Professional Documents
Culture Documents
Hot-Wire Calibration at Low Velocities Revisiting The Vortex Shedding Method
Hot-Wire Calibration at Low Velocities Revisiting The Vortex Shedding Method
Research Article
Hot-Wire Calibration at Low Velocities: Revisiting the Vortex
Shedding Method
Copyright © 2013 Sohrab S. Sattarzadeh et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly
cited.
The necessity to calibrate hot-wire probes against a known velocity causes problems at low velocities, due to the inherent inaccuracy
of pressure transducers at low differential pressures. The vortex shedding calibration method is in this respect a recommended
technique to obtain calibration data at low velocities, due to its simplicity and accuracy. However, it has mainly been applied in a
low and narrow Reynolds number range known as the laminar vortex shedding regime. Here, on the other hand, we propose to
utilize the irregular vortex shedding regime and show where the probe needs to be placed with respect to the cylinder in order to
obtain unambiguous calibration data.
y/D
St = 0.212 (1 − ), 300 < Re < 2000, (2)
Re
2
where 𝑓 is the fundamental vortex shedding frequency,
𝐷 the cylinder diameter, 𝑈 the flow velocity, and ] the
kinematic viscosity. Since the velocity, which is aimed at 0 I
being determined, appears in both the Strouhal and Reynolds
numbers, it has been proposed to employ the so-called 0
Roshko number defined as Ro = St Re = 𝑓𝐷2 /] [23]. −2
D
z/
The obtained Ro-Re—relation is usually favored over the 0 2 4 6
St-Re—relation given in (1) and (2), due to its linear and x/D
explicit nature [6]. When it comes to the exploitation of the
vortex shedding phenomenon behind a circular cylinder, it Figure 1: Schematic of the coordinate system centered around the
has been a common practice to restrict the method to the cylinder of diameter 𝐷. The streamwise, vertical, and spanwise
laminar periodic vortex shedding region or the so-called stable directions are denoted by 𝑥, 𝑦, and 𝑧, respectively. Hot-wire
measurements at constant free stream velocities were performed
range [12, 18–20], that is, 30 to 48 < Re < 180 to 200
at the intersection points of the red 𝑥-𝑦 grid. Positions indicated
[25] described through (1). Since this regime is susceptible to through blue crosses or crossed-circles and denoted by roman
oblique shedding which can alter the St-Re—relation [19, 26], numerals will be referred to and described throughout the paper.
special attention needs to be drawn to the end conditions
of the cylinder as well as the interpretation of the measured
frequency spectra [27].
While for flowmeter applications the Reynolds number long with a cross-sectional area of 0.8 × 1.2 m2 (height ×
range is commonly within the fully turbulent regime or the so width). The maximum speed with an empty test section is
called irregular range (although with quite noncircular shed- 70 m/s, and the streamwise velocity disturbance level is less
der geometries that exhibit a much less Reynolds number- than 0.025% of the free stream velocity. The air tempera-
dependent range), only very few have employed the region ture can be regulated within ±0.05∘ C by means of a heat
beyond the transitional range inheriting the discontinuities exchanger. For further information regarding the flow quality
in the St-Re—relation [25, 28], which occur between the in the MTL wind tunnel, the reader is referred to Lindgren
two Reynolds number regions defined through (1) and and Johansson [31]. Two circular cylinders with a nominal
(2). While there were practical reasons to favor the stable diameter, 𝐷, of 8 and 12 mm (giving a length-to-diameter
range for calibration purposes in the past, for example, due ratio of 150 and 100, resp.) were spanned at the middle of the
to the employment of zero-crossing counters [18, 29] or wind tunnel in terms of height and length. Throughout the
signal generators and oscilloscopes to determine the vortex remainder of the paper, the streamwise, vertical, and spanwise
shedding frequency by means of the so-called Lissajous (i.e., along the cylinder) directions are denoted by 𝑥, 𝑦, and
figure [5, 30], the shedding frequency can nowadays simply 𝑧, with the origin set at the centre of the cylinder as depicted
be determined by fast Fourier transform (FFT) algorithms. in Figure 1.
Also, the utilization of the irregular regime would avoid The hot-wire probe used for the present measurements
the aforementioned difficulties related to the appearance of is a standard straight hot wire with a platinum wire of
oblique vortex shedding. nominal diameter of 5 microns and a length of 1 mm. The
In the present paper the vortex shedding calibration hot-wire was calibrated in situ in the free stream against a
method for hot-wires is employed in the irregular region and Prandtl tube, which was connected to a micromanometer of
it is shown to be a robust and reliable technique. type FC0510 (Furness Control Limited), from which also the
ambient temperature and pressure were obtained. Calibration
3. Experimental Setup and curves were taken at the beginning and end of each set of
Measurement Technique measurements and fitted to a calibration relation proposed
by Johansson and Alfredsson [32]
The experimental data was obtained in the Minimum Turbu-
lence Level (MTL) wind tunnel located at the Royal Institute 1/𝑛 1/2
of Technology (KTH) in Stockholm. The test section is 7 m 𝑈 = 𝑘1 (𝐸2 − 𝐸02 ) + 𝑘2 (𝐸 − 𝐸0 ) , (3)
Advances in Mechanical Engineering 3
1.2
0.8 1 1.8
(fPee )∗
1 0 1.6
2
V)
100
E(
3
x/ 4 6 7
D 5 101 1.4
5 2 3 4
1 f (H 102
6 −1 0 y/D z)
0.8 0.9 1 1.1 1.2 Figure 5: Premultiplied power spectral density map for the voltage
signal from a hot-wire located 3D downstream of the cylinder
U/U∞
for different 𝑈∞ . The power spectral density is computed from
measurements within the wake (i.e., 𝑦/𝐷 = 2.5), while the mean
Figure 4: Surface plot of the evolution of the mean streamwise
voltage from the hot wire is obtained outside the wake region (i.e.,
velocity scaled by the free stream velocity in the 𝑥-𝑦 plane. White
𝑦/𝐷 = 6). The asterisk denotes normalization of the premultiplied
shaded areas denote 𝑈 within ±1% of 𝑈∞ .
spectral amplitudes to unity in order to visualize the fundamental
peaks as well as to ease visualization of the hot-wire calibration
relation. Note that the frequency axis is plotted in logarithmic scale
While the positioning of the hot wire for the deter- in order to emphasize the low velocity (i.e., frequency) region.
mination of the fundamental frequency has been clarified, Obtained calibration points for 𝐸 versus 𝑓𝑝 are highlighted by red
the position for the measurement of the mean voltage of circles, and the red solid line is (3) fitted to the data pairs.
the hot-wire that is related to the free stream velocity, that
is, 𝐸, needs to be checked. The surface plot of the mean
streamwise velocity depicted in Figure 4 clearly shows that
velocity. For smaller wind tunnels, where blockage might
irrespective of the downstream location, |𝑦/𝐷| > 5.5 appears
influence the quantitative results presented here, maps as
to be suitable to obtain the free stream velocity. Note also
shown in Figures 3 and 4 can easily be obtained and used for
that the turbulence intensity reaches the free stream level
future calibrations.
much quicker (not shown here) and therefore will not bias
the obtained mean voltage corresponding to the free stream
conditions. A less stringent lower threshold can also be used 4.2. Application of Vortex Shedding Calibration Method. Hav-
for further downstream positions. This rule of thumb is also ing established where to place the hot-wire probe in order to
valid when taking the results from the lower velocity as well pick up the fundamental frequency, 𝑓𝑝 , as well as an unbiased
as the smaller diameter into account. Hence, taking all results mean voltage from the hot wire, 𝐸, hot-wire calibrations with
with some tolerances into account, it is recommendable to both the 8 and 12 mm cylinders covering a velocity range
determine the fundamental shedding frequency for 1 < from as low as 0.15 m/s up to 10 m/s were performed. While
𝑥/𝐷 < 6 within 1 < |𝑦/𝐷| < 4, while the hot-wire voltage the low velocity end was restricted due to the stability of the
related to 𝑈∞ should be acquired at |𝑦/𝐷| > 5.5. These MTL wind tunnel, the higher end was mainly used to have
recommendations are also close to those given by Ardekani a sufficient overlapping region with the conventional hot-
[21], who gave 2.2 < |𝑦/𝐷| < 2.9 for 3 < 𝑥/𝐷 < 6, for the wire calibration against a Prandtl tube. For the performed
regime bridging, the stable and irregular regimes, namely, calibrations, the hot-wire was placed 3D downstream the
100 < Re < 430, in order to determine both 𝐸 and 𝑓 cylinder, and measurements were taken at 𝑦/𝐷 = 2.5 and
simultaneously. Nevertheless, it seems more appropriate to 6 (indicated with II and III in Figure 1) for 𝑓𝑝 and 𝐸,
comply to the limits given here, due to the extended Re— respectively, for each tunnel speed.
range employed, in order to obtain an unbiased hot-wire The result of such a calibration is depicted in Figure 5,
voltage which is related to the voltage in the free stream. where the premultiplied spectra are mapped for different
Having established the two regions in which (1) the fun- tunnel speeds and hence different 𝐸. Utilization of (1) and
damental frequency of the vortex shedding can be obtained (2) yields the free stream velocity, which coupled with the
from the voltage signal of the hot-wire and (2) the region measured voltage 𝐸 provides the data set for the calibration.
where the hot-wire is exposed to the mean voltage of the hot- Results from both cylinders have been depicted together with
wire that is related to the free stream velocity, it becomes a conventional calibration by means of a Prandtl tube in
obvious that a single hot-wire problem can be utilized for Figure 6. Note that for velocities beyond approximately 2.5
both purposes. Traditionally, one hot-wire is used to detect and 4 m/s for the 12 and 8 mm cylinders, respectively, the
the shedding frequency, while a second—the one that is relation given in (2) tends to underestimate the determined
calibrated—is measuring the voltage related to the free stream velocity, since the St-Re—relation reaches a maximum and
Advances in Mechanical Engineering 5
Table 1: Calibration cases depicted in Figure 6 with the utilized Reynolds number range and the coefficients for the best fit to the modified
King’s law, that is, (3). C3 corresponds to the conventional calibration method.
Fitting coefficients
Case Cylinder dia. (mm) Re
𝑛 𝑘1 𝑘2
C1 8 280–5500 0.4797 2.671 1.884
C2 12 140–8200 0.4896 2.723 1.775
C3 — — 0.4874 2.743 1.828
1.6 1.5 but such a deviation can still be considered fairly accurate
when compared to readings from pressure transducers at
1.4 differential pressures around or even below 1 Pa.
1.4
1.3
0 0.5 1 1.5 2 5. Summary and Conclusions
1.2
0 2 4 6 8 10 Hot-wire anemometry is since a long time the standard
U (m/s) measurement technique in turbulent air flows. Its indirect
working principal, however, calls for calibration against a
Figure 6: Conventional hot-wire calibration plot and magnified
view on the low velocity region (insert). Blue stars and dashed line known velocity to which the voltage from the hot-wire
are from a conventional calibration against a Prandtl tube, and red anemometer can be related. Commonly, the hot wire is cali-
circles and solid line are from the vortex shedding method (same brated against the pressure difference reading from a Prandtl
data as depicted in Figure 5), while the black squares are obtained tube; however, at low velocities, this becomes problematic,
similarly but with a cylinder diameter of 8 mm, instead of 12 mm. due to the inherent inaccuracy of pressure transducers at low
The lines are obtained from (3) (which inherently enforces that the differential pressures.
curve goes through 𝐸0 , denoted by +) fitted to the corresponding In the present work we have demonstrated the so-called
data. vortex shedding calibration method which has been known
in the past as the simplest and most accurate method to
calibrate hot wires at very low velocities but has only been
starts to decay from Re ≈ 2000 on. In order to utilize utilized to a very narrow Reynolds number regime. Here, we
the calibration points beyond this Re, the St-Re—relation propose to utilize also the irregular vortex shedding regime
for Re > 2000 was taken from well-established empirical and thereby extend the Reynolds number range.
data [25, 34, 35]. Although, “in practice, (2) is often also The flow field behind a circular cylinder has been mapped
applied for higher Reynolds numbers, [⋅ ⋅ ⋅ ], it should not be by means of hot-wire anemometry down to 6D. From these
used for Re > 5000” [35]. Nevertheless, it should be noted measurements, it could be inferred that both the fundamental
that for velocities above 2 m/s, the conventional calibration vortex shedding frequency as well as the hot-wire voltage
by means of a Prandtl tube is favored compared to any related to the free stream velocity can unambiguously be
other method, and therefore, the explained extension of determined irrespective of downstream position for 1 <
the vortex shedding method for higher velocities is barely 𝑦/𝐷 < 4 and |𝑦/𝐷| > 5.5, respectively. The proposed ex-
an esthetic operation. The Reynolds number range utilized tension of the well-established vortex shedding calibration
for the vortex shedding calibration method in the data method has been demonstrated to yield accurate calibration
illustrated in Figure 6 is presented in Table 1 for both cylinder data for hot-wires over more than a decade in velocity for a
diameters. The calibration constants, 𝑛, 𝑘1 , and 𝑘2 , for the selected cylinder diameter down to 0.15 m/s (restricted due to
fitted calibration data to the modified King’s law relation, that the employed wind tunnel and not the method itself).
is, (3), are also provided in the table for better comparison
between different calibration curves.
Coming back to Figure 6, the general agreement between Acknowledgments
the conventional Prandtl tube calibration and the results from
the vortex shedding calibration obtained with two different This research was done within KTH CCGEx, a centre
cylinder diameters over more than an order of magnitude supported by the Swedish Energy Agency (STEM), Swedish
in velocity can be appreciated. Hence, the utilization of the Vehicle Industry, and KTH. Professor P. Henrik Alfredsson is
irregular vortex shedding regime is justified. Although the acknowledged for stimulating discussions.
6 Advances in Mechanical Engineering