The biggest threat to network stability in Australia’s electricity grid is not from an increase in renewables such as wind and solar, but the control settings on the dominant fleet of “synchronous generators” – namely the country’s coal and gas fleet.
These conclusions come in a significant new study – submitted to government authorities, network operators and the market rule maker – that could help turn the current debate about renewables versus fossil baseload power on its head.
The report – prepared by Kate Summers, an electrical engineer with Pacific Hydro, and energy systems expert Bruce Miller – is important because it suggests that the biggest threat to Australia’s energy security comes not from an increase in wind and solar, but from the control settings of the coal and gas plants that have dominated the grid for the last few decades.
It is the result, the authors say, of an economist-driven decision at the start of the National Electricity Market in 1999 to adopt a market-based system for ancillary services, in contrast to nearly every other market where frequency services are mandated with fixed contracts.
The result is a perverse set of incentives and penalties that has resulted in the “deadbands” of the crucial control systems either being relaxed or switched off, leaving the grid effectively at the mercy of unforeseen events, because the assumed back-up or quick response is either too slow or non-existent.
It says these settings leave the power system exposed, particularly to a series of successive small events, and may have been a contributing factor to the major September 28 blackout in South Australia, even though the issue was not even considered in the subsequent analysis by the market operator.
It also left the NSW system perilously close to failure in the recent heatwave, because there was effectively little or no frequency control support within the state’s grid.
And because fossil fuel generators now have limited response or are unable to respond quickly to changes in frequency, the market operator is sometimes left with only one safety net option: load shedding of the sort that has caused such controversy in recent months.
“This should be a wake-up call that there is something seriously wrong in the NEM (National Electricity Market),” the report says. To illustrate this, it cites an event in South Australia in November 2015, when the inter-connector to Victoria tripped and local load shedding occurred.
Far from keeping the frequency controlled, the coal and gas generators – with the absence of appropriate governor controls – were pushing the frequency around, delaying the re-synchronisation with Victoria.
(Note: the green line illustrates where inter-connector tripped, the grey area is the frequency standards and the black line is the poor frequency control with the islanded region. The second deviation in frequency was a result of the “frequency control” services.)
And the report’s authors are surprised that this issue has not been even considered in analysis of the recent “system black” event.
“Given that a region has suffered a system black and the analysis to date has failed to question the logic of allowing local governors to be disabled or detuned to be made unresponsive, it is a sign that the prior emphasis on control philosophy has been lost,” it writes.
Not only have the regulators become cavalier on energy security, they are playing roulette by allowing the market in some cases to source all spinning reserve on the other side of interconnections, which can be lost, as it was in the September 28 blackout.
The issue has come to the attention of a small group of energy experts over the last 18 months, but has grown in importance during the recent events in South Australia and NSW as energy systems struggled, or failed, to deal with extreme weather events.
It is now a mainstream problem, given that wind and solar is being blamed for current and future system instability, whereas this report suggests the real problems lie with the control settings of existing baseload generation.
It is yet another argument that Australia’s electricity grid has been allowed to drift, and getting less responsive, by regulators and policy makers focused on economic theory and not good control engineering. This has led to sizeable inefficiencies.
The best analogy used by the paper’s authors is that of a truck approaching a hill. Normally, the driver would press the accelerator to make sure it had the momentum to get up the hill.
But in Australia’s electricity market, the driver is penalised for doing so. And by the time the market signal comes for him to take action, it is already too late – he is half way up the hill, the momentum has been lost, and he has to work the truck’s engine even harder to meet the challenge.
In electricity market terms, it means that generators have also got to work harder to maintain grid stability. And because the response has been delayed, grid stability is at risk, because the network is at greater risk of stalling.
“Integration of renewable energy has increased …. and the deterioration of the frequency control is often blamed on asynchronous machines associated with farm generation,” the report notes.
And, like in the film Hidden Figures, where a group of black female maths geniuses upset the all white male experts in NASA by pointing out that they had their maths wrong, and that their projections for the first manned flight would fail, this report suggests the maths supporting Australia electricity system’s constraint limits may also be wrong.
“The real-time market management of frequency services has downgraded the safety nets that were designed to avoid system collapse and may mean that existing system constraint limits are incorrectly calculated,” it says.
It blames more than 15 years of economic decision-making driven by inefficient market signals, and a deviation from “good electricity industry practice, resulting in more costs and less reliability.”
Two things happened as a result of the introduction of a separate market for FCAS – or frequency control and ancillary services – which is designed to keep the system operating within a frequency of 49.5 Hertz to 50.5Hz.
The first was the relaxation of the “deadband settings”, which had been tightly fixed at less than 0.1Hz, but this requirement was moved out to 0.3Hz or larger. That may appear to be a small adjustment, the report notes, but it is a significant change in the timing and response of synchronous units to contingency events.
It is, for example, 10 times the size of the mandatory governor deadband requirement in the UK.
Like that truck going up the hill, it means that much more contingency service is required to arrest a fall in frequency that would normally be required. This, the authors say, is an inefficient outcome and contradictory to the principles of the market.
That might make a significant pay-day for these units, but it could also mean that the cavalry arrives too late. Compounding this problem is a relaxation of the normal operating bandwidth, which meant that the governing response of the synchronous generators are further delayed and commence outside of the normal operating frequency band.
The report draws on examples taken from the recent System Black in South Australia, where tornadoes tearing down three main transmission lines triggered a sequence of events that led to a state-wide blackout.
The report says the removal of the primary governing response from within the normal operating band means that there is no primary governor control available for smaller contingent events, such as a 100MW loss of generation.
These events do not cause the frequency to fall low enough to trigger governor action, the response to these events is purely inertial and then controlled via second order regulation response only.
“This is a significant departure from the pre-market control philosophy, and it leaves the power system exposed when several smaller contingent events occur in quick succession such as occurred in South Australia on 28 September 2016.”
It cites this graph, showing the response of the TIPS B generator in South Australia as an example. It shows that when frequency plunged, the generators was still shedding power, because that was what the market was telling it to do. Because it no longer had a tight deadband about its response to frequency changes, it effectively provided “no response to the step changes”, the report noted.
“Examine T=9s and 14s the response of the unit to the event is defeated within a second the unit has returned to its pre-contingent loading, that is it provides no response to the step changes. The governors are disabled. Even during the collapse the unit is reducing its output in accordance with the market design. This illustrates that there was no spinning reserve was being held within South Australia, it was only available on the east side of the interconnector.”
And readers will remember that because the market operator had taken no precautions in the face of the upcoming storms, and was unaware of the ride through settings of wind farms, the blackout ensued.
Indeed, another report from RES and Lloyds Register also pointed to potential problems with gas generators: and suggested that if inverter controlled solar and storage had been in place instead of gas generators, then the blackout may have been avoided.
But while all the focus in the post blackout analysis was on the wind farm settings – which, incidentally, have been fixed, as was illustrated in last Friday’s gas generator failure when the grid rode through an even greater loss of generation, and an even greater overload on the inter-connector – the issue around fossil fuel generators has been ignored.
“Without mandatory governor deadbands it is no longer possible for system planners to rely on the response of the synchronous units as the response (and location) will depend on the market enablement,” it notes.
In other words, it is a market mechanism similar to the one that was unable to awaken the 200MW Pelican Point gas-fired power station from its slumber when the market operator got its weather forecasts wrong in the recent heatwave, causing thousands of customers to lose power.
“We do not have adequate control engineering in the biggest infrastructure assets in the eastern states. We’ve got economists and lawyers everywhere, but they have displaced engineering nous.”
“There is a real and immediate control problem on synchronous generator units.”
And the solution? The reports authors suggest that switching off governors, as most synchronous generators do to avoid perverse penalties in the market-based system, should not be allowed. This idea has gotten support in some ministerial and regulatory circles.
“Given that a region has suffered a system black and the analysis to date has failed to question the logic of allowing local governors to be disabled or detuned to be made unresponsive, is a sign that the prior emphasis on control philosophy has been lost.
Disablement of governors within a region means spinning reserve is being held in the other regions and the security becomes more dependent on the interconnection.
When the deadbands are wide, successive small contingencies which were once easily controlled are able to cause an interconnector to exceed the protection settings for angular stability.
For these and other reasons the FCAS markets require considerable redesign so that primary control can be re-instated within a safer and up until recently a more normal operating band.”
And the final irony is that it is not the wind turbines or solar PV inverter installations that are affected by the poor frequency control – it is the synchronous units themselves – which means that the deterioration of frequency control represents a real risk to 94 per cent of the power system.
Synchronous units, the authors suggest, are at risk of serious or catastrophic damage when undamped frequency oscillations occur or when the controls on the unit act contrary to the system forces.
Individually the generators have become used to simply following their dispatch targets (as the market has incentivised them to do so) and not having active frequency control.
This reduces some minor wear and tear that occurs with using constant governor control, but the problem now is that the practise is widespread and, as a result, the frequency has significantly deteriorated, causing undesirable and unwanted consequences.