Robust and Fragile Electricity Systems
By Paul Homewood
The recent South Australian blackout has triggered a debate about the manifest risks of wind farms to the security of electricity networks. National Grid’s 2016/17 Winter Outlook reinforces previous concerns that low-carbon policy mandates are resulting in electricity systems that are likely to be fragile in the face of external shock, and are therefore more difficult and consequently more expensive to manage.
The UK’s National Grid has just published its Winter Outlook for 2016/17, in which it describes the situation this winter as “tight but manageable” (Overview section, p. 14).
The margin of “derated capacity” over expected peak load is roughly 3.4 GW over about 52.7 GW expected peak load, or a margin that National Grid quotes as 6.6%. However, the constitution of this margin both undermines confidence in its resilience, and reminds us that security of supply is increasingly dearly bought in the United Kingdom.
The margin in fact is critically dependent on the 3.5 GW of contingency balancing reserves (defined on p. 14, as “additional capacity held outside the market”, meaning all sorts of odds and ends). It is interesting to note that, on page three, that National Grid observes that some units in the supplemental balancing reserve need more than one day’s notice of operation, which is not encouraging, either for reliability or for cost (plant brought in good time may become surplus to requirements, but will have to be paid in any case).
Excluding the (derated) contingency balancing reserve the margin is only 1% over a peak load of 52.7 GW, ie a margin of about 580 MW, with Loss of Load Expectation of 8.8 hours per year. This is not attractive, and shows how reliant the system has become on expensive contingency balancing reserves.
Furthermore, net interconnector imports are assumed as 2 GW. It must be questionable whether that is a safe assumption. We know from ample empirical evidence in Europe that interconnectors should not be relied upon in a tight corner, since they must reduce transit or even disconnect to protect themselves. In any case, if the market on the other side of the interconnector is also somewhat tight, they may be very little use at all. The news that the French system is likely to be itself experiencing tight capacity margins this winter, due to the safety inspections in the nuclear fleet is a reminder that this is no merely theoretical concern.
National Grid has de-rated grid connected windpower using the arguably generous Equivalent Firm Capacity of 21% (i.e. 0.21 x 10 GW = 2.1 GW). Given the overall narrowness of the margin, an error here could be critical.
Another point of concern is the fact that National Grid appears to have netted the derated capacity of embedded wind generation from the load estimate, i.e. 0.21 x 4.6 GW = approximately 1 GW. Again this could be badly wrong, and in any case, as one engineer has put it to me “This does not capture the combined probabilities of high demands and low availabilities of generation.”
In summary, the Outlook shows that the UK system is heavily dependent on costly contingency balancing reserves, the interconnectors, and on arguably optimistic assumptions about wind. As National Grid’s own summary, “tight but manageable”, suggests, it is now obvious that the UK has a fragile electricity system.
The potential consequences of such fragility, arising for similar reasons, have recently been made painfully evident in South Australia, which suffered a total system blackout on the 28th of September, with full service to all consumers only being restored completely by the 12th of October. The preliminary report and its update by the Australian Energy Market Operator (AEMO) present the current state of knowledge.
What we can infer at present is that the distal cause of the blackout was a state policy that discouraged conventional generation, and resulted in a system that was heavily dependent on wind turbines and on a single interconnector to the neighbouring state of Victoria. Experience has revealed this as a fragile system. The proximal causes of the blackout were a major storm with wind speeds in excess of those forecast. It appears that these winds caused grid damage resulting in voltage disturbances. The wind generators were not programmed to ride through such serious faults, and disconnected, resulting in the loss of 445 MW of generation, about 23% of the system load at that time (1,895 MW). This disconnection transferred the burden to the Victoria interconnector, which could not sustain that load, and itself disconnected to prevent damage, with a further loss of 900 MW of supply. The whole burden now transferred in a split second to the online conventional generation, which needless to say could not meet it and also disconnected, resulting in a black system. The entire event lasted about ninety seconds.
System operators world-wide will now be reviewing their policies on wind turbine Fault Ride Through, and AEMO itself is as a matter of urgency requiring the wind turbine operators to be programmed to provide more robust FRT, but this is more than changing the settings on a dial. Generators disconnect to prevent mechanical damage, which is a real hazard during a fault, so if higher levels of FRT are required, the wind turbines will probably have to be modified to make them less susceptible to gross physical harm in the event of major voltage disturbances. Conventional generation is, as a rule, already engineered to withstand fairly extreme faults, as witness the fact that none of the conventional generation that was online during the South Australian event disconnected until the interconnector came offline. Ensuring that the wind turbines are similarly robust may not be cheap, and there already signs that wind operators in South Australia are reluctant. The chief executive of one of the operators has been quoted Australian Financial Review that he was uncertain whether enhanced Fault Ride Through requirements would expose his equipment to damage: “It could have zero effect, it could have longer term operations and maintenance costs, it could have any number of issues”.
In fact the likelihood of it having zero effect on the wind turbines is small; these devices, already very expensive, will have almost certainly been engineered to be adequate, though not very much more than adequate to the levels of Fault Ride Through generally required. More demanding levels of FRT will thus mean that they are de facto under-engineered, with impacts on maintenance costs and reliability. Re-engineering will not be cheap, and improving the standards for new wind turbines will have a significant effect on their capital cost, pushing the hoped for independence of subsidy still further off into the future.
Read together, National Grid’s UK Winter Outlook and AEMO’s reports on the South Australian case, suggest that systems heavily exposed to wind generation tend to be fragile, and rendering such systems adequately robust is both difficult and, crucially, expensive.