High Voltage Concern at Wind Farms?

Visio-Smoky-Hills-One-Line-992x1023If we think about wind turbines as induction generators, one would assume that these would be VAR (reactive power) sinks, demanding vars from the grid to be able to deliver watts. However, that may be true from the point of view of only the wind turbines themselves. In reality, wind farms are far more than a group of small generators. Electrically, wind farms that deliver at bulk power levels to the grid behave more like a small urban subtransmission grid with characteristics that are far removed from those of a large power facility such as a coal, oil, nuclear or natural gas plant.To illustrate, consider the following observations from a system operator regarding a set of wind farms on their grid.

  • “High voltage is occurring on System-A’s 69kV system during periods of low wind generation at Wind-Farm-A and Wind-Farm-B. It looks like Wind-Farm-A is a little higher than Wind-Farm-B and can go as high as 74.8kV (108.4%). Wind-Farm-B peaked at 73.8kV (107%). ”
  • “There is no voltage control or power factor control at Wind-Farm-A or Wind-Farm-B. (The wind turbine generators are not configured with the ability to control voltage). As the real power increases, the wind turbines will absorb some reactive power which helps to lower the voltage.”
  • “With no wind, there is real power load at the wind farms. Real power load may go higher in cold weather since the wind turbines have oil heaters installed. The 34kV underground cables for the collector system have capacitance, so at times of no real power generation; we are seeing some VARS flowing from the wind farm to the transmission system.”
  • “We’re trying to improve our efforts on modeling and controlling these wind farms as their presence on our system is increasing. The main point is that our operators are seeing these wind farms dispatch MVARS when there is no real power output and absorb MVARS when there is real power output under certain conditions.”

A typical wind farm layout is shown above. The turbines are spread out all over the farm and are interconnected via a 34.5 kV system, also known as a “collector” system. The distances can vary and the cable sizes for the 34.5 kV lines can also vary. Transformers step up the voltage at the turbines to the 34.5 kV collector level, then to the transmission voltage (in this case, 230 kV) at the substation. The individual feeders have varying lengths and voltage drop issues may require capacitors to be installed at various locations (not shown in the diagram).

In terms of the reactive power utililation for the wind farm, we note that:

  • The wind turbines are inductive at all times when generating MW. Some newer designs will have voltage control that allows for a +/- 95% power factor range.
  • The wind turbines have transformers that step up to the collector system voltage, in this case, 34.5 kV, which are inductive.
  • The transformer at the substation is inductive.
  • Any capacitor banks will provide VARs as a function of the square of the voltage at the capacitor terminals. For example, a 100 kVAR cap bank will deliver 81 kVARs when the terminal voltage is 90% of nominal, and 121 kVARs when the voltage is 110% of nominal.
  • The collector feeders will generally be inductive when the wind farm is generating real power (watts). When the wind farm is not delivering power, the feeders are still energized but carrying power which is below their surge impedance loading. Under this condition, the feeders are capacitive. The capacitance is greater if the feeders are underground instead of overhead.

Based on the above, we would expect that the wind farm will absorb VARs when delivering MW, and the transmission voltage is at or below nominal. When the facility is not delivering MW, the feeder system may be sufficiently capacitive to overcome the inductance of the transformers, and we see high voltages on the transmission side.

In order to capture these effects in simulation software, a reasonable representation of the feeder system electrical characteristics is needed. Aggregate models (see Wind Farm Integration: On the Use of Agreggate Models) reduce the accuracy of the simulations or may not reliably capture the full range of response of the wind farm.