computational sciences center high-speed flow … · integrity ! service ! excellence computational...

28
Integrity Service Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace Engineer AFRL/RQHF Air Force Research Laboratory

Upload: lamthuan

Post on 05-May-2018

215 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Integrity « Service « Excellence

Computational Sciences Center High-Speed Flow Research

2014

Dr. Jonathan Poggie Senior Aerospace Engineer

AFRL/RQHF Air Force Research Laboratory

Page 2: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Hypersonic Technology

High-­‐Speed  Strike  Weapon  

Scramjet  Mach  5-­‐8  

Space  Access  

Rocket  Mach  20+  

Penetra6ng  Regional  ISR  /  Strike  

Scramjet  Mach  ≈5  

Tac6cal  Boost-­‐Glide  /    Prompt  Global  Strike  

Rocket  Mach  =  10-­‐20  

2  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 3: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

High-Speed Flow Research Team Computational Sciences Center

•  Government Employees –  Dr. Nicholas Bisek –  Dr. Ryan Gosse –  Mr. Eswar Josyula –  Dr. Joel Malo-Molina –  Dr. Jonathan Poggie

•  Basic Research Contractors –  Dr. Jon Burt –  Dr. Timothy Leger

•  Application Support Contactors –  Mr. D. Galbraith –  Mr. W. Humphrey

•  AFRL Partners –  Dr. Mark Hagenmaier (scramjet flows) –  Dr. Roger Kimmel (transition, SWBLI unsteadiness, flight tests)

•  CCAS Partners –  Dr. Datta Gaitonde (FDL3DI code, SWBLI unsteadiness, BL trips) –  Dr. Graham Candler (US3D code, thermophysics)

3  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 4: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Main Research Themes

1.  LES and DES as engineering tools 2.  Fatigue loading 3.  Inlet flows 4.  Selected topics in aerothermal prediction

4  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 5: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

1. LES and DES as Engineering Tools

•  Large-scale computations •  Computational methodology •  Verification and validation

5  Gosse,  US3D,  HIFiRE-­‐6   Sherer,  HIFiRE-­‐6  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 6: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

DoD HPCMP Frontier Project

6  

Year   CPU-­‐Hrs  (millions)  

2014   90  

2015   150  

2016   211  

2017   890*  

2018   890*  

Title:  Unsteady  Pressure  and  HeaPng  Environment  on  High-­‐Speed  Vehicles  with  Responding  Structures  PI:  Dr.  Ryan  Gosse,  AFRL/RQHV  

•  MulP-­‐disciplinary  simulaPons  •  Disparate  space  and  Pme  scales  •  UPlize  significant  fracPon  of  machine  

(10k  –  100k  cores)  •  Develop  procedures  for  large  jobs,  

large  data  

Garnet  /  ERDC:  150k  cores  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 7: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Hybrid Parallelism

7  

•  Exploit  fine-­‐grained  parallelism  with  vectorizaPon  and  threads  (OpenMP)  

•  Support  domain  decomposiPon  with  MPI  

•  Minimum  block  size  set  by  compact  difference  numerics  

•  Use  threads  to  scale  to  more  cores  at  same  accuracy  

Scaling  with  Blocks  and  Threads  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 8: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Turbulent Boundary Layer Flow

8  

Density contours, ρ/ρ∞

M  =  2.3,  2.9  Reθi  =  2000,  2500  106  to  109  cells  Up  to  32k  cores  Up  to  8  threads  Poggie  et  al.,  AIAA  Paper  2014-­‐0423  

Poggie,  AIAA  2014-­‐xxxx  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 9: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Comparison to Experiment: Boundary Layer Profiles

9  

Mean  Profiles   Streamwise  Fluctua6ons  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 10: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Comparison to Experiment: Spatial Correlations

10  

Data

Computation

Thus, the pseudo-intermittency can be expressed as:

c ! I1 " I y# $I1 " I2

! "#5$

Figure 7 shows the pseudo-intermittency profile cal-culated in this manner. An intermittency profile (Seliget al. 1989) derived using a VITA-based threshold tech-nique from hotwire data obtained in the same boundarylayer flow is included for comparison. The shape of theprofiles is quite similar, but they are offset by about 0.3d.The offset may well be due to differences in the thresholdfor defining turbulent fluid.

4.1.2CorrelationsA spatial correlation function was calculated over a set ofimages between the scattering intensity recorded at eachpixel in the field of view and the scattering intensity at areference location. The correlation coefficient was definedby the following formula:

R x; r# $ !PN

i!1 Ii x# $ " !I% & Ii x' r# $ " !I% &#########################################################################PNi!1 Ii x# $ " !I% &2

PNi!1 Ii x' r# $ " !I% &2

q #6$

where I is the intensity of scattered light recorded by thevideo camera, x is a reference location, and r is a relativedisplacement vector. This type of correlation mimics thecross-correlation between a pair of point probes at zerotime delay.

Figure 8 shows contour plots of the two-point cross-correlation for several heights in the boundary layer. Thedomain corresponds to the central third of the field of viewin Fig. 4. A set of N=400 images was used to calculate thecorrelation in each case. The correlation contours are seento be approximately elliptical in shape. Large regions ofhigh correlation are seen near the middle of the boundarylayer (Fig. 8b, c). In contrast, lower correlation levels andshorter length scales occur close to the freestream(Fig. 8a), where only the tips of the bulges reach, and nearthe wall (Fig. 8d), where fingers of freestream fluid rarelyreach.

Another feature of the plots is the changing inclinationof the principal axis of the ellipse-like correlation con-tours. This angle (h) was measured graphically from thecorrelation plots like the examples in Fig. 8, and is plottedin Fig. 9 as a function of height in the boundary layer. Theorientation of the principal axis varies from horizontal atthe mean boundary layer edge to about 45! at 0.6d, whereit stays constant until scatter is introduced due to lack ofresolution in the flow visualization near the wall.

Quite similar structure angles were observed (Spinaet al. 1991b) in space–time correlations of hotwire massflux data for small wire separations. For a wire separationof 0.09d, a structure angle of 45!–50! was obtained in therange 0.2 £ y/d £ 0.8 (Fig. 10).

A similar correlation analysis has been done of theplan-view images (Poggie 1991), but is omitted here for thesake of brevity. The results showed elliptical correlationcontours, elongated in the streamwise direction, withprincipal axes on the order of 1.0d and 0.5d at y/d=0.5.These results are also consistent with hotwire space–timecorrelations (Spina et al. 1991b).

Fig. 7. Pseudo-intermittency profiles through the Mach 2.9turbulent boundary layer. Solid line is image data, open circles ishotwire data from Selig et al. (1989)

Fig. 8. Two-point cross-corre-lations of side-view images ofthe turbulent boundary layer.Horizontal line indicates posi-tion of wall

444

D. Smith et al., AIAA 91-0651 Poggie et al., Experiments in Fluids, 2004

Flow

Experiment

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 11: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

kz

+

E(k

z)

10-3

10-2

10-1

10-9

10-7

10-5

10-3

10-1

101

Lz/!

0 = 5

Lz/!

0 = 10

Lz/!

0 = 20

Boundary Layer Fluctuations: Effect of Domain Width

Density  20  δ0  

10  δ0  

5  δ0  

Mass  Flux  (ρu)’  z-­‐Wavenumber  Spectrum  

Increasing  Lz  

x/δ0  =  100  y/δ  =  0.5  

Downstream  View  y-­‐z-­‐plane,  x/δ0  =  100  

ResoluPon:  Δx+  =  6  Δy+  =  0.5-­‐9  Δz+  =  5  Δt+  =  0.05  

11  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 12: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

2. Fatigue Loading

12  

Shock Interaction on Flat Panel

•  Fatigue loading from unsteady SWBLI very serious design concern

•  Typical for Mach 5 Cruise: 147 dB, 800 K (Zuchowski, AFRL TR, 2012)

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 13: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Separation Bubble as Amplifier

13  

AAmplifier

Input:Turbulence

!(f)Filter

Output:Shock Motion

•  Plotkin  (AIAA  J,  1975)  •  Linearly-­‐damped  Brownian  

moPon  •  Time  scales:  τu  <<  τR  •  Predicts  spectrum,  

autocorrelaPon,  fluctuaPon  intensity  

•  Frequency-­‐selecPve  amplifier  

•  Cut-­‐off  frequency  set  by  separaPon  bubble  characterisPcs  

•  Input  set  by  TBL  t/! R

Rp(t

)

0 2 4 6 8 10

0

0.2

0.4

0.6

0.8

1

2!f" R

Gp(f

)/("

R#

p2)

fGp(f

)/#

p2

10-2

10-1

100

101

102

10-4

10-3

10-2

10-1

100

101

0

0.1

0.2

0.3

0.4

0.5

Spectrum   AutocorrelaPon  Wall  Pressure  Sta6s6cs  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 14: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Flight Test Experiments

HIFiRE-1

•  HIFiRE Flight 1: Woomera Range, March 22, 2010 •  Documentation: Stanfield, Kimmel, and Adamczak,

AIAA Papers (2012-2013)

14  

Kulite Transducer Stations

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 15: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

2!f"R

G(f

)/("

R#

p2)

10-2

10-1

100

101

102

103

10-4

10-3

10-2

10-1

100

101

102

x/$0 = 67.0x/$0 = 68.0x/$0 = 69.0Plotkin (1975)

Wall Pressure Spectra

Computational Results Experimental Results

Various configurations, M = 2-5 Compression ramp, M = 2.3, 24 deg

HIFiRE-1 Flight Data Large-Scale

Shock Motion

Boundary Layer Turbulence

15  

LES captures large-scale unsteadiness J.  Poggie,  N.  J.  Bisek,  R.  L.  Kimmel,  and  S.  A.  Stanfield,  “Spectral  CharacterisPcs  of  SeparaPon  Shock  Unsteadiness,”  AIAA  Journal,  in  press,  2014.    

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 16: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Effect of Control

Reduced: •  Size of separation bubble •  Energy in low-frequency wall pressure fluctuations •  Fatigue loading

Baseline Case Control Case PSD of TKE

0.001 0.01 0.1 1 1010

!5

10!4

10!3

10!2

10!1

100

PowerSpectralDensityofTKE

Normalized Frequency, f l/U!

baselineperfectpulsedrealBaseline

Control

24° Ramp Mach 2.3 Reθ = 2000

16  

Bisek, Rizzetta, and Poggie, AIAA Journal, 2013

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 17: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Fluid / Hot Structure Interaction

Pressure  

Density  

•   Non-­‐equilibrium  LES  with  moving  grid  •   Non-­‐linear  structural  response  

Supersonic  Flow  over  Deflected  Panel  Large-­‐Scale  Separa6on  Shock  Oscilla6on  Drives  Structural  Dynamics  

M  =  5  

Panel  Responds  

Data:  Maestrello  and  Linden  (1970)  

R.  Gosse  US3D  

17  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 18: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

3. Inlet Flows

•  Corner flow interactions •  RANS modeling •  LES

18  

R.  Gosse,  US3D  Code,  HIFiRE-­‐6  

X-­‐51  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 19: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Corner Interactions in Inlet Flows

19  

x=0  

x=-­‐1/2  

x=0  

x=-­‐1/2  

Inviscid  flow

 incident  sho

ck    

impingement  lin

e  

Benek,  Suchyta,  and  Babinsky  (2014)  RANS  SimulaPons  (Overflow,  non-­‐eq.  k-­‐ω)  M  =  2.9,  α  =  13°,  W/H  =  1,  δ/W  =  0.07  

Mean  Flow  Streamlines  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 20: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Effect of Boundary Layer Blockage on Separation Scale

20  

0.00  

1.00  

2.00  

3.00  

4.00  

5.00  

6.00  

0.0000   0.0500   0.1000   0.1500   0.2000   0.2500   0.3000  

Δx/fδ  

δ/gW  

Scaled  M=2.9  W=13  

M=2.9  W=10  

M=2.9  W=8  

M=2.5  W=13  

M=2.5  W=10  

M=2.5  W=8  

M=2.9  W=13  

M=2.9  W=10  

M=2.9  W=8  

M=2.5  W=13  

M=2.5  W=10  

M=2.5  W=8  

Δx DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 21: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

LES for Corner Flows

M    =  2.3,  Reθ  =  2400  HFILES,  FDL3DI  Code  1.06x109  cells  13240  processors  ~2  weeks  

N.  Bisek,  AIAA  Paper  2014-­‐0558  

Mach  2.3  Turbulent  Corner  Flow  

21  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 22: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Corner-Ramp Configuration

22  

M    =  2.3,  Reθ  =  2400  HFILES,  FDL3DI  Code  750  Mcells  47040  processors  (5880  tasks  x  8  threads)  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 23: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

4. Selected Topics in Aerothermal Prediction

Pressure  Contours  RANS  for  3D  Interac6ons  23°  Sharp  Fin  M  =  5.0,  Reθ  =  7400  DLR  Experiments:  Schülein  et  al.  (2001)  

US3D  Code  

Streamwise  Profile  of  Wall  Heat  Flux  

23  

Leger  and  Poggie,  AIAA  2014-­‐0951  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 24: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Direct Numerical Simulation of Laminar-Turbulent Transition

24  

Wall  Heat  Flux  

Trip  

“Pizza-­‐Box”  Trip  

Flight  Test  ArPcle  

Trip  

Cost/run:  3200  processors  for  120  hours  

HIFiRE-­‐1  Boundary  Layer  Trip  Experiment  

Gronvall,  Bisek,  and  Poggie,  AIAA  Paper  2014-­‐0433  Kimmel  and  Adamczak,  AIAA  Paper  2011-­‐3413  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 25: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

State-to-State Kinetics: Influence on Wall Heat Flux

25  

( )[ ] ciciccici VVvt

ωρρ =++•∇+∂

∂ ~

DVEMDTDHC qqqqq +++=

Ø  Master equation for V-V/V-T ü 48 quantum levels for nitrogen"ü FHO rates"ü Self-diffusion in vibrational states"

Ø Transport coeffs from Wang-Chang Uhlenbeck eqn (Boltzmann equation with internal energy) Ø  Contributions to heat transfer: heat conductivity (HC), thermal diffusion (TD), mass diffusion (MD), and diffusion of vibrational energy (DVE)

Contribution of Heat Conduction to Total Heat Flux

Josyula  et  al,  2014  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 26: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Unsteady Scramjet Flameholder

26  Instantaneous  fuel  mole  fracPon  for  56SLPM  and  99SLPM  cases  

Mean  fuel  mole  fracPon  for  56SLPM  and  99SLPM  cases  

Streamwise  velocity  profiles  for  RANS  (red)  hybrid  RANS  and  LES  (black)  and  experiment  (symbols)  

• Mach 2 cavity flameholder simulated using hybrid RANS/LES (US3D)

• Compared to PIV data

• Characterization of turbulence and mixing in cavity for non-reacting cases

Numerical  InvesPgaPon  of  a  Supersonic  Cavity  Flameholder,  David  M.  Peterson,  Ezeldin  A.  Hassan,  Steven  G.  Tuxle,  Mark  A.  Hagenmaier,  Campbell  D.  Carter,  AIAA  2014-­‐1158.  

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 27: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

Summary

1.  LES and DES as engineering tools 2.  Fatigue loading 3.  Inlet flows 4.  Selected topics in aerothermal prediction

27  DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059

Page 28: Computational Sciences Center High-Speed Flow … · Integrity ! Service ! Excellence Computational Sciences Center High-Speed Flow Research 2014 Dr. Jonathan Poggie Senior Aerospace

DISTRIBUTION STATEMENT A. Approved for public release; distribution is unlimited. 88ABW-2014-3059