Measurement-Based Performance

Analysis of Wind Energy Systems

Marcelo G. Simões, Eduard Muljadi, Mohit Singh, and Vahan Gevorgian

W
ind measurements for identifying and chain of wind energy systems, and thus create local revenues
characterizing wind resources require instru- for supporting overall economic growth.
mentation and analysis techniques for defining To correctly and efficiently place wind turbines, it is nec-
signal processing features for gusts, turbulence, and inflow essary to study and observe the possible extraction of wind
winds. This paper describes the importance of those features energy at the desired rate at a chosen site. Usually, when wind
for determining responses of wind turbines. In addition, when speed is lower than 3 m/s (known as calm periods), the ex-
the system is deployed, the instrumentation that must be co- tracted power becomes too low to cover the wind turbine’s
ordinated to manage the utility connection requirements, losses and the system should not be put into operation. There-
such as assessing reactive power supply, fault ride-through, fore, such calm periods indicate the time to use energy storage.
and power quality monitoring, is also discussed. The control Sites with high average wind speeds do not have calm peri-
schemes for wind energy generation systems are discussed ods as often, and thus have less need for storage. The statistical
and associated with the system signal processing required to properties of wind velocity and speed are important in as-
control active power generation. sessing wind energy potential. It is important to establish the
duration curve of wind speed for every hour of the day and ev-
Introduction ery day of the year to obtain a total of 8760 data points.
Wind energy is derived from the movement of air masses that Three basic types of instrumentation are used for wind
drive wind turbines and their coupled generators. A scenario measurements: instruments providing data for the national
has been explored for the U.S. to reach 20% of wind energy meteorological services, instruments designed specifically for
penetration by 2030, in contrast with a scenario in which no determining and identifying the wind resources, and instru-
wind power capacity is installed [1], and recently another ments with high sampling rates (e.g., 10 samples/s) used for
analysis was made for 80% of renewable energy by 2050 [2]. studying gusts, turbulence, and inflow winds for determin-
In these analyses and studies, wind energy’s role in energy se- ing responses of wind turbines. Wind measurement systems
curity, economic prosperity, and environmental sustainability may require several towers and anemometers for turbulence
has been considered. Under the assumption that U.S. electric- characterization.
ity consumption will reach 5.8 billion MW·h per year by 2030, This paper describes how wind turbine controls are re-
investments in 300 000 MW of new wind generation capac- lated to instrumentation and measurement. In addition to
ity would be required. To make this scenario possible by 2030, measuring wind data for site surveys, it is necessary to coor-
wind energy production should increase by approximately dinate online instrumentation to manage utility connection
15%, and wind turbine costs should decrease by approx- requirements, such as: assessing reactive power supply, fault
imately 10%, considering no major breakthroughs in the ride-through capability, voltage control, power quality with
current wind technology. Of course, an overall enhancement monitoring of flicker and harmonics, and frequency control
and upgrade of the electrical transmission system would also so electrical variables are sampled at least at 600 samples/s. A
be necessary to relieve congestion in the existing system, im- proper control scheme must be implemented to control active
prove system reliability for all customers, and increase access and reactive power generation.
to lower-cost energy, particularly for new and remote genera-
tion resources. Both reports [1] and [2] point to some additional Wind Energy Capture
factors, e.g., that investments in wind energy would reduce To have sufficient confidence in characterizing the wind re-
water consumption, reduce nationwide natural gas use, create source at a site, a large sample of wind data collected over a
hundreds of thousands of jobs related to the manufacturing number of years is required. The data are often averaged on

April 2014 IEEE Instrumentation & Measurement Magazine 15

1094-6969/14/$25.00©2014IEEE
Authorized licensed use limited to: King Mongkut's University of Technology Thonburi. Downloaded on March 01,2021 at 07:02:47 UTC from IEEE Xplore. Restrictions apply.
12
c = 3 m/s
10 c = 5 m/s
c = 7 m/s
Yearly time (%)

8 c = 9 m/s
c = 11 m/s
6 c = 13 m/s

0
0 5 10 15 20 25
Wind speed (m/s)

Fig. 1. Rayleigh distribution of the wind speed.

a monthly basis and can be described by a Weibull probabil-

ity function [3]. The Weibull function expresses the fraction
of time the wind speed v is between a given v and v+Δv for a
given Δv. In practice, most sites present a wind distribution for Fig. 2. Speed control ranges for wind turbines.
the shape factor k within the range of 1.5 to 2.5. A typical wind
distribution found in most sites worldwide is known as the angle by the pitch controller. In active stall control, the blade
Rayleigh distribution, given by: pitch angle is continuously adjusted, based on the measured
parameters to generate the desired output power until the
maximum limit is reached. Electric stepper motors can be used
(1)
for this purpose [4], [5].
There are four speed ranges to consider in the operation of
where the scale factor c is related to the number of days with a wind turbine (Fig. 2). The first band goes from zero to min-
high wind speeds (Fig. 1). The higher c is, the higher the num- imum (cut-in) speed for power generation. The second band
ber of windy days. This parameter has at least three distinct has a performance coefficient (Cp) optimized by a system of
implications: a single speed at which the wind blows most of blade position control with pitch control, and the power in-
the time; the secondary predominant speed, which may also creases with increasing wind velocity. The third band is for
contain a lot of energy but is not taken into account; and an av- high-speed winds with a maximum constant output power
erage speed derived from h over the year: limited by the generator rating and maintained by blade con-
trol. The fourth band is a shutdown that is maintained with
(2) blade control and safety brakes, typically above 25 m/s [6], [7].
When the speed of a turbine is constant, the generated
power cannot be controlled and adjusted as the wind speed
Variable Speed Control for Wind Energy changes. So, only a fraction of the total energy is captured. On
Three different power control techniques are used in wind tur- the other hand, it is uneconomical to size a power plant to the
bines: stall regulation, pitch control, and active stall control. maximum power of the winds. There are benefits to incorpo-
Most small (kW-scale) wind turbines are passive stall regu- rating control in wind power turbines to increase wind energy
lated with the machine rotor blades bolted onto the hubs at capture and reduce dynamic loads [8], [9]. The turbine blades
a fixed attack angle. The rotor blade is twisted when mov- attack angle is commonly used to control speed and power
ing along its longitudinal axis to ensure the blade stalls. Stall when wind generators are directly connected to the grid. An-
means a gradual turbulence on the side of the rotor blade not other control approach is to use generators driven by turbines
facing the wind, rather than an abrupt stop when the wind with fixed attack angles of the blades and connected to the util-
speed reaches its critical value. If stall control is applied, there ity through a power electronic circuit and device. Control is
are no moving parts in the rotor and a very complex aero- applied to the load and power flow, which, in turn, acts on the
dynamic control system is needed to avoid stall-induced turbine’s rotation. Because the rotor speed must change ac-
vibrations. cording to the wind, the power converter control has to ensure
When a turbine is under pitch control, an electronic control- that the system follows the maximum power operating point,
ler checks the output power several times per second. When as indicated in Fig. 3 [10], [11].
the output power becomes too high, a command is sent to the The differences among turbine types are based on the elec-
blade pitch mechanism that immediately turns the hydraulic trical generation system that includes the generator, power
drive of the rotor blades slightly out of the wind direction (in- converter, and control algorithms. Strategies used to con-
creasing the angle of attack) – and vice versa, when the wind trol the prime mover use mechanical brakes and blade pitch
speed drops again. The rotor blades are kept at an optimum control to avoid runaway conditions and to keep mechanical

16 IEEE Instrumentation & Measurement Magazine April 2014

Authorized licensed use limited to: King Mongkut's University of Technology Thonburi. Downloaded on March 01,2021 at 07:02:47 UTC from IEEE Xplore. Restrictions apply.
Fig. 3. Turbine rotation power characteristics related to wind speed.

stresses of the wind turbine generator (WTG) in the rated

operating range. The pitch angle of the blades is usually con-
trolled during high wind speeds to keep the aerodynamic
power within the generator rating [12], allowing the output
power and rotor speed to be kept within their limits. The four
most common topologies are:
◗◗ Type 1: Induction generator – fixed speed.
◗◗ Type 2: Wound rotor induction generator with adjustable
external rotor resistance – variable slip.
◗◗ Type 3: Doubly fed induction generator (DFIG) – vari-
able speed.
◗◗ Type 4: Full converter system with a wound fed permanent Fig. 4. Typical wind generator systems.
magnet synchronous generator (PMSG) or asynchro-
nous generators – variable speed, direct drive or with a arrange the wind power plants to achieve the optimal power
gearbox. generation.
Fig. 4 illustrates these four types of wind turbine gener- Real performance is affected by local conditions, nearby
ators [8], [9]. WTGs of types one through three are based on obstructions, power demand profiles, and several turbine-re-
induction generators, and they require a gearbox to match the lated factors, in addition to performance deterioration with
generator speed (high-speed shaft) to the turbine speed (low- aging. Data that can be used for real-time performance assess-
speed shaft). WTGs of type four may be with or without a ment of wind turbines are limited to power, rotor speed, and
gearbox, depending on the type of the generator. wind speed. To gain enough data to obtain meaningful results,
three variables would have to be logged in real time: the out-
Wind Power Plant Energy Management put power of the turbine W, the rotational speed of the turbine
and Power Quality Control in rad/s, and the wind speed in m/s.
Each year, new wind power plant sites are commissioned The analysis would require instrumentation to obtain four
worldwide. However, it is difficult to accurately predict how system parameters: system voltage, system current, wind
well a turbine will perform in different atmospheric con- power plant turbines rotational speed distribution, and wind-
ditions. The data provided by manufacturers are only the power plant wind speed distribution. Therefore, wind power
individual turbine power curves, describing how output power plants require extensive data processing of the electrical net-
varies with wind speed. Although some atmospheric science- work characteristics and analyses of how the power quality
based approaches have been taken, using laser and sound impacts the performance.
versions of radar (LIDAR and SODAR) for measuring how It is possible to extend the instrumentation and signal pro-
the wind patterns affect performance, these instruments are cessing for wind turbine data aggregation of a large wind
expensive to install, and data analysis is difficult to under- power plant. The method of analysis used to measure the char-
stand, making it difficult to decide where to site and how to acteristics plays an important role in reflecting real conditions

April 2014 IEEE Instrumentation & Measurement Magazine 17

Authorized licensed use limited to: King Mongkut's University of Technology Thonburi. Downloaded on March 01,2021 at 07:02:47 UTC from IEEE Xplore. Restrictions apply.
Fig. 5. NREL’s National Wind Technology Center near Boulder, CO.

as closely as possible. From the power fluctuation perspective, available distribution network capacity. As a practical rule, the
the larger the area of the wind power plant, the more diverse output characteristics of wind turbine power do not exactly
the wind profiles driving each turbine. Therefore, the wind follow those of generator power, and they must be matched
fluctuations at one turbine will not be in phase with those at an- in the most reasonable way possible. Based on the maximum
other turbine farther away. A wind power plant with a number speed expected for the wind turbine, and taking into account
of small turbines creates less power and voltage fluctuations the cubic relationship between the wind speed and power, the
on the power grid than one with a few large turbines. As the designer must select the generator and gearbox to match these
number of turbines in a wind power plant increases over a limits. The most sensitive point is the correct selection of the
large area, the characteristics of the wind power plant are bet- rated turbine speed for the power plant. If it is too low, genera-
ter averaged. This lowers the impact of tower shadows and tion for high-speed winds will not be efficient. If it is too high,
wind turbulence on the output power. The more turbines that the capacity factor will be too low. There is an iterative design
are used to represent a wind power plant, the smaller the im- process to match the characteristic of commercially available
pact is from individual fluctuations. wind turbines and generators with regards to cost, efficiency,
Eventually, the collective behavior of the wind power plant and the maximum generated power. The maximum value of
farm is approximated by an average modeling, following a Cp should occur approximately at the same speed of the max-
similar pattern of an individual wind turbine, when the power imum power in the power distribution curve. Therefore, the
curve is such that the operation below the rated wind speed tip-speed ratio must be kept optimally constant at the max-
gives a positive slope ΔP/Δv > 0, while the operation around imum speed possible to capture the maximum wind power.
the rated wind speed is constant, i.e., ΔP/Δv ~ 0 and above is a This feature suggests that to optimize the annual energy cap-
negative slope ΔP/Δv < 0. ture at a given site, it is necessary that the turbine speed follows
The electrical output of a wind turbine is connected to a the wind speed variation, or tip-speed ratio (TSR) is kept con-
point of interconnection (POI), and then its power is transmitted stant, so as to keep Cp maximized. This is illustrated in Fig. 3,
by a transmission or distribution line. To analyze these spa- where the output power is controlled to follow the cube func-
cial effects, the wind speed applied to one group is time shifted tion of the rotational speed. Obviously, the design stress must
with respect to the other. The time shift can be calculated by di- be kept within the limits of the turbine manufacturer’s data
viding the distance between the centers of the two groups by because torque relates to the instantaneous power by: P = T.
the average wind speed. This simplified assumption can help
simulate the aggregation impact of wind turbines on a large Wind Generator Research Capabilities
wind power plant. at NREL
Several types of generators have been considered for NREL’s National Wind Technology Center (NWTC) is the na-
different applications of wind turbines. [10], [11]. After the in- tion’s premier wind energy technology research facility (Fig.
stallation site of the wind power plant has been selected, the 5) [13], [14]. The NWTC test site features approximately 11
next steps are to select the turbine rating, the generator, and the MW of variable renewable generation: wind and photovoltaic
distribution system. In general, the distribution transformer (PV) cells; dynamometers for wind turbine drivetrain testing
is sized on the peak capacity of the generator according to the (5 MW, 2.5 MW, and 225 kW); and a 7-MVA controllable grid

18 IEEE Instrumentation & Measurement Magazine April 2014

Authorized licensed use limited to: King Mongkut's University of Technology Thonburi. Downloaded on March 01,2021 at 07:02:47 UTC from IEEE Xplore. Restrictions apply.
Fig. 6. NWTC CGI for testing grid integration aspects of renewable generation.

interface (CGI) for testing grid integration aspects of renew- For example, the CGI allows wind turbine generator manu-
able generation. Accredited through the American Association facturers to test both the mechanical and electrical characteristics
of Laboratory Accreditation (A2LA) to perform certification of their machines in a controlled grid environment by replicat-
testing of wind turbines in accordance with the International ing many electrical scenarios that are only partially available in
Electrotechnical Commission (IEC) international standards, field-testing. The renewable energy industry now has a plat-
the NWTC performs tests required by wind turbine certifica- form on which to ensure that renewable energy systems meet
tion agencies, financial institutions, and other organizations stringent national and international electrical standards, and to
throughout the world (Fig. 6). test grid compliance of innovative electrical topologies and con-
The CGI specializes in testing multi-mega-Watt-scale wind trols. This will increase reliability and lower the cost of energy
and PV technologies as well as energy storage devices, trans- delivered by wind and solar power. CGI capabilities go beyond
formers, and protection equipment at transmission level fault ride-through to provide comprehensive electrical testing
voltages. The CGI’s capabilities allow industries to partner solutions at the multi-mega Watt level.
with NREL to test, optimize, and visualize the grid integration-
related performance of a unit before it is deployed in the field. Conclusion
Through a virtual link with the super-computing capabilities The power of a wind generator is directly affected by the wind
of the Energy Systems Integration Facility (ESIF), researchers speed. The majority of modern wind turbines are designed
and industry partners can enhance visualizations of complex with variable speed generators to maintain the operation at
systems in a virtual environment, and realize advanced real- a high efficiency level. In addition, high towers for wind tur-
time testing schemes combining CGI’s extreme flexibility with bines increase increase the energy capture but also increase the
ESIF’s grid simulator and smart grid capabilities. The CGI is a turbines’ exposure to turbulence and wind gusts that may af-
valuable asset for the energy industry that will save time and fect the generator’s performance. This paper described the
resources while minimizing integration issues, improving the instrumentation required for wind measurements for defining
reliability, and advancing the development of grid-friendly re- wind resources and emphasized the needs of (or, necessity to
newable and emerging technologies [15]–[17]. use) signal processing for determining gusts, turbulence, and

April 2014 IEEE Instrumentation & Measurement Magazine 19

Authorized licensed use limited to: King Mongkut's University of Technology Thonburi. Downloaded on March 01,2021 at 07:02:47 UTC from IEEE Xplore. Restrictions apply.
inflow winds for measuring the response of wind turbines. [9] M. G. Simões, S. Chakraborty, and R. Wood, “Induction
The requirements for instrumentation and measurement for generators for small wind energy systems,” IEEE Power Electronics
wind power plant energy systems have been discussed as Society Newsletter, vol. 18, no. 3, pp. 19-23, 2006.
well as the coordination to manage the utility connection, such [10] B. Kroposki, C. Pink, R. DeBlasio, H. Thomas, M. G. Simões, and
as assessing reactive power supply, fault ride-through, and P. K. Sen, “Benefits of power electronic interfaces for distributed
power quality monitoring. The paper showed schemes for energy systems,” IEEE Trans. Energy Conversion, vol. 25, no. 3, pp.
wind energy generation systems in conjunction with signal 901-908, Sept. 2010.
processing for the active power control. Several practical ex- [11] J. Ojeda, M. G. Simões, G. Li, and M. Gabsi, “Design of a flux
amples of the wind-related measurements were also discussed switching electrical generator for wind turbine systems,” IEEE
to illustrate how power quality is monitored in wind turbines. Trans. Ind. Appl., vol. 48, no. 6, pp. 1808-1816, Nov.–Dec. 2012.
[12] F. A. Farret, B. Palle, and M. Simões, “Full expandable model of
References parallel self-excited induction generators,” IEE Proc. Electric Power
[1] M. Hand, N. Blair, M. Bolinger, R. Wiser, R. O’Connell, T. Hern, Appl., vol. 152, no. 1, pp. 96-102, Jan./Mar. 2005.
and B. Miller, “Power system modeling of 20% wind-generated [13] National Renewable Energy Laboratory. [Online]. Available:
electricity by 2030,” IEEE Power Engineering Society General http://www.nrel.gov/esi/.
Meeting, Pittsburg, PA, USA, pp. 1–8, 2008. [14] B. Kroposki, B. Garrett, S. Macmillan, B. Rice, C. Komomua,
[2] Z. Shahan, 80% Clean, Renewable Energy by 2050: More Than M. O’Malley, and D. Zimmerle, “Energy systems integration
Possible, But Need More Political Will (& Public Demand). Scientific - A convergence of ideas,” NREL Technical Paper, NREL/TP-
American. [Online]. Available: http://www.scientificamerican. 6A00-55649, July 2012.
com/article.cfm?id=80-clean-renewable-energy-by- [15] M. Singh, V. Gevorgian, E. Muljadi, and E. Ela, “Variable-speed
2050-m-2012-07. wind power plant operating with reserve power capability,” IEEE
[3] M. R. Patel, Wind and Solar Power Systems: Design, Analysis, and Energy Conversion Congress and Exposition (ECCE), Denver, CO,
Operation, Washington, D.C., CRC Press, 2005. Sept. 2013, pp. 3305-3310.
[4] W. Hughes, D. K. McLaughlin, and R. Ramakumar, Applications of [16] IEC 61400-21, Ed.1: Wind turbine generator systems—Part 21:
Wind Energy Systems, 1976. Measurement and assessment of power quality characteristics of
[5] J. S. Rohatgi and V. Nelson, Wind Characteristics: an Analysis for grid connected wind turbines, Final Draft International Standard
Generation of Wind Power, Alternative Energy Institute, West Texas 88/144/FDIS International Electrotechnical Commission, IEC
A&M University, 1994. 2001-07-01.
[6] P. Novak, T. Ekelund, I. Jovik, and B. Schmidtbauer, “Modeling [17] IEEE Std 519–1992: IEEE Recommended Practices and
and control of variable-speed wind-turbine drive-system Requirements for Harmonic Control in Electrical Power Systems,
dynamics,” IEEE Control Systems Society, vol. 15, no. 4, pp. 28-38, pp. 1-112, Apr. 9 1993.
1995.
[7] M. G. Simões, B. K. Bose, and R. J. Spiegel, “Design and Marcelo Godoy Simões is an associate professor with the De-
performance evaluation of a fuzzy-logic-based variable-speed partment of Electrical Engineering and Computer Science at
wind generation system,” IEEE Trans. Industry Appl., vol. 33, pp. Colorado School of Mines, Golden, CO, and the director for
956-965, July/Aug. 1997. ACEPS – Center for the Advanced Control of Energy and
[8] M. G. Simões, B. K. Bose, and R. J. Spiegel, “Fuzzy logic based Power Systems at CSM.
intelligent control of a variable speed cage machine wind
generation system,” IEEE Trans. Power Electron., vol. 12, pp. 87-95, Eduard Muljadi, Vahan Gevorgian, and Mohit Singh are with
Jan. 1997. the National Renewable Energy Laboratory in Golden, CO.

20 IEEE Instrumentation & Measurement Magazine April 2014

Authorized licensed use limited to: King Mongkut's University of Technology Thonburi. Downloaded on March 01,2021 at 07:02:47 UTC from IEEE Xplore. Restrictions apply.