Appl. Sci. Res.

24 J u l y 1971

D Y N A M I C T H E O R Y OF A S H O R T HOT W I R E
N O R M A L TO AN I N C O M P R E S S I B L E A I R F L O W ,
CONSTANT R E S I S T A N C E O P E R A T I O N
R. E. DE HAAN
Laboratory for Aero- and Hydrodynamics
Delft Uiiiversity of Technology,
Delft, THE NETHERLANDS

Abstract

The aim of the investigation was p r i m a r i l y the c o m p u t a t i o n of the d y n a m i c a l

corrections on the static calibration curve. A brief aecount is given of t h e
available analytical expressions for the b_eat transfer of a h o t wire to the
a m b i e n t airflow. The appropriate form is inserted in the small p e r t u r b a t i o n
energy e q u a t i o n giving, with constant values of the characteristic para-
meters, a linear second order differential equation. A n u m b e r of numerieal
e x a m p l e s was e o m p u t e d f r o m the exact solution yielding the influence of
f r e q u e n c y on the sensitivity and the t e m p e r a t u r e distribution along t h e wire.
The analysis is supplemented b y a short review of some complicating factors
not included in the linear theory.

Nomenclature
A,B,B" coefficients of heat transfer; B ' = 0 . 4 B Ü °.4
b t e m p e r a t u r e coefficient of electric resistanee
4bRol
C
~d~~w
c velocity of sound; speeifie h e a t
d diameter
G overall t r a n s c o n d u c t a n c e
Gr Grashof n u m b e r
h h e a t transfer coeIficient
I heating current, fT + i = i q- { eäo~
l length of bot wire
le "cold length", ~C(V -- ~2)~-}
Kn Knudsen number
3/I Mach n u m b e r
N u , Nur Nusselt numbers

-- 231 --
232 R . E . DE HAAN

NT Temperature number
Pe Péclet n u m b e r
Pr Prandtl number
p 1 q- j~o/co*
q specific h e a t flow
Rwl, Rgl, Rol electric resistances of h o t wire per unit length,

Üd
Re= Reynolds number

o)d
Str = _ Strouhal number
U
Tw, Tl, Tg, To t e m p e r a t u r e
t time
u=Ü+ _u = Ü + ü eJ o)t v e l o c i t y of m a i n air flow
v A + B Ü o.4
X coordinate along hof wire
l
2le
K ~ Cp/CV ratio of specific heats at c o n s t a n t pressure and c o n s t a n t v o l u m e
äg, ,~r, Zw h e a t c o n d u c t i v i t y of air; ~v ditto of hot wire
A molecular m e a n free p a t h
,u d y n a m i c viscosity
'V k i n e m a t i c viscosity
p density
(1) angular f r e q u e n c y
Xw 1
CO*
pwCw 12e
P~w
H
Kg
x 2x l 2x
le 1 2le l
amplification factor of acoustic resonance
?~wl/Rgl = [ _ ( x ) / R g l
l~wl/Rgl

superscripts
A amplitude
-- m e a n v a l u e w i t h respect to t i m e
( ) m e a n v a l u e w i t h respect to position

subscripts
* characteristic q u a n t i t y
f film
g m e a n gas flow
w b o t wire
 DYNAMIC THEORY OF A SHORT HOT WIRE 233

0 zero degree Celsius, zero frequellcy

1 unit length
ullderscore dertoting complex quantity (only used when neces-
sary to avoid ambiguity)

§ 1. Introduction
In the past decades the hot wire has been the standard instrument
for measuring turbulent quantities in air and despite competitors
its position for fundamental measurements seems assured for the
near future. The constant resistance method of operation with up
to date electronic equipment is so far developed and offers so m a n y
advantages over the constant current method that one can expect
its almost universal use. Current practise is the use of the static
calibration curve as a base for the dynamical response. Even with
a high-gain feed back eorrections have to be applied for the finite
length of the wire because the temperature distribution along the
wire is frequency-dependent. This is especially important when
measuring dissipation characteristics whereby high frequencies and
a short wire are combined. The computation of these corrections is
always restricted to simplifying assumptions so it seemed worth
while to make an effort for a direct experimental dynamical cali-
bration. Generally speaking two procedures here offer themselves:
the first one is using a shock wave as a step function and the second
one is applying a harmonic plane sound wave superimposed on a
steady airstream.
The shock wave yields chiefly overall information concerning the
frequency range and the damping of the measuring system and in
this respect it has given valuable results. On the other hand one
can expect from harmonic sound waves detailed information under
well defined conditions which are moreover similar to the operating
conditions when measuring turbulence. The required plane wave of
sufficient amplitude can be obtained by exciting a resonance in
an open tube kept axially in a main air flow. Putting the hot wire
in an anti-node of the velocity fluctuations and determining these
fluetuations by means of the pressure gradient at this position
yields the dynamical response of the bot wire. The upper frequency
limit of this method is given by the radiation end-losses of the reso-
nance tube, it can be estimated at 4 or 5 kHz. Thence a substantial
frequency range of incompressible turbulent flow can be covered by
234 R . E . DE HAAN

it. We intend to describe an experimental set-up pertinent to the

matter in question next to a discussion of the results in an ensuing
article. As a first step we compute the dynamical corrections on the
static response curve in a form suitable for direct use. In view of
this restricted aim we will only consider a hot wire normal to an
incompressible air flow with superimposed small fluctuations. With
constant values of the charaeteristic parameters the energy equation
yields a linear second order equation with a subsequent analytical
solution. A comparison of the isentropic sound ware with the iso-
thermal incompressible concept of turbulence will offer, too, an
opportunity to discuss the effects of the temperature-dependency
of the relevant parameters of the ambient air.
Finally we remark that m a n y reports are already devoted to the
very complicated process of heat transfer from a hof wire at differ-
ent air speeds and frequencies. We mention in this respect especial-
ly some recent ones [1], E2J, [3? where a very complete review is
given of the whole situation; we derived much valuable information
from it.

§ 2. H e a t t r a n s m i s s i o n f r o m the hot wire

A logical analytical expression for the heat flow and a minimum
number of characteristic parameters are wanted here. The oper-
ational range of a hot wire intended for turbulence measurements in
subsonic airflows can be pur at 0.2 < R e < 40. The flow structure
in this fange is easily revealed by a simple qualitative description:
in the lower range of the airvelocities the field approaches that of a
pure viscous flow, so a well known and regular configuration with a
nearly identical deformation upstream and downstream of the cylin-
der is obtained, moreover extending over a large area. An increase
of the airspeed decreases steadily the area of the deformation in
front and enlarges the area at the back whereby gradually a sym-
metrical pair of vortices is formed in the wake. These vortices grow
increasingly less steady and at about R e ~ 40 the wellknown vortex-
street is formed; the value of the bot wire as a measuring device is
here questionable. The area of vorticity in the front half of the
cylinder is now so small that the boundary layer concept can be
used for say R e > 50.
This description illustrates that the flow mechanism is not only
complicated, it also changes eharacter in the operational range. One
 DYNAMIC T H E O R Y OF A SHORT HOT W l R E 235

cannot expect an exact theory here but at most an interpolation be-

tween the two more simple conditions Re ~ 1 respectively Re >~ 1.
As an introdnctory step we will eliminate the higher order terms in
the four basic equations by a survey of the numbers characterizing
the relative importance of the constituting terms.
In the first place K n = M / R e = A / d ~ 0.1, so although not
sufficiently small to reckon with a rigorous validity of the nonslip
condition of velocity and temperature (see [31) the continuum
concept is serviceable for our limited aim. Substantial simplifi-
cations are obtained by the validity of the low-frequency approach
justified by Str and Re Str ~ 1, by the neglect of radiation for
temperatures up to 300 or 400°C and by the neglect of free con-
vection at low airspeeds where Gr/Re <~ i. As for the compressi-
bility the effects of inertial forces are negligible because M 2, M~/Re
and Slr are all ~ 1. This is certainly not allowable for temperature
effects as A T / T , = 0(1). Besides p the quantities # and ~ of the
ambient air are also a strong function of temperature.
The conclusion of this elimination process is that we can pur
Nu---- ( (h)d)/2 ~ /(Re, Pr, NT) ; more information can be derived
for the case of a two-dimensional flow and extreme values of Re.
Re >~ 1 (Pr ~ P r Re ~ Re, so heat transport near the wire main-
ly by convection). The theory of laminar boundary layers stares
N u = / ( P r , NT) Re °'5 whereby the dependency of N u on P r can
be represented by a weak power say N u = / ( N T ) Pr°'3aRe °'a. More-
over it can be argued that the temperature-influence of # and ,t
have opposite signs and cancel exactly for P r = 1, whereas the
effect for P r = 0.73 is small. This suggests that by choosing a
suitable definition of the characteristic temperature T, the temper-
ature number N T can be made to disappear as an explicite quanti-
ty. Thus by putting T, : Tl, Nur : ((h)d)/Zf and Re : pgUgd/[,f
a sufficient accurate and very simple expression is procured by
Nur : a2PrO.a3Re o.5.
Re ~ 1 and ultimately Re -+ 0 (Pe -+ 0 so heat transfers near
the wire mainly by conduction).
Here it can be demonstrated for a two-dimensional flow that
N u - - / ( P e ) = / ( P r Re) ; a theoretical treatment for Re --> 0 based
upon the Oseen approximation and neglecting free convection leads
to a logarithmic decrease of N u as a function of Re, yielding N u = 0
for Re = O.
236 R . E . DE HAAN

H o w e v e r e x p e r i m e n t s with short wires d e m o n s t r a t e a finite value

of N u for Re = 0 with a horizontal t a n g e n t on the curve in this
point. This can be a t t r i b u t e d to the influence of free convection and
to the three-dimensional c h a r a c t e r of the heat flow.
S u m m i n g up a fair i n t e r p o l a t i o n in the range of a b o u t 0.3 <
< R e < 30 adjoining s m o o t h l y at b o t h ends the curves for R e >~ 1
and ~ 1 cited above is o b t a i n e d b y
Nur : a l P r °.2 + auprO.33Re °.4

where Nur ~--,(h)d Re-- pgUgd (1)

2f #f

The first t e r m and the value of the e x p o n e n t of R e being ~ 0.4 < 0.5
are consequences of the interpolation, the n u m b e r s al and a2 are
to be d e t e r m i n e d experimentally.
Applying to (1) the basic relations of heat transmission leads to
an expression suitable for direct use in hot wire problems. The local
specific heat flow in the air near the hot wire is defined b y

q ---- - - 2 w \ ~r ]w h(Tw Tg) or q Nu~tw T w d Tg

where N u = hd/2w. In the same w a y a m e a n specific heat flow is

defined b y

(q) = \ er / w = ( h ) ( T w - - Tg)
Tw -- Tg
or (q) ~ ( N u ) 2 w - -
d

where ( N u } ---- ( h ) d / 2 w . B y choosing a n o t h e r reference t e m p e r a -

ture we can formally likewise define

d Tw -- Tg
( N u ) =- N u r ~ ( h ) - - and (q) = Nuf~f
2r d
The heat flow per unit length of a cylinder is

ql ~ =d (q) = rc2~Nuf(Tw -- Tg) ----

= (r:2falPr 0.2 @ x2¢ra2Pr°'33Re 0.4) (Tw -- Tg) =
= 2~(A1 q- B 1 R e ° ' 4 ) ( T w - - Tg). (2)
in which A 1 = 7:alPr °'2, B1 ~ r:a2Pr °'a3.
 DYNAMIC THEORY OF A SHORT HOT WIRE 237

With R w l = R 0 1 [ 1 + b ( T w - - Te) ] we can put T w - - Tg

= ( R w l - - R g l ) / b R o l ; defining R e = o g U g d / # f and introducing
these expressions in (2) yields

ql =
[ ÄrA1
b R0-~-
@ ~«~~ (~)O~~o~] (Rwl -- Rgl) =

= [A 4- B U ° ' 4 ] ( R w l - - Rgl) (3)

where
~fA 1 ~f
A -= ~ - - a l n P r o.2 - -
b Rol b Rol '
2fB1 (pgd~°'4=a27:prO.a 3 )« ( p g d ~ °'4
B- b Rol \ #~-/ b Rol \ #-~-/ "

The dedueed expressions identify the characteristic data and

numbers relevant to the problem but the applied interpolation
process is not accurate. There are more (and large) uncertainties:
"constants" like b and R01 vary widely from wire to wire, depending
on the manufacturing process, the purity of the material, the heat
treatment etc. ; moreover the variation of the wire diameter is large.
This justifies the current practice of calibrating each wire individu-
ally as the only dependable procedure.

§ 3. Energy equation
The linear character of this equation is constrained to many as-
sumptions. The heat transfer to the ambient air is assumed to be a
two-dimensional process represented b y (3) with constant values
for the quantities ,tr, #f, Pc.
The hot wire is considered as a circular cylinder with a smooth
and clean surface, a homogeneous structure not affected by me-
chanical stresses, a negligible thermal expansion and constant
values of ~~, Cw and pw. The radial temperature gradient is neg-
lected, so are the thermo-electric Peltier and Thomson effects. A
linear relationship between the electric resistance and the temper-
ature is assumed; the influences of current density and the self-
inductance are neglected.
The end snpports are supposed to be massive and to possess the
temperature of the ambient air; temperature gradients in it and the
electrical resistance are taken zero.
238 R.E. DE HAAN

Most of these assumptions hold fairly in practice with the ex-

ception of 2r,/*f, Pc being independent of the temperature; we will
reconsider this aspect later on in § 7.
We can now write down the energy equation for small fluctu-
ations in an incompressible airstream for a wire stretching from
x = - ~ l to x = ½l
ö2Tw
I2Rwl = (Æwl - - R g l ) ( A 4- Æ U 0"4) - - '~w - - d2 4-
4 ~x2
7": äTw
4- pwCw _4 d2 _ _ôt (4)

where Rwl = ROl{1 @ b(Tw -- To)}.

u= Ü+_u= U + ü o ot
I=i+i=Z4-~e jC°t
Substituting Rw~ - - Rw~ 4- r_wl = F ( x ) + ~wl e m = [(x) eJO~t (5)
4b Rol
C--
z;d2;tw

and taking only first order terms, so (Ü 4- u)O-4 ~ { 1 4- 0.4(u/U)} U °.4 :

(i2 ÷ 2I_/)(Bw1 + _rwl) =

l(d2Æwl
C ~ dx
~2_rwl)
4- ~ ~x
1 pwCw ~wl
4- C 2w ~t
(6)

Separating the equation it follows for the time-mean quantities

d2/~wl
dx 2 C(A + B Ü 0.4 -- i 2) /~wl = --C(A 4- BÜ°'4)Rgl (7)

next for the fluctuations

1 62-/ { -- co pwCw}-/(x) =
C dx2 @ I 2 _ ( A +BU o.4)-j ~ 2~
ü
= (Bw1 - - R g l ) 0.4BÜ °'4 U -- 2I-Rwl$. (8)

These equations will now be solved successively and discussed.

 DYNAMIC THEORY OF A SHORT HOT WlRE 239

§ 4. Time-independent quantities
E q u a t i o n (7) can be written
d2/~wl
C ( V - - 12) Rwl = --CVRgl (7)
dxe

the b o u n d a r y conditions are

dJ~wl
x~---O, - - --0
dx (9)
3C = ½l, /~wl = Ægl.

The one-dimensional solution yields the particular integral/~wl =

= { V / ( V - - i2)} Rgl, the complete solution becomes after inserting
the b o u n d a r y conditions
cosh

/~wl = R g l - ~
J2Rgl
V -- i 2
{ 1 cosh~
}
V -- iz - -
cosh ~ = Rgl ~o~h~
V -- ]2
(10)

le is the "cold length" as it accounts for the cooling effect of the

supports (see [4]).
I n t e g r a t e d Irom x = --½l to ½l:
V -- i 2 tgh e
iZRg { tghe} E
/~w=Rg-r- V__i2 1 - =Rg (11)
e V -- i2

As the transconductance at zero frequency oI a high-gain feedback

is generally very large, say 100 to 200 A / V , we can pur/~w = constant.
E q u a t i o n (1 I) can also be written

/~w V
i2 -- (1 la)
/~w - - Rg Rg tgh e
1
/~w e
This is the torm conformable to the two-dimensional case e --> oo;
where Ü and i are both contained in e it is not possible to write
explicitely i 2 = / ( Ü ) .
The relation
cosh
V -- i2 - -
J~wl cosh e
Ægl V -- i 2 (10)
240 R. E . D E H A A N

can be characterized by a few parameters. /~wl is a maximum for

x ---- 0, thus from (10) and (11) it follows

represents a basic physical quantity (in the same way the correction
term in the denominator of (1 l a) has a physical meaning too; it
equals the quantity 2{q,/(i2/~w)}).
It will appear in the ensuing paper that the computations will
not be difficult but tedious. In order to get a clear review and con-
cise formulae it is essential to use groupings and parameters which
have a physical significance. In this respect quantities like le, s,
(tgh e)/s are the only practical choices. We will now compute some
important relations; in the first place the static sensitivity d i / d U
and secondly the change of -~wl and its end-tangent (d/~wl/dX)~z
with the speed U. (The latter two quantities are important as a
check on the asymptotic values of the corresponding dynamic forms).
 DYNAMIC T H E O R Y OF A S H O R T HOT W I R E 241

Writing ( 11 ) as
i2 tgh e
V s
/~w--Rg V-- [2 @Rg V-- T2 --0 or /(z% u ) = o

it follows d f / d U = --{(O]/~U)/(~//O])}, keeping in mind t h a t V is a

function of U and likewise s is a function of U and _? the differ-
entiating process is a simple procedure resulting in

3 tgh e 1
+
d] 1 2 e 2 cosh 2 s i2B '
dU 1 tghs I2 ( t g h e _ 1 ) 2VU
s 2V s cosh2 s

iB'
2rv
.
I1
V --
2V
i2
1 t g ~h ~
s
2-g
cosh 2 s
I2 ( t g ~--
h~ CO~~ 2 )
_°

(15)
F r o m (13) it follows :

d (d/~wl~_ 2Rgl d { ~2 }
d~ \dx--x/~l l dU V--i2 etghe

where

de ae dV ae d] sB' J d1
d U -- aV dU -4- a~- dU -- 2 ( V - - I ~) U V--i 2 dU

and where dI/dU is given by (15). Performing the calculations

yields

d (d/~wl ~ _ 2Rgl J2B'

X
d~ \ ~x-x-]¢l l 2V(V - I2) U
tgh 2 e s
tgh s -- 2 - - + - -
e cosh 2 e

x 1 cos.1 ) /16a,
242 R. E. DE HAAN

or in a nondimensional form

e zv \ e cosn ~ e /

The determination of d/~wl/dU is laborious, we will shortly outline

it. According to (10)"

Differentiating (10) with respect to U, inserting the already cited

form for de/dU and dI/dU from (15) yields a lengthy fraction.
Combining the terms in logical groups results finally in the following
form

A check on the correctness of our calculations can be made by

differentiating i17) with respect to ~ and then putting x = ½l or
= e; it leads again to the expressions (16a, b).
 DYNAMIC THEORY OF A SHORT HOT WIRE 243

§ 5. Time-dependentquantities
Equation (8) can be written

1 d2[ { ~o pwCw} ü
= (/~wi -- Rgl) B' _
U
-- 2[Rw1¢. (8)

We can apply two solving procedures: integrating (8) and subse-

quently introducing the feedback conditions gives i = _~exp jmt.
In this way the sensitivity ~_as a function of ü is obtained at once;
however an ensuing investigation of rwl as a function of x is very
intricate. We therefore prefer the alternative method of eliminating
in the equation (8) by imposing first the feedback condition and
then performing the integration.
From the definition of the transconductance of the feedback it
follows that
½1 }z
i=--_Gi2 S_rwtdx and ¢ = - - 2 _ G i I _/(x) dz (18)
0 0

in which G is the transconductance including the bridge- and the

operational factors.
Moreover as G0 -~ oo we put already 2 SôI/~wl dx = / ? w = constant.
Integrating (8) from 0 to ½ yields

1
-

C
-
d[
dx
- -
J C 2w t "
_

= ~1( R--w - - Ru) B ' LU -- iRw~

(1o pwC w
v_i2+ j c ~w
2I Rw +
GI
:(~w_Rg) B, ü 2 (d/)
Ü c -~ß-~ ~,.~"
The interpretation of the terms of this energy equation is clear, the
last term in which the differential quotient is still unknown repre-
sents the small fluctuating heat flow to the support. Multiplying by
244 R . E . DE HAAN

/~wi//~w:
o) pwCw
V --i2+j !
C 2w
2IRwfi_ 1 +
2G__I2Rw
/~wl
Rw
(/~w--Æg) B ' --
U C R%
ùwl(~) / (19)
where I__GI/~wis very large we can write the form between the square
brackets as E1 + 6] in which I~1 = / ( @ ~ x. Inserting (19) in (8)
yields

de[ { C(V . ~pwCw

-- t z) + joo » [ ( x )) =
dx 2 2w J

=C Rgl -_~ w 1 @~ /~w l ~-+

/~w~ (1 _ (~)(d/)
+ 2 -~w ~~~'
Substituting

z
~,wW/
- c(v - i~) =
lwcw _
z~o 7 w
1
~o. '

introducing from (10) and (11)

//~wl V---f2 . cosh~ . . . /~2 ( tgh s cosh ~ )

cosh e l/?wl e cosh e
-- 1
tgh tgh
V -- 12 e -~w V -- iŒ

next bringing the equation in a nondimensional form by multipli-

cation with lo/Rgl
2 and putting [(x)/Rgl = r:

2d2r - - ( 1 +
d~ j m--~~~-)r=

: 12cC + Õ I2 lRg 1
V -- i 2 tgh s s •
ê
 D Y N A M I C T H E O R Y OF A SHORT HOT W I R E 245

T h e b o u n d a r y c o n d i t i o n s are (d~/d$)o = O, (r)~ = O.

246 R.E. DE HAAN

Substituting x = const × e~~ in the left hand side gives

P=i
( 14-j co,
~ )~
further the particular integral is of the form {M 4- N(cosh ~/cosh s)}.
Performing the solving procedure yields after inserting the bounda-
ry conditions

~ (d~) ~ (d~)
MI~-+Ma -~- e Nl~-4-N2 ~ - ~ cosh
~= + 4-
co 6o cosh s
l+j-- j--
6O* 6O*

M1 N1 . +
+

+
II 14-j 6o,

M2
6O
J a).

N2
6O

/~
d~) cosh~p
(22)
60 09 ~ cosh et
14-j-- j--
60* 60*

The end-tangent (d~/d~)~ is yet unknown, therefore differentiating

(22), putting ~ = s, solving (d~/d~)~ and using the relations (21)
gives the final formula
p tgh ep
1
tgh ep tgh s
- - -F s(1 4- b")
P
j - -6o
co. ü
(d~)~ = M1
1 p tgh ep U
tgh sp P tgh s
1 -- Ma -~-- 4- ~ - t g h s m
j--
60*

(23)
where
iSB ' tgh s
MI= (1 4- (3'), (21)
tgh e .
 DYNAMIC THEORY O F A S H O R T H O T WIR]?; 247

V(1 -- 6)
M2 =
~(~ » ~~h~) . (21)

Of interest are the asymptotic values of (23); for co -+ oo: (d~/d~) --->
-+ 0, the heat exchange with the supports decreases steadily with
increasing frequency.
For (o -+ 0 the second fraction in numerator and denominator be-
comes indeterminate. Taking the limit gives
p tgh ep
1
lim tghe 1 e (tghe 1 )
o) 2 tgh e e cosh z s
j--
0)*

With this value substituted in (23) the relation becomes for ~o -» 0"
(d~) =(1+6,) -?ŒB'

tgh 2 8 e 6"e( -t-ghe~ ~ o ~1~ . )

2 tgh e + c~~h2~
e cosh z e ü
X Ü.
tghe i ~ (tghe 1 )@
e 2V e cosh 2 e
+~{tgh,~ ,2(<~e 2-V -/cosh2~
,)}
(23a)

With 6 = d' ~- 6" = 0 the corresponding static value (16b) is found

again.
The aim of this derivation is the determination of the dynamic
sensitivity; it is obtained by substituting (23) into (19): From (19)
it follows
/~w--RgB, ü 1 (d_B)
(1 +6)_~--
2IRw Ü CiRw ~l =
Rw--RgB, ü Rgl 1 (dr)
-- 2i/~w Ü /~w Cilc ~ ~
The two terms on the right hand side can be written according to
(11a)
248 R.E. DE HAAN

Inserting the values of M1 and M2 in (23) and next substituting the

resulting (d~/d~), in (19a) gives after some algebraics and rewriting
ot the formulae the Iollowing expression for the dynamic sensitivity:
 DYNAMIC THEORY OF A SHORT HOT WIRE 249

The asymptotic values of the dynamic sensitivity (24) will now be

considered.
Taking first o~/o). -+ oo.
With

~ve find for the limit values

250 R.E. DE HAAN

This limit value has only sense with an ideal feed-back or d = 0.

Then (24) becomes for m/~o. --> oo:

tgh s
f B' V - i2 e
ü 2V V 1
i2 tgh s
Ü V 8

tgh e ]
1

2V f2 tgh e
]. (2s)
V e
Taking now ~o/~. -~ 0.
Here we can at once write down: p ' = I, p " = 0, p = 1; the
fractions in (24) with j(o/m, in the denominator become undetermi-
nate, taking the limit values yields

lim(tghe . _
tghep)/j o) _ 1 (tghe_ 1 ) j

e ep to. 2 e cosh 2 e
l i m ( tghs_ p2 t g h e p ) / o) _ 1
s ep J ~o. 2 cosh 2 s

and the dynamic sensitivity becomes

i B' V -- [2
(1+~) ü 1 ×
2V V
U

tgh e 1 V
+ ~~VI@_~VII ~_ ~VII1]
2e 2 c o s h 2e V--] 2 tghe

X
1 --
tgh s [2 /' tgh e
8

i) (26)
e 2-12 k e cosh 2 e
 DYNAMIC THEORY OF A SHORT HOT WIRE 251

6vr =btghe(tghe_ 12 1 )
e e 2V cosh 2 e
V Bw--Rg(tgh2e_ 1 ) (26a)
6vH de i2 Rg e 2 cosh 2 e
( V)/?w--Rg 1
DV I I I = 6 1 -- e~- Rg 2 cosh 2 e

With d = 0 (26) reduces to the corresponding static value of the

sensitivity (15).
For sake of eompleteness we will conclude out theoretical treat-
ment by deducing the formula for r. In order to avoid too lengthy
formulae we will put ô = 0; we remark in this respect that in the
next paragraph will be mentioned that this is certainly allowed with
a high-gain feedback. The substitution of (23) in (22) is again a
tedious but trivial procedure, we will only mention the crucial
points of it.
The fractions (21) defining M1 and M2 are inserted in (23) after
which (23) is multiplied by {V -- ?2(tgh e/e)}(V -- f z) and brought
in its most concise form. In the expression (22) for r the terms are
arranged in a factor of ü/Ü and a factor of (d~/d~)e after which (23)
is inserted. Performing the relevant operations yields a relatively
simple expression :
1 ]2B'
o~ V ( V - i2) ×
j--
09*

(tg~e tghep~@( 1 tgh@~cosh~_(1 tgh e~cosh 8p

x
tgh@ ] ~ p2
(tg~e_ tgh ep~ Ü"
1 @ V . c0 @ /
]--
(D*
(27)
This result can be checked by differentiating with respect to ~ and
next putting ~ = e; the appropriate form (23) is again obtained.
Asymptotic values of ~. When ~ -~ oo nominator and denominator
of (27) remain finite, thus the factor 1/j(~o/(o,) decreases the value
of ~ steadily to zero at ~o = oo.
252 R . E . DE HAAN

W h e n co ~ 0 we have first to determine some limits:

( CO )~ 1 CO
= --~ 1 + j -- ,
p l+j co. ~ co.

6O
tgh e + j tg l e - -
co* co 1
tgh sp ~ -+ tgh e + j ½e
co co. c o s h 2 e '
1 + j tgh e tg ½e
CO*

tghepep ~ tghee J 21- ~ ~co- . (tghe_e

" cosh21/~
cosh@-+cosh(e-bj½e co.c°)-+c°she+j½e to.co sinhe,

(1)
cosh ~ + j ½~ - - sinh
cosh ~p co. cosh
-+
cosh ep co cosh e
cosh e -¢- j½e - - sinh e
CO*

co cosh~ co sinh
- - j ½ s co. c o s h e t g h e + j ½ ~ co. c o s h e

Carrying out the subsequent routine calculations in (27) the corre-

sponding expression (17) of the static quantities is reproduced so
r0 - + d}R.

§ 6. Numerical computations
The dynamic sensitivity is c o m p u t e d for a range of parameters which
is in c o m m o n use. The most clear procedure to demonstrate the
influence of frequency seems taking a specific wire of current di-
mensions and computing this influence for a n u m b e r of airvelocities.
For incompressible flow of air the m a x i m u m frequency when
measuring turbulence can be put at 10 k H z or co/co. < 200. With-
in this range the transconductance __Gis virtually a real q u a n t i t y
and of the order of 50 A/V for a high-gain feedback. A few tenta-
tive computations show that the influence of the correction terms
pertinent to the m a t t e r in question can be ignored, thence d = 0
in (24), (25), (26).
 DYNAMIC THEORY OF A SHORT HOT WlRE 253

T h e c o m p u t a t i o n s (in S.I. units) are p e r f o r m e d for a t u n g s t e n

wire, d i a m e t e r d = 5 ~ a n d the following c o m b i n a t i o n s :

l~0.5mm, H=1.4; l=lmm, H=l.4and2;

/=2mm, H=2; /=4mm, H=2; /=6mm, //=2.

T h e airvelocities are U = 3, 10, 30, 60 m / s ; o/o~, is v a r i e d be-

t w e e n 0 a n d oo. (in view of the finite b a n d w i d t h the highest values
only d e m o n s t r a t e the trend).

Physical constants o/tungsten. Specific electric resistance 0.5 ~f~-i

at 0°C,

b = 4 × 10 -8°C -1,2;v = 1 2 0 W m - I ° C -1,pw = 190 × 1 0 a k g m -3,

Cw = 143 Ws kg -1 °C -1.

D i t t o of air:

/~f = 0.20 × 10 -4 kg m -1 S- 1 , pg ~--- 1.21 kg m --a, 2f =- 0.028 W m -1 °C -1.

T h e n u m b e r s in (1) are t a k e n al = 0.5, a2 = 0.6, t h e n it follows

A = 40 × 10 -4 A 2, B ~- 30 X 10 -4 A 2 s °'4 111-0"4,
C = 43.5 x l0 s A - 2 m -2.

The basic m e a n q u a n t i t i e s ,Y a n d e are n e x t c o m p u t e d b y trial

a n d error f r o m (10a) a n d (11)"

l //--1
- - - - ½z(c(v - ~2))~, ~~ _ v.
2le H tgh e

We notice t h a t i c a n n o t be a d j u s t e d a r b i t r a r i l y ; clearly

i 2~<V=A +BÜ °.4.

For [2 __> V it follows le -~ oo so s --> 0 as necessarily l is a finite

q u a n t i t y . Thence e v e r y wire becomes a " s h o r t " wire (e -+ 0) w h e n
[2 ~ V. I n s e r t i n g in (11) (tgh e)/e -+ 1 - - ½s2 = 1 - - 1C12(V -- ~2)
yields after working out H = 1 + ~-~Cl2i 2. So we find t h a t

1I = -~w/Rg ~ 1 @ I~CI2(A @ B Ü 0.4) for i~ <~ A -}- B Ü °.4

(e.g. the wire w i t h l = 0.5 m m c a n n o t be h e a t e d u p to H : 2).

T h e physical i n t e r p r e t a t i o n is the steepening of the flanks of the
254 R . E . DE HAAN

temperature-distribution (19); the axial heat-flow becomes domi-

nant. The basic mean quantities are now collected in Table I.
The d y n a m i c sensitivity as a function of co/co, is next computed
from (24), (25) a n d (26). The results are t a b u l a t e d in Tables II
through VII, i.e. the form between the brackets { . . . . . . } in (24),
(25), and (26).
Commenting on the results we can state t h a t the values in the tables
can be represented b y smooth curves. The imaginary parts of the
correction shows a weak m a x i m u m in the middle frequency range.
This m a x i m u m however is very small, even for the shortest wires
it is no more then about 2% ; thence the phase shift is u n i m p o r t a n t .
The absolute value of the real part is strongly dependent on the
length diameter ratio: the shortest wire with a ratio of 100 has less
t h a n half the sensitivity of a long wire. All real parts decrease
monotonicly with frequency, the numerical value of this decrease
over the whole frequency range is Irom about 1.5% for a long wire
to about 7% for a short one. Generally speaking when measuring
dissipation in turbulence with a hot wire of a length diameter ratio
of about 200 a correction between 3% and 5% can be expected.
For practically all applications the range of the parameters given
in the tables will be sufficient.
In order to demonstrate the t e n d e n c y of the local fluctuation of
the resistance with the frequency we computed ~ and then d~R
from (27) and (17). These Iormulae can be written
i2B' ü
x F x - (27)
v(v - z~) u
i2B ' dU
d~ x Fo × _ . (17)
v(v - «~) u
The Iactor F b y which the dependency on x, co and e is determined
was eomputed for a wire with the approximate data: I = 1 mm,
d = 5 ~, H = 2, U = 10 m/s and m/co, = 0, 0.5, 20, 100. The trend
is clearly visible in Fig. 1" conspicious is the large phase shift, even
for the low value co/co, -- 0.5.

§ 7. Consequences of the temperature-dependency of ,ag, /gg and ,~g

In § 3 we stated t h a t it is questionable if this infiuence can always
be discarded. Next to p the properties 2 and ff of air are strongly
 DYNAMIC THEORY OF A SHORT HOT WIRE 2~~

do 8co i 8 ° ô~ 8ô~o~õ~ ôõ 8ôõo o o

~8~ dddddddddd
C',I

il
~ ~ ~ 0 ~
.? .?
~ 0
dddddddddddd
õo~5~~~oôõ
oôô oööôoo
dddddddddd
~~õ~~ h o
? T
::L
u3

dddddddddddd

O~
~dSS~s~sS~
ôooõõôoo
I I

0 8 dddddddddddd

- ~õ~~~~~~sõ
ooôoo õõoo
dddddddddd
~. o

dddddddddddd
tO
~o co o
=~~d~ °d-~~A°°o°õo° 8

ù<
0 0 0 ~ ~ ~ ~ ~ ~ 0
0 0 0 0 0 0 0 0 0 0
dddddddddd
I I

~õ~ ,~ d d d d d d d d d d d d

0 0 0 ~ ~ ~ ~ ~ ~ 0
0 0 0 0 0 0 0 0 0 0
dddddddddd
o
I I

d d d d d d d d d d d d
~õSõ~~=~~õ
u3 oooõoõõôo
dddddddddd
I I
0
0 ~.~ o
d d d d d d d d d d d d

O0 0 0 0 0 0 0
d d d d d d d d d d
TT ?T
O0 0 0 ~r
XX X X
d d d d d d d d d d d d

a,~~S5 °°-~~A~~ôôô8
 256 R.E. DE HAAN

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
dddddddddd dddddddddd
I I o I I

d d d d d d d d d d d d dddddddddddd
õ~o-oõ~~~~
oo~~~ooo~o
dddddddddd ädddddddd~
0 0 0 0 0 0 0 0 0 0

I I o I I
¢xl

H il
> d d d d d d d d d d d d d d d d d d d d d d d d
õ~-~~°~~ö~
ooôõõõoooo 0 0 0 0 0 0 0 0 0 0
< dddddddddd dddddddddd
0
I I 0
I I
C-I ".0
dddddddddddd dddddddddddd

~8~~~~oooo
dddddddddd
0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0
dddddddddd
cO cO
I I
Il I[
dddddddddddd d d d d d d d d d d d d

°~s-~~°°8°°o°8- oo °°-~~°°°°o°8o° 8
8 ~ Oxl W'~

ooôôõôõooo ~ õ~~~~~~~~oo
0 0 0 0 0 0 0 0 0 0
dddddddddd dddd~ddddd
~ 7 7 o
I I

d d d d d d d d d d d d d d d d d d d d d d d d

0 0 ~ ~ ~ ~ ~ 0 0 0 ~g~~~~~~~~oo
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
dddddddddd d d d d d d d d d d
~ 7 7 o
I I
[[ ir
> d d d d d d d d d d d d d d d d d d d d d d d d

~L ~ôôooooooo 0 0 0 0 0 0 0 0 0 0
dddddddddd d d d d d d d d d d
~ ? .7 o ~
T0
~ ~ ~ ~ ~ ~ ~ ~ 0 0
?
d d d d d d d d d d d d dddddddddddd

oo~~~~~~~~~õoo
8880000000 0 0 0 0 0 0 0 0 0 0
~I dddddddddd dddddddddd
cO
I I
[I
N~N~m~~NNNNN~~~~
, d d d d d d d d d d d d d d d d d d d d d d d d
t

~o~ °-~~°~~°°~8- oo
 DYNAMIC THEORY OF A SHORT HOT WIRE 257

d e p e n d e n t on T. This can be represented b y a simple power law:

- -- so Pr = constant. (28)

Reminding t h a t f~w -- Tg can a m o u n t to 300 or 400 °C for the center

p a r t of the wire one can expect a m a r k e d inIluence on the cosh-
function in (10), resulting in a flattening of the center p a r t and a
consecutive steepening of the flanks. This is clearly shown in [21 b y
e x p e r i m e n t and exact numerical c o m p u t a t i o n ; the m a g n i t u d e of
the deformation is from 10 to 20%. Virtually the same results are
obtained b y expanding in a series. We r e m a r k in this respect t h a t
the exact linear solution (I0) can be represented a c c u r a t e l y b y a
series of five even powers. Subsequently the power law of # and Jt
can be a p p r o x i m a t e d b y a series of two even powers and inserted in
the quantities A and B of (3).
Our main interest lies in the influence on the d y n a m i c a l corrections
as given in Tables II to VII and therefore we inspect the appropriate
equation (8). I n t r o d u c i n g here the m e n t i o n e d a p p r o x i m a t i o n s the
coefficient of _/(x) becomes a weak function of the a r g u m e n t x,
likewise the right h a n d side becomes a slightly corrected cosh
function. We m a y expect t h a t the only influence is a small change
of the correction factors in the tables of say 10 to 20% which can
be neglected.
A second question arises when we compare the acoustic wave
superimposed on the m e a n airflow as proposed for the d y n a m i c
calibration with the isothermal concept of incompressible turbu-
lence. We therefore consider a long hot wire with a perfect feedback
a l t e r n a t i v e l y situated in b o t h flows.
In a simple progressive plane sound wave d p - õc dU, in the
antinode of the velocity fluctuation in the resonance t u b e we can
pur dp = gõc dU in which ~ ~ 0. I represents the influence of a
resonance. According to (28) and the formulae for an isentropic
process we can write the following:

d2 dT d# dT
.
~0 . ca . { o ' . ~ c~ ~ - ' ca = c~ = 0.8
(29)
dp _ dU _ Ü
/5 - g ( K - I)M Ü inwhich M-- c
258 R.E. DE HAAX

dT dU

(29)
dp _ dU
- «M--
Ü
T h e heat transfer e q u a t i o n (3) can be written:

Inserting I = i + dI, U = Ü -+- dU, Tg = Tg @ dTg, 2r = #f + d~r,

#f = fif + d#f yields

-1- B174/?e°.4 1 + ~d,~~ + 0.4 dpg

/ôg @ 0.4 - -Ü --

(~~ d~~) o

Expressing all variations in dU/U b y means of (29) and p u t t i n g

dTr = ½ dTg yields after arranging the terms

dl
/ d U = 0.2,~r -Bl~eO.
-
4 (Tw - - Tg) X
-] j Ü i2Rwl
Nil 4_ «M ~A, -t- t3B1Be°'4 AI -}- B1Re°'4 "g ]
0 •4B1/~e °.4 -- ~3Tr(K -- 1) 0.4B1Æe °.4 Tw-- ~g
(30)
in which
Tg
= ~c~(~ - 1) ~ [ , ~ = ~ + 0.4 - o.2o,(,~ - 1) ~~ff

A few t e n t a t i v e c o m p u t a t i o n s show the influences to be certainly

less t h e n 5 0 , c o n s e q u e n t l y t h e y m a y be neglected.

§ 8. A three-dimensional effect of the airflow

On b o t h sides of its active p a r t the hot wire is c o p p e r p l a t e d up to
a d i a m e t e r of say 10 to 30 ~, these parts in t u r n are soft-soldered
to the prong. At the corresponding discontimfity in d i a m e t e r a
 D Y N A M I C T H E O R Y OF A S H O R T H O T W I R E 259

sF 2,
r

-2[
\

'\\ //

-5 J
_X
05 10 ~~

* Fig. 1. L o c a l f l u c t u a t i o n ~ of t h e r e s i s t a n c e .

"dead-water" zone can be expected where the airspeed Ü will be

nearly zero at the ends of the active part. This will eause a reduction
of the heat flow to the airstream and thence increase the slope of
the temperature curve on the hof wire. A marked influence on the
correction factor in Tables II to VII can be expected when this
effect is sufficiently large and when there is a distinct influence of
frequency. Concerning the latter point it appears that for the dead-
water zone Str ~ I for Re >~ 1 but on the other hand Re Str = o(1)
for Re ~ 1 ; so we have to investigate the effect for low airvelocities.
We remarked in § 7 that equation (7) eould accurately be solved
by introducing a short power series; this procedure can here be used
when in (7) the constant Ü 0.4 is replaced by Ü0.4[1 - (2x/l)m~. By
varying the value of the exponent m the width of the dead-water
zone can be changed. Performing in this way some computations
showed us that the slope of the flanks of the temperature curve of
the wire is increased with the maximal percentage of 10 to 20. So
again the effect on the dynamic correction factors is small and can
generally be discarded.

ReeMved 4 M a y 1970
260 DYNAMIC T H E O R Y OF A SHORT HOT W I R E

REFERENCES

[1] DÆVlES, P. O. A. L. alld M. J. FISHER,Proc. Royal Soc. A 280 (1964) 486.

[2] DAVI~S,P. O. A. L. and M. R. DAVlS, ISAV Report im. 155. 0ctober 1966.
[3] DÆVlS, M. R. and P. O. A. L. DAVlES, ISAV Technical Report no. 2. April 1968.
[4] BETCHOV,R., Proc. Koninkl. Ned. Akad. Wetensctiap/il (1948) 721.