PLCs can improve wind turbine performance
Most wind turbines use mechanical type anemometers and wind vanes to measure wind conditions, and wind speed and direction data are sent to the turbine controllers. However, there was much inherent vulnerability within the original equipment manufacturer (OEM) monitoring and control systems, leading to high maintenance costs and poor turbine power performance.
Harnessing the wind
Buffalo Gap wind farm, in central Texas, consists of 296 wind turbines (see Figure 1), constructed in three phases from 2006 through 2008. Phase one consisted of 67 1.8 MW turbines; phase two of 155 1.5 MW turbines; and phase three of 74 2.3 MW turbines. The capacity of the wind farm is 524 MW, currently making it the seventh largest wind farm in the world. It generates more than 1,600,000 MW/hr of clean, renewable energy per year.
The first problem was that the sensors use moving parts. The anemometer uses "cups" for wind speed measurements, and the wind vane uses a vane tail for measuring vector change, or wind direction. Physical inspection showed many failed sensors, with the majority of failures linked to bearing problems that led to accuracy degradation and shortened instrument life. Increasing bearing rolling resistance also affected wind measurement accuracy and hence turbine efficiency as this data is used to optimize turbine performance. Winter weather froze the exposed mechanical components of the sensors, adding to maintenance.
One brand of turbine suffered the greatest impact during freezing weather. A slight amount of moisture, when coupled with freezing temperatures, would cause the anemometer and wind vane to lock up, effectively shutting down the turbines. Another brand has heated sensors, which were slightly more tolerable to winter weather but experienced a much lower mean time between bearing failures due to the higher temperatures for the sensor bearings.
The second problem was sensor performance. Research and testing proved that the cup and vane type sensors were inaccurate in high turbulence or when steep wind shear conditions existed and were not suited for use downwind of the rotor, due to highly turbulent rotor wash.
Rotor acceleration increases gradually with wind speed during normal operation. However, when the anemometer under-reports wind speed due to a partial failure, rotor acceleration increases, indicating that a large amount of wind energy, around 10%, is not being converted into electrical energy. Instead, that energy is being absorbed through the main bearing and drive train and then dissipated by rotor motion.
The anemometer bearings typically will gain rolling resistance over time until they seize completely. Ideally, when the turbine recognizes abnormal loading due to increased rolling resistance, it will fault before a complete seizure. However, in many instances the turbine never recognized the problem, and this partial failure of an anemometer is more destructive than a complete failure because it can destroy the wind turbine drive train.
Improving the system
Wind turbines need accurate information regarding wind conditions to operate safely, as wind data is vital to record turbulence and protect the turbine. When an anemometer’s rotational speed is unable to change at a fast rate, as is the case when operating with faulty bearings, the turbulence cannot be measured accurately. The turbine control system adjusts blade pitch and rotor speed differently, depending on the degree of turbulence, to protect the blades and drive train from fatigue due to poor use of the blade airfoil. Bad anemometer measurements equal incorrect adjustments, which increases fatigue.
The need for accurate wind direction also is important. Power output dramatically decreases once the turbine exceeds +/-10 degrees misalignment. When compared to neighboring turbines with known good wind vanes, a 20% reduction in production was discovered, which could result from a GE turbine operating with a partially failed wind vane. This realization demonstrated that the measurement application was better suited for ultrasonic instrumentation, which would provide more accurate and reliable indication of wind conditions.
The wind measurement mechanical sensors were replaced with one ultrasonic sensor (see Figure 2) and used a programmable logic controller (PLC) to convert the signals from the sensors and better control the wind turbines.
The signal needed to convert from the ultrasonic sensor into a form that the turbine controllers could use. No available ultrasonic sensors provided the specific and various forms of information required for the various turbine controllers. Therefore, a Modbus adapter was installed, which accepts the input from the ultrasonic sensor signal and sends it to a PLC via Modbus.
The PLC acts as an emulator for the turbines; that is, specific ladder programs were written, depending on make/model of turbine, to emulate the digital logic sensor signal required by each turbine controller. We also were able to create algorithms to provide nonlinear corrections to dynamics such as the nacelle transfer function and yaw bias. All the equipment, plus a 240 W power supply, fit into the turbine nacelle, along with the existing OEM turbine controller (see Figure 3).
Using LIDAR, a light detection and ranging technology, to validate anticipated performance and the continued application of statistical tools and methods, the team realized it could further improve turbine performance. Yaw bias (alignment), along with wind speed correction (nacelle transfer function), was discovered to be inherently dynamic and nonlinear. This data allowed engineers to write algorithms to correct the distorted wind speed and direction data before interpretation by the turbine control system, thus increasing energy capture and reducing drive-train fatigue.
Data from the PLCs is transmitted to the control center (see Figure 4), where engineers perform statistical analysis, program the PLCs and turbine controllers, and analyze data to predict failures before they happen.
While the systems are still being fine-tuned, initial results include increased meteorological accuracy, less operational downtime, and reduced turbine-drive-train fatigue loading-along with increased turbine efficiency.
Tristan Lee is performance engineer at AES Corp. in Buffalo Gap, Texas. Edited by Eric R. Eissler, editor-in-chief, Oil & Gas Engineering, email@example.com.
- Physical moving sensors are not ideal for wind turbines.
- Measurements need to be accurate to ensure proper function; otherwise improper actuation can lead to severe damage.
- The need for accurate wind direction is also important. Power output dramatically decreases once the turbine exceeds +/-10 degrees misalignment.
When setting up a wind turbine it is important the PLC, sensors, and actuators are properly calibrated to ensure top performance.
– See related stories about PLCs and turbine design below.