Comment on GB Power Outage on 9 August 2019

Since the power outage on the GB transmission network on 9 August 2019, my question, from an entirely parochial perspective, has been - if the Hornsea Offshore Windfarm had been connected to the grid via HVDC rather than AC would the "de-loading" (a euphemism for trip) not have occurred, and could the subsequent trips, frequency events and "de-loadings" across the network potentially have been mitigated or avoided.

Spoiler alert: Yes

For reference, the initial scoping document of Hornsea One showing both AC and DC connection options can be seen here: Hornsea One Project Description.

The conclusion of Appendix D (Hornsea Technical Report) of the Final Technical Report is as follows:

"The de-load was caused by an unexpected wind farm control system response, due to an insufficiently damped electrical resonance in the sub-synchronous frequency range, which was triggered by the initial event."

The swings in V, P and Q could be sub-synchronous controller interaction or resonant overvoltages, which are very common with long AC, high voltage cable schemes. A publicly available report on the subject by myself (in a previous role), for EirGrid can be found here.

The Hornsea technical report states: "...Hornsea One’s systems identified a weak grid and then detected an unusual voltage disturbance which was subsequently discovered to have been caused by the circuit trip, itself caused by a lightning strike."

Whereas the main report states: "The configuration of the Hornsea network, with one SGT and one offshore transmission system user asset (OTSUA Circuit) on outage, was a contributory factor as it created a weak internal network environment. Subsequently Orsted have reviewed and reconfigured their network."

It appears that the configuration of the Hornsea network may have resulted in a sub-synchronous resonance condition on the offshore network which was just waiting for something to excite it causing Sub-Synchronous Oscillations (SSO).

Several types of SSO are possible, namely classical SSR (which includes torsional interaction (TI) and induction generator effect (IGE)), sub-synchronous torsional interactions (SSTI) and sub-synchronous control interactions (SSCI).

These can be summarised as:

SSR-TI: This occurs when a correlation exists between the electrical resonant frequency of the transmission system caused by the series capacitor, and the natural mechanical torsional modes of oscillation of a turbine-generator.

SSR-IGE: This occurs when the total resistance in the series resonant circuit of the generator and the transmission system is negative at sub-synchronous frequencies, creating the potential for negative damping.?

SSTI: These interactions occur between a power electronic controller and the mechanical mass system of a generator. The action of the power electronics controls at sub-synchronous frequencies can cause un-damped or growing torsional oscillation in the generator shaft system. This can also occur for wind turbine shaft systems, although the torsional modes are generally at low frequencies.

SSCI: These interactions occur between any power electronic device (including type 3 wind turbines controls). This is often confused with SSR but in fact does not involve a mechanical interaction of any kind. These oscillations can grow very quickly since they do not involve any mechanical dynamics.

It appears from the interactions of the Hornsea One equipment that SSTI or, perhaps more likely, SSCI occurred when the voltage impulse from the lightning strike excited the subs-synchronous resonance on the offshore network resulting in the turbine controllers and reactive compensation controllers.

The plot above shows that the Hornsea One system was damping oscillations shortly prior to the event. However, it only takes a small changes in network conditions, or indeed system frequency, to change the resonant point to an unstable condition.

It would be easy to criticise the developers of Hornsea One (or perhaps National Grid ESO) for not correctly identifying a potentially onerous condition, but SSO studies, particularly SSTI and SSCI, take a considerable amount of study resource as the response of the system at each frequency (with a small frequency resolution) must be undertaken for a very wide range of transmission system and offshore network configurations and conditions. A study for a single case at a single frequency can take up to an hour using standard, commercially available analysis software, and there are many frequency injections, modes of oscillation, and network configuration scenarios resulting in potentially thousands of cases, the vast majority of which would provide a negative result. The time to set up a properly configured time-domain model of all relevant equipment and their control systems is also significant.

The turbine software and network configurations for Hornsea One have been updated/modified to attempt to prevent this particular condition happening again. But, are there other conditions that may produce the same result? This is one of the major issues in using long, AC cables to connect large, offshore wind farms.

Which brings me back to my main point - would a connection via HVDC have prevented the "de-loading" due to SSO?

I believe it would, as the voltage dip would not have been propagated through the DC converters and the voltage and frequency on the offshore network would have been completely stable. The windfarm would not have "de-loaded" and the loss of generation across the network may have been such that the LFDD protection may not have been required.