boeing sonic boom prediction workshop results · boeing research & technology | aeromechanics...

Engineering, Operations & Technology Boeing Research & Technology

Boeing Sonic Boom Prediction Workshop Results 2nd AIAA Sonic Boom Prediction Workshop - Sci Tech 2017 - Grapevine, Texas

Todd Magee David Lazzara Boeing Research & Technology Boeing Research & Technology [email protected] [email protected]

January 7-8, 2017

mailto:[email protected]



Boeing Research & Technology | Aeromechanics Technology

Outline • FUN3D Results

– Method/Approach – Convergence – Nearfield Results for Required Grids – Comparison of Inviscid Flux/Limiter Methods

• Cart3D Results – Summary of Cases – General Cart3D Setup – Space-Marching Setup – Nearfield Results for Required Geometries and Adapted Grids

• Conclusions • Questions

| 2


�FUN3D Results

Magee | 3


FUN3D Method / Approach

�FUN3D flow solver Version-12.9-0555546 • Euler Equations • Conducted a flux methods and limiters trade for the NASA 25D with FTN • HLLC inviscid flux method with Van Leer limiter for all required cases

– This method has worked well for our preferred OVERFLOW method, so we wanted to see how the FUN3D implementation did for the required cases

• Initialized flow by running all cases 1st order �Utilized top three refined grids for each required case • Mixed grid for Axisymmetric body (1.00, 1.28, and 1.60) • Tetrahedra grid for JAXA WB (0.70, 0.83, and 1.00) • Mixed grid for NASA 25D with FTN (0.64, 0.80, 1.00)

Dual Intel Haswell 64-bit 14-core processors • NASA 25D Axisymmetric Body – 420 cores • JAXA WB – 420 cores • NASA 25D with FTN – 840 cores

Magee | 4


FUN3D was Utilized for all Required Cases

Magee | 5

NASA 25D with FTN

JAXA WB

NASA 25D Axisymmetric Body Equivalent Area


JAXA W/B Convergence History

Magee | 6

Iteration

R_1

0 3000 6000 900010-16

10-14

10-12

10-10

10-8

10-6

10-4

HLLC, Van Leer Flux Limiter, 0.70 gridHLLC, Van Leer Flux Limiter, 0.83 gridHLLC, Van Leer Flux Limiter, 1.00 grid

Iteration

C_L

3000 6000 9000

0.076

0.078

0.08

0.082

0.084HLLC, Van Leer Flux Limiter, 0.70 gridHLLC, Van Leer Flux Limiter, 0.83 gridHLLC, Van Leer Flux Limiter, 1.00 grid

Iteration

C_M

_y

3000 6000 9000-0.07

-0.068

-0.066

-0.064

-0.062HLLC, Van Leer Flux Limiter, 0.70 gridHLLC, Van Leer Flux Limiter, 0.83 gridHLLC, Van Leer Flux Limiter, 1.00 grid

Iteration

C_D

3000 6000 90000.006

0.0065

0.007

0.0075

0.008

0.0085

0.009

0.0095

0.01HLLC, Van Leer Flux Limiter, 0.70 gridHLLC, Van Leer Flux Limiter, 0.83 gridHLLC, Van Leer Flux Limiter, 1.00 grid


Undertrack Nearfield Results for NASA 25D Axi-Body At Multiple H/L

Magee | 7

X (meter)

dP/P

40 60 80 100 120 140 160 180 200 220-0.010

-0.005

0.000

0.005

0.010

X (meter)

dP/P

120 140 160 180 200 220 240 260 280 300-0.010

-0.005

0.000

0.005

0.010

X (meter)dP

/P200 220 240 260 280 300 320 340 360 380

-0.010

-0.005

0.000

0.005

0.010

H/L = 3

H/L = 1

H/L = 5


Undertrack Nearfield Results for NASA 25D Axi-Body

Magee | 8 X (meters)

(P-P

inf)

/Pin

f

100 150 200 250 300-0.006

-0.004

-0.002

0

0.002

0.004

0.006HLLC, Van Leer Flux Limiter, mixed 1.00 gridHLLC, Van Leer Flux Limiter, mixed 1.28 gridHLLC, Van Leer Flux Limiter, mixed 1.60 grid

H/L = 3


Undertrack Nearfield Results for JAXA WB at Multiple H/L

Magee | 9

X (meters)

dP/P

120 140 160 180 200-0.0050

-0.0025

0.0000

0.0025

0.0050

X (meters)

dP/P

40 60 80 100-0.0100

-0.0075

-0.0050

-0.0025

0.0000

0.0025

0.0050

0.0075

0.0100

H/L = 2.55

H/L = 0.85


Undertrack Nearfield Results for JAXA WB

Magee | 10

X (meters)

(P-P

inf)

/Pin

f

120 140 160 180 200 220 240 260 280 300-0.005

-0.004

-0.003

-0.002

-0.001

0

0.001

0.002

0.003

0.004

0.005

HLLC, Van Leer Flux Limiter, 0.70 gridHLLC, Van Leer Flux Limiter, 0.83 gridHLLC, Van Leer Flux Limiter, 1.00 grid

H/L = 2.55


FUN3D Undertrack Nearfield Results for NASA 25D FTN

Magee | 11

X (meters)

dP/P

40 50 60 70 80 90-0.015

-0.010

-0.005

0.000

0.005

0.010

0.015

X (meters)

dP/P

120 140 160 180-0.015

-0.010

-0.005

0.000

0.005

0.010

0.015

X (meters)

dP/P

200 220 240 260-0.015

-0.010

-0.005

0.000

0.005

0.010

0.015

H/L=1

H/L=3

H/L=5


FUN3D Undertrack Nearfield Results for NASA 25D FTN

Magee | 12

X (meters)

(P-P

inf)

/Pin

f

120 130 140 150 160 170 180-0.007

-0.006

-0.005

-0.004

-0.003

-0.002

-0.001

0

0.001

0.002

0.003

0.004

0.005

0.006


H/L = 3


Comparison of Several Inviscid Flux/limiter Methods for NASA 25D with FTN

Magee | 13


X (meters)

(P-P

inf)

/Pin

f

120 130 140 150 160 170 180-0.008

-0.006

-0.004

-0.002

0

0.002

0.004

0.006

0.008Van Leer flux construction, hvanalbada flux limiterVan Leer flux construction, no flux limiterVan Leer flux construction, Van Leer flux limiterHLLC flux construction, Van Leer flux limiterVan Leer flux construction, vanalbada flux limiterLDFSS flux construction, vanalbada flux limiter

Comparison of Inviscid Flux/limiter Methods - NASA 25DFTN

Magee | 14

Pg.15

Pg.16

Pg.17


X (meters)

(P-P

inf)

/Pin

f

132 134 136 138 140 142 1440.001

0.002

0.003

0.004

0.005

0.006


Magee | 15



Magee | 16


X (meters)

(P-P

inf)

/Pin

f

152 152.5 153 153.5 154-0.004

-0.003

-0.002

-0.001

0

Van Leer flux construction, hvanalbada flux limiterVan Leer flux construction, no flux limiterVan Leer flux construction, Van Leer flux limiterHLLC flux construction, Van Leer flux limiterVan Leer flux construction, vanalbada flux limiterLDFSS flux construction, vanalbada flux limiter


Magee | 17


X (meters)

(P-P

inf)

/Pin

f

156 158 160 162 164-0.005

-0.004

-0.003

-0.002

-0.001

0

0.001



�Cart3d Results

Lazzara | 18


Summary of Cases

�Non-adapted and adapted grids used for each case ▪ Axisymmetric Equivalent Area case ▪ JAXA Wing Body case ▪ NASA 25D with flow-through nacelle case

�CART3D flow-solver

▪ Finite volume, inviscid Euler solver, Cartesian Euler grid, with minor vertical stretching ▪ Grid-aligned shock, along with a small radius cylinder around the model for extraction of

flow state variables along linear sensors ▪ Parallel execution on single compute node

▪ Non-adapted cases: 20 core, 128 GB RAM nodes ▪ Adapted cases: 20 core, 512 GB RAM nodes

�Space-marching solver

▪ High-order finite difference Euler solver ▪ Propagates flow state variables from an inner cylinder boundary to an outer cylinder ▪ High density structured grid ▪ Parallel execution on a single compute node: 20 core, 128 GB RAM

Lazzara | 19


General Cart3D Setup Full Cylinder State Variable Extraction

▪ Full model geometry, cylinder H/L = 0.25, 2 deg spacing (Axisymmetric case used half-model)

▪ Cylinder centerline aligned with free-stream vector ▪ Cylinder length 2L—3L

Lazzara | 20

Radius, H

Length, L

Altitude


General Cart3D Setup (Cont.) Model Rotated for Grid-Aligned Azimuth Directions

▪ Model rotated about the free-stream velocity vector for each azimuth direction ▪ Requested azimuth directions are grid-aligned ▪ Non-adapted grids utilize pre-specified density boxes ▪ Model is rotated in pitch for grid-aligned shocks (including offset for sonic glitch) ▪ Adapted grids pursued to lower error on pressure signature functional

Lazzara | 21

0 deg. Grid-Aligned 30 deg. Grid-Aligned 50 deg. Grid-Aligned

Grid-Misaligned Direction


General Cart3D Setup (Cont.) Functional Convergence Monitored

▪ Functional convergence of all line sensors tracked ▪ Total functional convergence tracked for each adaptation cycle

▪ Same for non-adapted cases ▪ Adaptation functional weighting emphasized grid-aligned azimuth direction ▪ Adaptation cycles chosen until further error improvement not seen ▪ Smallest resolution is 2.4 cm for all cases

Lazzara | 22


Space-Marching Setup

�Concentric Cylinder Output ▪ High accuracy signature propagation for all

azimuth directions to a “frozen” signature with smaller computational expense

▪ MPI parallelism

▪ Ideal atmospheric model captures ambient change for large H/L

▪ Enables smaller CFD domain

▪ Introduced in AIAA 2016-2037

▪ Full-cylinder propagation and half-cylinder propagation available (with symmetry BC) ▪ Initial data sets (JAXA and NASA 25D cases)

incorrectly used the symmetry BC for full-model cases

▪ Analyses rerun without a symmetry BC using full-cylinder propagation and submitted

Lazzara | 23

Free-stream H/L = 1

H/L = 3

H/L = 5

H/L = 10


Axisymmetric Equivalent Area Case

Lazzara | 24

Non-Adapted Grid 22601287 cells

Adapted Grid #25 41394292 cells



Lazzara | 25

Non-Adapted Grid Adapted Grid #8



Lazzara | 26

H/L = 1 H/L = 3 H/L = 5

�Non-adapted and adapted signatures agree well


JAXA Wing Body Case

Lazzara | 27




JAXA Wing Body Case

Lazzara | 28



JAXA Wing Body Case: H/L = 0.85

Lazzara | 29 �Non-adapted and adapted signatures agree well


NASA 25D with Flow-Through Nacelle Case

Lazzara | 30




NASA 25D with Flow-Through Nacelle Case

Lazzara | 31



NASA 25D with Flow-Through Nacelle Case: H/L = 1

Lazzara | 32 �Non-adapted and adapted signatures agree well


Conclusions

�FUN3D Analysis • HLLC Inviscid flux method with Van Leer flux limiter provide nearfield

results with low dissipation – However some nearfield signature features are unphysical – Convergence stalled on some provided grids

• Results similar for all inviscid flux methods and limiters tried in the study – Van Leer flux construction with hvanalbada flux limiter yields the most

dissipative solution

�Cart3d Analysis • Non-adapted and adapted signatures agreed well • Full-cylinder propagation needed for full-models

– Setting a symmetry BC will yield incorrect signature shapes • Space-marching capability greatly reduces computational expense

– Full-cylinder propagation more efficient and accurate than conducting the same simulation in CFD alone

Lazzara | 33


�Questions?

Author, 7/10/2014, Filename.ppt | 34


FUN3D Cp Contours and Convergence for NASA 25D Axisymmetric Body


Iteration

R_1

0 2000 4000 600010-17

10-15

10-13

10-11

10-9

10-7

10-5

10-3

HLLC, Van Leer Flux Limiter, mixed 1.00 gridHLLC, Van Leer Flux Limiter, mixed 1.28 gridHLLC, Van Leer Flux Limiter, mixed 1.60 grid

Iteration

C_D

0 2000 4000 6000 8000 10000

0.001

0.0015

0.002

0.0025

0.003



FUN3D Cp Contours for NASA 25D with FTN

Magee | 38

Top

Bottom


Iteration

C_L

0 5000 100000.06

0.065

0.07


FUN3D Convergence Histories for NASA 25D with FTN

Magee | 39

Iteration

R_1

0 5000 10000

10-13

10-11

10-9

10-7

10-5

10-3


Iteration

C_M

_y

0 5000 10000-0.054

-0.052

-0.05

-0.048

-0.046HLLC, Van Leer Flux Limiter, mixed 0.64 gridHLLC, Van Leer Flux Limiter, mixed 0.70 gridHLLC, Van Leer Flux Limiter, mixed 0.83 grid

Iteration

C_D

0 5000 100000.008

0.009

0.01

0.011



Lazzara | 40

JAXA Wing Body H/L = 0.85: Symmetry BC Incorrectly Applied


Lazzara | 41

NASA 25D H/L = 1: Symmetry BC Incorrectly Applied

boeing sonic boom prediction workshop results · boeing research & technology | aeromechanics...

Documents