Application of Thermo-Calc/DICTRA in
Physics-Based Models to Support
Materials by Design®
Herng-Jeng Jou (Joe)
[email protected], http://www.questek.com
Computational Thermodynamics & Kinetics Seminar
2011-04-06
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
2
Outline
• Intro: QuesTek and Materials by Design
• Mechanistic Modeling With ThermoCalc and DICTRA
• Implementation With ThermoCalc and DICTRA
• Materials by Design Calculation Examples
• Future Directions
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
3
QuesTek Innovations LLC
• Founded 1997; privately-held; located in Evanston, IL
• 19 Employees, including 15 Engineers (9 PhD’s)
• Rapidly designing, developing, qualifying and inserting new materials using
computational methods on integrated basis
• Creates IP; licenses to OEMs or alloy producers/processors
• 4 alloys licensed: Ferrium® M54™, C61™, C64™ and S53®
• ~10 major new alloys in development, 30+ patents awarded or pending worldwide
• Working with many colleagues in industry and academia
• Serving industry and government
• Recipient of many business and technology awards
• Expanding our staff
Core Competencies and Key Services
• Mechanistic Modeling
• Systems-based Computational Design
• Material Invention / Obtaining IP
• Validating Performance through
Detailed Characterization
• Guiding Scale-Up of Manufacturing
• Obtaining Industry Certifications
• Licensing IP to Commercial Producers or
OEMs
• Commercializing Materials with Licensees
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
4
We Create Diverse Material Systems
• Iron-based
• Aluminum-based
• Nickel-based
• Copper-based
• Niobium-based
• Titanium-based
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
5
Ferrium C61 Ring and Pinion
Gear Steel
SCORE Off-Road Racing
Circuit, Baja 1000, etc.
Ferrium S53 UHS
Corrosion Resistant Alloy
Aircraft Landing Gear
A-10 Main Landing Gear
A-10 Nose Gear
Examples of QuesTek’s Computationally Designed Alloys
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
6
Modeling for the Purpose of Materials Design
Process
Structure
Property
Performance
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
7
Integrated Computational Materials Design
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
8
Example Flow-Block Diagram for UHS Gear Steels
Cas
t
Ingo
tFo
rge
Mac
hin
e
Carburize/Solution Treatment
Quench/Controlled Cooling
Tempering/Coating
Pow
der
Met
allu
rgy
Case
Dispersion Gradient
Residual Stress
Core
Matrix - Lath MartensiteNi - Cleavage Resistance
Co - SRO Recovery Resistance
Strengthening Dispersion(Cr, Mo, V, W, Fe)2CX
Avoid Fe3C, M6C, M23C6
Grain Refining Dispersiond/f
Micro-void Nucleation Resistance
Grain Boundary ChemistryCohesion Enhancement
Impurity Gettering
MicrosegregationMo, Cr secondary dendrites
PROCESSING PROPERTIESSTRUCTURE
SurfaceHardness
StressDistribution
ThermalResistance
CoreHardness
Toughness
Modulus
PERFORMANCE
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
9
Physics-based
Parametric Models
Use of Models in Materials by Design
Empirical ModelsMechanistic
Models
• Explorative Evaluation
• Trade-off Analysis
• Robust Analysis/Design
• Sensitivity Analysis
• Accelerated Insertion
of Materials
Type of Calculations:
Model Improvement
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
10
Outline
• Intro: QuesTek and Materials by Design
• Mechanistic Modeling With ThermoCalc and DICTRA
– Martensite/Bainite Kinetics Modeling
– Precipitation Modeling
• Implementation With ThermoCalc and DICTRA
• Materials by Design Calculation Examples
• Future Directions
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
11
Martensite Models (Ghosh and Olson, 1994)
FCC
BCC
x
G
DGchem
Quasi-Binary
DGchem
Critical Energy, DGcrit=DGs(n)+Wf
ss(Xi)+Wf^(T)
DGs(n=18) : Surface Energy, n is Defect Potency Size
• Wfss(Xi) : Frictional Work by Solute
Strengthening• Wf
^(T) : Frictional Work by Dislocation Forest
Balance
DGchem (T=Ms) = DGcrit (T=MS)Solve for MS
TC
Ni’(n,D), Pre-existing Potency Distribution
nnm*
P’(n), Autocatalytic Potency Distribution
nnm*
DGchem =DGs(n*)+Wfss(Xi)+Wf
^(T)
Volume Fraction (fM) Evolution:dfM/dn = (Ni’(n,D)+P’(n)fM)(1-fM)V(f)
with fM (n=)=0where:V: Volume of Martensitic Lath Sub-unitD: Grain Size
Solve for fM(n=nm*) and find fM(T)
Activated
fM(T) Model
MS Model
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
12
DGchem (xa, T=BS) = DGcrit (xa, T=BS)Solve for Bs and xa
Bainite Models (Olson, Bhadeshia and Cohen, 1989, 1990)
FCC
BCC
xa x
G
DGchem
DGchem (with interstitial partitioning)
Critical Energy DGcrit=DGs(n)+Wf
ss(Xi)+Wf^(T)
FCC
BCC
xa x
G
xI
DGd
DGchemDGcrit
• Carbon Diffusion, Vd (xa, xI)• Interfacial Mobility• Solute Trapping, Vk (xa, xI)
Set of Non-Linear Equations:• Vn(n) = Vd = Vk
• DGchem = DGd+Wfss(Xi)+DGs (n)+Wf
^(T)
Solve for Vn (n), xa, and xI
xax
xI
BCC FCC
Vn
BS Model
fB,stasis(T) Model can be formulated similar to fM(T) model
Balance
Bainite Subunit Growth Rate Model
Used to estimate bainite start C-curves
TC
TC
DICTRA
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
13
Martensite-Bainite Model Validation
MS (°C)
BS (°C)
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
14
Precipitation Growth and Nucleation Models
a: matrixb: precipitateDGm: effective driving forceDGs: misfit strain energy
G- where
)(2
exp)(
41 :Growth
12
s
2
0
e
kjk
ji
T
i
mmj
ji
T
im
m
m
v
CDCC
GC
CCCC
GCG
R
VRG
RT
QMRsR
RπNR
dt
dR
m
D
D
DD
DD
D
a
bab
a
b
re temperatuis , whereradius critical is
constant sBoltzmann' is number, sAvogadro' is
4t impingemen atomic of rate
3
16 nucleus critical a form work to
4factor h Zeldovitcwhere
)( Rate Nucleation StateSteady
*
2
3*
2/1
422
2
0
*
*
T0dR/dtR
kN
G
V
NR
V
GW
TRkN
VZ
dRRJeV
NZJ
C
Ba
m
m
aC
m
m
R
CBa
m
SSTk
W
m
aSS
B
R
D
D
b
b
b
a
b
b
Growth Model
Nucleation Model
TC
DICTRA
PrecipiCalc
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
15
PrecipiCalc g’ Results for IN100 Disk Alloy
10-9
10-8
10-7
10-6
10-5
0 10000 20000 30000 40000 50000 60000 70000 80000
<D>1<D>2<D>3
Dia
me
ter
(m)
Time (sec.)
10-9
10-8
10-7
10-6
10-5
0 10000 20000 30000 40000 50000 60000 70000 80000
<D>1<D>2<D>3
Dia
me
ter
(m)
Time (sec.)
T(t)
PrimarySecondary
Tertiary
0
0.1
0.2
0.3
0.4
0.5
0 10000 20000 30000 40000 50000 60000 70000 80000
Dist 1Dist 2Dist 3
Vo
lum
e F
raction
Time (sec.)
0
0.1
0.2
0.3
0.4
0.5
0 10000 20000 30000 40000 50000 60000 70000 80000
Dist 1Dist 2Dist 3
Vo
lum
e F
raction
Time (sec.)
T(t)
Primary
Secondary
Tertiary
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
16
M2C Model for Secondary Hardening Martensitic Steels
Enhancements to PrecipiCalc
• BCC-Cementite Para-Equilibrium Condition
• Heterogeneous Nucleation on Dislocation with pipe diffusion
• Simple Coherency Transition Model With A Modified HCP Description with Micromechanical Elastic Coherency Energy Parameterized by Aspect Ratio and Misfit (Compositions effects implemented into a TC’s TDB)
• Simplified Cementite Dissolution
• Simplified Dislocation Recovery Dynamics Diameter,d3nm
Elastic Relaxation Factor b
Surface Energy
Aspect Ratio a
1
35
0.3
0
incoh
coh
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
17
AF1410 M2C Precipitation
Isothermal M2C precipitation simulation of AF1410 steel at 510°C, along with the experimental results from G.B. Olson, T.J. Kinkus and J.M. Montgomery, Surface Science 246(1992) 238. Parameters used in the simulation include: surface energy coh=0.25J/m2, incoh=1.0J/m2, r=2.5x1014 (m/m3), Dpipe/Dvol=100.
significant lower coarsening rate than t1/3
t1/3
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
18
Outline
• Intro: QuesTek and Materials by Design
• Mechanistic Modeling With ThermoCalc and DICTRA
• Implementation With ThermoCalc and DICTRA
• Materials by Design Calculation Examples
• Future Directions
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
19
QuesTek’s Proprietary Software Platform (CMD)
• Modularized
• Object-oriented (C++, Python)
• Standardized communication to
enable customized integration
Command LineInterface
• Standardized I/O command line interface
• iSIGHT Persistent and NV API’s• integrator API
TCIPC
Model Implementations
CMD Programs
GUI
CMDProgram
Integrator API
CMDProgram CMD
Program
iSIGHT, DAKOTA
ThermoCalc/DICTRA
CALPHAD Thermodynamics and Mobility Databases
TC-APINumerical Library
Python/Shell Scripts or Console
Model developers, materials designers and developers
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
20
CMD GUI — Pre-Processor
• Operates All CMD Programs• Selects Models and CALPHAD Databases• Sets up Explorative, Full Factorial DoE, Sensitivity and AIM Analysis• Rapid Visualization of Results
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
21
CMD GUI — Post-Processor
Results Manager
Additional External Data Manager
• Generates Explorative Plots with one or two variations
• Combines/Overlays Multiple Results• Adds External Data
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
22
CMD GUI — Sensitivity and AIM Analysis
compositions and model inputs
(such as HT conditions)
described as uncertainty
probability distribution
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
23
Approaches for Integrating Thermo-Calc/DICTRA
Software Type Speed DICTRAExternal
Codes
Access-
ibility
Multiple
Instances
Programming
Facility
Error
DetectAuthor
PARROT Macro Slow No No Partial NoLimited
(PARROT)Yes TC-AB
TCLib IPC Slow No Yes Full No Full (C) Yes TC-AB
TQ API Fast No Yes Partial No Full (F77) Yes TC-AB
TCIPC IPC Slow Yes Yes Full YesFull
(C/C++/Python)Some QuesTek
TC API API Fast Yes Yes Full No Full (C/C++) Yes TC-AB
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
24
Outline
• Intro: QuesTek and Materials by Design
• Mechanistic Modeling With ThermoCalc and DICTRA
• Implementation With ThermoCalc and DICTRA
• Materials by Design Calculation Examples
• Future Directions
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
25
Overall Compositions
User Input
O2Content
Surface Energy L12/LIQ and
liquid diffusion
Surface Energy FCC/LIQ
TL (FCC)
LIQ Compositions
w/o L12
L12 Particle Size
Distribution (PSD)
Corrected L12PSD
dT/dt or T(t)
Thermo-Calc
PrecipiCalcPrimary L12
PrecipiCalcInoculation
PSD WidthCorrection
Grain Size
Model Integration Example: Castable
High Strength AA7xxx
DICTRATC
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
26
Model Integration Example: Design of Low Cost Ti Casting Alloys
DICTRATC
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
27
Design Exploration and Tradeoff Analysis
Matrix + Strengthening Dispersion Design
Grain Pinning Dispersion Design
DICTRA TCTC
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
28
Composition Sampling(wt%, ±6):C ± 0.01 Cr ± 0.2 Mo± 0.1 W ± 0.1 Co ± 0.3 Ni ± 0.1 V ±0.02
CMD/
iSIGHT
Variations of:Structure — carbide solvus Ts, martensite
Ms, precipitation control DG’sProperty — hardness HRc, toughness CVN
1000 Monte Carlo runs (12 minutes on a Pentium IV 2.2GHz CPU)
Sensitivity Analysis on S53 Compositions
TC
DICTRA
TC
TC
TC
TC
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
29
Example of CMD/iSIGHT Robust Optimization
Original Compositions 3% Failure
Fail
Robust Optimization
New Compositions 0% Failure
TC
TC
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
30
S53 — Design For Scale
S53-3 Segregation Experience
300 lb 8” VAR ingot
S53A Segregation Experience
3000 lb 17” VAR ingot
12 hours
Solidification Simulation
Homogenization Simulation
24
DICTRA
DICTRA
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
31
Probabilistic Modeling of Manufacturing Variation:
Forecast of Minimum Design Properties
PrecipiCalc YS Model YS Distribution
Mechanistic simulation+ (n=15) gives good prediction of 1% minimum YS
Process VariationDICTRA
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
32
Accelerated Insertion of Materials (AIM)
Application Example: Ferrium S53®
Predicted A-basis minimum = 280 ksi UTS A-basis minimum: 280 ksi UTS
• AIM methodology has demonstrated reliable predictions for design minimums
• Allows designers to apply design models to estimate property variation prior to
full design allowable development
• Reduces costs and risks of material design and development
AIM prediction
Experimental Data
Pro
pert
y
Mean value
+3
-3
3 10AMS
Specification
MIL-HBK 5
“A”- Allowables
AIM Predictions
DICTRA
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
33
Outline
• Intro: QuesTek and Materials by Design
• Mechanistic Modeling With ThermoCalc and DICTRA
• Implementation With ThermoCalc and DICTRA
• Materials by Design Calculation Examples
• Future Directions
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
34
Future Directions
• Model uncertainty and prediction confidence level
• Integration with additional system design framework
and methods
• New computer hardware and software architectures
• Further integration with component-level process
simulations
• Robustness improvement
• Cross platform
Computational Thermodynamics & Kinetics Seminar2011-04-06
Materials By Design®
35
Computational Materials Qualification Acceleration
•MMPDS handbook update issued
•Additional property data developed
•10th multi-ton full-scale ingot produced
•Aerospace Materials Specification issued
•Static property data developed
•3rd multi-ton full-scale ingot produced
•1st multi-ton full-scale ingot produced
•Static properties demonstrated at prototype
•System design chart (design goals) establishedJan-99
Jan-00Jan-01Jan
-02Jan
-03Jan-04Jan-05Jan
-06Jan
-07Jan-08Jan-09Jan-10Jan-11Jan
-12Jan-13Jan-14
Maj
or
Mile
sto
nes
Dates
5 design
iterations
1 design
iteration
S53
8.5yrsM54
~6yrs
First S53 landing gear field
service test flight occurred
on Dec. 17th, 2010