The previous article in this series explained that FPP-Reg-FCAS Recovery was designed so units that cause dispatch errors pay, while units that correct those errors are compensated. This was targeted at removing a cross-subsidy between those units causing dispatch errors and those units that would provide corrective primary response after implementing MPFR but not be paid for it. Please note that those units providing secondary response could choose to do so through their FCAS offer prices. This article explains how that works using 4-second data from AEMO’s FPP table in EMMS.

Dispatch errors occur

Quickly recapping dispatch errors, and whether they are good or bad, the following screenshots show the dispatch errors of a battery while charging. The first shows the scheduled dispatch instructions compared to measured SCADA data. The unit was scheduled to charge 200 MW, but at times charged only 150 MW, leading to an error of 50 MW. Dispatch errors are a normal feature of power system operation. Forecast errors, changing demand, unit ramping limitations and plant performance mean actual output rarely follows an ideal trajectory exactly. The important question is not whether dispatch errors exist, but how quickly they are corrected and who pays for the corrective work.

 

MPFR and AGC respond

MPFR provides the fast, local response that stabilises frequency, while AGC uses units enabled for Regulation FCAS to correct the remaining dispatch error and bring the system back toward balance. Primary response is immediate and proportional to local frequency deviation. Secondary response is slower and centralised, because AGC must receive measurements, process them, and send instructions to enabled units.

Primary response alone cannot fully rebalance the system. Because primary control responds in proportion to frequency deviation, a persisting dispatch error can leave primary controllers doing work while frequency remains away from 50 Hz. Secondary control, through AGC and Regulation FCAS, is therefore needed to correct dispatch errors, relieve the work being done by primary controllers, and restore energy balance.

Once a dispatch error exists, MPFR and AGC (using Regulation FCAS) work together to correct it. FPP-Reg-FCAS Recovery was designed to measure which units helped that correction and which units made the error worse.

How FPP identifies whether dispatch errors were good/bad

FPP does not simply ask whether a unit deviated from target. It asks whether the deviation helped or harmed frequency control at that point in time.

These error values are compared against the Frequency Measure (FM), which is shown in the next chart.

The FM is based on system frequency and represents a smoothed, balanced frequency signal. It was picked so units that implemented primary control would be reasonably confident they would be adjudged as good. Given the scale of the errors on this unit, and it was enabled for Regulation FCAS this is predominantly AGC signalling the unit. This is because with frequency close to 50 Hz, MPFR alone would not have required such a response, although the primary response can be quite significant if the unit has an aggressive droop. Droop is the setting that determines how strongly a unit responds to frequency deviation, in practical terms, how many MW of output change occur for each Hz change in system frequency. A lower droop setting means a larger MW response for the same frequency change. If the error is aligned with the FM, then it is good, otherwise bad. For this unit most of the dispatch errors were aligned with the FM and flagged as good.

How FPP measures corrective errors

Under FPP, the relevant regions, and the units within them, are defined by the applicable FCAS constraint. The following charts focus on Mainland and Global constraints, with Global also including Tasmania. They therefore exclude Tasmania-only constraints because Mainland and Tasmania are operated as two independent AGC control areas, connected by the frequency-sensitive HVDC link, Basslink. The chart overlays five-minute FCAS quantities, with LOWERREG always shown as a negative value. FPP separates measured deviations from generating units, loads and other participants, and then applies a residual category to balance the calculation. This residual mainly represents unscheduled quantities, such as demand forecast error and demand variation within the interval.

FPP classifies deviations as good or bad using the four-second Frequency Measure. The next screenshot shows these errors, the residual, and Regulation FCAS quantities. For completeness it also shows Requirement for Corrective Response (RCR) as the maximum MW value of good error in a dispatch interval, which is why RCR follows the peak of the deviation in the chart. RCR can be viewed as a measure of the volume of dispatch error that was corrected during the dispatch interval. It is fundamental to FPP because it determines the volume of corrective work that must be paid for.

In effect, RCR is the MW quantity of dispatch error corrected during that interval that must be paid for. At this stage, avoid comparing RCR to the Raise or Lower Regulation requirement. RCR is a gross measure of the dispatch error to ensure adequate settlements, while AGC-Regulation FCAS is used to correct the persisting net dispatch error.

How Reg-FCAS Recovery settlement occurs

Reg-FCAS Recovery recovers the cost of Regulation FCAS using the same performance snapshot as FPP, but only for the proportion of the Regulation FCAS deemed to have been used in the five-minute interval. As explained earlier, in some intervals, the Frequency Measure may be positive throughout, so there are only good deviations in the positive direction. In that case, the Lower RCR is zero and there is no current-interval data for allocating LOWERREG costs through Reg-FCAS Recovery. When this occurs, the relevant Regulation FCAS costs are recovered using a longer-run historical sample of performance data, known as Default Contribution Factors or DCFs.

The next chart shows used and unused dispatch errors, but only for units enabled for Regulation FCAS. It shows how Usage is calculated. Ideally, Regulation-enabled units should account for a substantial share of good deviations because AGC targets them to correct the underlying dispatch error, and because they have reserved capacity doing so that can also contribute to primary response. This was part of the design: if units enabled for Regulation FCAS earn more under FPP when controlled by AGC, they should have an incentive to lower their Regulation FCAS offer prices.

The following chart shows a detailed 4-second snapshot showing dispatch-error correction by Regulation-enabled units, compared to enabled and used quantities.

In this chart there are two cohorts of units: those enabled for Raise and those enabled for Lower, with some units appearing in both cohorts. The net error for each cohort is plotted as a grey line. To identify the gross good errors of enabled units, the chart plots raise-enabled units when FM > 0 and lower-enabled units when FM < 0. These errors are shown in green, both capped and uncapped by Regulation FCAS enablement. As shown, the capped good errors align with Usage. Usage is quite high for Raise, but not for Lower, meaning Regulation FCAS Recovery will allocate costs by historical DCFs instead of real-time FPP factors for that interval.

 

The following chart shows average values per dispatch interval over Q2 2026. RCR sets the volume in FPP transactions, enablement is the volume for Reg-FCAS Recovery, and used is the amount settled by real-time performance calculated in that interval. Unused is the amount recovered through historical DCFs. Note the variation between usage and enablement, particularly for lower regulation, indicating that much Reg-FCAS Recovery is settled using longer-run performance.

Settlement calculations and interval prices

Total settlement amounts are calculated for each five-minute dispatch interval. For FPP, the settlement value is based on RCR divided by 12, multiplied by the Regulation price. For Reg-FCAS Recovery, it is based on the cost of the relevant Regulation FCAS constraint, calculated from the enabled MW quantity and the marginal value of the constraint. These interval values produce a dollar amount that can be divided by good dispatch-error volume to calculate a $/MWh price for correcting dispatch error in each direction. Because the denominator can be highly variable, being the MWh errors in each interval, and the numerator is a dollar value based on a MW value, being RCR or Regulation enablement for FPP and Reg-FCAS Recovery respectively, the $/MWh values can vary greatly. Despite this, the marginal rate of the error can be priced.

The following chart shows this dynamic, with raise and lower separately presented. The $/MWh marginal prices for each dispatch interval are shown as stacked columns, with FPP and Reg-FCAS Recovery being additive for a bad performer, and only FPP applying for a good-performing unit. As can be seen from the chart, the errors in MWh are highly variable and lead to varying $/MWh marginal prices for these errors. However, a physical trader could assess this data during the trading day and respond to changing conditions — for example, by focusing on Regulation FCAS rather than energy because correction and usage are high.

These are more tractable over a longer time series, as shown in the following chart, which presents volume-weighted averages of FPP and Reg-FCAS Recovery, and compares them to the relevant raise and lower FCAS Regulation prices. As usual, the chart presents lower as negative for charting purposes only. Because this is a longer time series, a $/MWh value is also calculated for “Unused” across the whole period, being ∑Unused regulation costs divided by ∑ErrorMWh. This is an adequate way of representing the Default Contribution Factor approach, but has the benefit of showing when unused costs tend to accrue, albeit poor performance in any interval of the day contributes to a DCF.

The $/MWh rates are quite high, but remember this value is only for the measured ERROR (from scheduled trajectory) and not the full generation.  These values should be accounted for when considering Regulation FCAS bid prices, in addition to the FCAS price, the RRP and the unit’s marginal costs, and opportunity costs of not generating (or charging) at that moment to make FCAS available.  These calculations are probably suitable for a separate article, but these concepts informed the design of FPP and were originally set out in this CS Energy submission from 2019 to the MPFR rule.

Taken together, the FPP and RTCP calculations provide a practical way to translate dispatch-error performance into settlement outcomes. They show which units are contributing to or correcting frequency deviations, and how much those actions are worth in each interval.

Article 2 has shown how dispatch errors become measured performance, how corrective response is quantified, and how settlement outcomes are created. The next article examines the first year of practical outcomes, including which technologies received credits, which technologies paid charges, and whether the new arrangements appear to be changing behaviour.

 

---

About our Guest Author

---