Professional Documents
Culture Documents
Aerodynmics Article-Final
Aerodynmics Article-Final
Base Flow
Introduction:
The supersonic base problem appears in a wide range of objects that are being used
nowadays such as missiles, projectiles, launchers and airplanes. This problem is
characterized by the interaction between a supersonic flow with a subsonic flow at the
free-stream.
There are two ways to determine the properties and features of this type of flow. One is
to develop an experiment that could measure the pressure, temperature and velocity
inside the boundary layer and at the points of interaction. Despite the fact that this
provides results with better accuracy, it is not usual to predict such properties with
experiments because it is more expensive and demands more effort and time. Another
way is to perform a numerical simulation of the flow. The simulations are made with the
use of a CFD commercial package, such as the ANSYS 14 (Fluent). To perform the
simulations, a numerical method is needed. Some examples of numerical methods are
the RANS, LES, ZDES and RANS/LES. Their differences are basically the accuracy
given by them on the results and the computational cost involved with each one.
This type of problem is not well understood yet. This paper will include a proposed
research in order to investigate such flows and make improvements that will benefit the
population and the companies.
A complete understanding of this problem is of fundamental importance because, as it
is known, the supersonic flow introduces instability in the near region. If the problem
could be well understood, adaptations could be made and a decrease in the instability
generated and an increase in efficiency could be achieved.
1
Literature review:
A flow is considered to be supersonic when its velocity is greater than the speed of the
sound, that is 340.29 m/s. Commonly, the mach number (M) is used to identify if a flow
is supersonic or subsonic. If M > 1, the flow is supersonic and if M < 1, the flow is
subsonic. The Mach number, that is the ratio between the speed of the flow and the
speed of sound, is given in (1).
V
M= (1)
V sound
2
When an object travels at speeds greater than the speed of sound it creates acoustic
disturbances in the form of waves, which propagate with the speed of sound. If the
mach number of the particle is greater than one, the generated waves cannot propagate
upstream of the particle and the sound of an object in supersonic speed comes later
than the object.
In a supersonic flow, the pressure disturbance cannot travel upstream. Therefore, when
a fluid strikes an object at supersonic speed, it is forced to change its properties such as
the temperature, Mach number, density and pressure in an extremely violent and
irreversible way. This propagating disturbance caused by the transition from a subsonic
to a supersonic speed is called the shock wave. This is an ordinary wave that carries
energy and can propagate through the fluid. Across a shock wave there is an extreme
rise in density, pressure and temperature of the flow. The energy of the shock waves
dissipates very quickly with the distance.
As can be seen in figure 2, there is the presence of a mach wave, oblique shock and
normal shock when a flow passes a supersonic body.
Figure 2: Typical normal shock, oblique shock, and Mach wave pattern in supersonic flow past a body.
It is important to know what these three terms mean. The mach wave is a shock wave
of minimum strength, a normal shock is a plane shock normal to the direction of flow,
and an oblique shock is the shock wave which deviates at some arbitrary angle from the
flow direction.
It is known that the velocity upstream of a shock wave is supersonic, because the shock
wave only happens when a body exceeds the speed of sound. Analyzing the oblique
3
shock, it can be seen that two types of shock are possible. If the velocity downstream of
this type of shock is subsonic it will result in a strong shock, and if the velocity is
supersonic, a weak shock will occur.
Other properties that can be stated are that the downstream velocity component normal
to any shock wave is always subsonic and there is no change in the tangential velocity
component across the shock.
To represent all these disturbances, the Mach cone was created and can be seen in
figure 3. The cone OCD is called the downstream Mach cone and the cone AOB is the
upstream Mach cone. The inclination of this cone is given by (2).
(2)
When an object travels with speeds greater than the speed of sound it also generates
expansion waves that interact with the shock waves and eventually merges with them,
partially cancelling it out. The passage of an aircraft with M > 1 produces the sonic
boom which is the sound wave resulting from the degradation and merging of the shock
wave and the expansion wave produced by the aircraft.
When there is the presence of a shock wave, it is known that the drag force on the
object increases. This is due to the fact that the energy is extracted as work decreases
and the entropy increases, despite the fact that the total energy is conserved.
The supersonic base flow that is analyzed in this paper is not well understood yet. Their
difficulties arise in the fact that there is an interaction between a subsonic flow in the
free stream and a supersonic flow close to the object. It is difficult to develop
experiments that could measure the properties of this type of flow and it is also hard to
interpret the data obtained.
4
Such flows are not completely understood because of unsteadiness of features and
multiple constraints leading to very complex physics. Because of the complexity of this
problem we have a poor knowledge of unsteady features of the base flows and difficulty
in finding reliable experimental data.
One problem of this type of flow is that it causes a dramatic decrease of pressure at the
base, due to the massive separation of the flow on that region of the object. Other
problems are the increase in the base drag, because we do not have a flow control over
the body, and the detachment of the boundary layer that happens too early. Other
problems related with such flow are the generated instability, the shock waves and the
unsteadiness of the recirculation zone.
Complete understanding of these problems is of fundamental importance for us,
because when a missile, projectile, launcher or plane is being developed we need to
consider how this type of flow is going to affect the behavior of our object when in use.
Taking a look in figure 4, it is possible to see some features of this flow such as the
shear layer, the shock wave, the recirculation zone just behind the base and the
expansion of the flow. These features of the flow have an important role in the
properties as will be discussed later on.
7
Looking at figure 6, it is possible to see that the features of the supersonic base flows
are evidenced and point out the unsteady nature of such flows with the presence of
numerous turbulent scales. This picture was obtained using a ZDES simulation with the
mesh M2. Just behind the base it is possible to see the recirculation zone. Looking
more to the right it is possible to see the turbulent wake in the middle with a dark color,
and the recompression shock on the surroundings of the turbulent wake.
Comparing the e RANS/LES (M1) and LES (M1 and M2) simulations, the influence of
the mean velocity profile of the incoming boundary layer could be investigated. The
main discrepancy between these simulations is around the initial vorticity thickness,
which is the result from different boundary layer thickness at separation.
The results of [4] show that the simulation made with the RANS method poorly agrees
with experimental data from [2] and fails to reproduce the shear layer expansion behind
the base. In the LES calculation, the incoming boundary layer is responsible for the
under prediction of the mixing layer growth rate and the results obtained with this
method were more accurate. The simulation made with the ZDES method closely
matches the experimental data as the compression region is approached.
Concerning the turbulent quantities, Reynolds stress profiles in the mixing layer confirm
the better agreement of the RANS/LES results with the experiment while LES
simulations exhibit an excessive stability of the mixing layer.
Reynolds stresses in the last downstream station. According to [4], the base pressure
coefficient does not match the experimental value for both numerical methods.
In figure 7 is possible to see the Reynolds shear stress profiles.
8
Figure 7: Reynolds shear stress profiles. From [1].
Saracoglu, Paniagua, Sanchez and Rambaud [5] analyzed the effects of blunt trailing
edge flow discharge in supersonic regime considering flow discharge in the base region.
The aim of this study was to provide an analysis of the effects of the trailing edge purge
on the flow features, such as trailing edge shocks and base pressure. The flow
topology, base pressure and shock waves are analyzed, and this analysis provides
aerodynamic knowledge to the modulation of shock waves.
All the analysis of this study is based on time-averaged results of URANS and the
results were post-processed to quantify the shock intensity and angle.
9
Saracoglu et al. [5], utilize the URANS method, because it yields an improvement
compared with RANS, and the precision is similar to the provided by the Large-Eddy
Simulation without a high computational cost. This was shown by Garnier et al. [6].
To validate the numerical simulation the data obtained was compared to the
experimental results obtained by Smith [7] on a 2.5 Mach stream flowing over a
supersonic rearward-facing step.
The geometry, pressure and temperature used in the simulation are the same used by
Smith [7]. Saracoglu et al. [5] utilized a 2D triangular mesh with a non-uniform
rectangular boundary layer close to the wall. All the numerical simulation was performed
using the CFD package ANSYS 14 (Fluent).
The fluid was considered an ideal gas and the equations were solved explicitly using the
density-based solver due to the high compressibility effects. A second order upwind
differencing was used and the turbulence for the high Reynolds number was modeled
by a two equation k-epsilon model.
Other parameters were set in the Fluent such as the enhanced wall treatment with
pressure gradient effects, the pressure inlet; the walls were considered adiabatic and
convergence was assumed when the absolute residual value became less than 0.001.
Three types of mesh (N1, N2 and N3) were used.
In figure 8 it is possible to see the convergence history of the base pressure obtained
with the numerical simulation. It is possible to see that there are some oscillations
occurring at high frequency and this is related to the vortex shedding.
Figure 8: URANS convergence history of the base pressure in upper and lower side of the trailing edge. From [5].
10
Figure 9 shows the flow streamlines and vorticity in the trailing edge region. As can be
seen, the flow field was highly affected by the flow discharge rate. Figure 9a contains an
undisturbed base flow with two counter rotating recirculation bubbles neatly enclosed by
shear layers on each side of the trailing edge. In figure 9b the dead air zone was
penetrated by the low momentum purge jet, a new pair of vortices started to grow and
the shear layers on both side of the trailing edge were shifted outwards. In figure 9 the
bubbles disappear because of the higher momentum jet. The recirculation region was
reduced and captured by the shear layer. In figure 9d jet achieved supersonic speed
and started to push the recirculation regions towards the shear layers on each side of
the trailing edge due to a strong expansion at the jet outlet.
Figure 9: Changes in base region flow topology shown by streamlines (top) and vorticity magnitude (bottom) (a) no
blowing, (b) qpurge/qf = 1.05, (c) qpurge/qf = 1.4 (d) qpurge/qf = 13.5. From [5].
Figure 10 shows the changes in the trailing edge shock properties in function of density
ratios. Figure 10.1 shows the intensity and figure 10.2 shows the shock angle. The
flattening of the shear layer lowers the deflection of the flow on the wake and,
consequently, the degree of compression reduces.
11
Figure 10: Characterization of trailing edge shock waves: shock intensity (I) and angle variation (II) with respect to
density ratio. From [5].
In figure 11 it is possible to see how the extracted base pressure values were non-
dimensionalized by the no blowing base pressure.
Figure 11: Base pressure correlation in function of the density ratio. From [5].
The conclusions of Sacoglu [5] are that purging with density ratios between unity and
2.8 provided a beneficial playground to modulate the shock waves for lower intensity
without compromising the refrigeration requirements in turbine applications, and the
trailing edge discharge is beneficial in reducing static pressure variability and
consequently the flow structure interaction. Hence, the force exerted on the downstream
components by the shock waves may be lowered.
B. Fiorina, S.K. Lele [8] studied the artificial nonlinear diffusivity method for supersonic
reacting flows with shocks. The objective of this study is to develop and to validate a
numerical methodology adequate for resolving interactions between shocks, turbulence
and combustion. Although the Large-Eddy Simulation gives results with a good
accuracy for supersonic flows, they used a high order scheme in their simulations to
capture the physically important turbulent and chemical scales. This high order
12
schemes generate non-physical oscillations when solving for shock waves.
In order to solve this problem two artificial dissipation terms are introduced. First, a
nonlinear artificial viscosity, based on high-order derivatives of the strain rate tensor is
introduced. Then, an artificial diffusivity based on high-order derivatives of entropy is
also introduced. The addition of a nonlinear artificial viscosity and an artificial diffusivity
was demonstrated to be a good way to solve the shock-turbulence interaction problem
by Cook and Cabot [9,10].
The method with the nonlinear artificial viscosity only is denoted as NVI and the
numerical method with the artificial viscosity and diffusivity is denoted as NVIDI.
With this new numerical approach, a detailed analysis of the errors associated with
shock-capturing and contact-surface capturing was conducted, and the new scheme
proved to capture both weak and strong shocks without any degradation of
performance. It was also demonstrated that the numerical shock thickness and the
magnitude of the spurious wiggles are largely independent of the mesh size and the
shock/contact surface strength.
The numerical simulations were made for a shock tube problem. Simulations were
made for the NVI and NVIDI methods with different pressure ratios.
The first simulation was made with a pressure ratio of 2, used a uniform mesh of 160
grids points and was analyzed at the time t = 1.8. In figure 12 is possible to see a
comparison between the theory given by [9,10], the NVI and NVIDI the simulation
results for the pressure, the velocity, the density and the temperature. This simulation
showed that a good agreement between the numerical solution and the theory, in
particular both the positions of the contact surface and of the right propagating shock,
are well captured. As can be seen in figure 12, the NVI and NVIDI provide a nearly
identical solution except at the contact surface where small differences are observed.
13
Figure 12: Numerical simulations of the shock tube problem with an initial pressure ratio of 2. From [8].
A second shock tube problem was computed, but this time with a stronger pressure
ratio. The pressure ratio is now equal to ten. Simulation was performed with a 400 grid
points and analyzed at the time t=4.5. NVI had difficulties starting because of the sharp
pressure and density discontinuities, but the NVIDI overcomes this problem.
Figure 13 shows the comparison between the theory and the NVIDI method.
14
Figure 13: Numerical simulations of the shock tube problem with an initial pressure ratio of 10. From [8].
In the last shock tube problem, showed in figure 14, fluids of different initial entropy are
now separated by the diaphragm, and the free-stream Mach number is set to 1. The
simulation was performed in a 150 grid points and analyzed at the time t = 1.6. The
shock wave is well captured by both models. The results are almost the same for them
except at the contact surface region, where a large discrepancy is observed. The
comparison between NVI and NVIDI is plotted.
15
Figure 14: Numerical simulations of the shock tube problem with a density factor of 10. From [8].
With the results of the numerical simulation and the comparison with the theory, Fiorina
et al. [8] concluded that this new scheme is able to capture both weak and strong
shocks without any degradation of performance.
Ez Hassana, John Boles, Hikaru Aonoc, Douglas Davis, Wei Shyy [11] studied the
Supersonic jet and crossflow interaction with emphasis on the computational modeling
for this type of flow. The supersonic jet-in-crossflow problem represents essential
physics of fuel injection in SCRAMJET engines. This problem is complicated to deal
with because it has a highly unsteady regime, which leads to shocks, turbulent mixing,
and large-scale vortical structures that require a special turbulent treatment to obtain
accurate solutions.
16
A large revision on the numerical methods used for numerical simulations such as the
RANS, LES, ZDES, RANS/LES and multi-scale was made. Different turbulence
modeling techniques were reviewed and compared in terms of their performance in
predicting results consistent with the experimental data. There is also a large review of
papers and experiments that deal with supersonic flows.
The areas of fuel concentration and pressure and velocity fields were studied in [11].
Figure 15 shows the mean and variance of fuel concentration on the center plane for
the intermediate grid obtained from Hassan [12].
Figure 15: Mean and variance of fuel concentration from Hassan [12] for RANS, multi-scale and experimental NO-
PLIF images.
Analyzing the pressure and axial velocity with the RANS method using a coarse 0.6
million-symmetric, Hassan et al. [12, 13] discovered that they are really similar. Figure
16 shows pressure fields at the center plane for both RANS and multi-scale model
computations. Similar shocks and flow structures can be seen in both contours with the
RANS computation showing features that are less sharp due to the significantly coarser
grid. Figure 17 shows the axial velocity for RANS and multi-scale model computations.
In figure 20 is possible to see the pressure contours for the intermediate grid for RANS
and multi-scale computations and compare it with experimental PSP [14].
17
Figure 16: Pressure on the center plane. Comparison between RANS and multi-scale computation (baseline case)
from Hassan et al. [13].
Figure 17: Axial velocity on center plane. Comparison between RANS and multi-scale computation (baseline case)
from Hassan et al. [13].
18
Figure 18: Contours of P/PN on bottom wall. RANS and multi-scale predictions compared to PSP (baseline case)
from Hassan et al. [13].
After the study, Hassana et al. [11] concluded that the RANS methods are limited when
predicting flow structure because it cannot accurately capture unsteadiness in the flow.
This model is also inadequate for accurately modeling the large scale mixing in of
simulating jet in supersonic crossflow. Also, the LES method can predict turbulent
structures with reasonable accuracy but it requires high grid resolutions in the near wall
region, which makes it less practical for engineering flow analyses. Also, the hybrid
method RANS/LES combines reasonable accuracy and efficiency.
Research proposal:
My proposal research will be on the analysis of the supersonic base flow that occurs
with aircrafts and rockets where the flow that comes out of the jet engine has a
supersonic speed and the outer flow has a subsonic speed.
Details of the problem are as follows: every aircraft or rocket needs a jet engine to work
and their purpose. These objects normally travel with speeds greater than the speed of
sound, which means that they have a Mach number greater than one. But the flow just
gets a supersonic speed after passing through all the stages inside the engine.
For an aircraft, the flow gets into the engine with a subsonic speed and will be
compressed at the first stage inside the engine and will come out of this stage with a
high pressure. The next stage is the combustion chamber where the flow will suffer the
combustion process. At this stage, its speed is going to increase as the total energy of
19
the flow, measured by the increase the temperature. The next step is to pass through a
series of turbines. After all these stages the flow is released to the atmosphere with a
supersonic speed and will interact with the outer flow that has a subsonic speed at the
free-stream.
The principle of the rocket engine is very similar with the jet engine for propulsion, but it
does not require the atmosphere to provide oxygen for the combustion process as it
carries all the components needed for the reaction.
This problem of the interaction of supersonic and subsonic flows is really important, as
already mentioned, because with the supersonic flow comes instability, increase of
turbulence of the flow, unsteadiness of the recirculation zone, increase on the drag and
shock waves.
The analysis of this problem will be for aircrafts and rockets because they have an
important role in the economy today. The aviation sector does not stop growing. The
airline companies need to increase the number of flights every year to supply the
demand, and for this to happen they need to buy more airplanes. So, if the problem in
analysis in this paper were better understood, improvements on the engine and on the
fuselage could be made. These improvements could lead to a reduction in fuel
consumption, an increase in engine efficiency, a reduction in travel time since the speed
of the plane could increase and a decrease in the amount of pollutants released in the
atmosphere.
The analysis will be made by a numerical simulation of the problem. Due to the
complexity and importance of the problem, a numerical method that provides results
with a high accuracy is needed. All the numerical simulations will be performed using
the hybrid RANS/LES method that is capable of providing results with reasonable
accuracy and computational cost.
To perform this simulation some values such as the Mach number and the radius of the
engine need to be determined. Also, values like the velocity, pressure and temperature
in the free-stream need to be assumed. Another thing that has to be decided before
starting the simulation is the number of meshes and their number of nodes, and if any
grid refinements will be made.
To perform this simulation, the commercial package ANSYS 14 (Fluent) will be utilized,
because, despite not being an open source program, it is one of the most utilized
software for fluids.
To perform this numerical simulation the resources needed are the ANSYS 14 (Fluent)
commercial package, a computer with a good processor and a numerical code. Human
resources are not needed because an experiment will not be made.
In figure 19 is possible to see the problem of the interaction of a supersonic flow that
comes out of the jet with the subsonic outer flow.
20
Figure 19: Representation of the problem in analysis.
Conclusion:
This paper reviewed the properties of the supersonic flow and how such a flow can
affect a body exposed to it. Characteristics and properties of the numerical methods that
are being used to predict the properties and features of the supersonic base flow were
analyzed and their advantages and disadvantages were listed.
The current research for the supersonic base flow was discussed and some papers
such as the studies about Reynolds-Averaged Navier–Stokes/Large-Eddy Simulations
of Supersonic Base Flow, Effects of blunt trailing edge flow discharge in supersonic
regime were analyzed.
A research of the analysis and a numerical simulation of the interaction of a supersonic
flow coming out of a jet engine with a subsonic flow in the free-stream was proposed.
The aim of this research is to identify the properties and features of this interaction. As a
result of this, some improvements of engines could be made in order to reduce fuel
consumption, decrease the instability generated, reduce the noise produced and
increase the velocity of the airplane or rocket.
21
References:
[1] Smith HE. Experimental and theoretical investigation of the near wake behind
a rearward facing step in supersonic flow. PhD Dissertation. University of
Cincinnati; 1969.
[2] Herrin, J. L., and Dutton, J. C., “Supersonic Base Flow Experiments in
the Near Wake of a Cylindrical Afterbody,” AIAA Journal, Vol. 32,
No. 1, 1994, pp. 77–83.
[3] Tucker, P. K., and Shyy, W., “A Numerical Analysis of Supersonic
Flow over an Axisymmetric Afterbody,” AIAA Paper 93-2347,
June 1993.
[4] Franck Simon, Sébastien Deck, Philippe Guillen, “Reynolds-Averaged Navier–
Stokes/Large-Eddy Simulations of Supersonic Base Flow,” AIAA JOURNAL
Vol. 44, No. 11, November 2006.
[5] B.H. Saracoglua, G. Paniagua, J. Sanchez , P. Rambaud, “Effects of blunt trailing
edge flow discharge in supersonic regime,” Science Direct, September 2013.
[6] Garnier E, Pamart PY, Dandois J, Sagaut P. Evaluation of the unsteady RANS
capabilities for separated flows control. Comput Fluids 2012;61:39–45.
[7] Smith HE. “Experimental and theoretical investigation of the near wake behind
a rearward facing step in supersonic flow”. PhD Dissertation. University of
Cincinnati; 1969.
[8] B. Fiorina a, S.K. Lele b, “An artificial nonlinear di ffusivity method for supersonic
reacting flows with shocks,” Science Direct, September 2006.
[9] A.W. Cook, W.H. Cabot, A high-wavenumber viscosity for high resolution numerical
method, J. Comput. Phys. 195 (2004) 594–601.
[10] A.W. Cook, W.H. Cabot, Hyperviscosity for shock–turbulence interactions, J.
Comput. Phys. 203 (2005) 379–385.
[11] Ez Hassana, John Boles, Hikaru Aonoc, Douglas Davis, Wei Shyy, “Supersonic jet
and crossflow interaction: Computational modeling,” Science Direct, August 2012.
[12] Hassan E. Multi-fluid dynamics for supersonic jet-and-crossflows and liquid
plug rupture, in aerospace engineering. Ann Arbor: University of Michigan;
2012.
[13] Hassan E, Aono H, Boles J, Davis D, Shyy W. Multi-scale turbulence model in
simulation of supersonic crossflow. In 49th AIAA aerospace sciences meeting including
the new horizon forum and aerospace exposition 2011.Orlando, Florida.
[14] Crafton, J, A Forlines, S Palluconi, M Hsu, C Carter, and M Gruber,
Investigation of transverse jet injections in a supersonic crossflow using
fast responding pressure-sensitive paint. AIAA journal, submitted for
publication.
22
Peer Review:
The abstract comes before the introduction, moreover you should write on it the aims of the
report, the interpretation of the outcomes, the implications of it and if there is something still
unsolved.
The sentence that informs about the aims of the report in the introduction should be the last one
as a conclusion of it.
You called Mach number as M in the beginning of the report, thus you should keep with it.
Misspelling “Reynolds”
In the first paragraph of the current research, it is missing a comma before the “such”
Replace this analysis by these analyses because you are talking about more than one.
There is a need of human resource to use the software and do the analyses