Lifting Line Theory Tutorial Example
Lifting Line Theory Tutorial Example
Lifting Line Theory Tutorial Example
This theory was developed basically to answer the following important aerodynamic
question : Given the shape of a wing, how can we predict the aerodynamic properties of
the wing? What is the most efficient wing shape for a given task such that the induced
drag can be minimized?
The theory was developed by Prandtl in Germany and Lanchester in England at the
beginning of the twentieth century as described below on the following web site
http://www.centennialofflight.gov/essay/Theories_of_Flight/Prandtl/TH10.htm
The theory is based on the assumption that even though the flow around an aircrafts
wing is really 3-dimensional, it may be satisfactorily approximated by a linear summation
of flows around the elemental aerofoils, which makes up the overall wing, where the flow
around each aerofoil is assumed to be 2-dimensional. This approach gives a reasonable
result provided that the model flow takes into account the effect of the vortex sheet,
1
which is shed at the trailing edge of the wing. The trailing vortex sheet induces a
downwash velocity, which varies along the span wise direction.
The wing is assumed to be a flat plate lying on the x-y plane. Therefore, the theory does
not take into account the wings thickness distribution. It is also unable to handle any
dihedral or sweepback angle. However, it is capable of modelling a tapered wing with
geometrical and aerodynamic twists.
Given the above data the theory must find the solution in the form of span wise wing load
or lift per unit span length distribution, the overall wings lift coefficient and the induced
drag coefficient of the wing. From the Kutta-Joukowski Lift Theorem it is known that lift
is directly proportional to circulation or vortex strength. Therefore, the theory must be
capable of predicting the span-wise bound vortex strength per unit length distribution.
The unknown vortex strength distribution, (y), is approximated by a Fourier series as
follows:
N
( ) = 2bV An sin (n ) (1)
n =1
b
where: y= cos (2)
2
2y
or = cos 1 (3)
b
The basic problem is how to calculate the unknown Fourier series coefficients or
amplitudes, An . The approximation using a Fourier series becomes more accurate as the
number of terms, N, increases. However, for hand calculation we must limit the value of
N to a very small number of 4 or less.
The lifting line equation that needs to be solved is
4b N N
sin(n ( y ))
An sin(n ( y )) + nAn = G ( y) 0 ( y) (4)
a ( y ).c( y ) n =1 n =1 sin( ( y ))
N
4b n
+ sin(n ( y )) . An = G ( y ) 0 ( y ) (5)
n =1 a( y ).c( y ) sin( ( y ))
2
Let us now define the following quantities
4b n
C ( y , n) = + sin(n ( y )) (6)
a( y ).c( y ) sin( ( y ))
A(n) = An (7)
D(y) = G ( y ) 0 ( y ) (8)
N
C ( y, n). A(n) = D( y ) (9)
n =1
The above equation contains N unknowns, namely A(n) for n=1 to N. It is therefore
necessary to apply equation (9) at N different control points or values of distance along
the span, y, so that we have a system of equations that can be solved simultaneously to
calculate the values of A(n). The points chosen should not include the wing tips, since
regardless of the values of the Fourier coefficients, the vortex strength distribution
equation (1) is always satisfied at those points. Selecting those 2 points will not provide
any new information regarding the values of the Fourier amplitudes. It is also
recommended that the midpoint (y = 0) should also not be selected as a control point, for
similar reasons. To get the most accurate result for a given number of control points, the
following method for selecting the control points is recommended. Firstly, N should be
chosen to be an even integer, such that N = 2M, say.
The N points along the span are chosen so that they are equally spaced. In other words
the span is divided up into N equal intervals, and the midpoint of each interval is chosen
to be a control point. The port (left) wing tip is located at y = -b/2 whereas the starboard
(right) wing tip is located at y = b/2. The coordinates of the control points are then given
as follows. For each value of k, from 1 to N, we have
b 2k 1
y ( k ) = yk = 1 (10)
2 N
Note that the control points are symmetrical about the plane of symmetry, y = 0. In other
words we have y(k) = y(N+1-k). If the lift per unit span length distribution is symmetrical
about y = 0, then its value at y(k) is equal to its value at y(N+1-k). Furthermore, it can be
shown that the Fourier amplitudes with even indexes in equation (1) are all zero
Equation (1) can always be rewritten in a way, which separates the even terms from the
odd terms as follows
3
N M M
( ) = An sin (n ) = 2bV An sin ((2m 1) ) + An sin (2m ) (11)
n =1 n =1 n =1
M M
( ) = 2bV A2 m1 sin ((2m 1) ) = 2bV A ( 2m 1) sin ((2m 1) ) (12)
n =1 n =1
In the discussion below it is always assumed that the load distribution is symmetrical.
The Lifting Line Equation, which is the system of equations (9), that must be solved to
calculate the Fourier coefficients, can now be written as follows
M
C (k , 2m 1). A(2m 1) = D(k ) for k =1, 2, , M (13)
m =1
2k 1
k = ( k ) = ( yk ) = cos 1 1 (14)
2M
4b 2m 1
C ( k , 2m 1) = C ( k , 2m 1) = + sin((2m 1) k ) (15)
a(k ).c(k ) sin( k )
D ( k ) = G ( k ) 0 ( k ) = G ( yk ) 0 ( yk ) (16)
It should be noted that c(k) is the aerofoils chord length at the station y(k), or ( k ) ,
whereas G ( k ) and 0 ( k ) are the geometric and zero lift angle of attacks at y(k).
The geometric angle of attack may vary as a function of y if the wing is given a
geometric twist. A wing without twist is one where the geometric angle of attack is
constant for all values of y, such that the leading edge and the trailing edge of the wing
are straight lines, which lie on the same horizontal plane when G = 0. A wing may be
given a washout, where the wing is twisted such that the leading edge of the wing tip
aerofoil is now lower than the leading edge of the root aerofoil (the root aerofoil is the
aerofoil located at the plane of symmetry if it is imagined that the fuselage is not there
and the two halves of the wing meet at the plane of symmetry). A wing with washin is
one where the leading edge of the tip aerofoil is now higher than the leading edge of the
root aerofoil, whereas the trailing edge of the wing remains on the horizontal plane. It
follows, therefore, that the chord of the aerofoil at y may have negative or positive
geometric angle of attack values when G = 0 at the wing root, depending on whether the
wing has a washout or a washin.
4
Let the height difference between the leading edge of the wing tip aerofoil from the
leading edge of the root aerofoil is htip , which is negative for washout and positive for
washin. It should be noted that the leading edge of the wing is required to remain as a
straight line. Therefore, the twist angle or the geometric angle of attack at y relative to the
geometric angle of attack at the wing root can be calculated as follows
h 2y 2k 1 htip
( k ) = sin 1 = sin 1 .htip = sin 1 1 (17)
c bc N c
Let the angle of attack of the wing or the aircraft be denoted by the angle of attack at the
wing root, and is given the symbol of . This angle obviously can be varied and
represents the attitude of the aircraft (when the aircraft is at a level cruising flight this
angle may have a small positive value of not more than 3 degrees). Using this definition
we can now calculate the geometric angle of attack at y as follows
G ( k ) = + ( k ) (18)
D ( k ) = 0 (k ) + ( k ) (19)
A wing may be given an aerodynamic twist as well as a geometric twist. This means that
the aerofoil shape at the wing root is different from that at the wing tip. The shape of the
aerofoil in between the two limiting stations is then determined by insisting that the wing
cross-section should have a smoothly varying shape along the span wise direction. Since
the aerofoil shape at the wing tip is different from that at the root, therefore the value of
the sectional lift coefficient as well as its zero lift angle of attack may also vary along the
span wise direction. Provided the variation of a ( y ) and 0 ( y ) are given, equation (15)
can still be used to compute the matrix coefficients, C ( k , 2m 1) , thus the lifting line
theory can handle such a problem.
The theory can also handle the problem involving a variation in the chord length of the
sections as a function of y, as long as the functional form of c(y) is given. This means that
the theory is also applicable for analysing tapered wing shape, so long as the quarter
chord line is normal or almost normal to the aircrafts longitudinal axis. Obviously the
theory is not valid for a highly swept wing. For swept wings we should use vortex lattice
method, but this will not be discussed here.
The taper ratio is the ratio of the chord length of the wing tip aerofoil to that of the root
aerofoil. Normally the value is less than 1, i.e. the sectional chord length decreases with
increasing distance away from the plane of symmetry, or with increasing magnitude of y.
The taper ratio is given the symbol of and thus
ctip
= (20)
croot
5
The chord length at y is then given by the following equation
ctip croot
c ( y ) = croot 2 .y
b
For computational purpose the above equation is more specifically written as follows
1 ctip 1
c ( k ) = croot 1 2 y (k ) = 1 2 y (k ) (21)
b b
The specification for aerodynamic twist is quite complicated, since it requires knowledge
of the shape of the aerofoil section at each station y along the span. In the absence of such
information, we can simplify the problem somewhat by requiring that the tip aerofoil
differs only slightly from the root aerofoil such that the lift coefficient and zero angle of
attack at y are given by the following linear relationships
atip aroot
a ( k ) = aroot 2 y (k ) (22)
b
0,tip 0,root
0 ( k ) = 0, root 2 y (k ) (23)
b
Obviously the properties of the root and tip aerofoils must be supplied as inputs.
The procedure for the application of the lifting line theory in evaluating the aerodynamic
performance of any kind of wing shape (with certain limitations) can now be summarized
as follows.
The following data must be supplied:
1. The wings angle of attack, , and either the value of the wing span, b, the wing
area, S, or the Aspect Ratio. These quantities are related as follows
S b2 b
b= and AR = =
cav S cav
2. Any 2 of the following 3 quantities: root chord, tip chord and taper ratio or
c
croot , ctip and . The relationship involving the 3 parameters is = tip .
croot
The average chord can be calculated as follows cav = 12 croot (1 + ) = 1
2 ctip (1 + )
3. The lift curve slope and zero lift angle of attack of the root aerofoil, aroot and 0,root ,
and also those of the tip aerofoil, atip and 0,tip .
4. The value of either the tip aerofoil leading edge height relative to that for the root
aerofoil, htip , or the twist angle of the tip aerofoil relative to the root aerofoil, tip .
6
If the twist angle of the tip aerofoil relative to the root aerofoil is given, say tip ,
then the leading edge height difference should be calculated as follows
htip = ctip .sin tip
1. Select the value of the number of control points on the port wing M.
2. For k = 1, 2, 3, .., M calculate the following
b 2k 1
y (k ) = 1 (24)
2 2M
2 y (k )
( k ) = cos 1 (25)
b
1
c ( k ) = croot 1 2 y (k ) (26)
b
atip aroot
a ( k ) = aroot 2 y (k ) (27)
b
0,tip 0, root
0 ( k ) = 0,root 2 y (k ) (28)
b
2 y (k )
( k ) = sin 1 .htip (29)
b.c ( k )
D ( k ) = 0 (k ) + ( k ) (30)
4b 2m 1
C (k, m) = + .sin ( ( 2m 1) . ( k ) ) (31)
a(k ).c(k ) sin( (k ))
M
C ( k , m ) .A ( m ) = D ( k ) for k = 1, 2, , M (32)
m =1
A simple direct method for computing A(m), the solutions of (32), is the Gaussian
Elimination Method. Other methods, such as the Jacobi or Gauss-Seidel iterative methods
may also be used.
7
4. After the Fourier coefficients, A(m), have been calculated we can now compute
the non-dimensional wing load distribution as follows
( y) M
ND ( y ) = = A ( m ).sin ( ( 2m 1) . )
2bV 1
(33)
2y
( y ) = cos 1
b
5. The wings lift coefficient can be calculated as follows
2
1 M
A(m)
e= where = ( 2m 1) . (35)
1+ m=2 A (1)
CL2
CDi =
. AR.e
1
CDi = k .CL2 where k is the induced drag factor and k =
. AR.e
Worked Examples
The Problem:
A hypothetical, conventional small aircraft has a wing area of 50 square meter and an
aspect ratio of 8. The wing has no dihedral, is unswept at its quarter chord line and is not
twisted aerodynamically, or its cross section is the same aerofoil shape all along the span.
The tapered wing with a taper ratio of 0.6 is twisted geometrically. The geometric twist is
such that the tip aerofoil section is at an incidence of 2.9 degrees when the root section
is at zero degree incidence. The aerofoils zero lift angle of incidence is 2.0 degrees and
the aerofoil's lift coefficient curve has a slope of 6 per radian. If the wings angle of
incidence is 2 degrees while the aircraft is cruising, calculate the following
(i) Calculate the wings span and mean chord lengths. Also calculate the root chord and
the tip chord lengths.
8
(ii) At the span wise stations y = 7.071 m and y = 3.827 m calculate the local chord
lengths of the wing (note that the aircraft'
s plane of symmetry is located at the spans mid
point or at the station y = 0)
(iii) Calculate the geometric angle between the local chord line at each of the span wise
station stated above and the chord line at the root section, due to the twist.
(iv) Calculate the overall lift coefficient of the aircraft at cruise condition using the lifting
line theory. Hint : Use 2 of the y-stations given above, say y = 7.071 m and 3.827 m, or
y = -7.071 m and y = -3.827 m as your control points in the calculation. Would you get
different answers if you were to do the calculation using all 4 control points along the
whole span, as compared to only using 2 points along the semi span and taking advantage
of the wing loading symmetry? Explain.
Also calculate the Oswald factor and the induced drag coefficient.
The Answer:
b = S . AR = 50 x8 = 20 m
S 50
cav = = m = 2.5 m
AR 8
2 2
croot = .cav = x 2.5 = 3.125 m
1+ 1 + 0.6
croot ctip
Starboard wing (y positive): c ( y ) = croot 2 . y , for 0 <y< b/2
b
c ( 3) = 3.125 0.4784 = 2.6466 m = c ( 2 )
c ( 4 ) = 3.125 0.8839 = 2.2411 m = c (1)
9
(3). Twist angle distribution
3.827
( 2 ) = sin 1 0.009486 x = sin 1 ( 0.01372 ) = 0.790
2.6466
Due to symmetry we have
2 x3.827
( 2 ) = cos 1 = cos 1 ( 0.3827 ) = 67.50
20
2 x3.827
( 3) = cos 1 = cos 1 ( 0.3827 ) = 112.50
20
2 x7.071
( 4 ) = cos 1 = cos 1 ( 0.7071) = 1350
20
4b 2m 1
C (k, m) = + .sin ( ( 2m 1) ( k ) )
a.c ( k ) sin ( k )
10
13.333 2m 1
C (k, m) = + .sin ( ( 2m 1) ( k ) )
c ( k ) sin ( k )
D ( k ) = 0 + ( k ) = 2 0 ( 2 0 ) + ( k ) = 4 0 + ( k )
Taking advantage of the symmetry of the wing loading distribution we have M = 2 and
we can use m =1 and 2 (port wing only) or m =3 and 4 (starboard wing only). If port
wing control points only are chosen, then we have
13.333 1 13.333 1
C (1,1) = + .sin ( (1) ) = + x0.7071 = 5.2069
c (1) sin ( (1) ) 2.2411 0.7071
13.333 1 13.333 1
C ( 2,1) = + .sin ( ( 2 ) ) = + x0.9239 = 5.6544
c ( 2 ) sin ( ( 2 ) ) 2.6466 0.9239
13.333 3 13.333 3
C (1, 2 ) = + .sin ( 3 x (1) ) = + x0.7071 = 7.2069
c (1) sin ( (1) ) 2.2411 0.7071
13.333 3 13.333 3
C ( 2, 2 ) = + .sin ( 3x ( 2 ) ) = + x 0.3827 = 3.1706
c ( 2 ) sin ( ( 2 ) ) 2.6466 0.9239
D ( 2 ) = 4 + ( 2 ) = 3.210 = 0.056025rad
This can be solved using Gaussian Elimination method ( see Appendix) as follows
1.The following procedure should be carried out for each value of k, starting from k=1,
then k=2,3,up to k=M-1 after each loop is completed
11
Then
D(k )
D ( k ) =
C (k, k )
The result is
A (1) + C (1, 2 ) . A ( 2 ) = D (1)
C ( 2,1) . A (1) + C ( 2, 2 ) . A ( 2 ) = D ( 2 )
C (1, 2 ) D (1)
where C (1, 2 ) = and D (1) =
C (1,1) C (1,1)
1.2 Eliminate all elements in the k th column of all rows under the k th row
For each value of j, starting with j=k+1 then j=k+2 etc up to j=M
C ( j , m ) = C ( j , m ) C ( j , k ) .C ( k , m )
Then
D ( j ) = D ( j ) C ( j , k ) .D ( k )
The result is
A (1) + C (1, 2 ) . A ( 2 ) = D (1)
C ( 2, 2 ) . A ( 2 ) = D ( 2 )
where C ( 2, 2 ) = C ( 2, 2 ) C ( 2,1) .C (1, 2 ) and D ( 2 ) = D ( 2 ) C ( 2,1) .D (1)
Even though in the above example the matrix is only 2X2 or M=2, the procedure is quite
general and can be applied for any value of M.
The procedure described above is the elimination part of the Gauss method. Note that the
end result is a triangular matrix with all diagonal elements having the value of unity,
and all elements below the diagonal are zeros.
Now we shall describe the back substitution part of the method.
After the last elimination process has been done, we will have the last equation to be of
the form
C * ( M , M ) . A ( M ) = D* ( M )
This can immediately be solved for A(M) as follows
A ( M ) = D* ( M ) / C * ( M , M )
Knowing the value of A(M), we can now use the equation directly above the M th
equation, which has the following general form
A ( M 1) + C ( M , M ) . A ( M ) = D ( M 1)
The above equation can be solved immediately for A(M-1) as follows
A ( M 1) = D* ( M 1) C * ( M 1, M ) . A ( M )
This process can be continued until all A(m) for m=M, M-1, M-2,,1 have been
computed. For A(M-2) the equation is as follows
A ( M 2 ) = D* ( M 2 ) C * ( M 2, M ) . A ( M ) C * ( M 2, M 1) . A ( M 1)
The generalized formula for back substitution is as follows
12
First calculate A ( M ) :
A ( M ) = D* ( M ) / C * ( M , M ) and then
For k = 1,2,,M-1
k 1
A ( M k ) = D* ( M k ) C* ( M k , M j ).A ( M j )
j =0
Normalization
1 1.3841 A (1) 0.007643
. =
5.6544 3.1706 A ( 2 ) 0.056025
Elimination
1 1.3841 A (1) 0.007643
. =
0 10.9969 A ( 2 ) 0.012808
0.012808
Then A ( 2 ) = = 0.001165 and
10.9969
Back substitution
13
The Oswald efficiency planform factor is
1 1
e= 2
= = 0.9546
0.001165 1.04754
1 + 3x
0.009255
The induced drag coefficient is
CL2
CDi = = 0.04168CL2 = 0.002253
. AR.e
Let us now investigate the effect of the number of control points on accuracy of the result
Let a third point be chosen on the port wing such that ( 3) = 22.50 . The value of y(3) is
b
then y ( 3) = cos ( 3) = 9.2388 .
2
The chord length at the third station is c ( 3) = 3.125 0125 x9.2388 = 1.970
9.2388
The twist angle is ( 3) = sin 1 0.009486 x = 2.550
1.970
Therefore D ( 3) = 4 2.55 = 1.45 = 0.025307 rad
0 0 0
The additional matrix coefficients required are given by the following equation
13.333 2m 1
C (k, m) = + .sin ( ( 2m 1) ( k ) )
c ( k ) sin ( k )
13.333 5
C (1,3) = + x ( 0.7071) = 9.2069
2.2411 0.7071
13.333 5
C ( 2,3) = + x ( 0.3827 ) = 3.9992
2.6466 0.92388
13.333 1
C ( 3,1) = + x038268 = 3.5898
1.9702 0.38268
13.333 3
C ( 3, 2 ) = + x0.92388 = 13.4950
1.9702 0.38268
13.333 5
C ( 3, 3) = + x0.92388 = 18.3234
1.9702 0.38268
14
The elimination process gives the following results
0.007803
A ( 3) = = 0.0002661
29.3221
A ( 2 ) = 0.001165 ( 0.54551) x0.0002661 = 0.00102
A (1) = 0.007643 ( 1.76821) x0.0002661 1.3841x ( 0.000102 ) = 0.009525
1 1
e= = = 0.9631
0.00102
2
0.0002661
2
1 + 0.034403 + 0.039024
1 + 3x + 5x
0.009525 0.009525
CL2
Induced drag coefficient: CDi = = 0.04133CL2 = 0.002368
. AR.e
It can be observed that the values of lift coefficient and Oswald efficiency factor
computed using Fourier series with 2 terms, differ from those obtained using Fourier
series with 3 terms, by approximately 3 percentage points. However, the error of the
induced drag is aproximately 5 percentage points. Obviously, if greater accuracy is
required, then we have to use a Fourier series with more terms. From the results obtained
in this example, it appears that the Fourier series approximation is only accurate to, let us
say I percent, if the number of the Fourier coefficients is greater than 3.
15
Appendix
Solution of a system of simulltaneous linear equations using Gaussian Elimination
method
16
Normalization for second row
Now the values of the unknowns A(1), A(2) and A(3) can be calculated as follows
Back substitution
17