Paying for primary frequency response – Finally some light at the end of a long tunnel

For the last five years the Australian Energy Council (AEC) has been engaging in a long saga about how to maintain the National Electricity Market’s frequency close to 50 Hertz throughout our energy transition. This relates to managing frequency in normal conditions, i.e. continuously adjusting the NEM’s power balance to account for the natural random variations in demand and wind and solar output that occur between each five minute iteration of the dispatch engine.

By the mid 2010s, the NEM’s frequency performance had become poor and the industry was discussing new Frequency Control Ancillary Services (FCAS) markets to incentivise generators and batteries to fix it. Following a power system event on 25 August 2018, the Australian Energy Market Operator (AEMO) decided it had had enough of waiting for markets and successfully proposed a rule change obliging all capable generators and batteries to provide Primary Frequency Response (PFR) for free.

Whilst this rule delivered a much tighter frequency, it is seen as unfair, and, because it disincentivises investment and performance, unlikely to be sustainable in the long-term.

One of the FCAS designs the AEC was researching was “Double-Sided Causer Pays” (DSCP), which would have incentivised generators and batteries to voluntarily provide PFR. In 2022 the Australian Energy Market Commission (AEMC) adapted DSCP into the Frequency Performance Payments (FPP) rule for implementation in 2025. Whilst PFR is to remain mandatory, FPP compensates PFR providers in a way that broadly simulates a voluntary market.

AEMO is now designing the detailed implementation of the FPP and has performed some preliminary modelling of its likely outcomes. This work provides some early comfort, that, although a voluntary market was always the industry’s first choice, a FPP compensation regime might be the next best thing.

We take a look at this modelling and how it came to be.

How we got here  

The seeds of this saga go all the way back to 2001 when the FCAS markets were created, and when, consistent with the philosophy of efficient markets, mandatory obligations upon generators to support frequency control fell away. Instead, AEMO seeks competitive offers from all technologies to voluntarily provide FCAS, with only the cheapest providers used.

Those FCAS markets included six contingency services which pay providers for being ready to stabilise frequency after a major event. These work well and will soon be extended to eight.

It also included two regulation services that pay generators for receiving Automatic Generation Control (AGC) signals from AEMO in order to fine-tune the frequency. Because it uses a control system via AEMO, this is called secondary frequency control. It has a lag of at least 20 seconds between AEMO’s observation of a frequency deviation and its correction by a generator or battery.

A major oversight at the time was the need to also recruit primary fine-tune frequency control. PFR means action occurring directly within a generator/battery’s governor, which is much quicker. Nevertheless, for the first 10 years of the NEM the frequency characteristic remained quite good, and, at first glance, the NEM appeared to be disproving power system engineering theory by safely operating without fine tune PFR.

But sadly, that engineering theory ultimately proved correct. The NEM had inherited a lot of generation with legacy governors for whom PFR could not be switched off and so were in fact voluntarily providing it pro bono. As these governors were progressively upgraded to modern designs, the owners deactivated the response which the FCAS market permitted, and indeed intended, them to do.

This withdrawal of PFR supply coincided with an increase in the need for it. There had always been random movements in customer loads that required PFR correction, but the growth of stochastic wind and solar generation meant there were progressively ever more swings in supply/demand that had to be corrected. The result was an ever-uglier frequency characteristic.

Figure 1: Decline in frequency performance through the 2010s

Source: AEMC Frequency Control Frameworks Review

Fixing the Frequency Characteristic

Between 2015 to 2018 the industry was aware of the issue and much discussed market mechanisms that might correct it, but without much urgency. An AEMC review completed in July 2018 with the uninspiring conclusion of awaiting further research.

When a power system event islanded Queensland on 25 August 2018, the frequency was observed to take a very long time to stabilise. In response AEMO made a major change in gear by insisting that PFR become a mandatory service from all large batteries and generators (including renewable generators) as is the case in some overseas markets. Despite it clashing with the NEM’s market philosophy, given the sense of urgency that prevailed, AEMO’s rule was made, first temporarily and then enduring.

With such a wide-reaching, and, frankly oppressive, obligation to provide PFR, it is not surprising that the frequency is now much closer to 50 Hertz.

Figure 2: Monthly frequency distribution – January 2007 to January 2022

Source: AEMC Rule Determination PFR Incentives

Paying for Primary Frequency Response

Well before these events there had been different designs proposed to incentivise PFR.

The AEC supported research into the DSCP approach that initially appeared radical, and then as time progressed, became more mainstream. In 2021 the AEC and its membership managed a major investigation by Intelligent Energy Systems (IES), supported by a grant from the Australian Renewable Energy Agency (ARENA). This approach automatically rewards those who correct the frequency with PFR whilst recovering costs from those who distort it.

Figure 3: Causer Pays Deviation Quantities

Source: AEMO Causer Pays Procedure

It was designed to operate as a self-correcting voluntary regime, giving strong incentives for PFR when frequency performance is poor which weaken when performance is good. But with mandatory PFR, frequency performance is always good, so it could produce no meaningful incentive.

The AEMC recognised that a mandatory obligation without payment was unsustainable. Apart from obvious questions of fairness, a signal to invest in new PFR was needed as the dominant legacy providers, coal-fired generators, declined.

At this point the AEMC engaged IES themselves who adapted the DSCP into a new mechanism designed to fairly pay for PFR in the presence of mandatory PFR. This FPP design instead pays providers based on:

• The amount of “work” providers are observed to be doing to correct the frequency (rather than the frequency outcome itself), and
• The price of the Regulation FCAS service, as a proxy of the value that a competitive market would produce for this work.

This is an elegant theoretical compromise to:

• produce a financial outcome emulating a competitive market, thereby keeping providers happy, and
• retain mandatory PFR, thereby keeping AEMO happy.

This design was codified for implementation in 2025.

AEMO Implementation studies

AEMO is now consulting on the detailed implementation design of FPP. This requires intricate decisions on technical parameters, such as how to reasonably measure PFR provider performance.

To assist these decisions, AEMO has performed some interesting “backcasting” modelling. AEMO has looked at a period in recent history and applied the FPP calculations as if the rule was already operating. This has produced an estimate of FPP turnover that gives some confidence that payments will in fact reasonably replicate a voluntary, paid market.

As the existing FCAS Regulation market is used as a proxy for the value of PFR, the total gross turnover of the new FPPs will closely, but not exactly, match the Regulation market turnover.

Figure 4: Backcasting of gross payment amounts for Regulation vs FPP: Mainland Raise

Source: AEMO

During the backcast period above, the total amount of gross payments for Mainland and Global[1] requirements were just over $14m for FCAS Regulation whilst backcast FPP were just over $15m.

However, PFR providers have both positive payments (when they help correct the frequency) and negative payments (when, for whatever reason they are seen to be distorting the frequency). After netting these payments within each dispatchable unit, the total FPP turnover fell to about $5.5m.

Whilst FCAS Regulation provided a proxy value for FPP, the split of recipients of the payments is quite different.

For example, batteries, as excellent frequency responders, receive the strongest incentive from both arrangements. However, because AEMO buys only a limited amount of FCAS regulation, and despite their (currently) small installed capacity, batteries are competitively taking a large share of this market. In FPP the benefits are more evenly shared with other mandatory providers of PFR.  

Figure 5: Net settlement in FPP and Regulation aggregated by type: 20 Jul to 10 Oct 21

Source: AEMO. 

Note: “Residual” is settlement allocated to parties without detailed metering, which is mostly customers.

Conclusion

Whilst the NEM clearly needed to recruit PFR to restabilise its frequency characteristic, the AEC was highly critical of the mandatory approach that was taken in the aftermath of the August 2018 incident. In the AEC’s view, PFR should have been recruited through a competitive, voluntary market, which would have led to it being provided by the least-cost suite of PFR performers to the level needed to deliver a secure system. Instead, with all providers forced to deliver PFR, the system has gone to the other extreme of an unnecessarily tight frequency characteristic, the cost of which is hidden as an inefficient burden of participation in the NEM for generators and batteries.

But there is a silver lining to this cloud. An approach previously championed by the AEC was subsequently adapted by the AEMC into a mechanism that attempts to replicate the payments that would have resulted from such a competitive market. This is not a straightforward task, but early backcast modelling by AEMO gives some ground for optimism that it might achieve its intent.

This could be thought of as a kind of compensation for mandatory PFR, as well as providing an incentive to perform and invest. Whilst clearly a competitive voluntary procurement of PFR would be a better way to stabilise frequency, a payment that simulates that competitive procurement might be the next best thing.

It is also very satisfying that the AEC, amongst others, played a key role in the extended saga that is now bringing FPP to fruition.

[1] “Global” includes Tasmanian payments excluding local Tasmanian payments