A SPINNING HOT-WIRE ANEMOMETER FOB. SIMULTANEOUS MEASUREMENT CF u, v and w
by M. G. McLeod
A SPINNING Har-WlRE ANEMOMETER FOR SIMULTANEOUS MEASUREMENT OF u, v and w.
by
M. G. McLeod
'
..
Manuscript received December
1967
ACKNOWLEDGEMENTS
~his report was submitted as a thesis for the B.A.Sc. degree to the Department of Engineering Science in January
1967.
Gratitude is expressed to Dr. G. N. Patterson, pirector of the Institute for Aerospace Studies, for providing the opportunityland faciliti.es to do this research.
The author is indebted to Professor BEtkin for his guidance and many helpful suggestions.
Many people co-operated in the experimental work. Special thanks is due Dr.P. C. Hughes, Mr. T. R. Nettleton and especially t~Mr. D. Surry.
This work was made possible through the financial support of the United States Air Force under the research and technology contract
AF33(615)-2305,
of the Control Criteria Branch, Flight Dynamics Laboratory.SUMMARY
A rotating hot wire anemometer is a device for determining
simul-taneously the velocity components at a poipt in a fluid flow. This is accomplished
by analyzing the output waveform of an inclined hot wire which is rotated on a probe shaft. Two methods have been deve19ped for separating out the velocity
components. The first me~hod employs the first and second harmonies of the signal. The second method uses a calibration curve formed from the signal amplitude and
i t,s De level.
A rotating contact system using mercury was developed for noise-free tapping of the signal from the rot,ating hot wire.
1. 2.
3.
4.
5.
TABLE OF CONTENTS NOTATION INTRODUCTION EXPEI\IMENTAL APPAR.ATUS2.1 Description of the Probe
SUMMARY OF THE ROTATING HOT WlRE ANEMOMETER THEO~Y
DEVELOPMENT OF PROBE TECHNIQUES
v 1 2
3
4
104.1 The Thesis as a Design Problem 10
4.2 Experimen4al Verification of Hot Wire Signal 10
4.3
Two Techniques for t,he Determination of Velocity Components 114.4
Experimental Verification of the Techniques13
4.5
Accuracy of the Techniquesi4
4.6
Usefulness of the Techniques 15CONCLtJDIING REMARKS REFERENCES APPErqDIX
16
17
18
v
9 cp u v w a,z
i
i
w
l ... ... ... ... ...NOTATION
flow velocity vector
component of velocity vector perpendicular to hot wire
,
I
angle between V and the positive prob~ axis
angle made by projecting V into plane perpendicular to probe axis
x component of V
y component of V z component of V
rotational displacement of probe shaft
angle between the hot wire and the plane perpendicular to probe axis
I
I
THE FLOW VELOCITY VECTOR
I 1/ k---~ - - - - --- - - j Jf};;""'"
'I
..., ... I / U. _____---=-
....
:.=-1/ //
/
Ix..
PROBE GEOMETRY
E hot wire signal voltage, volts
De
level or average of hot wire signal, voltsfirst harmonie of hot wire signal, v0lts seeond harmonie of hot wire signal, volts
(voltage)2 intereept in King's Law equation, volts2
1. INTRODUCTION
There is a definite need for an instrument capable of determining
simultaneo~sly all of the components of the fluid velocity vector at a point.
Such an instrument would h~ve many practical applications, an example of which is the wake survey used in helicopter development.
A stationary hot wire probe is capable of measuring only the com-ponent of the flow velocity which is perpendicular to the wire. Furthermore, it is relatively insensitive to small changes in this velocity component produced by a yawing motion of the probe. By rotating the shaft of the probe it is theoretically possible to evaLuate all components of the flow velocity and at the same time maintain a high sensitivity to changes in the orientation of the velocity vector.
All previous work on this probe was done as a Bachelor's thesis tn
1965
and1966
by Mr. E. Y. Sarafian at the University of Toronto Institute for Aerospace Stud~es. A complete theoretical analysis of signal waveforms and their harmonies was carried out and a prototype probe was constructed. Be-fore any experimental work could be done with this probe, however, it was necessary to perfect a noise-free electrical contact system to conneet the terminals of the revolving hot wire to the bridge circuit.The purpose of this thesis was to complete the development of the prototype rotating probe, to compare the signals generated by the probe with theoretical signals, and to devise a technique for determining the com-ponents of the flow velocity from the signals.
2. EXPERIMENTAL APPARATUS
The apparatus used in this experimental analysis, with the exception of the rotating probe itself and the andlog computer consists of the same equipment as that used in normal hot wire anemometry.
(lJ Wind Tunnel
~he test section of the U.T.I.A.S. low-speed wind tunnel is used to create a smooth airflow whose velocity is accurately known. The calibration and testing of the probe is carried out here. Although higher speeds were available, tests were not conducted above 110 feet per second because of vibra-tion of the probe.
(2) The Rotating Probe
A hot wire probe which is fixed measures the component of velocity normal to the wire. By rotating the shaft of the probe a periodic signal is produced, since the normal velocity component varies at the same frequency as the shaft rotation. In order to indicate the rotational position or phase of the shaft at any instant, a second signal - the reference signal is also generated by the probe. This is explained in the detailed description of the probe which will follow.
(3) Disa Constant Temperature Anemometer
This instrument is used to keep the resistance of the hot wire at a constant value and thereby to fix its temperature. A bridge circuit is em-ployed to sense small changes in resistance. The output consists of a signal voltage which is that supplied by the anemometer to the hot wire to keep its temperature constant. Meters on the instrument give' a readout of the D.C. level of this signal voltage and the root-mean-square A.C. level of the signal voltage.
In order to analyze the output signal the following two electronic units are used.
(4)
Pace Analog ComputerBy means of a linearizing circuit in the computer (described in the Appendix), the output voltage is linearized and made to be numerically equal to the flow velocity component perpendicular to the hot wire. This is used in a comparison of the theoretical and experimental velocity waveforms.
(5) Bruel and Kjaer Wave Analyzer
Any periodic signal is composed of characteristic harmonics which are easily separated by means of a wave analyzer. Dnly the first two harmonics of the signal are used because the higher harmonics become progressively more difficult to measure with accuracy.
(6) Dual Beam Oscilloscope
For visualization, an oscilloscope with a Polaroid camera was
used to monitor and record all output waveforms. These included the linearized
and non-linearized signals, the first and second harmonies of the non-linearized
signal, and the reference signal. A schematic of the equipment used is given in figure 1.
2.1 Description of the Probe
The rotating probe has four major components: the hot wire,
its contact system, motor and reference signal generator. The hot wire is the
transducer element which is sensitive to the flow velocity. Rotation of the
shaft on which the hot wire is mounted is accomplished by means of a small
6
volt Pitman motor. Electrical contact between the ends of the hot wire onthe rotating shaft and the fixed part of the probe is made by means of a mercury
contact system. This is described in detail in the appendix. In order to
re-late the Disa signal to the rotational position of the probe at any instant, a small reference signal generator was placed on the free end of the motor. It
consists of a cylindrical permanent magnet on the motor shaft, rotating within
a fixed coil. This produces a sinusoid of the same frequency as the hot wire
signal. Ta any point on the Disa signal there corresponds simultaneously a unique voltage on the reference sinusoid. This unique value can be calibrated
against the known angular displacement of the shaft. For example, the minimum
signal voltage occurs when the flow velocity is most nearly parallel to the wire.
Yawing of the probe within the wind tunnel test section was
accomplished by rotating the base of the probe to the required angle. The probe
3 . SUMMARY OF TEE ROTATING HOT WIRE ANEMOMETER THEORY
1. King' s Law
The output of the hot wire anemometer is a voltage
E
which is related to the normal flow velocity Vn by King's Law:The constants A and B can he determined for any particular wire by a calibration. If
E
2 is plotted against ~n then a reasonably straight line results whose slope is Band whoseE
2 intercept is A.In order to predict the theoretical rotating hot wire signal it is first Eecessary to derive an expression for Vn as a function of the flow velocity v(v,e,~).
2. Determination of the Velocity Waveform Vn
Any given v(v,e,~) may be resolved into a component parallel to the wire and a component Vn in a plane perpendicular to the wire.
The reference frame is chosen such that the z axis coincides with the probe axis.
Notation:
w
velocity of the airflow inclination of hot wire
rotational displacement of hot wire ...k-_--+ _ _ ---..
v -
-
-
-
-
-
-
J
~
',
/ / /',I
//
- - -
-
- -- ~/ Vv(v,e,cp)
a
=
angle between hot wire and the plane perpendicular to probe axis=
~First, we resolve V into x, y and z components.
u V sin
e
coscp
v V sine
sincp
w V cose
/
/
y
/ :t./' , / IA , / , /I
·z
I
i
.
1'# \ \ \,
\ \,
\ V ~_--L.-.----i_ - - -y
rn IZ I I n'lp,-
,
-
---
-
--- J
\ \ \,
I'Y'\Next, w is resolved into compon-ents perpendicular and parallel to the wire.
w cos
ex
=
w sinex
Now, u, vare resolved into components perpendicular and parallel to the trace of the wire in the x y plane.
ln u sin ~-v cos ~
m u cos ~+v sin ~
The vector m is resolved further into components perpendicular and parallel to the wire
m m sin
ex
n
~
=
m cosex
The only component of velocity which affects the hot wire is the one which is normal to the wire. This component has the following breakdown:
Let the resultant velocity perpendicular to the wire be Vn0
Vn2 (wn - mn)2 + (ln)2
(w cos
a -
m sina)2
+ (u sin~
- v cos~)2
=
[V cos e cosa
(u cos ~ + v sin ~) sin aJ2 + [u sin~
- v cos~J2
[V cos e cosa
(V sin e cos ~ cos ~ + V sin e sin ~ sin ~)sJ..n
..
a
]2+ [V sin e cos ~ sin
~
_ V sin e sin~
cos~]2
[V cos e cosa -
V sine cos(~-~)
sina]2
+
[V sin e sin(~ _~)]2
For convenience we choose to measure ~ relative to the angular displacement ~ .
V
n V [ (sin
a
cos~
sin e - cosa
cos e)2 + (sin~
This is the fundamental formula of the rotating hot wire probe. There is one other component of velocity perpendicular to the wire. This component, the tangential one, is due to the rotation of the wire itself through the medium.
Y"oto..tiono./
I/elocit)' clish'-\ bu:tlon
Y"oto.t ion
It will be seen later that the contribution to the signal by this component is negligibly small.
Figure
5
shows the non-dimensional normal velocity Vn/V plotted against B, the angular rotation of the probe. As the probe is yawed (e increased) to45
0 , the amplitude of the waveform increases to a maximum. At45
0 the wire rotates from a position parallel to the flow to a position perpendicular to the flow.3.
The Anemometer SignalThe actual output signal voltage is given by King's Law.
(A +
BV~
[(sina
cos~
sine -
cosa
cos B)2 + (sin~
Thiswaveform, which will be called the unlinearized probe signal, is similar in shape to the linearized probe signalor velocity waveform. At any fixed velocity, the unlinearized signal increases in amplitude when the probe is yawed. This is demonstrated in figure
6.
Also, at any particular yaw angle 9, the amplitude of the signal may be increased by increasing the flow velocity. This can be seen in figure7.
It should also be pointed out that as the velocity increases theDÇ
level or average value of the signal also in-creases.A value of the wire angle
a
must now be chosen.4.
Optimization of the Wire Anglea
for Best Sensitivity to 9v
Consider a standard hot-wire probe mounted with the wire normal to the flow. In this position the instrument is very insensitive to changes in the angle of incidence of the flow velocity. If however, the wire is set with its direct-ion more nearly parallel to the flow, this sensitivity is
immediately improved. Suppose the wire is rotated so that its angle
a
increases from 00 to90
0 • Then the normal velocity component being measured is~iven by Vn
=
V cos a and the::::~t~:it~(:~ :~"1~~~I:
:v
iSGraphically,
v'"
= V Cos 0( S,: k VsinO(v
kV
~---~--~~~
o
~Oonor,.,..,a.l c.o".... pO I'"le ",t
It is easily seen that the sensitivity of the probe to
directional changes in the flow velocity is at a maximum when the probe wire is nearly parallel to the flow. For maximum sensitivity to changes in the flow direction then, it is to our advantage to mount the probe wire as nearly
parallel to the probe axis as possible.
There are two practical factors which must be taken into account before the best wire angle can be established. Both concern the probe wire
supports.
(1) As the wire is rotated
(a
is increased) from a position normal to theprobe axis, one probe support must be extended. There is a limit to this
extension - at high speeds the centrifugal force due to the rotation of the
probe causes this support to cantilever outwards. This puts extra: .stresses
on the hot wire itself a nd shortens its life. It can also be seen that in this case a much longer wire filament must be used to span the supports. This may introduce extra problems of vibration.
(2) As the wire angle
a
isincreased, the range of possible incident velocities which the probe can successfully
measure is reduced.· If the flow
is incident upon the probe at an
angle
e
greater than (900 -a)
it will at some point in the probers rotation pass over the wire itself. This introduces an undesired disturbance into the
only the flows with velocity
vec~or lying inside the cone
generated by the rotating wire
are not disturbed by the probe before measurements.
Therefore, 8max
=
90° -a
This leads to the question - How large a 8max must the instrument be capable of measuring for it to be useful? Suppose we
con-sider
2
flow with averagevel-ocity ~ and turbulence velocity
of ~ whosetime average is zero.
Then all possible flow
veloeities li~ within a cone
generated by ~ + ~max having
-l/V
maxI
angle 8
=
sin ~. lVI
If, for example we allowan angle 8
=
450 we are allowing the turbulencemag-nitude to be 1
lvi
or70.7%
of the mean velocity. The probe is useful formeasuring
flO~
wIth this amount of turbulence.The choice of a wire angle is q~i~e arbitrary and a compromise
must be reached between sensitivity to 8 and useful 8-range of operation.
The value of
a
for subsequent experiments was chosen to be 450 .This allows a high sensitivity in the range 0
<
8<
450• As shown before
this choice of
a
allows for a very adequate turbulence level of70.7%
of them.ean velocity.
e
It may be desired to ~ncrease the value of
a
depending upon theapplication of the probe. This will increase the sensitivity but reduce the
4.
DEVELOFMENT OF .PROBE TECHNIQUES4.1
The Thesis as a Design ProblemFundamentally, this thesis is a problem in instrument design. It is required to find a technique for separating out and determining the flow velocity components from the hot wire signal.
The probe makes available two signals - the linearized signalor velocity waveform, and the unlinearized signal as it comes from the Disa
instru-ment. In order to obtain the linearized signal, additional electronic compon-ents are required {see appendix). Since simplicity is an advàntage for any instrument it was decided to use the unlinearized signal provided it could furnish ~he necessary information about the flow velocity.
There are three units of information which must be o~tained from the signal, namely, the three flow components. This requires at least three measurements. The signal variables most readily available are DC level, first harmonic, second harmonic, phase dllgle, RMS value of the signal, and peak to peak signal voltage.
Each of the above can be determined fr om the theoretical wave-form. This has the advantage that a complete analysis of the probe can be carried out theoretically rather than empirically. The analysis, however, is useful only if the actual probe signal coincides closely with its theoretical coUnterpart.
The first requirements, then, is to examine the output voltage of the rotating hot wire and to compare it with the calculated theoretical signal.
4.2
Experimental Verification of Hot Wire SignalThe first step as in standard hot wire anemometry was to calibrate the probe to find the King's Law constants A and B. This was carried out in the wind tunnel test section with the wire normal to the flow and the probe shaft stationary.
The probe Shaft was then rotated at a speed of about
40
cycles per seconde (Speeds of60
cycles per second and its associated harmonics should be avoided for the noise level to be minimized.) Oscilloscope photographs of the probe signal were then taken for yaw angles of0,
10,
20, 30,
40 and45
degrees, and at each yaw angle flow speeds of0,
40
and80
feet per second were used. Since the wire anglea
was45
degrees, the yaw angle was limited to45
degrees in either direction.Figure
8
shows the photograph of a typical signal for values of 8 =40
degrees and V =81.1
feet per seconde In figure9
this waveform is com-pared with the theoretical waveform as calculated from King's Law using the calibration curve wire constants.Several observations are immediate. The DC level of the two curves is not the same. The cusp of the probe signal is not as pointed as it is theoretically predicted to be. THis can be attributed in part to the lack
of high frequency response of the electronics which is used to produce the signal. (The cusp caq be considered as made up of high frequency Fourier components
which are attenuated.) Another factor which may be involved here is an effect
observed in ordinary hot wire anemometry. If a hot wire which is normal to the
flow is yawed in the flow, the normal component is given by Vn ~ V cos 9. But
the response of the anemometer does not obey this law as the wire becomes more
parallel to the flow. At 9
=
90 degrees the calculated normal velocity componentis not zero but a small positive value.
~he dip in the signal at ~
=
180 degrees is also observed at yawangles down to 30 degrees. This dip is a characteristic 0f signals whose yaw
angle exceeds the hot wire _ '. angle
ex
and should only be present in signalswhere 9 is greater than 45 degrees. It cannot be attributed to errors in
ex
and6, for
ex
was determined to± 1 degree and
6 to ± ~ degree. Self-induceddistur-bances are not the cause. These would be created when the wire is parallel to
the flow and would affect the wire when it had rotated to a position perpendicular
to the flow. The probe shaft rotates once every 1/40th of a seconde In this
time a flow of 100 feet per second has moved 2.5 feet which is some three orders
of magnitude greater than the hot wire length. Therefore, all flow disturbances
created by the wire in the parallel position are downstream long before they can
be detected by the wire in a position normal to the flow where the irregularity
occurs. It is felt that the dip in the signal was actually caused by a flow
deflection, produced by the shroud over the probe. The effect of this deflection
is that the angle 6 relative to the hot wire is actually greater than the true
angle 9 relative to the probe.
Photographs for one hot wire were taken with the probe shaft
ro-tating first in one direction and then in the other. (Figure 11.) The second
photograph indicates that this wire was not mounted symmetrically with respect
to the mounting needles. This demonstrates the importance of precision in
fix-ing the hot wire to the probe.
Another observation made in the first experiment was that the
nor-mal component of velocity produced by the rotation of the wire itself is negligible.
The DC level with the probe stat~onary was 3.87 volts, while with the probe
rotat-ing, it was 3.92 volts. On the calibration cruve this corresponded.to a velocity of 0.3 feet per seconde
Although the theory and experiment do not agree closely, it was
decided to carry on with the analysis. The criterion for the usefulness of the
probe is not whether individual waveforms agree with theory, but whether the
changes which are produced in the waveform as V and 6 are varied can be used profi tably.
4.3 ~wo Techniques for the petermination of Velocity Components
JWo approaches were finally adopted for the separation of the V
and 6 components:
(1) analysis employing the signal harmonics El and E2
(2) construction of a set of calibration curves
and twice the fundamental frequency respectively. Since the fundamental fre-quency is equal to the rotational speed of the probe, the filters must be matched to this rotational speed.
The second approach offers simplicity. No special electronics is required over and above that used in normal hot wire anemometry. If the probe is accurately calibrated in a wind tunnel and is consistent, it will have corresponding accuracy when put into use. The problèm remaining is to find the best variables in terms of which the calibration is to be made.
No mention has been made to this point of the velocity co-ordinate
~. This is because ~ can always be determined quite easily and independently of V and 8 • A phase meter is employed to give the phase difference between the reference signal and the hot wire signal. These signals may requirefiltering at the fundamental frequency before their phase dif~erence can be measured. The reading of the phase mèter will always differ fr om ~ by a constant.
Since ~ is a relative angle, suppose it is chosen that ~
=
0 when the flow is incident upon the probe in the plane which is vertically below the shaft (the plane wtiich includes the probe support). Suppose, moreover, that the phase meter reads ~o for this orientation of the f.low. If the flow angle ~is changed to " then the phase meter will read ~o + , or ~o - , depending upon the direction of rotation of the probe shaft.(1) Analysis by Means of Signal Harmonics
It can be seen that when El is plotted against 8, a family of curves is generated, each curve being for a particular velocity. The curves are nearly linear and pass through the origine A similar set of curves is given ~y plotting E2 against 82 • The unique aspect of the two curve sets is that they appear to depend upon the flow velocity in the same fashion. This suggests a division of E2 by El to lose the functional dependence upon V and the result is a fUnction of 8 alone.
Supponse El = f(V) E2 = g(V) then E2
$t
El = f V 8 828 k8 where k is a constant by the hypothesis that g(V) - k f(V) This result is further supported by dimensional analysis. Since E2/El and 8 are dimensionless then k must also be dimensionless.
The theoretical values of the first and second harmonics were evaluated by computer from the defining formulae.
1 El (v,8)
=
7T
rE
cos t) dt) 0 27T E2(V, ) 1I
E cos 2t) dt3 7T 0Not only is the theoretical E2/El independent of velocity, but
its variation with ~ is linear. (See figure 13).
Once 8 is kno~, the flow velocity magnitude can be obtained
from the graph of (DC level) against V. This is a family of curves with a
curve for each 8 . (See fig~e 12)
(2) rConstruction of Calibration Curves
Suppose that for a given flow velocity, the probe is yawed so
that the angle 8 is varied. As 8 increases from 0 to 450, the peak to peak
amplitude of the signal increases (See figure 6). This peak to peak amplitude will be increased still further at any 8 if the flow speed is increased. (See figure 7.)
Af ter a careful analysis it was decided to plot the peak to peak
amplitude against the square of the DC level. Va~es of 8 ranged from 0 to 45
degrees and velocities from 0 to 100 feet per seconde Calculations using King's
Law were done again by computer.
2 The results are shown in figure 14. As velocity increased the
(DC level) increases ac c or di ngly . Lines of constant velocityare almost
ver-tical. By inereasing
e
at any velocity the peak to peak voltage is increased.Lines of constant 8 are almost linear and have an (E0
7
2 intercept of A.The significanee of this figure is that by obtaining simultaneously the peak to peak signal amplitude and the DC level, it is possible to determine
both V and 8 in one operation from the calibration eh art,.
4.4 Experimental Verifieation of the Techniques
(1) Harmonie Analysis
The probe was eali"Qrated a:s··hefore and the ea.libration c.urve
drawn. T~en with the probe in rotation, readings of DC level, first harmonie
and seeond harmonic were recorded. (The same values of 8 were used but this
time the intervals of the velocity span were shortened - 0, 8, 13, 35, 56, 72,
86, 95, 99 and 108 fee.:t per --s-eeond·. ) !.Phe-fr€-s·peeds w€rre· used beeause the wind tunnel motor speed was not continuously varia.ble, but opera.ted in steps. The
wind tunnel had to be stopped whenever it was necessary to change 8. As a
re-sult the same velocity was never aehieved twiee.
The curves of the. harmonies ~l and E2 were not plotted up
di-reetly. Instea.d, by plotting the harmonies agaihst V, their values were
inter-polated at velocities of 100, 81, 36 and 16 feet per seeond. These are shown
in figure 13.
The fundamental discrepaney between theory and exper"iment lies
in the slope of the curve. The eXperimental harmonie ratio is 1.4 times larger
than its theoretieal v&lue. Linea.rity was approaehed at high veloeities. The
non-linearity at low flow veloeities is probably due to fluetuations in the
(2) Calibration Curves )
J
Af ter calibration, oscilloscope ... ,15trot·ographs were taken of the
waveform for values of
e
and V as inpar~t
,
F -"Tne DC level was recordedsimultaneously with the photograph. Fig -_ '5-shows the amplitude plot and
should be compared wi th figure 14. FOT . ')si..ven V and
e,
the experimentalsignal amplitude is
a~oJJt b'tW.t
_:
:
~±~~l:
"
value. Th~s
amplitude reductionis sufferrd most by slgnals. f
e
-Lt q0+~:~.~e~s. ThlS effect has ~lreadybeen accounted for by the dlp at the top
-or
t'~al and the attenuatlon ofthe cusp. -,
It was not possible uo set up grid lirtes of constant velocity
since as explained previously, the wind tunnel could not reproduce any
particu-lar flow condition.
4.5 Accuracy of the Techniques
A precise analysis of sensi tivities of "the probe to V,
e
and cpusing the two techniques is beyond the scope of this thesis. However, a good
indication of sensitivity can be gained from an error analysis. The .accuracy of measurement of V,
e
and cp increases as V is increased. The amount of error in cp depends upon the ability of the phase meter to give the phase .angle between the reference signal and the first harmonie of the wire signal. Maximalerrors in
V
ande
are estimated using a combination of scatter of theexperi-mental points and the amount of separation this represents in the variable
be-ing determined from the graph. (1) Harmonie Method
(i)
e
is determined from the plot of E2/El againste.
(ii) !I'or this value of
e,
V is therr determined from the family of curves of (DC level)2 versus V.(iii) cp is determined by the phase angle between El and the reference signal. Maximum Error In
Velocity
e
V cp100 fps + -20 + - 2 fps + 10 60 fps + 2.50 +
-
2.5 fps + 1.50 30 fps + 3.50 + 1.5 fps +-
20 (2) Peak to Peak Signal Me'bhod{i)
e
is determined from the plot of peak to peak voltage arld (DC level)2 (ii) For this value ofe,
V is determined from the DC level curves.(iii) cp is determined by the phase angle between El and the reference signal.
Velocity 100 fps
60 fps 30 fps
4.6
Usef~lness of the Techniques ( Maximum Error 9 V ~3
0 +3
fps -+3
0 +3
fps ~3
0 +-
1 fps In cp +-
10 +1.5
0 -+-
20All experiments in this thesis were conducted under steady con-ditions. The effectiveness of these methods in unsteady conditions can only be postulated. High amounts of turbulence will cause the harmonics and the DC level to fluctuate. Since the filters and measuring devices have transients in their dynamic response it is not known whether the averaging processes involved would yield the true averages of the velocity components. This is one course for future investigation.
The photographic method when combined wi th a sensi tive averaging device for the DC level would give instantaneous velocity component values. For obtainirig mean values, statistical methods could be applied to a series of measurements.
For measurements in unsteady conditions, the probe shaft velocity
would have to be increased. If this is not done, the waveform will appear to distort before its cycle has bee~ completely traced out. Higher rotational
5 . CONCLUDING REMARKS
Spherical velocity components can be obtained simultaneously us-ing a rotatus-ing hot wire probe. A linearizus-ing network is not required for this determination.
A good set of mercury contacts is essential to the operation of the probe.
Although the voltage waveform of the probe has several anomalous characteristics, the variation of the signal with components V and 8 can be used profitably to identify these components.
TWo methods have been found for use and they appear almost equal in sensitivity to the velocity components.
One refinement required is the method of fixing the wire filament to the supports. If the filament was welded electronically on to the supports
it could be positioned with much greater prec.1.s1.on. When the wire is attached
by soldering, on the other hand, it must be placed to one side of the supports. Future probe designs could be reduced irl size if the motor was in an external unit and attached to the probe by means of a flexible cable. A small flywheel on the shaft could be used for reducing speed variations. By reducing the shaft size the mercury contacts could be miniaturized.
It is felt ~hat the rotating hot wire anemometer has great potential
1. Sarafiap., E. Y • 2. Kidron,. I.
3
.
Hinze, J.O.4
.
Schlichting , H.5.
Clauss, F.J. Kingery, M.K. REFERENCESRotating Hot Wire Velocity Probe, UTIAS Bach.
Thesis,
1966.
On the Measurement of Dynamic Flow Phenomena with the Cogstant Temperature Anemometer.
Disa Eiektronik
AlS,
Har lev , Denmark.Turbulence. McGraw Hill
1959.
Boundary Layer Theory.
New York, McGraw Hill
1965.
Sliding Electrical Contact Materials for Use in Ultrahigh Vacuum.
APPENDIX
The Rotatipg Contact Problem
The largest single technical problem encountered in the develop-ment of the probe was th at of devising a technique for establishing
noise-free electrical contact with the two hot-wire terminals on the rot,ating shaft.
It was imperative that a solution to this problem be found before experimental work could be resumed.
Many configurations and types of metal to metal contacts were
tried, but without success. Only during this experimentation was the importance
of clean electrical contacts realized. Firstly, the signal from the hot wire
had to be intelligible. With a varying resistance of contact this was impossible.
The second, and most important reason concerned the electronics of hot wire
anemometry. The bridge circuit of the Disa instrument responds to changes in
probe resistance. If the resistance sudden~y i~creases then the electronics
interprets this as an increase in temperature. In order to bring ~he
tempera-ture back to its normal value, the bridge circui~ reduces the probe current.
If contact is momentarily broken with the hot wire the current flow will stop.
Once contact is re-established, the low resistance of the cold w.ire is
respond-ed to with a large current surge. But the minute tungsten filament will with-stand only a limited number of these surges before it breaks. This explains the short wire life when poor metal contacts were used.
For test purposes it was essential that a single wire be used
under many conditions of incident air velocity. ~very wire has different
characteristics - a change of the probe wire in the middle of a test makes a comparison of the nonlinearized signal very difficult.
Mercury Contact System
The major obJect ion to the use of mercury in a contact system
lies in the nature of mercury itself. If it escapes from its container i~ will
corrode most metal surfaces - bearings and especial~y solder joints. For this
reason, tight mylar seals were used. The advantage of mercury is that as a
liquid it can be employed in such a manner as to give large surface contact area. Also, mercury readily forms an amalgam with silver and this results in excelleny conduction properties for the mercury-silver interface. For this
reason, silver discs were used for the rotating part of the contact. The
mercyry was placed in each of two cylindrical compartments enclosing the silver discs. Viscostty acting along with the whirling of the silver discs provided the centrifugal force necessary to hold the mercury to the cylindrical
retain-ing walls. In this position the chance of mercury escaping was considerably
lessened.
One slight disadvantage of this system is that as the speed of probe rotation changes, the contact surface area also changes resulting in a change of probe resistance. However, if the contact area is initially very large, the change of contact resistance will be very slight in comparison with the signal-producing resistance changes of the hot wire.
Linearization of the Hot Wire Signal
The hot wire signal can be made proportional to t~e normal
com-ponents of the velocity of flow by following the linearization technique
suggested by King's Law.
King's Law
s~ates tha~
B
In order to linearize the signal we must square it, subtract the E2 intercepB (A) and then square the result. This will be proportional to the component of flow velocity perpendicular to the hot wire.
The following program was used on a Pace Analog Computer:
-/OK,e ~/oo KIr: A
I----~/O TL 00 -/00 /(,I0E 2 _ /00 /(,~ A FSOOL - - - . FsooA
•
F500RE is the hot wire signal plot of 100 K12E2 with V the value of circuit by means of poteRtiometer POl. porportional to Vn0 By adjusting Tl on
AOI SM\ -A I="SOOL.
AO"
from the Disa instrument. By making a 100 K12A2 was determined and set in the
The output of amplifier A1H was then P02 a direct readout of Vn was obtained. A complete calibration was run with the probe in the wind tunnel. Errors in Vn were less than 2% in ~he range 10 fps
<
Vn<
100 fps, but as high as40%
for the range 0<
Vn<
10 fps.ROTATING PROBE ~
11
Referenee Signa IIYYY
Wave /\/V\-AnalyzerV\/V\/\.fV\
I
I
D
LJ--iLi nearizerHl
rvv-\
DCD
AC RMS Level Bridge Voltage DISA CONSTANT TEMPERATURE ANEMOMETER OSCfLLOSCOPE Unlinearized Signal I st Harmonie 2nd Harmonie11
Linearlzed Signalbercury motor slonol __ Shroud Not Icontoefs _g~nJlUJlo..r - _ _ ,..r ---~ ~, Shown
'"
/ I / ... / Reference Signal Varioble 6V Supply Disa Circuit Base RotatesCylindrical Plastic Shell
Mercury -
held to outside bythe
centrifugal force
supplied by
viscous acti on
I
1
_
.
. .
j~
-
-
e;j \
r
Ir
p;ij:t.:::::
Mylar Seals -
~etain
mercury
te
Disa
Circuit
Silver Discs -
silver forms
amalgam
MERCURY
CONTACT
Magnification;
1 1/2
times
V
nV
1-008
y~...-= =:...,,~ "~9=0
0,7
V ,0·6
9=10°
0(=
45°
0·4
9=20°
9=30°
0·2
0-0
K9=
40"
E
[VOlts]
7'(}1 LLL:/' ~ __---
~"
""
e=o
0 0e
=
10
,"-...
C(=45
0 0A =17'1
\,
" e =
20
B
=
3,68
\
"-.... e
=30
0 6'~I
V
=
100 fps
e
=
40
0 oe
=45
0
0 1800.1.3
o360
E
[VOlts]
7-0
6-0
50
0{ =45 oA= 17-1
B =3-68
o6=45
V=IOO fps V=80 V= 60 V=40
V= 20v=o
De
level
=6.93
volt s
vertical: 0.5 volts/cm
horizontal: 5 msec/cm
FIGURE
8
PHOTOGRAPH OF UNLINEARIZED SIGNAL
0(,
=
45
oe
=40
0v=
81.1 rpsA=17.1
E
[VOlts]
7
65
I,
I,
I I I I I I I I I,
I I I I I I,
,
I I I ""...
, , . , ,---
--, ,
"
De
Level ;' ;' ;' ;'"
"
"
"
"
0(=45 o V = 81·12 fps o 9=40 A = 17·1 B =3·92 \ \ \ \ \Theory---\
\ \ \ \ \ \ \ \ \ \ \Linearized Signal
DG
level =65.5 volts
vertical: 20 volts/cm
1 volt =1 fps
Unlinearized Signal
DG
level
=6.61
volts
vertical: 0.5 volts/cm
horizontal: 5 msec/cm
0(.=43
0e
=25
0 V=99.4
fps
A -15.0
B=).58
_.,' - =:;;;;;;- , ' 1 ' " ' - 1 ._""".
!
I ~ II!:;~ 11'lI!iIi
~
I
I ~ ~ II
II
I I /I
IIII!II !~
.
n' I,!l!!!'.~,
II
,
-,-
~",-
UI 0(.=45
oe
=45
0Probe shaft rotated in one direct ion and then the other
40
Callbratlon Curve 0 A=\7-\ B= 3-6830
DC Leve I Curves e.4~·•
e-30·
•
e.
O· •20
A-o
2
48
10§.
EI
I-O·
0-0'40'2
o 0(.=
45 A :r 17-1 B =3'68
o
IOOfps 6 81 fps • 36fps 16fpsE
peokta
peok3·0
[ VOlts]2.0
I-0(= 45 o A= 17·1B= 3-92
2-0 2e,
,
120fps I I,
I I,
130 fps I 1,
,
I I ,50fps I,
,
I,
170fps I I,
I I Ie
=
45
,
,
I I I 'IOOfps 1 o9=40
o o9= 30
o9
=
20
o I I , ....9
=
10
I I • ' • I , I . , I I+
T
I I I,
o9=0
30 3e40
45(oe
LEVEL)2Epeak
ta
peak2·
rVO
lts]
1'61·2
0'80·4
o ()(= 45 As l7'1 8=3'92e=45°
oe-40
1-e_30
0e_20
0•
e_lo
O•
•
o.oJ
~o
0 • 0 0 0 0 0 0.eco·
E
peak to peak 3·0 [ vOlts]2·0
1·0 O{= 45 0 A= 17·1 B=
3·92 THEORY EXPERIMENT -/ / /'"
'"
/ / / /"
"
/ / / /"
,,"
/ . , / ' / / ' / / ...."
/ '" / ... /"
"
'"
"
"
/,,"
/"
/.,,"
/'"
/ , , / / ." /"
I /'" / ~ ~ 20 25 30 /.,
/ / .... '" /"
/.,
/ / /"
.,
-
-=----35 /"
"
"
/ /"
"
/"
....- "....-40'"
/ ... ....-/ / / / ... .... ... /"
"
/ "....-... " 45"
/9=45°
,," 9=40°
9=20°
"
9=45
°
9=40°
9= 20°
.>(". 50 2 (De LEVEL)VrL'\.S TECHNICAL N(Y]$ NO. 12)
Institute for Aerospace Studies, University of Toronto
~
A Spinniug Hot-Wire Anemometer for Simultaneous Measurement of u, v and w.
~1. G. McLeod 19 pages
1. Anemometers
3. Velocity Measurement
1. McLeod, M.G.
16 figures 2. Hot-wire Anemometer 4. Turbulence
11. UTIAS Technical Note No. 125'
A rotating hot wire anemometer is a device for determining simultaneously the velocity
components· at a point in a fluid flow. T.his is accomplished by analyzing the output waveform of an inclined hot wire which is rotated on a probe shaft. Two methods have
been developed for separating out the velocity components. The first method employs
the first and second harmonies of the signal. The second method uses a calibration
curve formed from the signal amplitude and its DC level. A rotating contact system us
-ing mercury was developed for noise-free tapping of the signal from the rotating hot wire.
Avoilable eopies of this report are limited. Rehwn thls eard to UTIAS, if you require a eopy.
UTL'\.S TECHNICAL NOTE NO. 125
Institute for Aerospace Studies, University of Toronto
~
A Spinning Hot-Wire Anemometer for Simultaneous Measurement of u, v and w.
M. G. McLeod 19 pages 1. Anemometers
3. Velocity Measurement I. McLeod, M.G.
16 figures 2. Hot-wire Anemometer 4. Turbulence
11. UTIAS Technical Note No. 125'
A rotating hot wire anemometer is a device for determining simultaneously the velocity
components at a point in a fluid flow. This is accomplished by analyzing the output
·w3.veform of an inclined hot \lire which is rotated on a probe shaft. Two methods have been developed for separating out the velocity components. The first method employs the first and second harmonies of the signal. The second method uses a calibration
CUl've formed from the signal amplitude and its DC level. A rotating contact system us
-ing mercury was developed for noise-free tapping of the signal from the rotating hot
wire.
,ll'lAS TEC1U'lJICI\.L NarE NO. 12)
Institute for Aerospace Studies, University of Toronto
A Spinnillg Hot-Wire Anemometer for Simultaneous Measurement of u, v and >T.
~I. G. McLeod 19 pages 1. Anemometers
16 figures
2. Hot-wire Anemometer
4. Turbulence
~
3. Velocity Measurement
1. McLeod, M.G. 11. UTIAS Technical Note No. 125'
A rotating hot wire anemometer is a device for determining simultaneous1y the velocity components at a point in a fluid flow. This is accomplished by analyzing the output waveform of an inclined hot wire which is rotated on a probe shaft. Two methods have been developed for separating out the velocity components. The first method employs the first and second harmonies of the signal. The second method uses a calibration
curve formed from the signal amplitude and its DC level. A rotating contact system us
-ing mercury was developed for noise-free tapping of the signal from the rotating hot wire.