Académique Documents
Professionnel Documents
Culture Documents
Abstract— This paper describes a study of the lift effect parametric studies prevents the use of concept to it’s fullest
obtained by super circulation, an aerodynamic effect discovered potential.
by Henri Coandã that relies heavily on the Coandã effect. Data provided in [1] and [2] are of significant value
A parametric two dimensional CFD study has been carried out as starting points, however they can only model one aircraft
with two goals in mind, the primary goal was to see the impact of configuration and offering little details on how the
the ambient pressure on the super circulation effect and also a
T
mechanisms of super circulation work. Thuslly, in order to
secondary goal to investigate the super circulation processes
themselves. Early parametric studies have been performed by generalize the applications of lift achieved by super
various authors however the parameterizations provided in the circulation, parametric tests have to be made to insure at least
available literature is applicable only to some particular aircraft a semi-empirical set of basic design equations.
configurations. The value of this study is that it provides a bare Perhaps one of the most famous equations used to
geometric parameterization that can be used in a wider variety of describe super circulations is the momentum coefficient :
applications from aircraft lift and actuators to fluidic actuators
ES
and machinery. The tests showed no dependency between the
ambient pressure and the super circulation effect which
encourage us to state that an aeronautical application – that must
operate both at high and low altitudes- is feasible. Further study
Where
Cμ=T/qS (1)
has shown that the injector fluid is accelerated by the curved T represents the static thrust of the engine
ramp at higher velocities than those of the injector, providing providing the USB system
more leads for further refinement of our understanding of the q is the dynamic pressure of the free stream
phenomenon itself. S is the super circulated surface aria
Keywords-super circulation, Coandã effect, k-omega SST The denominator includes the dynamic pressure of
the free stream of air, which means it is more suitable for
describing aircraft landing and taking off than the hovering
I. INTRODUCTION
A
capability of Coandã’s original demonstrator – a lot of the
times this equation proves very valuable when dimensioning a
In 1932, the Romanian aerodynamicist Henri Coandã blown flap system, per se.
proposed a new heavier than air lift concept, the ― lenticular Key aspects such as injector stream velocities,
aerodyne‖. The principle used to achieve lift is now called curvature radii, ambient pressure must be taken into account in
― super circulation‖ and it is, in part, owed to the Coandã effect determining weather or not a super circulation application is
that helps maintain a stream of fluid to a nearby wall. preferable to a conventional lift system and under what
IJ
Although the Coandã effect is necessary to achieve super circumstances it is viable over the flight envelope of the
circulation, it is not sufficient, i.e. in order to achieve a application.
favorable pressure gradient we need to use curved surfaces Another aspect that make parameterization of this
such as cylinders. aerodynamic effect difficult is numerically modeling the
During the years, many attempts have been made to detachment of the boundary layer from the cylindrical ramp.
blend the lenticular aerodyne’s concept into conventional It is common knowledge that a turbulent boundary
tube-wing aircraft, the most famous examples are the Antonov layer is less likely to detach from a wall than a laminar
An-72 and An-74 and the Boeing YC-14. These aircraft used boundary layer, therefore various viscous models will yield
the cold by pass flow of their turbofan engines to generate a various points of flow separations.
combined Upper Surface Blow (USB) that provides significant In this paper we will try to investigate the influence
lift, yielding lower take off and landing velocities. of the ambient pressure, at zero true air speed (TAS) on the
Even if such aircraft have proved their commercial- pressure decrease over the super circulated ramp, considering
and often military- use, the lack of publicly available the same injector velocity using various viscosity models by
Computational Fluid Dynamics (CFD).
A. The setup of the CFD tests K-epsilon RNG with pressure gradient effect near wall
Although it is not the intended purpose of this paper, treatment
a brief analysis of the viscosity models has been made prior to
the full test in order to insure as much accuracy as a The flow immediately begins to split in two regions: a region
conventional viscosity model can offer. that attaches itself to the curved ramp and another one, clearly
In the literature [3], [4]. the viscosity model most affected by the pressure outlet boundary proximity. On first
commonly considered to predict boundary layer separation is glance, the re-attached flow may be neglected giving the false
the k-omega. This model has a lower turbulent production impression that this turbulence model predicts flow separation
than the k-epsilon and therefore can capture more accurately quicker than the k-omega, which is counter intuitive. Upon
the detachment of the boundary layer. closer inspection we can observe that the re-attached flow has
The general layout of the test can be seen in Fig.1, significant effects, remaining attached to the ramp for its
where we can observe the pressure outlet walls, the injector entire span – which was to be expected from this model.
defined as a velocity inlet and the cylindrical ramp. One indicator that shows this is not the best way to model the
A 90° arc was selected due to practicality reasons: flow is the fact that the boundary effect is quite intense, hence
1.from past experience, a 100m/sec flowing jet will the necessity to generate a larger domain which in turn implies
not stay attached for much longer than the 90° of the ramp a higher time expense.
2.if we were to consider that there is an even pressure A final remark that needs to be made is that the rapid pressure
T
distribution across the ramp’s span, drop visible at 72 cm of the ramp’s span is caused by a vortex
the resulting force will not be useful, as its lateral components meaning that, perhaps a more precise result may be obtained
will nullify each other as seen in Fig.4. by a nonstationary simulation.
A sensible argument can be made that the higher the
curvature, the higher the pressure gradient we will most likely Reynolds stress 5 equation model with pressure
obtain, however an optimum will be reached because of the gradient effect near wall treatment
ES
fact that a fast flowing jet will become detached quicker on a
highly curved wall than on a lower curved wall therefore the
parameter to be optimized in this case will have to be the
product between the pressure gradient obtained and the
The pressure plot indicates multiple vortices forming in the
immediate vicinity of the injector. It is the sensible thing to
assume that vortices are an expression of the Kelvin-
circumferential length of the attached fluid. Helmholtz interaction that manifests when two fluids with
Knowing the influence of the ambient pressure over different velocities have a common interface. This is the prime
the pressure gradient obtained over the span of the ramp noise generating mechanism for jet engines. Experience has
trough super circulation is important in two key aspects: shown that a time dependent nonstationary analysis yields
1.Calculating the effectiveness and efficiency of a more accurate results.
super circulation system with altitude Positive aspects of this model are the lack of influence both
2.Calculating the prospect of having the super from the boundaries and from the underside of the ramp.
circulation effect used by high pressure pneumatic systems However, the model does not predict attachment to the curved
A
such as fluidic actuators as described in [9]. wall at this velocity of 100 m/sec. Knowing that the velocity
The injector inlet velocity was intended as high as of the injected air is critical in achieving lift trough super
possible while within the incompressible domain of the circulation, lowering it further from 100 m/sec makes little
working fluid-which generally is though to be below a Mach sense.
number of 0.3, resulting in our case in an injector velocity of
100 m/sec. Hence a pressure based solver was employed. Spalart Allmaras strain/vorticity based production
The high of the injector for this, two dimensional
IJ
study, was chosen to be h=10 cm and the ramp radius R=50 In this model, the boundary effect is virtually nonexistent, the
cm, being close to 12.7% of the ramp span. fluid gets practically no parasitic influence from the pressure
Four viscosity models were initially tested and briefly outlets.
analyzed before the full test in order to decide which one The underside of the ramp has also no parasitic effect, leaving
would most likely give the best approximation for the flow the flow unaltered.
separation. Also, it is remarkable that the fluid remains completely
attached to the ramp for it’s entire span and that the drop in
static pressure generated almost identical to that predicted by
the k-omega SST model.
Apart form the fact that it cannot predict flow separation, this
model is very close to the k-omega.
K-omega Menter (shear stress transport SST) by the ambient pressure variations. It also has shown to be
independent of the ambient pressure both at low pressures and
This viscosity model is the most commonly used to model the at high pressures.
Coandã effect and related phenomena because it can predict Another remarkable fact is that the lift calculated for
both the attachment of the flow to the ramp and it’s eventual this super circulated ramp is significantly greater than the
separation from it. thrust of the injector bare flow. The significance of this
In this case we can observe some degree of boundary effect on finding can potentially be greater than just augmenting
the flow as it exits the ramp however it has no practical aerodyne lift, it could open the way for new types of jet engine
influence on the pressure gradient obtained on the ramp itself. nozzles, although experimental confirmation will be required
The trajectory of the fluid as it exits the ramp is not along the before further speculating on this last prospect.
tangent line to the ramp’s curve at the separation point but Further studies are required and may include non
rather slightly diverted away from the ramp. This is because of stationary analysis, thinner injector jets, ramp rugosity effects,
the influence of the jet thickness that will be discussed further. temperature effects.
Also the vortex near the right hand side of the domain
influences the exit trajectory in a converse manner.
III. RESULTS
A first observation that can be made is that the ramp
accelerates the injector fluid in a region of approximate one
T
third of the total jet thickness. The velocity increase is of
approximate 5-7% of the initial injected velocity.
Close observation of the pressure and velocity plots
shows close correlation between the velocity increase and the
pressure drop on the ramp.
Detachment of the flow is most likely influenced by
T
ES
Figure 3. Injector disctretisation
A
Figure 4. Lift distribution over a 180° ramp
IJ
T
Figure 5. K-epsilon pressure plot Figure 7. K-epsilon velocity plot
ES
A
IJ
T
Figure 11. Reynolds stress velocity plot
Figure 9. Reynolds stress pressure plot
ES
A
IJ
T
Figure 13. S-A stress pressure plot
ES Figure 15. S-A stress velocity plot
A
Figure 16. S-A velocity detail
IJ
T
Figure 19. K-omega SST velocity plot
ES
Figure 17. K-omega SST pressure plot
A
T
ES
A
Figure 21. Ramp static pressure distribution by viscosity models
IJ
T
Figure 22. pressure distribution over a supercirculated ramp k-omega SST
ES
A
IJ
T
Figure 24. Pressure drop plots at low ambient pressure also overlap
[5] US3971534
[6] Skavdahl, Howard; Wang, Timothy; and Hirt, William J.: „Nozzle
Development for the Upper Surface - Blown Jet Flap on the YC-14
Airplane.‖ Automot. Eng., Apr.-May 1974. [reprint] 740469, SOC.
[7] Spence, D. A.: „The Lift Coefficient of a Thin, Jet-Flapped Wing.‖ Proc.
Roy. SOC. (London), ser. A, vol. 238, no. 1212, Dec. 4, 1956, pp. 46-68.
[8] T.Welsh (Boeing) 1984 US4426054
[9] J. Glass (Cava Industries) US3589382
.