A Systematic Review of Grid-Forming Control Techniques for Modern Power Systems and Microgrids

Abstract

Looking toward the future, governments around the world have started to change their energy mix due to climate change. The new energy mix will consist mainly of Inverter-Based Resources (IBRs), such as wind and solar power. This transition from a synchronous to a non-synchronous grid introduces new challenges in stability, resilience, and synchronization, necessitating advanced control strategies. Among these, Grid-Forming (GFM) control techniques have emerged as an effective solution for ensuring stable operations in microgrids and large-scale power systems with high IBRs integration. This paper presents a systematic review of GFM control techniques, focusing on their principles and applications. Using the PRISMA 2020 methodology, 75 studies published between 2015 and 2025 were synthesized to evaluate the characteristics of GFM control strategies. The review organizes GFM strategies, evaluates their performance under varying operational scenarios, and emphasizes persistent challenges like grid stability, inertia emulation, and fault ride-through capabilities. Furthermore, this study examines real-world implementations of GFM technology in modern power grids. Notable projects include the UK’s National Grid Pathfinder Program, which integrates GFM inverters to enhance stability, and Australia’s Hornsdale Power Reserve, where battery energy storage with GFM capabilities supports grid frequency regulation.

1. Introduction

The worldwide concern about climate change on our planet is an indisputable fact. One of the factors contributing to the production of greenhouse gases is the energy sector. Due to the exponential growth of the population in recent years, more electricity is needed to meet demand. However, electricity production should come from non-polluting sources, such as solar and wind energy. Not long ago, the traditional power system was characterized primarily by the dominance of synchronous machines. However, with the deployment of Renewable Energy (RE) and its massive integration into the power grid, the traditional model is becoming outdated. Governments around the world are shifting their energy mix to achieve neutrality in the future. According to the report titled “Going climate-neutral by 2050” [1], the European government aims for zero emissions by 2050. Similarly, the United States government rejoined the Paris Agreement and committed to reducing greenhouse gas emissions by (50–52%) [2]. Likewise, the Ecuadorian government has introduced economic incentives to reduce emissions. Achieving these goals requires strong commitment and significant investment in renewable energy resources to drive decarbonization [3]. The integration of IBRs, from small to large-scale applications, is expanding rapidly in both continental and isolated power systems. In particular, in isolated power systems with heritage significance, reliance on fossil fuels for electricity generation is increasingly viewed unfavorably due to environmental fragility, and the introduction of high levels of IBRs has begun. With high IBRs penetration, the power system has changed from the classical model, now being dominated by non-synchronous machines. This paradigm shift has introduced new challenges in stability, resilience, and synchronization, making GFM control techniques essential for ensuring secure and reliable microgrids as well as large-scale power networks. Early approaches, such as droop control and Virtual Synchronous Machine (VSM) techniques, have been widely studied for their ability to mimic synchronous machine behavior. More recently, the application of Fractional-Order PI (FOPI) controllers has improved transient performance and fault recovery [4]. Ref. [5] provided one of the first full-order small-signal stability analyses of hybrid systems with both GFM and Grid-Following (GFL) inverters, demonstrating the stabilizing effect of even a single GFM. The authors in [6] proposed a hybrid GFM architecture that incorporates the Virtual Power Method into Voltage-Controlled GFM (VCGFM). This approach enables seamless non-switching transitions and significantly enhances system stability under large disturbances while maintaining robustness in small-signal behavior. Similarly, ref. [7] highlighted the need for standardized evaluation criteria and black-box testing procedures to ensure consistent performance and facilitate the transition from laboratory-based assessments to field-level deployment. At the system level, ref. [8] provided important guidance on how the proportion of GFM to GFL inverters influences overall stability margins, offering quantitative design recommendations based on grid strength indicators. Despite recent progress, GFM research continues to be fragmented, lacking broad agreement on key aspects such as scalability, interoperability, and fault tolerance of different control strategies. To address these challenges, this study employs the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) methodology to conduct a thorough review of recent GFM control strategies. It highlights key control techniques and various stability classifications and shares practical insights from real-world projects implemented in the latest years.

The remainder of this paper is organized as follows: Section 2 presents the literature review and methodology used. Section 3 presents the results and discussions of the GFM control techniques, and finally, the conclusions and recommendations are presented in Section 4.

2. Literature Review Methodology

2.1. Literature Search Strategy

To conduct this systematic review, an extensive and exhaustive search was performed across four well-established academic databases: Scopus, Web of Science, Lens, and IEEE Xplore. These databases were chosen for their high-quality research articles. Scopus is known because journals are reviewed for high quality according to four numerical measures: h-Index, CiteScore, SCImago Journal Rank (SJR), and Source Normalized Impact per Paper (SNIP). Lens, an open-access platform, offers tools to analyze search outputs by institutional affiliation and other metadata fields. Web of Science is a high-quality database and is one of the most preferred by researchers. IEEE Xplore, a powerful resource for electrical and computer engineering and related fields, provides access to the most cited publications, including journal articles, conference proceedings, and technical standards. The collective data retrieved from these databases provide a comprehensive and well-rounded understanding of the research landscape relevant to this review. The literature search was guided by the use of specific keywords applied uniformly across all four databases: “grid forming control”, “microgrid OR modern power system” and “stability”.

Table 1. Query strings for the literature search process.

Database	Query String	Document Count
Scopus	TITLE-ABS-KEY(“grid forming control”) AND (TITLE-ABS-KEY(microgrid) OR TITLE-ABS-KEY(“modern power system”)) AND TITLE-ABS-KEY(stability)	73
Web of Science	TS = (“grid forming control”) AND TS = (microgrid OR “modern power system”) AND TS = (stability)	26
IEEXPLORE	“grid forming control” AND (“microgrid” OR “modern power system”) AND stability	88
Lens	“grid forming control” AND (microgrid OR “modern power system”) AND stability	55

provides a comprehensive overview of the keyword combinations and the corresponding search parameters used.

Across the four selected databases, our initial search returned 242 documents relevant to the scope of this study. Given the high number of documents, we applied specific inclusion and exclusion criteria to filter studies that aligned the most with our research objectives. The inclusion and exclusion criteria are shown in

Table 2. Inclusion and exclusion criteria applied for selection of studies.

Inclusion	Criteria	Exclusion
Studies published from 2015 to 2025	Year of Publication	Studies published before 2015
Studied published in English	Language	Studies published in languages other than English
Peer-reviewed journal articles	Document type	Editorials, book chapters, and review articles
Studies addressing the following topics: grid-forming control, microgrid or modern power system, stability	Emphasis	Studies that do not specify microgrids or modern power systems.
Studies can include practical and experimental case studies applying grid-forming control	Direction	Studies that focus on microgrids and modern power systems are excluded if they do not address specific grid-forming control techniques. Additionally, research centered on energy management, economic analysis of power systems, and optimization is also excluded from our study

. The results were restricted to the last 10 years, 2015–2025, to provide a more current perspective on the literature, thus capturing the latest methodological advances and insights into the current state of research. Given that English remains the dominant language in scientific literature, we further limited our selection to English-language publications. To ensure the reliability of our findings, we limited the selection to peer-reviewed journal publications that specifically examine grid-forming control strategies, microgrids, or modern power systems, and stability. We excluded studies published before 2015 and document types such as reports and book chapters to maintain a focused search scope. Finally, since stability is a critical aspect of Inverter-Based Resource (IBR) deployment, studies that did not explicitly address stability within the context of grid-forming control were not included.

2.2. Study Selection Proccess

After defining the inclusion and exclusion criteria, the next step involved selecting studies that aligned with our research objectives. To ensure methodological rigor, we followed the PRISMA 2020 framework, and the completed PRISMA Checklist is provided in the Supplementary Materials. This process comprises four sequential phases: identification, screening, eligibility, and synthesis. By adhering closely to the PRISMA methodology, we ensured a transparent and systematic selection process. Figure 1 depicts the different steps applied for this literature review.

In the identification phase, the researchers performed extensive searches in specified databases to collect items without discrimination based on titles, authors, journals, page numbers, and citation numbers. To improve accuracy and reduce redundancy, duplicate entries were systematically detected and removed using bibliographic management tools such as Mendeley Reference Manager (version 2.130.2) and Zotero (version 7.0.15).

During the screening phase, the remaining records were assessed based on their titles, keywords, and abstracts. The inclusion and exclusion criteria outlined in Table 2 were systematically applied to ensure alignment with the research focus. As a result, only studies directly relevant to the review objectives progressed to the subsequent phase. In the eligibility phase, studies were assessed using a Likert scale to evaluate how closely they aligned with the objectives of this review. Only those that met a predefined relevance threshold were selected for inclusion. Finally, in the synthesis phase, we consolidated the findings from eligible studies, identifying key insights, emerging challenges, knowledge gaps, and notable real-world reference projects. These outcomes are discussed in detail in Section 3.

2.2.1. Identification Stage

At this phase, a total of 242 items were identified using the search strategy applied across four databases: 73 from Scopus, 26 from Web of Science, 88 from IEEE Xplore, and 55 from LENS. The higher number of items from IEEE Xplore is attributed to its specialization in technical fields such as electrical and electronic engineering. During this stage, 77 duplicate records were removed using bibliographic management tools, reducing the pool to 165 studies.

Figure 2 plays an essential role in visualizing the research landscape over the past decade. Figure 2a displays the year-by-year increase in academic output, while Figure 2b uses a heatmap to reveal the temporal density of publications by source. Finally, Figure 2c provides a comprehensive distribution of the selected studies. A clear rise in publication volume is evident: only 40 studies were published between 2015 and 2020, whereas 202 studies emerged between 2020 and 2025, reflecting a significant increase in research activity in recent years. This trend also highlights that grid-forming control is considered a hot topic in academia. The number of studies has generally increased over time. In 2021, the number of publications remained relatively constant compared to 2020, which could be attributed to the impact of the COVID-19 pandemic. However, from 2022 onward, the number of studies began to increase significantly as research activities resumed after the pandemic, reaching an increase of 77 studies in 2024. To maintain clarity and facilitate traceability throughout the review process, each study was assigned a distinct identifier linked to its originating database: Scopus-XXX (articles from Scopus), IEEE-XXX (articles from IEEE Xplore), WoS (articles from Web of Science), LENS-XXX (articles from Lens).

2.2.2. Screening Stage

The main objective of this stage was to apply the inclusion and exclusion criteria detailed in Table 2 and to assess whether the abstracts met these criteria. Two researchers independently screened each study. Studies that met the criteria were marked as “yes”; those that did not were marked as “no”. This process ensured a rigorous and unbiased selection of relevant studies. Following the application of the search filters and criteria outlined in Section 2.1, the screening process illustrated in Figure 3 resulted in the exclusion of 63 records and the inclusion of 102 studies. Of these, 52 were sourced from Scopus, 7 from Web of Science, 31 from IEEE Xplore, and 12 from LENS, with 102 total studies.

2.2.3. Eligibility and Inclusion Stage

At this stage, a rigorous evaluation process was conducted to ensure that only relevant studies were selected. A Likert scale evaluation was utilized to enhance the selection criteria. This scale, ranging from 1 to 3, was applied to each of the six defined criteria, enabling a systematic assessment of the studies’ quality and relevance. The six evaluation criteria used in this phase are as follows:

• Relevance to Study Goals
  The extent to which the study aligns with the objectives of this review.
• Technical Depth and Innovation
  The degree of innovation and thoroughness in the suggested grid-forming control methods.
• Methodological Rigor
  The robustness and soundness of the study’s methodology.
• Empirical Validation and Robustness
  The extent of experimental validation and the reliability of the results
• Grid-Forming Control Classification and Scope
  The classification of the proposed grid-forming control strategies and their applicability to modern power systems or microgrids.
• Practical Applicability and Scalability

To reduce selection bias and ensure methodological rigor, all studies were independently evaluated by two reviewers using a predefined six-criterion scoring rubric. Discrepancies were resolved through discussion and consensus. Only studies achieving a total score of 15 or higher (out of 18) were included in the final synthesis. The distribution of study quality scores is presented in Figure 4.

2.2.4. Synthesis of Selected Studies

This section presents the structured synthesis of the studies selected through the systematic review process. The analysis emphasizes the evolution of research trends and the deployment of various GFM control strategies in modern power systems or microgrids. Figure 5 illustrates the distributions of studies included during the eligibility phase. As shown in Figure 6a, the final set comprised 42 studies from Scopus, 5 from Web of Science, 21 from IEEE, and 7 from LENS.

Finally, after carefully following each stage of the review process, the PRISMA 2020 flow chart is presented in Figure 7. This diagram provides a structured visual summary of the systematic review methodology, outlining the phases of identification, screening, eligibility, and inclusion. It highlights the number of records retrieved from each database, the exclusions made based on the title and abstract reviews, and the final set of studies included for qualitative synthesis.

3. Results and Discussion

The literature review comprises a vast number of articles on different grid-forming control techniques, each designed to improve stability, reliability, and resilience in modern power systems and microgrids. These studies classify GFM control methods into several key categories, including GFM control techniques, including droop control, virtual synchronous generator, oscillator-based control, power synchronization control, synchronverter, matching control, and advanced control techniques. Many of these techniques have been discussed extensively in the literature by leading experts; however, only a subset has seen practical implementation in real-world projects. Droop control and Virtual Synchronous Generator (VSG) are two of the most widely used techniques in commercial projects. On the other hand, oscillator-based control, power synchronization control, synchronization converter, and matching control remain in the experimental phase through simulations, laboratory tests, or pilot projects. Following, we present the different GFM found in the literature.

In this section, various GFM control schemes, such as droop control, VSM, Power Synchronization Control (PSC), Dispatchable Virtual Oscillator Control (dVOC) are presented.

3.1. Droop Control

This kind of control was proposed two decades ago, making it the most widely used GFM technology for decentralized power-sharing in Microgrid (MG)s. The droop control concept originates from the governor action that enables the parallel operation of multi-synchronous generators. This control works by mimicking the behavior of a Synchronous Generator (SG) by adjusting the frequency and voltage based on active and reactive power variations. In [9], the authors implemented a droop-based power-sharing strategy carried out in an off-grid AC MG, ensuring balanced energy distribution among Electric Storage Systems (ESS). One of the key challenges when inverters connect to a pre-existing Point Common Coupling (PCC) is synchronization. Proper synchronization is essential to ensure the inverter operates in different operation modes. To address this, in [10], the authors propose an algorithm for the initial synchronization of a droop-controlled inverter. This approach was validated through simulations using Simulink/PLECS software, demonstrating its feasibility for safe and stable inverter integration into MGs. In some cases, the droop-controlled GFM may become overloaded under high-demand conditions. To avoid this problem, researchers in [11] present an overload mitigation control strategy that can autonomously transfer the additional load of the overloaded source to other sources without requiring communications links. In a complementary direction, in [12], researchers implemented a droop control-based dispatch strategy in a multi-source islanded microgrid, showing that proper droop tuning ensures proportional load sharing. The experimental results show that the dispatched GFM sources respond to the changed droop intercept to output the desired active power, and it is important to maintain the same frequency.

Recent work has also focused on enhancing droop control through supervisory and energy management strategies. For example, in [13], a multilevel droop control system based on virtual impedance was proposed for autonomous microgrids. The system was modeled in the MATLAB/SimPower System, where the authors, through simulations, demonstrated enhanced dynamic performance and load balancing. Similarly, ref. [14] applied droop control to develop a battery management strategy for Photovoltaic (PV) power plants operating in islanded MGs. Their approach enables smooth transitions between Maximum Power Point Tracking (MPPT) mode and Battery Saving (BS), ensuring stable voltage and frequency regulation.

A novel direction was presented in [15], where droop behavior was derived from the Andronov–Hopf theory. The method was tested and simulated, demonstrating that there is no necessary low-pass filter in the control loop. Frequency events are becoming increasingly common in modern power systems due to the growing integration of IBR and the gradual replacement of traditional synchronous generators. Because of this, in [16], authors analyzed the available frequency support capability of GFM converters. The findings indicate that during significant disturbances, saturation reduces the frequency support capabilities of inverters, revealing a compromise between limiting current and preserving dynamic stability.

Experimental validations play a crucial role in showcasing practical applicability. In the study carried out by [17], a droop-controlled inverter was tested in both the grid-connected and islanded modes. The results affirmed that GFM inverters can provide black start capabilities and voltage and frequency regulation. The authors in [18] conducted a co-simulation featuring more than 10,000 GFM and GFL inverters using droop control. The T&D co-simulation was carried out using the mini-WEEC system with 19 IEEE 8500-node distribution feeders and IBR. Open sources such as GridPACK and GridLABD and HELICS were used to carry out this deep study. The study emphasized that system frequency stability can still be achieved with 100% inverter-based generation, provided that at least 12% of the inverters operate in grid-forming mode with droop control. Finally, droop control has also been adapted for hydrogen production systems. In [19], the authors proposed a model of a GFM droop control approach to a reliable and constant power supply to the load.

Table 3. Summary of droop control-based GFM strategies.

Paper Title	Control Strategy	Stability Type	Advantages
[9]	Droop Control	Voltage and Frequency Stability	Facilitates power sharing without communication
[10]	Droop Control with Inner Voltage Loop	Transient and Synchronization Stability	Enhances transient synchronization via voltage loop dynamics
[11]	Droop Control with Current Limiting	Transient Overload Protection	Improves overload handling during grid faults
[12]	Multi-Master Droop Control	Power Sharing and Frequency Stability	Enables coordinated dispatch among GFM units
[13]	Multilevel Droop + Virtual Impedance	Voltage Stability and Power Quality	Improves reactive power sharing accuracy and voltage balance using virtual impedance
[14]	Droop Control with SoC-Based Priority	Energy Management, Frequency Support	Preserves battery health through SoC-aware control
[15]	Droop Characteristics with Oscillator	Small-Signal Stability	Fast synchronization and response without low-pass filters; improves dynamics
[16]	Droop with Current/Power Limits	Frequency Stability under Limits	Stabilizes frequency under power/current limits
[17]	Droop Control (PQ-based)	Voltage and Frequency in Islanded Mode	Achieves black start and voltage–frequency control in islanded mode
[18]	Droop + LADRC-MI	Frequency Regulation and Anti-Disturbance	Validated in hardware; demonstrates black start and voltage/frequency regulation

provides a comprehensive summary of the droop control studies reviewed, highlighting the control strategies adopted, associated stability objectives, and key advantages of each approach.

3.2. Virtual Synchronous Generator

Another kind of control is inspired by virtually emulating the dynamics and control of a Synchronous Machine (SM). This grid-forming technique is a more advanced control than droop control. The Virtual Synchronous Machine (VSM) incorporates the swing equation for the regulation of the active power generated by the SM. The rest of the control system, namely the measurement system, the reactive power controller, the active power synchronization, and the voltage and current controllers, remain unchanged compared to those used for the droop control. The concept of VSM has been applied in various studies to enhance grid stability and power sharing. In [20], the authors investigated this control to provide inertia support and damping, improving frequency stability, transient response, and power sharing in weak grids and isolated microgrids. The research uses Hardware-in-the-loop (HIL) environment for further validation and tests the VSM control in an island system operating with diesel generators. The results indicate significant improvements in frequency stability, particularly by reducing the Rate Of Change of Frequency (ROCOF). In [21], the authors established that substituting GFL inverters with GFM inverters significantly mitigates the severity of the ROCOF and leads to a higher nadir during under-frequency events.

In order to provide uninterrupted energy to loads in a microgrid, the authors in [22] presented two GFM control strategies to realize a smooth transition from the grid-connected operation mode to islanded mode. Based on the concept of VSM, they carried out simulations and practical applications using the proposed control and demonstrating the accuracy of the control. A practical demonstration was carried out in [23] using the Power Hardware-in-the-Loop (PHIL) platform, where the authors using the VSM GFM control demonstrated the controller’s capability to handle black start operations in hybrid Alternate Current (AC)/Direct Current (DC) grid. A methodology for developing a VSM controller as a Dynamically Linked Library (DLL) for Real-Time Digital Simulator (RTDS) is presented in [24] and integrated into a Controller Hardware-in-the-loop (CHIL). Using the Type-3 wind system within the MIGRATE model, the proposed control showed frequency and voltage support under wind variability or load changes.

In modern power systems, frequency control presents significant challenges. To address this, the authors of [25] integrated Angular Frequency Deviation Feedforward (AFDF) and Energy Recovery Control (ERC) techniques to mitigate frequency overshoot and improve overall energy efficiency.

In [26], the authors explored the behavior of short circuit and Fault Ride-Through (FRT) performance in both experimental and commercial GFM systems. Their findings demonstrated that reactive fault current requirements, as defined by current connection standards, can indeed be achieved with GFM VSM control. A VSM GFM strategy was proposed in [27], where the authors evaluated the impacts of the inverter on the adaptability of distance protection. The simulation results showed the effectiveness of this strategy in improving the adaptability of distance protection, regardless of low current scenarios or fault type.

Finally, in [28], frequency dynamics in a fully non-synchronous island grid were analyzed. The results showed that the VSM can simulate synthetic inertia, improving the frequency resilience.

3.3. Synchronverter

Besides droop and VSG control, the Synchronverter (SVR) is another well-known GFM concept. It was introduced with the idea of operating an inverter to mimic the dynamics of an SG. One of the key advantages of the SVR over the SG is that parameters such as inertia, friction coefficient, field inductance, and mutual inductance can be tuned. A detailed analysis of the small signal impedance of different variations in the syncronverter control was investigated in [29] and then compared against droop control and VSM based GFM inverters. The model and analysis were performed through MATLAB/Simulink coupled with the PLECS toolbox. A laboratory setup experiment was also conducted at the IAL MG test brench to validate the impedance model in a practical environment. The results showed that the power-related small-signal response of SVR was predominantly determined by the virtual inertia and damping characteristics, similar to other GFM controls, while the harmonic stability and grid interaction properties were strongly influenced by the implementation of inner voltage and current control loops.

3.4. Power Synchronization Control

Recent studies emphasize PSC as a widely adopted strategy for GFM converters, particularly in systems with high IBR. The authors of [30] propose a PSC-based current-limiting controller that preserves voltage-mode operation during faults without switching to grid-following mode. This approach enhances fault ride-through while ensuring safe current limitations. Ref. [31] validated PSC as part of a system-wide control framework for 100% renewable grids, showing that PSC-based converters can provide inertia emulation, frequency, and voltage support across large networks. Finally, ref. [32] proposes a universal synchronization method (U-FLL) that builds on PSC principles to enhance stability across multi-voltage and multi-inertia systems.

3.5. Virtual Oscillator Control

Although most conventional GFM approaches rely on droop control and VSM, researchers have proposed an alternative method to synchronize and control parallel single-phase inverters without communication. This approach is known as Virtual Oscillator Control (VOC) and emphasizes the fact that each inverter is digitally controlled to replicate the dynamics of a nonlinear oscillator.

In [33], an experimental validation of dVOC for inverter-dominated power systems is presented. dVOC is a control strategy that requires only local measurements to establish grid-forming behavior with droop characteristics. Using this method, the dVOC inverters achieve almost instantaneous dynamic synchronization and load sharing, making it a promising alternative for modern power systems. This kind of control provides better stability, faster synchronization, and transient response than VOC.

3.6. Advanced Control

Over the years, power systems have evolved significantly, and with the deployment of IBRs, the power system requires advanced GFM control strategies to maintain stability, flexibility, and resilience. Recent research introduces adaptive control, Artificial Intelligence (AI)-based, model predictive, and distributed architectures to enhance GFM inverter performance.

An adaptive GFM photovoltaic inverter control strategy is proposed in [34] for PV-based DC microgrids. A finite model predictive control (FCS-MPC) considers the characteristics of PV and eliminates the requirement for an external PI controller. Power sharing can be achieved with no necessary communication between Distributed Energy Resources (DERs). Simulations and HIL confirm the feasibility and dynamic performance of the proposed strategy to maintain stable bus voltage without overshoot, undershoot, or oscillation, even under dynamic conditions. A related decentralized GFM control is proposed in [35], integrating adaptive droop control for wind power to enhance frequency support. A model predictive voltage control method for energy routers in smart grids is discussed in [36]. The study highlights the bidirectional power flow management capability of GFM inverters in net-zero energy systems, optimizing voltage stability and reducing harmonic distortions. In [37], a unified nonlinear recursive control is presented to ensure transitions between GFL and GFM operation modes with constraint enforcement. The proposed GFM control was simulated and validated using GrinLink and OPAL-RT, where the results showed the efficiency of the proposed control in terms of preventing voltage, current, and power violations during mode switching. In [38], a double layer Model Predictive Control (MPC) is developed for Doubly Fed Induction Generator (DFIG)-based systems to enable frequency regulation and torque control for a microgrid.

Due to the rise of artificial intelligence, various authors have presented solutions for control using AI. For example, the researchers in [39] developed supervised deep learning AI-based controllers for GFM inverters. The AI controller is trained from the data collected from a simulated VSG. A comparison is also carried out between the aforementioned controller and the VSG on the same inverter. The result showed a faster and smoother response to changes for the proposed AI-based controller. Similarly, Ngamroo suggested a Recurrent Neural Network (RNN)-based controller for GFM PV microgrids in [40]. The RNN design integrates a convolutional network framework, providing better performance for voltage and frequency compared with other strategies.

Fuzzy logic was applied in [41] to mitigate the frequency deviation caused by load transients during island operation. The study revealed that incorporating adaptive droop control coefficients alongside fuzzy control significantly enhances both the frequency stability and dynamic response performance in MG.

In [42], a decentralized control approach based on adaptive droop was developed for wind energy systems. This method improves frequency support, enhances fault tolerance throughout the system, and requires minimal communication between components.

Table 4. Comparative summary of GFM control techniques based on the reviewed literature.

Control Approach	Grid Stability Performance	Inertia Emulation	Real-World Implementation	Advantages	Disadvantages
Droop	Supports voltage and frequency regulation under steady state and transients ([10,11,12,13,14,15,16,17,18,19])	Implicit via droop slope	Hornsdale, Dalrymple, Victoria Big Battery	Simple, scalable, widely used in practice	Sensitive to load variation; limited transient precision
VSG	Improves frequency response and ROCOF resilience ([20,21,22,23,24,25,26,27,28])	Explicit via swing equation	Wallgrove, Blackhillock BESS, Kapolei	Replicates SG behavior; improves transient behavior	Tuning complexity
SVR	Provides stable impedance and accurate frequency response [29]	Tunable virtual inertia	Tested only in laboratory settings; no real-world deployment to date	Flexible tuning; mimics SG behavior closely	High design complexity; field validation limited
PSC	Effective for synchronization in weak grids and fault ride-through ([30,31,32])	Partial via phase-locking dynamics	HVDC systems (e.g., Kriegers Flak)	Avoids PLL; good under weak conditions	May need additional voltage control strategies
VOC/dVOC	High-speed dynamic synchronization; decentralized [33]	Indirect via oscillator response	Pilot-stage deployments	Autonomous operation; ideal for decentralized systems	Requires grid-scale testing; performance under variability unproven
MPC/AI-Based Control	Adaptive to dynamic conditions; robust performance ([34,35,36,37,38,39,40,41,42])	Depends on controller architecture	Validated via HIL and simulation platforms	Handles nonlinearities; suited to DERs integration	Computationally intensive; limited field maturity

presents a comparative summary of the GFM control techniques compiled from the reviewed literature. This overview highlights the main characteristics, advantages, limitations, and real-world deployment status of each strategy.

3.7. Stability Analysis of GFMIs

Power system stability has been recognized in the literature as one of the main challenges to ensure reliable system operation. It is defined as the ability of an electrical power system to maintain or quickly restore a balanced state following a disturbance, ensuring that most variables remain within acceptable bounds so that the whole system continues to operate adequately. From the perspective of disturbance size, stability can be classified as small-signal stability and large-signal stability. Additionally, based on the duration of the disturbance, stability is categorized as short-term stability or long-term stability.

3.8. Small-Signal Stability

Generally, different types of converters belong to other areas of the power system and use different control methods, which may create system instability. To address this issue, the impedance model is used by [43]. Their study includes external and internal stability analyses supported by both simulation and experimental validation. In [44], authors analyze the impact of feedforward and decoupling loops on inverter dynamics and stability, comparing their findings with small-signal stability results from other voltage control designs. The study emphasizes enhanced oscillation frequency and robustness. Another study is presented in [45], where researchers analyzed the stability of multi-parallel inverters using the global admittance method. They reveal that the inner loop proportional coefficient and line inductance of the PQ-CI, as well as the outer loop proportional and integral coefficients of the VSG-CI, significantly affect system stability. Ref. [46] presents a sequence-based of individual converter systems toward a small-signal representation of complex power systems.

3.9. Large-Signal Stability

The transient stability in modern power systems is a critical aspect, particularly as IBRs replace conventional SG. Because of that, new stability challenges arise from IBR/IBR and IBR/grid interactions under transient conditions. Unlike SG based systems, where transient stability is largely governed by inertial response and synchronizing torque, IBR-dominated networks rely on control algorithms to maintain system stability during disturbances. In [47], for example, researchers analyze the performance of different IBR control schemes under transient line-ground short-circuit fault and load change. They used the four-bus power system model to assess transient stability: (i) GFL with a normal Phase Locked Loop (PLL) and −${\mathrm{i}}_{q,\max }$ current injection, (ii) GFL with a frozen PLL and −${\mathrm{i}}_{q,\max }$ current injection, and (iii) GFM with $Q_{\max }$ set point and compared in terms of IBR bus frequency stability and the PCC voltage stability. As a result, options (ii) and (iii) showed superior performance to option (i).

Further, the study evaluated four different GFM schemes under load step changes, comparing their grid frequency overshoot and settling time. The study found that VSM GFM incorporating linear $p-{\omega }^{*}$ y $q-v^{*}$ demonstrated superior transient performance compared to other control schemes. Another approach was carried out in [48], where the authors investigated the transient stability of power systems co-dominated by different types of GFM sources. The system analyzed was heterogeneous, consisting of a mix SG, VSM, and droop controller inverters. The study finds the transient synchronization behavior of such hybrid power systems can be described by a second-order motion equation, similar to the swing equation used for SG. Because of that, classical transient stability assessment methods can be extended to hybrid systems. In addition, the study highlights that droop-controlled inverters enhance system damping. Likewise, the droop frequency jump impacts the post-fault initial state, so a new factor should be addressed in transient stability assessment. Furthermore, hybrid power systems that integrate SGs and GFM inverters demonstrate improved transient stability compared to fully SG-based networks.

The activation of current-limiting protection affects the dynamic response of GFM converters, which is investigated in [49] using Hamiltonian theory. The study applies Hamiltonian mechanics to build a dynamic model of the GFM converter, a method often used in higher-order nonlinear control systems. The research highlights that proper tuning of current limits enhances transient fault response, minimizing excessive deviations in system frequency and voltage. The study further demonstrates that Hamiltonian energy function analysis provides a structured approach to evaluating system stability under current-limiting conditions. Through real-time simulations using the MT6040 platform, they validated the theoretical analysis.

Furthermore, the authors of [50] conducted an extensive study that analyzed the stability of the transient angle stability of paralleled GFM inverters, with particular attention to the effects of significant damping. Through a detailed transient model, they introduced the concepts of uniform damping energy and nonuniform damping energy. Based on that model, the authors demonstrated that large damping coefficients significantly expand the transient stability boundary by dissipating excess kinetic energy during disturbances and also that, by an adaptive damping strategy, they can optimize transient stability, extending the critical clearing time from 137 to 261 ms in simulation scenarios.

3.10. Experiences of GFMIs Projects

Multiple real-world deployments of GFM inverters have demonstrated their effectiveness in enhancing grid stability, resilience, and dynamic performance across diverse power systems and microgrids. This subsection reviews a selection of representative GFM projects implemented globally, highlighting their technical configurations, operational roles, and key lessons learned within varying regulatory frameworks and grid conditions.

Table 5. Complete GFM projects deployed or under construction.

Project Name	Location	Size (MW)	Technology	Year
Mountain View Solar	Hawaii, USA	7	BESS	2024
Project #1	Hawaii, USA	13	BESS	2018
Kauai PMRF	Hawaii, USA	14	BESS	2022
Bordesholm	Germany	15	BESS	2019
Provincetown BESS	Massachusetts, USA	25	BESS	2022
Dalrymple	Australia	30	BESS	2018
Waiawa Phase 2 Solar	Hawaii, USA	30	Solar + BESS	2025
Kupono Solar	Hawaii, USA	42	BESS	2024
Wallgrove	Australia	50	BESS	2022
New England BESS	Australia	50	BESS	2023
Broken Hill BESS	Australia	50	BESS	2023
South Fork Wind	New York, USA	75	GFM STATCOM	2024
Blackhillock, Phase II	Great Britain	100	BESS	2025
Terang BESS	Australia	100	BESS	2026
Hornsdale Power Reserve	Australia	150	BESS	2022
Riverina and Darlington Point	Australia	150	BESS	2023
Kapolei Energy Storage	Hawaii, USA	185	BESS	2023
Mackinac	Michigan, USA	200	HVDC BtB system	2014
Blackhillock, Phase I	Great Britain	200	BESS	2024
Blyth Battery	Australia	200	BESS	2025
Bungama BESS	Australia	200	BESS	2025
Western Downs Battery	Australia	200	BESS	2025
Victorian Big Battery	Australia	300	BESS	2024
Mortlake BESS	Australia	300	BESS	2026
Kilmarnock South	Great Britain	300	BESS	2026
TagEnergy BESS	Australia	300	BESS	2026
Hams Hall	Great Britain	350	BESS	2026
Wheatridge Renewable Energy Facility	Oregon, USA	380	Wind + Solar + BESS	2024
Eccles	Great Britain	400	BESS	2026
Kriegers Flak	Denmark/Germany	410	HVDC BtB system	2018
Maritime Link	Nova Scotia, Canada	500	HVDC bipolar system	2018
Liddell Battery	Australia	500	BESS	2025

presents a comprehensive summary of GFM projects that have been deployed or are currently under development worldwide.

3.10.1. Hitachi Energy

The world’s first large-scale back-to-back High Voltage Direct Current (HVDC) system is located on Michigan’s Upper Peninsula. Voltage Source Converter (VSC) technology was selected over classic HVDC because it provides the ability to operate under any disturbances, stability in weak grid conditions, power oscillation damping, and support islanded operation, and it has black-start capability. Hitachi Energy was selected by the American Transmission Company (ATC) to supply a 200 MW back-to-back HVDC light converter station. Before its implementation, several studies were carried out in PSS/E and PSCAD. In particular, the system stability was demonstrated by emulating the power-angle characteristics of an AC line during large disturbances [51].

Another relevant project, known as the Kriegers Flak Combined Grid Solution (KF CGS) was built by Hitachi in 2018. This project interconnected the eastern synchronous area of Denmark and Germany by extending the existing High Voltage Alternal Current (HVAC) offshore wind farm infrastructure in the Baltic Sea. The project used HVDC Back-to-Back (BtB) converter technology with voltage source converters.

Additionally, the Dalrymple project, with a capacity of 30 MW/8 MWh, was implemented in 2018 by Hitachi ABB Power Grids. This project is recognized as Australia’s first virtual synchronous generator. Within its first 18 months of operation, it successfully reduced outages in the area by 95%.

3.10.2. SMA Solar Technology

As a leading global and specialist system technology, SMA developed an energy solution to generate and store clean energy in Bordesholm, Germany. Utilizing a battery energy storage system with GFM inverters, the local utility, in collaboration with the Cologne University of Applied Sciences, demonstrated that the Battery Energy Storage System (BESS) is able to perform grid stabilization tasks. In their demonstration, they simulated a large power outage, disconnecting businesses, institutions, and households from the European utility grid for one hour. Simultaneously, they activated a battery backup grid. Therefore, all households and businesses were fed by this stand-alone grid with 100% of RE. Most of the 8000 households did not notice the transition.

Towards Net Zero for Great Britain (GB) is Europe’s largest transmission-grid-connected battery storage system, put into operation in 2024 in the locality of Blackhillock. The Blackhillock battery storage system consists of two phases. Phase I comprises 200 MW, while phase II comprises 300 MW. It is expected that the SMA GFM solution provides a stability service consisting of 116 MVA of short circuit level contribution and 370 MWs of inertia.

Mortlake is another interesting project that is being built in Australia. This project is a 300 MW/650 MWh BESS. With advanced GFM functionalities, it is expected that the will project play a large role in supporting the grid operation with 100% of RE. Bulk earthworks have been completed, and civil work has started.

3.10.3. Tesla

Hornsdale Power Reserve is a 100 MW battery project, owned by Neon, supplied by Tesla, built in 2017 near Jamestown, South Australia. This project uses a lithium-ion batteries energy storage system, with a discharge capacity of 100 MW and energy storage of 129 MWh. After a successful first phase, both in operational and economic terms, an expansion of 50 MW/64.5 MWh was completed in 2020. The expansion incorporated Tesla’s Virtual Machine Mode, which provides inertia support functionalities to the electric grid [52]. Another noteworthy project, the Wallgrove Grid Battery, is a 50 MW/75 MWh grid-scale lithium-ion battery. This project was constructed by Tesla and operates in Virtual Machine Mode (VMM), allowing that battery to mimic the “swing equation” of SM and provide synthetic inertia to stabilize the grid. This project gained attention because of valuable technical details, including the frequency at which a fast frequency response is required, its effectiveness in providing inertia during grid disturbances, and the amount of electricity it stores and distributes under various circumstances.

Another GFM project is the Kapolei Energy Storage, which has a power capacity of 185 MW/565 MWh.

The Victoria Big Battery (VBB) was developed by Neon, with a capacity of 300 MW/450 MWh, located next to AusNet Services’ existing Moorabool Terminal Station. In 2022, this project was successfully supported by the Australian Renewable Energy Agency (ARENA) under the Large-Scale Battery Storage (LSBS) Funding Round. Designed originally as a traditional GFL facility and due to the lessons learned from Hornsdale Power Reserve (HPR), it was decided that it would be a candidate to adapt the Tesla’s VMM to provide GFM capabilities [53].

Developed by Neon, the Western Downs Battery Energy Storage System (WDBESS) project began construction in 2023 with an initial capacity of 200 MW/400 MWh. WDBESS was designed with inverters in GFL initially. However, the project expanded its capacity to 255 MW/510 MWh and initiated the transition to GFM. This transition enhances the system strength and allows it to provide inertia services. One of the lessons learned from this project is related to quality studies, which revealed that the 14th harmonic exceeded its allocated limit, partly due to changes in the harmonic assessment methodology and emission allocations by Powerlink Queensland. Due to this, discussions with the Network Service Provider (NSP) were necessary to assess potential solutions, such as adjusting the inverter filtering parameters or installing a harmonic filter to ensure compliance with grid performance standards and maintain the power quality [54].

One of the most ambitious GFM battery projects is currently under construction in New South Wales. With a capacity of 500 MW/1000 MWh and a two-hour duration, this large-scale BESS aims to enhance grid stability by providing system strength, inertia services, and frequency regulation. With an investment of USD 750 million, AGL Energy Limited expects the project to commence operations by December 2025.

Table 6. Summary of lessons learned and challenges across GFM projects.

Project	Key Lesson Learned	Challenge or Gap Encountered
Mackinac HVDC	Local measurement-based controls enabled stability in weak-grid conditions.	Absence of high-speed communication links necessitated decentralized control; weak-grid dynamics became a dominant constraint.
Kriegers Flak	Validated multi-terminal offshore HVDC operation across national systems.	Integration delayed due to complex TSO coordination and incompatible national grid codes.
Dalrymple	BESS effectively provided fast frequency control and black start in islanded mode.	Regulatory uncertainty around classification and market participation of BESS in islanded mode.
Bordesholm	Demonstrated full grid-forming with 100% renewable island operation.	Lack of standardized inverter-based black-start protocols required customized control solutions.
Blackhillock Phase I	HVDC-Light stabilized a weak grid with high renewable penetration.	Interoperability between inverter and SG complicated the tuning process.
Mortlake BESS	Virtual impedance retriggering revealed control instability; addressed via simulation.	Proprietary OEM control logic limited transparency, hindering coordination and diagnostics.
Hornsdale Power Reserve	BESS outperformed traditional generators in Frequency Control Ancillary Services speed and accuracy.	Frequency regulation standards initially lagged the response capabilities of batteries.
Kapolei Energy Storage	Demonstrated fast response and black start services, replacing conventional thermal reserves.	Custom logic and alignment were needed for legacy thermal integration.
Victoria Big Battery	Retrofitting GFM required reworking legacy protection systems and control schemes.	Standards based on GFL control conflicted with GFM implementation.
Western Downs Battery	GFM transition revealed harmonic emission risks and importance of upfront spatial/filter planning.	Delayed harmonic allocation from NSPs limited mitigation time and posed compliance risks.

presents a comparative overview of the aforementioned GFM projects, highlighting key insights and implementation challenges identified throughout their deployment.

4. Conclusions

This systematic review thoroughly examines GFM control strategies within modern power systems and microgrids, emphasizing their importance in achieving stability, resilience, and energy systems primarily dependent on inverters. The review initially identified 242 studies across four databases: Scopus, Web of Science, IEEE XPLORE, and LENS. After removing 77 duplicate articles, the remaining studies were screened using well-defined inclusion and exclusion criteria, resulting in 102 studies. After a meticulous eligibility process that included a comprehensive quality review, a total of 75 particularly significant studies were selected for detailed evaluation. The study categorized and evaluated various GFM control techniques, such as droop control, VSM, PSC, VOC, and advanced control strategies. Traditional methods, such as droop control and VSM, are widely deployed because of their simplicity and good performance. On the other hand, advanced AI-based and model predictive techniques offer improved dynamic response and adaptability. However, these methods are in the initial development stage and are being tested in pilot environments.

Stability continues to be a critical focus, with both small-signal and transient stability addressed through theoretical modeling, real-time simulation, and experimental testing. Innovative approaches, including adaptive damping, hybrid GFM setups, and control strategies tuned for current-limiting conditions, show promise in improving the resilience of the system under high penetration of IBRs.

A critical part of this review involved analyzing several high-impact GFM deployments worldwide, including the Hornsdale Power Reserve, Blackhillock BESS, and Kriegers Flak HVDC project. These cases highlighted several important operational insights: (i) GFM inverters provide reliable frequency and voltage regulation in low-inertia systems, (ii) effective deployment is strongly influenced by compliance with grid codes and standardized testing procedures, and (iii) hybrid configurations combining GFM and GFL controls are gaining traction for optimizing both system stability and control flexibility. Furthermore, real-world projects have indicated that commissioning delays frequently stem from the lack of unified testing standards, highlighting the critical role of collaboration between manufacturers, system operators, and regulators in facilitating widespread, reliable GFM deployment. These insights emphasize the need to support simulation-based analyses with field-level implementation to verify real-world system behavior.