Claims:1. A wake detection system for detecting and responding to a wake condition at one or more wind turbines in a wind power generation plant (100), the system comprising:
a turbulence length scale estimation sub-unit (128) configured to estimate a turbulence length scale (208) of wind turbulence corresponding to at least a first wind turbine (104) and a second wind turbine (104) of a set of interacting wind turbines based on wind speed data (112) corresponding to the wind power generation plant (100) for one or more operating intervals;
a turbulence length scale analytics sub-unit (130) configured to:
identify a convergence of a first turbulence length scale (208) of wind turbulence corresponding to at least the first wind turbine (104) with a second turbulence length scale (208) of wind turbulence corresponding to at least the second wind turbine (104);
identify a divergence of the second turbulence length scale (208) corresponding to at least the second wind turbine (104) from the first turbulence length scale (208) corresponding to at least the first wind turbine (104) subsequent to the convergence of the first and second turbulence length scales (208) corresponding to at least the first and second wind turbines (104);
detect an occurrence of a wake condition corresponding to at least the second wind turbine (104) with respect to at least the first wind turbine (104) based on the identified divergence;
calibrate one or more operational settings corresponding to at least the first wind turbine (104) and the second wind turbine (104) based on the identified wake condition of at least the second wind turbine (104); and
a controller (134) configured to alter one or more operational settings corresponding to at least the first wind turbine (104) and the second wind turbine (104) based on the calibrated operational settings.
2. The system of claim 1, wherein the system is further configured to receive the wind speed data (112) corresponding to the one or more operating intervals from one or more measurement devices (106) disposed on the one or more wind turbines (104), and wherein the one or more measurement devices (106) comprise an anemometer, a LIDAR, or a combination thereof.
3. The system of claim 1, wherein the turbulence length scale analytics sub-unit (130) is further configured to identify one or more sets of interacting wind turbines based on at least one wind direction (108) corresponding to the wind power generation plant (100) for the one or more operating intervals and farm layout data of the wind power generation plant (100).
4. The system of claim 3, wherein the turbulence length scale analytics sub-unit (130) is further configured to:
compute a statistical metric configured to characterize a relationship between the first turbulence length scale (208) corresponding to the first wind turbine (104) and the second turbulence length scale (208) corresponding to the second wind turbine (104);
generate a divergence quantifier by processing the statistical metric via at least one analytic function; and
identify the occurrence of the wake condition based on a comparison of the divergence quantifier with a determined threshold.
5. The system of claim 4, wherein the statistical metric comprises a difference between the first turbulence length scale (208) and the second turbulence length scale (208), a ratio of the first turbulence length scale (208) and the second turbulence length scale (208), or a combination thereof.
6. The system of claim 4, wherein the analytic function comprises a slope function, a mean average function, a standard deviation, a logistics function, or combinations thereof.
7. The system of claim 1, wherein the turbulence length scale estimation sub-unit (128) is further configured to:
define a state vector corresponding to a wind turbine (104) of the one or more wind turbines based on one or more of a mean wind speed component, a wind turbulence component, and a turbulence length scale component, wherein the state vector models a turbulence corresponding to the wind turbine (104) of the one or more wind turbines;
define a process state model for the state vector based on the changes in one or more of the mean wind speed component, the wind turbulence component and the turbulence length scale, wherein the process state model models dynamics of the state vector;
define a measurement model corresponding to the wind turbine (104) based on a mean wind speed component and a wind turbulence component obtained from wind speed data (112); and
estimate a turbulence length scale (208) corresponding to the wind turbine (104) based on the process state model and the measurement model via a non-linear estimator.
8. The system of claim 7, wherein the non-linear estimator comprises an extended Kalman filter (EKF), an unscented Kalman filter, a recursive least squares estimator, a particle filter, or combinations thereof.
9. The system of claim 1, wherein the controller (134) is further configured to reduce an impact of the detected wake condition on the set of interacting wind turbines based on the calibrated operational settings.
10. A wind power generation plant (100), comprising:
one or more wind turbines (104);
one or more measuring devices (106) disposed on the one or more wind turbines (104), wherein the one or more measuring devices (106) are configured to measure wind speed at the one or more wind turbines;
a wake detection system comprising:
a turbulence length scale estimation sub-unit (128) configured to estimate a turbulence length scale (208) of wind turbulence corresponding to at least a first wind turbine (104) and a second wind turbine (104) of a set of interacting wind turbines based on wind speed data (112) corresponding to the wind power generation plant (100) for one or more operating intervals;
a turbulence length scale analytics sub-unit (130) configured to:
identify a convergence of a first turbulence length scale (208) of wind turbulence corresponding to at least the first wind turbine (104) with a second turbulence length scale (208) of wind turbulence corresponding to at least the second wind turbine (104);
identify a divergence of the second turbulence length scale (208) corresponding to at least the second wind turbine (104) from the first turbulence length scale (208) corresponding to at least the first wind turbine (104) subsequent to the convergence of the first and second turbulence length scales (208) corresponding to at least the first and second wind turbines (104);
detect an occurrence of a wake condition corresponding to at least the second wind turbine (104) with respect to at least the first wind turbine (104) based on the identified divergence;
calibrate one or more operational settings corresponding to at least the first wind turbine (104) and the second wind turbine (104) based on the identified wake condition of at least the second wind turbine (104); and
a controller (134) configured to alter one or more operational settings corresponding to at least the first wind turbine (104) and the second wind turbine (104) based on the calibrated operational settings.
11. The system of claim 10, wherein the one or more measurement devices (106) comprise an anemometer, a LIDAR, or a combination thereof.
12. The system of claim 10, wherein the controller (134) is further configured to reduce an impact of the detected wake condition on the set of interacting wind turbines based on the calibrated operational settings.
13. A method (200) for detecting and responding to a wake condition at one or more wind turbines in a wind power generation plant (100), the method comprising:
estimating (206) a turbulence length scale (208) of wind turbulence corresponding to at least a first wind turbine (104) and a second wind turbine (104) of a set of interacting wind turbines, based on wind speed data (112) corresponding to the wind power generation plant (100) for one or more operating intervals;
identifying (210, 212) a convergence of a first turbulence length scale (208) of wind turbulence corresponding to at least the first wind turbine (104) with a second turbulence length scale (208) of wind turbulence corresponding to at least the second wind turbine (104);
identifying (214, 216) a divergence of the second turbulence length scale (208) corresponding to at least the second wind turbine (104) with respect to the first turbulence length scale (208) corresponding to at least the first wind turbine (104), subsequent to the convergence of the first and second turbulence length scales (208) corresponding to at least the first and second wind turbines (104);
detecting (218, 219, 220) the occurrence of a wake condition corresponding to at least the second wind turbine (104) with respect to at least the first wind turbine (104) based on the identified divergence;
calibrating (224) one or more operational settings corresponding to at least the first wind turbine (104) and the second wind turbine (104) based on the identified wake condition of at least the second wind turbine (104); and

altering (226) one or more operational settings corresponding to at least the first wind turbine (104) and the second wind turbine (104) based on the calibrated operational settings.
14. The method of claim 13, wherein detecting the occurrence of a wake condition comprises:
computing (218) a statistical metric configured to characterize a relationship between the first turbulence length scale corresponding to the first wind turbine and the second turbulence length scale corresponding to the second wind turbine;

generating (219) a divergence quantifier by processing the statistical metric via at least one analytic function; and

identifying (220) the occurrence of the wake condition based on a comparison of the divergence quantifier with a determined threshold.
15. The method of claim 13, wherein estimating the turbulence length scale comprises:
defining (302) a state vector corresponding to a wind turbine (104) of the one or more wind turbines based on one or more of a mean wind speed component, a wind turbulence component and a turbulence length scale, wherein the state vector models a turbulence corresponding to the wind turbine (104);
defining (304) a process state model for the state vector based on the changes in one or more of the mean wind speed component, the wind turbulence component and the turbulence length scale, wherein the process state model models dynamics of the state vector;
defining (306) a measurement model corresponding to the wind turbine (104) based on a mean wind speed component and a wind turbulence component obtained from wind speed data (112); and

estimating (308) a turbulence length scale (310) based on the process state model and the measurement model via a non-linear estimator.

16. The method of claim 15, wherein altering one or more operational settings comprises reducing an impact of the detected wake condition on the set of interacting wind turbines based on the calibrated operational settings.
, Description:BACKGROUND
[0001] Embodiments of the present specification relate generally to wake detection in a wind
farm and more particularly to systems and methods for the real-time measurement and
characterization of turbulence conditions at a wind turbine to identify the occurrence of a wake
condition at the wind turbine.
[0002] Wind direction is highly variable at a wind farm system or a wind power generation
plant site that includes one or more wind turbines. Typically, layouts of a wind farm or a wind
power generation plant are based on an in-depth analysis of prevailing wind conditions in the
area and terrain of the area where the turbines are erected at specific locations relative to one
another in order to enable the turbines to capture as much wind as possible.
[0003] The wind farm layout, also known as farm geometry, serves as a reference point for
monitoring the location of the individual wind turbines relative to the wind direction. Since wind
direction is variable on a day-to-day basis, turbines are often re-aligned with the wind direction
via rotor movement. Depending on the direction of the wind flowing through the wind farm at
any given time, one or more turbines may be directly in line with one another relative to the wind
direction. In such a situation, the wind turbines that are located upstream from the other wind
turbines relative to the wind direction create a wake. As will be appreciated, wakes are invisible
ripples and/or waves and other disturbances that modify the prevailing turbulence conditions in
the atmosphere downstream and may result in damage to the wind turbines and decreased
efficiency of the wind farm. Wind turbines in the path of the wake may experience extra loads
on the blades, thereby causing more wear and tear. Additionally, the power produced by the
downstream wind turbines may be reduced considerably due to the resulting velocity deficit at
the downstream wind turbines. Furthermore, unduly turbulent conditions experienced by the
downstream turbines caused by the wake may result in non-uniform loading of the downstream
wind turbines.
[0004] Typically, presently available methods aid wind farm operators in estimating wind
direction via use of radar and weather information, and identifying the wind turbines that may lie
in a wake path of the most upstream wind turbines relative to the wind direction. Additionally,
the operation of the upstream wind turbines may be curtailed in order to reduce the wake
experienced by the downstream turbines and enable the downstream turbines to capture more
free-stream wind.
[0005] Some of these methods may fail to account for certain factors and characteristics of
wind turbulence. One such example includes a scenario where the wake produced by an
upstream wind turbine may experience a lateral shift at one or more downstream wind turbines
depending on prevailing turbulence conditions. This shift may be temporary compared to the
general wind direction. Also, in some situations, the lateral shift of the wake, also known as
meander, may result in minimal or no impact on the downstream wind turbines for intermittent
periods of time. The currently available methods may fail to verify/confirm any impact of the
wake on the downstream wind turbines prior to curtailing the operation of the upstream wind
turbines. In this situation, the needless curtailment of the operation of the upstream wind
turbines leads to sub-optimal harnessing of free-stream wind. Furthermore, in some examples,
the settings of the downstream wind turbines are based on a projection of wake occurrence,
which in turn is based on wind direction and/or turbulence intensity, thereby resulting in suboptimal
operation of the wind turbines.
BRIEF DESCRIPTION
[0006] In accordance with aspects of the present specification, a wake detection system for
detecting and responding to a wake condition at one or more wind turbines in a wind power
generation plant is presented. The wake detection system includes a turbulence length scale
estimation sub-unit configured to estimate a turbulence length scale of wind turbulence
corresponding to at least a first wind turbine and a second wind turbine of a set of interacting
wind turbines based on wind speed data corresponding to the wind power generation plant for
one or more operating intervals. Additionally, the wake detection system includes a turbulence
length scale analytics sub-unit configured to identify a convergence of a first turbulence length
scale of wind turbulence corresponding to at least the first wind turbine with a second turbulence
length scale of wind turbulence corresponding to at least the second wind turbine. The
turbulence length scale analytics sub-unit is additionally configured to identify a divergence of
the second turbulence length scale corresponding to at least the second wind turbine from the
first turbulence length scale corresponding to at least the first wind turbine subsequent to the
convergence of the first and second turbulence length scales corresponding to at least the first
and second wind turbines. Moreover, the turbulence length scale analytics sub-unit is further
configured to detect an occurrence of a wake condition corresponding to at least the second wind
turbine with respect to at least the first wind turbine based on the identified divergence and
calibrate one or more operational settings corresponding to at least the first wind turbine and the
second wind turbine based on the identified wake condition of at least the second wind turbine.
Furthermore, the wake detection system includes a controller configured to alter one or more
operational settings corresponding to at least the first wind turbine and the second wind turbine
based on the calibrated operational settings.
[0007] In accordance with another aspect of the present specification, a wind power
generation plant is presented. The wind power generation plant includes one or more wind
turbines configured to generate electric power. Additionally, the system includes one or more
measuring devices disposed on the one or more wind turbines, wherein the one or more
measuring devices are configured to measure wind speed at the one or more wind turbines.
Furthermore, the system includes a wake detection system configured to detect and respond to a
wake condition at the one or more wind turbines of the wind power generation plant. The wake
detection system includes a turbulence length scale estimation sub-unit configured to estimate a
turbulence length scale of wind turbulence corresponding to at least a first wind turbine and a
second wind turbine of a set of interacting wind turbines based on wind speed data
corresponding to the wind power generation plant for one or more operating intervals.
Additionally, the wake detection system includes a turbulence length scale analytics sub-unit
configured to identify a convergence of a first turbulence length scale of wind turbulence
corresponding to at least the first wind turbine with a second turbulence length scale of wind
turbulence corresponding to at least the second wind turbine. The turbulence length scale
analytics sub-unit is additionally configured to identify a divergence of the second turbulence
length scale corresponding to at least the second wind turbine from the first turbulence length
scale corresponding to at least the first wind turbine subsequent to the convergence of the first
and second turbulence length scales corresponding to at least the first and second wind turbines.
Moreover, the turbulence length scale analytics sub-unit is further configured to detect an
occurrence of a wake condition corresponding to at least the second wind turbine with respect to
at least the first wind turbine based on the identified divergence, and calibrate one or more
operational settings corresponding to at least the first wind turbine and the second wind turbine
based on the identified wake condition of at least the second wind turbine. Furthermore, the
wake detection system includes a controller configured to alter one or more operational settings
corresponding to at least the first wind turbine and the second wind turbine based on the
calibrated operational settings.
[0008] In accordance with yet another aspect of the present specification, a method for
detecting and responding to a wake condition at one or more wind turbines in a wind power
generation plant is presented. The method includes estimating a turbulence length scale of wind
turbulence corresponding to at least a first wind turbine and a second wind turbine of a set of
interacting wind turbines, based on wind speed data corresponding to the wind power generation
plant for one or more operating intervals. Further, the method includes identifying a
convergence of a first turbulence length scale of wind turbulence corresponding to at least the
first wind turbine with a second turbulence length scale of wind turbulence corresponding to at
least the second wind turbine. Additionally, the method includes identifying a divergence of the
second turbulence length scale corresponding to at least the second wind turbine with respect to
the first turbulence length scale corresponding to at least the first wind turbine, subsequent to the
convergence of the first and second turbulence length scales corresponding to at least the first
and second wind turbines and detecting the occurrence of a wake condition corresponding to at
least the second wind turbine with respect to at least the first wind turbine based on the identified
divergence. Moreover, the method includes calibrating one or more operational settings
corresponding to at least the first wind turbine and the second wind turbine based on the
identified wake condition of at least the second wind turbine and altering one or more
operational settings corresponding to at least the first wind turbine and the second wind turbine
based on the calibrated operational settings corresponding to at least the second wind turbine.
DRAWINGS
[0009] These and other features and aspects of embodiments of the present specification will
become better understood when the following detailed description is read with reference to the
accompanying drawings in which like characters represent like parts throughout the drawings,
wherein:
[0010] FIG. 1 is a schematic diagram of an exemplary wind power generation plant including
a wake detection subsystem for detecting a wake condition at one or more wind turbines, in
accordance with aspects of the present specification;
[0011] FIG. 2 is a flow chart illustrating a method for detecting a wake condition at one or
more wind turbines of the wind power generation plant, in accordance with aspects of the present
specification;
[0012] FIG. 3 is a flow chart illustrating a method for estimating a turbulence length scale
corresponding to turbulence conditions at one or more wind turbines of the wind power
generation plant, in accordance with aspects of the present specification;
[0013] FIG. 4A is a graph depicting an exemplary convergence of turbulence length scale
curves of one or more wind turbines in the wind power generation plant, in accordance with
aspects of the present specification, and
[0014] FIG. 4B is a diagrammatic representation of an exemplary occurrence of a wake
condition in a wind turbine, in accordance with aspects of the present specification.
DETAILED DESCRIPTION
[0015] As will be described in detail hereinafter, various embodiments of exemplary systems
and methods for enhanced wake detection in a wind farm are presented. Advantageously, the
systems and methods may be retrofit to an existing wind farm, thereby leveraging the existing
wind farm communication network, while circumventing the need for additional hardware and
associated expenditure. Moreover, the systems and methods provide wake detection in real-time
using sensors that are already on board the wind turbines.
[0016] In an effort to provide a concise description of these embodiments, all features of an
actual implementation may not be described in the specification. It should be appreciated that in
the development of any such actual implementation, as in any engineering or design project,
numerous implementation-specific decisions must be made to achieve the developer’s specific
goals such as compliance with system-related and business-related constraints.
[0017] When describing elements of the various embodiments of the present invention, the
articles “a”, “an”, “the” and “said” are intended to mean that there are one or more of the
elements. The terms “comprising”, “including” and “having” are intended to be inclusive and
mean that there may be additional elements other than the listed elements.
[0018] The terms “system” and “plant” may be used interchangeably in the present
specification in conjunction with the phrase “wind power generation,” and are used to refer to the
wind power generation system as a whole, including the wind farm and the balance of plant as a
power source to an electrical grid. The term “wind farm” is used to refer to a collection of wind
turbines that are disposed proximate to one another and used to produce electrical power.
[0019] In a wind power generation plant, the term “balance of plant” refers to all the
infrastructure and facilities of a wind farm and their associated cost with the exception of the
wind turbines and all corresponding elements, for example, the civil infrastructure and costs,
transformers, Supervisory Control and Data Acquisition (SCADA) systems, electrical cabling,
and substation electrical components such as switchgear, DC Links, rotor converters, medium
voltage (MV) transformers, grid converters, boost converters, power protection systems, electric
meters, and other components to facilitate the operational aspects of the wind farm. The balance
of plant may also include components that are used to facilitate the connection to the distribution
network such as switches, reinforcements, and protection and measuring elements.
[0020] The term “free-stream wind” refers to undisturbed, natural air flow, usually at the hub
height of a wind turbine. A wind turbine is said to be “upstream” relative to one or more other
wind turbines if the wind turbine experiences free-stream wind before the other wind turbines.
Correspondingly, a wind turbine is said to be “downstream” relative to one or more other wind
turbines if the wind turbine experiences air flow after the other wind turbines.
[0021] An exemplary wind power generation system 100 is illustrated in FIG. 1. In a
presently contemplated configuration, the wind power generation system 100 includes a wind
farm 102, a wake detection subsystem 120, and a controller 134. In some embodiments, the
wind power generation system may be controlled and/or manned by a system operator 132. In
the wind power generation system 100 of FIG. 1, the controller 134 is operatively coupled to the
wind farm 102. Also, the wake detection subsystem 120 may be communicatively coupled to the
wind farm 102.
[0022] The wind farm 102 of FIG. 1 may include one or more wind turbines, generically
referenced by reference numeral 104. The terms wind turbines and turbines may be used
interchangeably. One or more measurement devices 106 may be disposed on the wind turbines
104. These measurement devices are configured to measure wind speed, wind direction, and the
like. Also, some examples of the measurement devices 106 include anemometers, rotor speed
sensors, light imaging, detecting, and ranging (LIDAR) sensors and combinations and/or
variations thereof. In certain embodiments, the measurement devices 106 may be
communicatively coupled to the wake detection subsystem 120 directly or indirectly via use of
data repositories, data services, and the like. Moreover, the measurement devices 106 are
configured to convey measurement data 112 to the wake detection subsystem 120. The
measurement data 112 is generally representative of data gathered by the measurement devices
106 and may include measured wind speed, wind direction and the like.
[0023] As previously noted, the wind farm layout refers to the relative location of the wind
turbines 104, where the wind turbines 104 are erected at specific locations relative to one another
in order to enable the wind turbines 104 to optimally capture wind. The wind farm layout, also
known as farm geometry, serves as a reference point for monitoring the location of the individual
wind turbines relative to the wind direction.
[0024] The terms “interacting turbines” or “interacting wind turbines” or “set of interacting
wind turbines” as used herein refer to a set of one or more wind turbines that may be in
alignment with one another and with the prevailing wind direction, resulting in a situation where
one or more members of the set of interacting turbines may be in the wake of another wind
turbine.
[0025] As illustrated in FIG.1, reference numeral 108 depicts an example of a prevailing wind
direction of free-stream wind at the wind farm 102. Correspondingly, wind turbines referenced
by reference labels A1, A2 and A3 are depicted as being in mutual alignment and also in
alignment with the prevailing free-stream wind direction 108. The wind turbines A1, A2 and A3
form one set of interacting turbines. Moreover, in FIG. 1, the wind turbines A2 and A3 may be
in the wake of A1 with respect to the wind direction 108. In another example, wind turbines C1,
A2, and B1 may also be in mutual alignment. This mutual alignment of wind turbines C1, A2,
and B1 is generally represented by reference numeral 109. In this example the wind turbines C1,
A2, and B1 are not in alignment with the prevailing free-stream wind direction 108. Hence,
these wind turbines C1, A2, and B1 do not form a set of interacting wind turbines. In a similar
fashion, wind turbines A1, B1, and B2 are aligned with one another, but are not in alignment
with the prevailing wind stream direction 108 and hence, do not form a set of interacting
turbines.
[0026] With continuing reference to FIG. 1, reference numeral 110 depicts one example of a
temporary lateral shift in the wake with respect to the original free-stream wind direction 108.
As previously noted, this lateral shift or meander in the wake 110 may be due to lateral
components of wind turbulence or a systemic bias. In this example, even in the absence of a
change in the direction of free-stream wind 108, the lateral shift in the wake 110 may cause
turbine A2 to cease experiencing the wake caused by wind turbine A1 with respect to wind
direction 108 and instead experience the free-stream wind 108.
[0027] Turning now to the wake detection subsystem 120, in one embodiment, the wake
detection subsystem 120 may include a data repository 122, a memory sub-unit 124, and a
processor sub-unit 126. The wake detection subsystem 120 may be communicatively coupled to
one or more measurement devices 106 disposed on the wind turbines 104. Moreover, the wake
detection subsystem 120 may be configured to receive measurement data 112 from the
measurement devices 106 and store the data in the data repository 122. Additionally, the wake
detection subsystem may store historical wake condition data corresponding to wind farm 102 in
the data repository 122. Further, the wake detection subsystem 120 may also be
communicatively coupled to the controller or control subsystem 134.
[0028] In addition, the wake detection subsystem 120 may include one or more other units. In
a presently contemplated configuration, the wake detection subsystem 120 is depicted as
including a turbulence length scale (TLS) estimator sub-unit 128 and a TLS analytics sub-unit
130. In certain embodiments, the wake detection subsystem 120 and the sub-units 128 and 130
in particular may be implemented as software systems or computer instructions executable via
one or more processor sub-units 126 and stored in one or more memory sub-units 124. The subunits
128 and 130 may additionally or alternatively be stored and executed by other computing
devices such as a workstation, personal computer (PC), laptop, notebook, tablet, cell phone, and
the like. Further, the sub-units 128 and 130 may be implemented as hardware systems, for
example, via FPGAs, custom chips, integrated circuits (ICs), PIDs and the like.
[0029] The controller 134, as previously noted, may be operatively coupled to the wind farm
102. Further, the controller 134 may also be communicatively coupled to the wake detection
subsystem 120. The controller 134 is configured to control operations of the wind power
generation system 100. In one example, the controller 134 may control the operations of the
wind turbines 104 by controlling rotor speed and pitch set-points or by adjusting the yaw steering
of the wind turbines 104. Additionally, the controller 134 may control the operations of power
inverter circuitry, electrical filtering circuitry in the wind turbines 104, and the like.
[0030] Physical measurement of wind turbulence parameters at one or more downstream
turbines in real-time using appropriate instrumentation and techniques may verify the occurrence
of a wake at the turbines over a given interval of time or operating interval. As will be
appreciated from field observations and physics, the atmospheric turbulence experienced by a
wind turbine has different characteristics under wake and out of wake conditions. Furthermore,
characterization of the turbulence based on the spectral content of the turbulence in addition to
intensity of turbulence and/or wind direction may serve to enhance wake detection at the
downstream turbines.
[0031] In one type of graphical representation of the turbulence spectrum, the x-axis
represents the frequency of the turbulence and the y axis represents the power (for example,
magnitude squared) of the turbulence. Frequency spectrum parameters corresponding to a
downstream turbine in wake typically have a higher low-frequency content in the frequency
spectrum than that of the upstream turbine causing the wake.
[0032] With the foregoing in mind, it may be desirable to parameterize the frequency content
of the turbulence spectrum. One such parameterization of the turbulence spectrum is a
turbulence length scale. The turbulence length scale (TLS) is a derived quantity that describes a
size of large energy-containing eddies in a turbulent flow. In one embodiment, the TLS, in
addition to turbulence intensity and mean wind speed, captures the spectral content of the
turbulence. It may be noted that the TLS is estimated using quantities such as wind speed, wind
direction, and the like. These quantities are capable of being measured via use of physical
devices such as sensors, anemometers and the like. Moreover, measurement data corresponding
to these physical quantities may be retrieved from data repository 122. Also, the TLS is
calibrated in meters. A larger TLS typically implies higher power in the low-frequency region of
the frequency bandwidth of the turbulence. Hence, the larger TLS may be generally indicative
of a wake condition at the wind turbine. The estimation of the TLS of the turbulence
experienced by a wind turbine 104 is described in greater detail with respect to FIG. 3.
[0033] In accordance with aspects of the present specification, the wake detection subsystem
120 of the system 100 is configured to detect the occurrence of a wake condition at a wind
turbine 104 by tracking the wind speed measured at the wind turbine 104 by one or more
measuring devices 106 in real time. It may be noted that in the present specification, the wind
speed obtained from measurement devices 106 at a given wind turbine 104 may also be referred
to more simply as “wind speed data” or “wind speed measurement data” and represented by
reference numeral 112. The TLS estimator sub-unit 128 of the wake detection subsystem 120
estimates a TLS for each wind turbine in a set of interacting wind turbines based on the wind
speed measurement data 112. Additionally, the TLS analytics sub-unit 130 monitors the
turbulence length scales (TLSs) that have been estimated based on the wind speed measurement
data 112 corresponding to the set of interacting turbines to identify a convergence of the TLSs.
This convergence may be generally referred to as an initial convergence.
[0034] The TLS analytics sub-unit 130 also continuously tracks or monitors the estimated
TLSs of the set of interacting wind turbines for a subsequent persisting divergence. In one
embodiment, the TLS analytics sub-unit 130 is configured to compute a statistical metric
configured to characterize a relationship between the TLSs corresponding to the set of
interacting wind turbines. In one example, the TLS analytics sub-unit 130 is configured to
compute a statistical metric to characterize a relationship between the TLSs corresponding to a
first wind turbine and a second wind turbine of the set of interacting wind turbines, where the
first wind turbine is disposed upstream of the second wind turbine. Some examples of the
statistical metric include, but are not limited to, a difference between the turbulence length scales
of the interacting wind turbines, a ratio of the turbulence length scales of the interacting wind
turbines, and the like. Furthermore, the TLS analytics sub-unit 130 may process the statistical
metric via one or more analytic functions to generate a divergence quantifier. The divergence
quantifier is generally representative of a persisting divergence between the TLSs of the
interacting wind turbines over a determined time interval. Moreover, the TLS analytics sub-unit
130 is configured to detect a wake condition at one or more of the interacting wind turbines
based on a comparison of the divergence quantifier with a determined threshold.
[0035] Further, the system 100 may be configured to communicate the detected wake
condition to the system operator 132 and/or the controller 134. Subsequently, corrective
measures/actions may be deployed to reduce the impact of the detected wake condition on the
downstream wind turbines.
[0036] Turning now to FIG. 2, a flowchart 200 illustrating a method for detecting a wake
condition at one or more interacting wind turbines, in accordance with aspects of the present
specification, is presented. The method 200 is described with reference to the components of
FIG. 1.
[0037] In some embodiments, various steps of the method 200 of FIG. 2 may be performed by
one or more of the processor sub-unit 126 in conjunction with memory sub-unit 124, the TLS
estimator sub-unit 128, and the TLS analytics sub-unit 130. In certain embodiments, steps 202,
204 and 210-222 may be performed by the TLS analytics sub-unit 130 in conjunction with the
processor sub-unit 126 and step 206 may be similarly performed by the TLS estimator sub-unit
128. Furthermore, in one embodiment, steps 202 -228 of the method 200 may be performed
iteratively over a given operating interval. Moreover, in other embodiments, steps 202-228 of
the method 200 may be performed at definite times of a given operating interval.
[0038] It may be noted that flowchart 200 illustrates the main steps of the method to detect a
wake condition at one or more interacting wind turbines 104, and additional inputs and steps will
be described in greater detail with reference to FIGs 3 and 4.
[0039] The method 200 starts at step 202 where interacting wind turbines are identified for a
given operating interval. In one embodiment, the processor sub-unit 126 may obtain wind farm
layout data and wind direction data from the data repository 122. Using this data, the wind
turbines 104 may be grouped into one or more sets of interacting wind turbines by the TLS
analytics sub-unit 130.
[0040] At step 204, the TLS analytics sub-unit 130 retrieves at least one set of interacting
turbines identified at step 202. Additionally, the TLS analytics sub-unit 130 receives wind speed
measurement data 112 corresponding to that set of interacting turbines from the measurement
devices 106.
[0041] Furthermore, at step 206, one or more turbulence length scales 208 corresponding to
each member of the set of interacting wind turbines 104 is determined. Step 206 is
representative of one or more iterations of the method illustrated by flowchart 300 shown in FIG.
3 and may be performed by the TLS estimator sub-unit 128. The flowchart 300 will be described
in greater detail with respect to FIG. 3.
[0042] In some embodiments, steps 204 through 206 may be performed iteratively over the
operating interval to continuously update the estimated turbulence length scales. Consequent to
step 206, the turbulence length scales 208 corresponding to the interacting wind turbines 104 are
determined.
[0043] In some embodiments, the estimated turbulence length scale of the turbulence
experienced by a wind turbine 104 may be determined from one or more real-time measurements
such as mean wind speed, wind direction, and the like. Accordingly, the measurements may be
obtained or sampled by the measurement devices 106 at regular time intervals, for example,
every minute, and communicated to the wake detection subsystem 120. Also, in one
embodiment, the mean wind speed may be an input parameter used in determining the TLS of
the turbulence at a wind turbine by the TLS estimator sub-unit 128. In one example, the mean
component of the wind speed may be a ten-minute average of the aforementioned measurements
that are measured at every minute of the time interval.
[0044] Subsequently, the turbulence length scales corresponding to the interacting wind
turbines 104 are monitored by the TLS analytics sub-unit 130 to detect an occurrence of
convergence, as indicated by step 210. As previously noted, this convergence may generally be
representative of an initial convergence. A convergence occurs when the turbulence length scale
estimates corresponding to the set of interacting wind turbines converge to a close range of
values. The convergence of the turbulence length scales will be described in greater detail with
respect to FIG. 4. It may be noted that a convergence of the turbulence length scales 208 of at
least two members of the set of interacting wind turbines 104, where a first wind turbine is
located upstream of a second wind turbine with respect to the prevailing wind direction 108,
indicates that the first and second wind turbines 104 are experiencing the same mix of freestream
wind and turbulence components with a possibility of a wake occurrence.
[0045] Accordingly, in one embodiment, the occurrence of the convergence of the turbulence
length scales may be determined in accordance with equation (1).
0.98 ?? ??
????????
????????
?? ?? 1.2 ??1??
where convergence may be said to occur if equation (1) is satisfied.
[0046] In certain other embodiments, the occurrence of the convergence of the turbulence
length scales may be determined in accordance with equation (2).
????|???????? ?? ??????????|
????
?? ???????????????????????????????????????? ??2??
where TLS1 and TLS2 in equations (1) and (2) are generally representative of the turbulence
length scales of the first and second wind turbines 104 and ???????????????????????????????????????? is generally
indicative of a close range of values, for example, 1 m/min.
[0047] Accordingly, at step 212, a check is carried out by the TLS analytics sub-unit 130 to
determine if a convergence of the turbulence length scales 208 has occurred. At step 212, if it is
determined that the convergence has occurred, control passes to step 214. However, if no
convergence is detected at step 212, control passes back to step 210.
[0048] Subsequently, at step 214, the converged turbulence length scales 208 are monitored
by the TLS analytics sub-unit 130 for occurrence of a subsequent divergence. It may be noted
that a divergence of a previously converged set of turbulence length scales 208 is indicative of a
scenario where at least one wind turbine of the interacting wind turbines 104 experiences a wake
condition. The subsequent divergence of the turbulence length scales will be described in greater
detail with respect to FIG. 4. In certain embodiments, the converged turbulence length scales
208 may be monitored for occurrence of a persisting divergence or divergent trend over the
sample interval such as a ten-minute interval. It may be noted that the persisting divergence or
divergent trend, is generally a more dependable indicator of a wake condition being experienced
by at least one wind turbine of the interacting wind turbines 104 when compared to a divergent
trend corresponding to a shorter time interval.
Accordingly, at step 216, a check is carried by the TLS analytics sub-unit 130 to determine if a
divergence has occurred. In one embodiment, the occurrence of a divergence of the turbulence
length scales of a first and second wind turbine of the set of interacting turbines may be
determined in accordance with equation (3).
??
????????
????????
?? ?? 1.2 ??3??
where divergence may be said to occur if equation (3) is satisfied over a given time interval, for
example, ten minutes.
[0049] In certain other embodiments, the occurrence of a divergence of the turbulence length
scales of a first and second wind turbine 104 of the set of interacting wind turbines may be
determined in accordance with equation (4).
????|???????? ?? ??????????|
????
?? ?????????????????????????????????????? ??4??
[0050] The terms TLS1 and TLS2 in equations (3) and (4) are generally representative of the
turbulence length scales of the first and second wind turbines and ?????????????????????????????????????? is
generally indicative of a close range of values, for example, 10 m/min over a given time interval
of ten minutes. If it is determined that the divergence has occurred, control passes to step 218.
However, at step 216, if it is determined that the divergence has not occurred, control passes
back to step 214.
[0051] Referring now to step 218, a statistical metric configured to characterize a relationship
between the TLSs 208 corresponding to the first and the second wind turbine of the set of
interacting wind turbines is computed. Also, in one embodiment, the statistical metric may be
the ratio
????????
????????
of a first TLS TLS1 corresponding to a first, upstream wind turbine and a second
TLS TLS2 corresponding to a second, downstream wind turbine from the set of interacting wind
turbines 104. Additionally, in another embodiment, the statistical metric may be representative
of a difference of the TLSs 208 corresponding to the first and second wind turbines from the set
of interacting wind turbines 104 as represented in equation (5).
????|???????? ?? ??????????|
????
??5??
[0052] Furthermore, at step 219, a divergence quantifier is generated by processing the
statistical metric via one or more analytic functions. Some examples of the analytic functions
include, but are not limited to, a slope function, a mean average function, a standard deviation, a
logistics function, or combinations thereof. As previously noted, the divergence quantifier is
generally representative of a persisting divergence between the TLSs of the interacting wind
turbines such as the first and second wind turbines over a determined time interval. In one
example, the TLS analytics sub-unit 130 unit is employed to compute the statistical metric and
the divergence quantifier.
[0053] Subsequently, at step 220, the divergence quantifier is compared with a determined
threshold. The threshold may include a design parameter that is determined by offline data
analysis of historical wake condition data corresponding to wind farm 102, in one example.
Moreover, the threshold value or range of values may be retrieved from data repository 122. At
step 220, if it is determined that the divergence quantifier is greater than the determined
threshold, occurrence of a wake condition 222 is confirmed and control passes to step 224.
[0054] At step 224, operational settings for the wind turbines 104 in wake and the
corresponding upstream wind turbines 104 may be re-calibrated in response to the wake
condition 222 to reduce the impact of the detected wake condition 222 on the downstream wind
turbines. As previously noted, the operational settings may include settings for rotor speed,
pitch, yaw steering, and/or combinations and variations thereof. Furthermore, at step 224, data
corresponding to the identified wake condition 222 is used to determine the operational settings
for the upstream and downstream wind turbines. In addition, adjustments to the operational
settings may be calibrated based on one or more of the statistical metric, mean wind speed, wake
pair models, and combinations and/or variations thereof. Control may then be passed to step
226.
[0055] However, at step 220, if it is determined that the divergence quantifier is less than or
equal to the determined threshold, control is passed to step 228. This state where the divergence
quantifier is less than or equal to the determined threshold is generally indicative of one of two
conditions. In one condition, the divergence quantifier may be representative of a state where
the interacting wind turbines are going out of wake. Accordingly, the interacting wind turbines
may experience free-stream wind. In this example, operational settings of the for the interacting
wind turbines 104 may be correspondingly reset or re-calibrated to reflect the state of the
interacting wind turbines experiencing free-stream wind. In another condition, the divergence
quantifier may be representative of a state where the interacting wind turbines are not
experiencing a wake condition. In this example, no adjustment to operational settings of the
wind turbines based on a wake condition may be needed. However, in some examples, it may be
desirable to adjust certain operational settings based on other non-wake-related conditions of the
wind turbines. Control may then be passed to step 226.
[0056] At step 226, the operational settings may be communicated to the system operator 132,
in one example. Additionally, at step 226, the operation of the interacting wind turbines 104
may be altered in accordance with the re-calibrated settings via use of the controller 134. In
certain embodiments, control may be passed back to step 204 for one or more iterations of the
method 200 for the remainder of the current operating interval. In certain embodiments, the TLS
analytics sub-unit 130 may be used to perform step 220-228.
[0057] The turbulence length scale estimation at step 206 that is performed by the TLS
estimator sub-unit 128 is presented in FIG. 3. Turning now to FIG. 3, a flowchart 300 for
determining a turbulence length scale of turbulence at a wind turbine, in accordance with aspects
of the present specification, is presented. The method 300 will be described with reference to the
components of FIGs. 1 and 2.
[0058] In some embodiments, various steps of the method 300 of FIG. 3 may be performed by
one or more of the processor sub-unit 126 in conjunction with memory sub-unit 124 and the TLS
estimator sub-unit 128. More particularly, steps 302 to 310 of the method 300 may be performed
by the TLS estimator sub-unit 128. In one embodiment, the TLS estimator sub-unit 128 may
iteratively estimate the turbulence length scales at each wind turbine of the set of interacting
wind turbines by performing steps 302 to 310 for each wind turbine. Moreover, the TLS
estimator sub-unit 128 may perform step 308 via a non-linear estimator such as an Extended
Kalman Filter, an unscented Kalman filter, a recursive least squares estimator, a particle filter, or
combinations thereof.
[0059] At step 302, a state vector ???????? that models turbulence at a wind turbine 104 is defined
in accordance with equation (6).
???????? ?? ???????????? ?????????? ?????????? ??6??
where ?????????? is a state characterizing a mean component of wind speed, ???? ?????? is a state
characterizing a turbulence component of the wind speed, and ???????? is a turbulence length scale.
[0060] In one embodiment, ???????? may be set to an initial value. Also, in certain embodiments,
the initial value of ???????? may be determined using a heuristic based on a turbulence spectrum
analysis of historical turbulence data of the wind farm 102.
[0061] Subsequently, at step 304, a process state model is defined for the state vector. In one
example, the process model models dynamics of the state vector. Further, the process state
model is based on state changes in one or more of the mean wind speed component, the
turbulence component, and the turbulence length scale. In one embodiment, these changes may
be characterized via differential equations corresponding to state functions of the respective
components. As will be appreciated, differential equations may typically be used to model state
changes of processes in continuous-time. Accordingly, the process state model may be defined
in conformance with equations (7), (8), and (9).
?????? ?? ?? ??
1
??
?? ???? ?? ???? ??7??
?????? ?? ??
??????
2??
???? ?? ???? ??8??
???? ?? ???? ??9??
[0062] Referring to equation (7), ?????? is a derivative of a state function ?????????? that
characterizes the mean wind speed component of equation (3). Further, ?? is a time constant of
the mean wind speed component ??????????, and ???? is the zero mean Gaussian white noise process
value for the mean wind speed component ??????????.
[0063] Turning now to equation (8), ?????? is a derivative of a state function ???? ?????? that
characterizes the turbulence component of equation (6). In one embodiment, equation (8) may
be derived from the Kaimal spectrum model for wind turbulence. Moreover, L is the turbulence
length scale, ?????????? is the state function characterizing the mean wind speed component from
equation (6) and ???? is the zero mean Gaussian white noise process value of the turbulence
component.
[0064] In equation (9), ???? is the derivative of the turbulence length scale ???????? of equation (6).
In an embodiment, ???? is modeled as a random walk or slowly varying drift of the turbulence at the
wind turbine 104. Moreover, ???? is the zero mean Gaussian white noise process value for the
turbulence length scale. It may be noted that in certain embodiments, in equations (7) - (9), the
zero mean Gaussian white noise process values ????, ???? and ???? may include standard deviations as
design parameters.
[0065] Moreover, in certain embodiments, the process state model of the state vector
???????? may characterize the changes in one or more of the mean wind speed component, the wind
turbulence component, and the turbulence length scale via difference equations. As will be
appreciated, difference equations may typically be used to model state changes of processes in
discrete-time.
[0066] Turning now to step 306, a measurement model for wind speed at a wind turbine 104
is defined, based on a mean wind speed component and a wind turbulence component. In one
example, the mean wind speed component and the wind turbulence component may be obtained
from wind speed measurement data 112. Also, in one embodiment, the measurement model may
be defined in accordance with equation (10).
?????????????????????? ?? ???? ?? ???? ??10??
[0067] In equation (10), ?????????????????????? is a measurement model that characterizes the wind
speed measurement data 112 where ???? is the mean component of the measured wind speed and
???? is the turbulence component of the measured wind speed. Additionally, a state vector ???? may
be defined based on the measurement model at time instant ‘k’ in accordance with equation (11).
???? ?? ???????????????????????????? ??11??
wherein ???? is the state vector at time instant ‘k’ and ???????????????????????????? is the measurement model
of equation (10) at time instant ‘k’.
[0068] Subsequently, at step 308, a turbulence length scale corresponding to a wind turbine
104 for a given operating interval is estimated based on the process state model and the
measurement model. In one embodiment, the turbulence length scale may be estimated via use
of a non-linear estimator. Some non-limiting examples of the non-linear estimator include an
Extended Kalman Filter (EKF), an unscented Kalman filter, a recursive least squares estimator, a
particle filter, or combinations thereof. As previously noted, the process state model may be
defined in conformance with equations (7), (8), and (9) and the measurement model may be
defined in accordance with equation (10). In one embodiment, step 308 may be performed by
the TLS estimator sub-unit 128. Reference numeral 310 is generally representative of the
estimated turbulence length scale. It may be noted that the estimated turbulence length scale is
also represented by reference numeral 208 in FIG. 2. Furthermore, at step 312, the estimated
turbulence length scale 310 may be communicated to the TLS analytics sub-unit 130.
[0069] As previously noted, the initial convergence of the turbulence length scales of at least
a first wind turbine and a second wind turbine 104 of the set of interacting wind turbines to a
close range of values indicates that the first and second wind turbines 104 are experiencing the
same mix of free-stream wind and turbulence components with a possibility of a wake
occurrence. The turbulence length scales may subsequently diverge over a given operating
interval. A persisting divergence of the turbulence length scales of the first and second wind
turbines may indicate that the wind turbine having the larger turbulence length scale may be
going into wake.
[0070] With the foregoing in mind, a graphical representation 400 of turbulence length scale
values corresponding to a set of interacting wind turbines is presented in FIG. 4A. Also, FIG. 4B
is a graphical representation 420 of mean wind speed curves of the interacting wind turbines
corresponding to the turbulence length scale values of FIG. 4A over the same operating interval.
More particularly, FIGs. 4A-4B illustrate an exemplary correlation of wake condition detection
and mean wind speed, in accordance with aspects of the present specification. FIGs. 4A-4B are
described with reference to the components of FIGs. 1-2.
[0071] Graph 402 of FIG. 4A is a wind speed curve over a given operating interval. The yaxis
404 is representative of the wind speed measurement data 112 sampled by measurement
devices 106 at one or more interacting wind turbines 104, in meters per second. The x-axis 406
is representative of time stamps of the measurements collected over the given operating interval,
for example, every hour. Reference labels WA1, WA2 and WA3 are representative of windspeed
curves of the wind turbines A1, A2 and A3 in the wind farm 102. As previously noted, the
wind turbines A1, A2 and A3 are a set of interacting wind turbines.
[0072] Reference numeral 408 is indicative of a data point in the graph 402 where the value
of the wind speed measurement data 112 corresponding to the wind speed curve WA2 is lower
that the value of the wind speed measurement data 112 corresponding to the wind speed curves
WA1 and WA3. In a similar fashion, reference numeral 410 is representative of a data point in
the graph 402 where the value of the wind speed measurement data 112 corresponding to the
wind speed curve WA2 is in the same range as the values of the wind speed measurement data
112 corresponding to the wind speed curves WA1 and WA3. The data point 408 may be
representative of a point where wind turbine A2 is going into wake. Similarly, the data point 410
may be considered to be the point where the wind turbine A2 is going out of wake.
[0073] Referring now to graph 420, the x-axis 422 is representative of a time of an operating
interval at which various values of the turbulence length scales of the turbulence experienced at
wind turbines A1, A2 and A3 are computed. The y-axis 424 is representative of a turbulence
length scales in meters. The turbulence length scale curves TA1, TA2, and TA3 are
representative of turbulence length scales computed for the wind turbines A1, A2, and A3 of the
wind farm 102 of FIG. 1.
[0074] With continuing reference to the graph 420, reference numeral 430 is representative of
a data point in the graph where the turbulence length scales TA1, TA2 and TA3 achieve an initial
convergence. This initial convergence may be generally indicative of a state wherein the wind
turbines A1, A2 and A3 are experiencing the same mix of free-stream wind and turbulence
components with a possibility of a wake occurrence. Furthermore, reference numeral 426 is
representative of a data point in the graph 420 where the wind turbine A2 is going into wake,
where values of the turbulence length scale curve TA2 are transitioning to a higher range of
values in comparison to values of the turbulence length scale curves TA1 and TA3. Similarly,
reference numeral 428 is representative of a data point in the graph 402 where values of the
turbulence length scale curve TA2 are moving back into the range of the values of the turbulence
length scale curves TA1 and TA3. The data point 426 may be representative of a point where
wind turbine A2 is going into wake. Similarly, the data point 428 may be considered to be the
point where the wind turbine A2 is going out of wake. As clearly illustrated by FIGs. 4A-4B,
there is an observable correlation between the data points 408 and 426 and between the data
points 410 and 428.
[0075] The systems and methods for online wake detection presented hereinabove provide an
enhanced detection of a wake in one or more wind turbines in a wind farm. Advantageously, the
systems and methods estimate turbulence length scales using real-time measurement data
gathered at each wind turbine. Furthermore, a mean upward trend in a turbulence length scale
estimate indicates that a turbine is entering wake, and in a similar fashion, a mean downward
trend indicates that the turbine is exiting wake. Also, the estimation of the turbulence length
scales is performed online and in real-time or near real-time, thereby circumventing the need for
a system operator to monitor the wind farm and the need for any training data. Additionally, the
upward or downward trends are compared to a threshold to confirm or verify the wake condition,
therefore any over-correction of operating settings or set-points may be avoided in the event of
false positives and minor variations.
[0076] Moreover, the systems and methods may be implemented/retrofit for an existing wind
farm, thereby leveraging the existing wind farm communication network. This arrangement
obviates the need for any additional hardware and/or expenditure. The systems and methods
facilitate computationally simple wake detection in real-time using existing sensors on the wind
turbines. Furthermore, the systems and methods are self-learning when implemented for any
wind farm. In particular, these systems and methods obviate the need for any significant
engineering effort for configuration/tuning of wind farm wake models, running offline
optimization, and the like.
[0077] It is to be understood that not necessarily all such objects or advantages described
above may be achieved in accordance with any particular embodiment. Thus, for example, those
skilled in the art will recognize that the systems and techniques described herein may be
embodied or carried out in a manner that achieves or improves one advantage or group of
advantages as taught herein without necessarily achieving other objects or advantages as may be
taught or suggested herein.
[0078] While the technology has been described in detail in connection with only a limited
number of embodiments, it should be readily understood that the specification is not limited to
such disclosed embodiments. Rather, the technology can be modified to incorporate any number
of variations, alterations, substitutions or equivalent arrangements not heretofore described, but
which are commensurate with the spirit and scope of the claims. Additionally, while various
embodiments of the technology have been described, it is to be understood that aspects of the
specification may include only some of the described embodiments. Accordingly, the
specification is not to be seen as limited by the foregoing description, but is only limited by the
scope of the appended claims.