Professional Documents
Culture Documents
Related terms:
The area rule was first applied to a production aircraft in the Convair F-102A, the
USAF's first supersonic interceptor. Emergency application of the area rule became
necessary because of a serious problem revealed during the flight tests of the proto-
type aircraft, the YF-102. Its transonic drag was found to exceed the thrust produced
by the most powerful engine then available, which threatened to jeopardize the
entire program, considering that supersonic flight speed was an essential USAF
specification. The area rule was used to guide a major revised design of the fuselage.
This reduced the drag sufficiently for supersonic Mach numbers to be achieved.
The bursting occurs due to the positive pressure gradient bringing the low energy
fluid to rest, and the bursting can be eliminated by fluid suction applied just
downstream of the original position, as in the case of flow separation. The analogy
between a burst and a normal shock wave at supersonic speed is rather loose, but
both cases are associated with an instability in decelerating flow causing a sudden
increase of pressure.
Since the occurrence of the burst on a wing causes a loss of suction locally on the
flow surface and a modification of the position of separation of surface flow beneath
the vortex, the subject of bursting should be well understood for practical purposes.
The theoretical investigations were made by Legendre [34], Adams [35], Brown and
Michael [36], Roy [37], and Mangier and Smith [39].
As the listed references show, much has been done experimentally and analytically,
but as yet the phenomena of bursting of the leading-edge vortices are not fully
understood. Qualitative observations were made in order to understand the flow
behavior and to provide a sound basis for analytical prediction of steady and un-
steady bursting. The water-tunnel observations were concerned with how sweeping
back the leading edge could transform the highly turbulent flow of the leading-edge
separated flow on a stalled unswept sharp-edged wing into the stable vortex-flow
characteristics of a highly swept wing. Furthermore, the structure of the vortex was
studied; velocity components, pressure and noise level were measured, and the
location of the vortex core was determined.
In the analysis, by approximating the free vortex sheets by two line vortices, the flow
is presented simply [34, 35, 36], but the analytical results were not in good agreement
with experimental data. Mangier and Smith's solution based upon the slender wing
theory display, numerically, most of the experimentally observed features.
Each of the pair of vortices for a delta wing may be regarded as being generated by a
conically rolled vortex sheet from the leading edge, but the vortex can be considered
as a core of rotational flow fed by the vorticity from the leading edge, increasing
in size and strength with distance downstream. The vortex layers spring from the
leading edge S1 and form a line of secondary separation S2 on the upper surface.
The observations indicate that upstream of the burst, each stream line that enters
the vicinity of the vortex after bending around the leading edge has the form of a
nearly cylindrical spiral (Fig. 7).
FIG. 7. Axial filaments of dye flowing past sharp-edged delta plate. Water tunnel
velocity 2 in./sec [15]
Hall's [30] theory on vortex structure indicates that the rotational core consists of two
parts, i.e. an inviscid but rotational annulus, and an inner viscous core surrounded
by the annulus. The rotational annulus may be regarded as a region of approximately
constant swirl velocity, and the viscous core as the region in which the swirl velocity
falls to zero at the axis. The bursting appears to be a property of the core of the vortex,
and the axial flow within the core can be extremely sensitive to small retardations
of the external flow. The ratio of the swirling to the axial velocity component is a
very important factor for bursting. Nose-shape appears to have little effect on the
incidence at which vortex break-down occurs.
Measurements show that the total pressure at the axis is nearly constant along the
length of the vortex after a very steep fall close to the apex. The value of the total
pressure at the axis is very low due to the deficiency of rotational velocity caused by
viscous diffusion within the narrow inner core. The geometrical configuration of the
axis of the vortex on the wing is nearly straight and at an angle to the free stream
in the upstream portion of flow, but downstream, approaching the trailing edge, it
curves towards the free stream direction.
The various observations of burst at low velocity and Reynolds numbers indicate
that vortex flows from the sharp swept leading edge decelerate suddenly along the
vortex axis, deflect, and perform a regular whirling motion, then break-down of
turbulence occurs. The bursting is caused by an adverse pressure gradient along the
axis coupled with a low total pressure within the vortex core. On a wing, the axial
flow is easily brought to rest by pressure recovery associated with the trailing-edge
and the wake. In general, if the pressure recovery associated with the trailing edge
is not sufficient to stagnate the axial flow, a burst does not occur on the body
surface, although a burst may take place during a subsequent axial pressure rise due
to vorticity diffusion in the wake. Although the pressure rise in the outer regions of
flow surrounding a vortex is gradual, a sudden deceleration immediately ahead of
a burst appears to lie in the steeper pressure rise along a stream line near the axis.
Under certain conditions it is possible, depending upon the ratio of the rotational
to axial velocity components, to expand a vortex spontaneously and to provide the
pressure rise necessary for stagnation of the axial filament. The position of burst may
be determined by considering a balance of various independent factors dictated by
the geometry of the body. The burst location is sensitive to the pressure gradient
along the vortex. When the burst occurs upstream of the trailing edge, its position
depends upon a combination of incidence and leading-edge sweep back, but is
largely independent of Reynolds number. Once the conditions necessary to cause a
burst have been met, its final location may be determined by the extent to which fluid
from the turbulent region formed downstream of the burst can penetrate upstream
along the axis of the vortex. The observations reveal further, that for low values of
Reynolds number, between the positions of axial deceleration and the turbulent
break-down, a region of periodic flow exists, and the axial filament performs a
regular whirling motion. The vortex may expand in an axisymmetric manner about
the stagnation point in the axial flow, and because an axisymmetric arrangement
seems to be unstable, a strong tendency exists to collapse into a spiral configuration.
Therefore, the spiral configuration is a secondary feature of the process of bursting
[14, 15].
An investigation of the flow over an oscillating delta wing with 80° leading-edge
sweepback in a water tunnel [40] showed that considerable differences exist in the
position of vortices for steady compared to the unsteady motion, as qualitatively
shown in Fig. 8.
FIG. 8. Position of leading-edge vortex for a delta wing (pitching axis at 0·625 c0
from apex) [14]
With increase of incidence, the vortex gained its strength and moved inboard, but
did not reach the position of steady conditions until the maximum incidence was
attained. Conversely, with the decrease of incidence, the vortex lost its strength, but
moved inward beyond the position of steady flow.
Wing Theory
E.L. Houghton, ... Daniel T. Valentine, in Aerodynamics for Engineering Students
(Seventh Edition), 2017
Nature does not tolerate infinite velocities, and a more realistic vortex-sheet model
is shown in Fig. 7.34 (full lines). It is evident from this figure that the assumptions
leading to Eq. (7.23) cannot be made in the mid-span region even for high aspect
ratios. Thus, for swept wings, simplified vortex-sheet models are inadmissible and
the complete expression in Eq. (7.22) must be used to evaluate the induced velocity.
The bound vortex lines must change direction and curve around smoothly in the
mid-span region. Some may even turn back into trailing vortices before reaching
mid-span. All this is likely to occur within about one chord from the mid-span.
Further away, conditions approximate those for an infinite-span yawed wing. In
effect, the flow in the mid-span region is more like that for a wing of low aspect
ratio. Accordingly, the generation of lift will be considerably impaired in that region.
This effect is evident in the comparison of pressure coefficient distributions over
straight and swept wings shown in Fig. 7.35. The reduction in peak pressure over
the mid-span region is shown to be very pronounced.
Figure 7.35. Comparison of pressure distributions over straight and swept-back
wings.
The pressure variation depicted in Fig. 7.35(b) has important consequences. First,
if it is borne in mind that suction pressure is plotted in the figure, it can be seen
that there is a pronounced positive pressure gradient outward along the wing. This
tends to promote flow in the direction of the wingtips, which is highly undesirable.
Second, since the pressure distributions near the wingtips are much peakier than
those further inboard, flow separation leading to wing stall tends to occur near the
wingtips first. For straight wings, on the other hand, the opposite situation prevails
and stall usually first occurs near the wing root—a much safer state of affairs because
rolling moments are smaller and flow over the ailerons, needed to control rolling
moments, remains largely attached. These difficulties make the design of swept
wings considerably more challenging than the design of straight wings.
(9-90)
Brandt et al. [4] present the following expression to estimate the factor:
(9-91)
where
For three-dimensional turbulent flows, the eddy viscosity formulas require changes
to those for two-dimensional flows. Here they are defined according to Eqs. (5.7.4)
and (5.7.5).
Sample calculations for an infinite swept wing having the NACA 0012 airfoil cross
section with a sweep angle of = 30°, an angle of attack of = 2°, chord Reynolds
number Rc = 5 × 106 and transition location at x/c = 0.10 are presented in the
accompanying CD-ROM. See also Problem 8.7.
For three-dimensional turbulent flows, the eddy viscosity formulas require changes
to those for two-dimensional flows. Here they are defined according to Eqs. (5.7.4)
and (5.7.5).
Sample calculations for an infinite swept wing having the NACA 0012 airfoil cross
section with a sweep angle of = 30°, an angle of attack of = 2°, chord Reynolds
number Rc = 5 × 106 and transition location at x/c = 0.10 are presented on the
companion site, store.elsevier.com/components/9780080983356. See also Problem
8.7.
[74]
Figure 5.24 shows the calculated results (solid lines) with experimental results (sym-
bols) and those obtained by Bradshaw’s method [74] (dashed lines). The cross-flow
angle which represents the departure of the velocity vector within the bound-
ary-layer from the freestream velocity vector was computed from
(5.7.7a)
(5.7.7b)
(5.7.8)
(5.7.9a)
(5.7.9b)
Here and denote the wall shear values in the x- and z-directions, respectively,
obtained from the solution of the infinite swept wing equations.
(5.7.10)
(5.7.11)
The results in Fig. 5.25 show good agreement with experiment at two x-stations.
However, with increasing distance they begin to deviate from experimental values
and at x = 0.650 ft, the agreement becomes poor.
The above results indicate what was already observed and discussed in relation to
the shortcomings of the Cebeci-Smith algebraic eddy-viscosity formulation, that is,
it requires improvements for strong adverse pressure gradient flows. As discussed
in subsection 5.4.2, the improvements to this formulation were made for two-di-
mensional flows by allowing in the outer eddy-viscosity formula to vary. A similar
improvement is needed to the formulation for three-dimensional flows.
> Read full chapter
Figure 5.24 shows the calculated results (solid lines) with experimental results (sym-
bols) and those obtained by Bradshaw's method [74] (dashed lines). The cross-flow
angle which represents the departure of the velocity vector within the bound-
ary-layer from the freestream velocity vector was computed from
(5.7.7a)
(5.7.7b)
The streamwise component of the local skin-friction coefficient cfs was calculated
from
(5.7.8)
(5.7.9a)
(5.7.9b)
Here wx and wz denote the wall shear values in the x- and z-directions, respec-
tively, obtained from the solution of the infinite swept wing equations.
Data of Cumpsty and Head
In this experiment [73] the boundary-layer development was measured on the rear
of a wing swept at 61.1°. The boundary-layer separated at about 80% chord. The
measured profiles were affected by traverse gear “blockage,” probably because of
upstream influence of disturbance caused to the separated flow by the wake of the
traverse gear.
Fig. 5.25. Comparison of calculated (solid lines) and experimental (symbols) results
for the data of Cumpsty and Head on the rear of a swept infinite wing.
(5.7.10)
(5.7.11)
The results in Fig. 5.25 show good agreement with experiment at two x-stations.
However, with increasing distance they begin to deviate from experimental values
and at x = 0.650 ft, the agreement becomes poor.
The above results indicate what was already observed and discussed in relation to
the shortcomings of the Cebeci-Smith algebraic eddy-viscosity formulation, that is,
it requires improvements for strong adverse pressure gradient flows. As discussed
in subsection 5.4.2, the improvements to this formulation were made for two-di-
mensional flows by allowing in the outer eddy-viscosity formula to vary. A similar
improvement is needed to the formulation for three-dimensional flows.
Figure 5.27 shows a comparison between calculated and measured velocity profiles
on the line of symmetry (Fig. 5.27a) and off the line of symmetry as described in [76].
In general the agreement with experiment is satisfactory.
Fig. 5.27. Comparison of calculated (solid lines) and measured (symbols) velocity
profiles (a) on the line of symmetry and (b, c, d) off the line of symmetry for the East
and Hoxey flow.
The experimental data of Meier et al. consists of surface shear stress magnitude and
direction vectors and velocity profiles over a range of angles of attack. Figure 5.28a
shows a comparison of calculated surface shear stress vectors in laminar flow at =
10°. The magnitude of the shear stress vector is proportional to the shear intensity.
The agreement between the calculation and measurements on the windward side
is generally good, although there are some differences that are partly due to the
use of inviscid potential flow in the calculations, whereas the measured pressure
distribution shows viscous-inviscid interaction effects. It is clear that the laminar flow
is separated on the leeward side of the body at some distance aft of the nose. The
origin or nature of the high shear intensities leeward of the separation line cannot
be determined from calculations because calculations based on external flow that is
purely inviscid is not expected to account for strong interactions.
Fig. 5.28. (a) Measured (→) and calculated (→) distributions of wall shear stress vectors
(cf = w/1/2 u 2) for laminar flow and (b) for laminar, transitional and turbulent flow
on a prolate spheroid at = 10° [81].
Figure 5.28b shows wall shear vectors for laminar, transitional and turbulent flow
with natural transition. In general, the calculated and measured results are in
agreement with discrepancies (which are small) confined to the region close to
the specified transition. More quantitative comparison with the imposed transition
experiment is afforded by Fig. 5.29 which displays circumferential distributions of
wall shear stress at four axial locations. The calculated results display the correct
trends and are within 15% of the measured values with discrepancies tending to
diminish with downstream distance. A sample of the velocity profiles is shown in
Fig. 5.30 and corresponds to x/2a of 0.48 and 0.73, and again the agreement is
within or very close to the error bounds of the measurements, except in the regions
where the inviscid velocity distribution differed from the measured one. Additional
comparisons of calculated and experimental data are given in [81].
Fig. 5.29. Measured (dashed line) and calculated (solid line) resultant wall shear stress
values on the prolate spheroid at = 10° [81].
Fig. 5.30. Comparison of calculated (solid lines) and measured (symbols) streamwise
us/Us and crossflow us/Us velocity profiles [81].
> Read full chapter
11.5.2 1980s
• Stealth Fighter,
• Tacit Blue (Whale) Stealth Bomber testbed, new stealth approach leading to
B-2 Bomber,
• Joint STARS,
• Pilot's Associate,
Wing problems
T.H.G. Megson, in Aircraft Structures for Engineering Students (Sixth Edition), 2017