstream-Applied-Science/canteraJanaf
# reference about writing thermo file: https://github.com/jiweiqi/Cantera2Chemkin/blob/master/soln2ck.py
def build_nasa(nasa_coeffs, row):
"""
Creates string of nasa polynomial coefficients
:param nasa_coeffs
cantera species thermo coefficients object
:param row
which row to write coefficients in
"""
line_coeffs = ''
lines = [[1, 2, 3, 4, 5], [6, 7, 8, 9, 10], [11, 12, 13, 14]]
line_index = lines[row-2]
for ix, c in enumerate(nasa_coeffs):
if ix in line_index:
if c >= 0:
line_coeffs += ' '
line_coeffs += str('{:.8e}'.format(c))
return line_coeffs
def write(gas, thermo_fileName_new, plot):
"""Function to write cantera gas object to inp file.
:param gas:
gas: Cantera gas object,
thermo_fileName_new: Name of converted thermo file
plot: whether plot or not
"""
with open(thermo_fileName_new, 'w+') as f:
#f.write('THERMO ALL' + '\n' +
# ' 300.000 1000.000 5000.000' +'\n')
Tlow = 300
Tcommon = 1000
Thigh = 5000
deltaT = 100
T_low = np.arange(Tlow,Tcommon+deltaT,deltaT)
T_high = np.arange(Tcommon,Thigh+deltaT,deltaT)
temp_rangeAll = '{0:10.3f}'.format(Tlow) + '{0:10.3f}'.format(Tcommon) + '{0:10.3f}'.format(Thigh)
f.write('THERMO ALL' + '\n' + temp_rangeAll + '\n')
#write data for each species in the Solution object
for sp_index in range(len(gas.species_names)):
species = gas.species(sp_index)
species_name = gas.species_name(sp_index)
molecular_weight = gas.molecular_weights[sp_index]
nasa_coeffs = species.thermo.coeffs # orininal coeffs, size = 15, [0]=Tcommon, [1-7] is high coeffs, [8-14] is low coeffs
temp_range = '{:<10}'.format(Tlow) + '{:<10}'.format(Thigh) + '{:<8}'.format(Tcommon)
if (nasa_coeffs[0] != Tcommon):
print("fitting species: %s"%(species_name), "from ",nasa_coeffs[0], "to ",Tcommon)
cp_low = [species.thermo.cp(T)/molecular_weight for T in T_low] # [J/(kmol·K)] / [kg/kmol] = [J/(kg·K)]
cp_high = [species.thermo.cp(T)/molecular_weight for T in T_high] # [J/(kmol·K)] / [kg/kmol] = [J/(kg·K)]
cp_low_polynomials = np.polyfit(T_low, cp_low, 4)
cp_high_polynomials = np.polyfit(T_high, cp_high, 4)
R = ct.gas_constant/molecular_weight # [J/(kmol·K)] / [kg/kmol] = [J/(kg·K)]
#reverse order and divide by R (np.ployfit has calculated coeffs*R so we must divide by R to get just the coeffs)
a1_a5_low = np.flip(cp_low_polynomials,0)/R
a1_a5_high = np.flip(cp_high_polynomials,0)/R
T_ref = 298.15
h = species.thermo.h(T_ref)/molecular_weight # [J/kmol] / [kg/kmol] = [J/kg]
s = species.thermo.s(T_ref)/molecular_weight # [J/kmol/K] / [kg/kmol] = [J/(kg·K)]
a6_low = h/R - np.sum(np.array([a1_a5_low[i]*T_ref**(i+1)/(i+1) for i in range(5)]))
a6_high = h/R - np.sum(np.array([a1_a5_high[i]*T_ref**(i+1)/(i+1) for i in range(5)]))
a7_low = s/R - np.sum(np.array([a1_a5_low[i]*T_ref**(i)/(i) for i in range(1,5)]))-a1_a5_low[0]*math.log(T_ref)
a7_high = s/R - np.sum(np.array([a1_a5_high[i]*T_ref**(i)/(i) for i in range(1,5)]))-a1_a5_high[0]*math.log(T_ref)
nasa_coeffs = np.concatenate([[Tcommon], a1_a5_high, [a6_high], [a7_high], a1_a5_low, [a6_low], [a7_low]])
if plot:
Tlist = np.concatenate([T_low, T_high])
cp = [species.thermo.cp(T)/molecular_weight for T in Tlist] # [J/(kg·K)]
cp_fit_low = [ np.sum(np.array([a1_a5_low[i]*T**(i) for i in range(5)])) for T in T_low] # cp/R
cp_fit_high = [ np.sum(np.array([a1_a5_high[i]*T**(i) for i in range(5)])) for T in T_high] # cp/R
cp_fit = np.concatenate([cp_fit_low, cp_fit_high])*R
a1_a5_org_low = species.thermo.coeffs[8:13]
a1_a5_org_high = species.thermo.coeffs[1:6]
cp_bad_low = [ np.sum(np.array([a1_a5_org_low[i]*T**(i) for i in range(5)])) for T in T_low] # cp/R
cp_bad_high = [ np.sum(np.array([a1_a5_org_high[i]*T**(i) for i in range(5)])) for T in T_high] # cp/R
cp_bad = np.concatenate([cp_bad_low, cp_bad_high])*R
enthalpy = [species.thermo.h(T)/molecular_weight for T in Tlist] # J/kg
enthalpy_fit_low = [ a6_low + np.sum(np.array([a1_a5_low[i]*T**(i+1)/(i+1) for i in range(5)])) for T in T_low] # h/R
enthalpy_fit_high = [ a6_high + np.sum(np.array([a1_a5_high[i]*T**(i+1)/(i+1) for i in range(5)])) for T in T_high] # h/R
enthalpy_fit = np.concatenate([enthalpy_fit_low, enthalpy_fit_high])*R
fig = plt.figure()
ax1 = fig.add_subplot()
ax1.scatter(Tlist, cp, c='blue', label="cp_org, Tcommon=%d"%(species.thermo.coeffs[0]))
ax1.plot(Tlist, cp_fit, c='red', label="cp_fit, Tcommon=1000")
ax1.plot(Tlist, cp_bad, c='purple', label="cp_bad")
ax1.set_xlabel("T [K]")
ax1.set_ylabel("cp [J/kg]")
ax2 = plt.twinx()
ax2.scatter(Tlist, enthalpy, c='black', label="h_org, Tcommon=%d"%(species.thermo.coeffs[0]))
ax2.plot(Tlist, enthalpy_fit, c='green', label="h_fit, Tcommon=1000")
ax2.set_ylabel("h [J/kg]")
fig.legend(loc="lower right", bbox_to_anchor=(1,0), bbox_transform=ax1.transAxes)
plt.savefig("%s.png"%(species_name), dpi=500, bbox_inches='tight', pad_inches=0.1)
plt.close()
species_comp = ''
for atom in species.composition:
species_comp += '{:<3}'.format(atom.upper())
species_comp += '{:<2}'.format(str(int(species.composition[atom])))
species_phase = 'G'
# Ref: Table 3. Summary of the Rules for Thermo Data
# https://shepherd.caltech.edu/EDL/PublicResources/sdt/formats/chemkin.html#:~:text=The%20Chemkin%20thermo%20file%20format,%2C%20Sandia%20Report%20SAND89%2D8009.
line_1 = (
'{:<18}'.format(species_name) +
'{:<6}'.format(' ') +
'{:<20}'.format(species_comp) +
'{:<1}'.format(species_phase) +
'{:<28}'.format(temp_range) +
'{:<5}'.format(' ') +
'{:<1}'.format(' ') +
'{:<1}'.format('1') +
'\n')
f.write(line_1)
line_2_coeffs = build_nasa(nasa_coeffs, 2)
line_2 = line_2_coeffs + ' 2\n'
f.write(line_2)
line_3_coeffs = build_nasa(nasa_coeffs, 3)
line_3 = line_3_coeffs + ' 3\n'
f.write(line_3)
line_4_coeffs = build_nasa(nasa_coeffs, 4)
line_4 = line_4_coeffs + ' 4\n'
f.write(line_4)
f.write('END\n')
if __name__ == '__main__':
mech='mech.cti'
gas = ct.Solution(mech)
thermo_fileName_new = 'thermoCorrected.dat'
print(thermo_fileName_new)
plot=True # True or False
write(gas, thermo_fileName_new, plot)