What is FCAS? Australian frequency services and measurement

FCAS stands for frequency control ancillary services. They are the reserves that the Australian Energy Market Operator (AEMO) buys to hold frequency in the National Electricity Market (NEM) close to 50 Hz. On the mainland, the normal operating frequency band is 49.85 to 50.15 Hz. After a credible generation or load event, the Frequency Operating Standard requires frequency to stay within 49.5 to 50.5 Hz and to return to the normal band within 5 minutes. FCAS providers hold headroom or footroom for this, and AEMO enables it in each 5-minute dispatch interval together with energy.

There are ten FCAS markets. The very fast (1-second) raise and lower services started on 9 October 2023 with MASS version 8.0, and took the count from eight to ten. The current Market Ancillary Services Specification (MASS) is version 8.2, effective 3 June 2024. It sets the control, measurement and verification rules that this guide cites by section and table.

The ten NEM FCAS markets

A contingency assessment starts at the frequency disturbance time (FDT), when frequency leaves the normal band. It ends at Frequency Recovery or at the maximum period, whichever comes first. Early recovery also releases the slower services. If frequency comes back above 49.9 Hz within 1 s of FDT, the provider does not have to deliver fast, slow or delayed raise for that event. Recovery within 6 s releases slow and delayed raise, and recovery within 60 s releases delayed raise. The lower services use the same rule at 50.1 Hz (MASS Table 4).

How FCAS works: enablement, price and payment

Each FCAS market is a 5-minute spot market. Providers register with AEMO for each service and bid MW quantities in up to ten price bands. The band prices are fixed by 12:30 AEST on the day before; quantities can be rebid. In each dispatch interval, the National Electricity Market Dispatch Engine (NEMDE) enables enough of each service to meet AEMO's requirement, in increasing order of price, and co-optimises the result with energy dispatch. The highest-priced bid that is enabled sets the clearing price for that service in that region.

AEMO pays for enablement, not for energy delivered. For each interval:

Payment = MW enabled × clearing price ($/MW/h) ÷ 12

A battery enabled for 5 MW of fast raise in an interval that clears at $12/MW/h earns 5 × 12 ÷ 12 = $5 for that interval, whether or not frequency moves. AEMO recovers the cost from the parties that create the need. Contingency raise costs go to generators in proportion to their sent-out energy, and contingency lower costs to consumers in proportion to their consumed energy, in the regions concerned. Since 8 June 2025, regulation costs are allocated through Frequency Performance Payments, which pay units that help frequency and charge units that push it the wrong way. That arrangement replaced causer pays.

Regulation and contingency services

Regulation FCAS is central control. AEMO's automatic generation control (AGC) sends raise and lower signals through SCADA on a regular cycle, typically every 4 s. The enabled quantities form a band around the facility's load reference, and AGC moves the facility within that band. Only a centrally dispatched unit, such as a scheduled or semi-scheduled generating unit, a scheduled load or a scheduled bidirectional unit, can offer regulation, because AGC needs a centrally controlled MW level. The control system sends its controlled quantity, status flags and raise and lower limits to AEMO every 4 s with no more than 8 s latency. Its control response delay must stay at or below 150 s (MASS section 7.4). The smallest offer is 1 MW or 1% of the registered maximum, whichever is greater.

Contingency FCAS is local control. AEMO enables a quantity in advance, and the facility then responds to its own measurement of local frequency without further instruction. A facility enabled for both services adds them. MASS section 7.3 requires the plant controller to sum the contingency response and the AGC request. In the MASS example, a 100 MW unit has an AGC set point of 102 MW and a contingency response of 30 MW at 49.5 Hz, so it runs at 132 MW.

What raise and lower mean

A raise response increases generation or decreases consumption, which raises system frequency. For a battery, that is more discharge or less charge. A lower response does the opposite.

The MASS fixes the sign. Generation Amount is active power into the network, as a positive value in MW. Load Amount is active power from the network, as a negative value in MW. Its glossary describes Raise Response as an increase in Generation Amount or a decrease in Load Amount compared with the Initial Value before FDT. For a physical load reduction, the signed Load Amount moves towards zero, so check the arithmetic against the approved analysis convention rather than subtracting the glossary words literally.

Devices on one site rarely share that convention. Take a site that imports 3.0 MW. At a raise event, its battery discharges 2.0 MW and site import falls to 1.0 MW. In MASS terms, the Load Amount moves from -3.0 MW to -1.0 MW, which is a 2.0 MW raise response. A site meter that reports import as positive logs +3.0 MW and then +1.0 MW. A workbook that subtracts those values without conversion shows -2.0 MW, which reads as a 2.0 MW lower response. The inverter can log the same discharge as +2.0 MW or -2.0 MW, depending on the manufacturer. Before the first live event, command a known discharge and trace its sign through every interface to the analysis file.

Static and dynamic frequency response in the MASS

Engineers often call a relay-switched load "static" response and a proportional battery response "dynamic". The MASS uses the terms switching controller and variable controller. Great Britain uses "dynamic" as the name of a family of services, so use the MASS controller type in NEM design documents.

A switching controller delivers a fixed block of FCAS when local frequency crosses its Frequency Setting. On the mainland, the default setting is 49.8 Hz for raise and 50.2 Hz for lower. AEMO can allocate a raise setting from 49.80 Hz down to 49.60 Hz, or a lower setting from 50.20 Hz up to 50.40 Hz. It places larger switched blocks and facilities with higher availability closer to the normal band. The controller must apply its setting with an intrinsic uncertainty of less than 0.05 Hz (MASS section 6.2.2). The Tasmanian defaults are 49.00 Hz and 51.00 Hz.

A variable controller responds in proportion to the frequency deviation. Its deadband can be no wider than the normal band, so a raise response starts at 49.85 Hz at the latest and a lower response at 50.15 Hz. Outside the deadband, output follows the droop function.

Consider a battery enabled for 2 MW of fast raise, with its deadband edge at 49.85 Hz and full output at 49.5 Hz. The 49.5 Hz point is the mainland raise reference frequency, the lower edge of the containment band. The gain is 2 MW / 0.35 Hz = 5.7 MW/Hz. At 49.70 Hz, the battery delivers 0.15 Hz × 5.7 MW/Hz = 0.86 MW of raise.

The MASS sizes offers against a defined frequency ramp from 50 Hz to the reference frequency. The Standard Frequency Ramp on the mainland falls at 0.125 Hz/s. It crosses 49.85 Hz at 1.2 s and reaches 49.5 Hz at 4.0 s, so the battery above reaches full output at 4 s plus its own delays. Very fast FCAS uses the Fast Frequency Ramp of 1 Hz/s. That ramp crosses 49.85 Hz at 0.15 s and reaches 49.5 Hz at 0.5 s. At that rate, each 100 ms of delay is a fifth of the ramp.

Delay builds up at each stage: frequency acquisition and filtering, the controller's calculation, the link to the inverter and the inverter's own ramp. Output then stops at state-of-charge, inverter and site limits. The link is easy to underestimate. At 9,600 baud, a Modbus RTU write of two registers and its reply total 21 bytes, which is about 24 ms on the wire before either device processes it. A facility that uses both controller types must agree with AEMO how the response of each is separated (MASS section 6.5.2).

Distributed energy resources in frequency response markets

Batteries, generators and controllable loads can all provide contingency FCAS. The offer must match what the facility can physically deliver while enabled (MASS section 5.1). State of charge, site import and export limits, inverter rating, process constraints and other operating modes all reduce that figure. For a battery, energy is the usual limit. To hold 5 MW of delayed raise for the full 600 s assessment period, the battery needs 5 MW × 600 s = 0.83 MWh above its minimum state of charge, before losses.

Aggregation adds two measurement rules (MASS section 5.3.3). If an aggregated facility shares a connection point with a variable load or generator, the provider must measure the gross flow of the participating plant directly. Site net power cannot separate a battery's response from a chiller that starts at the same moment. The provider must also align each meter's data to the actual time the disturbance was detected, so that readings from many sites can be summed.

Table 5 relaxes sampling for large aggregations. A fleet of 25 or more facilities with no inertial response, and an initiation delay of 500 ms or less, can sample at 100 ms for very fast FCAS and 200 ms for fast FCAS. AEMO then discounts the registered quantity (MASS sections 6.5.3 and 6.5.4). For very fast FCAS sampled slower than 50 ms but no slower than 100 ms, the discount is 5% for a variable controller and 10% for a switching controller. For fast FCAS from 25 to 499 facilities sampled slower than 50 ms but no slower than 200 ms, the discount is 5%. So a fleet of 25 or more home batteries with a variable controller, sampling at 100 ms and enabled for 3.0 MW of very fast raise, must deliver at least 3.15 MW.

These rules replaced the VPP Demonstrations. From June 2019, AEMO let trial virtual power plants deliver fast FCAS with 1 s measurements at each connection point. That transitional arrangement ended on 30 June 2023 (MASS section 8.3).

FCAS measurement requirements

MASS v8.2 Table 5 sets the sampling rate for local frequency and for power by service and by facility type. Regulation has a power requirement only.

Switching controllers get one exception for fast FCAS. Power can be recorded at 4 s intervals if another measurement at 50 ms or faster fixes the timing of the delivery.

The same table sets accuracy and recording. For very fast and fast FCAS, local frequency needs an intrinsic uncertainty of 0.01 Hz or better and a resolution of 0.0025 Hz or better. Slow and delayed FCAS allow 0.02 Hz and 0.01 Hz. The frequency range must cover at least the operational frequency tolerance band. Power needs an intrinsic uncertainty of 2% or better and a resolution of 0.2% or better over the measurement range. Records must start at least 5 s before FDT and run 60 s after it for very fast and fast FCAS. Slow FCAS needs 20 s before and 5 minutes after, and delayed FCAS needs 20 s before and 10 minutes after. Recording must trigger whenever frequency moves by the Trigger Range, which starts 0.2 Hz either side of 50 Hz on the mainland and 0.8 Hz in Tasmania. All measurements are taken at or close to the connection point, with frequency and power on a common time scale (MASS section 5.3.1).

AEMO accepts a datasheet as evidence when it states intrinsic uncertainty and resolution for power and frequency. If it does not, MASS section 6.5.1 gives three routes. The first is type tests to IEC 61557-12 edition 2.1 by a laboratory accredited by NATA, IANZ or an ILAC MRA signatory, with Annex H where the meter is embedded in an inverter or protection relay. The second is certification to another relevant standard by one of those bodies. The third is test results or other evidence of compliance. An energy accuracy class, such as Class 0.5S to IEC 62053-22, describes energy registration. It does not state intrinsic uncertainty and resolution in the form Table 5 uses.

Records must be digital, in a format that AEMO can analyse with commercial spreadsheet software. The provider keeps them for at least 12 months from the frequency disturbance time and supplies them to AEMO on request (MASS section 5.4). For an aggregated facility, a report on the response of each member is due within 20 business days of AEMO's request (MASS section 4.2.1).

The FCAS verification tool

The FCAS Verification Tool (FCASVT) is AEMO's Excel workbook for calculating contingency FCAS delivery. Version 6.2 calculates all eight contingency services: R1, R6, R60, R5, L1, L6, L60 and L5. Where the tool and the MASS disagree, the MASS prevails.

The calculation follows MASS Table 10. For a scheduled unit, it first builds the reference trajectory that the unit would have followed without the disturbance, and measures the response against it. It then removes any inertial response. The difference from output just before the disturbance is the basic response. For a variable controller, the basic response is compensated for the difference between actual local frequency and the Standard or Fast Frequency Ramp. For a switching controller, it is compensated for the time that frequency took to reach the Frequency Setting, compared with the ramp. Delivery must at least equal the enabled quantity in each service.

The tool can only use what was recorded. It cannot correct a gap in the power record, an offset between the frequency and power clocks, or a record that starts after FDT. Keep the raw files, controller configuration, tool version and output together, so that the result can be reproduced.

FCAS registration

AEMO classifies each facility for the services it can deliver and allocates switching-controller Frequency Settings. It publishes contingency FCAS registration guides for battery energy storage systems and for wind and solar farms, and consulted on revised versions in 2026. The revised battery guide adds testing and measurement requirements for DC-coupled storage. Apply the version in force on the date of the application.

Regulation facilities are retested. The provider must notify AEMO and test at least every 4 years, within 8 weeks after a major overhaul or a change to active power controls, and within 8 weeks of an AEMO request. Results are due within 20 business days of the test (MASS section 7.6).

FCAS with the EpiSensor ZDR

The ZDR is EpiSensor's demand response controller. Its datasheet EPI-211-04 lists three variants:

All three sample frequency every 100 ms or faster and react within 100 ms. The datasheet gives 0.01 Hz frequency resolution and ±0.01 Hz accuracy, and Class 0.5S three-phase power measurement to IEC 62053-22. The relay is a changeover contact rated 4 A resistive at 250 V AC, so a large load is dropped through a contactor that the relay drives.

Against Table 5, the datasheet figures give a clear answer. For slow and delayed contingency FCAS, 100 ms sampling is well inside the 4 s limit, 0.01 Hz resolution meets the 0.01 Hz limit, and ±0.01 Hz is inside 0.02 Hz. For very fast and fast FCAS, a single facility must sample at 50 ms and resolve 0.0025 Hz, and the ZDR's 100 ms and 0.01 Hz meet neither. An aggregation of 25 or more facilities can use the 100 ms and 200 ms rows with the discount, but the 0.0025 Hz resolution still applies. The ZDR's own measurement therefore suits 60-second and 5-minute services. A ZDR-20 or ZDR-21 fits as a switching controller for a load that can drop on a relay, and a ZDR-22 fits as a variable controller for a battery. For power, the datasheet gives an energy class, so the provider needs section 6.5.1 evidence for intrinsic uncertainty and resolution.

The 20 ms GPS-stamped record on the ZDR-21 and ZDR-22 helps align meters across sites, as section 5.3.3 requires. It does not improve the frequency figures, which come from the 100 ms measurement. The ZDR-20 has no event recorder. Its data reaches the Gateway every second, which is inside the 4 s limit for slow and delayed FCAS. The provider must still show that the stored data meets the recording windows, time alignment and 12-month retention.

FCAS, FCESS and frequency response in other markets

FCAS is NEM terminology. Western Australia's Wholesale Electricity Market runs five Frequency Co-optimised Essential System Services (FCESS): regulation raise, regulation lower, contingency reserve raise, contingency reserve lower and a RoCoF control service. Each has its own accreditation. Great Britain procures Dynamic Containment, Dynamic Moderation and Dynamic Regulation. Continental Europe uses Frequency Containment Reserve and automatic and manual Frequency Restoration Reserve. A NEM registration does not carry over to any of them. The dynamic frequency response guide compares these services.

Technical requirements checklist

1. Pick the service and its Table 5 row. A single facility offering R6 needs 50 ms frequency and power sampling and 0.0025 Hz frequency resolution. R60 and R5 allow 4 s and 0.01 Hz.
2. Fix the document set. Record the MASS version (v8.2, effective 3 June 2024), the FCASVT version and the registration guide version in the design, and check them again before the site test.
3. Draw the measurement boundary. Where participating plant shares a connection point with a variable load or generator, meter the gross flow of that plant directly.
4. Prove the sign convention. Command a known discharge or load drop, and confirm that it arrives in the analysis file as a raise response.
5. Align frequency and power in time. Use one clock, or GPS time at every meter in an aggregation, and check the offset against a known event.
6. Set the controller within the rules. A switching controller uses its allocated Frequency Setting (mainland default 49.8 Hz raise, 50.2 Hz lower). A variable controller starts its response by 49.85 Hz and 50.15 Hz.
7. Reserve energy for the assessment period. Delayed FCAS runs to 600 s, which is 0.17 MWh per MW enabled, before losses.
8. Keep the records. Store digital, spreadsheet-readable records for 12 months from each disturbance, and be ready to report within 20 business days.