Advanced Analysis
The New Role for CFD in Industry
Canoga Park CFDPUB00485-1/mms
Advanced Analysis
The Engineering Challenge -
Reduce Development Cost without Compromising
Reliability
History 73% of program cost Where We Need to be
related to Test-Fail-Fix
cycle
“Concurrent Engineering &
Robust Design Practices” Reduce Development
• Streamline design & analysis processes Cost by a Factor of 8
• Identify all possible failure modes early
Target Goals
• Fully explore the design space
• Account for variabilities Reduce Development
• Quantify risks, sensitivities, margins, Time by a Factor of 4
system & component reliability
Canoga Park CFDPUB00485-2/mms
Advanced Analysis
In Concurrent Engineering & Robust Design Practices
CFD is one of the Key Enabling Technologies
• The new role comes with a price
• The “bar” has been raised - design tool vs color pictures
• Different environment
• Different customers
• Different expectations
• Different success criteria
• Requirements
• Ability to work real problems
• Engineering tool to be used by design engineers
• Turnaround times compatible with the design cycle
• Conceptual design (1-2 months)
• Preliminary design (4-6 months)
• Detail design (6-9 months)
• Quantified accuracy
• Acceptable cost
CFD tools have to function in this new environment
Canoga Park CFDPUB00485-3/mms
Advanced Analysis
Role Of CFD in the New Environment
• Expected to provide:
• Flow environment definition
• Performance assessment
• Structural and thermal load prediction (static & dynamic)
• Test guidance (facility, measurements, instrumentation, scaling)
• Performance code input (parameters, loss coefficients, shape factors)
• Approach - Use best tool available
• Multiple codes - general purpose (robustness), application customized (speed)
• Multiple providers - developed by or jointly with strategic partners, in-house,
commercial of-the-shelf, government
• Hierarchical physical models (turbulence/chemistry)
• Validated/calibrated/anchored (“degree of confidence”)
• Emphasis on - Provide engineering solutions and design guidance
• Use CFD in conjunction with engineering knowledge, other tools &
common sense - part of an overall integrated design & analysis system
• Support all phases of design - conceptual, preliminary, detailed (final)
• Deliver “value” within program budget and schedule constraints
Canoga Park CFDPUB00485-4/mms
Advanced Analysis
CFD Turnaround Time Requirement for 3D, Complex
Geometry, Complex Physics Analysis Results
First Time Through
First Time through CFD Analysis Results
Today
1 year --
Time from Geometry Definition to
Need to be by 2004
# of CFD Solutions/Day
6 mon. --
--3
1 mon. --
2 weeks --2
1 week --
2 days --1
1 day --
0.1
Year 94 96 98 00 02 04
Canoga Park CFDPUB00485-5/mms
Advanced Analysis
Use of CFD in Rocket Propulsion
System Development - Then & Now
Canoga Park CFDPUB00485-6/mms
Advanced Analysis
Typical Rocket Engine Components
Flow Devices (valves, manifolds, ducts)
Rotating Machinery
Thrust Chamber
Canoga Park CFDPUB00485-7/mms
Advanced Analysis
Thrust Chamber Components
Injector Main
Elements Injector
Combustion Nozzle
Chamber
Canoga Park CFDPUB00485-8/mms
Advanced Analysis
Then
(early eighties)
Canoga Park CFDPUB00485-9/mms
Advanced Analysis
First CFD Applications at Rocketdyne to Real
Hardware:
SSME Powerhead Flows (INS3D)
Canoga Park CFDPUB00485-10/mms
Advanced Analysis
Hot Gas Manifold
INS3D
• Detailed analysis of multiple designs (two-
& three-tube)
• Parametric analysis of turn-
around duct & fuel bowl contours
• Significant improvements quantified
• Verified with air flow tests
Canoga Park CFDPUB00485-11/mms
Advanced Analysis
Hot Gas Manifold
CFD Predictions Verified with Air Flow Tests
Air flow test results verify CFD-
based design improvements
Original Design
(FMOF)
Improved Design
FMOF
(T/A 16)
T/A 16
CFD predicts significantly reduced
pressure loss and more uniform flow in
redesigned manifold
Canoga Park CFDPUB00485-12/mms
Advanced Analysis
Main Injector
INS3D
SSME MI Two-Tube Design Analysis
SSME MI Single LOX
Post Analysis
• Main Injector inflow from CFD predicted
HGM/transfer duct analysis
• Simulated LOX post “core” via porous
media assumption
• Subsequent detailed analysis
about individual LOX posts
Canoga Park CFDPUB00485-13/mms
Advanced Analysis
HPOTP Ball Bearings
INS3D
• Investigate cause of ball bearing
discoloration after flight
• CFD analysis characterized heat transfer
around contact point where heat was
generated
Vorticity around Contact Point
Canoga Park CFDPUB00485-14/mms
Advanced Analysis
SSME Turbine Disk Cooling
REACT
• Exploring alternative HPOTP Temp
turbine blade cooling system
• Approximate & model complex
geometry (2-D)
• Understand flow environment
• Calculate thermal loads &
temperature distribution
• Suggest design changes
Canoga Park CFDPUB00485-15/mms
Advanced Analysis
CFD Used to Optimize High Performance Impeller
REACT
Canoga Park CFDPUB00485-16/mms
Advanced Analysis
Transient CFD Analysis Recommends Simple Fix to the
SSME Fuel Flowmeter Anomaly
REACT
f
1
f
2
Wake
location
Flow
Direction
• Abrupt shifting of flowmeter constant causes unreliable fuel
utilization reading
• Transient CFD analyses indicate
• Complex interaction between the flow straightener shed wakes and the
flowmeter rotor blades
• Anomaly is hydrodynamic in nature and not due to structural or duct vibration
• Origin of anomaly is unsteady forces imparted to the rotor at lower frequencies
than those experienced in bluff-body shedding (as was assumed previously)
• Simple fix is to move back the hexagonal straightener to weaken the
wake effects on the flowmeter blades (test and additional analysis
planned for confirmation)
Canoga Park CFDPUB00485-17/mms
Advanced Analysis
Now
(2001)
Canoga Park CFDPUB00485-18/mms
Advanced Analysis
Multiple CFD Codes
~
~
Production Codes New Codes
~
~
ENIGMA GALACSY
• General purpose, incompressible • Spray combustion code
• •
~
~
Navier-Stokes, steady/transient Navier-Stokes, steady/transient
• Finite difference, unstructured grid • Finite Volume, structured grid
• RANS turbulence model • Lagrangian/Eulerian
• Fixed and rotating reference frame • RANS turbulence models
• Customized version for FSI • Multiphase, multispecie, H2/HC finite-rate
chemistry
REACT TIDAL
• General purpose, low-speed code • General purpose, low- to high-speed code
• Navier-Stokes, steady/transient • Navier-Stokes, steady/transient
• Finite volume, structured grid • Finite volume, structured grid
• RANS turbulence models • TVD, shock capturing
• Fixed and rotating reference frame • RANS turbulence models
• Customized version for conjugate • Multispecie, H2/HC finite-rate chemistry
~
~
heat transfer • Multiphase (solid-gas)
USA ~
~ ICAT
• General purpose, high-speed code • General purpose, high-speed code
• Navier-Stokes, steady/transient • Navier-Stokes, steady/transient
• Finite volume, structured grid • Finite volume, unstructured grid
• TVD, shock capturing • TVD, shock capturing
• RANS turbulence models • RANS turbulence models
• Multispecie, H2/HC finite-rate chemistry • Multispecie, H2/HC finite-rate chemistry
Mach Number
Canoga Park CFDPUB00485-19/mms
Advanced Analysis
CFD Provides Flow Distribution in Turbine Discharge Duct
Enigma
• Flight discharge duct made compact
to save weight
• Creates highly nonuniform flow
Heat feeding into the heat exchanger (HEX)
Exchanger
• CFD predicted flow distribution
in the discharge duct and defined
the inflow to the HEX
• Predicted inflow conditions in HEX
design and analysis
• Resulted in a design that met
requirements for the compact duct
Canoga Park CFDPUB00485-20/mms
Advanced Analysis
Evaluation of Advanced Concepts
Enigma
Unshrouded Impeller - Design and Fuel and Oxidizer Valves -
Analyze Rotordynamics Analyze Showerhead Concepts
Short Length Jetpump - Design &
Analyze for Optimum Transfer
Efficiency - Validate Methodology
Tangential Force
25
Normal Force
20
Normalized Force
15
10
5
0
-2 -1 -5 0 1 2
-10
Whirling Ratio
Canoga Park CFDPUB00485-21/mms
Advanced Analysis
Preliminary Design and Redesign Studies
Enigma
Upper Stage Engine
Inducer+Kicker Design Inducer Back-Swirl
0.45
Head, water test (ft)
0.4
0.35
0.3
Upper Stage Engine Cross-over and Volute
0.25
0.2 Test 4-13-00 Ind+Kick. With 0.010 radial clearance
Design
INDANA prediction with Kicker, Radial Eq. Not conv.
0.15
Test 3-07-00 Ind+Kick. with .016 radial clearance
0.1 Head coefficient from cav. Tests w/0.010 rad.cl.
CFD Full-Scale (No Tip-Clearance)
0.05 CFD Water Test Model With 0.010" Tip Clearance
0 CFD Water Test Model With 0.016" Tip Clearance
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1 0.11 0.12
Flow Coefficient
Canoga Park CFDPUB00485-22/mms
Advanced Analysis
Future Design Environment
eTango
Integrated Design and Analysis Tool Consolidation
of Thirteen
Number of Codes
20 Codes
• eTango, An integrated centrifugal pump 15
design and analysis software package 10
• Runs on Windows based computers
• Incorporates Enigma™ CFD 5
Current Improved
Analysis Cycle Time,
100
Reduced Cycle Time
80
Hours
Templates
60
40 Training
20
Calendar Time
Canoga Park CFDPUB00485-23/mms
Advanced Analysis
Then, Now, and the Future (1/3)
Cycle Times
1988 2000 2005
Rotating Machinery Turnaround Time/ Turnaround Time/ Turnaround Time/
Labor (in hr.) Labor (in hr.) Labor (in hr.)
Turbines
Steady loads 160 / 120 20 / 16 4/4
Dynamic loads 320 / 240 40 / 20 8/6
Multistage No capability No capability 40 / 32
Pumps
Inducers No capability 60 / 40 8/4
Impellers 200 / 160 4/2 1/1
Diffusers/Crossovers 240 / 200 40 / 20 4/4
Volutes No capability 120 / 80 16 / 8
Bearings No capability 80 / 60 16 / 8
Complete Turbopump No capability No capability 160 / 160
Canoga Park CFDPUB00485-24/mms
Advanced Analysis
Then, Now, and the Future (2/3)
Cycle Times
1988 1998 2003
Thrust Chambers Turnaround Time/ Turnaround Time/ Turnaround Time/
Labor (in hr.) Labor (in hr.) Labor (in hr.)
Injectors
Gas/Gas 320 / 240 60 / 30 16 / 12
Liquid/Gas 480 / 360 120 / 80 24 / 16
Combustion Chambers
Flow environment 320 / 240 40 / 20 8/ 6
Thermal environment No capability 60 / 30 16 / 12
Combustion stability No capability No capability 120 / 60
Nozzle
Conventional 80 / 60 16 / 8 1/1
Aerospike No capability 80 / 40 8/6
Canoga Park CFDPUB00485-25/mms
Advanced Analysis
Then, Now, and the Future (3/3)
Cycle Times
1988 1998 2003
Flow Devices Turnaround Time/ Turnaround Time/ Turnaround Time/
Labor (in hr.) Labor (in hr.) Labor (in hr.)
Ducts 120 / 80 8/6 2/1
Manifolds 480 / 320 40 / 32 8/8
Valves No capability 80 / 60 20 / 12
Integrated Flow Path/ 1988 1998 2003
Turnaround Time/ Turnaround Time/ Turnaround Time/
Installed Performance Labor (in hr.) Labor (in hr.) Labor (in hr.)
VentureStar (RLV) No capability 160 / 120 40 / 40
NASP (hypersonics) No capability 240 / 120 40 / 40
Canoga Park CFDPUB00485-26/mms
Advanced Analysis
Believing the Predictions -
Validation/Certification/Calibration
Canoga Park CFDPUB00485-27/mms
Advanced Analysis
Lessons Learned on Validation/Calibration/Certification
• A general code validation procedure applicable for all codes and
applications is essential and can be developed
• Specific evaluation criteria are highly application dependent and it is
not possible to define a single general set of validation criteria
• Quantitative validation is only meaningful within limited classes of
applications
• The level of validation appropriate depends on the end application
• The validation process must be realistically achievable within the
engineering environment
Canoga Park CFDPUB00485-28/mms
Advanced Analysis
Code Validation is Essential for Engineering Design
• Validation is essential part of code development process
• Must be performed to ensure that analysis results are sufficiently reliable
and accurate for intended purposes
• Provides necessary confidence for code/analysis system to be used as
engineering tool
• Process offers means to quantify
• Code accuracy
• Code sensitivities
• Validation is a learning process
• Systematic approach to understand code capabilities and behavior
• Helps to identify code strengths and weaknesses
• Specifics of validation process depend on end application and intended
use of analysis results
Canoga Park CFDPUB00485-29/mms
Advanced Analysis
Two-Step Four Phase Validation Process
Step 1 - Select Flow Cases
Phase 1 Phase 2 Phase 3 Phase 4
Unit Problems Benchmark Cases Subsystem Cases Complete System
• Simple geometry • Special hardware • Subsystem or component hardware • Actual system hardware
• One element of complex flow physics • Two elements of complex flow physics • Moderately complex flow physics • Complete flow physics
• One relevant flow feature • Two relevant flow features • Multiple relevant flow features • All relevant flow features
• Experimental data with low • Experimental data with • Test data with moderate uncertainity • Limited test data with large
uncertainity or exact solution moderate/low uncertainity • Some IC’s and BC’s measured uncertainity
• All IC’s and BC’s measured or known • Most IC’s and BC’s measured • Most IC’s and BC’s unknown
Process Direction
Step 2 - Validate Code
Phase 1 Phase 2 Phase 3 Phase 4
Unit Problems Benchmark Cases Subsystem Cases Complete System
• Run unit problems • Run Benchmark Cases • Run simplified partial flow path • Run Actual Configuration
• Verify integrity • Assess Physical Models • Assess agreement with data • Compare With Test Data
• Assess accuracy, convergence, & • Establish Grid Distribution • Establish Grid Distribution
functionality Requirements Requirements
Process Direction
Process Direction
Canoga Park CFDPUB00485-30/mms
Advanced Analysis
Validation Requirements Depend on Intended Use of
Analysis Results
• Three design phases defined
• Conceptual - Initial definition concept layout
• Preliminary - Refined concept definition
• Detail - Final detailed design leading to hardware
• Different levels of code validation may be acceptable for each
design phase
• Conceptual
• Predict qualitative behavior of flow and parametric trends
• Phase 1 and 2 validation acceptable
• Preliminary - Refined concept definition
• Conceptual plus quantitative predictions (wider range of uncertainty)
• Phase 3 validation required
• Detail - Final detailed design leading to hardware
• Preliminary plus improved quantitative predictions (reduced range of
uncertainty)
• Phase 4 validation required
Canoga Park CFDPUB00485-31/mms
Advanced Analysis
Building Block Approach Uses Completed Validation
Cases for New Applications
Phase 1
Phase 2
Completed Validation
Flat Plate
Boundary Layer Phase 3
Curved Duct
(W/ W/o Rotation)
2D Duct
Diffuser
Boundary Layer Phase 4
Flow Over
2D Cylinder
Flow Over Airfoil Impeller
Impeller/Diffuser
Flat Plate Cascade
Annular Flow Rotor-Stator
Interaction
with Rotation Interaction
Turbine Cascade
(Rotation)
Acoustic Pulse Acoustics in
Complex Ducts
2D Acoustic Duct
Acoustic Duct
Validation Needed Turbine Blade
for New Application Cracking
Canoga Park CFDPUB00485-32/mms
Advanced Analysis
Looking into the Future -
CFD Technology Needs
Canoga Park CFDPUB00485-33/mms
Advanced Analysis
Technology Needs (1/5)
Preprocessing
• Defined as the process of going from CAD drawing to CFD geometry
model and eventually to CFD mesh
• Most time consuming (> 65%) and labor intensive (>70%) phase of
CFD analysis for most applications
• Can benefit from automating the mechanical portions of the process
through better links, scripts, and templates
Need
Reliable geometry repair tools, unstructured grid generators for
viscous flows & better coordination with solver developers for
dynamic grid adaptation capability
Canoga Park CFDPUB00485-34/mms
Advanced Analysis
Technology Needs (2/5)
Solvers
• Performance of next generation solvers being developed is critical for
CFD in industry
• Turnaround time has to be drastically reduced even though problems
to be analyzed will be much more difficult
• Solvers have to be compatible with the scalable heterogeneous
computing environment industry has adopted
Need
Solvers that can use both structured and unstructured
adaptive grids for steady-state and transient analysis
with a 100X improvement in computational time
Canoga Park CFDPUB00485-35/mms
Advanced Analysis
Technology Needs (3/5)
Physical Models (Turbulence, Chemistry, Transition)
• Turbulence modeling a major issue
• Workhorse models of the 1-, 2-equation RANS variety
• Better performance by higher order RANS models not yet fully
demonstrated on complex geometries
• Compressibility, heat transfer and transient flow issues not resolved
• Chemistry interaction modeling computationally very expensive
• For 100X solver speed-up, LES may be feasible for certain problems
• Chemistry modeling adequate for most applications
• Equilibrium and reduced kinetics models available for most fuels
• Reduced fast mechanisms needed for hydrocarbons
• Multiphase flow modeling still an art, but progress depends on
availability of fast solvers for model testing
• Transition modeling generally not an engine concern
(except for may be inlets)
Need
Robust and accurate turbulence models
Canoga Park CFDPUB00485-36/mms
Advanced Analysis
Technology Needs (4/5)
Postprocessing
• Defined as
• Diagnostic data interpretation
• Data reduction
• Graphics and visualization
• Data management and documentation
• Processing of the sheer size of data being generated already an
issue and will get worse (e.g. transient analysis)
Need
Software that can efficiently and accurately
access, reduce, manipulate, manage, and store data
in a multi-platform hardware environment
Canoga Park CFDPUB00485-37/mms
Advanced Analysis
Technology Needs (5/5)
Validation
• Lack of quality data for code validation biggest roadblock to more
extensive use of CFD
• The quality data can come from many sources
• Analytical solutions
• Very high fidelity simulations (e.g. DNS)
• Benchmark experiments
• Subcomponent tests (e.g. impeller)
• Component tests (e.g. turbopump)
• System tests (e.g. complete engine)
Need
High quality experimental data and databases for code & model
validation
Canoga Park CFDPUB00485-38/mms