Australia's Power System Frequency: Current Situation, Industrial Challenges, Efforts, and Future Research Directions

This industry-oriented article presents an overview and in-depth analysis of previous and current situations of power system frequency response and control in the Australian National Electricity Market (NEM) power system. It provides a comprehensive understanding of frequency support services in the NEM and their corresponding market and energy governance bodies and control mechanisms. The evaluation of different services and frequency control ancillary services (FCASs) provided by different electricity market players under the supervision of the Australian Energy Market Operator (AEMO) as an independent system operator provides lessons and an understanding of the current operation status with its real challenges and opportunities from frequency stability and security perspectives. Based on the evaluation of the current situation and future national planning, such as the integrated system plan (ISP), a number of research gaps and industrial technical issues are identified. The article also presents perspectives on future research directions to help the power system transformation toward almost 100% renewable-rich and more secure energy systems.


I. INTRODUCTION
M OTIVATED by energy security risks, environmental concerns, and issues related to fossil energy sources, Australia has recently approved a national plan, the wholeof-economy Long-Term Emissions Reduction Plan, to decarbonise the energy system and reach the net zero targets by 2050 [1].Therefore, it is essential to upgrade the existing power system to achieve Australia's goals.This includes increasing the hosting capacity for renewable energy sources (RESs) and inverter-based resources (IBRs) in the existing system to reach an almost 100% RESs contribution to electrification in Australia, upgrading the existing infrastructure to adapt to the high penetration of RESs, incorporating demand-side management (DSM) and emerging energy storage technologies in providing essential and ancillary services, and enabling the integration with e-mobility systems and other energy sectors.These necessary ongoing changes are bringing new opportunities and also various challenges to the Australian Energy Market Operator (AEMO), especially from a frequency stability and security viewpoint [2].
Australian power systems are rapidly progressing to reach almost 100% renewable-rich systems concept.Due to the generous incentive programs, Australia is currently at the forefront of the world in the installed rooftop solar photovoltaic (PV) per person, known internationally as PV per capita metric.Likewise, it has been expected the whole NEM will reach 100% renewable instantaneous operations in 2025.In 2020, 24% of the total electricity generation in Australia was from renewable energy sources, including solar (9%), wind (9%), and hydro (6%) [3], [4].The share of RESs in the total Australian electricity generation in 2020 was the highest since levels were first recorded in the mid-1960 s.The South Australia Power Region, as part of the National Electricity Market (NEM) power system, reached 100% instantaneous penetration from distributed photovoltaic renewable energy almost every day in October 2021.It is worth mentioning that on 27 November 2021, almost 150% of the SA's middle-of-the-day electricity needs were supplied by wind and solar.During this period, negative electricity demand was recorded in South Australia [4], [5].
It has been verified that increasing RES penetration and replacing conventional synchronous generating units with RESs significantly reduce the total rotating inertia, which directly affects the overall power system frequency stability and security.With a high power share from RESs, the frequency variation and fluctuation would increase, potentially resulting in frequency instability issues [6].Similarly, a significant increase in intermittency sources would result in a substantial increase in frequency and power flow fluctuations and variations during normal operational periods and high deviations in both frequency and rate of change of frequency (RoCoF) during large disturbances and contingency operation periods [7].These issues will become more complicated, especially in power systems with an increased share of small-scale RES-based distributed generating (DG) units in their distribution levels, like Australia's power system, due to the fact that the source of uncertainties, unknown inputs, and power variations in such systems would be related to both the demand-side and the power generation-side [8], [9].To solve these issues, services based on more flexible reserve sources and emerging technologies, such as virtual inertia, demand response (DR), energy storage systems (ESSs), virtual power plant (VPP), and aggregated electric vehicles (EVs), are theoretically suggested in the literature and research projects to support frequency, taking into account uncertainties and unknown inputs [7].However, the implementation of the recommended solutions is still in its infancy at the current time and needs advanced infrastructure and more research.Since most of these emerging solutions would be built based on aggregation methods using cyber and physical layers in the control centers by the independent system operator (ISO), there are also rational concerns regarding the resilience, frequency stability, and security of such systems.These concerns need to be discussed and solved from a frequency stability viewpoint, so that a stable operation mode can be guaranteed in case of an unexpected event related to the integration of emerging techniques and renewablebased DGs and demand or generation intermittency or uncertainties.
Many countries, including Australia, have started modernizing and digitising their energy system infrastructures, which is an essential step towards the smart grid concept to enable a 100% renewable power share.This includes upgrading the traditional supervisory control and data acquisition (SCADA) system with a new wide-area monitoring system (WAMS).Current frequency monitoring, control, and protection technologies are being upgraded to be based on the new WAMS infrastructure.Additionally, most conventional measurement systems are also being replaced by phasor measurement units (PMUs) to fit the new WAMS requirement for operating power systems with high penetration of RESs.These technological improvements bring new opportunities, such as the online monitoring of the frequency response, and real challenges, such as how to use these technologies to provide ancillary and control services from different sources effectively, economically, and reliably to maintain the frequency stability and the active power balance between generation and demand [10], [11].
These facts impose on the operators and service providers, e.g., transmission network service providers (TNSPs) in Australia, to procure and deploy new primary and secondary reserve amounts for providing services to support the frequency and keep the active power balance in the permitted range.Moreover, these challenges put pressure on system operators to seek new ways of compensating for the inertia shortfall and supporting the frequency, for example, by deploying synchronous condensers fitted with flywheels, adopting virtual inertia, deploying ESSs through the grid, and involving DSM in providing services like these to the system in the future.For instance, TNSPs have started the process of procurement of fast frequency response (FFR) services from different sources, e.g., BESSs and the providers of fast reserve such as hydro and gas turbines connected to the grid, for compensating the inertia shortfall in a number of power regions in NEM [12], [13].It has been verified that Australian NEM would encounter increased inertia and frequency response shortfalls in the future, which need to be compensated to avoid frequency instability issues.The challenges and their possible solutions are: i) the high RoCoF caused by low inertia requires unique services, i.e., virtual inertia, and fast frequency response (FFR) services; ii) the too-low frequency nadir caused by insufficient reserves and inertia due to the displacement and withdrawal of synchronous generation by cheaper fluctuating RESs requires FFR services which needs to have sufficient capacity as well; and iii) the slow recovery process of the frequency requires advanced frequency restoration services [14].
Power cuts and grid separation events between 2014 and 2022 in Australia confirmed the vulnerability and sensitivity of the existing Australian power system from the perspective of frequency stability due to the high penetration of IBRs, PE-based technologies, the volatility and unpredictable nature of RESs, and insufficient reserve and coordination between different power regions [15], [16], [17].For instance, the separation of the Queensland and South Australia power regions from NEM on 25 August 2018 has raised concerns about coordination and frequency control ancillary services (FCAS) in the NEM, which led to efforts and effective actions by AEMO and other energy industry players to improve stability and security [17].Despite the successful actions by AEMO and other related energy industry partners, like the Australian Electricity Market Commission (AEMC), for improving the frequency operation, especially after 2021, the Australian NEM system, especially due to its current transformation towards almost 100% RESs, needs continuous upgrades and actions to ensure stability and security of its frequency in light of different challenges related to the uniqueness of its design, operation policies, and adopted control approaches compared to international practices.
The aforementioned facts and discussions outline the need to continuously review and improve the frequency support services in Australia's power system.This article, therefore, aims to provide an understanding of frequency control services in Australia's power system and clarify the industry-oriented research gaps and the technical challenges within the system.This article also recognises the efforts of AEMC and AEMO in recent years, as they have, despite the power system's challenges, high uncertainty, and unique design, proposed and adopted improvements to maintain the system's frequency stability and security.This article discusses possible solutions to the system's challenges and technical issues and clarifies that in order to keep the frequency of the Australian grid within permissible limits, a high hosting capacity of IBR and RESs, and emerging technologies for providing services are required to help facilitate power system transformation.Moreover, industry-oriented research directions are clearly presented to motivate research on these real-world challenges and to help build a bridge between industry and academia so that these important challenges that delay the global energy transition around the world can be solved.The article's main contributions, advantages, and benefits are summarised as follows, r Providing a complete understanding of frequency control ancillary services in the Australian power system.
Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.r Building a bridge between industry and academia focuses on solving frequency control issues delaying the power system transformation to reach almost 100% RES-rich power systems.This article is organized as follows.Section II provides an overview of the NEM system, where Subsection I introduces NEM system and its unique topology, and Subsection I presents the different bodies and players responsible for frequency control in NEM.The frequency operating standard in NEM is briefly discussed in Section III, followed by an analysis of frequency support in NEM in Section IV.An analysis of the frequency stability situation in NEM, and AEMO's efforts to maintain it within an acceptable range are highlighted in Section and the technical aspects of frequency support in NEM are discussed in Section V. Industry-oriented research gaps and research directions are discussed in Section VI, and the conclusions are provided in Section VII.

A. Introduction to Australia's NEM System
There are two main power systems in Australia, NEM in the eastern and south-eastern states and the Wholesale Electricity Market (WEM) in Western Australia.Additionally, there are several isolated systems in various territories.In terms of demand, installed generation capacity, and number of consumers, NEM could be considered to constitute a large power system.Both NEM and WEM are operated by AEMO, which acts as an independent system operator.NEM is an interconnected grid comprising several interconnected state-wise networks and approximately 45,900 MW of installed generation.The NEM spot pool market is operated by AEMO across the eastern states of the mainland and Tasmania, including the power regions of Queensland (QLD), New South Wales (NSW), Victoria (VIC), South Australia (SA), and Tasmania (TAS).
Fig. 1 depicts the different power regions in NEM and the topology of the connections.The QLD power region is connected to NSW through DirectLink and QNI interconnectors, while the NSW power region is connected to VIC through SNO.Currently, the SA power system is connected to the VIC transmission network using high-voltage direct current (HVDC) MurrayLink and AC VIC-SA, so-called the Heywood Interconnector, transmission links.TAS is connected to the power regions of the mainland through BassLink, which is an underwater HVDC link.Generally, these interconnectors (tie-lines) between different power regions in NEM forms a unique challenging system from frequency control, stability and security viewpoints.The power regions are connected radially, forming a stringy network topology [18], instead of forming a strong mesh connection due to the location and distribution of human and industrial activities in Australia.The Australian power system encounters technical challenges related to its stability and security due to its unique structure, and the Australian leadership of the world's deployment of RESs and rooftop photovoltaic systems, resulted in frequency and active power challenges witnessed in NEM before these types of issues had (will) been seen in other international systems.
Although this work focuses on frequency control in the NEM, it is worth briefing the generation mix in the NEM to have a full picture of the current situation and the latest progress in the system.Fig. 2 depicts the annual generation by the fuel type in 2021 based on AEMO [19].Readers, who are interested in the generation mix and its change over the years in the NEM, are referred to the interactive maps in [20].Currently, Australia is at the forefront of the world in the installed renewable solar PV per person.The metric, known as solar PV per capita Watt/Capita, put Australia in position 1 globally ahead of the Netherlands and Germany.It is 1166 Watt/Capita for Australia, while it is 1040 Watt/Capita for Netherlands and 807 Watt/Capita in 2022 [21].This, in fact, highlights opportunities and challenges introduced to the NEM grid by installed rooftop PV, especially in the SA power region discussed further from the frequency control view of point in the next sections.It is worth mentioning that AEMO's Projections in the 2022 ISP Step Change scenario indicate that by 2025 there will start to be sufficient renewable resource potential in the NEM to at times meet 100% of demand [2].This might result in periods of operation in which the system will have an instantaneous renewable penetration of 100%.This highlights Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
challenges, especially from the frequency control perspective, where the G-PST project in Australia and this work are dedicated to investigating such challenges and providing suitable research directions.

B. Australia's National Energy Governance and Market Bodies
Australian National Electricity Market activities were initiated in all states in 1998 by the Energy National Cabinet Reform Committee (ENCRC).To enhance the management and operations of the energy market in Australia, ENCRC introduced three main bodies, namely the Australia Electricity Market Commission (AEMC) in 2005, the Australian Energy Regulator (AER) in 2005, and the aforementioned AEMO in 2009, to replace NEMMCO, which was responsible for operating the grid between 1998 and 2009.These active bodies are independent decision-makers with specified powers, boundaries, and liabilities that are designed to achieve the efficient operation of the electricity market in Australia.
AEMC makes the rules for electricity and gas services.It makes and amends the National Electricity Rules (NER), the National Gas Rules, and the National Energy Retail Rules and also provides market development advice to governments.AER is the regulator of the wholesale electricity and gas markets in Australia.AER can set prices and revenues for networks and is responsible for enforcing national regulations.AEMO is responsible for the real-time and day-to-day management of wholesale and retail energy market operations.AEMO serves as the ISO for the Australian energy systems and has statutory functions under the National Energy Laws.It is worth mentioning that AEMO is also the Victorian jurisdictional planning authority.In addition to the above technical and legislative bodies, the Reliability Panel and Energy Security Board (ESB) was created in 2017 to implement the recommendations of the Independent Review on the Future Security of the National Electricity Market, that is, the Finkel Review [22], [23].The ESB is made up of senior AEMC, AER, and AEMO leaders and plays a vital role in reviewing and updating grid operation requirements from reliability and security perspectives.
The aforementioned market bodies, panel, and board create integrity in the legislation, implementation, and enforcement of frequency control and active-power balance services in Australia's power system in terms of keeping it operating safely, stably, and securely.AEMC regularly reviews frequency stability and security in NEM based on technical reports it receives from AEMO, service providers, and stakeholders, especially TNSPs.It also receives rule change requests and reviews them based on technical advice from AEMO and experts.Its evaluation results are reported to ESB and higher legislative and governance bodies so that they can put forward and implement new rules or amendments to existing rules to keep the system operating in an best acceptable situation.On the other hand, AEMO, which is the technical operator of the grid, enforces the secure operation of the grid and takes all steps necessary to correct any frequency deviation and time error in both normal operations and during both credible and non-credible contingency events.To this end, AEMO utilizes several control schemes and management approaches to manage frequency fluctuations and deviations, which will be discussed in detail in the following sections.Furthermore, AEMO reviews frequency stability and security and publishes an informative periodic report known as the Power System Frequency Risk Review (PSFRR), which will be replaced by the general Power System Risk Review Report in 2023 [24].The continuous assessment of frequency stability and security by AEMO of Australia's power system along with the evaluation of the ancillary services contribute to the AEMC and ESB assessments that update the services and frequency control rules needed to keep the power system frequency operating securely in light of the current uncertainty, variability, and volatility in NEM.

III. FREQUENCY OPERATING STANDARDS IN NEM
An understanding of the frequency operating standard (FOS) that is currently in place in the Australian power system is crucial for appreciating the significant efforts that are being undertaken by different market bodies to maintain the stability and security of the NEM's grid.An evaluation of the frequency control of the power system and its effectiveness and challenges is also essential.In Australia, there are two different frequency bands in the FOS, one for the mainland power system and the other for Tasmania's grid.In fact, HVDC Basslink is the main connection between Tasmania and the mainland system.It is, therefore, partially acceptable as well as technically feasible to have two different FOSs for the time being, although the inconsistency in the FOSs can limit effective collaboration in terms of frequency support in the interconnected power system.
The FOSs in the mainland and TAS power grids contain two essential parts, i) frequency bands and ii) required frequency outcomes as defined by the required time scales for each response [25].The frequency bands adopted in both the mainland and the TAS systems are given in Table I.The frequency response time scales required based on FOS are depicted in Fig. 3.It is worth mentioning that in 2021/2022, AEMC and AEMO introduced a mandatory primary frequency response to enhance the frequency stability and power quality by keeping the frequency operating within 49.95-50.05Hz whenever possible, which will be discussed in the next sections.However, such changes have not yet been reflected in the FOS of NEM.
Since it is impossible to avoid the occurrence of disturbances in power system operations, credible or non-credible events, cascading events, and system separations due to natural events or other reasons and due to the unique design of the system in Australia, FOS introduces a large flexibility, known as required frequency outcomes, to the operator to have the required time for correcting the frequency in the system.In fact, the required frequency outcome is a compromise between the system operator capability and the dynamic constraints to achieve best frequency operation in order to maintain overall stability and security with minimum operation costs.The required frequency outcomes defined for Australia's power system are classified based on the event type and are summarised as follows: i) accumulated time error limit (ATEL): less than 15 seconds ATEL for both the Mainland and TAS except for island situations, during supply scarcity, or following a multiple contingency event; ii) generation event/load event: the frequency shall not be outside of the applicable normal operating frequency band (NOFB) for more than 5 minutes in the Mainland and 10 minutes in Tasmania, respectively; iii) network event: the frequency shall not be outside of the applicable operational frequency tolerance band (OFTB) and NOFB for more than 1 minute and 5 minutes in Mainland, while is allowable to be 10 minutes outside NOFB in Tasmania; iv) separation event: the frequency should be operated within the island band and should not be outside the NOFB for more than 10 minutes for both systems; v) protected event: the frequency shall be within extreme frequency excursion tolerance limit (EFETL) and should be returned to the NOFB within 10 minutes in both systems while there is no contingency event, however it is worth to mention that AEMO defines a protected event as an event with a low likelihood, high consequence non-credible contingency event, e.g. the loss of multiple transmission elements causing generation disconnection in a power region during periods where destructive wind conditions are estimated, for which AEMO must maintain the frequency operating standards following the occurrence of the event; vi) non-credible contingency event/multiple contingency event: the frequency shall be within EFETL and should be returned to the NOFB within 10 minutes in both systems while there is no contingency event.Readers interested in FOS in Australia are referred to [25].

IV. FREQUENCY SUPPORT SERVICES IN NEM
The services that are provided to Australia's power system that are aimed at keeping the active power balance and frequency within permissible limits include an inertia responsiveness service, primary frequency control, secondary frequency control known as automatic generation control (AGC), and tertiary frequency control.In addition to these main regulatory and ancillary services, emergency services are also being adopted to correct the frequency decline during larger disturbances, including during credible and non-credible events.
The Frequency Control Ancillary Services (FCAS) that are defined for NEM include regulation FCAS and contingency FCAS.The FCAS market is divided into eight different markets, which are summarised in Table IV.AEMO controls regulation FCAS, and the generators that provide regulation FCAS should maintain the frequency of between 49.85 Hz and 50.15Hz [25], while contingency FCASs are locally controlled and their availability and activation are monitored by AEMO.Measures to correct the frequency in contingency events include generator governor response, load shedding, rapid generation, and rapid unit unloading.Since the battery storage system can take charge and discharge, it can rapidly increase and decrease its power output/input.When there is a contingency event, e.g., a generation deficit due to a generator set accident or a sudden reduction in large-scale loads such as in factories, the electricity system's frequency will suddenly change.The grid can purchase electric energy in the contingency FCAS market to stabilise the frequency.In Australian NEM, different markets can act according to the scale of the crisis of the contingency event.The Contingency FCAS services are provided within the time required by their markets, respectively, such as the immediate raise service will be delivered within 6 seconds of a contingency event.When there is a small deviation in frequency, regulation FCAS can provide frequency regulation services by increasing or decreasing the active power injected into the system.The time response frames of different regulation FCAS and contingency FCAS markets are depicted in Fig. 3.It is worth mentioning that battery energy storage systems (BESSs), such as Hornsdale Power Reserve BESS, are currently being registered and are contributing to frequency control by participating in the eight FCAS markets defined in NEM.The over 100MW BESSs that are currently operated and integrated with the NEM grid include Victorian Big Battery (300MW ), Hazelwood BESS (150MW ), Hornsdale Power Reserve (150MW ), and Wandoan BESS (100MW ).The full list of BESS and VPP projects in the NEM system and their details including their to-date market total revenue are available in [26].In mid-2019, AEMO has initiated virtual power plant (VPP) trials, the AEMO's VPP Demonstrations, that are currently being implemented/tested in Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

TABLE III CURRENT INERTIA REQUIREMENTS IN MWS IN THE NEM
NEM in terms of their opportunities and challenges in order to officially approve the concept in NEM, which will be important for providing additional support to FCAS during the system transformation.The AEMO VPP Demonstrations are a collaboration between AEMO and the Australian Renewable Energy Agency (ARENA), AEMC, AER, and members of the Distributed Energy Integration Program.However, the main focuses of AEMO's VPP trial are on the contingency FCAS market, and on confirming the concept's capability to support the contingency FCAS market.It should be mentioned that in addition to the eight FCAS markets divided into either contingency or regulation services, there is an expectation that new markets will be introduced, such as inertia and very fast frequency response services.In fact, AEMC has already published National Electricity Amendment, i.e., Fast Frequency Response Market Ancillary Service, on Rule 2021 No. 8, which requires AEMO to set up two markets for very fast frequency responses.The uniqueness of the NEM system, the high levels of RESs and IBRs in the system, and system uncertainties lead to regular reviews of and updates to the FCAS markets.

A. Inertia Responsiveness Service in NEM
Inertia response service is essential for Australia's power system to sustain its operation in a secure and stable mode due to the fact that NEM is enriched with RES and is at the forefront of the world in regard renewable and power electronic-dominated power systems [27].Consequently, AEMO is encountering serious frequency stability and security issues before other international system operators.Generally, the aim of the inertia response in NEM is to enhance the frequency response during different types of events by limiting the RoCoF and frequency nadir, and slowing frequency decline, thus providing enough time for FCAS to act and arrest frequency excursions before reaching an unstable point [18].The other role of inertia in NEM is to provide an additional level of stability and reliability to the grid from the frequency control perspective.Due to its vital role in stabilizing and securing the operation of the grid, and due to the current unfavorable frequency situation in NEM, AEMO assesses the available rotating inertia in the system regularly and determines the minimum and secure levels of the required inertia in different power regions in Australia [14].Table III presents the currently required inertia in different power regions to securely operate the system, where the values are obtained based on a specific inertia requirement methodology adopted by AEMC and ESB in Australia [14], [28].Currently, the SA power region encounters severe inertia shortfall, and this will last until 2023, while the TAS region faces a potential inertia shortfall.Based on AEMO's estimation, the QLD power regions could face a potential inertia shortfall in 2025, while the TAS power region would encounter a severe inertia shortfall after 2024 [14], [29].It is worth mentioning that a number of power regions in Australia meet current inertia requirements by compensating for physical rotating inertia by fast frequency reserve due to the shortage in physical inertia in the renewable-dominated power system in NEM [14].However, this is a temporary solution in case there is an inertia shortfall in a specific power region.AEMC defines the (very) fast frequency response as the service of providing, in accordance with the requirements of the market ancillary service specification, the capability of very rapidly controlling the level of generation or load associated with a particular facility in response to the locally sensed frequency of the power system in order to arrest a rise or fall in that frequency [30].Detailed AEMO's report on the implementation options of FFR in the NEM is available in [31].
It is worth mentioning that the numbers given in Table III are calculated based on a specific methodology designed for determining the minimum and secure inertia levels at the NEM.Based on the inertia requirement methodology [14], AEMO calculates the minimum threshold level of inertia and secure operating level of inertia for each region when islanded.In determining the required level of inertia, AEMO has considered the following factors to reduce the level of inertia otherwise needed: r The largest credible contingency event when a region is operating as an island.
r The level of Contingency FCAS available in each region.
A complete description of the inertia requirements methodology adopted in NEM and approved by AEMC and ESB is available in [14], [28].The main theory basics behind the calculation of the minimum required inertia for each operational time interval (OTI) relies on the swing equation that relates the RoCoF to the power deficit.Fig. 4 shows the OTI for the NEM which is 5 minutes based on the current dispatching engine based on energy and FCAS markets.For instance, the required minimum synchronous inertia can be determined as follows, where H S,req t is the required minimum inertia (in MWs); f n is the nominal frequency (50 Hz for the NEM system); RoCoF allowed Region is the allowed maximum RoCoF after a contingency in a specific power region, where Region ∈ {V IC; NSW ; SA; QLD; T AS}; and |max{P G,i,max t }| is the largest possible contingency within a power region (in MW).
Based on (1), the required minimum level of inertia can be determined for each power region after obtaining the required information for the power region of interest.The main information includes the maximum possible contingency in the power Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
region and the allowed maximum RoCoF to ensure the secure operation of the power system.It is clear that these numbers would change in the future due to changes in the generation mix and due to the current energy transition in the NEM.With the increase of the penetration level of energy and inverter-based resources in the grid and the retirement of synchronous-based power plants, it will be an issue to satisfy the minimum requirements by synchronous inertia only.Thus, the future system would need a review and revision of the minimum and secure inertia requirements to consider other sources of providing inertia services such as virtual inertia from different emerging technologies and the consideration of possible future enhanced RoCoF ride-through capabilities that might partially relieve the high minimum inertia requirements.Readers who are interested in this topic, are referred to the recent research report by the authors as part of the Gloabl Power System Transformation (G-PST)-stage 2 project and the AEMO's reports [14], [28], [32]

B. Primary Frequency Control
Primary frequency control (PFC) is considered essential in NEM, as it supports frequency stability by keeping the frequency within NOFB in normal operational periods.Primary frequency control deals with both small and large disturbances.However, a frequency deadband is introduced to governors of conventional power plants and other primary frequency response (PFR) providers to avoid excessive or unwanted responsiveness to very small variations, which results in wear and tear in governor and turbine systems.In NEM, primary frequency control is implemented based on primary reserves that come from both fast and slow frequency response markets.All registered NEM generators must provide PFR, subject to energy availability.The primary frequency controller is being implemented using droop techniques in both conventional and emerging energy sources, e.g., BESSs, using locally measured frequency.The droop characteristic for conventional power plants is specified for NEM based on (2).
where ΔF is the frequency deviation beyond the upper or the lower limit of the generator deadbandin (in Hz); F N is the nominal frequency of the system; ΔP is the active power change (in MW); and P Max is the maximum operating level (in MW).The AEMO's current Interim PFR Requirements (IPFRR) enforce the droop coefficient to be less than or equal to 5%.It is worth mentioning that NEM permits flexible adjustments of the droop so that it may be asymmetrical for over-and under-frequency responses, and it may also differ for different levels of frequency change.Furthermore, additional flexibility is provided to adjust the droop based on the requirements of the power plant if the minimum PFR is achieved.The droop coefficient can be adjusted in the range of 1.7% − 5% [33].Currently in the NEM, primary control comprises the current mandatory PFR (MPFR) proportional droop response when frequency leaves a ±0.015Hz deadband around 50 Hz and contingency FCAS via proportional controls, or increasingly via switched controllers, when the frequency deviation is beyond the NOFB.Contingency FCAS is not designed to control, or capable of controlling, frequency to a 50 Hz setpoint.
In addition to the governor deadband and the droop coefficient, AEMO considers another important parameter, i.e. the time response of PFR, to ensure sufficient and correct PFR activation in NEM.Time response refers to how quickly the generator changes its active power in response to a frequency deviation outside its deadband.The IPFRR requires that generators be able to achieve a 5% change in active power output within no more than 10 seconds, in response to a positive or negative step change in frequency of up to 0.5 Hz.The speed at which various generation technologies can alter the MW output varies, with inverter-based resources (IBR) capable of a much faster response than synchronous generation technologies [33].
Due to its importance in frequency stability, PFR is one of the main issues for AEMC, and several policy pathways designed to lead to lasting PFR arrangements are currently either under discussion or being implemented.The options considered differ significantly in their effectiveness.The main criterion for AEMC adopting a PFR policy pathway is robust and effective aggregate frequency responsiveness in the long term.The constraints for provision of PFR to NEM as defined by AEMC include i) Decentralised-based on local detection and response, not impacted by communications unavailability, providing a dependable, robust and proportionate response; ii) Distributed-based on a large number of contributors over a geographically dispersed area, enabling responsiveness physically close to the disturbance, and reducing dependence on individual providers and prevailing network conditions, and reducing duty on individual plant; iii) Simple-based on a sequence of actions that can be handled within the control hierarchy of plant and, at the system level, can provide a stable base level of narrow band frequency responsiveness for other frequency control reforms being progressively overlaid; iv) Predictable-based on a level of consistent responsiveness to frequency deviations, reducing uncertainty in power system behaviour, system adequacy and frequency control need assessment; v) Flexible-can scale over time as the technology mix changes, and can be potentially extended to include new PFR sources and be overlaid with a headroom management mechanism in the future if needed [34].

1) Provision of PFR From BESSs in NEM:
A battery energy storage system provides the FCAS response by varying its active power when the local frequency exceeds the narrower of the lower or upper limit of the NOFB bands and its frequency control deadband.AEMO publishes clear guidelines to assist market participants in NEM to register a BESS for providing FCAS services.Information on the type of frequency controller to be used and the allowable droop settings when delivering FCAS is also provided to help participants determine the maximum ancillary service capacity that can be registered, subject to a successful FCAS assessment factors by AEMO, including metering, SCADA, negotiation of generator performance standards (GPS), and engagement with network service providers (NSPs).More information on BESSs and their role in the provision of PFR in NEM is available in [35].In NEM, the droop of the Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

BESSs used to adjust their contribution to PFR is determined as
where ΔF is the frequency deviation (in Hz); DB is the deadband of the frequency controller(in Hz); C is the registered capacity of the BESS (in MW); and SP is the capacity of the BESS used to provide an FCAS response (in MW).
The maximum ancillary service capacity provided by BESSs is rounded down to the closest integer based on (4).
) where F R is the absolute value of the difference between 50 Hz and the raise or lower reference frequency; and F B is the frequency deviation at which the maximum charge or discharge of the BESS is provided.
It is worth mentioning that the minimum allowable droop coefficient for BESSs in NEM is 1.7%.The deployment of BESSs in NEM and their use in providing FCAS, and PFR in particular, has helped stop frequency deterioration and improve frequency stability in the grid.The recent analysis of the impact of BESSs on the grid scale on frequency stability shows that energy storage systems are mandatory for the present and future power grid in Australia due to their high flexibility and quick responsiveness compared with conventional energy sources.For instance, the Hornsdale Power Reserve Energy Storage System located in the SA power region even demonstrates better performance following AEMO's PFR and central AGC set point and responds more quickly to contingency events than the traditional thermal generator in NEM [36].
2) Latest Updates in PFR for NEM: The Australian operator of NEM, AEMO, has started activities in adopting new providers of PFR for NEM.This step is based on the mandatory frequency response rule that was introduced to enhance the frequency stability and security of NEM [34].In Australia, PFR has historically been provided by synchronous generation.Recognizing the transition to high variable renewable energy (VRE), the mandatory PFR (MPFR) rule applies to all scheduled and semi-scheduled generation.Modern VRE systems can provide PFR through the implementation of a frequency droop response to an active power output within the control hierarchy of plant control systems.Frequency response capabilities are standard in all new VRE and battery energy storage systems, and are required for connection under the new considerations of NER.In fact, the inherent controllability of the active power of both VRE and BESS is typically significantly higher than conventional synchronous generation.Frequency response from VRE and BESS can typically be provided faster, and over a larger part of the MW operating range than from synchronous generation.Uncurtailed VRE is only able to provide an active power response in one direction, downwards from whatever its MW output is based on weather at the time.However, curtailed VRE and BESS are able to provide a response in both directions.Currently, AEMO is engaging with original equipment manufacturers as part of the MPFR rollout to ensure that the PFR is provided appropriately from the NEM VRE fleet.While inherent control capabilities exist at almost all sites, work to date indicates that a number of VRE generators, particularly at older sites, will require various updates to control software, particularly Power Plant Controllers or similar, to meet all NEM active power control requirements [34].This is materially different from the implementation of the MPFR rule for synchronous generators, almost all of which could meet the MPFR requirements by implementing setting changes with existing control systems.
Changes to VRE control system software are currently being trialed and validated at a small number of generating systems using equipment from each original equipment manufacturer (OEM), before moving to wider-scale implementation.PFR has now been fully implemented at a small number of wind sites, and testing is ongoing at a number of solar sites.The implementation of PFR from grid-scale BESS has generally proven to be more straightforward [34].

C. Automatic Generation Control in Australia's NEM
In NEM, the automatic generation control-AGC scheme plays a vital role in removing the frequency errors that resulted from the implementation of primary frequency control.AEMO considers its own approach for implementing AGC, which is different from general practices in international power systems.In most power systems around the world, AGC is implemented by the area control error (ACE), which is calculated based on the deviation of the frequency and tie-line power flow in each area, as mathematically presented in (5), while the deviation of the power flow of the tie-line is not considered in the current AGC of the NEM and is replaced by the time error as modeled in (6).AEMO calculates the ACE representing the MW-equivalent size of the current frequency deviation and the accumulated frequency deviation (time error) of the NEM system.Generally, ACE may be considered to represent a rough proxy for the required regulation FCAS volume.
where F is the measured frequency; F O is the time error, which is also known as the frequency offset that represents the accumulated frequency deviation; and β is the frequency bias, a tuned value that represents the conversion ratio between MW and 0.1 Hz of frequency deviation.
Fig. 5 depicts the Australian AGC scheme that is currently being adopted and implemented in NEM and operated by AEMO.It is clear from Fig. 5 that the main inputs for the AGC system are the measurements of both frequency and time errors to perform the ACE.In fact, AEMO considers only one ACE for the whole mainland grid within NEM, which means that there are no different ACEs for the different power regions in Australia's NEM system.This is a strictly centralized approach that might affect the reliability and security of the system and is less resilient if compared with the quasi-decentralized approaches that are being implemented internationally.Another observation is that the tie-line power flow deviation is not included in the Australian AGC system, which could negatively impact stability during both normal and event operations.However, the AGC system requires serious review to quantify the impact of the current AGC system on power flow fluctuations during normal and emergency periods.The centralised control approach adopted in NEM could pose several challenges, especially during events, e.g. during 2018 QLD and SA separation events in Australia.

D. Delayed/Tertiary Reserve in NEM
In addition to secondary frequency control, tertiary frequency control (TFC) in NEM is managed based on the FCAS market to ensure there is sufficient reserve for credible events and for operating AGC in the next horizontal time of system operation.This type of control/management is responsible for re-adjusting the operating point after an event and disturbance and is being checked based on the co-optimization of energy and reserve in energy and FCAS market every five minutes.In fact, TFC is the reserve generation capacity that can be used to reset primary and secondary frequency control services.This capacity does not automatically respond to frequency, rather it is available reserve that can be called on to restore the system to a secure operating state following contingency events.In NEM, the tertiary reserve is managed through energy market dispatch, which matches generation supply with forecast demand every five minutes [37].

E. FCAS Market in NEM
The design of the FCAS market is unique, as it is divided into regulation and contingency services.Regulation FCAS is responsible for operating the frequency inside NOFB, while contingency FCAS acts to quickly bring frequency to NOFB after a large disturbance or event either credible or non-credible one.The point is that the mainland is considered in the market as one region for both energy and FCAS, meaning that the tie-lines and major boundaries between the power regions are neglected in the market activities except their capability for transferring specific amount of power between different regions.It should be emphasised that both regulation FCAS and contingency FCAS are being procured for all regions in NEM without considering FCAS regionalisation and the binding network constraints.The exception is for the TAS power region, which can procure its needed FCAS without considering FCAS procurement on the mainland.This highlights that NEM has two separate or weakly connected systems, i.e. the mainland and TAS, from a frequency operations point of view, despite the fact that they are connected  together through the Basslink.The other exception is for the inertia requirements, which can be obtained in each power region separately based on the analysis of possible credible events and separation from other regions.In NEM, FCAS service providers are paid for making the capacity able to respond to frequency deviations regardless of whether FCAS is required and delivered.
The activated secondary ancillary reserve in MW from the FCAS market for the mainland and Tasmania systems are depicted in Figs. 6 and 7, respectively.The issued ACE signal of AGC in both systems for the same time period, Q1 in 2022, presents challenges especially in Tasmania's system from an AGC viewpoint and verifies that the systems are separated from each other from a frequency perspective [38].This indicates potential benefits of, and highlights the need for, involving tielines including HVDC links in making the connection between different regions powerful for supporting the frequency and smoothing the power flow fluctuations.In particular, HVDC links can greatly improve frequency stability and security by involving them in frequency response programs.However, only Basslink is considered to partially contribute to PFR while other HVDC links in NEM are not actively involved in providing such services, which might require more review and investigation of their capabilities to provide some frequency services during the future system transformation.

F. Emergency Frequency Control
The article focuses on frequency control ancillary services based on regulation and contingency FCAS market.The contingency FCAS market includes 6 different markets as discussed Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
in previous sections in detail, where all these markets are designed and dedicated to controlling the frequency in case of a contingency.It is worth mentioning that the contingency FCAS markets are detected to frequency deviation due to credible events, while the frequency of large deviations due to rare events and/or non-credible contingency events are managed through emergency mechanisms for controlling the frequency.It is to distinguish the frequency control issue between the Australian concept and international power systems due to the fact that the frequency control is treated as a market issue in Australia while it is a technical issue in international power systems like entso-e power grid in Europe and the interconnected power systems in USA.Therefore, the under-frequency load shedding (UFLS) scheme, in NEM and WEM systems, only operates during rare events where a drop in frequency has not been arrested sufficiently by FCAS control.Also, the load shedding is usually done over pre-contracted loads under emergency demand response schemes to adapt to the market rules.In the Australian NEM system, the yearly Power System Frequency Risk Review (PSFRR) provides an update on non-credible contingency events with a suitable analysis of the events and the actions taken to secure the system including UFLS schemes [24].For instance, a useful discussion about the impact of DER on UFLS, the UFLS remediation work in the SA power region, and the UFLS remediation work in other NEM regions can be found in the 2022 PSFRR report [24].Likewise, the different versions of the PSFRR periodically review emergency events in the NEM, provide AEMO's analysis around a protected event, and present suitable lessons based on an in-depth engineering analysis of non-credible events.It is worth mentioning that AEMO has recently replaced the PSFRR with the general power system risk review (GPSRR) starting from august 2023 which provides insights and analysis about emergency schemes and non-credible emergency events in Australia [39].Although this important topic is out of the scope of this work, it is to emphasis the importance of emergency schemes for frequency control especially for future renewable-rich power systems, like the NEM during energy transition and beyond.With the retirement of synchronous machines and moving power systems towards future deregulated smart power systems based on high deployment of distributed generating units like distributed PV in South Australia, there is a need to investigate the requirements for involving advanced over/under frequency relays (like 81 O/U) in the distribution level as-well-as in the transmission level to reach well-coordinated emergency control/protection schemes for the frequency deviation, especially in case of non-credible emergency events.

V. TECHNICAL ASPECTS, ANALYSIS OF FCAS IN NEM AND AEMO'S EFFORTS
It is a fact that Australia continues to lead the world in the per capita roll out of energy generated by renewable energy sources, e.g., solar and wind.While it offers opportunities, such as lowering greenhouse gas emissions, it presents serious challenges and technical operational issues related to FCAS and frequency stability and security in NEM.The deep analysis of the frequency situation during the last decade in NEM clarifies the technical challenges, which can be summarized into two aspects: i) the technical aspect, and 2) the policy aspect.
Technically, the adoption of Australian incentives approaches to increasing renewable energy integration into the grid on both grid and consumer scales led to several challenges, including high volatility and variations in both demand and generation sides.These variations in addition to increases in the weather and demand variations in Australia have resulted in high uncertainty in both demand and generation sides leading to higher active power imbalances in different time scales, which affect the frequency in NEM.From a policy prospective, NEM introduced a new FCAS approach in 2001, in which the mandatory ±50 mHz governor deadband was relaxed and market-based FCAS was adopted.The relaxation has incentivised power plants to reduce their obligations for a contribution to frequency response support.The aforementioned policies have not been thoroughly updated since 2001.This has led to a reduction in the adoption of the unique techniques needed to control the frequency of the modern power system in Australia and has also led to frequency events.Most of the frequency metrics adopted in NEM have been reporting worse frequency operations within the grid, especially before introducing the current MPFR arrangements.The frequency encountered severe deterioration after 2014 due to several reasons, as mentioned above.Fig. 8 shows that after 2014, the frequency distribution starts to flatten, and is not concentrated around 50 Hz, where this is clear from the reduction of the number of crossing the 50 Hz line, from 80 k to less than 40 k per month [38].In Fig. 8, three metrics, namely i) the number of 50 Hz line crossing; ii) the number of departures below 49.85 Hz; and iii) the number of departures above 50.15Hz, are used to evaluate the frequency performance in the NEM system by the AEMO, taking into account the frequency deadband is ±0.15 Hz.This means that if the 50 Hz line crossing number is high and the departure outside the frequency deadband is minimal, the frequency performance is on an acceptable level as requested by AEMO, otherwise, there is an issue with the frequency performance that might impact the system stability.As can be seen from Fig. 8, the frequency performance was acceptable due to the above-mentioned metrics before 2014.However, the performance started encountering serious deterioration after 2014 due to the increase in RESs penetration and other technical reasons related to the setting of frequency controllers and frequency deadbands.However, after AEMC and AEMO actions and rule changes discussed in the article introduced in 2020, the frequency performance has been corrected to ensure the power system frequency stability in the NEM.Other illustrative figures of metrics presenting the frequency deterioration 2013 and its performance correction after 2020 have been omitted due to the article's space limit consideration.The Australian operator of the system, AEMO, has taken serious and series of actions to correct the frequency performance in order to ensure the stability and security of NEM, including procuring more fast frequency reserve and enforcing narrow deadband based on incentive-based schemes.Similarly, the commencement of power plants and battery energy storage systems has forced technical aspects to evolve in order to improve frequency in the grid.The efforts by AEMC and AEMO improved grid performance and security, especially when taking into account that the FCAS is defined as a market issue in NEM.This puts restrictions on immediate remedial action by the operator of NEM, especially when compared with the international practice of seeing frequency control as a technical issue, which helps increase the flexibility for international ISOs to act when necessary.The efforts by the operator and energy bodies have corrected the frequency in NEM, where the aforementioned figures confirm that the frequency situation was relatively corrected after the immediate actions by AEMC and AEMO after 2020.Similarly, Fig. 9 shows the frequency distribution in the period Jan 2011-Jan 2019, which shows the deterioration of frequency after 2013.The comparison between Figs. 9 and 8 confirm that the recent efforts helped to make the frequency more concentrated around the nominal frequency in Australia, that is, 50 Hz.The aforementioned challenges and efforts have been generally highlighted and motivated by the server contingency event that occurred on 25 August 2018, in which two separation events, QLD and SA from NEM, resulted in the interruption of the electricity supply to industrial loads in VIC, NSW, and TAS, and a number of residential and commercial customers in NSW.The lessons of this event and more details are reported in [17].

VI. TECHNICAL-INDUSTRIAL AND MARKET CHALLENGES IN MODERN SYSTEMS AND FUTURE RESEARCH DIRECTIONS
In what follows, we discuss the technical challenges and industry-oriented research gaps of NEM's frequency and generalise the discussion for power systems under transformation in the aim of bringing attention to important issues that need to be adequately addressed.The discussion is followed by highlighting future industry-oriented research directions on each topic.
The discussion on FCAS services in previous sections, that have been adopted in Australia's power systems, highlights an important point that each power system is uniquely designed and operated based on several factors, including the geographical location, community behavior and its response to energy usage, economic growth, availability of energy and reserve sources, and interaction between policy and technical/market bodies.Therefore, a focus on real-case power systems would increase the impact of solved issues, especially in FCAS studies.The industry-oriented research gaps in NEM are summarized in the following points.
Inertia Response Services: Due to their importance, the required inertia and system strength should be determined on the basis of both stability and security analysis.For the NEM system, there is a need to thoroughly review and revise the inertia requirements to consider regional inertia to avoid any cascading events in case of separation of one or more power regions, which is one of the highest risks in NEM based on the events that occurred in the past.It is also mandatory to specify the required inertia for different time-horizons of operation due to the fact that the required inertia in the day-demand differs from those in night-demand, and the same is true for different operational seasons.Similarly, the minimum need for physical inertia should be determined and which inertia mix is required to be provided from different sources such as inertia emulation from wind farms and virtual inertia from ESSs.Likewise, the capability of emerging technologies such as Ultracapacitor (UCAP) energy storage, Hybrid ESS (HESS), and flywheel ESS need to be investigated for future NEM to help power system transformation toward almost 100% renewable and IBR-rich systems.Furthermore, the impact of different inertia levels in different power regions on power flow through tie-lines and stability in other connected regions needs to be investigated.Moreover, the compensation of part of the required inertia by fast frequency response should be studied to quantify its impact on the operation stability and security of the power system.The resilience of virtual inertia, the real-time estimation of available inertia and its type, and the design of the inertia market are future industry-oriented research directions on this important topic.
Primary Frequency Control Services: There are several practices for the implementation of PFC in different power systems around the world.It is suggested to work on one norm for implementing PFC internationally and the most important is to uniform the way of implementing PFC in one system.NEM might need to specify the deadband for governors in conventional power plants and for control actions from other participants, such as ESSs.The industry-research gap here is to optimally specify the needed deadband and droop for a specific system and operation time interval based on the operation social welfare.Furthermore, it is an industry-oriented research issue to evaluate the impact of adopting different deadbands and droops for different PFC providers in the same power region and in different power regions, and these interactions should be thoroughly studied to assess the impact on both frequency stability and security.The industry-oriented research gap is Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
how to allocate the PFC reserve in different power regions to improve operation stability and security and reduce operational and reserve procurement costs.As another research gap, the impact of adopting a central FCAS market on operation security needs to be discussed as well as what arrangements should be adopted to convert the system to be based on a fully decentralized FCAS market and operation.The important research-oriented direction is to investigate how to use different tie-lines to enhance primary frequency response, especially in power systems that have different types of tie-lines, e.g., AC tie-lines and HVDC links, like Australia's power system.
Secondary Frequency Control Services: The implementation of AGC in the Australian NEM system is completely different from international practices.NEM does not consider tie-line power flow deviations in online calculation of the area control error, and consequently, it would not be involved in the AGC system.The main concern in NEM is the high-power flow fluctuations through tie-lines between different power regions during normal operational periods because of the ignoring of tie-line control.In fact, NEM operates the whole power system as one power region even though the system is divided into several power regions due to the geographical locations and the long distances between the different power regions.This industry-oriented challenge affects the security of the power system and could lead to cascading events after one tie-line separation; the QLD power region separation in 2018 is a clear example of such issues.Furthermore, optimal allocation of secondary reserves in different power regions can help involve tie-line power flow deviation in AGC without more procurement of FCAS reserves, which is one of the future industry-oriented research directions.Moreover, the security and cybersecurity of AGC implementation is an industry-oriented research gap and future research direction for Australia's power system due to the fact that NEM still implements a centralized AGC system.The arrangements for converting the current AGC system to be quasior fully decentralized one is another important industry issue and research gap for NEM.Finally, the measurement sampling rate used for both primary and secondary control systems needs to be revised to be more accurately based on higher rates, especially for the AGC system, where the system infrastructure may convert to be based on wide-area monitoring (WAM) and PMU measurements instead of old SCADA systems.
Current FCAS Market Issue: NEM encounters technical and policy issues that affect the allocation of FCAS services in different regions.The main challenge is related to the regulation of FCAS services, where most of the amount offered for the operation of the power system comes from the VIC region.This might affect the overall security of the system because these services, i.e., regulation raise and regulation lower services, are crucial for solving imbalance issues in normal operational periods.The security issues raised from the fact that there are BESSs registered for providing the regulation FCAS, which are highly competitive to other sources, and since this regulation services will be provided by one or limited number of sources, this put the system under security issues.The other issues are the potential power flow fluctuation in interconnectors, and some time they can not be used to deliver the regulation services to other regions due to technical issues, e.g.BassLink, which also challenge the system stability and security.However, HVDC links other than Basslink do not provide frequency response services in NEM which might require review and upgrade.To avoid stability and security issues, AEMO has enforced FCAS allocations, under the Regionalization of Regulation FCAS Concept as an interim solution.The solution places constraints on the spot market that at least 25% of the total required regulation FCAS in NEM, i.e. 220 MW raise-regulation FCAS and 210 MW lower-regulation FCAS, should be delivered from outside of the VIC region, i.e.QLD + NSW + SA > 25% of the regulation FCAS requirement.However, this is an important technical and market issue that needs to be properly solved to maintain the transparency of the market, and operation stability and security of the system.
Current Technical Issues in NEM's FCAS: One of the important technical parameters in determining the required contingency FCAS is the load damping factor, known as the load frequency relief in NEM, because the procured contingency FCAS amount is equal to the size of the largest credible contingency minus the assumed load relief.Since 2001, the load relief has been selected as 1.5% and 1% for the Mainland and TAS, respectively.Analysis of events that occurred in 2020 confirmed the inaccuracy of the aforementioned values; therefore, AEMO reduced the Mainland load relief to 0.5% and the TAS load relief to 0-0.1%.However, this is one of the great challenges in today's power systems, especially for NEM, since the modern rotating loads are being connected to the grid through power electronic devices; therefore, there is a research gap and industry challenges for mid-term and short-term estimation/measurement of load relief, since it highly impacts the amount of the procured contingency FCAS, and thus the operational cost of the system.
Emerging Issues: The time response requirements from different FCAS markets need to be evaluated, and more adaptive time scales are needed to adapt to the new characteristic of the system to enable and smooth the energy transition.Two of the challenges in Australia's power system that would be great challenges in the future operation of the international systems as well are the uncertainties and variability.The industry-oriented research gap is how to quantify power variations and uncertainties in different power regions and how to assess and value their impacts on other power regions from the perspective of frequency stability and security issues, and from other perspectives as well.In Australia, the SA power region reached the high power share from renewable energy sources in October 2021 while other power regions connected to SA still operated with different levels of power generation mix.In such a case, a new market needs to be defined to address such issues, such as the causer of variability and the need for FCAS services, and consequently the reserve should be allocated based on one short-term estimation of uncertainties in each area to avoid excessive tie-line power flow fluctuation and high difference between local frequencies.Furthermore, the local frequency is becoming more obvious which might bring challenges to the operation stability and security in modern power systems such as NEM.Fig. 10 highlights the frequency differences, i.e., local frequencies measured in the third week of January 2022, between two regions connected to each other in NEM.The local frequencies should be qualified/quantified online in monitoring centers and controlled to keep the system in a good situation of connectivity between different power regions.Finally, Australian power systems including NEM and other international systems with high penetrations of RESs and IBRs are in urgent need of new metrics that can be evaluated online and in real-time, including assessing frequency and local frequencies and their stability and security.In fact, most of the existing metrics are designed for systems with low variability and are calculated based on the SCADA rate; therefore, developing advanced metrics for frequency in modern power systems would help enable the global power system transformation.

VII. CONCLUSION
In this article, the current frequency control situation in Australia's national electricity market (NEM) system was reviewed in-depth and the different frequency control ancillary services (FCASs) were evaluated.Moreover, the challenges in keeping the frequency within the permissible range in NEM were identified and discussed.Furthermore, AEMO's efforts and current updates in the FCAS market and the control actions by AEMC were evaluated, and their impact on the frequency operations was assessed.The article presented the challenges and technical issues of frequency control and the FCAS market from industrial perspectives and summarised research gaps and technical issues to motivate further research on real-world technical issues.Moreover, the article presented various future research directions and the actions needed to improve the frequency stability and security of power systems going through the energy transition.

rr
Clarify the industry-oriented research gaps and the techni- cal challenges related to frequency control within the NEM system.Reviewing current situation and efforts to improve fre- quency services in the NEM system.

r
Discussing possible solutions to industry challenges and research gaps.rProviding insightful and industry-oriented research direc- tions on frequency control for future systems.

Fig. 1 .
Fig. 1.Physical interconnection between power regions in the system of the national electricity market (NEM); i.e -HVDC and -AC Interconnector.

Fig. 2 .
Fig. 2. Annual power generation and their energy sources in the NEM in 2021.

Fig. 3 .
Fig. 3. Graphical illustration of FOS in NEM with the activation time of each FCAS market.

Fig. 4 .
Fig. 4. Calculation of the minimum required inertia for each operational time interval during a one day.

TABLE I FREQUENCY
BANDS DEFINED IN AUSTRALIAN FOS

TABLE II FCAS
MARKETS AND THEIR REQUIRED OUTCOMES IN NEM