computations of high enthalpy shock-waves in east using us3d
TRANSCRIPT
DURGESH CHANDEL
University of Minnesota
August 10th, 2017
Entry Systems and Technology DivisionNASA Ames Research Center
Computations of high enthalpy
shock-waves in EAST using US3D
Motivation
• High-speed entry vehicles experience immense surface-heating and ablation.
• Radiative heating is important, requires knowledge of gas-state and reaction kinetics.
10.16 cm dia. (blue)
Test section at 8 m
76 cm dia. (orange)East operational range:
V = 1.3 - 46.0 km/s
P = 0.002 - 76.87 psia
High-fidelity computations are required to interpret EAST data.
Typical conditions:
M = 23-32
V = 8-12 km/s
P = 3.87×10-3 psia
h0 = 33-60 MJ/kg
Shock-tube experiments inform about,
• Spectroscopic and Kinetics models
• Validation and uncertainty quantification for flight data
all images from nasa.gov
Earth
2
EAST flow
• Current interpretation: Flow over blunt-body
• Previous attempts
• Improved approach in US3D
Fluid Solver results
• Early evolution of shock
• Set-up for Radiation Calculations
Radiation solver results
Summary and Path forward
Contents
EAST: Flow over blunt-body
• Shock-deceleration and varied shock-strength causes strong non-equilibrium.
• Experimental data deviates from numerical predictions at certain conditions.
• Stagnation streamline analogous to shock-tube centerline.
• Shock-speed is matched with an appropriately chosen shock stand-off distance.
4
us d
Flow over cylinder
us =10 km/s
Test section
Shock-tube flow
10.4 km/s10.6 km/s10.7 km/s
~ 1.5m
d
Air-shocks
EAST: Previous Attempts
Contact-front lacks thermal equilibrium
7 days for 2m shock-travel on 2000 CPUs
• Kotov et al.(JCP 2014), 1T model, 2D duct
• Low-dissipation, high order shock-capturing
9.86 km/s Equilibrium
Shock
Front
Contact
Front
• Barnhardt et al., attempted full facility DPLR
simulation, 2009
• Aggressive CFL ramping caused instability
Radiance monotonically decreases behind shock.
Lower CFL cases in DPLR also get unstable.
COOLFluiD simulations are stable but expensive
(Bensassi, K.)
Time-accurate EAST simulations are expensive.
5
Ionization-fraction, near Test-section
tube d
ia.
x= 6.85mx= 6.8m
EAST: Improved Approach in US3D
Goal: 2D axi-symmetric flow simulations w/ real-gas effects.
(time-accurate yet computationally feasible!)
Required modifications:
• Implicit system of equations needs higher accuracy than stagnant shock-problems.
(more ‘kmax’ sub-iterations in computing Flux-Jacobians in each time-step)
• Catalytic recombination BC at wall
• Moving Grid and/or Numerical interpolation introduces errors.
• Moving frame-of reference? Frame-acceleration requires additional modeling.
6
Shock-frame is the best framework for moving-shock problems in US3D.
Moving-frame w/ constant frame-speed:
1. US3D code is heavily tested and verified for stagnant bow-shock problems.
2. Shock-frame calc. found to be more accurate than Lab-frame calc.
(moving-shock in perfect gas at M = 3, 5, 14)
3. Standing shock simulations at M = 20 keep the shock stagnant as expected.
Numerical set-up: Shock-frame calc.
Flux scheme 1st order MSW
Time-integration 1st order implicit FMDP
Frame-of-reference Shock-frame
Grid size Variable grid, 1.8 M
Grid resolution (min.) Δx = 10 µm near shock
Δr = 1 µm near wall
99.9% He + N2 79% N2 + O2
1.10546 kg/m3 3.096d-04 kg/m3
6000 K 300 K
Wave motion in shock tube at time t>0
7
The whole frame is moved with a constant speed close to the shock-speed
Driver gas (x<0) Driven gas (x>0)
r (m
)
x (m)Shock starts at x=0 at time, t = 0
Fluid Solver Results: US3D
Early evolution of shock
9
r (m
)
x (m)
• Shock-deceleration profile consistent w/ EAST.
• Translational and Vibration-electronic modes relax with
time.
r (m
)
x (m)
Shock-velocity
Translational Temperature
Shock-deceleration is stronger in 2nd order flux-scheme
Flow-field at later time
10
Axial-profiles are well-behaved near regions of strong gradients
shock-front
contact-surface
Centerline profiles
Lab-frame
Shocked-gas
~ 5cm wide
r (m
)
Set-up for Radiation Calculations
• Shock reached at ~1.29m
• Shock-front is kept in the
refined region by changing
the frame-speed.
Us = 10.02 km/s
• Thermo-chemical equilibrium is not fully
established.
• Radiation properties would be different
than the measurements at the Test-section.
11
t = 125µs LOS extraction using Bala’s code, Shock-frame
y (
m)
d
shock-front
contact-surface
Te
mp
era
ture
(K
)S
pe
cie
s m
ass-f
rac.
Radiation Solver Results: NEQAIR
VUV Radiation
LOS on Temperature contours, Shock-frame
d
EAST spectrum of interest
VUV = 120-215 nm
UV/Vis = 190-500 nm
Vis/NIR = 480-900 nm
IR = 700-1650 nm
13
Qualitative behavior is similar.
experiment
Us =10.3 km/s
Us =10 km/s
Radiation Spectra
Us = 10.02 km/s
P = 0.2 Torr
14
d = 2.48 cm
experiment
Us = 9.98 km/s
Vis/NIRUs =10.27 km/s
UV/Vis
Us = 9.98 km/s
IR
Summary
Progress in Summer:
• Improved approach in US3D gives stable solution w/ real-gas effects.(shock-front has to be kept in the refined region)
• Flow solution and Radiation profiles are consistent w/ test.
Path Forward:
• Propagate the shock till test-section. (Projected time: 14 days, 480 CPUs; 8 times cheaper than Kotov et. al.)
• Resolve BL features
• Include capability for variable frame-speed in US3D.
• Validate results w/ test-data.
Credits
Mentoring and Support
• Dr. A. Brandis, Dr. B. Cruden, Dr. D. Hash; (NASA Ames)
• Prof. G. V. Candler, Dr. I. Nompelis; (UMN)
Valuable discussions
• Dr. K. Bensassi, Dr. J. Schulz, Dr. R. L. Jaffe; (NASA Ames)
• Ames co-interns: Narendra, Bala, Maitreyee and others.
Questions?