Patent Application
20110158806
This application is related to the following publications, all of which are incorporated herein by reference:
This application is also related to the following patents and patent applications, all of which are incorporated herein by reference:
This application is also related to the following reference publications, all of which are incorporated herein by reference:
This patent application generally relates to a system for monitoring a wind turbine. More particularly it relates to a system for monitoring forces and moments on blades and tower of the wind turbine. It also relates to an energy harvesting system for providing power for uses such as monitoring and transmitting data.
The vast majority of the wind turbines on the market are composed of horizontal axis, upwind, three-bladed wind turbine designs, as shown in
However, systems for measuring forces on the blades that feed control system 46 have not been adequate, and the present patent application provides improvement.
One aspect of the present patent application is a turbine that includes a turbine blade, a plurality of sensors, a wireless sensor module, a data aggregator, and a blade pitch control unit. The plurality of sensors are distributed in a plurality of locations on the turbine blade suitable for determining a moment of the turbine blade. The sensor module is configured to transmit data to the data aggregator to determine the moment. The data aggregator is configured to provide an output to the blade pitch control unit. The blade pitch control unit is configured to adjust pitch of the turbine blade based on the moment.
a-2b are three dimensional views of prior art wind turbine blades showing the blade root with bolts extending for attaching the blade to a hub;
a is a front view of a wind turbine mounted on a tower illustrating how gravity loads are used to calibrate the load sensors under conditions when wind loads are minimal;
b is a cross sectional view of a blade of the wind turbine of
a-10c are graphs showing the flap and edgewise gravity induced blade moments as a function of blade rotation position for three different pitch angles of the blade;
a is a block diagram showing the components of the wireless bolt tension and compression sensing node that is included in each of the smart fasteners;
b is a block diagram showing the components of the wireless data aggregator node;
c is a block diagram showing the components of the wireless vibration sensing node that may also be included in each of the smart fasteners;
d is a block diagram of an energy harvesting, wireless system including components of a wireless sensor data aggregator (WSDA);
a is a top view of an embodiment of an instrumented load-sensor bolt smart fastener showing the instrumentation housing;
b is a cross sectional view of an embodiment of an instrumented load-sensor bolt smart fastener showing the location of the sensors in the pin region of the bolt and the instrumentation housing with its wire connectors, printed circuit boards, battery, and antenna;
c is a three dimensional view of an embodiment of an instrumented load-sensor bolt smart fastener showing the location of the sensors in the pin region of the bolt for a full bridge strain gauge set-up with two strain gauges on each side of the bore in the bolt;
c is a cross sectional view of the embodiment shown in
a is an exploded view of an embodiment of an energy harvesting instrumented load-sensor with a flexure mounted for receiving vibrational energy through a base, a printed circuit board and energy storage units for receiving and storing the energy;
b is an exploded view of the mounting portion of the energy harvesting instrumented load-sensor of
a is an exploded view of an embodiment of an energy harvesting unit with magnets that can move within a coil to generate electricity from vibrational energy in a shaft to which the energy harvesting unit is mounted;
b is a three dimensional view of the assembled vibration energy harvester of
a is a three dimensional view of another embodiment of a vibration energy harvester configured in the size and shape of a battery;
b is a three dimensional cross sectional view of the embodiment of a vibration energy harvester that uses magnets vibrating in coils to convert vibrational energy into electricity;
c is a schematic diagram of the magnetic flux lines of two magnets of
a is a block diagram showing the components of an energy harvesting circuit for receiving energy from an energy harvesting unit, converting to a useful voltage, storing the energy, and providing it to an application load;
b is a block diagram of a WSDA and timing control for its associated hard-wired and wireless sensor nodes;
a-24b are cross sectional views of an embodiment of an instrumented load-sensor bolt smart fastener showing a DVRT in the pin region of the bolt and the instrumentation housing with its wire connectors, printed circuit boards, battery, and antenna;
a is a front view of a wind turbine with wireless strain gauges mounted inside the tower; and
b is a side view of the wind turbine of
The present applicants recognized that the aerodynamic forces generated by moving blade 30 result in two moments at blade root 34 that are reacted by hub 36—a flapping moment generated by lift forces normal to the rotor plane, and an edgewise or lead-lag moment generated by aerodynamic forces parallel to the rotor plane, as shown in
Monitoring the real time status of the two blade moments can yield a great deal of useful information, especially combined with other turbine state data, such as the blade rotation angle, rotation rate, collective pitch setting, and wind velocity. Some of the operating characteristics that can be inferred from direct blade moments include:
The present applicants found that they could obtain a direct measurement of forces produced by a blade by using smart fasteners, such as instrumented load-sensor bolts. Instrumented pins, studs, or other smart fasteners can also be used. Strain sensors applied to the interior surface of the blades could also be used. The measured forces could then be used to determine blade moments. Applicants recognized that this directly measured determination of blade moments was superior to calculating moments based on other measurable parameters, such as the average wind velocity.
In this application a phrase such as “the sensor is on the turbine blade” means that the sensor is mounted to any portion of the blade, including an interior surface. In addition the phrase “the sensor is on the turbine blade” includes the sensor being mounted on a device that is itself mounted on the blade, such as a bolt that holds the blade to a hub.
Applicants also recognized that the improved measurement of operating loads on the blade and better load control allows designers the freedom to operate each of the blades closer to its limits, for example at higher wind conditions. This results in reduced electrical generation cost because more electricity can be generated from a given investment in equipment (electrical generation cost is largely amortization of the original equipment cost). Furthermore, improved measurement of applied blade load and better load control leads to reduced scheduled and unscheduled maintenance and increased wind turbine reliability. Both of these lead to decreased cost of electrical generation.
In addition, by monitoring the change in blade moments over time at different operating conditions, the present applicants could also infer:
Additional sensors, such as accelerometers, can be used to detect abnormal vibration in the gear box that may indicate an abnormal condition, such as a broken tooth. These sensors are synchronized, as described herein below, with sensors measuring loads on the blades so the load conditions on the rotor can be related to conditions measured on the gear box.
In one embodiment, at least three smart fasteners, such as instrumented load-sensor bolts 50, are used to measure moments blade 30 is applying to blade root 34, as shown in
Referring to
F1, F2, F3 Axial forces in the three bolts
R Radius of the bolt circle
θ1, θ2, θ3 Angular position of the instrumented bolts (θ1=0 for convenience)
Fc Blade centrifugal force
Mf Blade flapping moment
Me Blade edgewise moment
It can be shown that
Where:
a=—R cos θ2
b=—R sin θ2
c=—R cos θ3
d=—R sin θ3
In the general case of 4 instrumented load-sensor bolts 50 oriented at θ1, θ2, θ3, and θ4, the equation to be solved is:
Since measurement errors in the load sensors F1 . . . Fn, are randomly distributed about a zero mean, the more load sensors in the aggregate measurement, the greater the self-canceling effect of these errors, and the better the measurement accuracy of the moments of interest.
In any bolted joint, there is load sharing between the bolt itself and the contact interface. In the current wind turbine application, the contact interface is where surface 52 of blade flange 42 at root 34 of blade 30 contacts hub flange 40 on hub 36. Measuring only the load in instrumented load-sensor bolt 50 does not provide a complete estimate of the load in the bolted joint overall. The load in the bolt will be some fraction of the overall joint load. The value of this fraction will depend on the relative stiffness of the bolt in comparison to that of the contact interface. This fraction can be determined through analysis, for example using finite element analysis techniques. It can also be taken into account through calibration. In the above descriptions, the applicants recognized that a correction for this load sharing is made to compute the desired aerodynamic blade loading. In particular, each of the bolt loads, Fi, is scaled by the inverse of its load sharing fraction as determined by the calibration coefficients or by the finite element analysis.
The solutions defined above assume that the bolted connection between blade 30 and hub 36 consist exclusively of instrumented load sensor bolts 50. In one embodiment only a small fraction of the bolt attachments are instrumented load sensor bolts. The non-instrumented bolts 32 share in the load transfer between blade 30 and hub 36. Therefore, the equations noted above apply only for that portion of the loads being carried by instrumented load-sensor bolts 50. The measured centrifugal load determined above is proportional to the total centrifugal load according to the ratio of the number of instrumented load-sensor bolts 50 to the total number of bolts 32.
FC=Total centrifugal load
Fc=Centrifugal load carried by bolt sensors per Eq. 2
N=Total number of bolts
n=Number of sensor bolts
To determine the total flapwise and edgewise moment, it is first necessary to solve for all the bolts' loads. It can be shown that:
F
i
=A*sin(θi+β)+K where Eq. (4)
Fi=bolt load of the ith bolt
θi=Orientation of the ith bolt
β=phase angle
K=Mean of F1, F2, F3 measured instantaneous AXIAL sensor loads
Once all bolt loads are determined, the total flapwise and edgewise moment arising from the bolts can be determined:
MF=Total flapping moment
Fi, θi=Bolt loads and positions for i=1 to N (total number of bolts)
Similarly, for the edgewise moment:
ME=Total edgewise moment
Retaining bolts 32 are bolted into hub 36 with a preload. In one embodiment this preload is measured in a calibration step, and calibration coefficients so determined are used to properly measure blade operating load. The bolt forces calculated above are those forces in the bolts above and beyond those created by preloading the bolts upon installation. Furthermore, the 1 g gravity loads on the blade result in significant root moments as well, and calibration is used to properly isolate the blade aerodynamic loads in operation. Calibration can be performed as described below, with reference to
For a given blade rotation position θ and pitch angle φ, it can be shown that:
M
f
=Wd(sin θ cos φ+b cos θ) Eq. (9)
M
e
=Wd(−sin θ sin φ+a cos θ) Eq. (10)
Note that for the sake of simplicity, the blade coning angle and the shaft tilt angle have been assumed to be zero. These equations can be modified to account for these effects.
As shown in
To calibrate the load sensors for the effect of preload and 1 G gravity loads: in one embodiment instrumented load-sensor bolts 50 are factory calibrated before installation, and calibration coefficients found in the factory calibration are used directly to determine the preload applied to each bolt upon installation. These values are stored.
To measure 1 G gravity loads, sensor measurements are taken when no aerodynamic loads are present (i.e. when the wind velocity is zero or some very small value (perhaps 1 knot). Applicants recognized that since aerodynamic loads are proportional to the square of the wind velocity, loads at 1 knot wind speed are about 0.44% of the loads at 15 knots.
To determine 1 G gravity loads as per
In operation (i.e. in the presence of wind and with the rotor turning), the sensor forces are measured as shown in
Applicants recognized that individual conventional bolts 32 may be removed after blade 30 has been bolted in place. In one embodiment, conventional bolts 32 are replaced with instrumented load-sensor bolts 50 without removing the rotor and blades 30 from tower 60, thus avoiding a disassembling and reassembling operation for previously assembled wind turbines. Applicants recognized that this retrofit capability can therefore save time and money. In one embodiment instrumented load-sensor bolts 50 are recalibrated after installation to take into account preload, gravity, and joint load sharing as described above.
In one embodiment, the load sensor outputs of instrumented load-sensor bolts 50 are processed in processor 62 and transmitted to data aggregator 90 for computation of flapping moments Mflp, as described herein above and in equations 5-8. These moments are then used by data aggregator 90 to determine blade servo pitch commands for each blade 30, as illustrated in the block diagram in
The rotating flapping moments are converted into the stationary d-q (direct and quadrature) axes using the “d-q axis transformation”, which permits classical SISO (Single Input Single Output) control theory to be applied, as described in References 1-5.
Blade pitch control is used for two purposes. First, adjusting pitch enables extracting energy more efficiently from the prevailing winds. This improves the “capacity factor” (i.e. the fraction of the full electrical generating capacity of the turbine that it actually produces given variable ambient winds. Higher capacity factor and higher AEP (annual energy production) reduces the COE (cost of energy) produced, since the majority of this cost is amortization of the fixed turbine capital expense over the total energy produced.
A second purpose of load-based blade pitch control is to permit reduced safety factors in the design of the blade, which leads to lower blade weight and cost. In the absence of a load measurement and active control scheme, blades are designed using more conservative design margins to account for gusts and other randomly distributed wind loads. With load monitoring and control, these margins can be reduced because the system actively manages blade loads. Rotor blade costs are proportional to weight, so more structurally efficient (i.e. lighter) blades reduce the turbine capital cost and the COE.
In one embodiment the wireless bolt tensioning and compression sensing node 64 within each instrumented load-sensor bolts 50 each includes circuit board 66 with components as shown in
Data is transmitted over a wired or wireless link such as RS-232 port 86 or RF-transceiver 88 to wireless sensor data aggregator node (WSDA) 90 that includes transceiver 92 or USB or RS 232 port 94, as shown in
Use of the FRAM is particularly useful for applications that use high speed sampling. A high sample rate (HSR) node was designed by MicroStrain, Inc. to log data at 50 kilosamples per second (kSPS) with 12-bit A/D converter resolution. The HSR node can be configured to sample a full differential Wheatstone bridge input (strain gauges, accelerometers, pressure sensors, etc.) or a single-ended 0-3 volt input. Each high rate logging event may consist of 125,000 samples, or ˜2.25 seconds of record time when sampled at 50 kSPS. The node can store up to 1 million samples on its embedded, non-volatile memory chip, as shown in
The HSR node measures only 31×31×5 mm including mounting tabs and operates from a 3 volt DC supply. When logging data continuously at 50 kSPS, average current consumption was measured at 8.8 mA (26.4 mW). When actively transmitting digital data from non-volatile memory (along with framing and checksum bytes), the average current consumption was 25 mA (75 mW). When in sleep mode, current consumption was only 20 microamperes (0.060 mW). To test performance, a signal generator was used to produce a 1 kHz sine wave input (reference) signal, and this reference waveform was sampled using the HSR node's single ended input.
In one embodiment, the instrumented axial load sensor bolts are very similar to Microstrain's shear bolt sensor, as described in the '111 application. However, whereas the shear bolt strain gages of the '111 application are located and oriented to measure shear load, for connecting wind turbine blades 30 to hub 36 the primary load is axial tension. Because bolts 32 project through clearance holes 38 in hub 36 the shear load in bolts 32 is theoretically zero. In some cases, however, there is contact between bolt 32 and the inner bore of hub clearance hole 38, and some shear load may be introduced in bolt 32. But this shear load does not enter into the equations for calculating the two moments.
Therefore, in this embodiment, strain gauges 110 are oriented along the bolt/bolt axis to measure the axial load, as shown in
The present inventors recognized that by using FRAM for non-volatile memory and by changing their A/D converter to one such as the Texas Instruments ADS8319 they could enhance the sampling rate significantly, at least to 50 ksamples/S.
In one embodiment sensor node 64 is independently powered with long life batteries, as shown in
Various embodiments of flexures 130 are described herein. Each of the types of flexures can be made from a material such as superelastic NiTi (nickel titanium) to allow for large strains without damage. To adjust the resonant frequency up or down the flexure can be made thicker or thinner. Each of the flexures can have stops to protect it from overloads.
In one embodiment, flexure 130 is a cantilever beam energy harvesting device, as described in the '117 application, mounted on a vibrating surface. In another embodiment base 140 of energy harvesting device 142 of
In certain wind conditions the wind turbine can start to shake or oscillate at an amplitude that can cause damaging strains to its structural elements. The wireless accelerometer sensing node shown in
The piezoelectric material and rectifier/energy management circuit with energy storage that is part of the wireless bolt tension and compressing sensing circuit or that is part of the wireless vibration sensing node provides power for operating the sensor and its electronic support circuit, including its transceiver.
With any of the energy harvesting devices, the mass, dimensions and stiffness of the flexure are selected so the natural frequency of vibration of the flexure is tuned to the predominant frequency produced by the wind turbine surface to which it is mounted. Stiffness in the embodiment of
For installation, flexure 130 is compressed, opening the space between the ends where PZT stack 162 will be installed. Then when PZT stack 162 is installed and the pressure is released, PZT stack 162 is held under a residual compressive stress. During installation, PZT stack 162 is bonded at only one end so PZT stack 162 is protected from overloading while mass 164 is vibrating. Glue, such as Loctite, is provided on base screws 166a and housing screws 166b to keep them from loosening while flexure 130 is vibrating.
The structure of
In one test, energy harvesting device 142 was excited by vibrations that reproduced those measured from a helicopter main gear box. For the test, output of energy harvesting device 142 was collected in a capacitive energy storage circuit, and “consumed” by a fixed load resistor. The absolute and normalized power output values for energy harvesting device 142 are shown in Tables I and II. These data provide a guide to design of the harvester for a given high sample rate vibration monitoring application, given the power consumption specifications as provided in the previous section. For example, assuming a vibration harvester of 4.3 cubic centimeters (cc) and 38.5 grams is suitable for the application, we can expect this harvester to generate 37 milliwatts continuously when mounted on this particular Helicopter Gear Box.
The duty cycle of sensing node 64 can be adjusted, or the harvester can be re-sized, to meet the needs of the particular monitoring application. The output of our vibration harvesters increases approximately linearly with both mass and volume. The power consumed by the sensor node decreases significantly as the interval between samples is increased. The calculated average power consumption of our wireless sensor nodes 64 for a range of sample intervals are provided in the next section.
High Sample Rate Energy Consumption w/Duty Cycling
A high sample rate node with an accelerometer is particularly useful for the detecting problems in the gear box of the wind turbine where a gear or output shaft is spinning at a high rate or is subject to high frequency vibration. The high sample rate node can also be energy harvesting, as shown in
The high sample rate node, as well as the other nodes in the network, will support scheduled data logging according to programmable time intervals (duty cycles), along with periodic RF transmission. Shorter duration sampling duty cycles will result in longer sleep durations, as well as reduced “on” time for radio transmissions. Therefore, periodic time interval sampling or “duty cycling” can greatly reduce the average power consumption, as we illustrate in the following example.
In this example, 50 kSPS data are collected for a relatively short (1 second) duration, or “burst”, at a time interval of once per minute (60 seconds). The time duration required for wireless data transmission can be calculated by dividing the number of samples recorded by the wireless data transmission rate. Digital wireless data transmission, using an efficient 802.15.4 protocol with framing and error checking, can be accomplished at a rate of 4000 SPS. This information, along with power consumption for various operational modes, allows the average power consumption to be calculated by the following expression:
Average power consumed=(data logging duty cycle)*(power consumed when logging)+(wireless data transmission duty cycle)*(power consumed when transmitting)+(sleep duty cycle)*(power consumed when sleeping).
Where the power consumed when logging=26.4 mW; transmitting=75.0 mW; and when sleeping=0.060 mW. And where the data logging duty cycle=1 sec/60 sec=1.67%; the wireless data transmission duty cycle=12.5 sec/60 sec=20.83%; and the sleep duty cycle=77.5%.
Therefore, the average power consumed=1.67%*26.4+20.83%*75 mW+77.5%*0.060 mW=16.1 mW. This power consumption is ˜44% of the gearbox vibration harvester's output (37 mW). Therefore, sampling at 50 kSPS rates for 1 second every minute can be perpetually sustained, without primary batteries, in the gearbox application. Rechargeable batteries are used to store energy from the harvester.
Our system was designed to be fully programmable, and therefore, fully adaptable to the needs of specific vibration monitoring applications. In Table III below, we provide the average power consumption (in mW) for 50 kSPS data collected for 0.1 sec, 0.5 sec, and 1 second durations at time intervals of once per minute, once per 10 minutes, once per hour, and once per day. Longer time intervals between data acquisition bursts would allow smaller, lower mass energy harvesters to sustain operation.
Thus, energy requirements from a battery or energy harvester are reduced with shorter sample duration and longer acquisition time intervals between sampling. Because the nodes have a real time clock, data obtained from the vibration monitoring nodes on the gear boxes is time stamped with a clock synchronized with clocks on the smart bolts and so the data for each can be correlated.
In another embodiment, piezoelectric material is replaced with one or more magnets 180 vibrating inside one or more coils 182, as shown in
Several separated coils 182 are wound around each bobbin 190 with coil winding directions alternating clockwise and counterclockwise. In this embodiment three such coils 182 are included. Two bobbin and coil assemblies 192 are clamped to front clamp 188a with coil clamps 194.
Passing through each bobbin 190 is a stack 196 of permanent magnets 180 arranged with multiple oppositely directed magnetic fields to induce current in each of oppositely wound coil 182. One embodiment, illustrated in
Tuning weights 202 are provided with each stack of magnets 196, and magnet stacks 196, tuning weights 202, and copper spacers 200 are all mounted to two circular spring elements 204a, 204b connected to the rear post clamp 188b. Circular spring elements 204a, 204b form a linkage that constrains stack of magnets 196 to move in curvilinear fashion up and down within bobbin and coil assembly 192, preventing binding. As rear post clamp 188b vibrates with post 186 to which it is mounted, spring element 204a, 204b also vibrates, and vibration amplitude increases with the distance from the pivot point 205a, 205b at rear post clamp 188b, the maximum amplitude directly across from the pivot point. If tuning masses 202 are adjusted so that spring 204a, 204b vibrates in resonance with the frequency of vibration of post 186, amplitude of vibration will be large and a substantial amount of electricity will be harvested at coils 182 and brought along wires (not shown) to power conditioning and signal conditioning boards 206.
The magnetic induction energy harvesters of
E=−dφ
B
/dt
By taking minor input vibrations and amplifying those vibrations via a resonant-tuned mass-spring system, we create the desired large magnet oscillations with the coils. Our specific magnet arrangement is configured with two specific parameters in mind: (1) Non-ferrous spacer 200 is sized to specific dimensions to optimize the rate of flux change in coil 182 as magnet 180 passes though, thereby optimizing the current and power induced. This size is determined via computer modeling of the magnetic interactions; and (2) repelling magnets are placed within coil 182. The opposing polarity concentrates the flux lines thereby increasing the rate of flux change as magnets 180 oscillate. This configuration allows even minor oscillations to produce significant current.
Magnetic flux lines imaging the magnetic field produced by opposing magnets 180 inside coils 182 from FEA software are shown in
The prototype built by applicants used 0.012″ thick stainless steel spring elements 204a, 204b with an OD of 2.3″ and an ID of 1.9″. The prototype used bobbins 190 with 2 coils 180 each (therefore 3 magnet arrays 198 on each stack 196). Coils of various wire gauges (32-48) were tested with an ID of 0.25″, an ID of 0.5″ and a length of 0.25″. As the wire gauge was decreased (wire diameter increased) the output voltage decreased while power remained constant. Magnets 180 used in the prototype were N50 (NdFeB), 0.1875″OD, 0.063″ID, x 0.1875″ long. Tuning weights 202 were adjusted such that the vibrating mass, including magnets 180, weights 202, spacers 200, and springs 204a, 204b was 22 grams. This resulted in a resonant frequency of approximately 22 Hz. Input vibrations for these tests ranged from 0.1 g to 1.0 g and the energy produced ranged from 3.5 mW to 35 mW respectively.
Mechanical stops could be included to protect springs 204a, 204b against overtravel and premature failure. These stops could be magnetic (opposing the last set of magnets) to create a damper effect as opposed to a hard stop.
In another embodiment with one or more magnets 210 moving in coil 212, flexure element 214 is internal to housing 216, as shown in
AC voltage produced from energy harvesting element 236 is rectified by rectifier 238 and stored in a large capacity storage element, such as input storage capacitor 240, as shown in block diagram of
Supercapacitor 248 is included in parallel with TFB 246 to provide high peak current capability. Applicants found that TFB's 246 internal series impedance increases as temperature decreases, and this high impedance limits the peak instantaneous current. They found that using supercapacitor 248 in parallel with TFB 246 mitigates this issue. Thin film battery 246 can be part number MEC07, IPS Inc., Littleton, Colo. Supercapacitor 248 can be HW207, CAP0xx, Ltd, Myrtle Beach, S.C. Buck converter 244 can be part number LT1934-1 from Linear Technology, Milpitas, Calif.
As long as TFB 246 is adequately charged battery undervoltage/lockout switch 250 provides power to the application load. If TFB 246 is discharged below a critical level, such as 2 V, application load 252 is disconnected from TFB 246 until it is charged. Voltage sensitive switch 242 and battery undervoltage lockout switch 250 can both be part number LTC1540 from Linear Technology.
The wireless sensor data aggregator is responsible for time synchronized data collection from wireless and hard wired networks. Important design criteria for the WSDA were:
One of the challenges for a distributed multi-network topology is synchronizing all the data acquisition nodes throughout the entire system.
The real time clocks 78 on all wireless and wired sensor nodes 64 are synchronized at the beginning of a test to the WSDA base station's time reference, using a wired beacon and a wireless beacon to communicate that reference. The WSDA base station uses a Global Positioning System (GPS) 1 pulse per second (PPS) clock as the default timing reference. In the event that GPS is not available, the WSDA switches to its internal +/−3 PPM real time clock as the timing reference to insure synchronization of all the remote sensor nodes to the WSDA's clock. With either timing source, the WSDA's timing engine provides a stable 1 Hz reference for the transmitted synchronization beacon.
On wireless sensor nodes 64, the same timing engine is slightly modified to provide the real time clock to provide adjustable output from 1 Hz to 4096 Hz, which is used to drive a sensor-sampling interrupt on the host processor.
Precise timing enables the aggregated data from the network to be accurately time stamped, but it also enables scaling of the wireless network. Combining time division multiple access (TDMA), carrier sense multiple access (CSMA), and frequency division multiple access (FDMA), the synchronized network can support a large number of wireless sensor nodes. The system's aggregate sensor sampling rate with continuous digital wireless communications may be estimated at 10,000 samples/sec per radio channel (at up to 16-bit sensor data resolution). For example, assuming a network of wireless strain nodes were configured to sample a 3-axis strain gauge rosette at a rate of 33 samples per second, this system will support up to 100 distinct wireless nodes (300 strain gauges) using only TDMA and CSMA techniques on a single radio communication channel. By adding radio transceiver chips, or by scanning radio channels within the WSDA base station, the system will theoretically support 16 of these strain sensing networks, or as many as 300 strain gauges*16 radio channels=4,800 individual strain gauges.
The results of the load sensor measurements are ultimately sent to the pitch control system, which in one embodiment is located inside nacelle 270 of wind turbine 37, i.e. on a non-rotating part of wind turbine 37.
a) In one embodiment, each sensor, such as strain sensors, accelerometers, and temperature sensors in instrumented axial load-sensor bolt 50, or gearbox sensors 272 in gearbox 274, are connected to a wireless transceiver. The data from each sensor is wirelessly transmitted to a transceiver (not shown) in hub data collection unit 276 in rotating hub 36. From hub data collection unit 276 in rotating hub 36 the data is transmitted to WSDA box 278 inside stationary nacelle 37, as shown in
b) In another embodiment, each sensor is connected by wire to hub data collection unit 272 that has both wired and wireless transceiver capability in rotating hub 36. From rotating hub data collection unit 276 the data is then wirelessly transmitted to WSDA box 278 in stationary nacelle 37. A CAN bus or an RS485 network can be used for wired transmission from sensors to hub data collection unit 276, minimizing the amount of wire used.
c) In another embodiment, each sensor is connected to a wireless transceiver and the data from each sensor is wirelessly transmitted directly to WSDA box 278 in stationary nacelle 37, as shown in
One of the technical challenges that affects this choice is that the transmit antennae of load sensors in bolts 50 are located within metallic hub 36. Hub 36, shown in
In addition to the strain gage method for instrumented sensing bolt loads as described herein above, alternate instrumented sensing bolts can be used. One approach marketed by Rotabolt (http://www.sound-connections.co.uk/rotabolt/index.html) is shown in
In one embodiment applicants provide a sensor, such as an air gap capacitance sensor to provide an electrical signal indicating size of the gap between the normally freely spinning cap and the bolt head.
In another embodiment, applicants provide a sensor, such as a DVRT, to provide an electrical signal indicating the change in length of the bolt when the bolt is properly tightened. That change in length is related to strain and the amount of tension in sensing bolt 300, as shown in
Another bolt load sensing scheme is shown in
d=Incremental displacement=lunder load−lno load
P=Bolt load
l=Length of bolt under load
E=Youngs modulus of bolt material
A=Cross sectional area of the bolt
Therefore, the bolt load can be determined as
In another embodiment, strains on blades are monitored with strain gauges bonded to the blades, such as an interior surface of blades. In one embodiment, three strain gauges 320a, 320b, 320c are oriented within hollow blade 30 at 120 degree angles to each other, as shown in
In one embodiment, four sensors are used, placed at 12:00, 3:00, 6:00, and 9:00 o'clock, and this simplifies calculations because bending moments can be derived directly from measurements. The sensors at 12:00 and 6:00 are differenced to amplify flapwise bending and cancel temperature effects. Sensors at 3:00 and 9:00 are differenced to amplify edgewise bending. One problem is that there are huge temperature gradients on wind turbine blades and the material is not isotropic. In one embodiment, a secondary piece of material is used within each sensor to cancel temperature. Such strain gauges are available from Columbia Research Labs, Woodlyn, Pa. The present inventors recognized that to cancel readings by differencing 12:00 and 6:00 sensors, for example, advantage is provided by synchronizing them precisely or wiring them together.
A package as described in the '244 application, “Strain Gauge with Moisture Barrier and Self Testing Circuit,” can be used at each of the three positions in each blade. Precision timekeepers can be included in each package for time synchronization. A compliant bonding method can be used to conform the package to the curved shape of the blade.
In one embodiment, the wireless communication protocol is scalable and time-synchronized, facilitating correlating load data. In one embodiment, the communication protocol supports both wireless and hard-wired sensor networks. Data from different sensors are collected at multiple sampling rates and time stamped and aggregated within a single scalable database on a base station, such as a WSDA. The WSDA is connected to the pitch control unit. It can also be integrated with the pitch control unit. In one embodiment, the WSDA's processor supports a Linux server, web interface, eight (8) Gigabytes secure flash memory, CAN, IEEE 802.15.4, Ethernet, RS232/422, and mobile phone. In one embodiment the data is relayed over mobile phone networks to a secure server. Software permits access to the aggregated data over the internet, using the time stamp as a unifying reference for the various types of sensor information.
The WSDA, in addition to providing a central location for collecting wireless and wired network sensor data, also provided a beaconing capability to synchronize each sensor node's embedded precision timekeepers. For testing, a saw tooth analog voltage waveform was provided as an input to two wireless nodes to provide a reliable means of determination of the system's timing accuracy. With the synchronization beacon provided only at the start of a 2 hour long test, and with two wireless nodes exposed to multiple temperature cycles of −40 to +85 degrees C., the system demonstrated a 5 millisecond timing accuracy. Thus, for a wind turbine for which temperature changes more slowly, transmitting a beacon once every 20 minutes would result in a sub-millisecond timing accuracy.
In one embodiment, a time beacon is provided from the base station or WSDA that reaches all sensor nodes simultaneously, as shown in
While sampling of data from sensors is thus synchronized in all the sensors receiving the beacon, transmission from the nodes is staggered, as shown in
In one embodiment, 10 nodes are transmitting, two of which are shown in
We estimate power consumption for each smart bolt with a Nordic radio transmitting data updating at 20 samples per second will be less than 120 uA on average. With the smart bolt equipped with the D cell from Tadiran the battery life can be calculated by the capacity of the battery, which is 19A-hours divided by 120 E −6=158,000 hours which is 18 years. The DecaWave radio will provide improved range when it becomes available for purchase in a few months.
Signals in addition to strain (or load) can be measured with the smart bolts of the present patent application. Triaxial accelerometers and rate gyros included in the instrumented sensor bolts permit determining:
While other sensors are currently used to measure shaft rotation angle and rate and blade pitch angles, providing sensors in bolts for measuring these parameters may be useful and may lower cost because integrating a variety of sensors in one easily added location can reduce installation and system complexity.
In addition, in one embodiment, blade root (or hub) vibration modes are also measured using strain gauges or accelerometers. Almost all other elements of the wind turbine such as the generator shaft, gearbox and tower already use vibration sensors (strain gages or accelerometers) for structural health monitoring. The present patent application thus provides a way to provide vibration data for a portion of the system not otherwise measured.
In addition, structural fatigue monitoring to prevent potentially catastrophic faults in rotorcraft and other structures is obtained by tracking load history with the synchronized energy harvesting wireless sensor nodes of the present application. A paper, “Extending HUMS to rotor systems,” by Kieran Daly, incorporated herein by reference, and published at http://www.flightglobal.com/articles/2009/02/16/322531/extending-hums-to-rotor-systems.html describes the problem with health and usage monitoring of rotors that is solved by the present patent application.
The health of vibrating machinery, such as gearboxes connected to rotors for helicopters and wind turbines, is also tracked using the sensor nodes of the present patent application.
Finally, temperature sensing can be important. In one embodiment the load sensors are temperature compensated. In addition, numerous temperature sensors are provided in the wind turbine system to monitor load sensor temperature, external air temperature, gearbox temperature, etc. In one embodiment, hub temperature is measured as a diagnostic, to early identify a maintenance problem or, for example, to identify a pitch servo controller that was overheating or that caught fire.
In a book: EMC for Product Designers, fourth edition, by Timothy Williams, ISBN 0-7506-8170-5 pages 421 to 423, several modes of lightning energy ingress to exposed electronic circuits are described.
Each of these entry modes has an individual protection solution. In an assembly like a bolt with embedded sensors that is to be mounted on a wind-turbine tower, lightning exposure potential is very high.
In the present invention, where electronics are entirely contained within a steel bolt, the exposure is predominantly limited to two categories.
In one category of entry mode, transient voltages are caused by current flowing through the resistance of the bolt. This produces a potential difference across the conductive surface of the bolt.
In another category of entry mode, electric and magnetic fields are induced by the current through the bolt and its inductive reactance at the frequencies where the maximum lightning transient energy exists.
For the case of the conducted voltage drop across the bolt resistance, if no direct connection is made between the bolt and the electronics, this threat is minimized. Strain gages used have a polyamide substrate with several thousand volts dielectric strength, making it possible to electrically isolate all of the electronics from the bolt structure.
For the high frequency electromagnetic radiated field components of a lightning transient event, a shielding sub-housing on the circuit board(s) is necessary. The exposure of the electronics from these effects is dependent on the size of the assembly and the length of any wiring, or circuit board traces used. By using a shielding sub-housing, current induced in the traces and wiring by this electromagnetic component, is shunted around the electronics.
Where battery operated electronics are used, conducted energy is not a concern for the power supply path. Where conductors from strain gauges or other external sensors enter the electronics, care must be taken to protect from electromagnetic transients induced on wiring. Various commonly used components, such as gas discharge tubes, Zener diode type transient voltage suppressors, and metal oxide varistors can be used to protect from these transients.
To summarize, mitigation of lightning is accomplished by isolating the electronics, i.e. no direct connection to the bolt, employing shielding sub-housing on circuit boards, and transient voltage limiting suppression devices applied to sensor wiring that must enter the sub-housing. In addition, when all the electronics is contained within an outline that is small relative to the wavelength of the electromagnetic transient, exposure is minimized.
In one embodiment, a strain gauge package 330, as described in the '244 application, “Strain Gauge with Moisture Barrier and Self Testing Circuit,” is applied at three positions inside blade tower 60, as shown in
Tower 60 has variable geometry, so a calibration is performed. In one embodiment, calibration is performed by measuring response of tower 60 before and after erection of generator assembly 334 that includes nacelle 37 within which is gearbox 274 and a generator. As a crane mounts generator assembly 334 onto tower 60 it's known weight and known center of gravity is used to calibrate response of strain gauges 330 in tower 60. Similarly measurements can be taken as other components are added to tower 60, such as rotor blades 30 with their known weight and moment arms. Alternatively, calibration can be performed by using a crane to provide known weights or forces to the top of tower 60 in known directions, for example known weights hung from pulleys. The response of strain gauges 330 in tower 60 is recorded in response to these known forces and moments. An appropriate transfer function is then stored in a memory in WSDA box 278. In yet another alternative, calibration of sensors used in the tower is performed on the ground prior to erection, or in the factory prior to shipment with known weights and moments.
In one embodiment, bending moments on tower 60 are measured. Since tower 60 supports blades 30 and the structure is in static equilibrium the bending moments on tower 60 must be equal to flapping moments on blades 30. Thus the present applicants recognized that the measurement of the bending moments in the tower can be used for continuous, in the wind calibration of the blade flap moments as measured by blade sensors 320a, 320b, 320c or by instrumented axial load-sensor bolts 50. The total moments measured by either of the blade sensors should be equal to the total moment measured by the tower sensors. This is useful for checking the validity of the blade mounted sensors.
In another embodiment, torque and RPM sensors are provided along the generator shaft, as described in the commonly assigned '505 patent, incorporated herein by reference. These sensors are used to obtain information about the lead-lag (edgewise) moments on the blades that cause the torque in the generator shaft. Thus, measurements of generator shaft torque and RPM can be used to validate the bolt sensor's measurements in this plane. Since the reaction force of the generator is unknown, this calibration check cannot be performed on a continuous basis. However, when the system is no longer spinning, then the laws of static equilibrium dictate that the torsional (edgewise) moments induced by the wind against the blades are completely resisted by an equal and opposite reaction torque in the main generator shaft. Thus, measuring the torque in the clamped generator shaft provides a measure of the torsional moment induced by the wind. If the turbine has been stopped, for example with a clamp located beyond the shaft torque sensors so as not to interfere with their measurement, then this torque measurement can be made and used as a check of the blade sensors' calibration in the edgewise plane.
In a further embodiment, by providing an RPM or angular velocity sensor on the generator shaft, the net input power at the generator's shaft can be determined. The power is equal to the torque times the angular velocity of the shaft. The input power measurement allows checking the overall efficiency of the generator itself, since the electrical output power provided by the generator depends on the input power and can also be measured. In the event of a significant drop in efficiency of conversion of the generator shaft power to electricity, the WSDA is programmed to send an e-mail or other alert via the internet or cell phone/satellite networks to the maintenance providers. In addition, the input power can be compared to a separate measurement of wind speed to determine whether the blades are properly transferring available energy to the shaft.
While the disclosed methods and systems have been shown and described in connection with illustrated embodiments, various changes may be made therein without departing from the spirit and scope of the invention as defined in the appended claims.
This patent application claims priority of U.S. Provisional Patent Application 61/169,309 filed on Apr. 15, 2009 entitled “Wind Turbines and Other Rotating Structures with Instrumented Load Sensor Bolts or Instrumented Load Sensor Blades.”
This invention was made with Government support under contract number N68335-08-C-0099, awarded by the US Department of the Navy. The Government has certain rights in the invention.
| Number | Date | Country | |
|---|---|---|---|
| 61169309 | Apr 2009 | US |
