Five to six years ago the installed amount of power from photovoltaic (PV) systems in Germany increased exponentially. This was due to strong incentives provided under the German Renewable Energy Sources Act (EEG). Although connected to the low-voltage distribution system, the contribution from this large number of low power systems was already starting to influence the overall system, introducing previously unseen risks.
At that point the technical directive in place for validating grid compliance of PV inverters - Generating Plants Connected to the Medium-Voltage Network, Guideline for generating plants’ connection to and parallel operation with the medium-voltage network; June 2008 - already took into account many component based grid requirements, such as transient stability (fault-ride-through). However, it fell short on at least one aspect that could not have been found by looking at or testing only a single PV inverter: the alignment of tripping events.
According to the interface protection rules of that time an immediate shut-down of the PV inverter was required if the grid frequency should at any point in time reach or exceed 50.2 Hz. In itself this is an appropriate rule to prevent over generation until the grid’s primary control systems have had time to recover the situation.
However, considering that the combined power contribution from the numerous PV inverters have reached proportions of several gigawatts, especially during high production periods, the implementation of this interface protection rule at a fixed 50.2 Hz, unwittingly instigated an instantaneous loss of generation that can be significantly larger than the balancing power available Europe-wide for primary frequency control, rendering the overall system unstable.
Likewise, if the grid frequency is restored and the PV inverters are again allowed to connect and deliver the available power to the grid at exactly the same time, the frequency limit will again be exceeded, thus causing the generators connected to the low voltage distribution network to shut down again, an effect often referred to as the ‘yo-yo effect’.
The technical directive has since been updated with a droop control in power at increased frequency levels, but this situation still indicates the value in system testing on top of component testing. For example, hardware-in-the-loop (HIL) testing and model validation would have shown the 50.2 Hz problem before it happened. Individual component testing could not do that.
For further information on HIL testing and model validation please see DNV GL’s position paper on Power Cybernetics.