·
,
THEORY OF AIRFOIL
RESPO~~SE IN A GUSTY ATMOSPHERE
PART II - RESPONSE TO DISCREl'E GUSTS
uR
CONTINUOUS 'I'URBLJ'LENCE
b~
L. '1'
0I'
ilotas
ti
THEORY
OF
·
AIRFO
IL
RESPONSE IN A GUSTY ATMOSPHERE
PART 11
-
RESPONSE
TO
DISCREI'E GûSTS OR CONTINUOUS TURBULENCE
by
L. T. Fi10tas
Manuscrip~
received
October,
1969.
ACKNOWLEDG EMENT
This investigation was suggested by Prof. H. S. Ribner; his
super-vision
and helpful
comments are gratefully
ack~owledged.
I am also indebted
to
Mr.
John
Leung for his consideraQle assistance with the numerical
compu-tations.
Financial
support was received from the National Research Council of
C
a
nad
a
under Grant No.
A2003 and by the Air Force Office of Scientific Research,
Office
of Aerospace Research, United States Air Force, under GrantNo.
67-o672A.
SUMMARY
The
response
of an infinite
span
airfoil in an arbitrary two-dimensional
upwash field is
studied analytically on the basis of the
'aerodynamic
transfer
function'
giving the pressure distribution in an inclined sinusoidal gust derived
in
Part I of this report.
When the upwash field ean be completely specified
(discrete gust case) the lift
response
is expressed
as
a quadrature
i~volving
the transfer function
.
When
only statistical
properties of the upwash can
ge
specified
(atmospberic turbulence) mean
square
val~es,
correlation functions
and power spectra
ane
similarly
expressed. Certain results are given in closed
form.
Expressions
giving
;
the the transfer function in terms of correlations
which may
be determined experimentally
are
given.
Previously published
experi-mental
data are compared with analytic results based on Part I. The detailed
expressions
deal
with
the lift response; however, the modification required
for
TABLE OF CONTENTS
NOTAT
ION
1
INTRODU
C
TION
RESPONSE TO
DISCRETE GUSTS
2
RESPONS
E
TO .cONTINUOUS TURBULENCE (Correlation Functions)
4
~R
A
NSF
ER FUNCTION
IN
TERMS OF EXPERIMENTALLY DEl'ERMINABLE
6
QUANT
ITIES
ONE-
DlMENSIONAL LIFT SPECTRUM
6
ANAI,.YTIC RESULTS FOR THE
EFFECTIVE
SEARS FUNCTION
7
SOME EXPLICI
T RESULTS
:
VERY
NARROW LIFTING STRIP
8
MEAN
SQUARE LIFT COEFFICIENT
9
T
HER
S
TA
+
I
~
TICAL
PROPERTIES
10
NJTE
ON
PI
TC
HING
MQMENT
11
CONCI..lJl~ING
R
EMA..."RKS
12
-
Ef~ER
EN
C
E
S
13
b
.
CL
CM
f
l
f
2
h
(
i,y)
NQI'ATION
airfoil semi-chord
lift
coefficient
time derivative
of CL
=
d
CL/dt
pitching moment
coefficie~t
typ
i
cal frequency
(eqn.
42)
"
"
(eqn.
43)
indicial response (eqn
.
10)
chordwise wave
number, non-dimensionalized by b
spanwise
"
"
"
"
"
b
L
turbulence integral scale, non-dimensionalized by
b
R
LL
li
ft
coefficient auto-correlation
(eqn.
11)
Rww
upwash
auto-correlation
(eqn.
13)
R~W
l
i
f
t-upwash cross correlation (eqn.
17)
S(k
l
}
Sears function
(Ref.
15)
Se(k
l
)
e
ff
ective Sears function (eqn
.
27)
T
(
k
l
,
k
2
)
l
ift t
r
ansfer function
(eqns.
5
and
6)
T
M
(k
l
,
k
2
)
m
o
m
ent
transfer
function
(eqn. 45)
u
x,
y
5
E
CP
LL
f
lig
h
t ve
l
ocity
gust upw
a
sh field
Four
i
er transfer
of
w
(eqn.
2)
s
pac
e
co-ordinates
at
rest relative
to
atmosphere,
non-dimen~ionalized
by b.
i
aspect ratio
of lift
sensiR~
element
nume
r
ical constant
=
1.2
Y\M
upwash power spectrum (eqn. 21
)
CP
L
W
lift up-wash cross spectrum (eqn. 21
)
cp
L
CPw
one-dimensiona1 lift spec
t
rum
(eqn
.
22
)
INTRODUCTIO
N
The
con~lnuingevolutionary trend
towa~dlarger and more flexible
ai
r
craft
is reflected
in a corresponding need for more refined analytical
tech-niques.
In this respect, d.ynamic response
charact.eristics associated with
flight
through gusty air
are particularly significant: problems associated with more
numerous
anel
easily excitable vibraticnal modes are aggravated
by
requirements
for
sU3tained
f
l
ight
t::'1
!'
ough
regions
of severe gustiness.
Calculation
of the response to random inputs, such as atomospheric
turbulence,
becm~.es
feasible
through use of power spect
r
al
meth
ods
(generalized
harmonie analysis)0
Tllese methods circumvent direct
treatm
ent
of random time
funct
ions
by relating
the statistics of a linear syste
m'
s response to those of
the excitation tffi-ough
a
non
-sta
tistical
'transfe
r
fu..11.ctiOl'l'.
The aerodynamic
aspect of
flight
through 8usts
-
discrete or
conti:nuous
-
is contained entirely
in
this
transfer function (re
lating the instantaneous
respons
e
to the downwash
in
a single spectral
compo:qent of the gust)
.
The fou.."1dations
fOT
the spectral approach to the prediction of aircraft
gust
res
po
nse
were
laid by
Lieprüann
1,2 and elaborated by Ribner 3: the subject
has
rece
iv
ed contirlU
al attention ever since
o
Reviews and
recent
developments
may be found in References
4
to
8
and
their
bibliographies.
A
funda:.'llEmta
l aspect of
the
analysis is linear superpositicm of
(weighted) responses
to all frequencies.
Cer"cain questions may however be raised
concerning the applicability
of usual aerodynamic methods at
very
high frequencies
-
where,
for
instance, a tÜne lag in fulfilJ.ment of
the Ktltta
condition is
pos
si
ble9o
When the turbulence scale
is relatively
large,
these
considerations
are
usuall
y
unim;>ortant,
"
as
ev~d8nced
by
excellent agreement of certain
calcu-lationE
with I'light test
datA.
,
.
Nevertheless, detailed experimental
yerifi
-cation
of the
th
eory
would be
very desirable; pa.rticularly if
h;
is
recognized
that
shor
t
take
-
off/landing
aircraft operating at low speeds and altitudes may
encoUt'1-cer tlu
'1m
lence wi th
au
o:::-der
of
roagni
tude sffialle
r
scale
(and
greater
intensity)
t!lan
is gene;::
'
ally considered t;ypicaL
The possibility of using
aeroelastically
scale~
mode
l
s in
specially fitted wind tunnels
for
determination
of frequency re.3ponse
provides
further
incentive for detailed comparison of
line
arized theory
and
measureme~t.Initial
expe~iments
using a simple configuration made by Lamson
lO
and
Hakkine:q and I\ichardsonll
were somewhat inconclusiveo Available instrumentation
prevented
accure.te determination
of
~'Urbulenceproperti
es;
accurate theoretical
estimates
for the
aerodynamic transfer
function
were
lacking.
A renewed attempt
to' put the power
spectral
technique
on
solid
experi-mental fcoting has been
tmdertaken at the University of Toronto
5Institute for
Aerospace
S
tudies
.
In
Part I of this report12
a
suitable transfer function
was
derived;
mOTe
refined
experiments are currently underway13.
The
present
work
concerns the
lift
response
of an infinite span
airfoil:
the main motivation was
to explicitly
link
up the transfer function with
the experiments
.
Toward
this
end, an
eÀ~ressiongiving the
response
to an arbitrary discrete upwash field is
first
formulat~d;
this
is
used to generate relations suitable for study
of
the
response
to
continuous
turbule~ce.
The transfer function is then
exp
ressed
in
terms of
correlations
which are anticipated from the experimentso
.
10 11
The
nrevious
experlmental
results
3are also compared with the
l
i
ft response, extension to any ot her linear response is straightforward
:
in
p
ar
ti
c
ular
,
a note giving the explicit modification required for calculation of
pitc
h
i
n
g moment response i
s
included.
RES
P
O
NS
E 'RO DISCREI'E
G
UST
S
.ó,
s
note
d
in the int
r
o
d
uc
t
ion
, t
he aerodynamic aspect of flight through
g
u
s
t
s is entir
el
y
con~ainedi
n
the 't
r
a
n
sfer function' relating instantaneous
r
e
s
po
nse
t
o u
pwa
s
h in an elementary spectral (Fourier) component of the gust.
A
l
t
ernat
e
ly~
th
e 'influe
nc
e functio
n
' expressing the response to a unit upwash
impulse
m
a
y b
e
cons
ide
r
ed as basic
.
(The two equivalent approaches a
r
e compared
in R
e
f.
7).
Co
nsi
de
r
an airfoil of infinite span flying through an atmosphere which
i
s initially in
a
dis
turbed state. Since the air-foil responds only to the
verti-c
a
l
c
om
p
onen
t
of v
eloc
i
t
y
in its own plane (li
n
earized the
o
ry is implicit
through-o
ut
), the d
i
s
turba
n
ce may be regarded as a two-dimen
s
ional upwash pattern w(x,y)
in the airf
o
i
l pl
an
e (Fig.l)
.
We
b
egi
n
b
y
deriving the fo
r
mal solution for the lift when w(x,y) is
co
m
p
let
el
y s
pecif
i
ed (discrete gust)
ru~d
the transfer function (or the influence
f
un
c
t
i
o
n)
i
s
fu~own.
In the following section a random pattern that is specified
only statis
tically (continuous turbulence) will be taken up
.
Let us initially suppose that the Fourier integral representation
( 1)
-co
is va
l
id.
(
A s
uf
f
icient condition would be, for example, if w were absolutely
int
e
grabl
e o
v
e
r
t
h
e plane). Application of the inverse transformation to (1)
giv
e
s
00+
k
y)
2
dx
dy
(2)
1
:::
- - -
2
(
2
7T)
J
'
J
w(x,y)
-00E
q
u
a
t
ion
(1
)
e
x
p
r
esse
s
the arbitrary field as the superposition of elementary
si
nu
soid
al c
omp
onents of the type (Fig.
2
,
af ter Ref
.3
)
w
(x,y)
e
wh
e
r
e th
e
amp
l
itude is given b
y
A W
"
w
·i(k
x
+
k
y)
1
2
e
(4)
B
ut
f
o
r
a
n
elementary component of the form (
3
), the lift coefficient on a strip
o
f wid
th
dy
centered on the point (x,y) may be written in the form
27T
w
(x,y)
e
u
An
analytical expression12
for the 'transfer function' T (generalized here to
apply
to
both positive
and
negative
arguments)
is
given by
-iksgnkl[sin~
-
]
e
(6)
[1
+
nk(l
+
sin2~
+
~
k
cos~)]
1/2
where
k
Since
responses
may be superposed the
lift
coefficient due to an
arbi-trary pattern of
upwash is formally given by
- 00
A similar formalism
for
the
pressure
di
s
tribution
may
be readily obtained in
the
same way
from
the result
s
of
Ref.
12.
We shall
also
req-y.ir'e the total
response of a
finite
wing segment.
Let
CL5
be the total
lift
coefficient carried by
a
strip of
w
idth
2b5 (Fig.
2
). Thus
if the strip
is centered
on y
=
y
(x,yj
==
1
25
CL(x,y) dy
Using equatioll
(7),
thi
s
may be
wri
tten
as
00C
L5
(x,y)
==
J J
h(x
-s
,
y-T])
w(s,T])
dsdT]
- 0 0where
00h(x,y)
==
~~U
J
J
T(kl ,k2)
- 0 0(
8)
(10)
The
function
h
may
be interpreted as the response of the strip to a unit impulse
of upwash located at
(x,y) (i.e. the 'influence function').
In
principle,
equation
(:7)
completely determines the lift history of
the
airfoil
as it
passes through some specified gust (a limited patch of uniform
downwash, for examp
le): in prac4ice, the poor convergence of the integrand
severley restr
icts this
application
-
whether
numerical or
analytical integration
methods are
contemplated.
Nevertheless, the
above
expressions form the basis
for
useful
results
applying
to flight through continuous turbulence: this
application
is
eonsidered next.
It
might
be noted, in passing, that expressions pertaining to lift on
a finite segment,
obtained
from
equation
(8),
can be interpreted as resulting
from application
of
strip
theory to a firüte span
wing
of aspect ratio 5.
F
o
r
e
x
ample, the
r
esult
for isotropie
turbulence will (in contrast to procedures
some-times labelled
's
trip
theory'
that assume uniform instantaneous downwash along the
spa
n)
be the
s
trip
theory
res
ult
without restrietion
on the ratio of turbulence
s
eale to
span.
RESP
O
NSE
TO CONTI
N
UOUS TU§BULENCE (Correlation Funetions)
If
the
upwash field
encountered by the
wing
is an indefinitely extended
field of turbulence
the
relations
derived in the previous sections are
inapplica-b
le
i
n
their p
r
e
s
ent form*.
If, however, the turbulent field is
assumed
to be a
sta
tiona
r
y
r
andom
tunetio~
of position in the
x-y
plane, the theory of generalized
harmonie analysis
l
shows
that relations
~hat
can be derived for statistical
pro-pe
r
tie
s
remain
valid
.
For
e
xampl
e,
the
correlation of the lift coefficients carried by two
identical
strips
of
wi
ng at different spapwise locations (Fig
.2
)
may
be defined
as
R
=
LL
CL
5
(x,Y) CL
5
(x + 6
x, y +
Ay)
(11)
Here the overbar signifies an ensemble
ave
ra
ge.
Up
o
n..
inserting
(
9
)
there results
00(12)
- ( ) ( )w
h
ere
-x=x+&
y=y+6y
If the
turbu
l
ence
is
homogeneous, the velocity correlation can be
expressed
in form
of
a Fourier
transform. That is
-()()
(13)
Inserting (13
)
into (12) and
performing
some
elementary manipulations
leads
to
*
A valid formulati
on
would
involve writing the integrals in the
Stieltjes form;
00
i
(kl~x
+
k
2
6y)
E
LL
11
dkl dk2 CP
~w
(kl,k2) e
-00 00[kl(x- s )
+
k2(Y-~)
)
{ 11
-i
}
X
h(x-s, y-Tj) e
d(x-S)d(y-~)
- 00x
{IJ
h(x-"y-~)
e
i[k1
(x-,)
+
k2(Y-~)]
d(x-,)
d<Y-~)
}
-00
(14)
But
equation
(10)
is in
the form of
a
two-dimensional Fourier transform; taking
the
in
vel'
se
co11
h(x,y
)
-co
21T
U
(15)
U
si
ng thi
s
relation, equation
(14)
becomes
~L
(16)
It
is
apparent
that
RLL
is a
function
of the
separatiouP
~ a
~
d
6y and not of
the
location
or
ori
eLltation of the elements. From its
definition
(11),
i t is
also
apparen
t that the
mean square
lift coefficient is oQtained from (16)
with
6x
=
6
y
=
O
.
Expressions involving correlations
between other quantities
may
be
ob
t
ai
ne
d i
n
a
s
imilar way.
For the present purposes,
we
shall
require
the c
r
oss
cor
:re
la
tion of lift
coefficient
wi
th vertical
velocity,
defined as
Using
(9)
and (10),
as
0021T
11
R
LW
U
CPww
-00CL
(x +6x, y +
6y) w
(x, y)
6
in
the derivation of (16), there results the relat
i
on
i(k
l
&
+
k
2
l::,y)
(k
l
,k
2
)
sink
2
6
T(k
l
,k
2
)
(18)
k
2
6
e
dk
l
dk
2
For a given
type of turbulence the power spectral density would be
:
known;
sLnce
the transfer
function
is also
k~own
(6',
the correlations (16)
a
nd (18) are
completely determinate. Alternately, if the correlations
were
determined experimentally,
the equations could ge inverted to express the
tr~s
fer
functi
o
n
in
terms of experimentally determi1able quantities - providing the
experimental
equivalent to the analysis of Ref. 12.
This aspect will be taken
up
next.
TRANSFER FUNCTIDN
IN TERMS OF EXPERIMENTALLY DETERMINABLE QUANTITIES
Equations (16) and (18) are both in form
of
two-dimensional Fourier
transforms:
they may be inverted to give respectively
5k
2
1/2
h~(kl ,k
2 )
I
=
21T
U
sin5k
['P
LL (k1 ,k2 )
]
2
cp
ww
(k k
l ' 2
'
]
.
IT(kl~k2)
U
5k
2
CPLW(kl ,k2
)
=
21T
sin5k
2
cp
ww (
kJ!t2)
(20)
where
00-i(k
6x
+
k
b,y)
CPpQ
(k
l ,k2 )
1
11
RpQ (6x, f::..y) e
1
2
d6x
df::..y
47T
2
-00(21)
P~
Q
=
L or W
Equation
(19)
gives the magnitude of the
transfer
function; equation
(20)
gives the
phase
as
well.
The power spectral densities in the right
hand
m
embe
rs
may
be determined by numerical integration
of experimentally
determined
corrleation
functions.
The
experimental apparatus
would
consist (as in References 10 and 11)
of
an airfoil spanning
a
wind
trunnel test section; turbulence producing grid
located
upstream . .-AJ:'.he
lift can be sensed by instrumented strips along the
span;
the velocity
by hot
wi
r
e
.
Under
the assumption of a frozen turbulence pattern,
the separati
on
6x
is identified
with a (dimensionless) time delay f::..tb/U in a
w
ing fix
ed
(labo
ra
tory)
reference
frame.
The
correlations
are then given by
~ltiplication
of
signals,
with adjusta9le time delay, from two instrumented
strips
at
vari
able
spanwise
spacing
(R
L
Ü
) or
a single strip and moveable hot
wire
(R
LWÁ.
(For
experimeRtal details, as well as a scheme whereby one
of
the
integrations
for the
power
spectral
density can qe performed directly by use
of
an el
e
ctronic
wave analyzer see
(Ref.13)).
Determination of
the lift spectrum
CltL
requires
two separate lift
sensing
elementso
In
their
experiments,
using a
single
element, LamsonlO and
Hakkinen
and Richardsonll measured
a
'one-dimensional'
spectrum containing less
information
; this spect
r
um
is
~he
subject of the next sectiono
ONE-DlMENSIONAL
LIFT SPECTRUM
We
ean
define (e.g. - Ref.
3
) a one-dimensional power spectrum obtained
from
the two-dimensional
functions
~yintegrating out
one
of the variables. Thus
00
(22)
- 0 0
For
the
l
ift, using
equation
(19 ),
the one-dimensional
spectrum
may be written as
00 ' )(k
)
=.4·i
J
2
(
Sink25
)L-(
kl,k)
CPL
1
if
I
T
( k1
,
k2)
I
k
2
5
CPww
dk
2
-00(23)
B;ut from
equation
(16
)
with f1y
==
0, it is
apparent
that
00
ik
l
&.
R
LL
(6.x,0)
:=J
CPL (kl)
e
dk
1
(24)
-·00'I'hus
CP
J
J
(
kl ) is the Fourier transform
of the
lift auto
-
correlation
on
a single
lif- s
en
si
ti
v
e strip (bf1x
==
f1t/U
1
and
may
then be compared to the experiments
of
References
10
and
11.
~he relatio~hi
p
between the
o
ne
-d
imensi
o
nal lift
and
v
e
locity
spectra
ma
y
be
formall
y written
as
2
CPL (kl)
=
I
H (kl)
I
CPw (kl
)
Under
t
h
e a
ss
umption
of
unif
o
rm instantaneous spanwise velocity (lifting
point
assump
t
ion
3),
~he
'on
e-
dimensional
transfer
function
'
H(k
l
)
would be given
by
1
b
2Tr
S
u
(~
I
2
(2
6)
where
S is th
e
well-known sinusoidal gust function
of
Sears
l5
(the fac
to
r b
in
(26
)
ari
s
e
s
becaus
e
the
wave
number k2
appearing
in
(22)
is non
-
dimensional
)
.
Comparing
(25
)
and
(26
)
we may define (as in Ref.ll)
an
'effective
Sears function'
t
y
(
Thus
S
allows for
the
two-dimensionality of the actual t,urb;ulence
on
the
one-dimension~l
respo~se).
The right hand side
of
(27)
wa
s
obtained from experimenta1 measurements
in References
10 and
11.
Using
(23)
and
an
appropriate analytical
express
i
on
for
the turbulence spectrum CPw'
the
corresponding
theo
r
etica
l
results
C~D
also
be
ob.tained.
This will be
taken up
~ext•
.Al~AL·Y'rIC
RESU
.L1'S
FOR
THE EFFECTIVE
SEAgS
FUN
CT
I
O
N
To obtain
explicit
results it is
necessary to
specify
thé
form
of the
tu.rgulence
spectrum
CPww'
A
commonly used
form
appropriate
to isotr
op
ic
turbulence
is
(28)
has
pointed out
2
that, provided the scale is consistently defined, analytic
r
esults (i
nv
olving weighted
integrals)
should not depend too greatly on the
exact spect
r
al shape assumed.
Using
(22),
(23)
and
(28) in (27) we get
=
(29)
where
from
(6
)
=
(30)
Us
in
g approp
ri
ate values for the parameters L and
5,
equation
(29)
was
evaluated numerically
for comparison with the experimental data of References
10
a
nd
'
,11. Th
e
results
are shown on Fig.
3.
•
The
two
data
points at the smallest values of k
l
from References
11
t
e
nd
to conflict
with the
results
of Reference
10
and the present theory which
indicate a marked flattening
for small kl' The experiment al results seem less
sensitive
to
scale length than the analysis would indicate. Despite a broad
gener
al
ag
r
eement~
scatter and uncertainty in the experimental conditions still
leave the issue
uncertain,
SOME
EXPLICIT RESULTS :
VERY NARROW
LIFTING STRIP
dk
2
Num
erical
integration
of
(29)
shows that / S
/2
is a very weak funct
i
on
of
5
(
=
aspect
ra
tio
of
lift sensing element) for
val~es of
5
less than about
.5 -
that
is~
for values
of practical interest. Simplified results for a
diff
e
renti
all
y
narrow
strip, obtained by setting
5
=
0
in the previous expressions
,
will
accordingly apply
approximately to experimental cases.
Taking
5
=
0
in
(29),
( 31)
dk
2
Results obtained by numerical
integration of (31) are compared with the absolute
squa
r
e of the
Sea
r
s function on Fi&.
4.
If the turbulence integral scale is
much
la
rger
than
the
chord
(L~oo),
I
S
1
2 behaves like the absolute square of the Sears
function
only
fo
r
low
values
of tEe frequency parameter kl : that is, the 'lifting
po
int
'
approximation2
,3
is valid when
L~oo
and k
For
large
v
al
u
e
s
o
f
k l ,
t
he lifting
po
int
theory
o
v
e
r
est
imat
es
the
lift
power spectrum for
a
ll
L
a
s
would
be
expected
(because
o
f
spanwise
cancel-latio~s
which
are
n
eglected).
In
~act,
it
may be
shown
from
(25)
that
(3
2)
in epen
d
e
nt
of th
e
scal
e
l
e
ngth
.
For small valu
e
s
of
k
l
, Fig
.
4 indicates
Th
e
limi. ti
n
g result
can
b
e explici
tly
obta
ined by s
etti
ng
kl equal t
o
zero
in
(31)
a
nd
carrying thr
ough the
integration. The resulting expression
is rath
e
r
c
omp
lex~
a
simp
ler
resul
t
is
o~tain
through
use of
th
e
fact
that
th
e
constant
po
!'tion o
f
t
he
curves
for
ISe
1
(Fig
.4)
int
er
s
ect
the
curve
f
o
r.L
-7.00at kl
~
l
/
L
.
When L
00and k
=
0, we
must
ha~e
Is 1
2
=
1.
A
n
expres
si
on
with
this
limi
t
fo:..
k
l
=:0
and higB freque
ncy
exp
ansi
on
~iven
by
(
32)
is
ea
sily
con-trhed~
Is
(k ) /2
~
e
1
2 2
$!} (
E
+
TT k
l
)
n
3TT
2
k
1
2
,c,
.
n
E
+
,
L
-700(3
4
)
Good
agreeme
n.t
with th
e
nu.>n
e
rica
l
results is obt
ai
n
ed
for
E
=
102 (Fig
.
4
).
fo~
approximate closed
f
o
rm
expression for the effective
Sears function
2
2
$
!}
(
E
+
TT
k
l
)
,
k
l
>
1
$n
E
+
3
TT
2
k
2
L
l
If:
e
(k
l)
12
-
(35)
$n
(E
+
TT
2
/L
2
}
1
_1, k
l
<
$.n
E
+
3';
/
L
2
L
This
expreseion is also
indicated
on Fig.
4.
MEAN
SQUAR
E
LTll'T COBF'FICIENT
The mean
square
lift c
o
ef
ficient
is obtained by setting
6x
and
~
bo
t
h equ
a
l to zer
o in
the
cor
rel
atio
n
eq~ation
(16
).
Or
eq~
ival
e
ntly,
from
(
2
4
)
For
t
h
e infinitesimal strip (5
=
0), with use of (28) for the turbulence
spectrumj t
hi
s may
be reduced to the form
1
+ nk(l +
sin2~
+Hk
c
o
s~
)
(37)
V
a
lu
es obtained
by
numerical integration of (37) are compared with
the lif'ti
:
i.1g p
oi
nt theory o
f Liepmann
1
on Fig. 5
.
The two results agree asymptoti
-c
a.
l
ly
for
larg
e
L (t
ur
b
u
l
ence
l
integra
l
scale much larger thaB the chord).
~pan
sion of
Liepmann's ex
p
r
es
s
ion g
i
ves
C 2
L
::::::
1
-which is
also in
d
ic
ated on Fig
.
5.
2
L
L (3 log 2H +
1
)
, L
-700(38)
Ifj
o
n th
e
ot
h
er hand j the turbulence integral scale is much smaller
than
t
he
chord$
equation (
3
7) gives the asympototic formula
'
C 2
L
2
L
log L
, L
-70
The
r
ru1ge
of val
idi
t
y
can be extended with the aid of (35)
:
the result
corres-ponding to the
leadirtg two te
rms
for small L is
:::::: 1
2"
log
(E
+
H
2
/L
2
)
log
E
+
3H
2
/L
2
This expres sio
n
ag
r
ee
s
w
i
th
v
al
ues obta
i
ned by direct numerical
i
ntegrat
i
on of
(37
)
for va
lues
of
L
up to
abou~G
3
-
as indicated on Fig.
5.
Wh
e
n
t
h
e
in
teg
r
al
s
cale L
i
s of the order of a chordllength - as
i~
may
be in lm., leve
l
~
c.
urb
ulence
-
the one
-
dimensional assumption o
v
erestimates
the mean
s
quare lif
t
by a f
ac
t
o
r o
f about 3.
o
'
rHER STATISTICAL
PR
O
P
ER
T
IE
S
Fur
t
her s
t
a
t
i
st
i
cal prope
r
ties of the lift response may be obtained
as weighted
in
tegrals
o
f
the tran
s
fe
r
function(e.g. Ref.
4)
.
The mean square
lif
t
&.~ri
-
ati v
e j
for
ins
tanee j can be w
ri
tten a
s
=
00U
2
J
k 2
b2
1
- 00(40
)
If the
lift
fluct
uation
s
are
a Gaussian process, on
the
average
the lift coefficient
exceeds some level CL
per
Q~it
time
N
times,
where N is given by
the expres
s
ion
4
fN
==
1
27T
C 2
L
2~
L
l
(41)
Ir
CL
is
large enough (CL>
2i CL
2
),
then
N
is the average number
of
peaks per
unH time
g
reat
er than CL.
From
(38)
it may be seen that the
ratio
~1
=[
é
L
2
/
C
L
2
1
1/2
(42)
w
hich
has
the
dimensions
of frequency is proportional to
the
number
of
zero cr
os
s-ings
w
ith
p
osi
tive
slope
p
O
er
unit time. T-hus
f l
may be considered a tY}>ical
fr
eque
ncy
associated
with
tqe lift
fluctuati
ons.
Alternately, since
the
i
ntegral
scale
L
may be interpreted as a typical
(dim.
en
si
onlt~ss
here) wavelength associated wi
th the
turbulence, a
point
o
n
the
w
ing sho
u
ld
experience a
tY}>
i
cal
frequency
"
U/Lb
(43
)
Th
u
8
f l and f
2
are
differe~t
estimates of the typical
lift
fluctuation freque
ncy
;
if
th
e
y
are
compa
tib
le (viz
:
proporti
o
nal), their
ratio should
be
r
e
latively
in
de-pend
e
n
t
of
the
scale
l
ength.
Results for the
5
=
0 case
-
obtained by
~u.me
ri
cal
integration
of
(3
7)
and
(49),
with
(28)
for
the
tu
rbulence spectrum -
are plotted
on Fig
.
G
.
If
re
qui
red
,
the data of
Figures
5
and
6
may
oe combined
to
get
N
.
NCJl'E ON
PITCHING
MOME...l'fr
The methods
applied in prev
io
u
s
sections
to t
h
e calculation of lift
res
p
onse
are
al
so
applicable to calculation of
a:ny
other
linear
response.
I
t
is
necesGary only
t
o r
eplace the lift tra...llsfe
r func
tion
'1'
b
y
one appropriate to
the
response in
question.
The
p
i
teh
ing m
o
men-I;
re
sponse
i
n
part
ic
ular
might
be
of
in"Gerest
as
the 'lif't
ing
poino~' analysisl
p
rediets
zero i
ns
tal1taneous
pitching
moment about
the quarter chorda
Th
e
pitchillg
moment coefficient abput the
le
a
ding
edge
(nose down
po
sitive
)
aS80ciated
with flight
thro~h
8...11y
elementar
y
sinusoidal component
of
the f
o
rm
(3
)
is given by the theory 2 as
=
1
(
4
4
)
whe
re
I
and
I
are
m
o
dified Bessel
~~ctions
of the
first
kind.
A
transfe
r
7TW
ju
e
=
1
TÇr
( 45)
This
t
r
ansfer function may
be
evaluated
in terms of experimentall
y
d
et
ermin
ed
co
rr
e
l
at
io
ns
of pitching
moment and
upwash by use of equations of the
form
(1
9)
~nd
(
20) with the
correlations
suitably redefined.
CO
NC
L
UDING
REMARKS
Expressions r
equired for the calculation of the
response
of
an infinite
span airfoil
t
o
an arQitrary
two
-
dimensional upwash field
were
formulated.
In
part
icu
la
r
~
the res
ponse of
an
airfoil
spanning
a turbulent wind tunnel
was
pred
i-c'ted
.
It
is
hoped
that
co
m
pa
riso
n of the theory
with
experimental data
(currently
being
gathered
)
will
pave the
way
toward
a
more
fundamental
understanding of
rand
o
m
gust phenom
ena.
The
fla
t po
r
tion
of
the one-dimensional lift spectrum observed
experi-ment
al
ly a
t
lo
w
values of
the frequency parameter
was
shown
to
r
esult
from
the
sp
a:üw
ise var
iations in
turlmlence velocity.
The smoothing
influence
of the
span-wi
se
variations
was also demonstrated
quantitatively by comparison of 'lift
i
ng
1
.
Liep~~,
H.
W.
2.
Li
epma
un
,
H. W.
3
.
Ribner
,
H.
S
.
4.
Houb
o
lt
,
J
.
C.
St
einer,
R
.
Pratt,
K.
G
.
5.
A
'
Harrah,
R
.
C
.
6
.
Ei
ch
e
nbaum
, F.
D
.
7.
Et
kin
,
B
.
8
.
s
teiner,
R
.
Prat
t,
K
.
9
.
Sears,
W.
10
.
Lamson,
P.
1
1
.
H
a
kkinen, R
.
J
.
R
ic1ar
Q
son,
A. A.
1
2.
Fi
l
ot
as,
L.
T.
13.
Nettl
et
o
n, T. R.
REF ER ENC E
S
On the
Applieati
on
of
St
at
isti
cal
Con
cepts
to
th
e
Buffeti
ng
Prob
l
em.
J
. Aero. Sci.
Vol.19,
No.12,
pp.793-800,
D
ecember 1952.
Extension of
the Statistical
Approach
to Buffeting
and Gu
s
t Response of
Wi
ngs
of
Fini
te
Span.
J.
Ae
r
o.
S
ei.
Vol.22, No.3,
pp.197-200,
March 1955.
S
pect
r
al
Theory of
Buffet
ing
and
Gu
st Res
p
o
ns
e
:
Unificatio
n
and
Extensio
n
.
J
. Aero.
ScL V01..23,
No.12,
pp.1
075-
10
78, De
cember 195
6
.
Dynamic Response
of A
ir
planes
to
Atmospheric
Turbu
le
nce
Including Flight
D
ata on
Input and Response.
NASA
TR R
-
199,
J
une 1964.
Man~uverabi1ity
and Gust Response Problems
of
Low
-Al
titud
e, High-
S
pee
d
Flight
AG~BD Rep.556
19
6
7.
A New Method of Computing
the
Dynamic
Response
of
Ai
r
craft to Three
-
Dimensional Turbulence.
AIAA
Struc
tur
a
l
Dyn
amics
and
Ae
r
oelast
ici
t
y
Special-i
s
t
s
Conf
e
r
e
n
ce, Wew
Orleans,
Louisianna.
Ap
ril
1
6
-17,
1967.
Theory
o
f
F
light
of
Ai
r
planes
in
Isot
r
opic
Turbu1enc
e
;
Revi
ew an
d
Extension,
A
GARD R
ep.
372,
April
1961.
(
also
L~IAS
R
ept.
72,
F
e
br
ua
ry
19
61
)
Some
Ap
plications
of
Power
Spectra t
0
Turhulence
Prob
lems
.
J.
Airc
raft,
V
o1.4
No.4
pp
.3
6
0
-365,
July-August
19
67.
.
Some Re
cent Development
s in
Ai
rf
oil Theory,
J.
Aer
o.
Sci
.
V
Ol.23, No.5,
p
p.
490
-
498,
May 1956.
MeasuremerrGs
of
Lif
t
Fluctuations Due
to
Turbulence,
NACA TN
.
43880 ,
March 1957.
The
or
etical
and
Expe
rim
ental
I
nv
estigation o
f
Rand0m
Gust Loads
:
Part
I,
Ae
r
odynamic Transfe
r
Func tion
o
f
a Simple Wing Configuration in
Incompressible Flow.
NACA TN. 3878, May 1957.
Tneory of Airfoil Response
in
a
Gusty
Atmos~he
re:
P
art I, Aeredynamic
Transfer Function.
UTIA
S
Report 139, October, 1969.
14
.
Wiener~
No
15.
Sears,
Wo
R.
The
Fourier
lntegral and Certain of
lts
Applications
:
Chapto IV,
Generalized
Harmonie Analysis,
pp.150.
~
over,
1933
.
Some
Aspects of Nonstationary Airfoil Theory
and
lts
Practical Applications.
J.
Aero.
Sci. Vol.B,
y
w(x,
y)
FIG.
1.
INFINITE SPAN AIRFOIL PASSING THROUGH AN ARBITRARY
TWO-DIMENSIONAL UPWASH PATTERN.
by
ELEMENTARYSPECTRAL
COMPONENT OF GUST:
w
=
we
'"
i(kl
x
+
k 2
y)
bx
AIRFOIL POSITION
TIME:
t
TIME:
t-
.ó
t
2b
b
.óy
t
(=
U.ót)
FIG. 2. FLIGHT OF AN INFINITE AIRFOIL THROUGH AN ELEMENTARY
SPECTRAL COMPONENT OF GUST.
.~
.
1.0
IS
e
(k
l
)1
2
. 10
1
•
Ol
.
Ol
ST
RIP
THEO RY
(REF. 11)
•
•
•
•
•
REF. 10
•
REF
.
10
o
REF. 11
--
...
~L
-
8
Ó=
.
28
(L
=
.
4
•
••
•
•
•
..
Ó=
.28
L
=
.4
Ó=
•
28
L
=
.8
Ó=
•
33
L
=
.
3
...
•
•
- -
PRESENT RESULTS
(
EQN
.29)
. 10
... ...•
"
"
...
"-...
"
SEARS n
;
NCTION
I
S(k 1)
1
2
"'"
""'-...
"
o
•
I
...
"
•
• re
•
"-"
"-"
"
"
"
"
o
•••
•
"
.
"
"
"
"
~
'"
"
"
o
•
~
"
'\
ó
=
.
28
L
=
.
8
'
"-
"
..
.
o
""
,,-"
~\
ÓL
=
.4
1.0
o
,,"-"-\
o
"-FIG
.
3
.
'EFFECTIVE SEARS FUNCTION'
-
COMPARISON OF
ANALYSIS AND EXPERIMENT.
k 1
1.0
5
-2 - - - -__
SEARS FUNCTION
.10
IS(k 1
)12
t - - - -
1 _______________ _
L
.01
- - EXACT, EQN. (31)
- - - - APPROXIMATE, EQN. (34)
.001
.01
. 10
1.0
10.
FIG. 4.
'EFFECTIVE SEARS FUNCTION' - ANALYTICAL RESULTS FOR VERY
2 - / 2
4'1r w 2
U
1.0
.,
...
.
-
.
""
'LI
FT
ING
POINT' THEORY
LIEPMANN,
REF. 1
"
LARGE
SCALE EXPANSION, EQN.
(
38
)
.
10
.01
/,
/,
;,
;,
/,
.001
.
10
iJ
'"
'"
/,
/,
;"
~/J
~ . / . / . / . /Y' / " "
SMALL
SCALE EXPANS
IO
N,
EQN.
(
39
)
/,
PRESENT RESULT, EQN
. (
3
7)
1.0
L
10.
100.
T
U
RB
ULENCE
INTEGRAL
SCALE
AIRFOIL SEM
I
- CHORD
FIG. 5. MEAN SQUARE LIFT COEFFICIENT CARRlED BY NARROW
SPANWISE
STRIP -
AIRFOIL PASSING THROUGH HOMOGENEOUS
TURBULENCE.
f
1
/f
2
10
6
4
2
1.0
10.
100.
1
.1~0---~'---~---
I
• u ••
L
=
TURBULENCEINTEGRALSCALE
AIRFOIL SEMI CHORD
Unclassified
Securitv Classification
DOCUMENT CONTROL DAT A -
R
&
0
(Security clllssification ol tltle, body ol abstract and Inde1Cln~ annotation must be .ntered when the overIlIl report Is cta •• U/ed)
\. ORIGINATING ACTIVITY
(Corpora te author) 211.REPORT SECURITY CLASSIFICATION
Institute for Aerospace Studies. ,
Unclassified
University of Toronto,
2b.GROUP
Toronto
5,
Ontario,Canada.
3.
REPORT TITLE
THEORY OF AIRFOIL RESPONSE IN A GUSTY ATMOSPRERE.
PART
II:
RESPONSE TO
DISCRETE GUSTS OR CONTINUOUS TURBULENCE
4.
DESCRIP TI V
E NOTES
(Type ol report IInd Inc/usive dIltea).
Scientific
Interim
15·
AU THOR(SI
(FIrat nllme, mlddl. Inltllll, 11I8t nllme)L.
T.
Filotas
11·
REPORT DATE
'11.TOTAL NO. OF PAGES
T'b.NO
.
OF1SEFS
November 1969
13
811.
CONTRACT OR GRANT NO
.
M'
-Al''Ul::ih 0 ( -UO
(CA 9a.ORIGINATOR'S REPORT NUMBERIS,
b.
PROJECT NO
.
9781-02
UTIAS Report No. 141
61102F
c. gb.
OTHER REPORT NOIS,
(Any other numbera thllt may be aaa/l/<led thla report)d.
681307
AFOSR 69-3089TR
10.
DISTRIBUTION STATEMENT
l .
This document has been approved for public release and sale;
its distribution is unlimited
11.
SUPPLEMENTARY NOTES
12.SPONSORING MILITARY ACTIVITY
Air Force Office of Scientific Research,
TECR,
OTHER
1400 Wilson Ave,
(SREM)
Arlington,Virginia,22209, USA
13.
ABSTRACT
The response of an infinite span airfoil in an arbitrary two-dimensional upwash
field is studied analytically on the basis of the
Iaerodynamic transfer function'
giving the pressure distribution in an inclined sinusoidal gust deri ved in Part
I of this report. When the upwash field can be completely specified ( discrete
gust case) the lift response is expressed as aquadrature involving the transfer
function.
When only statistical properties of the upwash can be specified
(atmospheric turbulence) mean square values, correlation functions and power
spectra are similarly expressed.
Certain results are given in closed form.
Expressions giving the transfer function in terms of correlations which may be
determined experimentally are given.
Previously published experiment
al
data
are compared wi th analytic results based on Part
I.
The detailed expres si ons
deal with the lift response; however, the modifications required for the pitching
moment response is explicitly stated.
Unc1assified
Security Classification
t 4. LINK A LIN K B LINK C
KEY WORDS
ROLE WT ROLE WT ROLE WT