DIRECTION SENSITIVITY OF ANEMOMETERS
Harald Mellinghoff, Olaf Haack, Renhard Kluin
DEWI GmbH, Germany
Summary
This paper looks at the impacts that different types of sensors experience when they are subjected to
tilted inflow and effects following a rotation. We see that not only sonic anemometers with the transducer
pairs in wake position show significant directional sensitivities. Moreover, classic cup anemometers
which feature an asymmetric shape around the vertical axis have been tested. Depending on the
closeness of the asymmetry in structure to the rotor, high impacts on the measurement behavior have
been observed. As a consequence, we suggest to assess these features for anemometers in use in the
wind energy and to account for the effects in the mounting uncertainty contribution of the measurements.
1. Introduction
1.1 MEASNET guideline development
With the revision No. 2 from October 2009 of the
MEASNET guideline for anemometer calibrations,
the procedure has been opened to other sensor
types as well. The basic intention was to open up for
the use of 2D and 3D sonic anemometers.
The guideline notes that the operational
characteristics shall be assessed for directional
sensitivity and other effects that could influence the
measurement behavior like temperature.
The MEASNET procedure is currently written for the
assessment of a single directional orientation
measuring the horizontal wind speed component
only. Yet, if we start studying the directional
sensitivity, not only a rotation around the vertical
axis is of interest, the tilt angle can be studied as
well. Both measurements have been performed at
DEWI in the past. A series of tests with different cup
and sonic anemometers has been performed in the
past.
1.2 Increased scope of sensors
The classic cup anemometer is a complex
instrument when it comes to describing the details of
the flow in the rotor volume. So far the empirical
measurmements with try and error optimization of
the design are the standard apporach to know the
measurement behaviour of a cup anemometer. The
sensors are pretty simple when it comes to the
principle of operation and the signal forming
process. The calibation function gives a linear
relation between the sensor output signal, usually a
frequency and the wind speed, averaged at least
over a full revolution of the cups.
Other types of sensors are under discussion for the
revision of the standard [2] for the measurement of
power curves. Most likely we will see sonic
anemometers in the future, but basically other
measurement principles can be considered. They all
have in common that the internal signal processing
of the sensor is major contribution to the signal
delivered. We will be talking to configurable devices
which can be operated in different modes. This
raises questions on the traceability of the data and
proposal to come up with a classification scheme
similar to the one introduced for cup anemometers
will have to take that into account.
A further common feature of such sensors is, that
they are not symmetric around the vertical axis.
Thus it can be expected that the relative position to
the inflow has some impact on the reading. Usually
these sensors have internal correction functions and
the following experiments give an indication to which
extend they can compensate these effects.
2. Measurement program
In the framework of a house internal research
program two sets of measurements have been
conducted in the wind tunnel. The open wind tunnel
of University of Oldenburg with an area of 8000 cm²
and a turbulence intensity of about 0.2 % has been
used.
2.1 Azimuth rotation tests
These tests rotated the anemometer around the
vertical axis. The stepping width is 5 degrees. The
starting point is either defined with respect to the
North marker of the sensor or with reference to the
sensor cable if it does not inside the supporting tube.
As a standard setting, DEWI places the sensor
cables downwind the tube. The tests have been
conducted at three wind speeds in the typical wind
speed range of wind energy use at 5, 10 and 15 m/s.
The ratio of the wind speed at the assessed azimuth
angle and the reference position has been
calculated and plotted over direction.
2.2 Tilt angle tests
Anemometers are subjected to the flow in the wind
tunnel at three different wind speeds. The
anemometer is mounted tilted in such a way that the
center of the measurement volume of the
anemometer is kept at the fixed reference position.
The conditions at this position are very well
assessed and the relation to the location of the pitot
tubes is known quite well. The Accuwind [3] project
has shown that the approach to the measurement
position has significant impacts. We will see that
keeping the position fixed, leads to good
reproducibility of results.
In these tests the angle of inflow has been modified
in a range from -30 to +30 degrees in steps of 2
2.3 Assessed sensors
The following sensors have been subjected to the
measurement program.
Cup anemometers
 Thies First Class Advanced, 4.3351.10.000
 Windspeed A100L2
 Windspeed A100LM-HE1
 NRG Max. 40
Sonic anemometers
 3D Gill Windmaster 1590-PK-20, Firmware
2329-112, digital output signal
 2D Vaisala WMT701, standard settings,
digital output signal
3. Measurement results for the azimuth rotation
tests
The first anemometer presented is the Thies First
Class Advanced. It is symmetric around the vertical
axis and the cable runs inside the mounting tube.
We therefore see the impact that experiment
configuration leaves us a fluctuation bandwidth. Part
of the variability that we see is the fact that cup
anemometers do not show a constant rotation, even
if they are subjected to very low turbulent
homogeneous flow. Since the reference point
selection has some element of randomness we look
at the span between the majority of the maximum
values and majority of the minima measured.
Expressed as an amplitude this is a measure for the
magnitude of the impact on a standard wind
measurement. For this case we get an amplitude of
1 %. All following diagrams use the same scaling in
the x and y axis to allow direct comparison.
Azimuth response anemometer TFC Advanced (Serial No. 1208016)
0.02
0.015
Azimuth response anemometer Windspeed A100L2 (Serial No. 5747 / cup 5JN)
0.02
0.015
0.01
0.005
0
-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
360
380
360
380
azimuth angle [°]
dev. V anemo bin average 5.1 m/s
dev. V anemo bin average 10.3 m/s
dev. V anemo bin average 15.1 m/s
Figure 2 Windspeed A100L2, Azimuth
experiment over 360 degrees at 5.1, 10.3 and
15.1 m/s.
Azimuth response anemometer Windspeed A100LM-HE1 (Serial No. 5460 /cup TJV)
0.02
0.015
0.01
0.005
0
-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
azimuth angle [°]
dev. V anemo bin average 5.2 m/s
dev. V anemo bin average 10.3 m/s
dev. V anemo bin average 15.1 m/s
Figure 3 Windspeed A100LM-HE1, Azimuth
experiment over 360 degrees at 5.2, 10.3 and
15.1 m/s.
The next cup anemometer is the NRG Max 40 with
the standard rubber boot and the cable again placed
downwind of the anemometer. We get the following
azimuth scan result:
Azimuth response anemometer NRG Max 40 (Serial No. OTC8671)
0
-0.005
0.02
-0.01
0.015
0.01
-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
360
380
azimuth angle /°
dev. V anemo bin average 5.0 m/s
dev. V anemo bin average 9.8 m/s
dev. V anemo bin average 14.3 m/s
Figure 1 TFC ad. Azimuth experiment over 360
degrees at 5.0, 9.8 and 14.3 m/s.
rel. deviation to 0° azmuth angle
rel. deviation to 0° azmuth angle
0.01
0.005
This cup anemometer is also available with an
external heater applied to the anemometer shaft. In
this case we see a very pronounced directional
sensitivity with amplitude up to 3.5 % in wind speed
[fig. 3].
rel. deviation to 0° azmuth angle
Since both types of measurement took almost an
hour a correction has been applied that accounts for
the remaining fluctuation in the change in incoming
wind speed. This change has been at maximum
0.2 %.
looking at the minima near 70 and 300 degrees we
observe something that more related to 120 and 240
degrees. We assume that the two screws fixing the
mounting adapter at an angle of 120 degrees is this
element. The amplitude is on the order of 1.5 % in
this case.
rel. deviation to 0° azmuth angle
degrees. At each position data has been collected
for 30 seconds. The vertical position has been used
as the scaling factor. For each tilt angle tested, the
relation has been between the tilted and the vertical
position has been calculated.
0.005
0
-0.005
-0.01
-0.015
-0.02
-0.025
-0.03
-0.035
-0.04
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
360
azimuth angle [°]
The next sensor is the Windspeed A100L2 (fig. 2)
cup anemometer fixed with a mounting adapter to
the 25 mm tube, cable located downwind. Because
of the cable the expectation is that one would see an
effect associated with 180 degrees turns. However,
dev. V anemo bin average 5.2 m/s
dev. V anemo bin average 10.4 m/s
dev. V anemo bin average 15.1 m/s
Figure 4 NRG max 40, Azimuth experiment over
360 degrees at 5.2, 10.4 and 15.1 m/s.
380
In contrast we look at the sonic anemometers results
[fig. 5]. The Gill 3D data base has only one
measurement at 10.3 m/s available. We observe a
clear pattern with a 120 deg maximum impact with
pronounced maximum values which are related to
the transducers and the supporting structures in the
flow. The general amplitude is 1.5 % and up to 3 %
in the peak regions which have a width of
approximately 20 degrees each.
Below -14 degrees and above +18 the difference to
the cosine response is of magnitude 1 % and more.
In contrast the Windspeed A100L2 is giving readings
smaller than the cosine response of 2 % and more
compared to the vertical orientation.
tilt response anemometer TFC Advanced (SN 1208016)
0.1
nozzle
nozzle
0.08
0.06
0.04
rel. deviation of anemoemter frequency
The observed amplitude is on the order of 2 % in
speed.
Azimuth response anemometer Gill Windmaster 1590-PK-020 (Serial No. 095015) at 10.4 m/s dataset 0054_10
-20
+20
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
-0.14
0.040
-0.16
0.035
-0.18
0.030
-0.2
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
0.025
-6
-4
-2
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34
rel deviation to 0° azimuth angle
tilt angle /°
0.020
dev. V anemo bin average 5.2 m/s
0.015
dev. V anemo bin average 10.3 m/s
dev. V anemo bin average 15.1 m/s
cosine response
Figure 7 Tilt angle test of the Thies First Class
Advanced.
0.010
0.005
0.000
-0.005
-0.010
-0.015
tilt response anemometer Windspeed A100L2 (body: 5747, cup: 5JN)
-0.020
0.1
-0.025
nozzle
nozzle
0.08
-0.030
deviation bin average
-0.035
deviation at 1 Hz
0.06
rel. deviation of anemoemter frequency
0.04
-0.040
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
360
380
azimuth angle [°]
Figure 5 Azimuth test for the Gill 3D sonic
anemometer. The dots mark the individual 1
second measurements and the blue line is the
bin average.
-20
+20
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
-0.14
-0.16
-0.18
The 2D Vaisala sensor is the last one assessed in
this series. Looking at the plot, it is hard to identify
an individual pattern that repeats every 120 degrees.
Instead we see amplitude on the order of 1.5 % to
2 %.
Azimuth response anemometer Vaisala WMT701 (Serial No. F3420001)
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
-6
-4
-2
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34
tilt angle /°
dev. V anemo bin average 5.1 m/s
dev. V anemo bin average 10.3 m/s
dev. V anemo bin average 15.1 m/s
cosine response
Figure 8 Tilt angle test for the Windspeed
A100L2.
It is interesting to see, that the A100 version of the
same anemometer with the heater attached has a
more moderate tilt response [fig. 9]. It should be
noted that presented picture indicates that the
sensitivity close to the vertical position is high.
Especially for slight downwind inflow conditions the
anemometer might actually be running faster than a
standard sensor of same type,
0.030
0.025
0.020
rel. deviation to 0° azmuth angle
-0.2
0.015
0.010
0.005
0.000
-0.005
-0.010
-0.015
-0.020
-0.025
tilt response anemometer Windspeed A100LM-HE1 (body: 5460; cup: TJV)
-0.030
-20
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
320
340
360
0.1
380
azimuth angle /°
dev. V anemo bin average 9.9 m/s
nozzle
0.08
0.06
dev. V anemo bin average 14.4 m/s
Figure 6 Azimuth test for the Vaisala 2D sonic
anemometer.
4. Measurement results for the tilt angle tests
The results for the tilt angle test are presented in a
similar manner.
The typical situation for a cup anemometer is the
tilted inflow from below, i.e. the wind has an upward
component. In this presentation we use the
convention that positive inflow angles are associated
with upward flow.
We start with the Thies First Class Advanced [fig. 7].
It has a very good agreement with the cosine
response which is defined as the ideal behavior.
0.04
rel. deviation of anemoemter frequency
dev. V anemo bin average 5.0 m/s
nozzle
-20
+20
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
-0.14
-0.16
-0.18
-0.2
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
-6
-4
-2
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34
tilt angle /°
dev. V anemo bin average 5.2 m/s
dev. V anemo bin average 10.3 m/s
dev. V anemo bin average 15.1 m/s
cosine response
Figure 9 Tilt angle test for the Windspeed
A100LM-HE1.
For the NRG Max 40 we get the following result
[Fig. 10]. In the inflow range 0 to 7 degrees the
anemometer reports wind speed higher than the
cosine response in a magnitude up to 1.7 %. Near
+8 degrees the anemometer readings are lower than
the cosine response. At the outer tilt inflow ranges
behind + 14 and - 14 degrees the anemometer
changes its response dramatically and can even
report winds that are clearly higher than the
incoming wind even if the vector wind speed
definition is used.
tilt response anemometer NRG Max 40 (OTC8671)
0.1
0.08
0.06
rel. deviation of anemoemter frequency
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
nozzle
-0.12
nozzle
-0.14
-0.16
-20
+20
-0.18
-0.2
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
-6
-4
-2
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34
tilt angle /°
dev. V anemo bin average 5.2 m/s
dev. V anemo bin average 10.4 m/s
dev. V anemo bin average 15.1 m/s
cosine response
Figure 10 Tilt angle test for the NRG max 40.
The strength of the sonics is that they do not have
moved parts which complicate the flow through the
measurement volume. We see this in the response
of the 3D Gill Windmaster in figure 11.
tilt response anemometer Gill Windmaster 1590-PK-20 (095015)
0.1
nozzle
nozzle
0.08
0.06
0.04
rel. deviation of anemoemter reading
The 2D sonic Vaisala (settings WMT701
A1A0A1A2B1A1) shows a similar pattern for the
downward flows associated with negative tilt angles
here. It actually has the closest agreement with the
sonic response of all measurements taken in this
series. On the other hand the readings for the
upwind inflow situation give a different picture.
Between 0 and 5 degrees upwind inflow we get a
reading of up to 2 % below the cosine response. The
reading approaches the cosine for higher tilt angles
but it stays below the cosine response on the order
of magnitude of 1 % to 1.5 %.
-20
+20
0.02
0
-0.02
-0.04
-0.06
-0.08
-0.1
-0.12
-0.14
-0.16
-0.18
-0.2
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
-6
-4
-2
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34
tilt angle /°
dev. V anemo bin average 5.2 m /s
dev. V anemo bin average10.4 m/s
dev. V anemo bin average 15.1 m/s
cosine response
Figure 11Tilt angle test for the 3D Gill Windmaster evaluating the horizontal component of
the signal. Internal signal correction has been
turned on.
This sonic is very close to the cosine response in a
range -6 to +6 degrees. For larger tilt angles in upor downwind direction the readings are reported to
be about 1 % above the cosine response, for large
positive angles it is almost 2%. Yet, the behavior is
not showing jumps only minor impact with wind
speed
5: Summary and Outlook
A number of four cup anemometers and two sonic
anemometers has been subjected to wind tunnel
measurements to study the impact of sensor rotation
and tilted inflow. Both measurement types will play a
significant role in the understanding and
classification of anemometers of any type.
5.1 Conclusions
 Asymmetry around the vertical axis has
measureable impact on the reading of
anemometers.
 If a standard mounting position is not used
during calibration, it will increase the
scatter/uncertainty of the serial calibrations.
 A direction correction or increased the mounting
uncertainty should be used for sensors with
directional dependencies.
 For sensors with internal data processing unit,
the settings of the device shall be documented. The
status signal of the unit shall be monitored.
 Sonics have a tendency for reduced scatter at
constant wind speeds when tested in the tunnel.
 Sonics can be very good on the tilt angle
response.
 Sonics shall be assessed for azimuth
dependency especially for conditions with one
transducer being in the wake of another.
5.2 Outlook
The tests have been performed keeping either tilt or
the azimuth rotation constant. The combination of
both should be studied. The experiments shall be
repeated with other wind tunnels to ensure that a
robust methods can be defined in the
standardization work that allow reproduction of the
results.
Tilt response anemometer Vaisala WMT701 (Serial No. F3420001)
0.1
0.08
nozzle
nozzle
6. References
0.06
rel. deviation to 0° tilt angle
0.04
-20
+20
0.02
[1] MEASNET, Anemometer calibration procedure,
Version 2, October 2009
0
-0.02
-0.04
-0.06
-0.08
[2] IEC 61400-12-1, Power performance
measurements of electricity producing wind turbines,
1st ed. 2005-12
-0.1
-0.12
-0.14
-0.16
-0.18
-0.2
-34 -32 -30 -28 -26 -24 -22 -20 -18 -16 -14 -12 -10 -8
-6
-4
-2
0
2
4
6
8
10 12 14 16 18 20 22 24 26 28 30 32 34
tilt angle /°
dev. V anemo bin average 5 m/s
dev. V anemo bin average 9.9 m/s
dev. V anemo bin average 14.4 m/s
cosine response
Figure 12 Tilt angle test Vaisala WMT701 using
the default settings.
[3] Accuwind Summary Report, T. F. Petersen et al.,
Risø R-1563(EN), July 2006