Next Article in Journal
Study on the Development of an Optimal Heat Supply Control Algorithm for Group Energy Apartment Buildings According to the Variation of Outdoor Air Temperature
Previous Article in Journal
A Smart Green Building: An Environmental Health Control Design
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Cup Anemometers’ Loss of Performance Due to Ageing Processes, and Its Effect on Annual Energy Production (AEP) Estimates

1
IDR/UPM, ETSIA, Universidad Politécnica de Madrid (Polytechnic University of Madrid), Pza. del Cardenal Cisneros 3, Madrid 28040, Spain
2
Aerospace Propulsion and Fluid Mechanics Department, ETSIA, Universidad Politécnica de Madrid (Polytechnic University of Madrid), Pza. del Cardenal Cisneros 3, Madrid 28040, Spain
3
Department of Aerospace Materials and Production, ETSIA, Universidad Politécnica de Madrid (Polytechnic University of Madrid), Pza. del Cardenal Cisneros 3, Madrid 28040, Spain
*
Author to whom correspondence should be addressed.
Energies 2012, 5(5), 1664-1685; https://doi.org/10.3390/en5051664
Submission received: 22 March 2012 / Revised: 30 April 2012 / Accepted: 16 May 2012 / Published: 23 May 2012

Abstract

:
The deviation of calibration coefficients from five cup anemometer models over time was analyzed. The analysis was based on a series of laboratory calibrations between January 2001 and August 2010. The analysis was performed on two different groups of anemometers: (1) anemometers not used for any industrial purpose (that is, just stored); and (2) anemometers used in different industrial applications (mainly in the field—or outside—applications like wind farms). Results indicate a loss of performance of the studied anemometers over time. In the case of the unused anemometers the degradation shows a clear pattern. In the case of the anemometers used in the field, the data analyzed also suggest a loss of performance, yet the degradation does not show a clear trend. A recalibration schedule is proposed based on the observed performances variations.

1. Introduction

Today, there is a massive use of wind speed anemometers by industry. The growth experienced by the wind energy sector has increased and spread their use, but leaving aside the applications related to wind energy it should be said that they are being increasingly used in other fields potentially affected by wind (bridges, big cranes). The accuracy of the measurements is extremely important for the wind energy industry as the extractable wind power is proportional to the third power of the wind speed [1,2]. In order to ensure the accuracy of an anemometer it should be recalibrated after being in use some time. The calibration method consists of placing it in an incoming flow with a known speed, uniformity and turbulence level, and measuring its output signal at various given wind speeds. It is thereby possible to obtain a relationship between the velocity of the wind flow and the anemometer’s output signal (the control parameter of the output signal normally being its frequency). It is widely accepted that the relationship between the measured wind speed and the anemometer’s output frequency is linear [3], that is:
V = × f + B
where V is the velocity of the flow (wind speed), f is the anemometer’s rotation frequency output, and A (slope) and B (offset) are the calibration coefficients corresponding to the tested anemometer. However, it must also be said that this behavior is not exactly linear, although as mentioned, for most purposes the linear approximation is sufficiently accurate [3,4].
The aim of the present paper is to analyze the variations of the calibration constants A and B over time. This particular study has been proposed to the IDR/UPM Institute many times by enterprises in the wind energy sector, interested in anemometers’ potential loss of performance due to wear and tear. At present, the ageing of the anemometers is addressed through frequent recalibration [5]. Some effort has also been done with regard to the anemometers’ recalibration in the field [6,7]. However, as far as the authors know there seems to be a lack of results and research in the available literature concerning anemometers’ loss of performance.
It is reasonable to assume that once an anemometer is in service, the loss of performance, if it happens, should modify both calibration constants, A and B. On the one hand, this degradation could affect the anemometers’ rotational speed, that is, the anemometers’ capacity to transform energy from wind into rotation of the shaft should be reduced if energy losses increase (friction, for example), or the rotor’s moment of inertia or its aerodynamics are changed by the mass addition of dirt. The reduction in the rotational speed can be translated into an increase of the constant A value. On the other hand, the degradation could also affect the starting speed of the anemometer, that is, as it is longer in service the wind speed necessary to start its rotation could be higher if the friction has increased, and that effect can be translated into an increase of the constant B. Together with the aforementioned considerations, it should also be mentioned that, as common in complex mechanisms, the anemometer’s rotor could have a transitional period of time at the beginning of its service life before reaching its stable working condition.
Two different anemometers’ degradation cases have been studied. The first one is the degradation of anemometers not used in field and just stored. This case was studied with the data from many calibrations performed on three different single individual anemometers. These calibrations are periodically carried out as part of the internal quality procedures at the IDR/UPM Institute. Each one of these three anemometers is stored in its own box between calibrations, the average seasonal variation of the climatic conditions in the calibration lab being around 20–30 °C, 928–955 hPa, and 18–44% relative humidity. No maintenance program was performed on these anemometers. The second case is related to the degradation of anemometers used in the field. The data from calibrations performed on the same anemometers, sent several times to the IDR/UPM Institute, were collected and analyzed in order to study the degradation of five different models of anemometers. In order to obtain statistically more significant results, the calibrations period covered in a previous study [4] (January 2003 to August 2007) has been extended to January 2001 and January 2011.
Five enterprises of the wind energy sector (Barlovento, Cener, Dekra Ambio, Ecosem, and Ges-Siemsa) collaborated with the IDR/UPM Institute in order to complete the information and strengthen the study with regard to the anemometers’ behavior once in service. Thanks to the information provided by the aforementioned enterprises, the maintenance work on several individual anemometers was traced. Some of these anemometers were subjected to high level maintenance, normally consisting of changing the bearings (sometimes together with the change of the anemometer’s electronics and the cups’ rotor, if damaged).

2. Testing Configuration and Anemometers Studied

The anemometer calibrations analyzed in the present paper were performed by strictly following the MEASNET recommendations [8,9], that is, over 13 points, from 4 to 16 m·s−1. More details concerning the facility and the calibration process are included in reference [4].
Six different cup anemometers were studied: Risø P2546 (WindSensor, Risø DTU: Roskilde, Denmark); Thies Clima 4.3350 and 4.3303 (Thies Clima: Göttingen, Germany); Climatronics 100075 (Climatronics Corp: Bohemia, NY, USA); Vector Instruments A100 L2 and LK (Windspeed Limited, trading as Vector Instruments: Rhyl, UK).
As stated in the introduction, two different analyses were carried out, the first one was based on calibrations performed on three anemometers used for internal procedures at the IDR/UPM Institute (Climatronics 100075, Vector Instruments A100 L2, and Thies 4.3350), the last two of them being first class anemometers in accordance with IEC-61400-12-1 [10]. These anemometers were calibrated periodically in order to test the quality of the calibration process (more information of the requirements fulfilled by the IDR/UPM Institute in order to ensure a high level of accuracy can be found in reference [4]). The second analysis was based on the data available concerning anemometers sent several times to the IDR/UPM Institute for periodic calibrations. Those anemometers have mostly been used in the field (e.g., installed on wind turbines), and each one under different climatic conditions.

3. Results and Discussion

3.1. Variations in Calibration Constants from Stored Anemometers (not Used in the Field)

In Figure 1 the values of the calibration coefficients corresponding to the IDR/UPM Institute anemometers (Climatronics 100075, Vector Instruments A100 L2, and Thies Clima 4.3350 –from now on these anemometers will be referred to as Cl-100075, A100 L2, and Th-4.3350 in the text), are shown as a function of the number of days after their first calibration. A linear fitting to the data has been included in all graphs. The average values for every 300 days, together with the standard deviation bars, have also been included in the graphs (see Table 1). As said, these three anemometers were used just for IDR/UPM Institute internal procedures and apart from their periodic calibrations, they were not used at all. Despite the scattering of the data shown in Figure 1, some trends can be observed by analyzing the 300-day average results. In some cases the variation of the coefficients seem to clearly fit a linear behavior (Cl-100075 and Th-4.3350 A coefficients, and A100 L2 B coefficient), whereas in others (Cl-100075 and Th-4.3350 B coefficients, and A100 L2 A coefficient) the correlation with the linear fit is worse (see in Figure 1 that the coefficients of determination, R2, are in these cases significantly lower, from 7.47 × 10−3 to 1.56 × 10−2, than the ones previously mentioned, from 3.53 × 10−1 to 6.93 × 10−1).
In Figure 2 the output frequency at 7 m·s−1 wind speed of the studied anemometers is plotted as a function of the number of days after their first calibration. There seems to be a transitional period after the starting of life, where the anemometers were more and more efficient in terms of translating the wind speed into rotation (although the practical differences are negligible, around 0.1–0.2 Hz). After this transitional period, the anemometers tend to be less efficient, their output frequency at constant wind speed being decreased with the use. The transitional period observed is around 450 days after the first calibration, for the studied anemometers. The number of calibrations performed on each anemometer within this period was respectively 9 (Cl-100075), 18 (A100 L2) and 30 (Th-4.3350).
If a linear behavior over time after the first calibration, Δt, is considered, both calibration coefficients, A and B, can be expressed as:
A = A 0 + d A d t Δ t ± σ A
B = B 0 + d B d t Δ t ± σ B
where (A0, dA/dt) and (B0, dB/dt) are respectively the linear fit of both coefficients, and the terms σA and σB are a measure of the scattering of the data (as known, the interval ±σ indicates a 68.2% confidence error limits in a Gaussian process).
Bearing in mind what was mentioned in the Introduction with regard to the anemometers’ performance degradation (that is, coefficients A and B—one of them or both—tend to increase if degradation is produced), and the data from Figure 1, the behavior of the three considered anemometers has been estimated as follows:
Figure 1. Calibration coefficients’ variation with regard to IDR/UPM Institute anemometers: Climatronics 100075 (top), Vector Instruments A100 L2 (middle), and Thies Clima 4.3350 (bottom) cup anemometers. These coefficients were measured in different calibrations from January 2001 to June 2006 (Climatronics 100075), from September 2003 to September 2007 (A100 L2), and from November 2006 to August 2010 (Thies Clima 4.3350). The 300-day average value has been included, together with the standard deviation bars. The linear fitting to the data has been also included.
Figure 1. Calibration coefficients’ variation with regard to IDR/UPM Institute anemometers: Climatronics 100075 (top), Vector Instruments A100 L2 (middle), and Thies Clima 4.3350 (bottom) cup anemometers. These coefficients were measured in different calibrations from January 2001 to June 2006 (Climatronics 100075), from September 2003 to September 2007 (A100 L2), and from November 2006 to August 2010 (Thies Clima 4.3350). The 300-day average value has been included, together with the standard deviation bars. The linear fitting to the data has been also included.
Energies 05 01664 g001
Table 1. Mean and standard deviation values of calibration coefficients A and B every 300 days, corresponding to the IDR/UPM Institute Climatronics 100075, Vector Instruments A100 L2 and Thies Clima 4.3350 anemometers.
Table 1. Mean and standard deviation values of calibration coefficients A and B every 300 days, corresponding to the IDR/UPM Institute Climatronics 100075, Vector Instruments A100 L2 and Thies Clima 4.3350 anemometers.
Climatronics 100075
Period consideredNumber of calibrationsA meanσAB meanσB
First 300 days74.6917 × 10−27.5214 × 10−52.4124 × 10−18.9279 × 10−3
Between 300 and 600 days44.6925 × 10−22.6458 × 10−52.2824 × 10−17.9439 × 10−3
Between 600 and 900 days104.7015 × 10−27.7683 × 10−52.3398 × 10−11.0886 × 10−2
Between 900 and 1200 days114.7103 × 10−27.8608 × 10−52.2276 × 10−11.1274 × 10−2
Between 1200 and 1500 days134.7175 × 10−28.5312 × 10−52.3057 × 10−11.1498 × 10−2
Between 1500 and 1800 days144.7325 × 10−21.1543 × 10−42.3905 × 10−11.9730 × 10−2
Between 1800 and 2100 days54.7254 × 10−28.4135 × 10−52.4437 × 10−11.8364 × 10−2
Vector Instruments A100 L2
Period consideredNumber of calibrationsA meanσAB meanσB
First 300 days125.0151 × 10−21.7868 × 10−41.5813 × 10−11.6276 × 10−2
Between 300 and 600 days125.0005 × 10−21.1291 × 10−41.7469 × 10−11.2122 × 10−2
Between 600 and 900 days155.0022 × 10−28.4847 × 10−51.9887 × 10−11.3416 × 10−2
Between 900 and 1200 days84.9991 × 10−21.0809 × 10−41.9644 × 10−11.9603 × 10−2
Between 1200 and 1500 days215.0060 × 10−21.2157 × 10−42.0680 × 10−13.1423 × 10−2
Thies Clima 4.3350
Period consideredNumber of calibrationsA meanσAB meanσB
First 300 days104.8200 × 10−27.8535 × 10−52.5722 × 10−16.4219 × 10−3
Between 300 and 600 days364.8198 × 10−28.8757 × 10−52.5058 × 10−11.6461 × 10−2
Between 600 and 900 days334.8237 × 10−21.3275 × 10−42.4841 × 10−12.6559 × 10−2
Between 900 and 1200 days454.8329 × 10−29.0911 × 10−52.4362 × 10−12.1235 × 10−2
Between 1200 and 1500 days334.8398 × 10−28.4043 × 10−52.4344 × 10−11.5727 × 10−2
Between 1500 and 1800 days104.8361 × 10−21.2206 × 10−42.6393 × 10−11.9459 × 10−2
  • Cl-100075. This anemometer seems to degrade decreasing the rotation speed, but no clear effect can be observed on the offset speed, so only degradation due to the loss of rotation speed was considered (A0 = 4.684 × 10−2, and dA/dt = 2.547 × 10−7: both the linear fit from Figure 1; B0 = 0.2505: value from the initial calibration in 2001, and dB/dt = 0; σA = 7.7548 × 10−5, and σB = 1.26607 × 10−2: average values of the scattering from Table 1).
  • A100 L2. Only degradation due to the increase of the offset speed was considered for this anemometer (A0 = 5.044 × 10−2: value from the initial calibration in 2003, and dA/dt = 0; B0 = 1.5857 × 101, and dB/dt = 3.7815 × 10−5: both the linear fit from Figure 1; σA = 1.21218 × 10−4, and σB = 1.85679 × 10−2: average values of the scattering from Table 1).
  • Th-4.3350. As in the case of the Cl-100075 anemometer, only degradation due to the loss of rotation speed was considered (A0 = 4.8120 × 10−2, and dA/dt = 1.880 × 10−7: both the linear fit from Figure 1; B0 = 0.26358: value from the initial calibration in 2006, and dB/dt = 0; σA = 9.9509 × 10−5, and σB = 1.7644 × 10−2: average values of the scattering from Table 1).
Figure 2. Variation of the output frequency at 7 m·s−1 wind speed, f 7m/s, as a function of the number of days after the first calibration, for the IDR/UPM Institute’s anemometers, Cl-100075 (top left side), A100 L2 (top right side), and Th-4.3350 (bottom). The 300-day average value has been included, together with the standard deviation bars.
Figure 2. Variation of the output frequency at 7 m·s−1 wind speed, f 7m/s, as a function of the number of days after the first calibration, for the IDR/UPM Institute’s anemometers, Cl-100075 (top left side), A100 L2 (top right side), and Th-4.3350 (bottom). The 300-day average value has been included, together with the standard deviation bars.
Energies 05 01664 g002
Taking into account the Equations (1) to (3), the variation in the measured wind speed as a function of the time (days) after the first calibration can be estimated as:
Δ V = ( 1 A 0 d A d t V + ( d B d t B 0 A 0 d A d t ) ) Δ t ± | V B 0 A 0 σ A + σ B | ,
where the second term is the 68.2% confidence error limits (obviously, this confidence level can be extended by increasing the respective confidence levels of A and B, that is, σA and σB).
Once the variation of the measured wind speed regarding an individual anemometer has been defined as a function of the time from its first calibration, it is possible to establish some criteria in order to decide when it should be necessary to recalibrate it. The first criterion could be based solely on the deviation of the measured speed. In this case, the recalibration should be programmed when, at a certain wind speed, V, the difference between that velocity and the measured wind speed, ΔV, has reached a certain level. Let us suppose that the limit is established at X% of the reference wind speed, V. Then, the recalibration of the anemometer should be scheduled at:
Δ t = X 100 V + | V B 0 A 0 σ A + σ B | 1 A 0 d A d t V + ( d B d t B 0 A 0 d A d t ) ,
where Δt is the number of days after the initial calibration. At that time, the measured wind speed has X% deviation with respect to the wind speed, with 84.1% confidence (supposing a Gaussian process). Obviously, if it is decided to increase the confidence level to 97.7% the values of σA and σB would have to be increased by a factor of 2. On the other hand, if a 50% confidence level is considered the term related to σA and σB must be ignored in Equation (5).
In Table 2, Table 3 and Table 4, the proposed recalibration schedules of the Cl-100075, A100 L2, and Th-4.3350 IDR/UPM anemometers, for reference wind speeds V = 4, 10, 16 and 22 m·s−1, have been respectively included as a function of the accepted difference between the measured and the reference wind speeds, and the confidence level. See also in the Figure 3, the recalibration diagrams corresponding to the reference wind speed V = 10 m·s−1 as a function of the confidence level, for 1, 0.5, 0.3 and 0.1% error with respect to the reference wind speed.
Table 2. Recalibration schedule (days after the initial calibration) of the IDR/UPM Institute Cl-100075 anemometer for reference wind speeds 4, 10, 16 and 22 m·s−1 as a function of the accepted deviation from the reference wind speeds, and the confidence level.
Table 2. Recalibration schedule (days after the initial calibration) of the IDR/UPM Institute Cl-100075 anemometer for reference wind speeds 4, 10, 16 and 22 m·s−1 as a function of the accepted deviation from the reference wind speeds, and the confidence level.
1% deviation from reference wind speed 0.5% deviation from reference wind speed
Confidence levelReference wind speedConfidence levelReference wind speed
4101622 4101622
50.0%1962188618681860 50.0%981943934930
84.1%2887243023212272 84.1%1906148613861342
97.7%3813297327732683 97.7%2832203018391753
99.9%4738351632253095 99.9%3757257322912165
0.3% deviation from reference wind speed0.1% deviation from reference wind speed
Confidence levelReference wind speedConfidence levelReference wind speed
4101622 4101622
50.0%589566560558 50.0%196189187186
84.1%151411091013970 84.1%1122732639598
97.7%2439165214651381 97.7%2047127510911009
99.9%3365219619171793 99.9%2972181815441421
The second criterion suggested to recalibrate the anemometers can be based on the variations of the Annual Energy Production (AEP) estimations due to the error in the wind speed measurements caused by the anemometer’s loss of performance. In Figure 4 the variations of the AEP due to the Cl-100075, A100 L2 and Th-4.3350 anemometers’ degradation (with 84.1% confidence level) are shown for 4, 7 and 10 m·s−1 hub height annual average wind speed (see also Table 5). These AEP calculations have been made by following the procedure recommended by the International Electrotechnical Commission (IEC) [10], using the General Electric GE2.5 wind turbines power curve as a reference (see [4]). The recalibration of the anemometer should then be ordered when the underestimation of the AEP reaches a certain critical level.
Table 3. Recalibration schedule (days after the initial calibration) of the IDR/UPM Institute A100 L2 anemometer for reference wind speeds 4, 10, 16 and 22 m·s−1 as a function of the accepted deviation from the reference wind speeds, and the confidence level.
Table 3. Recalibration schedule (days after the initial calibration) of the IDR/UPM Institute A100 L2 anemometer for reference wind speeds 4, 10, 16 and 22 m·s−1 as a function of the accepted deviation from the reference wind speeds, and the confidence level.
1% deviation from reference wind speed 0.5% deviation from reference wind speed
Confidence levelReference wind speedConfidence levelReference wind speed
4101622 4101622
50.0%1058264442315818 50.0%529132221162909
84.1%1793376157297697 84.1%1264243936134788
97.7%2528487772279576 97.7%1999355551116667
99.9%32635994872411455 99.9%2734467266098546
0.3% deviation from reference wind speed0.1% deviation from reference wind speed
Confidence levelReference wind speedConfidence levelReference wind speed
4101622 4101622
50.0%106264423582 50.0%106264423582
84.1%841138119212461 84.1%841138119212461
97.7%1576249734194340 97.7%1576249734194340
99.9%2311361449166219 99.9%2311361449166219
Table 4. Recalibration schedule (days after the initial calibration) of the IDR/UPM Institute Th-4.3350 anemometer for reference wind speeds 4, 10, 16 and 22 m·s−1 as a function of the accepted deviation from the reference wind speeds, and the confidence level.
Table 4. Recalibration schedule (days after the initial calibration) of the IDR/UPM Institute Th-4.3350 anemometer for reference wind speeds 4, 10, 16 and 22 m·s−1 as a function of the accepted deviation from the reference wind speeds, and the confidence level.
1% deviation from reference wind speed 0.5% deviation from reference wind speed
Confidence levelReference wind speedConfidence levelReference wind speed
4101622 4101622
50.0%2740262926022591 50.0%1370131413011295
84.1%4478362234193328 84.1%3108230821182032
97.7%6216461542354065 97.7%4846330129342769
99.9%7954560850514802 99.9%6584429437503506
0.3% deviation from reference wind speed0.1% deviation from reference wind speed
Confidence levelReference wind speedConfidence levelReference wind speed
4101622 4101622
50.0%822789781777 50.0%274263260259
84.1%2560178215971514 84.1%201212561077996
97.7%4298277524132251 97.7%3750224918931733
99.9%6036376832302988 99.9%5488324227092470
There seems to be some discrepancy with regard to the recalibration based on the deviation of the measured speed at 10 m·s−1 reference wind speed (Figure 3), and the recalibration based on the underestimation of the AEP (Figure 4).
Figure 3. Recalibration diagram proposed for the IDR/UPM Institute’s Cl-100075 (top left side), A100 L2 (top right side), and Th-4.3350 (bottom) cup anemometers.
Figure 3. Recalibration diagram proposed for the IDR/UPM Institute’s Cl-100075 (top left side), A100 L2 (top right side), and Th-4.3350 (bottom) cup anemometers.
Energies 05 01664 g003
Based on the loss of performance at 10 m·s−1 wind speed, the Cl-100075 anemometer should be recalibrated before the A100 L2 anemometer, however, for 4 m·s−1 hub height annual average wind speed the underestimation of the AEP is higher in the case of the A100 L2 anemometer than in the case of the Cl-100075 anemometer, no matter the time elapsed since the first calibration. This can be explained as both anemometers degrade differently, according to the data in Figure 1. The Cl-100075 anemometer loses performance by decreasing the rotation speed for a given wind speed, whereas the A100 L2 anemometer does it by increasing the offset speed. This represents a different degradation of the anemometers’ performances at different reference wind speeds. In Figure 5 the underestimation of the measured wind speed, ΔV, is plotted as a function of the reference wind speed for the three anemometers considered 300, 900 and 1800 days after the first calibration. As can be observed in the aforementioned figure, for low wind speeds the loss of performance of the A100 L2 anemometer is greater than the one from the Cl-100075 anemometer, whereas the situation is different for higher wind speeds.
Figure 4. Variation of the Annual Energy Production (AEP) underestimation caused by the error in the wind speed measurement due to the loss of performance of the IDR/UPM Institute’s Cl-100075 (top left side), A100 L2 (top right side), and Th-4.3350 (bottom) cup anemometers. AEP related to General Electric GE2.5. Underestimation calculated with 84.1% confidence level.
Figure 4. Variation of the Annual Energy Production (AEP) underestimation caused by the error in the wind speed measurement due to the loss of performance of the IDR/UPM Institute’s Cl-100075 (top left side), A100 L2 (top right side), and Th-4.3350 (bottom) cup anemometers. AEP related to General Electric GE2.5. Underestimation calculated with 84.1% confidence level.
Energies 05 01664 g004
Figure 5. Underestimation of the measured wind speed, ΔV, as a function of the reference wind speed for the three anemometers considered (left: Cl-100075; middle: A100 L2; right: Th-4.3350) 300, 900 and 1,800 days after the first calibration of the anemometer. Confidence level: 84.1%.
Figure 5. Underestimation of the measured wind speed, ΔV, as a function of the reference wind speed for the three anemometers considered (left: Cl-100075; middle: A100 L2; right: Th-4.3350) 300, 900 and 1,800 days after the first calibration of the anemometer. Confidence level: 84.1%.
Energies 05 01664 g005

3.2. Variations in Calibration Constants from Anemometers Used in Field

As mentioned previously, the behavior of five different anemometers (Risø P2546, Thies Clima 4.3303 and 4.3350, Climatronics 100075, and Vector Instruments A100 LK), has been studied using the data from individual anemometers calibrated at least two times at the IDR/UPM Institute. The number of anemometers analyzed with regard to the number of calibrations performed is as follows, Risø P2546: 29 anemometers calibrated 2-times, 18 anemometers calibrated 3-times, 11 anemometers calibrated 4-times, four anemometers calibrated 5-times, and one anemometer calibrated 6-times; Climatronics 100075: five anemometers calibrated 2-times, and three anemometers calibrated 3-times; Vector Instruments A100 LK: 90 anemometers calibrated 2-times, 45 anemometers calibrated 3-times, and seven anemometers calibrated 4-times; Thies Clima 4.3303: 33 anemometers calibrated 2-times, 11 anemometers calibrated 3-times, three anemometers calibrated 4-times, and one anemometer calibrated 5-times; Clima 4.3350: 227 anemometers calibrated 2-times, 54 anemometers calibrated 3-times, and four anemometers calibrated 4-times.
Table 5. Underestimation of the Annual Energy Production (AEP) caused by the deviation in the wind speed measurement due to the loss of performance of the IDR/UPM Climatronics100075, Vector Instruments A100 L2 and Thies Clima 4.3350 anemometers. AEP related to General Electric GE2.5. Underestimation calculated with 84.1% confidence level.
Table 5. Underestimation of the Annual Energy Production (AEP) caused by the deviation in the wind speed measurement due to the loss of performance of the IDR/UPM Climatronics100075, Vector Instruments A100 L2 and Thies Clima 4.3350 anemometers. AEP related to General Electric GE2.5. Underestimation calculated with 84.1% confidence level.
Climatronics100075
Days after first calibrationHub height annual average wind speed
4 m·s−17 m·s−110 m·s−1
3001.87%0.86%0.47%
6002.42%1.14%0.62%
9002.97%1.42%0.78%
12003.52%1.70%0.93%
15004.07%1.98%1.09%
18004.62%2.26%1.24%
21005.17%2.54%1.40%
24005.72%2.82%1.55%
27006.27%3.10%1.71%
30006.82%3.38%1.87%
Vector Instruments A100 L2 Thies Clima 4.3350
Days after first calibrationHub height annual average wind speed Hub height annual average wind speed
4 m·s−17 m·s−110 m·s−1 4 m·s−17 m·s−110 m·s−1
3002.61%1.12%0.60% 2.14%0.97%0.52%
6003.29%1.38%0.73% 2.54%1.17%0.63%
9003.96%1.65%0.87% 2.93%1.37%0.75%
12004.64%1.92%1.01% 3.33%1.57%0.86%
15005.32%2.18%1.15% 3.72%1.77%0.97%
18005.99%2.45%1.29% 4.12%1.97%1.08%
All anemometers studied were used in the field under different climatic conditions during different time periods. The elapsed time between calibrations is not exactly equal to the periods of service due to transportation, installation and disassembly times, but it seems reasonable to assume that this time is minimal when compared to the period of service. The percentage variation of the calibration constants A and B of the studied anemometers is shown in Figure 6 and Figure 7 as a function of the time elapsed from the first calibration.
Figure 6. Percentage variation of calibration constants A and B from the initial values with regard to Risø P2546 (top), Thies Clima 4.3303 (middle), and Thies Clima 4.3350 (bottom) anemometers calibrated several times at the IDR/UPM Institute. The 300-day average value has been included, together with the standard deviation bars.
Figure 6. Percentage variation of calibration constants A and B from the initial values with regard to Risø P2546 (top), Thies Clima 4.3303 (middle), and Thies Clima 4.3350 (bottom) anemometers calibrated several times at the IDR/UPM Institute. The 300-day average value has been included, together with the standard deviation bars.
Energies 05 01664 g006
Figure 7. Percentage variation of calibration constants A and B from the initial values with regard to Climatronics 100075 (top), and Vector Instruments A100 LK (bottom) cup anemometers calibrated several times at the IDR/UPM Institute. The 300-day average value, together with the standard deviation bars, have been included for the second model (the data corresponding to the Climatronics 100075 model is clearly not sufficient for an equivalent statistic).
Figure 7. Percentage variation of calibration constants A and B from the initial values with regard to Climatronics 100075 (top), and Vector Instruments A100 LK (bottom) cup anemometers calibrated several times at the IDR/UPM Institute. The 300-day average value, together with the standard deviation bars, have been included for the second model (the data corresponding to the Climatronics 100075 model is clearly not sufficient for an equivalent statistic).
Energies 05 01664 g007
In these figures, the 300-day average variation and the standard deviation bars have also been included as a way to filter the great scattering from the data (see also in Table 6 the mentioned average and standard deviation values of each anemometer model, with the exception of the Climatronics 100075, as not enough data is available with regard to this model—only 11 calibrations were carried out in the studied period, see Figure 7). Taking into account this 300-day average evolution, there seems to be a similar behavior between all anemometers, the value of constant A decreasing over time since the first calibration, and the value of constant B increasing. The only exception to this rule is the Thies Clima 4.3350 anemometer, whose 300-day average variation seems to increase slightly with regard to both coefficients (see Figure 6).
Table 6. Percentage variation of coefficients A and B from the initial calibration each 300 days, corresponding to Risø P2546, Thies Clima 4.3303, Thies Clima 4.3350, and Vector Instruments A100 LK anemometers respectively calibrated at the IDR/UPM from June 2003 to January 2011, from September 2003 to September 2010, from November 2003 to January 2011, and March 2005 to January 2011. Average and standard deviation values (see also Figure 6 and Figure 7).
Table 6. Percentage variation of coefficients A and B from the initial calibration each 300 days, corresponding to Risø P2546, Thies Clima 4.3303, Thies Clima 4.3350, and Vector Instruments A100 LK anemometers respectively calibrated at the IDR/UPM from June 2003 to January 2011, from September 2003 to September 2010, from November 2003 to January 2011, and March 2005 to January 2011. Average and standard deviation values (see also Figure 6 and Figure 7).
Risø P2546
Period considered after the first calibrationNumber of calibrations in the considered periodΔA mean
[%]
σΔA
[%]
ΔB mean
[%]
σΔB
[%]
First 300 days50.34%0.34%1.62%7.61%
Between 300 and 600 days390.27%0.40%1.21%12.79%
Between 600 and 900 days210.40%0.44%5.06%22.26%
Between 900 and 1200 days200.70%0.47%4.83%13.42%
Between 1200 and 1500 days120.96%0.43%12.31%17.27%
Between 1500 and 1800 days130.63%0.33%17.05%18.50%
Between 1800 and 2100 days60.75%0.37%3.35%13.03%
Between 2100 and 2700 days30.98%0.36%5.62%11.42%
Thies Clima 4.3303
Period considered after the first calibrationNumber of calibrations in the considered periodΔA mean
[%]
σΔA
[%]
ΔB mean
[%]
σΔB
[%]
First 300 days140.46%0.73%5.58%13.57%
Between 300 and 600 days190.19%0.88%2.00%5.92%
Between 600 and 900 days80.04%0.34%2.43%6.10%
Between 900 and 1200 days150.36%1.82%10.21%33.20%
Between 1200 and 1500 days60.17%0.48%9.51%5.45%
Between 1500 and 2100 days61.02%1.41%7.86%20.17%
Thies Clima 4.3350
Period considered after the first calibrationNumber of calibrations in the considered periodΔA mean
[%]
σΔA
[%]
ΔB mean
[%]
σΔB
[%]
First 300 days520.06%0.66%4.26%42.73%
Between 300 and 600 days1120.03%0.48%3.91%13.08%
Between 600 and 900 days840.16%0.71%3.10%39.46%
Between 900 and 1200 days580.07%0.52%6.76%22.86%
Between 1200 and 1500 days120.03%0.53%5.20%16.38%
Between 1500 and 1800 days160.43%1.01%10.22%19.39%
Between 1800 and 2100 days50.42%0.55%5.62%9.04%
Between 2100 and 2400 days80.25%2.04%60.81%147.46%
Vector Instruments A100 LK
Period considered after the first calibrationNumber of calibrations in the considered periodΔA mean
[%]
σΔA
[%]
ΔB mean
[%]
σΔB
[%]
First 300 days410.13%0.59%6.47%17.59%
Between 300 and 600 days770.15%0.55%1.55%17.22%
Between 600 and 900 days290.01%0.52%12.15%17.38%
Between 900 and 1200 days220.06%0.67%10.36%27.80%
Between 1200 and 1500 days220.30%0.51%5.34%14.16%
Between 1500 and 2100 days100.31%0.26%12.37%23.29%
Table 7. 68.2% confidence limits (ΔAlower, ΔAupper, ΔBlower, and ΔBupper) for the variation of the constants A and B in the intervals 0–300, 300–600 and 600–1000 days from the initial calibration, with regard the anemometers studied (Risø P2546, Thies Clima 4.3303 and 4.3350, and Vector Instruments A100 LK).
Table 7. 68.2% confidence limits (ΔAlower, ΔAupper, ΔBlower, and ΔBupper) for the variation of the constants A and B in the intervals 0–300, 300–600 and 600–1000 days from the initial calibration, with regard the anemometers studied (Risø P2546, Thies Clima 4.3303 and 4.3350, and Vector Instruments A100 LK).
Risø P2546
Interval [days after first calibration]ΔAlower [%]ΔAupper [%]ΔBlower [%]ΔBupper [%]
0–300−0.45%0.03%−17.40%32.36%
300–600−0.67%0.14%−12.84%18.30%
600–1000−0.73%0.19%−16.93%22.17%
Thies Clima 4.3303
Interval [days after first calibration]ΔAlower [%]ΔAupper [%]ΔBlower [%]ΔBupper [%]
0–300−1.18%1.51%−25.01%24.77%
300–600−0.68%1.07%−7.55%1.89%
600–1000−1.29%0.85%−17.05%32.12%
Thies Clima 4.3350
Interval [days after first calibration]ΔAlower [%]ΔAupper [%]ΔBlower [%]ΔBupper [%]
0–300−0.74%0.60%−44.85%36.07%
300–600−0.45%0.52%−12.18%20.61%
600–1000−0.25%0.69%−11.94%13.13%
Vector Instruments A100 LK
Interval [days after first calibration]ΔAlower [%]ΔAupper [%]ΔBlower [%]ΔBupper [%]
0–300−0.75%0.65%−10.95%29.08%
300–600−0.40%0.74%−15.53%18.97%
600–1000−0.64%0.39%−14.30%36.10%
In Table 7 the 68.2% confidence limits (assuming a Gaussian process), for the variation of constants A and B in the intervals 0–300, 300–600 and 600–1000 days from the initial calibration are included. With these variations of the constants A and B it is possible to estimate the same confidence limits for the measured wind speed deviation, Vlower and Vupper, at any wind speed, V:
V lower , upper = ( 1 + Δ A lower , upper ) A 0 f + ( 1 + Δ B lower , upper ) B 0
where ΔAlower, ΔAupper, ΔBlower, and ΔBupper are respectively the variation limits of calibration coefficients from Table 7, A0 and B0 are the calibration constants corresponding to each anemometer (from reference [4]: A0 = 0.627 and B0 = 0.179 for the Risø P2546; A0 = 0.047 and B0 = 0.499 for the Thies Clima 4.330; A0 = 0.0483 and B0 = 0.248 for the Thies Clima 4.3350; A0 = 0.0505 and B0 = 0.195 for the Vector Instruments A100 LK), and f is the frequency output, which can be obviously expressed as a function of the reference wind, V, speed as:
f = V B 0 A 0
In Figure 8 the 68.2% confidence limits for the variation of the measured wind speed at 10 m·s−1 reference wind speed, and for the selected anemometers are shown. The results included in Table 7 show that the loss of performance is greater for the Thies Clima 4.3303, Thies Clima 4.3350, and Vector Instruments A100 LK in the first period considered (0–300 days after the initial calibration) than in the second (300–600 days after the initial calibration). This suggests the existence of a transitional period after installation where the anemometer is adjusting before reach a stable working point. In the case of Thies Clima 4.3303 and Vector Instruments A100 LK, the deviation from the initial calibration in the third period studied (600–1000 days after the initial calibration) returns to the level of the first period. This indicates that the aforementioned adjusting period is over. In the case of the Thies Clima 4.3350 anemometer it seems that the adjustment period is extended to the third period studied. With regard to the Risø P2546, it also seems that there is a transitional period. However, this effect is less clear in this case as only calibration coefficient B seems to increase in the third period its deviation from the first calibration.
Figure 8. 68.2% confidence limits for the variation in the interval between 0 and 300 days after the first calibration, of the measured wind speed at 10 m·s−1 with regard to the anemometers studied.
Figure 8. 68.2% confidence limits for the variation in the interval between 0 and 300 days after the first calibration, of the measured wind speed at 10 m·s−1 with regard to the anemometers studied.
Energies 05 01664 g008
However, these tendencies that could be applied to a large number of anemometers may not describe the behavior of a single individual anemometer. In Figure 9 the behavior of the anemometers that were calibrated more times at the IDR/UP Institute during the studied period of time is shown. No clear pattern for the loss of performance of an anemometer can be extrapolated from these graphs, that is, with the available information it is not possible to estimate, with the procedure described previously for the stored anemometers, the variations of the measured wind speed over time since the first calibration. It should also be said that together with the different conditions in which each anemometer was in service, these anemometers were subjected to different maintenance processes before their calibrations. The maintenance performed on each anemometer referred to in Figure 9 has been included in Table 8, according to the available information. No clear effect of the maintenance on the calibration constants variation with the time elapsed from the initial calibration can be observed. Some anemometers that were not subjected to any maintenance before their calibration showed a change in the A constant variation pattern that, as far as the authors know, could only be attributed to a change in the service conditions (see in Figure 9 variations of A constants with regard to R-1—3rd calibration, and R-2—5th calibration; or in Figure 10 with regard to Th50-1—3rd calibration, and Th50-4—3rd calibration).
Table 8. Maintenance works performed on the more times calibrated anemometers at the IDR/UPM Institute in the studied period of time.
Table 8. Maintenance works performed on the more times calibrated anemometers at the IDR/UPM Institute in the studied period of time.
Risø P2546
AnemometerMaintenance before calibration
2nd3rd4th5th6th
R-1NoNoNoYes 0No
R-2NoNoNoNo-
R-3Yes 0NoNoNo-
R-4NoNoNoNo-
R-5Yes 0NoNoNo-
Thies Clima 4.3303
AnemometerMaintenance before calibration
2nd3rd4th5th6th
Th03-1Yes 1Yes 1(*)Yes 1-
Th03-2Yes 1(*)Yes 1--
Th03-3Yes 1, 2Yes 1Yes 1,2--
Th03-4Yes 1NoYes 1--
Thies Clima 4.3350
AnemometerMaintenance before calibration
2nd3rd4th5th6th
Th50-1NoNoNo--
Th50-2NoYes 1(*)--
Th50-3(*)(*)(*)--
Th50-4NoNoNo--
AnemometerMaintenance before calibration
2nd3rd4th5th6th
LK-1NoYes 1Yes 1--
LK-2NoYes 1No--
LK-3NoYes 1No--
LK-4NoYes 0No--
LK-5NoNoNo--
LK-6NoNoNo--
LK-7NoYes 1Yes--
* No information is available with regard to any possible maintenance before the calibration; 0 No information is available with regard to the maintenance performed to the anemometer, but probably change of bearings; 1 Change of bearings; 2 Change of the cups’ rotor.
Figure 9. Percentage variation of calibration constants A and B from the initial values with regard to (from top to bottom): five specific Risø P2546 anemometers (R-1 to R-5) calibrated more than four times; four specific Thies Clima 4.3303 anemometers (Th03-1 to Th03-4) calibrated more than three times; four specific Thies Clima 4.3350 anemometers (Th50-1 to Th50-4) calibrated more than three times; and seven specific Vector Instruments A100 LK anemometers (LK-1 to LK-7) calibrated more than three times. All anemometers calibrated at the IDR/UPM Institute.
Figure 9. Percentage variation of calibration constants A and B from the initial values with regard to (from top to bottom): five specific Risø P2546 anemometers (R-1 to R-5) calibrated more than four times; four specific Thies Clima 4.3303 anemometers (Th03-1 to Th03-4) calibrated more than three times; four specific Thies Clima 4.3350 anemometers (Th50-1 to Th50-4) calibrated more than three times; and seven specific Vector Instruments A100 LK anemometers (LK-1 to LK-7) calibrated more than three times. All anemometers calibrated at the IDR/UPM Institute.
Energies 05 01664 g009
Going one step further, the residuals concerning the calibrations performed on the single individual anemometers whose maintenance was traced were analyzed. The residuals of a calibration are calculated as the difference between the reference wind speed and the transfer function at all the calibration points. No conclusion that could lead to foreseeing the maintenance requirements for an anemometer can be derived from this particular analysis. In Figure 10, the percentage deviation of the residuals’ standard deviation, σres, from the initial calibration has been plotted for the traced Risø P2546 anemometers as a function of the time elapsed since the initial calibration. In order to have consistent data, only calibrations performed on anemometers never subjected to any maintenance are included in this graph. Although in the figure it seems that the residuals tend to stabilize some time after the initial calibration, no clear conclusion with regard to the loss of performance can be extrapolated from the data.
Figure 10. Variation of the standard deviation of the residuals, σres, as a function of the time elapsed from the initial calibration, with regard to calibrations performed to 27 Risø P2546 anemometers that were not subjected to maintenance. These anemometers were calibrated several times at the IDR/UPM Institute (16 anemometers calibrated 3-times, nine anemometers calibrated 4-times, and two anemometers calibrated 5-times). The symbols indicate second calibration (squares), third calibration (circles), fourth calibration (triangles), and fifth calibration (rhombi). A natural logarithm line has been fitted to the data.
Figure 10. Variation of the standard deviation of the residuals, σres, as a function of the time elapsed from the initial calibration, with regard to calibrations performed to 27 Risø P2546 anemometers that were not subjected to maintenance. These anemometers were calibrated several times at the IDR/UPM Institute (16 anemometers calibrated 3-times, nine anemometers calibrated 4-times, and two anemometers calibrated 5-times). The symbols indicate second calibration (squares), third calibration (circles), fourth calibration (triangles), and fifth calibration (rhombi). A natural logarithm line has been fitted to the data.
Energies 05 01664 g010

4. Conclusions

In this study, the variation of the calibration coefficients over time has been analyzed for six anemometer models (Climatronics 100075, Vector Instruments A100 L2, Vector Instruments A100 LK, Risø P2546, Thies Clima 4.3303, and Thies Clima 4.3350). Two kind of analyses have been carried out, the first one being with anemometers unused in the field, that is, just stored (Climatronics 100075, Vector Instruments A100 L2, Thies Clima 4.3350), and the second one being with anemometers used in the field (Climatronics 100075, Vector Instruments A100 LK, Risø P2546, Thies Clima 4.3303, and Thies Clima 4.3350). The major conclusions resulting from this work are:
  • The stored anemometers showed clear degradation trends, affecting both calibration coefficients, A and B. This degradation of the anemometer’s behavior is translated into a loss of rotation speed (increase of coefficient A), and/or an increase of the offset speed (increase of coefficient B). Depending on the anemometer the degradation can affect both calibration constants differently, thus changing the degradation pattern. The stored anemometers analyzed seemed to have a 450 days transitional period in which the anemometer’s behavior is adjusted.
  • The loss of performance of anemometers used in the field is affected by a great level of scatter. The data analyzed suggest that, in general, the studied anemometers tend to accelerate the rotation speed and increase the offset speed. However, based on the data from anemometers calibrated more than three times over a large period of time (more than 600 days), this conclusion can not be applied to predict the behavior of an individual anemometer. In terms of the data with regard to variation between two consecutive calibrations, the level of scattering was higher for the calibrations done within 300 days than the one for the period between 300 and 600 days. This suggests that as far as normal climatic conditions are concerned, the anemometer has a transitional period after the first calibration before reaching the stable performance range.

Acknowledgments

The authors are indebted to Enrique Vega, Encarnación Meseguer, Patricia Pérez and Luis García for their friendly help and dedication in getting the data concerning the calibrations from the stored data records since 2001. The authors would like to thank also Alvaro Molia (Barlovento), Iván Zaratiegui (Cener), David Pazos (Dekra Ambio), Andrés Brezmes (Ecosem), and Javier Suarez (Ges-Siemsa), for the information provided with regard to the maintenance works performed on the traced anemometers. The authors would like to thank Alain Wery and Chris Lacor, from the Vrije Universiteit Brussel, for sharing their experience in anemometers testing and for their support. Finally, the authors are grateful to Brian Elder and Tania Tate for their kind help in improving the style of the text.

References

  1. López, P.F.; Duro, R.J. A virtual instrument for automatic anemometer calibration with ANN based supervision. IEEE Trans. Instrum. Meas. 2003, 52, 654–661. [Google Scholar] [CrossRef]
  2. Kristensen, L. Can a cup anemometer “underspeed”? Bound. Layer Meteorol. 2002, 103, 163–172. [Google Scholar] [CrossRef]
  3. Kristensen, L. Cup anemometer behavior in turbulent environments. J. Atmos. Ocean. Technol. 1998, 15, 5–17. [Google Scholar] [CrossRef]
  4. Pindado, S.; Vega, E.; Martínez, A.; Meseguer, E.; Franchini, S.; Sarasola, I. Analysis of calibration results from cup and propeller anemometers. Influence on wind turbine Annual Energy Production (AEP) calculations. Wind Energy 2011, 14, 119–132. [Google Scholar] [CrossRef]
  5. Papadopoulos, K.H.; Stefanatos, N.C.; Paulsen, U.S.; Morfiadakis, E. Effects of turbulence and flow inclination on the performance of cup anemometers in the field. Bound. Layer Meteorol. 2001, 101, 77–107. [Google Scholar] [CrossRef]
  6. Paulsen, U.S.; Mortensen, N.G.; Hansen, J.C.; Said, U.S.; Moussa, A.E.-S. Field calibration of cup anemometers. In Proceedings of European Wind Energy Conference and Exhibition, Milan, Italy, 7–10 May 2007.
  7. Kristensen, L.; Jensen, G.; Hansen, A.; Kirkegaard, P. Field Calibration of Cup Anemometers; Risø National Laboratory: Roskilde, Denmark, January 2001. Available online: http://130.226.56.153/rispubl/VEA/veapdf/ris-r-1218.pdf (accessed on 20 May 2011).
  8. Measuring Network of Wind Energy Institutes (MEASNET). Cup Anemometer Calibration Procedure, 1st ed.; MEASNET: Madrid, Spain, 1997. [Google Scholar]
  9. Measuring Network of Wind Energy Institutes (MEASNET). Anemometer Calibration Procedure, 2nd ed.; MEASNET: Madrid, Spain, 2009; Available online: http://www.measnet.com/wp-content/uploads/2011/06/measnet_anemometer_calibration_v2_oct_2009.pdf (accessed on 9 March 2012).
  10. International Standard IEC-61400-12-1. Wind Turbines. Part 12-1: Power Performance Measurements of Electricity Producing Wind Turbines, 2005–2012, 1st ed.; International Electrotechnical Commission: Geneva, Switzerland, 2005. [Google Scholar]

Share and Cite

MDPI and ACS Style

Pindado, S.; Barrero-Gil, A.; Sanz, A. Cup Anemometers’ Loss of Performance Due to Ageing Processes, and Its Effect on Annual Energy Production (AEP) Estimates. Energies 2012, 5, 1664-1685. https://doi.org/10.3390/en5051664

AMA Style

Pindado S, Barrero-Gil A, Sanz A. Cup Anemometers’ Loss of Performance Due to Ageing Processes, and Its Effect on Annual Energy Production (AEP) Estimates. Energies. 2012; 5(5):1664-1685. https://doi.org/10.3390/en5051664

Chicago/Turabian Style

Pindado, Santiago, Antonio Barrero-Gil, and Alfredo Sanz. 2012. "Cup Anemometers’ Loss of Performance Due to Ageing Processes, and Its Effect on Annual Energy Production (AEP) Estimates" Energies 5, no. 5: 1664-1685. https://doi.org/10.3390/en5051664

Article Metrics

Back to TopTop