Paris 2020

A1-101

Is reliance on synchronous machines holding back

D.VAUGHAN*

Entura
Australia

SUMMARY
This paper explores the need for a changed paradigm in power system thinking. The age of the large-
scale fossil fuel thermal generating unit is rushing to a close. This paper provides discussion on a
number of aspects relating to the transition to more inverter-based generation. The discussion is
intended to provoke thinking as to whether current philosophy and practices relating to power
system control, protection and design remain best practice when the generation fleet, and to some
extent the customer loads, are changing or have changed.

The electrical power grid is energy-source agnostic. It cannot distinguish between energy converted
from water, wind, sun, hot rocks, coal, gas or uranium. The way these sources of energy are
controlled and converted to electrical energy does matter to the grid but the grid has been designed
to work well for a certain type of generator, namely 3-phase synchronous machines.

The genius and flexibility of the AC machine and transmission system facilitated massive growth in
carbon emissions because it was easy to adapt steam technology, which powered the massive
growth of industrialisation through the 20th century and beyond, ushering in greater wealth and
prosperity.

To limit the damage wrought by climate change, reducing carbon emissions is increasingly urgent.
This challenge may require multiple actions but the likely energy mix will inevitably play a role. Solar
and wind will no doubt comprise a large share of the new mix. These energy sources invariably
require a form of inverter technology to convert the raw power source: either asynchronous AC from
wind turbines or DC from photovoltaic solar panels. These inverters interact with the power system
in a way that is physically different from the way of synchronous machines, and so react to power

[email protected]
system disturbances in different ways. This difference leads to questions about the appropriateness
of current power system paradigms and raises the need to consider alternative approaches.

This paper explores alternative approaches in the areas of loads, generator and grid controls, and
system security.

Just as generator technology is advancing, power electronics are forming a significantly greater share
of the load, leading to the possibility of flexibilities in terms of frequency and voltage, at least on a
domestic level. Additionally, smart-grid technology can be used to vary demand in real time to
manage mismatches between generation and demand.

Generator controls have traditionally been autonomous or at least localised. Grid controls
(specifically protection) have been local and compartmentalised (with some notable exceptions, such
as special or system protection schemes). Modern communication techniques coupled with high-
speed data processing can provide alternative approaches.

A combination of advanced generation and network control, smart use of demand control, and
changes to supply standards and approaches can start to migrate the power system away from the
grid of the 20th century and into a more flexible and secure grid for at least the next few decades of
the 21st century.

KEYWORDS

Inverter – Renewables – Power System Stability – Frequency Control – Voltage Regulation – Fault
Ride Through – Demand-side management

[email protected] 2
Generator inverter level control

The AC system relies on controlling two fundamental quantities: frequency and voltage. A third
requirement is for power generation to be shared. Inertia and fault level have been a regular topic
over recent years. Essentially, they represent the means of controlling frequency and voltage. Inertia
operates to restrict the rate of change of frequency, allowing control time to drive frequency back to
nominal (within the limits of AC generators). Fault level is a measure of how strongly the voltage at a
point in the network is regulated by remote sources.

Frequency is perhaps one of the most romantic aspects of power systems. Machines spinning at
disparate ends of the grid spin in synchronism with each other. They speed up and slow down
together through good times. In bad times they may separate. Sometimes they fly apart, resulting in
darkness.

Voltage is only a localised indicator of health, but that health can be gone in an instant. At low fault
levels, voltage is about beauty and form but has no substance.

Everything depends on these two fundamental quantities and the philosophy of control that is
adopted. The brilliance of synchronous machines lies in their ability to operate autonomously, relying
on similar units elsewhere to cooperate. As the synchronous machines are disconnected or
displaced, the ability to rely on each other is reduced until all are brought down.

The major difference between inverter-based technology and synchronous machines is the ability to
track system frequency, and the means by which the inverter can then control the amount of real
and reactive power it generates and hence its contribution to frequency and voltage control.

The mechanical inertia in the power system will diminish as inverter-based generation replaces
synchronous generation. This will increase the risk of instability due to load and/or generation
changes since the system will become less robust to these changes. The result will be larger
deviations in frequency and less margin with which to control frequency to its nominal value.

There are two possible future scenarios: a low-inertia grid or a zero-inertia grid. In a low-inertia grid,
the most likely scenario in the near term is that some sort of spinning plant will be available to
provide enough inertia to allow other inverter-based controls to provide the additional fast
frequency response required. There is then some chance of regulating frequency using high-speed
communications or some sort of off-grid synchronising signal or metronome.

Considering how a zero-inertia grid works, there need to be even grander methods for frequency and
voltage-angle control than the wide area observation [1] required at low inertia. With no inertia, that
is only inverter-based plant, frequency is just a number for an inverter. Their controls will attempt to
lock onto a signal and determine what the frequency of the network is and produce power
accordingly. The problem is these controls (to date) have been created such that they rely on inertia.

To explore how an off-nominal frequency could arise in a zero-inertia grid, as well as its impact, it is
necessary to consider how power sharing and demand will be controlled across the power system.
This is not simple at zero inertia. Voltage magnitude and angle are the key to regulating power flows.
All is well if all the inverters are outputting to their power setpoint and the power system is balanced
through some sort of miraculous control. The question arises as to how to detect an imbalance.

This assumes the presence of a mix of voltage source and current source inverters. Current source
inverters rely on some other device maintaining the voltage’s integrity both in terms of frequency
but also shape. In a zero-inertia (or more importantly zero-synchronous-machine grid) this must be

[email protected] 3
provided by a voltage source inverter in grid-forming mode. That is, it has a stiff control of the
frequency of the voltage that it generates and a similar stiffness to its voltage magnitude. This means
that imbalances in the network are mopped up by these units. They can operate in isochronous
mode (that is, with infinite frequency droop) or they can use a finite droop control. This finite droop
is more compatible with the way a grid with inertia works. If it is adopted at zero-inertia, frequency
deviation can remain as a signal of power imbalance.

This system arrangement relies on the strength of the voltage source inverters within the system to
regulate the voltage waveform at all points in the power system. A very large installed capacity of
these inverters will be needed in order to provide sufficient system strength. This appears to replace
the benefits achieved from the synchronous machines with an alternative source of those benefits.
However, it is important to question whether that is the most efficient response or whether it results
from a bias towards a synchronous machine network.

This bias is widespread. In particular, it is inherent in the design of current source inverters. They rely
on undistorted voltage waveforms to allow accurate imprinting of current but do nothing to minimise
their own effect on the distortion of that voltage waveform. It is interesting to consider whether
some of the system strength requirement could be reduced if the level of power injection was
dependent on the shape of the resultant voltage waveform just as inertial response for synchronous
machines is dependent on rate of change of frequency. This would allow local stability to be
maintained and the inverter equivalent of transient stability to self-manage with less reliance on grid-
forming inverters to provide system strength.

This is hypothetical, but it seems as though the limitations of inverter technology could be overcome
by adding alternative controls. Inverters are not rotating AC machines and so offer different
capability. This different capability may provide a new controls, impossible for Tesla, Edison and
Westinghouse to envisage at the dawn of AC transmission, that might arrive at a similarly robust
result. That is, rather than try to bend inverter technology to emulate synchronous AC machines or
continue to rely on them.

Ma et al. [1] suggest ways of sharing the frequency regulation burden across multiple inverters by
calculating and communicating a system frequency target (centre of inertia frequency). This might
allow greater stability of inverter controls because it would provide inverters with a first-stage
approximation of what frequency should be and so improve the inverter’s ability to track voltage.
Hammer et al. [2] demonstrate ways of sharing the inverter-based inertial control in a stable way.
These are examples of expanded system thinking, in which the flexibility of the inverter is used to
greater advantage than it would otherwise be if limited by the synchronous machine paradigm.
However, even discussing inertia betrays some continued adherence to that paradigm.

This paper has so far outlined some of the benefits in changing the thinking relating to inverter
controls. In order to realise these benefits, it is necessary to stop thinking that the grid can continue
to rely on synchronous machines. Consequently, it is necessary to stop thinking that all the ways of
controlling the grid that make sense for a grid with synchronous machines would continue to be
appropriate in a grid without them. This is not easy for an industry that has thought this way for 4 to
5 generations.

[email protected] 4
Compatibility of loads

There is a slow but inexorable migration of power usage, particularly in non-industrial settings, from
directly AC-connected loads such as radiators and incandescent light globes to inverter-fed heat
pumps and LED lights. Almost all electronic devices are now immune to large variations in frequency
and voltage. Even an inspection of the input specifications for a phone charger will indicate that they
are very adaptable. The questions then arise of whether grid standards may able to be relaxed over
time, and why the standards are the way they are.

Certainly, frequency standards vary widely across the globe already. The European standard of +/–
0.8 Hz is stricter than the Australian mainland standard of +/– 1 Hz, which is stricter than the
Tasmanian standard of +/– 2 Hz. This suggests that customers can tolerate wider variation than the
standard to which grids are operated now. In a grid in which inertia or inertia-like controls come at a
high cost, it may not be sustainable for the standard to be maintained at these high levels.

The supply standard appears to be set to suit the highest practicable requirements and to then
provide that level of supply to everyone. With the ability of power electronics to tailor supply
conditions for sensitive loads, there may be value in maintaining a lower grid-wide standard and only
reinforcing or providing special arrangements for localised supply points. This may increase the
margins for control in the main network.

This leads to consideration of time-of-use compatibility. It has always been a struggle to align
consumption with production and vice versa. Various jurisdictions have enforced time-of-use tariffs
or set timer off-peak meters to attempt to better align production and consumption. Storage of
various forms has been used, such as heat banks and now batteries. The big step in thinking that will
yield efficiencies is user behaviour. If it is recognised that some industrial processes are more
efficient as a ‘base load’, then variability in renewable power production must be dealt with by other
users.

Storing power is inefficient due to cycling or decay of stored energy. Alignment of use and generation
is more efficient. Various jurisdictions rely on price or forecasting to signal possible surplus periods
and then expect demand to react. This is too passive. Strong alignment between available generation
and demand will only be achieved through a more active regime. This regime is as much behavioural
as it is technological.

Some energy-conscious behaviour already exists, such as doing the laundry when it is sunny, or
closing the curtains when it is hot. If consumers are given the right signals and controls in a timely
fashion, then patterns of energy usage could be seamlessly or frictionlessly adapted to suit the
supply balance.

In the past, there was a problem with sufficient night-time demand to keep thermal units running
above minimum loads overnight. The problem (if indeed it is a problem) now is to keep them running
at all in some circumstances. Where once customers were incentivised to help overcome the short-
comings of generation plant, now some generating units compensate customers for misalignment of
demand and supply.

Understanding the nature of renewable generation patterns and then designing controls and
facilitating behavioural change in energy use patterns may be one of the paradigm shifts required to
make the transition to renewable generation more efficient and sustainable. Storage is an
alternative, expensive solution; however, large efficiencies will be achieved by developing
mechanisms that expand the ‘smart’ charging concept beyond dedicated batteries and electric
vehicles and into other appliances and processes.

[email protected] 5
Power system security

Synchronous machines deliver massive fault currents when the network fails, which has made
detection and discrimination of fault events easy. The network has been designed to manage these
fault levels, with fault-limiting devices often installed to decrease the cost of lower voltage networks.
This paradigm is challenging to maintain if synchronous machines are not operating. Inverter design
cannot efficiently deliver equivalent fault levels. Something must change.

There are already a number of effective coping methods. Differential protection will detect faults
almost regardless of fault level. Other systems such as RCDs can detect irregular current flow and
protect equipment and people. It is simply a matter of being smarter. It will be expensive in places
but it is increasingly apparent that the cost of not decarbonising is also very high.

Power system security is essentially a matter of stability and reliability of supply while ensuring safe
operation of power system equipment. That is, to keep as much of the network supplying customers
without overloading anything and to ensure that faults are detected and removed as quickly as
practicable, before injury to people or the wider environment. Speed and discretion is required.

Speed and discretion of protection relates to the ability to detect, discriminate and then switch. All of
these elements may include a communication component. Rapid advances in processor capability
have led to greater precision and speed in protection operation but these systems are still designed
for a particular set of circumstances with assumptions built into their settings and operation. These
assumptions may be appropriate in the bulk transmission network, but in radial networks with
embedded renewable generation the number of scenarios and system conditions may expand.
Security of supply may be enhanced by considering what inverter-based technology can withstand.

In theory, inverter technology should be more resilient than synchronous machines to some system
imbalances. An inverter-fed radial connection could quite easily operate a single-pole auto-recloser
system maintaining more of customer demand. A synchronous machine could not handle such an
imbalance for any length of time. Granted, there are many system events for which inverters are not
more resilient. Indeed, the controls and signal processing available from the inverter may be helpful
information for the protection system to understand what sort of event has occurred and then act
accordingly. Greater distribution of generation may also lead to a lower level of required reliability
for some network elements since islanding proportionally small generation sources may not have a
massive impact on system security.

Protection settings and system design should consider and leverage the contribution of inverter-
based technology to system security rather than continue to limit functionality to what a
synchronous machine might withstand or demand that the inverter behave in a similar way to the
synchronous machine.

Previous discussion in this paper about inverter controls was more about individual controls rather
than a system-wide view. If the focus is now shifted to a system-wide view, a number of challenges
of inverter controls come to the fore, as well as virtues that are tempting but ultimately perilous.
These relate to the speed at which the controls can adjust the output of the inverters to affect
voltage and power output and the coordination of these controls across the network. Hammer et al.
[2] show that increasing virtual inertia can lead to instability in load sharing and frequency control –
the very things that virtual inertia is intended to assist. This is classic control theory where high gains
and slow feedback leads to instability.

[email protected] 6
If a system variable cannot be measured and then controlled fast enough, this limits the strength
with which controls can be applied. While speed of sensing is improving, Ma et al. [1] doubt that
communications are fast enough to allow high-inertia gains to be set across the network. Peng et al.
[3] conclude that better modelling of entire control loops (including external communications) will
ultimately be required to ensure optimal system-wide tuning of power and frequency control
responses from distributed inverter systems. While proposing that there is great potential in
exploring these controls, they caution that the problem is not trivial.

All of this presupposes that the voltage waveform and its unifying presence across the power system
is strong enough to hold the system together.

Voltage regulation and control is essential for system security. The example of South Australia has
demonstrated that sufficient voltage regulation and control can be achieved with relatively small
numbers of synchronous machines in strategic locations, even though the network is relatively
unmeshed. This gives sufficient fault level to allow the current source inverters to function properly
and for synchronous machines to remain synchronised. However, it is possible that this could yet be
another example of limited thinking. There may be another source of fault level that might be more
flexible than synchronous condensers and lead to better transient stability outcomes.

In the synchronous grid, system tuning techniques are used to tune power system stabilisers on
synchronous machine excitation systems to damp out power swings in the system. Tolerable
performance requirements are set for the tuning of step responses from these units to ensure they
respond adequately to system faults and switching events. These same rules are applied to inverters
but not to the fundamental controls of the inverters themselves. The mechanism that the inverters
use to detect the voltage waveform phase (the phase-locked loop, PLL) is more important from a
transient stability perspective. Pico and Johnson [4] show that careful modelling of the PLL loops and
protection systems is vital to allow prediction of performance. Taul et al. [5] further show that the
incorrect coordination of PLL gains can lead to unstable operation under some grid conditions. This
means that thinking in terms of performance requirements and proxies for stability observation
needs to change in line with the radically different generating plant that inverter-based technology
presents.

The increasing predominance of inverter-based generator controls requires rapid change in the
thinking about transient stability, protection and fault management in the power system to protect
against stifling of renewable development, adding unnecessary cost for consumers or detrimentally
affecting the robustness of the power system. This new thinking must incorporate an understanding
of the various levels of control embedded in inverter-based generators and the degree to which they
affect the wider network.

Conclusions

In conclusion, it appears that reliance on synchronous machines is impeding the evolution of the
power grid to facilitate renewables. The ways this occurs include:
1. focusing on inverters emulating synchronous machines rather than leveraging the strengths
of the inverter technology
2. designing inverter systems to rely on the existence of synchronous inertia and system
strength rather than developing controls that allow adaptation of operation at lower system
strengths
3. ignoring more complex wide-area monitoring and control techniques that may enhance the
operation of large numbers of inverters across the grid
4. continuing to adhere to high standards of frequency regulation rather than exploring ways to
relax these requirements for the majority of load points

[email protected] 7
 5. over-relying on technology alone to solve imbalances in demand versus generation, leading
to inefficient and costly solutions. Public information and clear signalling of times of surplus
generation may lead to behavioural change and reductions in surpluses
6. lack of consideration of strengths of inverter controls in terms of resilience to imbalance and
their possible role in fault location detection
7. assessing stability of inverter controls using benchmarks more suitable for synchronous
generation
8. relying on synchronous machines to provide power system stability and inverters to do no
harm.

Despite some academic and industry work being undertaken across many of these areas, the general
thinking still appears to be dominated by a synchronous-machine-based paradigm. This is not always
misplaced. There is no denying the spectacular success of the synchronous-machine-based grid. The
grid of the future may need different solutions to enable inverter-based or inverter-dominated grids
to provide adequate or satisfactory performance in the absence of the support from synchronous
machines. Since inverter-based technology is fundamentally different to the rotating machines that
have dominated the grid until now, it is reasonable to expect that the grid of the future will be
fundamentally different in terms of its capability and flexibility.

At this point in the transition from synchronous-machine-dominated grids to a grid in which inverter-
based generation pervades, the focus is still on what is being eroded rather than what is being built.
The electricity industry is being forced to adapt quickly. Efficient adaptation is a challenge that, if
met, will bring great value to customers now and into the future. This paper argues that such
adaptation cannot occur efficiently unless the grid design begins to leverage the strengths of
inverter-based technology while mitigating shortcomings.

Many challenges remain; however, the sooner a new grid paradigm is embraced, the sooner the
capabilities of inverter technology can be leveraged unfettered by the insistence that they emulate
synchronous machines.

BIBLIOGRAPHY

[1] Y. Ma et al., ‘Enhancing Power System Transient Stability by Virtual Synchronous Generator
Control Using Wide-Area Measurements’ (IEEE Energy Conversion Congress and Exposition,
2019, pp. 2546–2551)
[2] B. Hammer et al., ‘Effects of Virtual Inertia on Stability and Power Sharing in Inverter-Based
Power Systems’ (International Conference on Clean Electrical Power, 2019, pp. 375–381)
[3] Q. Peng et al. ‘On the Stability of Power Electronics-Dominated Systems: Challenges and
Potential Solutions’ (IEEE Transactions on Industry Applications, 55(6), 2019, pp. 7657–7670)
[4] H.N.V. Pico & B.B. Johnson, ‘Transient Stability Assessment of Multi-Machine Multi-Converter
Power Systems’ (IEEE Transactions on Power Systems, 34(5), 2019, pp. 3504–3514)
[5] M.G. Taul et al. ‘Systematic Approach for Transient Stability Evaluation of Grid-Tied Converters
during Power System Faults’ (IEEE Energy Conversion Congress and Exposition, 2019, pp. 5191–
5198)

[email protected] 8