Skip to main content

WIND FARM SCADA DATA ANALYSIS

Wind Farm Maintenance

Why install additional wind turbine condition-monitoring hardware when your existing SCADA data can be used as a condition-monitoring system? Wind farm maintenance can be more effectively planned and failures prevented (or in the worst case, anticipated).  Many academic papers have been published on wind turbine SCADA data analysis. Most focus on early failure detection of major components (e.g. gearbox and generator) and demonstrate that in many cases it is possible to anticipate failures. SCADA Miner creates neural network models to predict a watched value (e.g. main bearing temperature, drivetrain vibration or pitch motor current) based on predictor values (e.g. ambient temperature, active power, wind speed, pitch angle etc.).  Alerts are produced upon deviation from these typical models. The effectiveness of such techniques is well-documented. There are however, many simple tests that can avoid downtime which are not well-documented. Examples are given below.

Panel Temperature

Wind turbine technicians sometimes turn panel cooling fans off while they service a panel. The fans are noisy, so this is understandable. Sometimes technicians forget to turn the fans on when they put the turbine back into service. Depending on ambient temperature and wind conditions, the turbine can operate for days with no panel cooling fans.

A test which examines the previous hour of ambient temperature, active power and reactive power can predict the expected temperature of this panel with reasonable accuracy. The panel with no cooling fans will quickly deviate from its expected temperature characteristic and generate an Exception Report which is emailed to the lead service technician. In many cases this alert can save the turbine from being out of service overnight due to panel over-temperature.

Wind Turbine Maintenance Events

Wind turbines execute periodic maintenance routines such as lubrication and pitch battery tests. Periodic routines are easy to audit with automated tests. Once a week, SCADA Miner counts the number of occurrences of periodic event codes (exceeding a minimum qualifying duration) and compares this count (normalised by turbine uptime) to a minimum acceptable count. If the wind turbine is found not to have executed a sufficient quantity of qualified events, an Exception Report is generated.

Failure to execute the expected quantity of a periodic routine seems unlikely but it can happen for a number of reasons including wrong parameterisation, a wiring fault or repeated turbine downtime. Such a test can avoid damage.

Sensor Failure

Wind turbines use a number of temperature sensors to monitor various components and shut down the wind turbine if the component exceeds an alarm level. Most SCADA historians collect 10-minute minimum, maximum, average and standard deviation analogue values. A temperature sensor will typically fail open or short circuit. When this happens, the measured temperature will be a limit value (e.g. 250 or -50°C). When the sensor begins to fail, it can exhibit short duration excursions to these limit values. Because of the short duration of limit value occurrences, the wind turbine control system does not shut the turbine down (it uses an average temperature to reduce noise susceptibility). These limit values are however, visible in the 10-minute minimum and maximum values and their frequent occurrence always precedes sensor failure as they represent an intermittent open/short circuit.

SCADA Miner’s suite of sensor failure tests increase the likelihood of receiving an early warning of sensor failure and performing preventative maintenance to avoid downtime due to sensor failure.

Wind Farm Performance

SCADA Miner’s tests achieve great repeatability using alerting based on deviation from an historical wind turbine power curve characteristic or a wind-farm-average characteristic. Groups of measurements can also be tracked such as blade pitch/active power/wind speed and generator speed/power. Exception Reports are produced upon detection of a deviation. Below is a small sample of SCADA Miner tests for problems which are often overlooked.

Nacelle Direction Drift

On wind farms with wind sector management, it is important to ensure the wind turbines do not lose their sense of direction. This can lead to generation when the turbine should be shut down (e.g. for noise or load mitigation reasons) or unnecessary downtime because the turbine thinks it is in a shutdown wind sector when in fact it should be generating.

Inefficient Reactive Power Exchange

Inefficient levels of reactive power can occur both at the turbine terminals and the point of connection. SCADA Miner offers tests not only to ensure the wind farm complies with its connection agreement, but also to produce alerts if the wind farm generates or consumes reactive power exceeding the minimum level required by the connection agreement. Excessive reactive power exchange causes real losses.

Unexplained Wind Farm Constraint Detection

Unnecessary wind farm constraints occur much more frequently than may be expected. These can be due to wrong parameterisation, software updates, operator error, and poorly implemented active power control routines.

SCADA Miner has tests to detect suspect constraints and alert wind farm owners to their occurrence.

Generation when the Spot Price is Negative

For a number of reasons, wind farms sometimes generate electricity when the spot price is negative enough that the wind farm owner has to pay for the electricity generated (even when government renewable energy rebates are considered). SCADA Miner offers tests which alert on detection of generation when the spot price makes it unprofitable as well as on detection of conditions which can lead to this occurring.

Wind Farm Compliance

Existing SCADA systems offer simple alarming based on repeating alarms that are generated by other devices or analogue thresholds. SCADA Miner can analyse data to produce Exception Reports when a problem requiring attention is identified . Some example compliance tests offered by SCADA Miner are listed below.

Reactive Power Compliance

Tests can be configured to generate alerts upon detection of excursions (of configurable severity) beyond the limits of the connection agreement.

Sensor Correlation

Signals such as wind speed, temperature and wind direction are sent to third parties to aid forecasting. High signal quality may be required to ensure compliance with the connection agreement.

Wind speed measurement from a met mast can be checked against nacelle wind speed measurements, as can ambient temperature. Generation of an Exception Report following loss of correlation between the primary signal and its predictor ensures timely awareness of the problem and can avoid breach notices from the grid authority.

Grid Event Detection

Wind turbines typically auto-restart if they trip due to a grid event (e.g. voltage dip or phase shift due to upstream network switching). This can result in very little downtime which often means the event goes unnoticed. Successful fault ride through events are even easier to miss. Wind turbine performance during grid events should be examined to ensure behaviour is compliant with the connection agreement. Upon detection of grid-related alarm codes being generated by multiple turbines within a short timeframe, an Exception Report is created. This prompts the recipient to examine the wind farm performance.

Wind Turbine Availability Auditing

The downtime report is often the basis of calculation of service payments from wind farm owner to wind turbine manufacturer or service contractor. Due to the commercial significance of this data, it makes sense to run automated tests to ensure its accuracy and identify potential problems.

Detect Excessive Alarm Remote Resetting

A high count of an alarm code may indicate the alarm is being remotely reset rather than the problem being investigated. If a wind turbine shuts down due to over-speed twice in one day, has a recurring pitch alarm or shuts down multiple times due to failure to successfully execute a scheduled safety/maintenance routine, an Exception Report can be generated. Excessive alarm remote resetting can cause permanent damage. By alerting the wind farmer to its occurrence, the Exception Report can help prolong equipment life.

Detect Wrongly Allocated Downtime

Wind turbines may perform scheduled routines which result in owner-allocated downtime. Examples include lubrication and cable untwist events. Periodic tests can be created to ensure the frequency and duration of such events is in accordance with typical operation. These tests alert the owner to a wind turbine stuck in an owner-allocated downtime state. They will also generate an alert if a wind turbine is performing a greater number of scheduled events than anticipated. This could be due to wrong parameterisation or a sensor failure.

Temperature Alarms vs. Outdoor Temperature

Turbines can shut down or constrain due to high component temperature when the outdoor temperature is within specification for unconstrained operation.  Sometimes the turbine measures  high (or extremely low) outdoor temperature and shuts down when in fact, the temperature sensor has failed. SCADA Miner checks the temperature recorded by an alternative measurement source.  An Exception Report is generated if temperature shutdown or temperature-initiated power reduction occurs within wind turbine rated operating temperature.  This information can be particularly valuable. If a turbine constrains due to a high temperature component, that turbine is still 100% available according to most availability-based contracts.  While the wind farm owner suffers lost revenue, there is little motivation for the OEM to address the problem.  In countries where the electricity spot price increases sharply on hot days, these constraints occur when electricity is most valuable. There can be significant cost benefits in early detection and rectification of temperature-constraining defects.

Routine Reporting

SCADA Miner can automate the creation of routine reports. Example reports include:

  • Partial/custom availability calculations not supported by wind turbine manufacturer
  • Comparison of MTTR/MTBF for the same alarm at different wind farms used as a performance metric for teams of wind turbine technicians.
  • Top 10 alarms causing downtime at each wind farm
  • Availability
  • Production
  • Lost production factor
  • Remote alarm reset performance – how quickly monitoring technicians address alarms