Frequency response by wind farms in power systems with high wind power penetration
Doctoral thesis, 2017
The integration of variable speed wind turbines (VSWT) in power systems is increasing in order to reduce the emission of green house gases. This increase of power electronic converter interfaced generation causes a decrease of power system inertia, increasing the risk of load shedding and system frequency instability. This thesis investigates the impacts on frequency stability that can be caused by a decreased inertia in weak power systems.
Firstly, a simplified algorithm for estimating the kinetic energy in the Nordic power system has been presented and compared to the TSOs’ estimation based on breaker status. The average error between the TSOs’ estimate and the proposed estimate is 0.83%. Furthermore, the standard deviation of the error is 13 GWs, about 7% of the system average kinetic energy. The algorithm manages to capture the alternating behaviour of the system kinetic energy with a correlation coefficient of 0.94 when compared to the existing TSOs’ estimate. By using the algorithm, the system kinetic energy in a 2025 scenario is estimated and possible contributions with additional kinetic energy from VSWT is considered.
Secondly, the frequency behaviour in the Nordic power system is evaluated and the roots and causes of the decreasing frequency quality is analysed. In terms of the large power imbalances, about 70 events, spread through 2012-2014 were evaluated and out of theses, 20% caused Ro-CoF values steeper than -0.1 Hz/s. Furthermore, the evaluated imbalances had an average time to nadir of 8.7 s and an average frequency change of 0.3 Hz.
Thirdly, in the case analysed with a hydro unit and a 50% wind penetration ratio, the frequency duration outside [49.9, 50.1] Hz was reduced from 81% to 53% when the VSWT is controlled with the proposed (frequency controlled pitch actuator&droop)-strategy, as a complement to primary frequency control. The wind energy reduced by this frequency support (due to suboptimal operation on theCp(l ,b )-plane) is about 6%. The absolute requested system frequency reserve has been reduced by 49%. Lastly, in an islanded, hydro power based test system with 50% wind power subjected to a generation trip. The proposed algorithm, with a fixed support phase and a frequency dependent recovery phase, the best frequency nadir with a smaller impact on rotor speed (4.2% less) than a f-independent algorithm. The proposed algorithm has the lowest frequency error of the algorithms evaluated, 5.2% better than the f&df/dt-dependent algorithm and 15.5% better than the f -independent algorithm. The proposed algorithm is then evaluated in PSS/E with the Nordic-32 bus model with a WPR of 34%, improving the frequency nadir from 48.91 (without support) to 49.22 Hz.
power system stability
wind power integration