Tag: academic paper

  • Ancillary services from wind turbines: automatic generation control (AGC) from a single Type 4 turbine

    This paper was published by the Wind Energy Science journal. The journal is open-access and is a run by the EAWE – the European Academy of Wind Energy. The publication charge is reasonable at around EUR 70 per page and there is no page limit. Even nicer is that the processing charge is lower if you use Latex. That puts a smile on my face 🙂

    Thankfully, you can read the full paper here: https://wes.copernicus.org/articles/5/225/2020/

    https://doi.org/10.5194/wes-5-225-2020

    TL;DR Version: This work uses a single, IEC Type 4 wind turbine to provide AGC. The turbine is an 800 kW Enercon E-53 machine located at the Cowessess First Nation site east of Regina, Saskatchewan. 10% of rated power is offered on the regulation market and up-regulation (power increase) is provided via continuous curtailment. Anemometer wind speed is averaged over 30 seconds and used to estimate power in the wind via a persistence model – power is assumed to match the average value. Power targets are updated every four seconds. Performance is good – as you might expect with any inverter-based technology. The question of income is more challenging although PJM regulation market rates would lead to profitable operation.

    Image showing time series data of the wind turbine active power following an external power target.
    (a) Providing 40 kW of regulation from a single 800 kW wind turbine when operating at rated power. Note that the offset observed is discussed in Sect. 3.2. (b) Providing 100 kW of regulation. Blue shaded region in (a) and (b) is the range of possible regulation. The diagram labelled “. . . (a)” continues.

  • Performance analysis of a 10 MW wind farm in providing secondary frequency regulation: Experimental aspects

    I wrote this while at WEICan – the Wind Energy Institute of Canada. This is an IEEE Transactions on Power Systems paper, published in 2018 and presented at the PES GM 2020. The paper is behind the IEEE paywall but as with all papers summarized here, you can request a copy by email.

    The key reason this was published is to make data public. Wind turbines can vary their active power output – this is a well-established fact. What is missing is an objective analysis of this ability – how well it works, what trade-offs are involved, how one wind turbine compares to the next and so on. This is empirical data and publishing data is always challenging with the academic community. The academic focus is on advancement – usually via something “new” – a new application, new ideas, new theories, new combinations. Much of the “new” exists in the ideal world of computer simulations with numerous assumptions and simplifications. This is in stark contrast to empirical data from the real world – data whose value should be obvious.

    TL;DR Version: This paper analyses five hours of data from a grid-connected wind farm providing AGC – automatic generation control. Very simply, the 10 MW wind farm varies its active power up and down in a 1 MW range. The upper limit of this range is the maximum power currently available via prevailing winds. Continuous power curtailment is used to provide room for power increases. The performance score – how accurately the wind farm follows the power signal – is 65% on average over the five hours of data, calculated via the PJM method. Using PJM’s market price data for AGC, participating in the AGC market is profitable with the performance score calculated.

    More details and nuance:

    1. WEICan’s wind farm uses Type 5 wind turbines. These are rare as they are no longer in general production. They are unique in that the design uses a synchronous generator with a direct connection to the power grid. No power electronics are used in the grid interface. The generator operates at 60 Hz (North America) and uses a hydraulic torque converter to maintain constant rotor speed. The specific term used by Voith is a “hydrodynamic torque converter”.
    2. The technology internal to the wind turbines should not affect results as these results are representative of general wind turbines. The key point is that the power output is variable and depends on the prevailing wind speed. This fact is true for any wind turbine design. Power accuracy changes depending on the technology used. An inverter can track active power more accurately – to cite one point of difference.
    3. The key point with profitability is market income – how much is frequency regulation worth in relation to energy. The PJM market for example had (in 2017) regulation market prices high enough to make this profitable.
    4. The control scheme used is simple – curtail power enough to create room for up-regulation. The active power estimate is achieved by a look up table and average wind speeds (ten minute). As you might expect, this sometimes produces an estimate above the power available in the wind. In these situations, error increases as the power target will not be reached.
    5. On the flip side, curtailing to power levels below what is available is generally not a problem. DeWind’s design appears to have a lower limit to power curtailment. Anecdotally, this is around 500 kW for individual wind turbines. The reason might be minimum torque margins to maintain generator shaft speed. The turbine themselves are able to operate at lower power levels, down to approximately 100 kW. Since DeWind no longer exist, the precise reason is not known.