You are on page 1of 10

An Analytical Representation of Delta Wing Aerodynamics

Shiang-yu Lee, Ph. D.*


The Boeing Company, Seattle, WA, 98124-2207

The tangible evidences of available experimental data for delta wings are examined for
their physical significances and implications. Several primary aerodynamic features appear
to have not been clearly identified and represented properly in existing analysis models.
Most apparent is that test data indicates that, at the moderate aspect ratio of 2.31 for an
equilateral triangle, the aerodynamic force is normal to the wing surface with no obvious
non-linear component. In addition, the total lift value is significantly higher then the classical
linear theory prediction. Based on these observations, a mathematical model is established
by extending the circulation formation model proposed by this author in an earlier paper.
This, together with an empirical formula for the non-linear component, produces a result
closely matching the experimental values for all cases evaluated. A comparison to the
circulation parameters also confirms the current model accuracy. Based on these results, a
discussion is extended into the slender wing theory and suction analogy approaches. This
paper also points out the use of slenderness ratio as a better representation of aerodynamic
properties for planforms other then rectangular wings.

Nomenclature
A = wing aspect ratio
b, c = wing span and chord
CL, CD, CN = lift, drag and normal force coefficients
CT = leading edge thrust or suction force coefficient
A, B, etc. = points on the wing surface
g = circulation similarity function
K = conical flow (Sychev) similarity parameter, tan / tan
KV = Suction Analogy vortex lift parameter
l = leading edge length
q = dynamic pressure
S = wing surface area
V = main stream velocity
w = velocity normal to the wing surface
x, y, z = Cartesian coordinates
wE, wF, wG = velocity components contributing to circulation formation toward the wing edges
= angle of attack
= apex half angle of a delta wing planform
= half base angle of the delta planform or half of the wing sweep angle ( /2)
, , = local coordinate measures
= circulation or vortex strength
= fluid density

I. Introduction

T his paper presents an analytical representation of lift and drag parameters for the delta wing planform. Because
of their geometrical feature, delta wings exhibit excellent lifting capability and also, in high speed range, incur
much lower sonic pressure drag than straight wings. Furthermore, with the presence of separation vortices inherent

*
Engineer Scientist, Boeing Information Technology, PO Box 3707, Seattle WA, 98124-2207/MS 2R-97, AIAA
Senior Member.

1
American Institute of Aeronautics and Astronautics
to the planform shape, they also have excellent stall characteristics for high angle of attack maneuvers. In practice,
the planform concepts have evolved into more elaborate forms, such as leading edge extensions, ogee and Gothic
variable sweep wings, etc. Because of its practical importance as well as general academic interest, this planform
geometry has been studied extensively since the early days of aerodynamics developments. An exhaustive literature
listing can easily be found, for example, in Schlichting1 and we will only cite the references that have close bearing
to the current study.
It is generally accepted that the aerodynamic forces on delta wings, as on other slender geometry, includes the
linear, or potential flow lift and drag, as well as a significant contribution from non-linear or separation vortex flows.
For the linear force, the popular “slender wing theory” has been considered as standard representation for small
aspect ratios. This formula has been established through several different reasoning’s by Jones2, De Young3,
Kuchemann4, etc. More precise methods1 through lifting surface analysis or extended lifting line theory yield
modification for larger aspect ratios. There are also several empirical formulae representing the lift coefficient. Most
recently, Traub5 provided a modification of the
classical equation for moderately higher aspect ratios.
One assertion2 made regarding the slender wing
theory is that only the wing area with expanding span
would contribute to lift. Also, as indicated by De
Young3, the formula is insensitive to the real planform
geometry and other literature further quoted that the
induced drag is “always” at minimum and projected
from the lift at half the angle of attack. These claims
have been shown6 as not necessarily reflective of
physical realities.
The classical studies by Legendre7, by Brown and
Michael8 and by Mangler and Smith9 are formal
analyses, utilizing conical similarity to model the
cross flow plan including the separated vortex flow
fields. Typically, a flow field and pressure distribution
is determined at one cross section and the sectional lift
Figure 1, Delta Wing Lift Coefficients
coefficient is obtained by an integration over the span.
The lift is then applied to the entire wing planform for
the overall lift coefficient. Since the separation
vortices are considered within the analyses, these
results are taken as total lift values including non-
linear effects. These results generally overestimate the
lift values.
Bollay10established a trailing vortex effect model
for the narrow rectangular wings and Gersten11
followed with a lifting surface method for analyzing
arbitrary planforms. Both methods included the non-
linear aerodynamic force contributions and their
results approach the Newtonian force value in the
limit. However, the results are only presented as
numerical values and no separation of linear and non-
linear effects appear to be available.
The well known “Suction Analogy” approach was
proposed by Polhamus12 which prominently highlights
the vortex forces. His results demonstrate a very good
match to experiments for the overall lift. The approach
later was extended by many investigators to other Figure 2. Delta Wing Drag to Lift Ratio (A = 2.31)
planform geometry with equal success.

2
American Institute of Aeronautics and Astronautics
II. Assessment of Available Experimental Data
In order to establish a rational approach to the problem, this study attempts first to carefully examine the physical
characteristics from established experimental results which then guide a mathematical model consistent with the
observations. Following Kuchemann13, the experimental results for the over-all lift value at 15 degrees angle of
attack is compared to the classical lifting surface results for the linear lift component, Fig.1. At a glance, it appears
that the over-all lift is nearly parallel to the predicted linear value. If the difference is to be made up by a non-linear
component, it would require a coefficient value of around 2. At smaller ranges, say at A = 0.5, the deviation is
larger, which requires a non-linear component close to 3 to fill the difference.
Question arises when the Newtonian non-linear force, or what is commonly known as the sine square law, is
taken into consideration. It is well known that this force coefficient has a limiting value of 2 at zero aspect ratio
(which, at 15 degrees, gives a lift coefficient of 0.129) and diminishes when aspect ratio rises. Clearly, if such a non-
linear component is used in conjunction with the classical linear lift, the experimental results would not be matched.
In comparison, it is noted that the Suction Analogy approach produces a “Vortex Lift” coefficient (KV) close to
for small aspect ratio which then rises to higher values. This coefficient is obviously higher then the required value
of 2 observed. The reason that the total lift matches the experiment is that a “potential-flow lift for the condition of
zero leading-edge suction” is applied in the lift calculation, which amounts to reducing the classical lift coefficient
2
by a factor of cos .
In order to understand the problem better, the
experimental results for an equilateral triangle, A =
2.31, due to Winter14, is examined in detail. First, in
Fig. 2, the ratio of drag over lift is compared to the
value of tan . Surprisingly, the values are in
perfect match, indicating that all forces are normal
to the wing surface. This evidence shows that there
is no pure “potential lift” component or, in other
words, no leading edge suction force.
In Fig. 3, the lift and drag coefficients are
plotted directly against the angle of attack. Here, it
is observed that there is no obvious non-linear effect
but a linear normal force coefficient of
C N = would project a lift and drag matching
the experiment perfectly until, at higher angle of
attack, the lift begins to drop. This force coefficient
value of , although a pure coincidence, is Figure 3. Delta Wing Drag and Lift Coefficients
significantly above those predicted by the classical (A = 2.31, CNL = )
analyses. For this geometry, there is still a
significant presence of separation vortex over the
wing. The common view is that a separation vortex flow contributes primarily to the non-linear lift and drag forces.
But here we observe that clearly this is not the case as there is no obvious non-linear lift component.
Finally, included in Fig. 3 is also the plot of the induced drag predicted by the traditional drag polar relation,
C D = C L2 / A . Here it is seen that the test drag value is well above the prediction. This same observation has also
been asserted by Kuchemann13.
In summary, we have observed that:
1. The total experimental lift is significantly higher then those predicted by the classical theory linear lift
coefficients.
2. No appreciable non-linear lift component is observed for the equilateral triangle (60o sweep) wing and
thus it should not be expected for higher aspect ratios. This is observed even though a pair separation
vortices is indeed present in the flow.
3. The overall aerodynamic force is normal to the wing for the 60o sweep case and therefore this should
also be true for smaller aspect ratios. This implies that there is no drag free “potential lift” and
consequently, no leading edge suction for smaller aspect ratios.
4. The classical “Drag Polar” expression is not consistent with experiments in the delta wing
configuration.
These observations provide a guide to the mathematical model development.

3
American Institute of Aeronautics and Astronautics
III. Analytical Model
In developing the theory for rectangular wings, Lee15 established that circulations on a wing are the direct
consequence of the reflective flows on the wing surface and their strengths are determined by the wing planform
geometry. The circulation is, in fact, dictated by the orientation of reactive circulatory flows and does not depend on
the “induced flow” caused by the trailing vortices. The approach may be described as a “kinematically determinant”
problem where the circulation distribution is solved by prescribing the flow rule restrictions. By applying similar
reasoning, we attempt to establish the aerodynamic parameters for the delta wing.
Before going into the derivation, we can
first make some “guesstimate” observations.
As indicated by Lee15 for rectangular wings,
the centerlines and bisect lines play an
important role in determining circulation
developments. In the delta wing case, we can
also assume that the bisect lines issuing from
the vertices form bifurcation lines, separating
the reactive flows directing them toward the
nearest edge, as indicated in Fig. 4. Now, if
we assume that the circulation production is
“fully effective”, Fig. 5a, that is, in the region
near the trailing edge, the circulatory flows
are directed to the trailing edge with the
circulation circumferential velocity equal to
the plate reactive flow,

( ) = 2 bw tan (b / 2 ) (1)
Figure 4. Delta Wing Circulation Formation
By assuming the wing circulation to be the
same as the starting vortex left in the wake, which would be the base triangle circulation, this then leads to the lift
coefficient,

b/2
2 V ( )d
C L = L / qS sin = 0
= A tan (2)
(1 / 2) V 2 S sin

It is seen from this relationship that the lift coefficient converges to the value of 2 for very large aspect ratios,
consistent with the Kutta-Joukowski theory. However, when
A is small, the lift value is too high: for example, with A =1
and at 15 degrees, the value would be 0.63, well above the
experimental total lift coefficient of around 0.5.
As a second case, if we assume that the circulations
follow in the radial direction, Fig 5b, the lift coefficient, as
dictated by the stream-wise component of circulation, then
becomes,

b/2
V bw tan (b / 2 )d
CL = 0
(3)
(1 / 2) V 2 S sin
= (1 / 2) A tan

This value obviously converges to the slender wing Figure 5. Delta Wing Circulation Formation
theory and is consistent with the classical lifting surface Candidate Configurations
results. However, as discussed earlier, the results would
obviously be too low to adequately fit the experiments with a

4
American Institute of Aeronautics and Astronautics
reasonable non-linear component. On the other hand, when the aspect ratio becomes large, the lift coefficient
approaches to , half of the expected value. Even though Eq. 2 and 3 are both inadequate, they do provide the upper
and lower bounds to the lift coefficients and a guide to proper expression development.
In developing the new analytical representation, as guided by the experiment results, we assume that there are
indeed only two aerodynamic force components, the linear and the non-linear forces, that is,

C L = C NL sin cos + C NN sin 2 cos (4)

The linear force, like the potential aerodynamic lift, is assumed to be a consequence of circulation on the wing,
but, in accord with the experimental evidence, we assume the resulting force is normal to the wing instead of the
free stream. For the non-linear force, we utilize an empirical formula which gives the limiting coefficient value of 2
at zero aspect ratio and diminishes at the other extreme.
The separation vortices over a delta wing, initiated at
the apex, is a common flow feature. From a different
perspective, they could be viewed as the manifestation of
the circulation formation process. Here, after the
circulatory flows are created by the wing surface, they
separate and drift into the wake as vortices. And in fact,
the vortex strength variation along its axis should reflect
the magnitude of creation. It is therefore reasonable to
express, as shown in Fig. 6, that the strength for vortices
out-flowing toward either leading edge at point D at a
distance from the apex as,

( ) = 2 w tan (5)
This is to say that the full strength of the reactive flow
is expanded, near the apex, on forming the two separation
vortices. However, if we move further downstream,
because of the presence of the trailing edge, there is
nothing preventing the formation of a third circulatory Figure 6. Delta Wing Flow Configuration
flow around the rare edge as well. This third component of
circulation must cut into the strength of the side flowing
vortices. In fact, if we assume, as in Ref15, that the three
circulations have equal strength, then,

E = F = G (6)

Or,

2 DE × wE = 2 DF × wF = 2 DG × wG (7)

Since the reactive flow velocity is the only variable in


this equation, and since the third circulation must reduce
the strength of leading edge vortices, we can assert that,
wE + wF + wG = 2 w (8) Figure 7. Delta Wing Circulation
Equations 7 and 8 establish the basic relationships Flow Geometry
governing the circulation strength on the wing. Since the
purpose is to derive the lift coefficient for the wing planform, we shift our focus to the bisect line issued from the
base vertices, Fig. 7. Here we assume that the bound circulation remaining on the wing is the same as the base
triangle circulation after it drifts into the wake and after solving for the proper flow component, in Eq. 8, and
simplifying, we obtain the circulation at span location ,

5
American Institute of Aeronautics and Astronautics
2 w(b / 2 ) tan (b cos (b / 2 ) cos( ) / cos )
( )= (9)
b cos (b / 2 ) cos( ) / cos + (b / 2 ) tan / 2

Which, after applying the definition for lift and integrating over the span, leads to the slope of linear normal
force coefficient,
C NL = N L / qS sin
2 w(b / 2 ) tan (b cos
b/2 (b / 2 ) cos( ) / cos )
=2 V d /(1 / 2) V 2 S sin (10)
0 b cos (b / 2 ) cos( ) / cos + (b / 2 ) tan / 2
b/2 (b / 2 ) tan (b cos (b / 2 ) cos( ) / cos )
= (16 / S ) d
0 b cos (b / 2 ) cos( ) / cos + (b / 2 ) tan / 2
And, finally, upon normalization of the coordinate variable, we obtain,

1 (1 ) tan (2 cos (1 ) cos( ) / cos )


C NL = 4 A d
0 2 cos (1 ) cos( ) / cos + (1 ) tan / 2 (11)

Equation 11 is an analytical expression for the linear normal force on a delta wing, derived without any
approximation assumptions and it is believed that there is no previous recording of a similar equation. It can be
shown that the lift distribution is similar to a elliptic profile but the value is distinctly different from the slender wing
theory. Also, the expression does converge to the Kutta-Joukowski theory in the large aspect ratio limit. Since the
integral is relatively complex, a numerical integration is performed and the results presented in Table 1.
To facilitate a better understanding and utilization of the current result, an approximation expression is also
presented by interpolating the circulation generating velocity component as a linear function of the span to obtain,
for the circulation strength,

( ) = (4 / 3) bw tan (1 /b 2 2
/ b2 ) (12)

which leads to,

b/2
C NL = 2 V (4 / 3) bw tan (1 /b 2 2
/ b 2 )d /(1 / 2) V 2 S sin
0 (13)
= (7 / 9) A tan

This is an approximation formula easily evaluated with a calculator and the results are also listed in Table 1.

Aspect Ratio 0.5 0.7 1.0 1.5 2.0 2.31

Integration (Eq. 11) 1.07 1.43 1.9 2.55 3.05 3.49


Approximation, (Eq. 13) 1.08 1.44 1.91 2.54 3.01 3.26
Classical Solution12 (KP) 0.78 0.97 1.3 1.75 2.25 2.45

Table 1, Linear Normal Force Coefficients (CNL )

As can be seen, the approximation values stay very close to the numerical integration results but becomes
slightly lower in value as aspect ratio rises. Over all, the force coefficients are quite a bit higher in value then the
lifting-surface theory12 also listed in the table. In order to compare with experimental results, we utilize a simple
empirical relation for the non-linear force component, which is also assumed to produce a force normal to the wing,

C NN = 2 /(1 + A) (14)

6
American Institute of Aeronautics and Astronautics
The factor 2 is, of course, based upon the Newtonian force limiting value and the function is monotonically
decreasing as aspect ratio rises. Finally, the force coefficients, Eq. 13 and 14, are used in Eq. 4 to evaluate the
overall lift coefficient and plotted against experimental data12 for several aspect ratio cases in Fig. 8 through Fig. 12.
In Figure 10, one extra set of test data16 is shown. As can be seen, our results agree very well with experiments.

Figure 8. Lift Coefficient, A = 0.5 Figure 9. Lift Coefficient, A = 0.7

Figure 10. Lift Coefficient, A = 1.0 Figure 11. Lift Coefficient, A = 1.5

In Figure 13, the Winter test result is again presented comparing with current analysis, for which, in view of the
earlier assessment that there is no obvious non-linear contribution, it has been left out of the plot. Here, the
approximation result is still a bit higher then the experimental data. However, realizing the exploratory nature of the
current formulation, this situation should encourage further investigation into improvements in the representation.
Figure 14 shows the aspect ratio correlation for the current theory. Here, it is demonstrated that the diminishing
non-linear component, Eq. 14, complements the linear force to produce the overall lift in alignment with
experiments.

7
American Institute of Aeronautics and Astronautics
IV. Discussions
It is demonstrated in this study that, by applying the circulation formation theory, we are able to establish an
analytical expression for the delta wing linear aerodynamic force parameter. The primary premise is that, at any
point on a bisect line, a circulation would emanate toward each of the three boundary edges, all of which having the
same strength. The results show that the delta wing is capable of producing very strong circulations leading to high
values of linear lifting force, much higher than the classical slender wing or lifting surface theories predict. By
complementing the results with an empirical representation of the non-linear lift, we obtain the total lift values that
match experimental results closely, a situation indicating that these equations serve quite adequately as a candidate
analytical representation of delta wing aerodynamics.

Figure 13. Lift Coefficient, A = 2.31


( CNL = 3.26, Non-linear lift not included)
Figure 12. Lift Coefficient, A = 2.0

These results calls for a re-examination of the


classical analysis approaches. First, the lift
coefficient of slender wing theory can be obtained
by assuming an “elliptical load distribution” based
on w, the projected velocity component of the free
stream normal to the wing. The elliptic distribution
relationship dictates a maximum circulation value
of 2wb, which is only dependent on the span and not
any other geometrical feature. This, however, is not
a justification for geometry independence since it is
a consequence of the assumption. The circulation
value is not only smaller than the full circulation
capability afforded by the chord, wc, but also less
than what the span dimension is capable of
producing, wb.
To determine the adequacy of the theories, we
Figure 14. Current Theory Total Lift Coefficient
compare the circulation strength with the similarity
16
parameters established by Hemsch and Luckering .
For our analysis, Eq. 9, can be expressed in terms of
the g function based on the conical flow similarity (Sychev) parameter defined in that reference as,

g= /(Vl tan 2 cos ) = 4 sin tan sin /(3 tan 2 cos ) (15)

8
American Institute of Aeronautics and Astronautics
Similarly, we can also estimate the parameters for
the classical lifting surface analysis used by the suction
analogy12 approach which converges to the slender wing
theory at small aspect ratio. The results for aspect ratio
of unity are shown in Fig. 15. It is seen that the current
analysis model appear to be in very good agreement
with the experimental data presented in the reference16,
whereas the other models do not. And it is apparent that
the classical theory circulation is too low, whereas the
Smith9 theory is too high.
The vortex lift coefficient derived in the suction
analogy is debatable as normally one would expect an
aerodynamic parameter to vary according to the
geometry instead of being a constant. As it turns out, we
can trace how such a constant parameter comes about. It
has been shown13 that, according to the slender wing
theory, the leading edge suction force, or thrust, is equal
to the induced drag. In suction analogy, this suction
force is reverse projected to the leading edge by a factor
cos 2 to produce the vortex force. Since, at small
aspect ratios, cos 2 has almost the same value as Figure 15. Circulation Strength Parameters
tan or A / 4 , the net result is that the vortex lift
coefficient assumes the constant value of , independent of aspect ratio,

C LV = CT cos / cos 2 = C D cos / cos 2 =C L2 cos / A cos 2


A (16)
=( sin ) 2 cos / A cos 2 sin 2 cos
2
Knowing this, one would question how is the reverse projection of suction force justified, to which Polhamus
offered no detailed explanation.
The fact that, for the delta wing cases, the
aerodynamic forces are exerted normal to the wing
chord plane is also of interest. This means that the
potential flow draft free lifting force is not present. Of
course it has long been argued that, for a flat plate with
sharp edges, it is physically not possible for the flow to
exert a leading edge suction force on the wing.
However, since the normal force is linear to the sine
function of the angle of attack, it would have to be
related to the potential flow theory. We are interested in
finding out when, or if, in any situation, a drag free lift
element would become possible for delta wings. In
examining the Winter test data for aspect ratio of 4, it is
evident that the force is still normal to the wing. These
questions need to be resolved by more experimental
investigations.
As a final observation, it is seen that the aspect ratio
parameter is not really a very good measure for delta
wing properties or other non-rectangular planforms. Figure 16. Aspect Ratio and Slenderness Ratio
Since the aspect ratio is defined as the square of the span
divided by the planform area, which, for delta wings, ends up as twice the span divided by height. As illustrated in
Fig. 16, a delta wing of aspect ratio one is really much narrower a geometry than a square plate. Also, considering
other geometries, a circular wing would have an aspect ratio close to 0.75 instead of being unity. Although the
aspect ratio is merely a parameter of convenience, when it is used for comparison, it often provide a misleading

9
American Institute of Aeronautics and Astronautics
impression on the planform aerodynamic effectiveness. We can demonstrate this effect by comparing the delta wing
against rectangular wings based on the “slenderness ratio” scale, Fig. 17. Here, equal values are more representative
of the shape characteristics but what stands out is that a delta wing shows much higher lift values then rectangular
wings at the same ratio. It is therefore recommended here that, for general geometries, the slenderness ratio should

Figure 17. Lift Coefficients Represented in Slenderness Ratio

be used in place of aspect ratio for an undistorted representation of aerodynamic properties.

References
1
Schlichting, H, E. Truckenbrodt and H. Ramm, “Aerodynamics of the Airplane”, MacGrow-Hill, New York,
1979.
2
Jones, R. T., “Properties of low-aspect ratio pointed wings at speeds below and above the speed of sound”, NACA Report
1340, Washington, D.C., 1946.
3
De Young, N, “Spanwise Loading for Wings and Control Surfaces of Low Aspect Ratio”, NACA Tech Note 2011, 1950.
4
Kuchemann, D., “A simple method of calculating the span and chord wise loading on straight and swept wings of any given
aspect ratio at subsonic speeds”, Rep. Memor. Aero. Res. Coun., London, 2935, 1952.
5
Traub, L. W., “Extending Sleder Wing Theory to Not So Slender Wings”, Eng. Notes, J. Aircraft, 40-2, 2003.
6
Katz, J. and A. Plotkin, “Low Speed Aerodynamics,” Cambridge University Press, Cambridge, UK, 2001.
7
Legendre, R., “Flow in the Neighborhood of the Apex of a Highly Swept Wing at Moderate Incidences”, Aer. Re. Council
Report 16, London, 1954.
8
Brown, C.E. and W.H. Jr., Michael, “On Slender Delta Wings with Leading Edge Separation”, NASA TN 3430, 1955
9
Mangler, K.W. and J. H. B. Smith, “A Theory of the Flow Past a Slender Delta Wing with Leading Edge Separation”, Proc.
Roy. Soc. A, 251, 1959.
10
Bollay, W, “A Non-linear Wing Theory and its Application to Rectangular Wings of Small Aspect Ratio“, Z. Angew.
Math. Mech., 19, 21, 1939.
11
Gersten, K., “A Nonlinear Lifting-Surface Theory Especially for Low-Aspect Ratio Wings“, AIAA J. 1, 1963.
12
Polhamus, E. C., “Predictions of Vortex-Lift Characteristics by a Leading-Edge Suction Analogy”, AIAA J. Aircraft, 8,
1971.
13
Winter, H., “Flow Phenomena on Plates and Airfoils of Short Span”, NACA, TM 798, 1936.
14
Kuchemann, L. M., “Theoretical Aerodynamics”, 4th Edition, Constable and Company, Ltd., London, G.B., 1973; Printed
by Dover Publications, Inc., New York, N.Y., 10014, 1982.
15
Lee, S., “ The Vortex Impulse Theory of Wing Aerodynamics”, AIAA Paper AIAA-2004-4733, Presented at the 12th AIAA
Applied Aerodynamics Conference, Providence RI, 2004.
16
Levin, D. and J. Katz, “Dynamic Load Measurements with Delta Wing Undergoing Self-Induced Roll Oscillations,” J.
Aircraft, 21, 1984
17
Hemsh, M. J. and M. M. Luckering, “Connetion Between Leading-Edge Sweep, Vortex Lift and Vortex
Strength for Delta Wings”, J. Aircraft, 27, 1990.

10
American Institute of Aeronautics and Astronautics

You might also like