The Australian Energy Regulator is considering options to impose more controls over the dispatch of wind and solar farms, after alarms were raised over the increasing number of wind and solar projects that suddenly withdraw capacity when wholesale prices are negative.
In the last year or two an increasing number of wind and solar facilities have introduced automated software and trading systems that switch off the generators according to certain algorithms – the most common being the growing number of occasions when prices go negative. Many wind and solar projects are required – in the terms of their power purchase agreements with customers – to switch off when that occurs.
The wind and solar farms can do this under current rules because are what is known as “semi-scheduled”, and required only to observe a cap on output, not a floor. But the Australian Energy Market Operator is not pleased. The sudden loss of output requires other generators providing frequency control to fill in the gaps. Normally they would do this only for sudden changes in demand, or the unexpected failure of a big generator.
The market operator is now fearful of what may occur once the share of wind and solar capacity increases from the current capacity of more than 8,000MW to levels of 30,000MW and up to 45,000MW or more in the next two decades, or even to instantaneous penetration levels of up to 75 per cent or even 100 per cent over the next five years. It worries that the current rules may prevent it from properly managing the system.
So the Australian Energy Regulator has been asked to canvass options and has published an issues paper outlining its thinking to date and how wind and solar farms can be made to follow dispatch targets in the same way that fully scheduled generators do.
The solution, whichever one may be finally agreed, will be one of the biggest changes to market rules in decades, and is likely to have implications for wind and solar farms, their contracts, and the possible need to add battery storage. It will be a debate fiercely contested by many in the sector.
One of the more immediate responses was that the switch to five-minute settlement should largely fix the issue, because there will be no incentive for wind and solar farms to crash out of the market.
But the very same institutions worried about this situation are proposing a 12-month delay to 5MS, despite many saying that the market – having been given four years notice – is or should be ready and any further push back will delay the clean energy transition.
Several proposed options are considered difficult or impractical, but have not been entirely ruled out. These include changes to the “causer pays” factor, the removal of the “semi-scheduled” category, or the outright banning of automated systems that react to price movements, rather than resource availability.
Three other options that are being pursued reflect possible changes to the semi-scheduled category.
One may be a “megawatt” target and a ramp rate similar to that imposed on scheduled generators (coal, gas and hydro), that would take into account the availability of the wind or solar resource. The second option is an “energy target” within a five minute period, while the third is operating as an “inflexible generator” that operates on a fixed dispatch target.
“Semi scheduled generators should be expected to follow their dispatch instructions linearly across a dispatch interval, subject to the availability of the resource, unless there is a demonstrable hazard to their physical generation or personnel or as part of providing a system service,” the paper says.
This graph above is provided as an example of the problem. The orange line is the target dispatch according to its bidding. The blue line is what this particular wind farm actually delivered. The grey line is the price. It shows the wind farm was switched off earlier than its target in response to the falling prices.
“During this period the dispatch price in the region ranged from around $36/MWh to – $1000/MWh,” the paper says. “In this case the participant changed their output more rapidly than was represented in the ramp rate in their offer. Supporting wind data indicates that the observed reductions were unrelated to fluctuations in the wind resource.”
It is common practice for generators – be they wind farms, solar farms, or even coal generators – to bid at the price floor (minus $1,000/MWh) to ensure they get priority over dispatch. The actual prices are actually decided by the price of the last megawatt dispatched.
That can be manipulated according to many factors and bidding strategies, which in turn can depend on a number of different factors -demand, supply, availability, contracts and hedges. Once locked in, fully scheduled generators must deliver – with danger to life and equipment being the only excuse. The semi-scheduled generators have no such obligations.
This next graph is another example of the difference between target and delivery of a wind farm. Again, the orange line is the target, the blue line the delivery. The grey line represents frequency variations the market operator says is caused by the difference.
“In this example, the early rapid reduction to 0 MW, before a corresponding dispatch instruction was issued with a 0 MW target, appears unrelated to resource availability or technical limitations. SCADA7 data indicates the number of turbines available remained relatively constant and the wind speed fell only slightly,” the paper says.
The work is also being combined with another proposed rule change that would require wind and solar farms to be more forthcoming about changes to anticipated output due to physical factors, such as the extreme heat that caused mass “high temperature” cut outs among wind farms in Victoria and South Australia at times in the last summer.
What will be interesting is to see how battery storage emerges as a complication, or a potential solution to the issue. At the moment it is seen as a compilation, AEMO prefers to register batteries separately, but the regulator recognises that this may not always be possible or feasible from a developer’s perspective.
The existence of a co-located battery means that a wind developer can divert the output to the storage facility if prices go negative, rather than switch the facility off. But it doesn’t necessarily help AEMO’s management of the grid in those situations.
On the other hand, the paper suggests, it could help the wind or solar plant be more predictable in its output. It suggests that this could lead to a change in category.
“If these hybrid systems intend to operate differently to that based on the availability of the renewable resource, then perhaps they are more suited to being classified as scheduled as their output is not then directly linked to the availability of an intermittent resource and they can predict and inform AEMO of their output capabilities with greater certainty.”
How common is this problem? InSouth Australia, the paper says, the number of events where the output of wind and solar farms were more than 20MW has increased dramatically over the last 12 months.
At this stage, most of these variations are not exceeding the capabilities of the system,” the paper says.
“However, if a number of semi scheduled generators were to reduce their output, during a dispatch interval (which could be equivalent to, or more than, the magnitude of a credible contingency event), this would place the power system at a greater level of risk than for credible contingency events that are currently considered. This concern is compounded by the forecast growth in the volume of semi scheduled generation in the future.
“While the number of occasions where semi scheduled generators shut down during negative priced periods is currently relatively low, it has grown and could continue to increase while their dispatch obligations remain less stringent compared to scheduled generators.”
Have any thoughts about this initiative? Please either add to comments section below, or email [email protected]
