Scanning Laser Epitaxy (SLE) – Technology for Engine Hot Section Components

Scanning Laser Epitaxy (SLE) – Technology for
Additive Repair and Manufacturing of Turbine
Engine Hot Section Components
Suman Das
Professor and Director
Direct Digital Manufacturing Laboratory
Woodruff School of Mechanical Engineering &
Georgia Tech Manufacturing Institute
Georgia Institute of Technology
Atlanta, Georgia
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
DIRECT PART MANUFACTURING DEFINING THEMES
IN GEORGIA TECH’s DDM LAB
Direct Part
Manufacturing and
Remanufacturing of
Turbine Engine
Components
Large Area Maskless
Photopolymerization (LAMP)
 Digital 3-D printing of ceramic
cores and integral cored molds.
 Disrupts conventional investment
casting.
 Foundry compatible.
 No change to conventional
metallurgy.
 Eliminates variation.
 Enables advanced cooling
designs.
 Enables breakthrough
efficiencies.
GEORGIA INSTITUTE OF TECHNOLOGY
Scanning Laser Epitaxy (SLE)
 Disruptive direct metal additive
manufacturing and repair
process.
 Processing of non-weldable,
and non-repairable hot-section
nickel-base superalloys.
 Repair and refurbish fielded
engine components to “as good
as new” or “better than new”.
 Produce functional fully dense,
3-D components from metal
powders.
 Enables surge production.
DIRECT DIGITAL MANUFACTURING LABORATORY
DIRECT PART MANUFACTURING PROJECTS & SPONSORS
Cyber-Enabled Direct Digital Manufacturing of Heterogeneous Multifunctional
Components (DDM of HMCs) – ONR
Program Manager: Dr. Khershed Cooper
Scanning Laser Epitaxy of Single-Crystal Nickel Superalloys – ONR
Program Managers: Dr. Khershed Cooper and Dr. Ralph Wachter
Direct Digital Laser Manufacturing of Nickel Superalloy Single-Crystal
Components – ONR
Program Managers: Dr. Ralph Wachter and Dr. Khershed Cooper
DURIP: High Power Fiber Laser for Direct Digital Manufacturing – ONR
Program Manager: Dr. Khershed Cooper
Direct Digital Manufacturing of Airfoils – DARPA, Defense Sciences Office
Program Manager: Dr. Bill Coblenz
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Additive Repair and Manufacturing of Hot Section
Turbine Engine Components Through
Scanning Laser Epitaxy (SLE)
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
TEAM
Dr. Khershed Cooper
Program Manager
Dr. Suman Das (PI)
Rohan Bansal
Ranadip Acharya
Justin J. Gambone
Paul Cilino
Office of Naval Research
CYBER ENABLED MANUFACTURING SYSTEMS (CeMS)
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Science and Technology Development Goals
• Develop thermally-controlled laser additive
manufacturing (AM) of nickel-base superalloys for
the repair of turbine engine hot-section hardware.
• Develop AM methods for nickel-base superalloy
components with predictable and controlled
microstructure directly from powder feedstock for
small lot functional prototyping and for surge
production.
• Investigate solidification control during laserbased additive manufacturing (AM) of nickel-base
superalloys and other high temperature materials,
particularly of hot-section alloys.
• Establish process physics understanding for
controlling the solidification microstructure as a
function of process parameters with predictive
models, sensors and real-time process control .
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Motivation – DoD Repair Needs
Applications: Repair Needs for In-service Superalloy Turbine Hardware
AV-8B Harrier
Rolls-Royce F-402 Pegasus Engine
CMSX-4 single crystal turbine blades
V-22 Osprey
Rolls-Royce AE1107 Engine
Hercules C-130J
Rolls-Royce AE2100 Engine
E-2C Hawkeye
Rolls-Royce T56 Engine
CMSX-4 single crystal turbine blades
GEORGIA INSTITUTE OF TECHNOLOGY
Apache AH-64
GE T700 Engine
DIRECT DIGITAL MANUFACTURING LABORATORY
Motivation – DoD Surge Production Needs
• Surge Production of turbine airfoils for long-range standoff
weapons (LRSO) through additive manufacturing.
• Engine OEM studies indicate up to 60% cost reduction
relative to investment casting in surge production scenario.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
SLE Application Development Approach
Additive Repair of Turbine Components
– Develop repair technology for nonweldable or non-joinable superalloys
including equiaxed, DS and SX.
– Repair of manufacturing defects on
surfaces: Porosity, hot tearing, inclusions,
misruns.
– Repair of aging stationary and rotating
engine components.
Additive Manufacturing of Turbine Components
– Apply the learnings from repair to metal-powder
bed-based layer-by-layer 3-D additive
manufacturing.
– Produce fully dense, defect free components
without tooling or post-processing.
– Process advanced material systems considered
extremely difficult or expensive conventionally.
– Produce functionally graded components that are
currently difficult to manufacture.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Processing Challenges for Hot Section Alloys
• Solidification Cracking/Hot tearing.
• Grain Boundary Liquation Cracking.
• Strain Age Cracking.
• Recrystallization and Loss of
Underlying Microstructure.
• Preserving nominal alloy
composition.
• Tramp elements in alloy
composition.
• Oxidic cleanliness and impact on
properties.
• Powder contamination.
• Unmelted powder particles
trapped in the melt.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Processing Challenges for Hot Section Alloys
Area of interest
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Scanning Laser Epitaxy (SLE)
•Process enabling direct additive
manufacturing of three-dimensional
objects with specified microstructure.
•Based on controlled melting and resolidification of a pre-placed metal powder
on a base substrate of same or similar
alloy composition.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Scanning Laser Epitaxy (SLE)
•
•
•
Tightly focused scanning laser’s rapid movement induces surface tension gradients
and incorporates powder into the melt pool
Partial remelting of underlying substrate to initiate nucleation off prior grains.
Epitaxial solidification through heterogeneous nucleation.
Galvanometer
scanners
Stroboscopic video
microscope
Fiber laser beam
Laser beams
Laser Head
Reflected
radiation
Focusing optic
Pyrometer
Power
control
signal
Xenon strobe
Analog voltage
output
Data
acquisition
Processing and controller
Probe beams
Laser
Power supply
Solidification microstructure control as a
function of laser processing parameters
and thermal boundary conditions on
deposition substrate.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
TECHNICAL APPROACH
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
SLE ALLOY PROCESS DEVELOPMENT





•
Demonstrate SLE of Ti-6Al-4V
Demonstrate SLE of Equiaxed MARM-247
Demonstrate SLE of DS MARM-247
Demonstrate SLE of SX CMSX-4
Demonstrate SLE of Rene 80
SLE Development of Other Equiaxed, DS and SX Alloys
• IN 100
• GTD 111
• PWA 1480
• PWA 1484 SX
• IN 738
• IN 738LC
• CM247LC
• Rene 108 EQ
• Rene 108 DS
• Rene 220
• TiAl
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
SCANNING LASER EPITAXY OF MARM-247
SOME EXPERIMENTAL RESULTS
BB
10mm
MARM-247 deposited on Cast MARM-247 plate
AA
Fully dense, crack-free and pore free microstructure of MARM-247 deposit
• Fully dense deposits produced
• No cracking upon stress relief heat treatment at 1067 C
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
SCANNING LASER EPITAXY OF MARM-247
SOME EXPERIMENTAL RESULTS
BB
CC
AA
MARM 247 deposit
1 mm
MARM 247 deposit
Bond line
Bond line
MARM 247 plate substrate
GEORGIA INSTITUTE OF TECHNOLOGY
MARM 247 plate substrate
DIRECT DIGITAL MANUFACTURING LABORATORY
SCANNING LASER EPITAXY OF MARM-247 and CMSX-4
SOME EXPERIMENTAL RESULTS
AA
2 mm
CMSX-4 deposited on Cast CMSX-4 plate
AA
Fully dense, crack-free and pore free single-crystal microstructure of CMSX-4 deposit
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
SCANNING LASER EPITAXY OF CMSX-4
EBSD CHARACTERIZATION RESULTS
AA
Upper Figure: Optical micrograph along cross-section AA of CMSX-4
deposit
Lower Figure: EBSD orientation color map showing growth orientation of
deposited material in good agreement with parent structure.
AA
2 mm
GEORGIA INSTITUTE OF TECHNOLOGY
Top View of CMSX-4
deposit on CMSX-4
subatrate. Section AA is
shown in red.
Inverse Pole
color map
Figure
DIRECT DIGITAL MANUFACTURING LABORATORY
SCANNING LASER EPITAXY OF RENE 80
SOME EXPERIMENTAL RESULTS
Fully dense, crack-free and pore free microstructure of Rene 80 deposited
on Rene 80 plate, as processed
Fully dense, crack-free and pore free microstructure of Rene 80 deposited
on Rene 80 plate, post heat treatment
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Process Modeling
Key Modeling Needs
•
•
•
•
•
•
•
•
•
Accurate thermal modeling and melt depth prediction to better
control the process.
Microstructure evolution modeling as a function of processing
parameters and solidification conditions.
Prediction of the onset of microstructural transformations including
CET and OMT as a function of processing parameters for DS and
SX alloys.
Modeling of thermal stresses.
Hot tearing criteria to predict crack formation during processing.
Modeling of phase transformations during solidification.
Development of microstructure-processing-property maps.
Models for accurate modeling of process physics.
Models for real-time process control.
DIRECT DIGITAL MANUFACTURING LABORATORY
GEORGIA INSTITUTE OF TECHNOLOGY
Review of Prior Modeling Work
• Rosenthal Solution was used by several researchers.
• FEM approach was used to solve the conduction based model for
LENS, Laser Cladding processes.
• CFD simulation is used to incorporate melt pool convection as well as
Marangoni convection.
Gaps Identified
• Simulations in prior work are all based on a single-bead approach and
lack the generalization required to incorporate the raster scanning and
associated microstructure modeling in the SLE melt pool.
• Prior works do not incorporate local property variations with
temperature. These can change the simulation results significantly
while providing greater accuracy.
• Time tracking of the melt pool position is required for raster scan
approach
M. Gaumann, et. Al., Acat Materialia 49, pp.1051
Childs, T.H.C. et al, CIRP Annals - Manufacturing Technology, 2004. 53(1): p. 191-194.
Liu, W. and J.N. DuPont, Acta Materialia, 2004. 52(16): p. 4833-4847.
DIRECT DIGITAL MANUFACTURING LABORATORY
GEORGIA INSTITUTE OF TECHNOLOGY
SLE Model Schematic
Prior work: Single Bead ellipsoidal Profile
Our work: SLE Model schematic showing powder
placed on the top of a substrate and the laser
beam rastering normal to the plane and stepping to
the right.
• Melt pool modeled as an ellipsoid.
• Analytically normal component of solidliquid interface at different location can be
calculated based on position vector.
• Not suitable for the evaluation of the
geometry parameters related to the melt
pool in a raster scan pattern.
Line source
motion
Instantaneous
Laser position
Liu, W. and J.N. DuPont, Acta Materialia, 2004. 52(16): p. 4833-4847.
Our work: SLE simulated line heat
source melt pool profile
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Overall SLE Modeling Methodology
Immediate Consolidation Approach (CFX) Slow Consolidation Approach (FVM)
Thermal
Conduction
o/p
Boundary
Convection
Property=
f(T) [*]
Microstructure
prediction
o/p
Saunder’s
Database
Melting and
Resolidification
i/p
Sintering Model
(Level Set)
Powder bed
Radiation (Ray
Tracing)
i/p
Scan
parameters
Radiation
Scan
parameters
Conjugate Heat
Transfer
o/p
o/p
Stress
Analysis
Substrate
Temperature
Distribution
Fluid Convection
Marangoni effect
* Property values for CMSX-4 taken from - Matsushita et. al.,
Journal of Chemical & Engineering Data, 2009. 54 (9): p
2584-2592
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Results - Melt Depth Prediction
Representative operational parameters used
Sample
Powder Thickness (mm)
Power (W)
Scan Speed (mm/s)
Number of Repeat Scan
Sample 1
Sample 2
Sample 3
1
1
1.5
480
510
584
200
200
250
100
50
50
Sample 1
Sample 2
Sample 3
Animation of the transient temperature
distribution and evaluation of the solid-liquid
interface, and melt depth in SLE (click on
the play button at bottom of picture)
Plot of simulated melt depth (shown in red line) superimposed on the actual micrograph
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Microstructure Prediction
Oriented-to-Misoriented Transition (OMT) Prediction
Oriented-to-Misoriented transition occurs when the SX columnar orientation flips
from 001 to 100 at some height above the substrate-deposit interface plane.
Contour of orientation vector - Blue contour level
indicates desired (001) orientation while red contour
level indicates (100) orientation or possible OMT.
The intermediate plane (dark blue) indicates the
current laser position.
DIRECT DIGITAL MANUFACTURING LABORATORY
GEORGIA INSTITUTE OF TECHNOLOGY
Results – OMT Prediction
Orientation vector plot showing OMT for two different operating parameters
Observation:
1. The orientation vector for (001) is assigned a value=1. The same is assigned a
value of 10 for (010) and a value of 100 for (100). A value of zero indicates
conflicting columnar orientation possibility.
2. Green contour level indicates desired (001) orientation while Orange contour
level indicates (100) orientation or possible OMT.
DIRECT DIGITAL MANUFACTURING LABORATORY
GEORGIA INSTITUTE OF TECHNOLOGY
Microstructure Prediction
Columnar-to-Equiaxed Transition (CET) Prediction
1.
2.
3.
4.
5.
6.
The Temperature gradient at the solid-liquid interface
(G) is evaluated from CFD model.
The solid-liquid interface orientation(θi) is evaluated
from θi =cos-1(Gi/G) (for i=x, y, z).
The growth rate of the solid-liquid interface is given by
V= S cos θ (S= scan speed or Line source speed).
the temperature gradient parallel to the dendrite
growth direction, Ghkl, is calculated using the equation
Ghkl =G/cos ;
The dendrite growth velocity, Vhkl, is calculated using
the equation V hkl = Scosθ/ cos .
Rappaz Modification to predict CET:
n= material constant(= 3.4 for CMSX-4), =
equiaxed fraction( Critical value = 0.066%),
N0= Nucleation density, ∆Ttip= Tip
undercooling, ∆Tn= Nucleation undercooling
7.
A lower value on the left hand side increases φ for a
given nucleation density. Hence, a lower value of this
ratio indicates likeliest location of columnar to
equiaxed transition.
DIRECT DIGITAL MANUFACTURING LABORATORY
Contour of CET probability criterion
The higher the value of Gn/V, the higher
the probability of obtaining SX
microstructure, while lower values indicate
the probable locations of the CET. The
intermediate plane (dark blue) indicates
the current laser position.
GEORGIA INSTITUTE OF TECHNOLOGY
Results – CET Prediction
(a1)
(a2)
(b1)
(b2)
CET criterion plot for same sample at (a1) heat source position 1= 7.9 mm, (b1) heat source position 2= 10.1 mm,
(a2) and (b2) same with Marangoni effect
Observation:
1. Time tracking is important!!
2. The CET criterion value varies with laser scan position. Marangoni convection reduces this value!!
3. From the contour plot, position 1 is likely to show a higher value of columnar height.
DIRECT DIGITAL MANUFACTURING LABORATORY
GEORGIA INSTITUTE OF TECHNOLOGY
CET – Experimental Observations
Experimentally observed CET at heat source position 1= 7.9 mm and (b) heat
source position 2= 10.1 mm
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
CET – Experimental Observations
Experimentally observed CET at heat source position 1= 7.9 mm and (b) heat
source position 2= 10.1 mm
Observation:
1. Position 1 shows CET at higher z-value.
2. Good agreement with the CET criterion value deduced from thermal and
microstructure modeling.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Flow Field
(a)
(b)
(a) Velocity vector observed in the liquid region and (b) Temperature map of the
velocity vector in the melt pool.
Observation:
1. Flow field consists of two rotational vortices, with zero velocity at center and
scan source separating them from each other.
2. Loose powder at lower temperature is drawn from the front at some distance
beneath the top surface.
3. The higher temperature melt pool expands at the top surface.
4. Natural convection observed at lower velocity with high temperature current
going upward and then following the scan source.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Digital Micrograph Image Analysis-Based Automated
Microstructural Feature Tracking
Single Crystal
Deposit Height
Total Deposit
Height
Meltback Depth
Sample
Deformation
Main microstructure features to be tracked in every sample for CMSX-4 (top
and left) and Rene-80 (right).
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Digital Micrograph Image Analysis-Based Automated
Microstructural Feature Tracking
Original Substrate Location
Bond Line Tracking
• Bond line tracking relative to the
original substrate location (top)
allows for the meltback depth
across the sample to be found.
• Detailed dendrite tracking (left)
allow for the [001] SX width, angle
and termination point to be found
across the sample length. 34
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Digital Micrograph Image Analysis-Based Automated
Microstructural Feature Tracking
Bond Line Tracking
• Bond line tracking in Rene-80
samples (top) allows for the
meltback depth across the sample
to be found.
• Pore tracking (left) allow for the
location, degree of circularity and
size of each pore to be saved.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Digital Micrograph Image Analysis-Based Automated
Microstructural Feature Tracking
SX Termination Line
• The SX termination line shows the point where the [001]
single crystal microstructure ends in the sample.
• Single crystal buildup is necessary for CMSX-4 samples.
• Rene-80 and Marm 247 are concerned with the total deposit
height.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Thermal Imaging-based Real-time Feedback Control
Model-reference Adaptive Control Scheme (MRACS)
• Nearly all thermophysical and optical properties vary with
temperature, density and phase.
• Thermal imaging-based absolute temperature itself is inaccurate
due to unknown emissivity.
• Simple PID control alone cannot account for variations in these
properties during processing.
• Using the MRACS scheme, the laser power and/or the scan speed
will be adjusted in real-time to regulate the melt pool temperature.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Thermal Imaging-based Melt Pool Tracking
• Use thermal camera to determine average melt
pool temperature and width for use in feedback
control
• Thermal camera frame rate is 50-60fps and
ideally would not be reduced significantly by
tracking algorithm
• Implemented algorithm based on Canny edge
detection takes 1-2ms and reduces frame rate to
40-45fps
• Algorithm is not yet fully optimized and running
on old hardware, leaving room for improvement
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Thermal Imaging-based Melt Pool Tracking
• Use thermal camera to determine average melt pool temperature and width.
• Thermal camera frame rate is 50-60fps and ideally would not be reduced significantly
by tracking algorithm.
1.) Acquire thermal image
2.) Convert to gray scale
3.) Apply Gaussian blur
4.) Binary threshold
5.) Canny edge detection
6.) Find bounding box
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
CONCLUSIONS
• Fully dense, defect-free, porosity free deposits have been produced in wellknown hot section superalloys MARM-247, Rene 80, CMSX-4, .
• Epitaxial solidification and microstructural continuity has been
demonstrated on investment cast substrates of same alloy chemistry.
• A comprehensive multiphysics modeling framework for the processing of
alloy powders through SLE has been developed.
• The predictive capability of the modeling framework has been validated for
melting, solidification and microstructural evolution in single-crystal alloy
CMSX-4 processed through SLE.
• A quantitative metallography-based computational tool has been developed
for microstructural characterization and DoE optimization.
• Thermal-imaging based real-time adaptive control scheme should provide a
tight control of the melt pool temperature, temperature gradients and
cooling rates.
• Coupled with detailed and approximate models, the real-time control
scheme will enable layer-by-layer control of melting and microstructure
uniformity, as well as the ability to produce functionally graded parts.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
SLE FUTURE APPLICATIONS AND IMPACT
• Monolithic metal components of extreme geometric complexity
and high efficiency that cannot be conventionally
manufactured.
– Turbine airfoils with non-castable complex cooling passages.
• Monolithic metal components with dual or functionally graded
microstructure and properties.
– Next-generation functionally graded turbine airfoils with spatially tailored
single-crystal, directionally solidified, and equiaxed microstructure with
locally optimized properties.
• Metal components with functionally graded alloy composition
and properties.
– Functionally graded turbine disk
• Metal components with functionally graded distribution of
ceramic nanoparticles
– Oxide dispersion strengthened alloys
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
DoD Relevance
• Repair and restoration to flightworthy condition of high value
legacy components made of non-weldable and non-joinable
alloys using same or similar alloy powders.
• Directly digital manufacturing or remanufacturing of improved
versions of various “in service” or legacy components from
digital design/inspection data and feedstock powders.
• Reduction in acquisition/repair costs and savings in lead time
through the elimination of tooling, and reductions in processing
steps, material waste, and energy consumption.
• Future enabling technology for manufacturing next-generation,
higher performance components of weapons systems
exploiting advanced materials and microstructural
configurations.
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY
Thank you
• Questions?
43
GEORGIA INSTITUTE OF TECHNOLOGY
DIRECT DIGITAL MANUFACTURING LABORATORY