Phase transformation time
The aggregated average error for simulation of the time in seconds on which transformation of the phases Ferrite, Pearlite, Bainite and Martensite are starting and finishing respectively. The benchmark is based on empirically measured TTT diagrams for 14 steel grades (chemical compositions listed below) that are not included in the machine learning database and where benchmarking TTT data has been extracted for randomly selected temperatures. In the table, _s and _f refer to the start and finish time of the transformation of a phase, e.g. Fe_s being Ferrite start time. TTT M_s model is temporary and will be improved in V 2.0 when consuming the main ferritico Ms model.
Ferritico error
Fe_s
Fe_f
P_s
P_f
B_s
B_f
M_s
3.57
3.06
3.06
4.69
3.55
4.10
17.37 °C
Competitor error
9.14
3.26
3.26
8.22
2.26
3.90
13.52 °C
STEEL GRADES INCLUDED IN THE BENCHMARKING
912,7 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,06
0,43
0
0
0
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
910,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,54
0,46
0
0
0
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
910,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,64
1,13
0
0
0
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
871,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,40
0,57
0
0
3,49
0,01
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
927,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,60
0,52
0
0
5,00
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
927,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,21
0,78
0
0,99
1,09
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
927,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,22
0,77
0
1,91
1,08
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
843,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,33
0,53
0
0,90
0
0,18
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
899,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,11
0,45
0
1,52
3,22
0
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
1.038,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,33
0,41
0
0
0
1,96
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
857,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,35
0,80
0
0
0
0,25
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
871,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,40
0,42
0
0
0
0,53
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
815,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,48
0,94
0
0
0
0,25
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
899,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,68
0,87
0
0
0
0,24
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
899,0 °C
Auestenitization temp
C
Mn
Si
Cr
Ni
Mo
V
Co
Al
W
0,11
0,38
0,44
5,46
0
0,42
0
0
0
0
Cu
Nb
Ti
N
B
P
S
Sn
Zr
As
0
0
0
0
0
0
0
0
0
0
TTT SIMULATION
Machine learning based prediction of isothermal transformation in low-alloy steels.
We are happy to provide the market with improved TTT simulation capabilities and to support additional TTT data use cases as the simulation accuracy has significantly improved compared to conventional tools.
The Ferritico TTT module is consumed as a SaaS where the TTT data is presented in a TTT diagram or in a aggregated data file.
Input:
-
Alloy composition
-
Austenitization temperature
Output:
-
What phases are formed for each temperature and their corresponding on- and offset time
-
Time for phase fractions (10%, 50%, 90%, 100%)
-
Hardness (HV) for each transformation temperature
-
AC1 and AC3
BENCHMARKING REPORT
The Ferritico TTT module simulation accuracy has been benchmarked through comparisons to empirical TTT measurements based on either Jominy or Dilatometer. The benchmark includes comparisons on phase formations and on- and offset time . The benchmarking report below also compares the Ferritico simulation accuracy with market leading simulation software built on physical models.
Phase formation
The aggregated average percentage error for simulation of whether the phases Ferrite, Pearlite, Bainite and Martensite are formed. The benchmark is based on empirically measured TTT diagrams for 14 steel grades (chemical compositions listed below) that are not included in the machine learning database and where benchmarking TTT data is extracted for randomly selected temperatures .
Ferritico %-error
Ferrite start
Pearlite start
Bainite start
Martensite start
5.07 %
4.91 %
7.60 %
28.85 %
Competitor %-error
23.04 %
31.13 %
4.17 %
0.00 %
Ferrite end
Pearlite end
Bainite end
Martensite end
4.91 %
3.37 %
12.75 %
NA
31.13 %
27.73 %
12.43 %
NA