Professional Documents
Culture Documents
2019 - RoCoF Limits
2019 - RoCoF Limits
DOI: 10.1002/2050-7038.12128
RESEARCH ARTICLE
KEYWORDS
frequency response, power system inertia, synthetic inertia
LIST OF SYMBOLS AND ABBREVIATIONS: ▪▪▪Pgen – active power of a synchronous generator;f – frequency;t – time;f0 – frequency at the start of
the transientevent;Pk – active power of the kth synchronous generator;H – inertia constant;S – apparent power of a synchronous generator;Pmech –
mechanical power of a synchronousgenerator;EK,E – kinetic energy of a synchronous generator;J – moment of inertia;ωgen – angular speed of a
synchronous generator;ωsys – synchronous angular frequency;Ssys – apparent power of the electrical system;ωn – nominal angular frequency;Ta –
acceleration time constant (starting time constant), ▪▪▪Rate of Change of Frequency – RoCoF;European Network of Transmission System
Operators for Electricity – ENTSO‐E;Frequency Containment Reserve – FCR;Transmission System Operator – TSO;Transmission System Operator
of Ireland ‐ EirGrid;System Operator of Northern Ireland – SONI;High Voltage Direct Current HVDC;Australian Energy Market Operator ‐ AEMO;
Electricity Reliability Council of Texas – ERCOT;Automatic Load Shedding ‐ ALS;Fast Frequency Response ‐ FFR;Synchronous Inertial Response –
SIR;Fast Post‐Fault Active Power Recovery (FPFAPR);Ramping Margin ‐ RM;Enhanced Frequency Response, EFR;Pennsylvania – New Jersey –
Maryland Interconnection – PJM
--------------------------------------------------------------------------------------------------------------------------------
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the
original work is properly cited.
© 2019 The Authors International Transactions on Electrical Energy Systems Published by John Wiley & Sons Ltd
1 | INTRODUCTION
The share of distributed generation, especially of intermittent renewables such as wind and solar energy, has increased
significantly in Europe during the last decade. At first, standards and regulation regarding the terms of network connec-
tion were not as elaborated as today, as a result of which most of the distributed generators are operating appropriately
between 49.5 and 50.5 Hz. If the frequency decreases or increases outside this range, these units simply disconnect from
the network. At the same time, conventional generators must remain connected3 between 47.5 and 51.5 Hz.
Another aspect is that these generation units are connected to the grid via power electronic converters; therefore,
they do not have a directly coupled rotating mass as conventional synchronous generators. This fundamental difference
leads to further challenges in system operation. As an example, the increasing penetration of nonsynchronous sources
could bring forth large power swings in the system, with frequency deviations, and the rate of change of frequency
(RoCoF, which is the time derivative of the frequency) has also become larger than before. These tendencies together
have a negative effect on the security of supply, even in large, interconnected power systems like the Continental
Europe area of the European Network of Transmission System Operators for Electricity (ENTSO‐E). The changes pose
a challenge in consideration of a fundamental planning rule of the Continental Europe network: The system has to be
able to withstand frequency deviations as the result of a significant change in load or generation (3000 MW as the ref-
erence event—which is the simultaneous loss of the two largest generation units) by utilising Frequency Containment
Control (FCR) (also known as primary reserve before the introduction of the System Operation Guideline.)1,2
Insufficient FCR does not automatically lead to more severe intervention (eg, under frequency load‐shedding),
but improper settings of distributed generation protection relays may cause further unnecessary loss of generation, lead-
ing to increased frequency deviations, which may result in large power swings. To avoid such disturbances, the
frequency operation range of distributed generators must be revised, as it was identified by many transmission system
operators (TSOs).3
Calculations based on statistical principles have shown that market‐induced imbalances (due to the schedule follow-
ing) and associated deterministic frequency deviations pose a considerable risk by procuring available reserves for other
uses.4 To reach the probability rate of significant dynamic events measured to 1 in 32.5 years (as in 2002, before the non-
synchronous generation growth), FCR would need to be increased3 by 120 MW. According to optimistic estimates, this
change would cost 31 M EUR annually.3 However, it would only reduce the risk of significant dynamic events resulting
in large frequency deviations within the range where older distributed generation units tend to disconnect, but it does
not constitute an effective alternative to long‐term corrective measures.
Such corrective measures were initiated by ENTSO‐E lately: Member states were called upon to take appropriate
actions to ensure that newly installed distributed generators meet the frequency cut‐off requirements set out in the Net-
work Code Requirement for Generators5 as soon as possible (but at latest before entry into force). Member states were
also invited to initiate a negotiation between system operators, regulatory authorities, and market participants in order
to determine the scope and implementation of the necessary adjustment programme at the level of member states.3
Partly as a reaction to the ENTSO‐E approach and partly because of ongoing research activities, TSOs have published
various limitations (introduced in Section 2.2. in detail) for RoCoF with usually determining boundaries of safe opera-
tion. As the penetration of nonsynchronous generators is expected to increase further, RoCoF limitations and calcula-
tion methodology must be periodically revised. In most cases, such revisions include extensive computer simulations of
dedicated power system disturbances while monitoring RoCoF, frequency extremes, and the level of available inertia in
the system. Several such publications are cited in the present paper, but there is a high research activity both concerning
theory and the analysis methods and mitigation opportunities. The aim of this paper is to overview the system dynamics
interpretations and TSO experiences (both with technical aspects and marketisation actions) worldwide and assess the
possible mitigation approaches from a practical viewpoint.
Tang et al6 examine the participation of wind power plants in frequency control processes of the power system. The
authors introduce natural and actual turbine time constants and use analytical expressions for power reduction. Iner-
tial support provided by wind turbines is in the focus of the work of Zeni et al,7 who model how variable speed tur-
bines could enhance primary frequency control. Their results include scenarios of reaching 50% penetration of wind
power and highlight that wind speeds near cut‐off range are the most critical from the security of supply point of
view. Similar ideas motivate the paper of Fernandez‐Bernal et al.52 The aim of the authors is to determine maximal
possible penetration level of wind turbines in the Spanish power system. Their conclusion is that contribution of var-
iable renewable energy resources is not necessary to maintain adequate values of primary reserve and inertia even in
high (71%‐82% and 66%‐79%, respectively) share of wind generation. In contrast, Jauch proposes a flywheel energy
HARTMANN ET AL. 3 of 14
storage system integrated into a wind turbine to support frequency control of the power system.53 A comprehensive
review on the most important issues was presented by Tielens and Van Hertem.26 The authors have taken the per-
spective of the TSO to identify challenges and potential solutions and have proposed a new definition of inertia. Sapari
et al12 address load‐shedding schemes for power systems with high nonsynchronous penetration, which is a widely
used countermeasure for increased frequency response, typically a result of decreasing inertia. Rokrok et al8 review
and compare various primary control methods that can be applied in microgrids with high renewable energy penetra-
tion and thus low inertia.
The remainder of the paper is organised as follows. Section 2 gives an overview on the theoretical background of the
inertia decrement issues and discusses operational experiences from power systems. Section 3 details the technical
and market options that are capable of providing a solution for the low inertia level. Advantages and drawbacks of
the solutions are shown. Finally, conclusions are drawn in Section 4.
2 | I N T E R P R E T A T I O N O F R O C O F AN D S YS T E M S P E C I F I C E X P E R I E N C E S
Synchronous inertia is associated with the instantaneous physical response of conventional generators having directly
coupled rotating mass, which acts to overcome the imbalance of supply and demand by changing the rotational speed
(and the electrical frequency as well), thus the kinetic energy of the unit. If measured frequency decreases, speed and
active power (Pgen) of the unit increases, and vice versa.
df df
Pgen > 0 IF < 0; Pgen < 0 IF > 0: (1)
dt dt
Large df/dt values may endanger secure system operation because of mechanical limitations of individual synchro-
nous machines (inherent capability), protection devices triggered by a particular RoCoF threshold value, or timing
issues related to load shedding schemes. Initial df/dt is the instantaneous RoCoF just after the disconnection of either
a generator or load from a power system, before any controls become active. This is theoretically the highest system
RoCoF. Its average for an interconnection of the number of N synchronous loads and generators can be computed as
follows9:
df f 0 Pk
¼ N
: (2)
dt t¼0þ 2∑i¼1;i≠k H i Si
If a power system's generation portfolio fully consists of synchronous generators, in case of a notable sudden change
in the active power (eg, generator outage, loss of significant load, and system split), RoCoF can be calculated by the
change in the kinetic energy stored inside the rotating masses of the machines.
1
d Jω2sys
dðE K:E: Þ 2 dωsys
Pgen ¼ Pmech − Pload ¼ ¼ ¼ Jωgen ; (3)
dt dt dt
where Pgen is the active power of all the generators in the system, Pmech is the time varying turbine power (mechan-
ical power) of these machines, Ploadis the frequency and voltage dependant, time‐varying consumption (including
losses), J is the moment of inertia, ωsys is the synchronous angular frequency, and EK,E is the kinetic energy of
the system.10
Considering a fully synchronous generation mix, the synchronous inertia constant (H) can be expressed with the
aggregated sum of apparent power (Ssys), each machine's Hi, Ji, and Si parameters and the system's ωn nominal angular
frequency as it follows:
1
H i Si ¼ J i ω2n : (4)
2
4 of 14 HARTMANN ET AL.
From which an aggregated synchronous system, inertia constant can be expressed as follows:
∑ H i Si Jω2n
H¼ ¼ ; (5)
Ssys 2Ssys
where Ssys is the cumulative apparent power of each generator in the system. It is important to note that the kinetic
energy of the system is an energy dimension parameter, usually given in MWs, while the synchronous inertia constant
is a time dimension (usually given in seconds) attribute. H is given by the ratio of the kinetic energy (synchronous iner-
tia) and the nominal active power of the machine (equal to the nominal apparent power as in Equation 4 above).
The acceleration time constant (or starting time constant, Ta) defines the time period that is needed to accelerate the
synchronous machine to its nominal speed, with the accelerating power equal to the machine's nominal apparent
power.
By substituting these expressions in Equation (2):
where dfsys/dt is the RoCoF of the system frequency. It can be observed that smaller H values result in higher RoCoF
and therefore more rapid frequency changes during the same time span.
Existing studies on the probable effects of decreasing inertia in the power system were widely published in the last
decade. During this period, technical and scientific work on this field has become more focused, and a number of meth-
odologies have become widely utilised. The primary question is not the possibility of decreasing inertia anymore but the
extent of this decrease, which can be handled by the examined power system without a need for extensive interactions
or disturbance of the market players. The group of simulated cases has become formulated and mostly includes the crit-
ical generator, the load or import/export point outages and resulting RoCoF values, and frequency. The latter values are
largely dependent on the characteristics of the examined power system, as inertia is a determining parameter immedi-
ately after the disturbance takes place, until the controllers (eg, FCR) can react. Since this time constant is directly
related to the technology of power plants, characteristic values are relatively easy to be determined.
It has to be emphasised although that results of such simulations should not be adapted to power systems with dif-
fering characteristics (eg, size, generation mix, and international connections), which also limits the coherent compar-
ison of such studies. From a general perspective, many large power systems—such as Continental Europe—have been
studied lately, but systems with such levels of installed (and interconnected) capacity do not show significant vulnera-
bility to low inertia. (A loss of 3000 MW in generating capacity would result in 0.8 Hz frequency decrease in the Con-
tinental Europe synchronous system, assuming that self‐regulating effect of load is 1%/Hz.) Three notable exceptions in
the range of 1 to 10 GW order of magnitude peak load are Ireland, Britain, and south‐eastern parts of Australia. In case
of smaller power systems, operational vulnerability is considered to be higher; the upscaling of simulation results and
solutions, however, has many limitations due to the often specific nature of these systems, like island operation.
In the following, an overview of both groups is presented, and a high‐level comparison of the most important results
is presented. It is worth noting that while professional literature mostly refers to low levels of inertia as the main oper-
ational issue of power systems, the real problem is the change in dynamic behaviour as a result of low inertia. Three
such problems are identified by transmission system operators1,9,11:
• Increased RoCoF values: Unbalance of supply and demand results in faster frequency changes during disturbances or
loss of mains protections can activate relays or cause damage in equipment.
• Lack of adequate reserves: As proportion of synchronous machines decreases, the number of power plants participat-
ing in ancillary services has become less, while increasing penetration of intermittent sources would also necessitate
reserves with faster response.
• Frequency extremes: Current protection schemes (especially under frequency load‐shedding and over frequency
generation‐shedding) may exacerbate effects of disturbances; thus, a change of protections settings might be
necessary.
HARTMANN ET AL. 5 of 14
In the following, an overview on operational experiences and regulations is presented, defining three categories for
the power systems in question.
Ireland and the power systems of EirGrid and System Operator of Northern Ireland (SONI) were the firsts in the world
where higher RoCoF values were identified as potential problems, in 2010. Early modelling indicated that penetration of
nonsynchronous producers has to be kept under 50% in order not to exceed 0.5 Hz/s levels, or settings of distributed
generator RoCoF relays have to be modified.13 In the SONI subsystem, such RoCoF rules are in force since 2001, but
the 1.5 Hz/s limitation has not been controlled rigorously.14,15 In October 2011, EirGrid and SONI have suggested to
increase generator RoCoF limits to 4 Hz/s, but due to significant opposition from the plant owners, a 1 Hz/s limit
measured over 500 ms time window has been accepted for all power plants (including incumbent ones).14-16 Further,
computer simulations have shown that a severe disturbance could result in 2 Hz/s RoCoF values in the SONI system,
which may eventually lead to the splitting of the system.61 Based on these results, a controversial decision was made, as
for generators of the SONI system, a temporary limitation of 2 Hz/s was proposed. But as the decision was unjustified
from the perspective of the generators, and the limit for them remained 1 Hz/s.17 The latest study on this topic was
published18 in 2016, in which it was stated once again that the desirable RoCoF limits are 1 Hz/s for the EirGrid and
2 Hz/s for the SONI system. Results of the DS3 (Delivering a Secure Sustainable Electricity System) were also intro-
duced, focusing mostly on technological alternatives to maintain RoCoF levels at 0.5 Hz/s.
National Grid, the transmission system operator of the British Isles, also faces increasing issues with nonsynchronous
generation technologies. Approximately one‐fourth of electricity demand is produced by renewable sources, half of
which is considered to belong to this category. The System Operability Framework,19 published in 2015, has identified
high RoCoF values as immediate challenges, prioritised at the highest level. The reason for this was twofold: Current
settings of RoCoF relays and decreasing volumes of ancillary reserves were both part of the problem. The study has
forecasted that by 2025 to 2030 due to the increasing penetration of solar photovoltaic and wind power and power flows
on high voltage direct current (HVDC) lines, outages may result in RoCoF levels exceeding the current limitation of
0.5 Hz/s. Experts have also expressed the need for fast frequency control reserves, reaching volumes three or four times
as big as today, by 2025 to 2030.
Australia has accelerated its research (Future Power System Security Program) in recent years, as the first complex
studies have highlighted that certain regions of the Australian power system might face significant operational issues
in a few years. The findings of initial research have identified that widespread use of solar photovoltaic generation
can result in unusually high RoCoF values, primarily in South Australia, secondarily in Tasmania and Queensland.20
The South Australian system—besides showing the highest share of solar photovoltaics in the country (58%)—is con-
nected to the main system via transmission lines with relatively low capacity. Simulated cases have assumed the outage
of Heywood Interconnector, which has led to increased RoCoF levels. (It is worth noting that recent capacity upgrade of
the line has only made the situation worse, as increased import‐export flows create higher supply‐demand imbalances
in the area.) Australian Energy Market Operator (AEMO) has taken immediate action, which includes adaptation of the
under‐frequency load‐shedding system, overview of disturbance protocols, and possible island operation of the South
Australia subsystem. Until the time of the writing of this paper, no RoCoF values have been published by AEMO,
but they are expected to be revealed, making the Australian power system another example of large systems, where
low inertia levels are identified as a threat to secure operation.
The New Zealand power system (operated by Transpower) has faced the most recent disturbance in 2011, when fre-
quency nadir was 47.5 Hz on the Northern Island. RoCoF values reached 0.73 Hz/s, but disconnection of generators
was not observed. Partly based on the experiences of this disturbance, a new under frequency load‐shedding scheme
was designed for Northern Island, with the installation of RoCoF relays in the fourth (last) step, set at 1.2 Hz/s.21
The effect of the future spread of wind power was also examined. Simulations showed that with the installation of
approximately 1300 MW of wind power, RoCoF would increase to 2.1 Hz/s. Nonetheless, there are no RoCoF require-
ments in force by Transpower.
6 of 14 HARTMANN ET AL.
In Hawaii, consecutive power system events have resulted in the disconnection of distributed generators due
to high RoCoF values. In July 2010, frequency nadir and RoCoF were 57.7 and 0.373 Hz/s, respectively, which led
to the modification of power plant control parameters, and position controllers were installed at gas turbine
fuel valves.59
The island of Cyprus also has a notable history of recent events. The most significant one was in January 2012, when
68 MW of wind power was shut down due to RoCoF levels exceeding relay thresholds (1.3 Hz/s vs 1.25 Hz/s). As a
result, RoCoF relay settings were increased59 to 1.5 Hz/s.
The Nordic countries of Europe initiated the Future System Inertia project in 2013 with four goals: (a) to elaborate on a
methodology for the examination of frequency disturbances and inertia; (b) to harmonise calculation of inertia; (c) to
develop a real‐time inertia estimator; and (d) to examine the effect of future supply and demand changes on system iner-
tia. Besides assessing the inertia constant for several power plant technologies, the study also highlighted that inertia of
the Nordic power system is expected to remain between 124 and 305 GWs, which is higher than the historical minimum
of 2009 (115 GWs); thus, severe RoCoF issues are not expected.22,23
Several studies are available for the interconnections of Northern America as well. Research conducted by National
Renewable Laboratory in 2014 showed that if the combined penetration of wind and solar photovoltaic panels reach
33%, RoCoF values are expected to increase24 by 18%; however, these values are still in the range of 0.1 Hz/s. Similar
simulation was performed for the California Independent System Operator area: The outage of Palo Verde nuclear
power plant (the loss of 2690 MW) could be handled by the operator even in case of renewable levels25 between 37%
and 50%. Results from the Electricity Reliablity Council of Texas (ERCOT) area are comparable, as even the most severe
single outage (2750 MW) would not increase RoCoF levels27 above 0.4 Hz/s.
Germany, a leading country in the field of renewable energy use, has not identified near future vulnerabilities related
to low level of inertia.
Denmark, another good example for widespread use of renewables, purchased synchronous compensators in 2013,
which could be used to provide inertial response, although this is not their primary task. As the Danish system has a
very high proportion of nonsynchronous generation, new power plants with a capacity above 1.5 MW have to with-
stand RoCoF values of 2.5 Hz/s, which is a slight increase compared with the former 2 Hz/s limit. The regulation
also defines that wind turbines and photovoltaic units must withstand such frequency changes28,29,60 for at least
200 ms.
TSOs assess the effects of the synchronous inertia reduction worldwide, mostly focusing on RoCoF limitations. There-
fore, RoCoF constraints could be seen as the main metric until today. However, it is important to note that in frequency
disturbances, the extremes and the time period while the system frequency is below a predefined value must also be
included when countermeasures are examined. Another aspect is that RoCoF is a noisy signal due to its derivative
nature. Consequently, a consistent definition is needed on the measurement of this parameter. Based on the literature,
500 ms time window averaging is the most commonly used solution, because it is both reliable and fast enough for prac-
tical applications, but, eg, another ENTSO‐E document suggests limits (2.5 Hz/s for HVDC, 2 Hz/s for photovoltaics,
and states that 4 Hz/s is feasible for wind turbines).9 Conventional generators were not planned to withstand high
RoCoF values; therefore, measurements and tests should be carried out to identify the technically feasible limits. RoCoF
limits could be defined in three ways effectively9:
• Short circuit powers: prefault and postfault minimum short circuit capacity at a connection point, prefault active and
reactive power, and voltage at the connection point are needed for calculations.
• Moving averages: eg, 2Hz/s for moving average of 500‐ms window, 1.5 Hz with 1‐second window, and 1.25 Hz/s for
2‐second window.
• Creating a frequency‐time profile (see Figure 1) with extremes (zenith, nadir), steady‐state offset, disturbance ride
through capability, and consecutive df/dt ramps (with different time windows).
HARTMANN ET AL. 7 of 14
In large systems, the system splits, while in smaller synchronous areas, usually the outages (generation, significant
load, or import/export connection) are the most severe events. Table 1 covers the main conclusions of the TSO experi-
ence comparative analysis. RoCoF is a local parameter (frequency itself is different at distant buses during electrome-
chanical transients); therefore, different limits come up in the same synchronous area as well (eg, Denmark above).
By reviewing the actions taken by TSOs, three group of solutions can be formed, which are all considered to prevent
further increase of RoCoF levels in power systems. Two of these groups focus on technical solutions, primarily the
use of synthetic inertia, while the third aspect emphasises the importance of proper market structures and services. It
is a common position of the energy industry that the combination of these solutions is necessary to achieve optimal
results.
The need for inertial response emulation grows as more and more generators are being connected to the system via
power electronic converters. The concept of synthetic inertia was first proposed by Beck and Hesse.30 Since then, many
control approaches and physical model implementations have been published. Synthetic inertia is a combination of
Large scale, Continental Europe, Multiple AC connections Calculations do not System splits
interconnected Nordic countries within a synchronous area show significant
operational challenges,
further studies on
increasing non‐synchronous
share is in progress
Large scale, Ireland, Great No synchronous Calculations show Outages (HVDC,
island Britain, Australia connection (only operational challenges, generator)
through HVDC or TSOs already address
other converter solution) solutions
or very limited, weak
connections
Small scale, Hawaii, No synchronous connection Wider limits, severe Outages, faults
island New Zealand (only through HVDC or disturbances already
other converter solution), observed, different
or very limited, weak measures for operation
connections
8 of 14 HARTMANN ET AL.
algorithms that emulate a mathematical model of synchronous generation, renewable energy generators, or storage
devices as a source and power electronics.10 All the approaches use a feedback of grid‐side measurements for the con-
troller, usually voltage, current, and frequency at the connection point. However, there are studies which evaluate the
operation of cooperating synthetic inertia providers; therefore, these may investigate other signals and communication
infrastructure.10,31
Three main separated concepts are presented in the literature for synthetic inertia controllers that are summarised in
Table 2.
Virtual synchronous generator model‐based solutions—or so‐called “synchronverters”—consist of equations that
define the relation of the machine's electrical and mechanical parameters, such as torque and flux equations and virtual
impedances in different depth. In some variants, exciters are also included to provide a complete model of synchronous
generators based on measured frequency and voltage. An important advantage of these methods is that RoCoF is not
used as an input, so the measuring issues related to this parameter are avoided. The moment of inertia and damping
factor parameters could be tuned to effectively model the electromechanical transient behaviour of the emulated
generator, and these parameters are well‐known in stability studies.32-34
The second group consists of the swing equation‐based topologies. This approach leads to reduced complexity com-
pared with synchronverters. A disadvantage is that RoCoF is used as an input, so proper tuning of the control and signal
processing parameters are important. Lots of variants neglect the damping part of the equation35,36—as a worst‐case
approach from the perspective of stability—but if it is considered, another controller type could be formulated (some
sources call them virtual synchronous generators).10,37,38
Other frequency dependent active power responses form the third group. This means that instead of using models of
physical devices (machines or swing equations as a subsystem aggregation), the scope is on the provision of positive
effects by controlling. The most common variations are P‐f droop controllers,38,39 synchronous power controllers,40,41
and virtual oscillation controllers.42,43 These types of synthetic inertia providers are only constrained by the generation
or storage technology and the power converter capabilities. There are several approaches which use soft computing
methods in the optimisation of the emulation.44-46
Considering the conventional grid control approaches, a possible way to handle the reduction of inertia is to imple-
ment operation constraints for transient stability processes.47 As an example, EirGrid and the Irish regulator carried
out a research project which implements a minimum available kinetic energy constraint in the unit commitment and
economic dispatch calculations. Frequency stability is ensured by the synchronising torque, rotating mass, and fre-
quency control of synchronous generators. But these units are being substituted with zero fuel cost, nonsynchronous
solar, and wind generators. Additionally, there are a lot of high voltage direct current connections in Ireland as well,
“Synchronverters” Accurate model which is coherent with the Numerical instability issues
electromechanical transient studies
RoCoF not needed Voltage source model no inherent
PLL is only for synchronisation overcurrent protection
Swing equation Easier to apply and tune than the whole synchronous Poor tuning could lead to swings
generator type
Still connects to stability terminology Voltage source model no inherent
overcurrent protection
Uses RoCoF as an input
Other active power‐ Adaptivity and wide‐range of applicability Non‐common approaches in grid control
frequency relation Easy to apply if droop control is used Communication infrastructure might be
needed
Technology limits could be considered Use of PLL could lead to instability
HARTMANN ET AL. 9 of 14
which results in extremely low synchronous inertia in several cases when the load is low while nonsynchronous
generation and import share are high. Therefore, the project's conclusion was a suggestion to integrate a minimum
kinetic energy criterion in the dispatch, which ensures that RoCoF does not exceed a defined value (0.5 Hz/s in
Ireland at the time of the project). The calculation of the limit is based on the swing equation and deals with the
most severe single fault (loss of the largest generator or HVDC infeed), and hence, it is an energy‐based, topology‐
independent approach.48
Consequently, new services are being developed, and in general, the existing structure of the ancillary services mar-
ket is being reviewed to prepare marketisation. The common idea behind these processes is to distinguish three separate
service categories in the form of synchronous inertial response (SIR), FFR, and frequency containment response. In the
following, the market example of four regions is introduced, and a comparison is given in Table 3.
TSOs of the Irish Island have introduced a total of six new services lately, including SIR, FFR, Fast Post‐Fault Active
Power Recovery (FPFAPR), and three different Ramping Margins (RM1, RM2, and RM8), two of which are closely
related to the issue discussed in the present paper. SIR is a response in terms of active power output and synchronising
torque that a unit (generator, synchronous condenser, and load) can immediately provide after disturbances. The service
is defined as the kinetic energy of the unit, multiplied by the SIR factor. The SIR factor of a generator is calculated as the
ratio of kinetic energy to the lowest sustainable power output at which the unit can still provide reactive power control.
The threshold of the factor is between 15 and 45 seconds, where the higher value is considered for condensers and loads.
Severe criticism has surfaced, as the thresholds limit the remuneration of inertia in the system (eg, if a power plant can-
not provide inertia at a low active power output, its factor might be below 15 s), but the regulatory authority did not
change the limits in order to maintain economic signals and motivation to provide such service. According to the Irish
regulation, synthetic inertia is considered to be distinct from SIR, as most of the technologies capable of providing an
emulated inertial response are not synchronously connected to the grid. The other newly introduced service related
to inertia is the FFR, which is defined to bridge the gap between SIR and existing FCR service and is expected to mit-
igate RoCoF in case of disturbances. Technically, FFR means additional increase of active power output from a gener-
ator or decrease of demand by consumers within 2 seconds of the start of the event. Volume of available FFR is
measured based on the additional power output provided. If a market player is capable of delivering both SIR and
FFR, it is allowed to participate with both services.58
National Grid initiated work on new services in 2010, which has led to the introduction of the Enhanced Frequency
Response (EFR) service. EFR requires active power increase in 1 second, sustained for 15 minutes.55 EFR is a continu-
ous service, aiming to manage system frequency prefault through a droop response characteristic; thus, it is fundamen-
tally different from other services discussed in this section, as they are all postfault services. A similar service is found on
the PJM (Pennsylvania‐New Jersey‐Maryland Interconnection) market in the United States, but it performs control
based on central measurements, while the EFR utilises local frequency values. The relatively long sustainment require-
ments exclude certain technological options, notably synthetic inertial response from this market, but on the other
hand, it is fit for energy storage. National Grid has already launched tenders for EFR, with two types of services, which
only differ in the size of the deadband (±0.05 Hz or ±0.015 Hz); the narrower deadband was more favoured.
AEMO has been taking significant steps in the process of introducing new services on the market; the first proposals56
were published in 2017. Similar to other power system operators, AEMO emphasises that these new services are not equiv-
alents of conventional inertial response but are faster than the existing FCR service. The plans in Australia include seven
new services, for all of which necessary response time, frequency‐deadband, and the range of possible technologies were
defined. These services are emergency response, contingency FFR, primary frequency control, fast response regulation,
simulated inertia, and grid‐forming technologies. All services have to be activated within 1 second either in proportion
to frequency deviation or in an event driven manner. Detailed plans for the services are found in Tielens.49
ERCOT started the examination of possible development of new services on the frequency control market in 2012. A
broad range of services were published in 2013, which are expected to be implemented27,57 by 2019.
SIR is defined as an instantaneous response from synchronous machines, calculated as the stored kinetic energy
thats extracted from the rotating mass of the machine after a disturbance. Accordingly, the response of a given
machine is independent of the active power output of the unit. ERCOT also notes that synthetic inertial response
cannot be considered as an SIR service. As high RoCoF values are not identified as a problem in the supply area of
ERCOT, marketisation of the service is expected to be slow, so implementation is handled separately from other
products.
An unusual solution has been chosen in connection with FFR as well, since active power output has to be sustained
for a significantly long period, which practically excludes units capable of providing synthetic inertia from this market.
The product definitions of ERCOT also create an overlap between FFR and PFR services in terms of response time,
which usually decreases efficiency. Requirements for these services are to be determined for each operating hour and
procured on the day‐ahead market.57
4 | CONCLUSIONS
The decrease of rotational inertia in power systems has initiated various approaches at system operators around the
world, leading to a number of published solutions. The most common way of thinking is to apply synthetic inertia,
which is a combination of algorithms that emulate the mathematical model of synchronous generation based on
physical models. Other technical and nontechnical alternatives can be identified as well. One of the main directions
suggest using a purely control‐based approach instead of complex dynamic equations; droop control of cooperating
power plants provides very similar response before the activation of FFR. Another direction sees rotating masses as
an indispensable part of power systems, and either prescribes a minimal level of inertia to be kept in the system or uti-
lises synchronous compensators to provide a technical alternative. This latter solution enables the use of conventional
dynamic equation‐based calculation in future systems as well, providing a smoother transition for the personnel of sys-
tem operators. A third alternative is to exploit the potential of demand‐side management, which—despite the increasing
share of converter‐based consumer devices (including electric vehicles)—is a viable alternative. Marketisation of all pre-
vious options is also a crucial challenge for regulators and market operators in the near future. To preserve the stability
of power systems in the future, the disappearing rotating mass should be handled with a complex approach, considering
technology (generation and storage), reserves, market models, and regulation frameworks as well.
Among the technical solutions, synthetic inertia has clearly gained the most attention lately. However, several issues
are to be discussed in connection with this technology. One of the main questions is how efficiently synthetic inertia can
substitute direct physical response of the system (synchronous inertia). Published results show that response times are
in the range of 50 to 200 ms, which are very fast, but in relation to time constants of dynamic power system events, they
are far from being instantaneous. This suggests that in order to limit RoCoF levels during disturbances, SIR is necessary
to a certain extent. However, exact control techniques, parameter settings, and efficiency of different technological alter-
natives are questions of the future, while other options are being evaluated as well, like market and dispatch constraints,
synchronous condensers, demand side inclusion, and adaptive protection schemes.
Further analysis is needed on the transient behaviour of systems with a high share of nonsynchronous generation.
The modelling and calculation methods must consider the characteristic differences of such technologies compared with
the synchronous generation, which leads to the evaluation of the stability definitions and criteria as well.
12 of 14 HARTMANN ET AL.
Generating synthetic inertia can be a solution to some problems—further steps are needed to maintain system
stability (Figure 2).
Authors of the present paper have initiated a joint research project with experts of MAVIR Hungarian Transmission
System Operator, to examine related issues in the Hungarian power system and to share the most important results with
the power system community.
A C K N O WL E D G E M E N T
This work was supported by the ÚNKP‐2017‐4 New National Excellence Program of the Ministry of Human Capacities,
Hungary.
ORCID
Bálint Hartmann https://orcid.org/0000-0001-5271-2681
R EF E RE N C E S
1. ENTSO‐E, Need for synthetic inertia (SI) for frequency regulation, ENTSO‐E guidance document for national implementation for
network codes on grid connection, 2017
2. European Commission, Commission Regulation (EU) 2017/1485 of 2 August 2017, establishing a guideline on electricity transmission
system operation, 2017.
3. ENTSO‐E, Dispersed Generation Impact on Continental Europe Region Security, ENTSO‐E Position Paper, 2014.
4. ENTSO‐E & EURELECTRIC, Deterministic frequency deviations—root causes and proposals for potential solutions, 2012.
5. European Commission, Commission Regulation (EU) 2016/631 of 14 April 2016, establishing a network code on requirements for grid
connection of generators, 2016.
6. Tang C, Soong WL, Frere P, Pathmanathan M, Ertugrul N. Dynamic wind turbine output power reduction under varying wind speed
conditions due to inertia. Wind Energy. 2013;16(4):561‐573.
7. Zeni L, Rudolph AJ, Münster‐Swendsen J, Margaris I, Hansen AD, Sørensen P. Virtual inertia for variable speed wind turbines. Wind
Energy. 2013;16:1225‐1239.
8. Rokrok E, Shafie‐khah M, Catalão JPS. Review of primary voltage and frequency control methods for inverter‐based islanded microgrids
with distributed generation. Renew Sustain Energy Rev. 2018;82:3225‐3235.
9. European Network of Transmission System Operators—Electricity (ENTSO‐E), Rate of Change of Frequency (RoCoF) withstand
Capability, ENTSO‐E guidance document for national implementation for network codes on grid connection, 2018
10. Tamrakar U, Shrestha D, Maharjan M, Bhattarai BP, Hansen TM, Tonkoski R. Virtual inertia: current trends and future directions. Appl
Sci. 2017;7:1‐29.
11. ENTSO‐E, Frequency stability evaluation criteria for the synchronous zone of Continental Europe—requirements and impacting factors,
2016.
12. Sapari NM, Mokhlis H, Laghari JA, Bakar AHA, Dahalan MRM. Application of load shedding schemes for distribution network
connected with distributed generation: a review. Renew Sustain Energy Rev. 2018;82:858‐867.
13. EirGrid/SONI, All Island TSO Facilitation of Renewables Studies, 2010
14. EirGrid/SONI, RoCoF Modification Proposal—TSOs' Recommendations, 2012
15. EirGrid/SONI, DS3 Rate of change of frequency modification recommendation to the CER, December 2012
16. Commission for Energy Regulation, Rate of change of frequency (RoCoF) modification to the grid code decision paper, 2014
17. Commission for Energy Regulation, Rate of change of frequency (RoCoF)—review of TSO and generator submissions—final report, 2013
18. EirGrid, RoCoF Alternative & Complementary Solutions Project, Phase 2 Study Report, 2016
19. National Grid, System Operability Framework 2015—UK electricity transmission, 2015
20. Australian Energy Market Operator, Future power system security program progress report, 2016
21. Transpower New Zealand, System Operator TASC ‐ TASC 033 Report, 2014
22. ENTSO‐E, Future system inertia, 2015
23. Spahic E, Varma D, Beck G, Kuhn G, Hild V. Impact of reduced system inertia on stable power system operation and an overview of possible
solutions. Boston, MA: 2016 IEEE Power and Energy Society General Meeting (PESGM); 2016.
24. National Renewable Energy Laboratory, Western Wind and Solar Integration Study Phase 3—Frequency Response and Transient Stabil-
ity, 2014
HARTMANN ET AL. 13 of 14
57. Synapse Energy Economics, Synapse Comments on FAST Proposal in ERCOT, Cambridge, Massachusetts, 2014
58. Commission for Energy Regulation, Utility Regulator, Single Electricity Market—DS3 System Services Technical Definitions—Decision
Paper—SEM‐13‐098, 2013
59. Australian Energy Market Operator, DGA Consulting, International Review of Frequency Control Adaptation, 2016
60. Energienet.dk, Technical regulation 3.2.2 for PV power plants with a power output above 11 kW, 2015
61. EirGrid/SONI, Northern Ireland Separation Studies, 2012
How to cite this article: Hartmann B, Vokony I, Táczi I. Effects of decreasing synchronous inertia on power
system dynamics—Overview of recent experiences and marketisation of services. Int Trans Electr Energ Syst. 2019;
e12128. https://doi.org/10.1002/2050‐7038.12128