Professional Documents
Culture Documents
, S.M. Dash
, and A. Hosangadi
Combustion Research & Flow Technology, Inc. (CRAFT Tech)
Email: gfeldma@craft-tech.com
Recently developed gas liquid methodology for cavitating flows is applied to higher-speed
flows dealing with bulk liquid venting and flyout. The new methodology improves upon
earlier VOF methodology, using a unified multi-phase thermodynamic framework and
preconditioning, and, it operates in a multi-element UNS grid. An earlier bulk flyout study
analyzed using VOF methodology is reviewed and a repeat of this calculation using the new
methodology is described, including application of grid adaptation to improve resolution at
the captured gas/liquid interface. Additional jet in cross-flow and venting problems are
described which demonstrate current capabilities.
I. Introduction
HIS paper discusses the application of recently developed gas/liquid methodology for cavitating flows,
1-5
to
higher speed problems dealing with bulk liquid venting and flyout. The analysis of bulk liquid interactions with
a gaseous stream is complex and requires the ability to: (1) fit or capture the deforming gas/liquid interface; and, (2)
to predict primary breakup processes along this interface producing droplets or ligaments. In conventional
approaches, the gas and bulk liquid are treated as distinct, immiscible phases. For problems with complex interface
geometries and/or where breakup may occur, capturing approaches are more practicable, with volume-of-fluid
(VOF) methodology typically being used for applications such as fuel injection in combustion chambers, with
empirical breakup relations applied along the captured interface.
T
Earlier applications of density-based VOF methodology (using the CRAFT CFD
methodology was the void fraction approach. Work prior to performing bulk-liquid fly-out problems entailed the
analysis of liquid propellant guns (LPG) where the pressures were very high and the liquid behaved in a
compressible manner. However, in analyzing liquid venting or bulk fly-out from missiles, the pressure was much
American Institute of Aeronautics and Astronautics
2
AIAA-2007-0922
lower and the liquid behaved in an incompressible
manner, which limited the applicability of the gas/liquid
framework in the CRAFT CFD
code.
Figure 2. Results From Bulk Liquid Fly-Out
Simulation Showing That Important Physical
Mechanisms Were Captured.
The original problem modeled was a liquid blob
flying through air at Mach 2. The liquid was a
cylindrical blob 50 cm long and 50 cm in diameter and
used properties extracted from our LPG work
9
. To
perform this analysis with the CRAFT CFD
code,
pressures had to be "artificially elevated" to make the
liquid behave in a compressible manner. The simulation
was run in a blob fixed framework where the grid moved
and decelerated with the blob. The liquid was
impulsively started at a velocity of 700 m/s. A snapshot
of the results is shown . This snapshot shows
the important physical characteristics captured in the
simulation. First, pressure lines show the bow shock
that formed ahead of the blob. Also evident is the
gas/liquid interface highlighted by the dark black lines.
Liquid ligaments are shown to have formed at both the
leading and trailing edges of the blob. This is the region
where primary breakup would first occur.
Figure 2
The simulation was first run without a primary breakup model, which made it possible to demonstrate the
capabilities of the gas/liquid formulation alone. A primary breakup model was then added to include more physics
into the simulation. Refs. 7 and 8 describe details of how the breakup and formation of droplets along the gas/liquid
interface was modeled. Figure 3 shows snapshots of the liquid blob case run without the primary breakup model. It
is clear that without breakup included, the liquid rolls up in an unrealistic manner. The views at 2.0 ms and 2.4 ms
show that without a breakup model, the ligaments formed at the gas/liquid interface continue to grow and elongate
without breaking off from the original blob. The case was then run again, this time making use of the primary
breakup model with results shown in Figure 4. The snapshots of this simulation taken at 0.45 ms and 1.14 ms look
similar to the case without primary breakup with some ligaments seen at the leading edge of the blob. However, the
remaining snapshots show very different results from the case without primary breakup. There are liquid droplets
seen breaking away from the ligaments, and by 3.81 ms there is very little of the original liquid blob still remaining.
Figure 3. Gas/Liquid Contours From The Simulation Run Without A Primary Breakup Model.
American Institute of Aeronautics and Astronautics
3
AIAA-2007-0922
Figure 4. Gas/Liquid Contours From The Simulation Run With A Primary Breakup Model.
III. Improved Gas/Liquid Formulation
Based on the comments above regarding the need for an improved gas/liquid framework for bulk flyout and
venting problems, new methodology developed for cavitating flows was examined. The new methodology entails
extensions to the multi-element UNS code, CRUNCH CFD
(5)
For efficient operations over a wide range of Mach numbers, this matrix may be further preconditioned to get
well-conditioned eigenvalues that improve convergence and reduce round-off errors. The following results indicate
that the upgraded formulation provides a robust methodology for predicting high-speed liquid venting and bulk fly-
out problems without the previous limitations on system pressure and required liquid compressibility.
IV. Bulk Liquid Fly-Out Studies With Improved Formulation
A. Liquid blob with comparison to earlier work
The earlier liquid blob case was repeated (same conditions) using the new formulation in the CRUNCH CFD
code. The only difference was that in this case we worked in ground fixed coordinates. Thus, a Mach 2 flow was
blown over an initially stationary liquid blob, which then moves downstream and eventually equilibrates with the
free stream flow. The solution with CRUNCH is remarkably similar to that earlier obtained with CRAFT.
Mass fraction contours are shown in Figure 5. The same roll-up of the liquid is seen at the edges, followed by
the elongation of liquid ligaments like the CRAFT CFD
code was
the ability to perform grid adaptation. The liquid blob case was run again, this time starting with a coarse grid and
then making use of grid adaptation. The current capabilities of the CRISP CFD
grid
adaptation feature will allow us to resolve the interface without the size of the grid becoming too computationally
expensive, and without the need for excess user intervention.
1 msec
4 msec
3 msec
2 msec
Figure 9. Liquid Blob Moving Through Time With Grid
Adaptation.
1 msec
4 msec 3 msec
2 msec Grid adapts to flow-field features as
the solution marches through time.
Figure 10. Changing Grid As It Adapts To The Blob Moving Through The Domain.
American Institute of Aeronautics and Astronautics
8
AIAA-2007-0922
C. Low pressure liquid blob
In this simulation, the liquid blob was analyzed at a pressure of 1 atm. As stated earlier, a deficiency in the
original VOF formulation was the need to artificially boost the pressure to make the liquid behave in a compressible
manner. Running the blob at a pressure of 1 atm is a precursor to cases that will be simulated at higher altitude, and
in general, pushes the limits of the gas/liquid formulation. Due to the stiffness of the equation of state currently used
for the liquid phase, it was necessary to add a small amount of artificial dissipation to the solution for numerical
stability. A more broadly applicable equation of state may be necessary for simulations run at higher altitudes or
lower pressures. Mass fraction contours are shown in Figure 11. Similar to the high pressure cases, the liquid rolls
up and forms ligaments at the trailing edge. The pressure contours are shown in Figure 12. As seen in the higher
pressure blob case, a bow shock forms upstream of the blob. The lower pressure case does look different towards
the trailing edge; however, as pressure waves show there is not as much expansion as in the high pressure case.
1 ms 2 ms
Gas Liquid Interface
Highlighted In Black
1 ms 2 ms
Gas Liquid Interface
Highlighted In Black
3 ms 4 ms 3 ms 4 ms
Figure 11. Mass Fraction Contours For Liquid Blob At 1atm.
1 ms
2 ms
3 ms 4 ms
1 ms
2 ms
3 ms 4 ms
Figure 12. Pressure And Liquid Mass Fraction Contours Of Liquid Blob At 1atm.
American Institute of Aeronautics and Astronautics
9
AIAA-2007-0922
V. Jet-In-Cross-Flow Studies
A. 2D and 3D Simulations at Atmospheric Conditions
Another class of gas/liquid interaction problems studied was liquid jet-in-cross-flow problems. These problems
show the versatility of the gas/liquid formulation in their ability to handle another complicated flow regime, of the
liquid jet flowing transverse to the gas. The first cases were simulated at atmospheric pressure (P =1 atm) and
temperature (T =300 K) analogous to the liquid fly-out problem. In this case, the cross flow gas was moving with a
velocity of 100 m/s and the liquid jet was moving at a velocity of 5 m/s. The liquid stream was 100% liquid by
volume.
This study was first conducted using a 2D slot jet, with liquid volume fraction contours shown in Figure 13.
Despite the low speed of the liquid, the momentum of the jet allows the liquid to penetrate to a height of about 2.5
nozzle diameters. Figure 14 shows velocity vectors for the 2D jet. This gives good perspective on the dynamics of
the flow field, and shows some important physics that were captured such as the recirculation region downstream of
the slot. An upstream air boundary layer was not included in this solution, which would have produced a separated
zone upstream of the liquid jet.
For completeness, the 2D case was extended into 3D with a round liquid jet and the same flow conditions as
described earlier. Figure 15 show contours of liquid volume fraction at the nozzle centerline. It can again be seen
that the momentum of the jet causes the liquid to penetrate to a height of about 2.1 nozzle diameters. For gas/liquid
flows the acoustic speed will vary greatly as the two phases mix, and this is shown in Figure 16. The acoustic speed
is highest in the regions where there is a pure gas or liquid, and drops dramatically as the two phases mix. Finally,
the shape of the jet can be seen in Figure 17 which shows iso-surface contours of constant liquid volume fraction.
The jet can be seen rolling-up as the gas and liquid mix downstream.
Liquid Penetration Height
~ 2.5 Nozzle Diameters
Liquid Penetration Height
~ 2.5 Nozzle Diameters
Figure 13. 2D Slot Jet Liquid Penetration Height.
Recirculation region
downstream of the jet
Recirculation region
downstream of the jet
Figure 14. 2D Slot Jet Velocity Vectors.
American Institute of Aeronautics and Astronautics
10
AIAA-2007-0922
Liquid Penetration Height
~ 2.1 Nozzle Diameters
Liquid Penetration Height
~ 2.1 Nozzle Diameters
Figure 15. 3D Round Jet Liquid Volume Fraction And Jet Penetration Height At Nozzle Centerline.
Acoustic speed is highest in pure
gas and pure liquid regions.
Acoustic speed drops dramatically
in regions of gas/liquid mixing.
Acoustic speed is highest in pure
gas and pure liquid regions.
Acoustic speed drops dramatically
in regions of gas/liquid mixing.
Figure 16. 3D Round Jet Showing Acoustic Speed Variations As A Result Of Gas/Liquid Mixing.
Liquid Penetration Height
~ 2.1 Nozzle Diameters
Jet rollup can be seen as liquid jet
and gas cross-flow mix.
Liquid Penetration Height
~ 2.1 Nozzle Diameters
Jet rollup can be seen as liquid jet
and gas cross-flow mix.
Figure 17. 3D Round Jet Iso-Surface Contours Of Constant Liquid Volume Fraction Shows Jet Penetration
Height And Jet Roll-Up As The Gas And Liquid Mix.
American Institute of Aeronautics and Astronautics
11
AIAA-2007-0922
B. Case 2 Missile Venting Simulations
The second liquid jet-in-a-crossflow problem was simulated with a generic missile body flying at an altitude of
10 km. A six-inch diameter hole at the missile mid-point was vented a liquid/gas mixture. Flow conditions and
problem setup are shown in Figure 18. For this case, the jet was 50% gas/50 % liquid by volume. The jet/free-
stream interaction, jet acoustic speed, and computational stability are all highly dependant on the liquid volume
fraction of the jet.
As shown in Figure 19, there is a significant difference in mass fraction and volume fraction at low free-stream
jet pressures. While the jet is a 50/50 mix of gas and liquid by volume, it is nearly all liquid by mass. Figure 20
exhibits pressure contours, and a shock is seen to form in front of the liquid jet due to interactions with the free
stream. As shown previously for the slot jet in Figure 16, the acoustic speed is highest for a pure liquid, drops off
sharply as the liquid concentration decreases, and then rises sharply again when the mixture becomes a pure gas.
This is seen clearly again for the missile at altitude in Figure 21 as the acoustic speed varies by more than an order
of magnitude from the free-stream to the jet, and then rises again as the liquid in the jet diffuses out downstream. In
Figure 22, Iso-Surface contours of liquid volume fraction show how the jet rolls-up as it mixes with the free-
stream.
Free-stream:
Altitude = 10 km
Mach Number = 1.6
Velocity = 500 m/s
Pressure = 28008 Pa
Temperature = 233.1 K
Jet:
Liquid Volume Fraction = 51.7%
Liquid Mass Fraction = 99.8%
Mach Number = 0.32
Velocity = 33 m/s
Pressure = 12P
inf
(336100Pa)
Temperature = 300 K
Missile
Gas/Liquid Jet
Figure 18. Missile Liquid Jet Venting Problem Setup And Definition.
Volume
Fraction
Mass
Fraction
Figure 19. Liquid Volume And Mass Fraction Contours For A Missile At 10 km.
American Institute of Aeronautics and Astronautics
12
AIAA-2007-0922
Shock seen
ahead of the jet.
Figure 20. Pressure Contours Around The Missile And Liquid Jet.
Shock seen ahead of the jet
Large variation in the
acoustic speed seen across
the jet. Then shown to
increase as the liquid
diffuses out down stream.
Figure 21. Missile Liquid Jet Mach Number And Acoustic Speed Contours.
Looking Downstream Looking Upstream
Jet shown rolling up as
it mixes downstream.
Figure 22. Iso-Surface Contours Of Constant Liquid Volume Fraction Of Liquid Venting From Missile.
American Institute of Aeronautics and Astronautics
13
AIAA-2007-0922
VI. Conclusion
The cases described above were all exploratory problems to assess gas/liquid capabilities of the CRUNCH
Code
used for cavitation, to bulk flyout and venting problems. It was shown that using the new formulation
available in CRUNCH
, gas/liquid studies could be performed at atmospheric pressure and low altitudes. This was
not possible using the earlier VOF formulation that required the pressure to be boosted artificially. The added
flexibility of the unstructured numerics in the CRUNCH
code.
Acknowledgments
This work was primarily supported by CRAFT Tech IR & D funding.
References
1
Hosangadi, A., and Ahuja, V., "Simulations of Cavitating Flows Using Hybrid Unstructured Meshes," Paper No.
FEDSM2000-11082, 4
th
ASME/J SME J oint Fluids Engineering Conference, Boston, MA, J une 11-15, 2000.
2
Hosangadi, A. and Ahuja, V., A Generalized Multi-Phase Framework For Modeling Unsteady Cavitation Dynamics And
Thermal Effects, Paper No. AIAA-2003-4000, 33
rd
AIAA Fluid Dynamics Conference, Orlando, FL, J un 23-26, 2003.
3
Hosangadi, A., Ahuja, V., and Ungewitter, R.J . Simulations Of Cavitating Flows In Turbopumps, Journal of Propulsion
and Power, Vol. 20, No. 4, pp. 604-611, J uly-August, 2004.
4
Hosangadi, A. and Ahuja, V., Numerical Study Of Cavitation In Cryogenic Fluids, Journal of Fluids Engineering, Vol. 127,
pp. 267-281, March 2005.
5
Hosangadi, A., and Ahuja, V., A New Unsteady Model for Dense Cloud Cavitation, FEDSM2005-77485, 2005 ASME
Fluids Engineering Summer Conference, Fifth International Symposium on Pumping Machinery, Houston, TX, J une 19-23,
2005.
6
Hosangadi, A., Sinha, N., and Dash, S.M., "A Unified Hyperbolic Interface Capturing Scheme for Gas/Liquid Flows," AIAA-
97-2081, 13
th
AIAA CFD Conferences, Snowmass, CO, J une 29-July 2, 1997.
7
Tonello, N.A., Dash, S.M., and Perrell, E.R., Advanced Computational Framework For Dynamic Bulk-Liquid Gas
Interactions, AIAA Paper-99-2205, 35th AIAA/ASME/ SAE/ASEE J oint Propulsion Conference and Exhibit, Los Angeles, CA,
J une 20-24, 1999.
8
Perrell, E., Tonello, N.A, Hosangadi, A., Sinha, N., and Dash, S.M., CFD of Complex Three-Dimensional Multi-Phase
Flowfields, Edgewood Chem-Bio Center, ECBC-CR-045, J anuary 2002.
9
Hosangadi, A., Sinha, N., Dash, S.M., Wren, G., DeSpirito, J ., Coffee, T., and Thompson, "3D Puddle Ignition Study for
Liquid Propellant Guns," 32nd JANNAF Combustion Mtg., Oct. 23-27, 1995.
10
Hosangadi, A., Ahuja, V., and Ungewitter, R.J ., Simulations of Inducers at Low-Flow Off-Design Conditions, FEDSM2005-
77395, 2005 ASME Fluids Engineering Summer Conference, Fifth International Symposium on Pumping Machinery, Houston,
TX, J une 19-23, 2005.
11
Ahuja, V., Hosangadi, A., Shipman, J ., and Cavallo, P.A., Simulations of Instabilities in Complex Valve and Feed Systems,
AIAA-2006-4758, 42
nd
AIAA/ASME/SAE/ASEE J oint Propulsion Conference & Exhibit, Sacramento, CA, 9 - 12 J ul 2006.
12
Cavallo, P.A., and Grismer, M.J ., A Parallel Adaptation Package For Three-Dimensional Mixed-Element Unstructured
Meshes, Journal of Aerospace Computing, Information, and Communication, Vol. 2, No. 11, pp. 433-451.
13
Cavallo, P.A., Arunajatesan, S., Sinha, N., and Baker, T.J ., Transient Mesh Adaptation Using an Error Wake and Projection
Method, AIAA Paper 2006-1149, 44
th
Aerospace Sciences Meeting, Reno, NV, J anuary 9-12, 2006.
American Institute of Aeronautics and Astronautics
14