Report Metodo Schrenk

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

U.

S, DEPARTMENT OF COMMERCE
National Technical InformationService

N62-57948

A SIMPLE APPROXIMATION METHOD FOR OBTAINING


THE SPAN'WISE LIFT DISTRIBUTION

National Advisory Committee for Aeronautics


Washington, D.C.

Aug 40
NOTICE

THIS DOCUMENT HAS BEEN REPRODUCED FROM

THE BEST COPY FURNISHED US BY THE SPONSORING

AGENCY, ALTHOUGH IT IS RECOGNIZED THAT CER-

TAIN PORTIONS ARE ILLEGIBLE, IT IS BEING RE-

LEASED IN THE INTEREST OF MAKING AVAILABLE

AS MUCH INFORMATION AS POSSIBLE,


,11,
NATIONAL ADVISORY COMMITTEE FOR AERONAUTICS

TECHNIGAL MEMORANDUM NO. 948

A SIMPLE APPROXIMATION NETHOD FOR OBTAINING

THE SPANWISE LIFT DISTRIBUTION"

By 0. Schrenk

PRELIMINARY P_EMARES

In this paper a simple approximation method is pre-


sented for rapidly computin_ the lift distributions of
arbitrary airfoils. The numerical results are compared
with those obtained by an exact method and for many pur-
poses show a satlefac_ory degree of accuracy. The latter,
_e for all practically occurring cases, can be estimated at
the start of the computation work with the aid of the com-
parison examples _iven.

The method described below enables the approximate


determination of the lift diztributlons in a few minutes
with an accuracy sufficient for many purposes. It is also
characterized by a certain simplicity which is useful in
the clarification of many questions and is in accord with
the englneerls point of view. Finally, the method is ap-
plicable to cases for which all other methods are entirely
unsuitable (for exampie, wings with end plates).

Simil_r methods, as the author subsequently found,


have already appeared elsewhere. The surprising acc-aracy
of such simple methods is, however, generally unrecognized,
so thmt a presentation of comparlson computatLons which
provide a measure of the degree of accuracy obtainable,
should lead to an extended application of the method.

The author wishes to express his appreciation to his


co-worker at G_ttlngen, Mr. N. Hiorth, for carrying out
the laborious computations required for the comparison.

-, I!
*"Ein einfaches l_aherung_verfahren sur Ermittlung yon Auf-
triebsverteilungen lanes der Tragfl_elspannwefte."
Luftwissen, Bd. 7, Nr. 4, April 1940. pp. 118-120.
2 NACA Technical Memorandum No. 948

YUNDA_ENTAL IDEA 0¥ THZ METHOD

The plausible assumption is made that the real lift


distribution lies between an_ideal distribution independ-
ent of the wing shape and a distribution determined in a
simple manner by the wing shape. The ideal distribution
is that with minimum induced drag and constant induced
downwash velocity - that is, for the usual monoplane, the
elliptic distribution; while the distribution dependent
on the the shape is proportional to _ t at each position
of the wing.

COMPUTATION PROCEDURE

Inthe case of the untwisted wing the angle of attack


is not absolutely necessary. There is drawn instead (for
the monoplane) the semiellipse of equal area with the
chord-dlstribution curve, and the llft distribution is ob-
tained by forming the arithmetical mean between the two
curve s.

In the case of twisted wings (and similarly for wln_s


with aileron or flap deflection) there is first determined
the zero llft direction of the entire wing. A sufficient
approximation for this is the direction of the mean aero-
dynamic twist

8(x) tCx)
S=
t

Where the bars denote mean values, 8(x) and _ are


the twist angles between the reference direction of the
entire airplane and the local and mean zero lift direc-
tions, respectively. For all further computations, the
angles of attack and twist are reckoned from the zero llft
direction _Iven by 8.

_ith angles reckoned as indicated above, the llft is


decomposed into a component without twist and a twist com-
ponent without lift. The second component is determined
on the basis of a mean value which has the zero line in- A
stead of the ellipse as the ideal distribution, and for
which the twist angle must always be considered.

The general case with tWAst is most conveniently com-


NACA Technical _emorandum No. 948 3

puted with the aid of the following formula, which requires


no explanation:

dA i
q d__c_
dm
4
_ tCx) +-._
W
1- -_"'_

1 dca 8(x) t(x)


+_ q _--i

In the above relation dc_


-- and _Ca can generally
d_ d_co
be taken as constant along the span. A somewhat more accu-
rate formula is given below.

dC a
The trial computation of _ca
dm and d.--_ may be some-
what refined by the introduction of a correction factor
fe
(1 + _) setting:

%
de a

de_
1 + dca F (l + _)

The constant _ for various taper ratios was determined


by trial in s_ch a way that the lift determined with our
approximation method is _iven as correctly as possible
(fi_. 1). In the case of the rectangular wln_ the value
thu_ determined a_rees with the theoretically determined
value of Glauert. For other taper ratios the agreement
has not been checked and in this connection is not required.
A further refinement in the value of K, nevertheless,
seems unnecessary.

The computation procedure thus consists of the follow-


ing steps :

a) Coz_putat_on of 8 by forming the mean.

b) Trial computation of dca/d_ and dCa/dO_.

c) Computation of the lift distribution by the


formula for d.A/dx.
4 NACA Technic_l Memorandum No. 94B

ACCURACY OF THE METHOD

The comparisons _iven in figures 2 to 13, between the


accurate value_ computed by the method of Multhopp and the
results of the approximation method, show in _eneral a sot-
isfactory, and to some extent even surprising, agreement.
The error arising through the assumption of a constant
value of dCa/dC_ along the span in unfavorable cases, can
be eliminated by computing a mean value

_c,(x) t(x)
dc a dc_

and then computin_ the lift by the only slightly altered


formula

dx 2 q _ dc_. t(x) + d_ T, t 1 - J

I dCa(X)
+ _ q d_= 8(x) tCx)

The sharp difference between lift without twist and twist


without lift, to be sure, no longer arises. This correc-
tion has not been applied in the example here given.

Further refinement through additional computations,


at the expense of briefness and simplicity, does not ap-
pear to be of advantage. The deviations, moreover, occur
at such positions where the exact theoretical solution
does not agree with actuality.- for example, for unrounded
wing tips and at transition positions of ailerons and
flaps.

The deviations at the transition positions of ailer-


ons _nd flaps can be readily balanced by hand with the aid
of examples _iven in the figures.

POSSIBILITIES OF APPLICATION

The method, as mny be seen, is suitable for all prob-


NACA.Technical Memorandum N0..948

lems where too great accura'cy is not required; that is, in


general, for investigations with regard to the maximum
lift coefficient Camax, stalling, and static equilibrium
problems. By the decomposition into an ideal, plan form,
and twist distribution, simple and time-savin_ relations
may be set up for frequently repeated computations of bend-
ing moments, transverse forces, and torsional moments along
the sp_n. For the computation of the induced drag and for
the downwash computation, the method is not directly appli-
cable.

The method is suit,__ble for the determination of lift


distributions also in cases for which the usual methods
f_il completely. Thus, it is applicable to monoplane
wings with end plates. Ideal distributions that take the
place of the ellipse c_n be computed on the basis of an
investigation of the Aerodynamic Experimental institute
(reference l) - the distributions there given being for
smallest induced drag and constant downwash. The relation
for the lift distribution now becomes:
@

&A 1 dca __ _I_ ....


_i : _ q 2_- _ t(x) + i ----
b

3_. f f(x) d(x)


b

I dC a
+ _ q -_-= 8(x) t(x)

_here f(x) is the function denoted by Mangler as the


"ideal function" for the giwBn case with end plates,

The method should likewise find application to bi-


planes and other arrangements.

SUMMARY

The approximation method described makes possible


lift-distribution computation_ in a few minutes. Compar-
ison with an exact method shows satisfactory agreement.
The method is of greater applicability than the exact
#L

6 NACA Technical Memorandum No. 848

method and includes also the important case of the win_


with end plates.

Translation by S. Reiss,
National Advisory Committee
for Aeronautics.

REFERENCE

i. Man_lor, W.| The Lift Distribution of Wings with End


Platos. T.M. No. 856, NACA 1938.

!
|
NACA Technical Memorandum No. 948 Figs.l,2,3,4,5,6

0.2

\i Figure

the formulas
1.-
The values
that occur
for dA/dx.
of K
in
For

\ [ /"

ta, tl
non-trc_ezold
the n_rest
taper ratio
shaped wings
value of the
il used for the
0 0°2 0.4 0.6 0.8 1.0 determination of K.

Approximation
Erect by Mu_Ithopp's method

o._ _.I

0 C2 0.6 1.0
_

%
0.'1 _-b,...
" _ Trapezoidal _
| t a. i:i=0.25 .._-_
0
0 0.2 0.6 1.0 0.2 0.6 1.0

Fi_:es 2 to 5.- Comp_.rlson curves with b2/¥=6.67 for various


ttp_ ratios. The comparison foz ta. ti=0.5 is
given on Fig i. 6. 1% !t_'ist" ti

/_.Z::.r._ --._ _cal e "


1.4 / "-_._..-. _ en/ar_ed -- Approximation
--. "'-. 3' =0.I 1
1.0 Exact by
Mul thopp 's
method
0.6 ,o '\' L =0._ J
• o_ _. ,./ b2

0.2
Trapezoidal ta:ti=0.5 _--->b/Z
I t I I I I I

0 0.2 0.4 0.6 0.8 1.0

figure 6.- Comparison curves for ta_ti=0.5 for varioud values


of F/b 2. No twist.
NACA Technical Memorandum No. 948 Figs.7,8,9,10

.0.16

Figure 7.- Comparison curves


Jl for a wing without
--o.o8 _- : twist with cut-out in center.
All three curves enclose
] I I , t I, ,, -._
equal area•
Key for Fig. ?.
o 0.2 0.4 0,6 0•8 1,0
.... Chord distribution
Approximation
Exact by Mul_hopp's
method.
Figure 8.- Comparison curves
0.16
for a rectangular
wing with semicircular

0.08 o -k rounding, without twist. The


deviation of the approximation
method at the wing tip for the
non-rounded rectangular wing
..... [ _ t I ,, ! here almost completely
0 0.2 0.4 0•6 0•8 1.0
disappears.

Key for Figs. 8,9,10.


Approximation
Exact by Multhopp's method

Figure 9•- Comparison curves


for a wing wlth
ta:ti=O.5 and bB/F=6.67 for
various values of Ca with a
linear twist which is 0 ° at
the center and S ° downwards
at the wing tip.

Joo,____.%,__
o.2 o.4 T2"_o

- O. 016 Ca=0
1 Figure I0.- Larger scale
_ _ __> X
-'__0.2_--_._ O.6 b'72 1.0 representation
of the curves for Ca=0 fr@m
v

_ _. Fig, 9.
U

--0.016
)
NACA Technical Memorandum No. 948 _tge.ll,12,13

M O. 024 _a=O
4_

Figure ii.- Results for Ca--O


as in Fig. I0 with
1.0 a twist which similarly amounts
,, ,,[
/ to S ° at the wing tip but
increases parabalically.
0.016-

-- Key for Figs. Ii,12,13-

Approximation
Exact by Multhopp's method

tkl=O. 2 "t ._ .c
9 8kl =14.3 o _J 0.16

0.08
/
--'-'_6" I I I

),.-0.8 _' ' '- 0.2 0.6 1.0


.0.08

Figure 12.- Comparison curves for a rectangular wlng with b2/F=5 for
an aileron deflection B=0.25_14 ° with and without lift.
The corners of the approximation curve can practically be well rounded
off by hand following the example given and thus considerably better
agreement 18 obtained• The rolling momentm of the non-rounded off
approxlmatlon curve very well agree with the exactly computed values.

0.32

0.24

\ 8kl 60 tkl O.
.Figure IS.- Comparison
O. 16 curves for a
tapered wt.ng with ta:tt=
0.5 and b2/F=6.67 with
0.08
60 ° deflected split flap
along center half of
0 span.
0.2 0.4 0.6 0.8 1.0

You might also like