Professional Documents
Culture Documents
Application Challenges GFM
Application Challenges GFM
Flicker,
Robert H. Lasseter, Hugo N. Villegas Pico, Gab-Su Seo, ©SHUTTERSTOCK.COM/JITTAWIT21
Brian J. Pierre, Abraham Ellis, Jeremiah Miller, and Guohui Yuan
Pathways to the
Next-Generation
Power System With
Inverter-Based
Resources Challenges and recommendations.
Authorized licensed use limited to: University of Texas at Austin. Downloaded on April 04,2023 at 22:19:51 UTC from IEEE Xplore. Restrictions apply.
ANAGING THE STABILITY OF TODAY’S Synchronous generators regulate their terminal volt-
M
electric power systems is based on decades ages and respond to changes in grid frequency through
of experience with the physical properties changes in their power output. We refer to these genera-
and control responses of large synchronous tion sources as grid forming. Today’s inverter-based genera-
generators. Today’s electric power systems tion sources generally use phase-locked loops (PLLs), which
are rapidly transitioning toward having an increasing pro- rely on externally generated voltages from synchronous
portion of generation from nontraditional sources, such as machines to operate. We refer to these types of invert-
wind and solar (among others), as well as energy storage er-based generation sources as grid-following inverters. In
devices, such as batteries. In addition to the variable nature case of unintended separation of the power system or after
of many renewable generation sources (because of the a blackout, islanded systems comprising only these types of
weather-driven nature of their fuel supply), these newer inverters cannot operate autonomously. This limitation of
sources vary in size—from residential-scale rooftop sys- the grid-following inverters has inspired an investigation
tems to utility-scale power plants—and they are intercon- into grid-forming control methods for power electronic
nected throughout the electric grid, both from within the inverters, which provide functionalities that are traditionally
distribution system and directly to the high-voltage trans- provided by synchronous machinery. Early work on this
mission system. Most important for our purposes, many of topic started in the 1990s, focusing on power systems with
these new resources are connected to the power system small footprints (e.g., microgrids) and on small islands (such
through power electronic inverters. Collectively, we refer to as Kauai, Hawaii). Today, grid-forming controls are being
these sources as inverter-based resources. considered for deployment in bulk power systems because
The operation of future power systems must be based of their ability to enhance the stability of these grids when
on a combination of the physical properties and control loads are largely being served by inverter-based resources.
responses of traditional, large synchronous turbine This article reviews the challenges involved in integrat-
generators as well as those of inverter-based resources ing inverter-based resources into the electric power system
(see Figure 1). The major challenges stem from the recog- and offers recommendations on technology pathways to
nition that there is no established body of experience inform the academic community, industry, and research
for operating hybrid power systems with significant organizations. We will 1) discuss the difference between
amounts of inverter-based resources at the scale of grid-following and grid-forming control approaches for
today’s large interconnections. inverter-based resources; 2) review relevant research and
To operate such large hybrid power systems, the outline research needs related to five grid-forming inverter
assumptions that underlie current generation design and topics: frequency control, voltage control, system protec-
control approaches must be reexamined and, where tion, fault ride through (FRT) and voltage recovery, and
appropriate, modified or even redefined to take explicit modeling and simulation; and 3) introduce a road map
account of the new challenges and opportunities present- that outlines an evolutionary vision in which grid-forming
ed by these inverter-based forms of generation. We inverters play a growing role in power systems that, in
should expect that new control approaches, operational turn, leads to the identification of nearest-term priorities
procedures, protection, and planning tools and processes for research. This article builds upon the Research Road-
will be required. map on Grid-Forming Inverters (Lin et al. 2020). Interested
Present Future
Generator Generator
Inverter Inverter
(a) (b)
Figure 1. (a) The present power system has historically been dominated by synchronous generators having a large rotational inertia with a rela-
tively modest amount of inverter-based resources, such as photovoltaics, wind, and batteries. (b) Future systems will have a significant fraction
of generation interfaced with power electronics and might be dominated by inverters. This implies a need for next-generation grid-forming control-
lers that ensure grid stability at any level of penetration with inverter-based resources.
θ
Pref Reference PLL
Calculation
iref
dc-Side Current
Control Control
i v
PWM
PQ
Current
Source V ω
Q P
(a)
Droop
Voltage
Source V ω
Q P
(b)
Figure 2. Functional diagrams of (a) grid-following and (b) grid-forming inverters. Grid-following inverters mimic current sources at their output
terminals, whereas grid-forming inverters act like voltage sources whose output abides by droop laws. PWM: pulsewidth modulation.
Authorized licensed use limited to: University of Texas at Austin. Downloaded on April 04,2023 at 22:19:51 UTC from IEEE Xplore. Restrictions apply.
Grid-Forming Controllers physically originates from the rotating masses of ma-
From here forward, the term “grid forming” acts as an chine-based generators. Since inverter-based resources
umbrella for any inverter controller that 1) regulates ter- do not contribute inertia to a power system, it follows
minal voltages, 2) can coexist with other grid-following that the replacement of machines with inverters will
and grid-forming inverters and synchronous generation reduce the system inertia and may increase the risk of
on the same system, and 3) does not require a PLL or large frequency swings. Figure 3 illustrates the relation-
communications to operate together with multiple grid- ship between decreased machine capacity and in-
forming assets. As shown in Figure 2, grid-forming con- creased frequency deviations across time. To address
trollers can be broadly categorized as droop, virtual this concern, grid-forming inverters may be used to
synchronous machines, and virtual oscillator controllers. counteract both the loss of inertia and primary frequency
Droop control is the most well-established grid-forming control provided by retired synchronous generation.
method; it was conceived in the early 1990s. Its key fea- Similar to the natural behavior of synchronous ma-
ture is that it exhibits linear tradeoffs, often called droop chines, grid-forming inverter-based resources would au-
laws, between real power versus frequency and reactive tonomously react to frequency swings and adjust their
power versus voltage. This mirrors how synchronous power injections during a low-frequency event.
machines operate in steady state. They give rise to the fol- Reduced inertia may result in a larger rate of change
lowing properties regardless of whether they are of frequency and increasingly volatile system dynam-
machines or inverters: ics, and it also necessitates faster control actions to arrest
xx system-wide synchronization: all units reach the frequency swings. Because the magnitude of the frequen-
same frequency cy swing after a disturbance is largely tied to the imbal-
xx power sharing: each unit provides power in proportion ance between generation and load, enough untapped
to its capacity. capacity must be reserved as headroom for frequency
Virtual synchronous machine control replicates the control. A drawback is that unused capacity could repre-
dynamic behavior of a synchronous machine with sent an opportunity cost for both renewable and fossil-
an inverter. The complexity of the emulated virtual fueled generation because power output must be
machine can vary greatly, from detailed models to sim- throttled to less than the available amount.
plified swing dynamics. Implementations that close Referring to the controllers shown in Figure 2, we pro-
ly match machine characteristics have both Q-V and vide a brief survey of the existing frequency control strat-
P-omega characteristics and are often called synchron- egies. The P-omega droop offered by grid-forming units
verters. Virtual inertia methods are simpler and capture would govern the steady-state frequency deviation after
only the dynamics of an emulated rotor and its steady- the initial transients have subsided. Typically, these
state P-omega droop.
Virtual oscillator control is another inverter control
TABLE 1. A comparison of grid-following
method that emulates nonlinear oscillators. As illustrat-
and grid-forming controls.
ed in Figure 2, the model takes the form of an oscillator
circuit with a natural frequency tuned to the nominal Grid-Following Control Grid-Forming Control
ac grid frequency and its remaining parameters tuned
Assumes grid already Assumes converters must
to adjust the nominal voltage and control bandwidth. formed under normal actively form and regulate grid
Despite its unconventional appearance, it exhibits the operations voltages
Q-V and P-omega droop laws in steady state that the
Control of current injected Control of voltage magnitude
other grid-forming methods also offer. However, its sim- into the grid and frequency/phase
ple time-domain implementation and dynamical prop-
Decoupled control Slight coupling between
erties offer enhanced speed.
of P and Q P and Q
Inverter Control State of the Art Needs PLL It may use PLL control to
and Open Research Questions switch between modes
In this section, we review relevant research and outline Needs voltage at the point Can black-start a power system
research needs related to the following five topics: fre- of common coupling to
quency control, voltage control, system protection, FRT deliver P and Q
and voltage recovery, and modeling and simulation. Cannot operate at 100% Can theoretically operate at
power electronics penetra- 100% power electronics pen-
Frequency Control tion; instability thresholds etration; can coexist with grid
Frequency control refers to generation control actions (tipping points) exist following
designed to maintain system frequency near the nomi- Not standardized, inadequate
nal value. In machine-based grids, the system inertia operational experience at a
systems perspective
strongly influences the frequency dynamics and
4,000
3,500
3,000
Annual Mean Primary
MW/0.1 Hz
1,000
500
1994
1995
1996
1997
1998
1999*
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
Figure 3. Decreasing total system inertia for the Eastern Interconnection. (Source: NERC, 2017; used with permission.)
Authorized licensed use limited to: University of Texas at Austin. Downloaded on April 04,2023 at 22:19:51 UTC from IEEE Xplore. Restrictions apply.
drop or rise, takes place under various tune volt-VAR control actions are
load conditions. All conductors in
This must be known to introduce undesired interac-
power systems have an intrinsic balanced by the tions between grid-following inverters
impedance that results in a variable and voltage-regulation equipment.
voltage profile along the length of the need for system Methods to mitigate interactions
line when current flows. Active voltage
intervention (increasing or reducing
resilience because between all types of inverter controls
and other generation should be inves-
voltage to preferred operational limits) islanded operation tigated for inverter-dominated grids.
might use electromechanical or elec-
tronic components, from generators to is a key benefit of System Protection
other devices. Such devices include
load tap changers, voltage regulators,
grid-forming inverters The effect of grid-forming inverters on
power system protection is funda-
capacitor banks, synchronous con- as a response mentally different than that of grid-
densers, and others, along with early- following inverters and has not been
stage commercialization of solid-state to widespread extensively studied. In theory, the
technology (power electronics), e.g., a
static synchronous compensator.
catastrophic events. fault current from grid-forming invert-
ers, though it may vary by the control
For synchronous generators, their schemes, may have a subtransient
automatic voltage regulator adjusts behavior that more closely mimics
the output voltage either by adjusting power delivery via synchronous machines and is significantly larger than
the main field or real and reactive power output by modu- that supplied by grid-following inverters. The short circuit
lating the exciter field current. Voltage stability and reac- currents from grid-forming inverters can be equivalent to
tive power sharing among parallel-connected synchronous synchronous generation but are normally constrained to
generators is achieved via Q-V droop control such that each 4–6 p.u. for short time periods (<10 cycles) before steady-
machine follows a linear relationship between reactive state limits (<2 p.u.) are imposed. A larger short circuit
power and voltage. subtransient response will be limited primarily by the
Grid-forming inverters natively provide voltage regula- short circuit response of componentry in the grid-forming
tion via their Q-V droop laws, often called volt-volt inverters, related to its internal impedance. The short-time
ampere reactive (volt-VAR) control, which closely match- response is limited by semiconductor ratings, whereas the
es the behavior of synchronous machines. Mirroring ter- steady-state response is limited by inverter hardware
minology from frequency control, this is generally called parameters, e.g., thermal management.
primary voltage control to emphasize that these control By design, traditional three-phase grid-tied inverter
actions are done locally, without communication. Thus, controls will not provide zero- or negative-sequence cur-
grid-forming inverters can be especially helpful in pro- rents, which can be used to identify unbalanced fault con-
viding voltage support in weak grids. ditions more easily; inverter controls are designed to
Recent advances in voltage control for inverter systems suppress negative-sequence currents. It is recommended
are mostly concentrated in microgrid systems with droop to program grid-forming inverters to source zero and
control. Virtual impedances have been used to improve negative currents, mimicking a fault behavior of synchro-
reactive power sharing and mitigate parameter sensitivity. nous machines, in an unbalanced fault condition. This
To further enhance reactive power sharing and reduce would yield an increase in the efficacy of traditional pro-
steady-state errors, communications-based secondary- tection mechanisms compared to the pure grid-following
level controllers have been proposed. However, novel control case and would significantly simplify the identifi-
methods should be devised for deployment in bulk power cation of unbalanced faults.
systems to reduce communication dependency for scal- A protection issue unique to grid-forming inverters is
ability and resilience. operation in islanded/microgrid mode when a portion of
Recent findings have also uncovered adverse interac- the power system is disconnected from the bulk grid. Tra-
tions between grid-forming inverters and synchronous ditional grid-following inverters automatically shut off in
machine excitation systems that regulate voltages, and an islanded condition, with the absence of an external
similar issues have been observed on grid-following con- voltage, but grid-forming inverters can continue to oper-
trol types. These interactions can destabilize hybrid sys- ate islanded from the area grid (in many cases, such
tems and appear to be common to both grid-following resilient microgrid operation is a primary benefit of grid-
and grid-forming inverter controls. forming inverters). To maximize the benefit, some form of
Interactions and voltage oscillations may occur in islanding protection still will be needed for grid-forming
systems of grid following with grid-support functions. inverters to safely operate in islanded mode while ensur-
Here, the piecewise linear volt-VAR relations on grid-fol- ing the safety of electrical personnel and other bystanders.
lowing inverters and the time delays and filters used to This must be balanced by the need for system resilience
Authorized licensed use limited to: University of Texas at Austin. Downloaded on April 04,2023 at 22:19:51 UTC from IEEE Xplore. Restrictions apply.
model and positive-sequence simulations are a desirable simplified and often are not able to accurately capture
tool for bulk power system studies. Thus, there is an the behavior of a system. Recently, the importance of
important need for appropriate inverter-based genera- improving the grid-following model to capture this
tion models for existing positive-sequence simula- behavior has been recognized, and it also applies to grid-
tion tools. Currently, such models are usually highly forming inverters.
1.25
1.2
Positive-Sequence Voltage (per Unit)
1.15
1.1
1
“No Trip Zone” ∗
0.9
0.85
0.75
0
2.5
0 0.1 0.5 1 2 3 4 5 30 300
0.15 Time (s)
0.033
(b)
Figure 4. Voltage no trip boundaries of the North American interconnections. (a) Eastern, Western, and ERCOT. (b) Quebec. ERCOT: Electric Reli-
ability Council of Texas. (Source: NERC, 2020; used with permission.)
Grid-Forming
Grid-Forming Standardize Wide
Inverter Key Tools for
R&D Validation and Grid-Forming Grid-Forming
Research Practitioners
Demonstration Deployments Adoption
Questions
Authorized licensed use limited to: University of Texas at Austin. Downloaded on April 04,2023 at 22:19:51 UTC from IEEE Xplore. Restrictions apply.
microgrids and progressing to island or other smaller systems. Therefore, there is a staging of implementation
weak grids, and eventually to entire interconnections. This between different usage levels as well as stages of usage
multiyear perspective recognizes that the scale and scope within application areas (denoted by a color gradient
of the types of power systems for which inverters will be within an application area).
called on to provide grid-forming services will begin mod- We are currently seeing (and will continue to see for
estly. Specifically, it recognizes that the dominant form of the foreseeable future) the incorporation of grid-forming
inverter control today is grid following and that future inverters in island microgrid environments. These
power systems will involve a mix of inverter-based sourc- microgrids, which are already being incorporated in a vari-
es with both grid-following and grid-forming control. ety of areas (for example, rural villages in Alaska, univer-
Growth over time will be paced or enabled by how well sity campuses, and military bases) run hybrid diesel-
grid-forming inverters perform and what advantages they renewable grids with grid-forming inverters on energy
bring. This recognition, in turn, establishes a natural storage. As the technology of grid-forming inverters
sequence of priorities for the research questions that matures, we will begin to see the emergence of 100% grid-
must be addressed. forming islanded microgrids with scalable multi-inverter,
multiple grid-forming-based architectures, and energy
From Microgrids to Isolated Power Systems sources. Such microgrids, although small, can still provide
to Continental-Scale Power Systems a wealth of practical knowledge in the deployment of grid-
Grid-forming implementation will occur through phased forming inverters.
implementations of grid-forming inverters, starting with As the technology for grid-forming-based microgrids
smaller, more constrained microgrids and eventually matures, grid-forming-based implementations will begin
moving toward larger grids (Figure 7). Even within to appear in larger island grid settings (3–15 years), such as
application areas, phased implementation is likely to in Hawaii and the Caribbean. These grids have more inter-
occur, with initial grid-forming implementation being sea- operating sources and loads, are geographically larger, and
sonal or taking place during situational periods when exhibit a much larger and more complex behavior com-
additional firm sources are needed (e.g., instantaneous pared to site-level microgrids. Additionally, although
inverter-based generation periods or the provision of volt- site-level islanded microgrids are primarily built at distribu-
age regulation under local, specific contingencies) before tion-level voltages, many island grids have subtransmis-
the widespread adoption of grid-forming-dominated sion-level voltages.
Grid
Forming in
Bulk Grids
10–30 Years
Instantaneous
Grid Forming Challenges in
Weak Portions of Bulk Grids.
Western Electricity Coordinating
Council, ERCOT (and so on)
3+ Years
Figure 7. Incorporating grid-forming controls into the bulk electric grid will take place gradually after key functionalities have been demonstrated
and confidence has been gained by operating them in smaller microgrids and island power systems.
distribution-level assets emphasize current harmonics tems before carrying out maintenance?
limits, reactive power limits, and anti-islanding functions. ■■ Should islanded subsystems within a larger grid
By contrast, grid-forming units are predominantly used remain energized to enhance reliability and facili-
for voltage regulation instead of current regulation, reac- tate a system black start?
tive power can vary for voltage support, and grid-forming
inverters natively provide uninterrupted power during The Need to Begin Developing Appropriate
islanded conditions. Models for Existing Simulation Tools as Well as
The following aspects should be considered in the con- Enhanced Modeling and Simulation Tools
text of ongoing efforts to modernize standards for grid- Modern state-of-the-art grid analysis tools have been
connected inverters within IEEE, the International tailored toward grids dominated by synchronous
Electrotechnical Commission, Federal Energy Regulatory machines. A widely adopted assumption within these
Commission, and NERC: tools is that the synchronous speed of power systems
xx Current versus voltage waveform quality: Under the remains relatively close to nominal during and after a
existing regulatory paradigm, inverters are con- transient; however, this widely adopted assumption
trolled to inject sinusoidal currents with minimal might not be valid in systems with high penetrations of
harmonic content. This aligns with prevailing prac- inverter-based generation because they lack synchro-
tices in which harmonics demanded by nonlinear nously connected rotor inertia. Hence, researching
loads are provided by machines and reactive com- appropriate modeling and simulation tools that are suit-
ponents, whereas inverters inject only 60-Hz com- able for studying the transition from machine-based
ponents. If machines are displaced by grid-forming systems to inverter-based ones is a high priority. In fact,
inverters, however, then the inverters will need to some independent system operators and developers
Authorized licensed use limited to: University of Texas at Austin. Downloaded on April 04,2023 at 22:19:51 UTC from IEEE Xplore. Restrictions apply.
require this capability immediately, highlighting the Mag., vol. 17, no. 6, pp. 89–98, 2019, doi: 10.1109/MPE.2019.
importance and urgency of appropriate modeling and 2933072.
“Reliability guideline: BPS-connected inverter-based
simulation tools.
resource performance,” North American Electric Reliability
Another important challenge is to simulate inverter Corporation, Atlanta, GA, USA, 2018. [Online]. Available:
controllers as implemented in the field to accurately https://www.nerc.com/comm/PC_Reliability_Guidelines_DL/
predict undesirable performance. This problem was Inverter-Based_Resource_Performance_Guideline.pdf
highlighted during the recent Blue Cut Fire, Canyon Fire A. Yazdani and R. Iravani, Voltage-Sourced Converters in
Power Systems: Modeling Control and Applications. Hoboken, NJ,
2, and subsequent events, where disconnections of
USA: Wiley, 2010.
inverters were not predicted by current simulation. This Q. C. Zhong, “Virtual synchronous machines: A unified
problem can extend to the future incorporation of interface for grid integration,” IEEE Power Electron. Mag.,
inverters with grid-forming controls; hence, appropriate vol. 3, no. 4, pp. 18–27, 2016, doi: 10.1109/MPEL.2016.
models for existing transient simulation tools should 2614906.
“State of reliability,” North American Electric Reliability
be implemented.
Corporation, Atlanta, GA, USA, 2017.
Generator Frequency and Voltage Protective Relay Settings,
Conclusions NERC Standard PRC-024-3, Atlanta, Georgia, USA, 2020.
For the next decade and beyond, the large interconnec-
tions will comprise both electromechanical and inverter- Acknowledgments
based resources. Inverter-based, grid-forming resources This work was authored in part by the National Renew-
will be necessary for the stable operation of the bulk power able Energy Laboratory, operated by the Alliance for Sus-
grid. This article envisioned the key short- and long-term tainable Energy, LLC, for the U.S. Department of Energy
R&D needs for inverter-based resource grid-forming con- (DOE) under contract DE-AC36-08GO28308. Funding is
trols, protection, and modeling as part of hybrid grids. We provided by the U.S. DOE Office of Energy Efficiency and
have also provided a comprehensive analysis of the chal- Renewable Energy Solar Energy Technologies Office. The
lenges in integrating inverter-based resources and offered views expressed herein do not necessarily represent the
recommendations on potential technology pathways views of the DOE or the U.S. government.
to inform the academic community, industry, and re-
search organizations. Biographies
Additional short-term and long-term roadmapping Yashen Lin (yashen.lin@nrel.gov) is with the National
and detailed system performance metrics will need to Renewable Energy Laboratory, Golden, Colorado,
be developed to support this transition to the next- 80401, USA.
generation power system with inverter-based resourc- Joseph H. Eto (jheto@lbl.gov) is with the Lawrence
es. Future assessment will also need to address the Berkeley National Laboratory, Berkeley, California, USA.
many areas beyond the scope of this article that more Brian B. Johnson (brianbj@uw.edu) is with the Universi-
broadly correlate to long-term grid modernization ty of Washington, Seattle, Washington, 98195, USA.
efforts, for instance, distribution system operations Jack D. Flicker (jdflick@sandia.gov) is with the Sandia
with grid-forming inverter controls, the evolution of National Laboratories, Albuquerque, New Mexico,
sensing and communications systems, economic anal- 87185, USA.
ysis, and cybersecurity. Robert H. Lasseter (lasseter@engr.wisc.edu) is with the
University of Wisconsin, Madison, Wisconsin, 53706, USA.
For Further Reading Hugo N. Villegas Pico (hvillega@iastate.edu) is with the
M. C. Chandorkar, D. M. Divan, and R. Adapa, “Control of par- Iowa State University, Ames, Iowa, 50011, USA.
allel connected inverters in standalone ac supply systems,” Gab-Su Seo (gabsu.seo@nrel.gov) is with the National
IEEE Trans. Ind. Appl., vol. 29, no. 1, pp. 136–143, Jan./Feb. 1993,
doi: 10.1109/28.195899.. Renewable Energy Laboratory, Golden, Colorado, 80401,
“IEEE Standard for Interconnection and Interoperability of Dis- USA.
tributed Energy Resources with Associated Electric Power Sys- Brian J. Pierre (bjpierr@sandia.gov) is with the Sandia
tems Interfaces,” IEEE Standard 1547-2018. National Laboratories, Albuquerque, New Mexico, 87185, USA.
B. B. Johnson, M. Sinha, N. G. Ainsworth, F. Dörfler, and S. V. Abraham Ellis (aellis@sandia.gov) is with the Sandia
Dhople, “Synthesizing virtual oscillators to control islanded
inverters,” IEEE Trans. Power Electron., vol. 31, no. 8, pp. 6002– National Laboratories, Albuquerque, New Mexico, 87185,
6015, 2016, doi: 10.1109/TPEL.2015.2497217. USA.
R. H. Lasseter, “Control of distributed resources,” in Proc. Jeremiah Miller (jeremiah.miller@ee.doe.gov) is with
1998 Int. Conf. Bulk Power Syst. Dyn. Control IV-Restructuring, the U.S. Department of Energy Solar Technologies Office,
pp. 323–330. Washington, D.C., 20585, USA.
Y. Lin et al., “Research roadmap on grid-forming inverters,”
National Renewable Energy Laboratory., Golden, CO, USA, No. Guohui Yuan (guohui.yuan@ee.doe.gov) is with the U.S.
NREL/TP-5D00-73476, 2020. Department of Energy Solar Technologies Office, Washing-
J. Matevosyan et al., “Grid-forming inverters: Are they the ton, D.C., 20585, USA.
key for high renewable penetration?” IEEE Power Energy