Boeing VG Template

1
The New Role for CFD in Industry
2
The Engineering Challenge - Reduce Development
Cost without Compromising Reliability
Where We Need to be
History
73 of program cost related to Test-Fail-Fix cycle
New Processes
Concurrent Engineering Robust Design
Practices
Reduce Development Cost by a Factor of 8

• Streamline design analysis processes
• Identify all possible failure modes early
• Fully explore the design space
• Account for variabilities
• Quantify risks, sensitivities, margins,
• system component reliability

Reduce Development Time by a Factor of 4
3
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
4
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/chemist
  ry)
• 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

5
CFD Turnaround Time Requirement for 3D, Complex
Geometry, Complex Physics Analysis Results First
Time Through
Today
1 year --
Need to be by 2004
6 mon. --
--3
of CFD Solutions/Day
1 mon. --
Time from Geometry Definition to First Time
through CFD Analysis Results
--2
2 weeks
1 week --
--1
2 days
1 day --
0.1
04
Year
98
94
96
00
02
6
Use of CFD in Rocket Propulsion System
Development - Then Now
7
Typical Rocket Engine Components
Flow Devices (valves, manifolds, ducts)
Rotating Machinery

Thrust Chamber
8
Thrust Chamber Components
Main Injector
Injector Elements
Nozzle
Combustion Chamber
9
Then (early eighties)
10
First CFD Applications at Rocketdyne to Real
Hardware SSME Powerhead Flows (INS3D)
11
Hot Gas ManifoldINS3D

• 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

12
Hot Gas ManifoldCFD Predictions Verified with
Air Flow Tests
Air flow test results verify CFD-based design
improvements
13
Main InjectorINS3D

• 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

14
HPOTP Ball BearingsINS3D

• Investigate cause of ball bearing discoloration
  after flight
• CFD analysis characterized heat transfer around
  contact point where heat was generated

15
SSME Turbine Disk CoolingREACT

• Exploring alternative HPOTP
• turbine blade cooling system
• Approximate model complex
• geometry (2-D)
• Understand flow environment
• Calculate thermal loads
• temperature distribution
• Suggest design changes

Temp
16
CFD Used to Optimize High Performance
ImpellerREACT
17
Transient CFD Analysis Recommends Simple Fix to
the SSME Fuel Flowmeter Anomaly REACT
Wake
location

• 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)

18
Now (2001)
19
Multiple CFD Codes
Production Codes
New Codes


Physics

• GALACSY
• Spray combustion code
• Navier-Stokes, steady/transient
• Finite Volume, structured grid
• Lagrangian/Eulerian
• RANS turbulence models
• Multiphase, multispecie, H2/HC finite-rate
  chemistry

• ENIGMA
• General purpose, incompressible
• Navier-Stokes, steady/transient
• Finite difference, unstructured grid
• RANS turbulence model
• Fixed and rotating reference frame
• Customized version for FSI

Chemistry

• REACT
• General purpose, low-speed code
• Navier-Stokes, steady/transient
• Finite volume, structured grid
• RANS turbulence models
• Fixed and rotating reference frame
• Customized version for conjugate heat transfer

• TIDAL
• General purpose, low- to high-speed code
• Navier-Stokes, steady/transient
• Finite volume, structured grid
• TVD, shock capturing
• RANS turbulence models
• Multispecie, H2/HC finite-rate chemistry
• Multiphase (solid-gas)

Turbulence

• ICAT
• General purpose, high-speed code
• Navier-Stokes, steady/transient
• Finite volume, unstructured grid
• TVD, shock capturing
• RANS turbulence models
• Multispecie, H2/HC finite-rate chemistry

• USA
• General purpose, high-speed code
• Navier-Stokes, steady/transient
• Finite volume, structured grid
• TVD, shock capturing
• RANS turbulence models
• Multispecie, H2/HC finite-rate chemistry

Mach Number
20
CFD Provides Flow Distribution in Turbine
Discharge Duct Enigma

• Flight discharge duct made compact
• to save weight
• Creates highly nonuniform flow
• feeding into the heat exchanger (HEX)
• 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

Heat Exchanger
21
Evaluation of Advanced Concepts Enigma
Fuel and Oxidizer Valves - Analyze Showerhead
Concepts
Unshrouded Impeller - Design and Analyze
Rotordynamics
Short Length Jetpump - Design Analyze for
Optimum Transfer Efficiency - Validate
Methodology
22
Preliminary Design and Redesign StudiesEnigma
Upper Stage Engine InducerKicker Design
Inducer Back-Swirl
Upper Stage Engine Cross-over and Volute Design
23
Future Design Environment eTango
Integrated Design and Analysis Tool
Consolidation of Thirteen Codes

• eTango, An integrated centrifugal pump
• design and analysis software package
• Runs on Windows based computers
• Incorporates Enigma CFD

Reduced Cycle Time
24
Then, Now, and the Future (1/3) Cycle Times
1988
2000
2005
Rotating Machinery
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Turbines Steady loads Dynamic loads Multistage
160 / 120 320 / 240 No capability
20 / 16 40 / 20 No capability
4 / 4 8 / 6 40 / 32
Pumps Inducers Impellers Diffusers/Crossovers
Volutes Bearings
No capability 200 / 160 240 / 200 No
capability No capability
60 / 40 4 / 2 40 / 20 120 / 80 80 / 60
8 / 4 1 / 1 4 / 4 16 / 8 16 / 8
Complete Turbopump
No capability
No capability
160 / 160
25
Then, Now, and the Future (2/3) Cycle Times
1988
1998
2003
Thrust Chambers
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Injectors Gas/Gas Liquid/Gas
60 / 30 120 / 80
16 / 12 24 / 16
320 / 240 480 / 360
Combustion Chambers Flow environment Thermal
environment Combustion stability
320 / 240 No capability No capability
40 / 20 60 / 30 No capability
8/ 6 16 / 12 120 / 60
Nozzle Conventional Aerospike
1 / 1 8 / 6
16 / 8 80 / 40
80 / 60 No capability
26
Then, Now, and the Future (3/3) Cycle Times
1988
1998
2003
Flow Devices
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Ducts
120 / 80
8 / 6
2 / 1
Manifolds
40 / 32
8 / 8
480 / 320
Valves
No capability
80 / 60
20 / 12
1988
1998
Integrated Flow Path/ Installed Performance
2003
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
Turnaround Time/ Labor (in hr.)
VentureStar (RLV)
No capability
160 / 120
40 / 40
NASP (hypersonics)
40 / 40
240 / 120
No capability
27
Believing the Predictions - Validation/Certificati
on/Calibration
28
Lessons Learned on Validation/Calibration/Certific
ation

• 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

29
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

30
Two-Step Four Phase Validation Process
Step 1 - Select Flow Cases
Phase 1
Phase 2
Phase 3
Phase 4
Benchmark Cases
Complete System
Subsystem Cases
Unit Problems

• Subsystem or component hardware
• Moderately complex flow physics
• Multiple relevant flow features
• Test data with moderate uncertainity
• Some ICs and BCs measured

• Special hardware
• Two elements of complex flow physics
• Two relevant flow features
• Experimental data with
• moderate/low uncertainity
• Most ICs and BCs measured

• Simple geometry
• One element of complex flow physics
• One relevant flow feature
• Experimental data with low
• uncertainity or exact solution
• All ICs and BCs measured or known

• Actual system hardware
• Complete flow physics
• All relevant flow features
• Limited test data with large
• uncertainity
• Most ICs and BCs unknown

Process Direction
Step 2 - Validate Code
Phase 2
Phase 1
Phase 3
Phase 4
Benchmark Cases
Complete System
Subsystem Cases
Unit Problems
Run simplified partial flow path
Run unit problems
Run Benchmark Cases
Run Actual Configuration
Assess Physical Models
Compare With Test Data
Assess agreement with data
Verify integrity
Assess accuracy, convergence,
functionality
Establish Grid Distribution Requirements
Establish Grid Distribution Requirements
Process Direction
Process Direction
31
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

32
Building Block Approach Uses Completed Validation
Cases for New Applications
Phase 1
Completed Validation
Phase 2
Phase 3
2D Duct Boundary Layer
Phase 4
Flow Over Airfoil
Impeller
Impeller/Diffuser Interaction
Annular Flow with Rotation
Rotor-Stator Interaction
Acoustics in Complex Ducts
Acoustic Pulse
Acoustic Duct
Validation Needed for New Application
Turbine Blade Cracking
33
Looking into the Future - CFD Technology Needs
34
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 (gt 65) and labor intensive
  (gt70) 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
35
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
36
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
37
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
38
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