Computational Fluid Dynamics Simulation of Hypersonic Engine Components

busyicicleMechanics

Feb 22, 2014 (3 years and 5 months ago)

114 views

1

Computational Fluid Dynamics Simulation of
Hypersonic Engine Components


by


Jack R. Edwards


Associate Professor

Department of Mechanical and Aerospace Engineering

North Carolina State University, Raleigh, NC


2

Overview



Computational fluid dynamics simulation of hypersonic engine
components


a major research thrust area in Aerospace Engineering at
NCSU since the Mid 1980s.


Current areas of emphasis
:


Nose
-
to
-
tail simulations of complete engine flowfields (NASA
Glenn; Edwards and McRae)


Modeling of turbulent Schmidt number and Prandtl number effects
in supersonic combustion (NASA Langley; Hassan and Edwards)


Modeling of supercritical
-
fluid and barbotage injection of
hydrocarbon fuels (AFRL/PRA; Edwards)


Algorithmic enhancements to NASA’s VULCAN flow solver
(NASA Langley; Edwards and McRae)


Hybrid large
-
eddy / Reynolds
-
averaged modeling of scramjet
component flowfields (NIA Seed Grant; Edwards)



3

Personnel



Dr. Jack R. Edwards,
Associate Professor


CFD algorithm development for reacting / multi
-
phase flows


Dr. Hassan A. Hassan
, Professor


Transition and Turbulence Modeling


Dr. D. Scott McRae,
Professor


Solution Adaptive Gridding Methods



Jason Norris
, Keith McDaniel, Ming Tian: Ph.D. students


Ana Pinto, Michael Schoen: M.S. students


Adam Amar: Undergraduate research assistant


4

Unique Contributions



Low
-
Diffusion Flux
-
Splitting Schemes (LDFSS)


High
-
resolution upwind
-
differencing methods


Extensions for real fluids, gas
-
solid flows, multi
-
phase mixture
flows, chemically reacting flows, etc


Several parallel, multi
-
block, implicit flow solvers built around
LDFSS techniques



k
-


Transition / Turbulence Models


Coordinate
-
invariant two
-
equation model for wall
-
bounded and
free
-
shear flows at all speeds


Transition model accounts for Tollmein
-
Schlicting, crossflow,
bypass, and second
-
mode disturbance growth


Predicts onset and extent of transition and has been coupled with
the Spalart
-
Allmaras and the
k
-


model


5

Unique Contributions



Dynamic Solution
-
Adaptive Gridding Techniques


Improved feature resolution through point
-
clustering


Extensions for time
-
accurate flows, multi
-
block grids with non
-
contiguous interfaces, unstructured grids


Recent applications to high
-
speed inlet unstart and pollutant source
tracking in air
-
quality models



Hybrid Large
-
Eddy / Reynolds
-
Averaged (LES/RANS)
Simulation methods


Techniques combine RANS strategies near solid surfaces with
LES strategies further away


Transition facilitated by flow
-
dependent blending functions


Applications to shock / boundary layer interactions in internal
flows


6

Resources



NCSC IBM SP
-
2 (720 processors, 1 teraflop; soon to be
replaced with a linux Beowulf cluster)


4
-
processor Compaq ES
-
40


2
-
processor Microway DS
-
20


1
-
processor Compaq XP
-
1000


Several Sun, SGI workstations


Several PCs


LaTEX, Tecplot, Ensight, animation software


VULCAN (NASA Langley), CHEM3D (Dow Chemical)


REACTMB variants (NCSU)


All codes parallelizable with MPI message
-
passing

7

High
-
Speed Propulsion



Time
-
dependent simulations of Scramjet inlet / isolator /
combustor interactions


Nose
-
to
-
tail simulations of NASA Glenn’s GTX Rocket
-
Based Combined
-
Cycle engine concept


Addition of time
-
derivative preconditioning and parallel
implicit schemes to NASA’s VULCAN flow solver


Simulation of injection of supercritical fuels


Simulation of aerated
-
liquid injection of hydrocarbon
fuels (Barbotage)



8

Independent Ramjet Stream Cycle in RBCC Engines


Injectors add fuel to the incoming air.


Mixing in ramjet stream precedes ignition.


Thermal throat is present.


Location of thermal throat can be modulated by variations in fuel injection.

Thermal Throat

Flame Front

Rocket exhaust

Fuel injection and

premixing

9

Rocket
-
Based Combined
-
Cycle Simulations

10

Rocket
-
Based Combined
-
Cycle Simulations

11

Rocket
-
Based Combined
-
Cycle Simulations: Rocket
-
shutoff with
Nitrogen Purge

12

Aerated
-
liquid (Barbotage) injection experiments


The Air Force Research Lab
(AFRL) aerated
-
liquid injector is
schematically illustrated in Fig. 01;



Rectangular configuration with a
dimension of
6.4

mm x
2.0

mm;



A square cross section with
dimension,
D
, of
2.0

mm used for
the final discharge passage,
L/D=20
, converging angle
θ
=
50
°
;



Water as the test liquid, and
nitrogen as the aerating gas.


Fig
.

01
,

Schematic

of

the

injector

assembly

and

internal

flow

structure

13

Volume fraction contours (GLR = 0.08%)

Bernoulli inflow B.C. for the liquid phase

14

GLR=2.45% Photos and simulations

15

Hybrid LES/RANS Simulation Techniques



General approach: unsteady RANS (Reynolds
-
Averaged Navier
-
Stokes) near solid surfaces


LES (large
-
eddy simulation) in outer part
of the boundary layer and in free
-
shear layers



Transition between RANS / LES based on flow
-
dependent blending
functions based on ratios of turbulence length scales


best results
when transition occurs in outer part of log layer



RANS models: k
-


and Menter’s k
-




LES subgrid model: Yoshizawa’s one
-
equation SGS model



Applications to cavity flameholder configurations, flow behind
projectiles, shock / boundary layer interactions



16

Hybrid LES/RANS Simulation Techniques

Instantaneous axial velocity (25 degree compression / expansion
corner
)

17

Hybrid LES/RANS Simulation Techniques

x
'
,
c
m
p
w
/
p

-
5
0
5
0
.
5
1
1
.
5
2
2
.
5
3
3
.
5
4
4
.
5
5
E
X
P
H
y
b
r
i
d
(
k
-

,
F
=
F
3
,
f
i
n
e
g
r
i
d
)
H
y
b
r
i
d
(
k
-

,
F
=
F
3
,
c
o
a
r
s
e
g
r
i
d
)
R
A
N
S
(
k
-

)
Wall pressure distributions (25
degree compression/ expansion
corner)

u
/
u
e
y
'
,
c
m
0
0
.
1
0
.
2
0
.
3
0
.
4
0
.
5
0
.
6
E
x
p
H
y
b
r
i
d
(
k
-

,
F
=
F
3
,
f
i
n
e
g
r
i
d
)
H
y
b
r
i
d
(
k
-

,
F
=
F
3
,
c
o
a
r
s
e
g
r
i
d
)
R
A
N
S
(
k
-

)
x
'
=
1
.
2
5
c
m
2
.
3
5
c
m
3
.
1
0
c
m
1
1
1
0
Velocity profiles in recovery
region (25 degree compression /
expansion corner)


18

NIA
-
Sponsored Work


Primary Goal: to extend earlier work in hybrid LES/RANS simulations to
three
-
dimensional flows characteristic of dual
-
mode scramjet engines



Year 1 accomplishments


Addition of generalized multi
-
block capability to hybrid LES/RANS
solver


Addition of full reactive
-
flow capability


Development of better blending functions to shift modeling from
unsteady RANS to LES



Test cases underway:


Investigation of separation
-
shock unsteadiness in compression
-
corner
interactions


Simulation of reactive flow downstream of UVA single
-
ramp, dual
-
mode injector using hybrid LES/RANS












19

NIA
-
Sponsored Work: Separation
-
Shock Unsteadiness



Prediction of response of turbulent boundary layer to shock interaction
(representative of high
-
speed flows within inlet / isolator configurations)



Large
-
scale, low
-
frequency unsteadiness of regions of shock
-
separated
flow observed in experiments



Can hybrid LES/RANS methods predict this type of unsteadiness?

20

NIA
-
Sponsored Work: Separation
-
Shock Unsteadiness

Time
-
dependent surface pressure contours

21

NIA
-
Sponsored Work: Separation
-
Shock Unsteadiness

Average surface pressure distributions

PDF of separation
-
shock position

22

Leveraging NIA
-
Sponsored Work




Hybrid LES/RANS Simulations of Complex Internal Flows with
Multiple Shock / Boundary Layer Interactions” Edwards and Hassan;
AFOSR; pending



“Database and Model Development for Combined
-
Cycle Mode
Transition” McDaniel, Cresci, Edwards, Goyne, O’Brian, Riggins,
Schetz; NASA NGLTP; pending (submitted by NIA)



MURI White Paper on Combined Cycle Engines, Frankel, Edwards,
McDaniel, Goyne, Hanson, Sung, Dutton, Loth; AFOSR; pending







23

Challenges



Demise of North Carolina Supercomputing Center (July 1, 2003)


loss of 720 processor IBM SP
-
3



Mitigation strategies:


32 processor IBM P690 (NCSU)


32 processor IBM Bladecenter (NCSU)


128 processor IBM Bladecenter (NCSU; under construction;
expandable)


Access to 1024 processor IBM SP
-
3 at Oak Ridge National
Laboratories








24

Simulation of a time
-
dependent coatings process

25

Pollutant Capture in Circulating Fluidized Beds



Three
-
phase system: two solids phases, one multi
-
component gas phase


Sub
-
models for fine particulate matter agglomeration,
sulfur dioxide sorption, mercury capture onto activated
carbon


High
-
resolution LDFSS extension for separated gas
-
solid
flows




26

Solids voidage time evolution

27

Fine PM number density time evolution

28

Fine PM flow rates

29

Supercavitating water flow about a projectile

30

New Directions



Atmospheric turbulence modeling and solution
-
adaptive
meteorological simulations


Level
-
set methods and immersed
-
boundary algorithms


Human
-
induced contaminant transport


Diesel engine injector simulations


Two
-
phase bubble dynamics


Hybrid LES/RANS simulations of


Shock
-
train propagation


Ramped
-
injector flowfields


Biological systems (lung bronchii, aortic aneurisms)




31

Level
-
Set / Immersed Boundary Methods: 2
-
D
Simulation of “feet” moving in a box filled with air