MDO Investigation of

1

• MDO Investigation of
• Advanced Design Concepts
• Applied to the
• Blended Wing-Body Configuration
• Twelve Month Progress Review
• NASA Langley Research Center, Grant NAG 1-02024
• January 22, 2003
• Hampton, VA

2
Our Team

• Students
• Vance Dippold, III
• Leifur Leifsson
• Andy Ko
• Serhat Hosder
• Faculty
• B. Grossman (NIA)
• R.T. Haftka (University of Florida)
• W.H. Mason
• J.A. Schetz

Meeting weekly via telecon with web
presentations and archiving
3
Overview and Tasks
The Job Use MDO to integrate advanced
propulsion concepts that address the problems of
emissions (including hydrogen) and noise Use an
advanced concept as the baseline the BWB

• Task 1 Models for Distributed Propulsion
• Task 2 MDO with Distributed Propulsion and Noise
• Task 3 Noise

4
In the Initial 6 month effort

• Developed an MDO methodology BWB and benchmarked
  for a conventional BWB (1994/96)
• Fuel volume/tank weight issue of Hydrogen fuel
• Structural weight of pressurized cabin
• Looked for a noise model for conceptual design
• Chose Lilleys model
• Developed an approach for distributed propulsion
• Developed baseline data for hydrogen propulsion

5
In recent 6 months

• Developed an understanding of aero-propulsion
  integration for distributed propulsion systems
• Developed and implemented an approximate model
  for distributed propulsion in our BWB methodology
  based on jet flap theory
• Our understanding of distributed propulsion
  concepts and associated computational issues
  matured.
• We obtained quantitative insight into noise and
  distributed propulsion using CFD

6
Reconsidered Propulsion Concept
Original concept
Engines
Internal Ducts
NASA Concept (Mechanical Engineering, Nov. 2002)
Current Thinking
7
Presentation Outline

• Introduction - Bill Mason
• Distributed Propulsion BWB MDO Program
• Andy Ko
• Analysis and Modeling of Distributed Propulsion
  and Noise for MDO
• Vance Dippold
• Parametric Noise and Distributed Propulsion
  Studies with GASP
• Serhat Hosder
• Future Work
• Joe Schetz

8
Distributed Propulsion BWBMDO Program

• Andy Ko

9
Configuration details

• Reasonable number of engines distributed along
  the span
• Considering buried engines with boundary layer
  inlets.
• Part of engine exhaust is ducted to exit the
  trailing edge
• Amount that is ducted can be controlled to
  optimize performance and noise
• Jet Wing
• Could extend along entire span
• Longitudinal control by changing the jet angle

10
MDO Program Flowchart
11
Validation Mission

• 800 Passengers
• Based on 1994 report mission to validate
• Mission parameters can be changed for further
  studies

12
Distributed Propulsion BWB MDO

• Objective function
• Different objective functions can be used
• Takeoff Gross Weight
• Noise
• Objective functions can also be specified as
  constraints
• Design variables
• A total of 21 design variables available for
  preliminary code
• Geometric Properties (Chord, t/c, Sweep)
• Fuel Weight
• Average cruise altitude
• Thrust
• Parameters
• Various parameters to allow for different
  configurations include
• Mission parameters
• Engine parameters (number of engines, position of
  engines)
• Distributed propulsion specific parameters
• Duct weight factor
• Propulsive efficiency factor
• Duct efficiency

13
MDO Constraints

• Constraints in place
• Fuel volume
• Balanced Field Length
• Landing distance
• Second segment climb gradient
• Missed approach climb gradient
• Approach velocity
• Cabin planform area
• Cabin height
• Top of climb rate of climb
• Stability Control
• Conventional elevons
• Distributed Propulsion Jet deflection
• Geometric
• Simulated cabin egress constraint

• In progress
• Noise (can also be specified as an objective
  function)
• Fuel volume for hydrogen fuel

14
Distributed Propulsion BWB Program Summary

• Framework for conventional BWB completed
• Used as a comparator
• Optimized results have been verified with 1994
  BWB design
• Distributed Propulsion BWB
• Modifications to conventional BWB code
• Distributed propulsion effects
• Control by using jet wing deflection angle
• Ducting issues
• Aerodynamic effects
• Jet coefficient is an important parameter
• CJ Jet Thrust / (qSref)
• Jet flap theory as a model for distributed
  propulsion
• CFD used to verify and adjust analysis methods
• Following discussion will focus on distributed
  propulsion analysis methods

15
Effect of Distributed Propulsion on Propulsive
Efficiency
16
Froude Propulsion Efficiency

• To access the benefits of distributed propulsion,
  we will look at the limiting cases and represent
  the savings in terms of Froude Propulsion
  Efficiency
• The Froude Propulsion Efficiency is defined as
• Ratio of useful power out to the rate of kinetic
  energy added to the flow
• A typical value of the Froude Propulsion
  Efficiency for a high bypass ratio turbofan at
  Mach 0.85 is 80.

17
Efficiency and Engine SFC

• The increase in Froude Propulsive Efficiency is
  reflected in the reduction in specific fuel
  consumption (SFC) of the engine
• The required equations are
• Therefore,

18
2-D non-lifting case

• Lower limit
• Body and engine separate
• Engine jet has no influence in body wake
• We will assume 80 efficiency for high bypass
  ratio turbofan
• Upper limit
• Engine jet perfectly fills in the wake
• Therefore efficiency 100

19
Reality

• In reality, it would be difficult to perfectly
  fill in the wake with the engine jet
• Therefore, only a fraction the maximum savings
  due to filling in the wake is attainable.
• But, what fraction of that savings?
• We will perform a parametric study to see the
  effect of this fraction of savings on aircraft
  design.
• CFD results will provide better understanding and
  level of savings

20
Lifting case

• At low Mach numbers, we now have both viscous
  drag and induced drag
• If the induced drag viscous drag (which is the
  case at maximum L/D)
• Perfect scenario
• Part of thrust is diverted to exhaust out of
  trailing edge
• Thrust out of body to overcome viscous drag
  perfectly fill in the wake
• Remaining thrust to overcome induced drag
• Therefore, the system has an average of 90
  efficiency
• 100 for thrust out of body, 80 for remaining
  thrust
• The limits are now 80 - 90 efficiency
• This example shows that the limits on the
  efficiency depends on the ratio of viscous drag
  to the total drag

21
Impact of filling in wake

• First cut analysis
• Applied to a conventional transport aircraft
• Boeing 777 class aircraft
• Part of thrust diverted to trailing edge of wings
  and fuselage.
• Results
• Savings in SFC of almost 10 and takeoff gross
  weight of 6 for a perfect distributed propulsion
  system
• Savings in SFC of 7 and takeoff gross weight of
  3.7 if only 60 of the savings due to a perfect
  distributed propulsion system can be achieved.
• Although only a first cut analysis, results are
  encouraging

22
Longitudinal control with jet wingSpences Jet
Flap Theory extension
23
Implementing Spences Jet Flap Theory

• Purpose
• To replace elevon control by changing jet
  deflection angle of the distributed propulsions
  system
• Spences Jet Flap theory
• Provides analysis method to estimate 3D CL
• Modified to account for sweep
• Analysis method for 2D Cm
• Extended to obtain approximation to 3D Cm
• Function of CJ and jet deflection angle
• Tested with several geometries to demonstrate
  validity
• Results
• For CJ 0.03 (typical value at cruise)
• Jet wing deflection provides sufficient control
  for the BWB
• To be adjusted with CFD results

24
Issues

• A typical CJ value at approach condition is 0.2
• Compared to CJ 0.03 at cruise
• This is because the dynamic pressure at 110 knots
  (M0.166) at SL is much smaller than that at Mach
  0.85 at 35000 ft.
• Therefore, for the same thrust, we have more
  control at approach than at cruise
• Cruise condition then might be the limiting
  condition
• However,
• Should only the jet be used for control?
• What happens at engine out?
• Or, diverter/ducting system fails?
• Perhaps a variable-angle jet flap would be better
• Combination of a mechanical flap and a pure jet
  flap
• Prevents the complete loss of lift control in
  case of total failure of the jet-flap blowing
  system.

Ref Gainer, Long, Volger, Comparison of
Aerodynamic Theory and Experiment for Jet-Flap
Wings
25
Other distributed propulsion effects
26
Distributed propulsion Induced Drag

• Induced Drag
• From Spences Jet Flap Theory,
• Therefore,
• However, at cruise
• Typical values of CJ 0.03
• Therefore, savings in induced drag is small

27
Ducting losses

• A penalty would be applied to the thrust of the
  aircraft due to ducting losses
• Since a portion of the engine exhaust will be
  ducted for the jet wing, the loss in thrust will
  be a function of the jet wing thrust

Ratio determining jet wing thrust
Useful thrust
Thrust out of engine
Duct efficiency
28
Weights Other issues

• Structures (Wing weight)
• The wing weight in FLOPS allows up to 8 engines
  on the wing
• For more than 8 engines, equivalent 8 engines
  will be used.
• Duct weight
• Could not obtain information or methods to
  estimate duct weights
• To simulate, a factor is applied to the
  propulsion weight
• Also accounts for associated systems and vectored
  thrust mechanism
• What about effects of jet wing on drag of
  aircraft?
• Increase in drag is accounted as loss in thrust
• Throttle dependant drag is accounted to thrust
• All the effects have been put together in the BWB
  program
• Initial results

29
Initial Results
Distributed Propulsion BWB
Conventional BWB
30
Initial Distributed Propulsion Results

• Duct Weight Factor
• Factor applied to engine weight to simulate duct
  weight
• Duct efficiency
• Factor that is applied to the thrust to account
  for losses in ducts
• Distributed propulsion factor
• Fraction of savings in propulsive efficiency that
  can be achieved by filling in the wake

31
Status Continuing work

• We have
• An MDO program for a conventional BWB for a
  comparator
• An MDO program for a distributed propulsion BWB
• Simplified analysis methods to be modified with
  CFD results
• Continuing work
• Implement Lilleys noise formula for wings
• Correlation for clean wings No jet wing effects
• Optimize BWB for noise
• As CFD results become available,
• Will be implemented for higher fidelity
  aerodynamic data
• Will provide more accurate noise modeling
• Hydrogen fuel

32
Continuing work Hydrogen Propulsion

• We have all the pieces
• Engine data received from Mark Guynn
• Fuel tank/volume data from literature search

33

• Detailed Distributed Propulsion Modeling
• Vance Dippold, III

34
Detailed Distributed Propulsion Approach

• Our Motivation
• Simplified Distributed Propulsion models were
  developed for MDO
• Produce detailed models of Distributed Propulsion
  and Noise to help validate MDO models
• Review Distributed Propulsion Approach Evolution
• Engine Modeling
• Embedded Region
• Jet Wing
• Representative 2D test case
• Airfoil selection
• Induced drag estimation
• Jet Wing calculations
• Current Status

35
Distributed Propulsion Approach
Conventional Aircraft with Distributed Propulsion
(NASA Aeronautics Blueprint, 2002)

• Use actuator volume to model propulsor
• ANSYS Flotran includes a Force Field model
• Solve entire wing in ANSYS Flotran
• ANSYS Flotran is a Finite Element N-S code

36
Early Problems

• Encountered problems
• Trouble running transonic cases
• Lack of shock formation
• BL and wall functions
• Must pick grid to match turbulence model
• First grid point at y15
• Cycle pick grid, run, adjust grid, run, etc

37
Embedded ANSYS Region Approach

• Use Embedded Region approach
• Use quick, proven tools for general airfoil flow
• Locally solve around Distributed Propulsion
  system
• Reduce time cost
• Solve Upstream Region
• MSES with BL interaction p(x) and inviscid
  inflow to embedded region
• Virginia Tech BL codes BL part of inflow to
  embedded region
• Solve Embedded Region containing Modeled Engine
• Use ANSYS Flotran with Force Field
• Iterate using flap approximation for aft engine
  effects in Upstream Region

38
Revised Distributed Propulsion Approach

• MDO and more thought lead to Jet Wing concept for
  Distributed Propulsion
• Inject fluid/exhaust from wing TE
• Model jet by applying BCs to wing TE
• ANSYS force field not necessary
• Use same Embedded Region solution approach

39
Representative 2D Model

• Select 2D Airfoil based on wing section at 62
  span
• Chord 22.2 ft
• CL2D 0.69
• M2D 0.72
• Re 35 Million
• Modified SC(2)-0410
• Reduced aft loading
• MSES results
• CL0.690
• CD0.00725

Ref Boeing 1996 BWB report
40
Account for Induced Drag

• MSES predicts pressure drag, viscous drag on
  airfoil
• Must estimate local induced drag to determine
  accurate engine thrust
• Used idrag
• Local induced drag very small at representative
  wing section

41
Jet Wing Calculations

• Thrust based on pressure and viscous drag
• Calculate momentum coefficient for jet-wing at
  cruise conditions
• Jameson, Attinello significant increase in CL
  when CJO(1.45)
• CJ small at cruise assume very small effects of
  jet on CL first-iteration results may be very
  close to actual
• Duct height on 2D representative airfoil
• Chord 22.2 ft
• TE thickness 1.3 in
• Can accommodate duct height 0.5 to 0.9 in

42
Problems with MSES/ANSYS

• Data extraction from MSES
• MSES outputs Ue and surface Cp and graphically
  displays flowfield
• New code added to output all flowfield data, but
  inconsistent with native data output
• Little user support found
• Running ANSYS Embedded Region (with approx input
  from MSES)
• ANSYS does not consistently run with small
  changes in grid
• Airfoil TE pressure distribution consistently
  differs from MSES solution by 10-15
• Little user support found
• Trying to use 2 Commercial Software packages in a
  non-standard ways ???
• Reconsider this Approach

43

• Parametric Noise and Distributed Propulsion
  Studies With GASP
• Serhat Hosder

44
Introduction

• Parametric noise studies with CFD for the MDO
  model
• Objective to minimize the airframe noise
• Lilleys Approach to airframe noise and the role
  of maximum turbulent kinetic energy (max. TKE)
• To formulate the design problem, we need to know
  how max. TKE changes with the design variables
• 2-D Airfoil studies on TKE
• 3-D Wing studies on TKE
• CFD studies for the distributed propulsion
• 2-D airfoil calculations with jet injection at
  the trailing edge

45
Lilleys Approach to Airframe Noise

• Turbulent flow near the trailing edge on upper
  surface of the wing is critical for noise
• Equation for noise intensity I (Lilley, 2001)

(Lilley, 2001)

• Note that
• Maximum TKE at the trailing edge as a conceptual
  design noise metric

46
2-D airfoil studies on TKE

• Parametric studies on the max. TKE at the
  trailing edge (T.E.) of airfoils
• Used NASA Supercritical airfoils SC(2)-0706,
  SC(2)-0710, and SC(2)-0714
• Same family of airfoils with different thickness
  ratios
• Investigations performed to determine
• The effect of lift coefficient (CL) on the max.
  TKE at the T.E
• Effect of thickness on the max. TKE at the T.E
• Used Menters TKE-? turbulence model in GASP

47
SC(2) airfoils with different max. (t/c)
SC(2)-0714
SC(2)-0710
SC(2)-0706

• Runs performed for different CL at Mach0.75,
  Rec6.2?106
• Design CL0.7 for all airfoils
• Trailing edges closed for the calculations

48
Thickness and CL effect on max.TKE
SC(2)-0714
SC(2)-0710
SC(2)-0706

• At the same CL, max. TKE gets larger with the
  increase of thickness
• For each airfoil, max. TKE vs. CL behavior
  similar to drag rise
• Max. TKE gets larger when the flow is close to
  separation

49
Max. TKE, CP, and CF comparisons at the same CL
50
3-D Wing Studies on TKE

• Baseline Geometry ONERA M6 Wing
• Wilcoxs 1998 TKE-? turbulence model
• Used wall functions in the turbulence model
• Grid wall spacing from a published grid for ONERA
  M6 wing
• For the noise minimization problem
• Different wing design variables including twist,
  sweep, (t/c), etc.
• We already know the effect of (t/c) and CL on
  max. TKE from 2-D airfoil studies
• For the initial study, change the twist
  distribution and keep the other variables fixed.
• Mach0.84, Rec11.72?106 for different CL
• Compare section Cl, spanload, and max. TKE
  distributions along the span

51
Max. TKE distribution along span
Wing without twist
Wing with twist
filled symbols upper surface unfilled symbols
lower surface
52
Jet wing studies with CFD

• Modified SC(2)-0410 airfoil with blunt trailing
  edge
• Airfoil thickness 10
• Trailing edge thickness0.49 of the chord
• Used Menters TKE-? turbulence model
• Performed runs at Mach0.721, a2.66 deg.,
  Rec38.4?106
• Without jet
• With jet injected from trailing edge
• Jet angle same as the angle of attack

53
Grid Used in jet wing calculations
Zone 1
Zone 2

• Zone 1 (464x64 cells), wraps the airfoil and
  extends to outflow boundary
• Zone 2 (84x40 cells), in the wake cut, extending
  to outflow boundary
• Jet boundary conditions imposed at the trailing
  edge

54
CL, CD, and Cp comparisons
Shock moves upstream with jet injection,
changing CL
55
Velocity profiles in the wake
56
Flowfield in the T.E. region
Velocity vectors in the T.E. region obtained for
the case with subsonic jet
Streamlines of the T.E circulation region
obtained for the case with no jet
57
Future Work

• Noise study
• Complete parametric studies on max. TKE
• Formulate the design problem for noise
  minimization
• Create response surfaces from the CFD results
• Test the design approach on a single wing
  geometry
• Implement design approach to BWB MDO framework
• Distributed propulsion study for BWB
• Continue 2-D and 3-D parametric studies on jet
  wing
• Refine theoretical approaches used in MDO process
• Noise and TKE investigation for the distributed
  propulsion

58
Future Plans

• Task 1 Models for Distributed Propulsion
• Insight through more 2-D/3-D CFD studies
• Formulation to include in MDO
• Task 2 MDO with Distributed Propulsion And Noise
• MDO studies with distributed propulsion
• MDO with Hydrogen for conventional and
  distributed propulsion
• Integrate refined and higher fidelity models into
  MDO
• Task 3 Noise
• Noise minimization alone on clean wing model
  problem
• Formulation for MDO
• Investigate validation of Lilleys method as a
  surrogate for more complete noise calculation