Operational Challenges of Digital Real-Time Simulation (DRTS) for the Management of Distributed Energy Resources (DERs)

Abstract

The transition toward active and converter-dominated distribution networks is increasing the need for Digital Real-Time Simulation (DRTS) platforms capable of supporting the validation, operation, and planning of modern power systems with high penetrations of Distributed Energy Resources (DERs). This review aims to critically examine the technological and methodological barriers that continue to limit the evolution of DRTS from a validation tool toward an operational cyber–physical infrastructure for future intelligent power systems. Particular attention is devoted to computational scalability, communication latency, Hardware-in-the-Loop (HIL), model conversion, proprietary “black-box” devices, digital twins, and the computational strategies required to preserve deterministic real-time execution. The reviewed literature indicates that maintaining real-time determinism while preserving high-fidelity electromagnetic transient (EMT) models remains one of the principal technological challenges for future DRTS platforms. The analysis further shows that many of the current limitations associated with digital twins, grid-forming technologies, and large-scale industrial deployment originate not from isolated technological deficiencies, but from the interaction among computational, communication, interoperability, synchronization, and model-management constraints. The evidence indicates that overcoming these limitations requires coordinated advances in computational architectures, communication infrastructures, model automation, interoperability, and cyber–physical integration rather than isolated hardware improvements. Overall, this review argues that the future impact of DRTS will depend not only on improvements in simulation performance, but also on its evolution into an interoperable and experimentally oriented cyber–physical infrastructure capable of supporting phenomenological analysis and the next generation of intelligent, resilient, and autonomous power systems.

1. Introduction

Electrical distribution networks are undergoing profound changes as Distributed Energy Resources (DERs) technologies become increasingly integrated into modern power systems. This evolution is transforming distribution systems from passive infrastructures into actively managed networks with bidirectional power flows and more demanding operational requirements [1,2]. At the same time, Distribution System Operators (DSOs) are no longer limited to delivering electricity to consumers but are increasingly required to coordinate flexibility services and maintain system stability under operating conditions characterized by reduced inertia and increasing operational uncertainty, highlighting the need for advanced real-time simulation environments capable of supporting their analysis, validation, and operation [1,2,3,4,5].

This transformation introduces significant technical challenges. Conventional power systems relied heavily on synchronous machines, whose mechanical inertia contributed to frequency stability and fault response. However, the progressive displacement of synchronous generation by inverter-based resources (IBRs) has fundamentally altered the dynamic behavior of modern distribution networks. To compensate for the loss of natural inertia and maintain grid stability, advanced control strategies based on grid-forming (GFM) converters, distributed control architectures, and flexibility services have become key enabling technologies [6,7,8]. These developments require simulation environments capable of accurately reproducing fast EMT, communication delays, controller interactions, and cyber–physical dependencies that cannot be adequately represented using conventional offline simulation tools [1,9,10,11]. These emerging challenges motivate the need for a comprehensive assessment of the current capabilities and limitations of DRTS technologies. Accordingly, this review examines the computational, communication, modeling, and interoperability barriers that continue to limit the evolution of DRTS toward advanced cyber–physical infrastructures for future active distribution networks. DER management is considered not as a standalone control or DERMS problem, but as an application domain whose increasing reliance on real-time monitoring, coordination, communication, and power-electronic interfaces places additional demands on DRTS infrastructures. The review therefore focuses on the technological constraints that determine whether DRTS can reliably support DER-oriented applications rather than on the architectures or functions of DER management systems themselves.

In this context, Digital Real-Time Simulation (DRTS) has emerged as one of the most promising technologies for the development, validation, and future operation of distribution systems. By combining real-time execution with HIL and Power Hardware-in-the-Loop (PHIL), DRTS platforms enable the closed-loop interaction between physical devices and their corresponding digital models within a deterministic simulation environment, supporting the safe evaluation of advanced controllers, DERs, and GFM technologies before field deployment [11,12,13]. Furthermore, the convergence of DRTS with digital twins, artificial intelligence (AI), and cyber–physical systems has expanded its role from equipment validation toward cyber–physical experimentation environments capable of supporting advanced controller validation, digital twin implementation, AI-assisted decision support, and phenomenological analysis of future converter-dominated power systems [14,15,16].

Despite these advances, the widespread adoption of DRTS by DSOs continues to be constrained by a combination of computational, communication, and interoperability challenges. The literature consistently identifies limitations associated with computational scalability, overrun phenomena, model partitioning, communication latency, synchronization requirements, model conversion, and proprietary (“black-box”) devices. At the same time, important questions remain regarding the scalability of EMT simulations, the practical limits of grid-forming converters, the maturity of digital twin architectures, and the transition of DRTS from research laboratories toward large-scale industrial deployment [1,10,15].

Several review studies have already examined important aspects of DRTS, HIL and PHIL, cyber–physical power systems, and digital twins. However, these studies generally focus on specific technologies, simulation approaches, or application domains rather than on how their limitations interact within a common DRTS framework. This distinction becomes increasingly important in active distribution networks, where computational scalability, communication latency, synchronization, model interoperability, and the representation of physical devices are no longer independent issues. The present review therefore focuses on these interactions and on how they affect the evolution of DRTS from a simulation and validation tool toward a broader cyber–physical experimentation infrastructure. To make this distinction explicit, representative previous reviews are compared with the scope of the present study in

Table 1. Comparison of Previous Reviews and the Distinct Perspective of the Present Review.

Review	Primary Focus	Scope Covered by the Previous Review	Distinct Perspective of the Present Review
Faruque et al. (2015) [11]	Real-time simulation technologies	DRTS architectures, hardware and software, I/O systems, modeling and solution techniques, HIL, computing capabilities, and available simulator platforms.	Moves from describing DRTS technologies toward examining how computational, communication, modeling, and interoperability constraints interact and affect their evolution toward cyber–physical experimentation.
Gaitán-Cubides et al. (2022) [2]	DRTS theory and energy-transition applications	DRTS classification, solution methods, sampling, HIL/PHIL, multi-rate and distributed simulation, and applications related to the energy transition, Smart Grids, and distribution systems.	Provides a critical infrastructure-oriented perspective that links computational scalability, communication, synchronization, model integration, and advanced applications rather than primarily cataloging energy-transition applications.
Sidwall and Forsyth (2022) [10]	Real-time simulator development and best practices	Power-electronics modeling, converter simulation, HIL testing, IEC 61850 [17] simulation and interfacing, and simulator development from the RTDS manufacturer perspective.	Extends beyond a manufacturer-specific platform perspective to compare cross-domain constraints affecting DRTS scalability, interoperability, Digital Twins, and industrial transition.
Nasab et al. (2024) [1]	DRTS applications and future trends	DRTS/HIL applications, transmission systems, HVDC, protection, WAMPAC, TSO–DSO interactions, BESS, and laboratory implementations.	Focuses on the operational bottlenecks and interdependencies that limit DRTS scalability and evolution, including communication, model conversion, black-box devices, Digital Twins, and industrial adoption.
Montoya et al. (2020) [13]	Advanced laboratory testing	RTS, PHIL, CHIL, PSIL, co-simulation, geographically distributed testing, interoperability, cybersecurity, DERs, and laboratory/industrial testing experiences.	Places laboratory testing within a broader chain of DRTS constraints, examining how computational execution, communication, synchronization, model integration, and physical interfaces collectively limit experimentation and deployment.
Vogt et al. (2018) [18]	Smart-grid co-simulation	26 co-simulation frameworks, simulation tools, synchronization methods, computational effort, problem size, research topics, and identified co-simulation trends.	Treats co-simulation as one component of a broader DRTS infrastructure problem and connects synchronization and computational constraints with physical hardware, interoperability, Digital Twins, and industrial deployment.
Strezoski (2023) [5]	DER management systems (DERMSs)	Centralized and decentralized DER management, DER aggregation, real-time grid management, forecasting, flexibility, communication with DERs, vendors, and pilot projects.	Addresses the simulation infrastructure required to validate and evolve DER-oriented functions, rather than the architecture and functionality of DERMS themselves.
Thwe et al. (2025) [15]	Digital Twins for power systems	DT definitions, applications, functional and non-functional requirements, enabling technologies, data federation, interoperability, academic and industrial practices, and associated challenges.	Examines the DRTS capabilities and limitations that condition the transition from digital representations toward experimentally coupled and operational Digital Twins.
Aslam et al. (2024) [19]	Integrated control–communication modeling and Smart Grid co-simulation	Communication infrastructure, control and network simulators, co-simulation platforms, cyber–physical smart grid modeling, comparative platform analysis, and communication-related challenges.	Integrates communication and computational constraints with DRTS execution, HIL/PHIL, model interoperability, converter-dominated applications, Digital Twins, and industrial transition.
Tozak et al. (2024) [8]	Grid-forming converter modeling and control	GFL/GFM structures, modeling approaches, control objectives, applications, and GFM installation projects involving BESS, wind, hybrid systems, and HVDC.	Treats GFM as one advanced application through which broader DRTS constraints—model fidelity, computational burden, HIL/PHIL, synchronization, and validation—can be examined.
Present review	Operational challenges of DRTS for DER-oriented power systems	Computational scalability, communication and synchronization, HIL/PHIL, model conversion, interoperability and black-box devices, Digital Twins, GFM applications, and industrial deployment.	Provides a cross-domain critical perspective on the interactions among these constraints and their role in the evolution of DRTS from a simulation/validation platform toward a cyber–physical experimentation infrastructure.

.

Against this backdrop, this review examines how computational, communication, interoperability, and experimental constraints interact to shape the evolution of DRTS for DER-oriented power systems. Particular attention is given to the technological contradictions and unresolved challenges that arise when DRTS is extended from real-time simulation and equipment validation toward cyber–physical experimentation and advanced applications.

1.1. Review Methodology

The review follows a thematic and critical literature-search strategy rather than a bibliometric or fully systematic review protocol. The literature search was conducted using IEEE Xplore, Scopus, and Web of Science, with the search extended through 2026. The search focused primarily on peer-reviewed journal articles published in English and addressing topics directly related to DRTS, active distribution networks and DSOs, DERs, HIL, PHIL, Digital Twins, Grid-Forming converters, communication and synchronization, and cyber–physical power systems.

Because the review also addresses standards, interoperability, and documented experimental infrastructures, selected technical, institutional, and normative sources were additionally considered when they provided information that could not be adequately represented by journal articles. These complementary sources included relevant standards, technical specifications, institutional reports, and documented laboratory infrastructures directly related to DRTS deployment and interoperability.

The literature was identified through an iterative search process structured around the four review questions and the main technical dimensions of the review. These dimensions included computational execution and scalability, communication and synchronization, model integration and interoperability, HIL/PHIL, Digital Twins, grid-forming converters, and advanced DRTS applications. The search combined terms related to DRTS, power systems, distribution networks, and DERs with domain-specific terms such as computational scalability, overrun, partitioning, multi-rate simulation, communication latency, jitter, synchronization, IEC 61850, GOOSE, Sampled Values, interoperability, model conversion, Digital Twins, HIL/PHIL, grid-forming converters, energy storage, electric vehicles, and flexible loads. Boolean combinations using AND/OR operators were progressively refined across the databases to account for variations in terminology and to capture the interconnected research domains rather than relying on a single fixed search string. Duplicate records were removed during the screening process. Although the search emphasized the recent literature, particularly studies published from approximately 2020 onward, earlier studies were retained when they provided foundational methodologies, established simulation approaches, or review perspectives directly relevant to the research questions.

The screening process was conducted in successive stages. First, 145 records were retrieved from the selected databases. Title and abstract screening, together with duplicate removal, resulted in 129 retained records. Second, the full texts were evaluated according to their technical relevance to the four research questions, resulting in 113 eligible studies and complementary sources. Finally, the selected literature was consolidated thematically and verified for relevance, scope, and evidentiary contribution, resulting in a final corpus of 109 references used throughout the review.

Studies were considered eligible when they were peer-reviewed journal articles published in English and made a direct technical contribution to at least one of the four review questions concerning computational execution, communication and synchronization, model integration and interoperability, HIL/PHIL, and advanced DRTS applications. Conference and workshop papers, non-English publications, and studies without a sufficiently direct connection to the technical scope of the review were excluded. The resulting literature was then analyzed thematically to identify recurring technical constraints, interactions among research domains, and unresolved research gaps.

Figure 1 summarizes the literature-search and thematic selection strategy adopted in this review. The process combines the use of three major scientific databases with an iterative thematic search, duplicate removal, full-text assessment, and final thematic consolidation. The selected sources were organized according to the four review questions while maintaining the cross-domain perspective adopted throughout the review.

The main stages of the literature selection and screening process are summarized in

Table 2. Literature Search and Selection Process.

Stage	Description	Result
Identification	Search in IEEE Xplore, Scopus, Web of Science	145 records retrieved
Screening	Title/abstract screening, duplicate removal	129 retained records
Eligibility	Full-text assessment for technical relevance	113 included studies
Final Corpus	Thematic analysis across four review questions	109 references

, which details the identification, screening, eligibility, and final inclusion criteria applied throughout the review.

1.2. Contribution and Organization of the Review

Based on the comparison presented in Table 1, the contribution of this review lies not simply in combining several DRTS-related topics, but in examining how limitations in one part of the simulation infrastructure affect other parts. Computational execution, communication and synchronization, model integration and interoperability, HIL/PHIL, and advanced applications are considered as connected elements of the same DRTS problem. This perspective is particularly relevant for active distribution networks, where the increasing presence of DERs and power-electronic interfaces makes these dependencies more pronounced.

From this perspective, the present study examines the computational and communication constraints associated with large-scale DRTS implementations, the impact of model conversion, interoperability, and proprietary (“black-box”) devices on simulation fidelity, and the role of DRTS as an enabling technology for HIL validation, digital twins, grid-forming converters, and cyber–physical experimentation. More importantly, it identifies the principal technological contradictions, unresolved challenges, and research opportunities that continue to limit the transition of DRTS from research laboratories toward large-scale industrial deployment.

To structure the critical analysis, the review is guided by four questions:

• RQ1. What computational constraints limit the scalability and deterministic execution of DRTS in active distribution networks with increasing penetration of DERs?
• RQ2. How do communication latency, synchronization, and multi-domain interactions affect the fidelity and real-time execution of DRTS?
• RQ3. How do model conversion, interoperability, proprietary devices, and HIL/PHIL interfaces influence the practical fidelity and scalability of DRTS-based experimentation?
• RQ4. What role can DRTS play in advanced applications such as Digital Twins and grid-forming converter validation, and what factors limit their transition from laboratory environments toward broader industrial deployment?

To facilitate this analysis, the review is organized into three interconnected layers. The first layer introduces the evolution of active distribution networks and the flexibility services enabled by DER integration. The second layer focuses on the computational and communication infrastructures required to support deterministic real-time simulation. The third layer examines advanced DRTS applications, including model conversion, digital twins, HIL validation, and the emerging role of DRTS as a cyber–physical experimentation platform. The paper concludes with a critical discussion of the principal technological contradictions, research gaps, and future directions identified throughout the reviewed literature, followed by the main conclusions and perspectives for future research.

Figure 2 illustrates the conceptual framework adopted throughout this review and the relationships among its three analytical layers. Rather than representing independent research domains, the framework highlights how active distribution networks, DRTS infrastructures, and advanced applications evolve as tightly interconnected components of the same cyber–physical ecosystem. This perspective provides the organizational backbone of the paper and establishes the conceptual context for the cross-domain interactions, current limitations, and future opportunities discussed throughout the paper.

2. Computational Constraints and Infrastructure

The rapid digitalization of electrical distribution networks has been accompanied by the widespread integration of DER and advanced power-electronics-based control strategies into aging power system infrastructures. This transformation has significantly increased both the computational complexity of power system models and the real-time execution requirements imposed on DRTS platforms. Unlike conventional offline simulations, where execution time mainly affects computational efficiency, DRTS platforms must complete every numerical iteration within strict real-time deadlines while simultaneously reproducing EMT, communication processes, and control system dynamics. Real-time computational performance is therefore no longer determined solely by processor speed, but also by the ability to sustain deterministic execution under increasingly complex cyber–physical operating conditions. As a result, computational capacity, hardware architecture, and workload distribution become fundamental factors for ensuring both simulation fidelity and numerical stability in advanced distribution network applications [1,2,11].

These computational demands extend beyond processor performance alone. Modern DRTS platforms must simultaneously address large-scale model scalability, synchronization among subsystems operating at different temporal resolutions, and the efficient coordination of distributed computing resources without compromising deterministic real-time execution. These challenges have motivated the adoption of advanced techniques such as model partitioning, multi-rate co-simulation, heterogeneous computing, and distributed execution frameworks, all designed to improve computational efficiency while preserving simulation fidelity. However, no single strategy completely resolves these challenges, since improvements in computational performance are frequently accompanied by new synchronization, communication, and interoperability constraints. These trade-offs have become a fundamental design consideration for next-generation DRTS platforms capable of supporting increasingly digitalized active distribution networks [20,21,22].

The computational challenges outlined above indicate that achieving deterministic real-time simulation depends not only on increasing computational power but also on balancing scalability, synchronization, and numerical accuracy. The following subsections examine these computational bottlenecks, review the principal mitigation strategies adopted in modern DRTS platforms, and identify the remaining research challenges associated with future large-scale DER-oriented power systems.

2.1. Temporal Determinism and the Overrun Phenomenon

The defining characteristic of DRTS, compared with conventional offline simulation, is its strict synchronization with physical time. Accordingly, one second of simulation must correspond exactly to one second in the real physical system, which implies that all numerical calculations and operations must be completed within a predefined time interval. This requirement, known as temporal determinism, is the fundamental requirement for ensuring that the simulator can interact consistently with external physical devices in environments such as HIL, PHIL, and advanced driver validation platforms [1,10,11,12].

Determinism requires that the differential equations representing the dynamic behavior of the electrical system be completely solved within each simulation time step. In electrical power system applications, particularly those based on EMT, this requirement becomes computationally demanding because of the large number of state variables and the need to employ time steps on the order of microseconds to accurately capture the fast electromagnetic phenomena associated with power electronic devices, converters, and energy storage systems [1,11,23].

The main operational challenge arises when the computational cost required to solve the model exceeds the time available within each simulation cycle, a condition commonly referred to as overrun [11,23]. As a result, the simulator loses its real-time execution capability and effectively behaves as a non-real-time simulation. In this scenario, the results are no longer suitable for applications requiring real-time interaction, since the resulting delays alter the dynamic response of controllers, physical devices, and communication systems connected to the simulator [23].

The occurrence of overrun can be attributed to multiple factors, including excessive model complexity, inadequate selection of the simulation step, the use of models with a high level of electromagnetic detail, complex control algorithms, and an inefficient distribution of the computational load among the available hardware resources [24,25]. The migration of models originally developed for offline simulations to real-time platforms can generate additional difficulties when the time constraints imposed by deterministic execution are not considered [24,25,26]. These factors rarely occur independently, and in practical applications they often interact, making overrun a multidimensional computational problem rather than the consequence of a single hardware limitation [20].

For modern distribution networks, this challenge becomes even more relevant because of the increasing penetration of DERs into conventional power grids [9,27]. DER technologies increasingly incorporate power electronic converters that operate at high switching frequencies and require finer temporal resolutions than those traditionally adopted in electrical power system stability studies. As a result, the computational load increases significantly, reducing the number of nodes that can be simulated on a single processor and increasing the likelihood of overrun events [27,28].

Despite the continuous increase in computational performance, the overrun problem has not disappeared. Rather than representing a limitation that can be eliminated through successive hardware generations, the reviewed literature suggests that computational demand has expanded at a comparable pace as simulation models have incorporated higher switching frequencies, larger electrical networks, and tighter cyber–physical integration. Consequently, improvements in processing capability have often been accompanied by increasingly ambitious simulation objectives, preventing overrun from becoming merely a hardware-related issue [21,22,26].

This persistent imbalance establishes a direct relationship between model fidelity and the available computational capacity. Although higher temporal resolution allows a more accurate representation of the fast electromagnetic phenomena present in DERs, it also increases the number of operations that must be executed during each simulation cycle, making the available computational resources one of the main bottlenecks for implementing large-scale EMT simulations [1,2,9]. This trade-off also explains why reported computational performance cannot be compared solely through processor specifications or simulation step size. Similar execution times may correspond to substantially different model complexities, making the practical scalability of DRTS platforms strongly dependent on modeling decisions rather than computational power alone [21,22,29].

Several approaches have been proposed to mitigate the overrun phenomenon, including multi-core architectures, FPGA-based platforms, specialized hardware accelerators, and advanced computational load balancing techniques [21,26]. Increasing computational resources alone is not sufficient to address this challenge. As simulation models continue to grow in size and complexity, maintaining real-time performance increasingly depends on how efficiently the computational workload is organized rather than solely on the available hardware. Model partitioning and workload distribution across multiple processing cores have evolved from optimization techniques into essential design requirements for modern DRTS platforms [21,26,29]. These approaches also introduce new challenges related to synchronization, communication overhead, and subsystem coupling, which must also be carefully addressed to preserve simulation fidelity [29].

From this perspective, model partitioning should not be interpreted as a universal solution to overrun. Although partitioning reduces the computational burden assigned to individual processors, it simultaneously increases data exchange and synchronization requirements between subsystems. As a result, strategies intended to improve computational efficiency may introduce additional communication overhead or numerical inaccuracies, particularly in strongly coupled electrical networks [21,29].

The reviewed literature indicates that overrun should no longer be interpreted solely as a consequence of insufficient computational resources. It reflects the increasing difficulty of preserving deterministic execution while simultaneously increasing model fidelity, system scale, and cyber–physical interaction. Addressing overrun requires a combination of computational optimization, model partitioning, and synchronization strategies rather than isolated hardware upgrades. This observation explains why recent research has progressively shifted from hardware-oriented solutions toward scalable computational architectures capable of supporting increasingly complex DRTS applications [21,22].

Table 3. Comparison of Representative Strategies for Preserving Temporal Determinism in DRTS.

Mitigation Strategy	Primary Objective	Main Advantages	Principal Limitations
Model partitioning	Distribute the computational workload across multiple processing cores or simulation nodes.	Improves computational scalability, enables larger EMT models, and reduces processor overload.	Requires efficient synchronization among subsystems and may introduce communication overhead.
Multi-rate simulations	Execute different subsystems using time steps adapted to their dynamic behavior.	Reduces computational burden while maintaining high temporal resolution where required.	Selecting appropriate time steps is challenging for strongly coupled electromagnetic systems and may affect numerical stability.
FPGA-based acceleration	Offload computationally intensive numerical tasks to dedicated hardware.	Enables deterministic execution with very small simulation time steps and high processing performance.	Increases hardware complexity, development effort, and implementation cost while offering limited flexibility for model modifications.
Computational load balancing	Distribute simulation tasks evenly among available processing resources.	Improves processor utilization and minimizes local computational bottlenecks.	Performance strongly depends on model structure and the efficiency of the partitioning strategy.
Real-time model adaptation	Modify or simplify offline simulation models to satisfy deterministic execution constraints.	Facilitates migration from offline environments to DRTS platforms while preserving essential system dynamics.	Model simplifications may reduce simulation fidelity, particularly for fast EMT and converter-dominated systems.

provides an overview of the principal computational challenges associated with the overrun phenomenon together with the representative mitigation strategies most frequently reported in the literature.

Consequently, overrun should be understood not only as a computational constraint but also as an indicator of the growing complexity associated with integrating detailed electrical models, advanced control strategies, and cyber–physical interactions within a deterministic real-time framework. This perspective provides the basis for the following sections, where scalability and synchronization are examined as complementary challenges rather than independent computational limitations [21,22,29].

2.2. Structural Scalability, Partitioning, and Distributed Co-Simulation

The ability to represent increasingly complex electrical distribution networks is one of the main challenges for DRTS [1,10]. Although advances in computing hardware have significantly increased the available processing power, the accelerated growth of modern distribution networks has caused model complexity to outpace the improvements in computational performance [11,22].

Unlike transmission systems, where equivalent network models are often sufficient, distribution networks require a much more detailed representation owing to their high penetration of DERs, heterogeneous loads, bidirectional power flows, and the extensive use of power-electronics-based devices. As a result, EMT models associated with distribution systems usually contain a significantly larger number of differential and algebraic equations, increasing the computational burden and reducing the size of the systems that can be simulated within a single processor. This growing model complexity has become one of the primary factors limiting the scalability of modern DRTS platforms [1,2,22,30].

This observation highlights an inherent trade-off in large-scale DRTS. Increasing the size of the simulated network rarely depends solely on additional computational resources; it often requires simplifying portions of the model or reducing the level of electromagnetic detail. Consequently, structural scalability should not be interpreted as the ability to simulate larger systems without compromise, but rather as the ability to balance computational feasibility against model fidelity according to the objectives of the simulation [1,21,22,24].

These scalability constraints often force engineers to simplify network models or replace portions of the system with Thevenin or Norton equivalents to preserve deterministic execution [24,25].

Traditionally, DRTS platforms have addressed this problem using multi-core architectures [20,21], where different tasks can be executed simultaneously. However, the simple availability of multiple processors does not guarantee a commensurate improvement in performance. For parallelization to be effective, the model must be divided into smaller subsystems that can run independently, while minimizing dependencies between them. This process, known as partitioning, is one of the fundamental mechanisms for increasing simulation capacity without compromising temporal determinism [21,31].

According to the level of user intervention, partitioning strategies are commonly categorized into implicit and explicit approaches. In implicit partitioning, the simulation software automatically distributes computational tasks among the available processing resources [20,21]. Although this approach simplifies the modeling process, its performance often deteriorates as network complexity increases, making explicit partitioning strategies increasingly necessary. In explicit partitioning, the user manually defines the boundaries between subsystems to optimize load distribution, reduce communication times, and maximize processor utilization [21,32].

Table 4 compares the representative partitioning strategies adopted in DRTS platforms, highlighting their implementation principles together with their principal advantages and remaining limitations. Although each strategy contributes to improving computational scalability, their effectiveness ultimately depends on the characteristics of the simulated network, the selected synchronization mechanism, and the communication architecture supporting the real-time execution.

Among the most commonly used techniques to enable parallel execution are transmission line decoupling techniques based on distributed-parameter models. These techniques introduce a computational and mathematical separation between different regions of the network, allowing each subsystem to execute independently on a dedicated core or processor without simultaneously solving the complete system of equations. When the propagation time associated with the transmission line is compatible with the selected simulation step, this approach preserves EMT accuracy while significantly reducing the computational load on each processing unit [21,32,33].

However, the selection of appropriate partition points is critical for maintaining simulation stability and accuracy. Inadequate partitioning may increase communication delays, synchronization errors, numerical inconsistencies, and spurious energy exchanges between subsystems, ultimately compromising simulation stability and convergence [32,33,34,35].

Table 4. Comparison of representative partitioning strategies used in modern DRTS platforms.

Partitioning Strategy	Mathematical/Physical Principle	Representative Platforms	Main Advantages	Principal Limitations	Representative References
Implicit partitioning	Automated DAG (Directed Acyclic Graph) task scheduling and state-space matrix decomposition by solver.	RTDS–HYPERSIM	Simple implementation and reduced manual setup; automatic core allocation.	Suboptimal load balancing in highly asymmetric/non-linear distribution networks.	[16,18]
Explicit partitioning	Manual subsystem boundary assignment at user-selected nodes/busses.	OPAL-RT RTDS	Optimal CPU/FPGA pipeline utilization; controlled inter-core latency.	Requires deep system knowledge; manual reconfiguration needed upon topology changes.	[16,18]
Distributed-parameter decoupling lines	Traveling wave propagation delay based on Bergeron/ULM model	RTDS, OPAL-RT, HYPERSIM	Full EMT matrix decoupling without loss of dynamic fidelity; enables parallel execution.	Constrained by physical transmission line length; unsuited for short lines without artificial stubbing.	[18]
Distributed multi-rack partitioning	High-speed inter-chassis communication via optical/FPGA backplanes (Aurora, PCIe, SFP+).	RTDS, NovaCor, OPAL-RT, eMEGAsim.	Scales to ultra-large EMT systems (>10,000 active nodes/DERs).	Increased packet communication overhead, jitter, and inter-rack sync constraints.	[11,32,35]

Table 4. Comparison of representative partitioning strategies used in modern DRTS platforms.

Partitioning Strategy	Mathematical/Physical Principle	Representative Platforms	Main Advantages	Principal Limitations	Representative References
Implicit partitioning	Automated DAG (Directed Acyclic Graph) task scheduling and state-space matrix decomposition by solver.	RTDS–HYPERSIM	Simple implementation and reduced manual setup; automatic core allocation.	Suboptimal load balancing in highly asymmetric/non-linear distribution networks.	[16,18]
Explicit partitioning	Manual subsystem boundary assignment at user-selected nodes/busses.	OPAL-RT RTDS	Optimal CPU/FPGA pipeline utilization; controlled inter-core latency.	Requires deep system knowledge; manual reconfiguration needed upon topology changes.	[16,18]
Distributed-parameter decoupling lines	Traveling wave propagation delay based on Bergeron/ULM model	RTDS, OPAL-RT, HYPERSIM	Full EMT matrix decoupling without loss of dynamic fidelity; enables parallel execution.	Constrained by physical transmission line length; unsuited for short lines without artificial stubbing.	[18]
Distributed multi-rack partitioning	High-speed inter-chassis communication via optical/FPGA backplanes (Aurora, PCIe, SFP+).	RTDS, NovaCor, OPAL-RT, eMEGAsim.	Scales to ultra-large EMT systems (>10,000 active nodes/DERs).	Increased packet communication overhead, jitter, and inter-rack sync constraints.	[11,32,35]

These limitations indicate that no partitioning strategy can be considered universally optimal. Approaches that maximize computational efficiency under one network topology or operating condition may become less effective as subsystem coupling, communication requirements, or model objectives change. Consequently, partitioning should be viewed as a model-dependent design decision rather than a standardized procedure applicable to all DRTS studies [21,31,32,33,34,35].

The evolution of DRTS platforms has made it possible to extend partitioning strategies to distributed architectures composed of multiple simulation racks interconnected through high-speed communication networks [13,34]. These architectures enable the representation of large-scale systems that exceed the computational capacity of a single simulation platform while supporting the coordinated execution of complex models for active distribution networks and highly digitalized microgrids. As the level of model distribution increases, so does the need to efficiently manage synchronization and data exchange between separate subsystems, leading to significant communication challenges [34,36].

Although partitioning strategies partially overcome the limitations associated with processing capacity, an additional challenge arises from the coexistence of phenomena evolving at different time scales. While some subsystems require time steps on the order of microseconds to accurately represent power electronic devices, others can be modeled using considerably larger time steps [37]. These different time scales require the development of multi-rate and multi-domain co-simulation approaches that seek to optimize computational resource utilization without compromising the stability or fidelity of real-time simulation [21,38,39].

The reviewed literature suggests that partitioning has evolved from a computational optimization technique into a fundamental architectural principle for large-scale DRTS [21,31,39]. Its effectiveness depends not only on the selected partitioning strategy but also on the interaction among synchronization mechanisms, communication latency, and model topology. This interdependence also explains why improvements in scalability frequently shift computational challenges toward synchronization and communication management rather than eliminating them altogether [21,31,34,36,39].

2.3. Multi-Rate Co-Simulation and Multi-Domain Synchronization

Although partitioning strategies enable computational load distribution and increase the simulation capacity of DRTS platforms, modern distribution networks still contain phenomena evolving at significantly different time scales. Power electronic converters, particularly those associated with photovoltaic systems, grid-forming converters, and primary control loops, may require simulation time steps on the order of microseconds, whereas energy management systems (EMSs), secondary control, grid supervision, and electricity market applications evolve over time scales ranging from milliseconds to several seconds [40]. Using a single time step for the entire system therefore requires a temporal resolution small enough to capture the fastest dynamics, even when such a resolution is unnecessary for slower components. This increases the computational burden and the risk of real-time overruns, often forcing model simplification or a reduction in the number of represented components [1,20,21,22,27].

Although partitioning strategies enable computational load distribution and increase the simulation capacity of DRTS platforms, a fundamental challenge associated with the different time scales of modern electrical systems remains. Modern distribution networks integrate devices exhibiting significantly different dynamic behaviors. Power electronic converters, particularly those associated with photovoltaic systems, grid-forming converters, and primary control loops, may require simulation time steps on the order of microseconds, whereas energy management systems (EMSs), secondary control, grid supervision, and electricity market applications evolve over time scales ranging from milliseconds to several seconds [24,33]. This coexistence of fast and slow phenomena prevents the use of a single simulation time step for the entire system, making time management one of the principal challenges in DRTS.

The use of a single simulation time step requires selecting a temporal resolution small enough to accurately represent the fastest phenomena present in the network. Although this approach ensures high simulation accuracy, it also considerably increases the computational load because the entire system must be solved using a time step that may be unnecessarily small for many modeled components, thereby increasing the risk of real-time overruns. As a result, the ability to represent large-scale networks with high levels of electromagnetic detail is ultimately constrained by the available computational resources. In many cases, this limitation forces model simplification or a reduction in the number of represented components to maintain compliance with real-time requirements [20,21,22,41].

Beyond these computational limitations, maintaining high simulation fidelity in large-scale power systems has driven the adoption of High-Performance Computing (HPC) architectures in DRTS platforms, often requiring substantial investments in computational infrastructure. To address these limitations, distributed real-time laboratories, also referred to as collaborative or SuperLab environments, have emerged, enabling multiple institutions interconnected through high-speed communication networks to jointly execute the same real-time simulation. Under this collaborative framework, each laboratory contributes computational resources, specialized models, or experimental equipment according to its expertise, thereby expanding the scale and complexity of studies that can be performed under deterministic real-time conditions [13,36,42].

To overcome these limitations, Multi-Rate Real-Time Simulation (MR-RTS) has become one of the principal strategies for improving computational efficiency in DRTS platforms. This approach divides the system into subsystems with different temporal requirements and assigns each subsystem an appropriate simulation time step according to its dynamic behavior [21,32].

The adoption of multi-rate simulation, however, does not eliminate computational constraints; instead, it redistributes them. While assigning different time steps improves processor utilization, it also increases the complexity of coordinating information exchanged across temporal boundaries. As a result, the computational gains achieved through temporal decomposition must be balanced against the additional synchronization requirements introduced by the multi-rate architecture itself [21,32,38].

Complementary to multi-rate simulation, multi-domain approaches represent different subsystems of the power system using mathematical models selected according to the dominant physical phenomena. One of the most widely adopted approaches is EMT–RMS co-simulation, in which subsystems containing power electronic converters, grid-forming converters, or DERs are represented using EMT models, whereas the remaining portions of the network are modeled using RMS representations. This hybrid modeling strategy concentrates computational resources on regions requiring high dynamic fidelity, thereby reducing the computational effort associated with representing the entire system in the EMT domain [9,11,25,43].

Figure 3 illustrates the conceptual framework of multi-rate and multi-domain co-simulation adopted in modern DRTS platforms. The figure shows how different dynamic domains operating at distinct temporal scales are coordinated through synchronization and communication mechanisms to balance computational efficiency with simulation fidelity.

The increasing diversity of DER applications also expands the scope of the systems that DRTS platforms may need to represent. DERs are increasingly coupled not only with stationary storage and flexible loads, but also with transportation-energy systems in which charging demand, battery swapping, renewable generation, and network conditions interact across time and space. Recent work on solar-powered highway systems, for example, considers the coordinated management of electric-vehicle charging and battery-swapping loads to increase the utilization of solar generation [44]. Although such studies are not themselves DRTS applications, they illustrate the growing complexity of DER-oriented scenarios that may require real-time simulation for control validation, communication assessment, and coordinated operation. This broader coupling reinforces the need for multi-rate and multi-domain simulation strategies capable of representing electrical, control, and transportation-related dynamics at their respective temporal and spatial scales.

These interactions also introduce additional requirements for DRTS-based validation. EV charging, battery storage, flexible loads, and renewable generation operate across different temporal scales and may be coordinated through communication-dependent control strategies [44]. Their integration therefore requires simulation environments capable of representing not only the electrical dynamics of DERs, but also the control, communication, and coordination mechanisms that determine their aggregate behavior. In this sense, the increasing coupling between DERs and transportation-energy systems strengthens the need for multi-domain DRTS environments in which electrical, control, and communication phenomena can be evaluated under coordinated operating conditions.

Beyond hybrid modeling approaches, several computational techniques have been proposed to further improve computational efficiency in DRTS platforms. Among them, Shifted Frequency Analysis (SFA) translates the frequency content of the signals toward lower frequencies, allowing larger simulation time steps while preserving the representation of low-frequency oscillations and dynamic phenomena associated with low-inertia systems [38,39]. Another widely adopted strategy is the use of substep-based architectures, in which fast-switching devices are solved using internal simulation time steps smaller than the main simulation step, thereby improving computational efficiency without compromising the fidelity of critical electromagnetic phenomena [38,39,45].

Average Value Models (AVMs) are widely employed in studies where the primary objective is energy management, operational stability, or the evaluation of long-term operating scenarios. These models eliminate the need to explicitly represent individual switching events, substantially reducing computational cost while enabling the simulation of large-scale systems with high DER penetration [43,46]. As a result, DRTS platforms can analyze complex scenarios associated with active distribution networks, storage systems, and low-inertia energy ecosystems without requiring the computational power of a fully detailed EMT simulation [13,47].

The diversity of these computational strategies reflects the absence of a single modeling approach capable of simultaneously maximizing computational efficiency, numerical accuracy, and physical fidelity. The choice of modeling technique therefore depends not only on the available computational resources but also on the specific objectives of the study, the phenomena of interest, and the level of detail required for each subsystem [21,39,43,46].

These modeling strategies also introduce additional synchronization requirements because subsystems operate with different simulation time steps. Maintaining physical consistency therefore depends on the timely exchange of information among controllers, communication systems, and power electronic converters operating across multiple temporal scales. Delayed or inconsistent data exchange may produce estimation errors, incorrect controller responses, and numerical instabilities during real-time execution [21,32,46].

As DRTS platforms integrate an increasing number of distributed computational nodes, maintaining deterministic communication becomes increasingly challenging. Communication latency, signal transmission delays, interpolation errors, and algebraic loops at co-simulation interfaces may introduce artificial damping, numerical artifacts, or dynamic responses that do not exist in the physical system, thereby compromising simulation fidelity and numerical accuracy. These effects are particularly critical in HIL applications, where even small timing deviations can significantly alter the interaction between the real-time simulator and the connected physical devices [14,34,47].

To address these challenges, several synchronization strategies have been proposed, including hierarchical synchronization, predictive delay compensation, nested modeling, and iterative information exchange between coupled subsystems. These approaches improve simulation stability even when different domains operate with non-integer simulation time-step ratios [21,32,48]. These interactions highlight that synchronization performance cannot be considered independently from the underlying communication infrastructure, motivating the analysis of communication architectures and latency constraints in the following section.

The reviewed literature indicates that multi-rate and multi-domain co-simulation have become key enabling strategies for overcoming the computational limitations of large-scale DRTS platforms. However, these approaches do not remove the underlying execution constraints; they redistribute them across computational, communication, and synchronization domains. Reducing the computational burden through temporal decomposition or subsystem partitioning increases the need to exchange information across temporal and spatial boundaries, making communication delay, synchronization accuracy, and coupling strength relevant determinants of the resulting simulation fidelity. Conversely, tighter synchronization and higher communication frequency can increase computational and communication overhead, reducing part of the scalability gained through decomposition. Consequently, computational load, communication delay, synchronization accuracy, and model fidelity should be considered as interdependent variables rather than as isolated DRTS constraints. This interaction represents a fundamental design trade-off for large-scale real-time simulation and motivates the communication and latency analysis developed in the following section [20,31,33,36,39].

3. Communication Infrastructure and Latency Challenges in DRTS

The increasing adoption of distributed simulation, HIL, PHIL, and collaborative real-time laboratories has transformed communication infrastructures into a core element of DRTS. While computational performance determines whether a model can be executed within the required simulation time step, the communication infrastructure determines whether geographically distributed subsystems remain temporally synchronized throughout execution. As DRTS platforms continue to scale beyond single-machine implementations, deterministic communication becomes as critical as computational capability, since the overall fidelity of the simulation depends on both the timely execution of numerical models and the reliable exchange of information among distributed components [13,34,36].

Unlike conventional offline simulations, where communication delays have little influence on the numerical solution, real-time environments require data to be exchanged within strict temporal constraints. Latency, jitter, packet loss, clock synchronization errors, and communication overhead can disrupt the interaction among partitioned subsystems, controllers, and physical devices, introducing numerical inconsistencies that compromise both simulation stability and HIL validation [14,34]. The impact of these effects becomes particularly evident when communication delays approach or exceed the simulation time step, reducing temporal consistency between interacting subsystems [47].

As DRTS platforms continue to evolve toward geographically distributed architectures, communication constraints become a limiting factor for scalability, potentially offsetting the computational gains achieved through model partitioning and distributed execution [13,34,36,49,50].

Therefore, communication infrastructure should not be regarded as a secondary component but as an integral part of the real-time simulation process. The selection of deterministic communication protocols, latency and synchronization management, and the integration of electrical and communication network models are essential for developing reliable DRTS platforms capable of supporting large-scale cyber–physical energy systems [51,52].

Accordingly, this chapter examines the principal communication challenges affecting modern DRTS platforms. It first discusses deterministic communication protocols used in substation and real-time environments, then analyzes the impact of latency and jitter in geographically distributed simulations, and finally reviews network co-simulation and Information and Communication Technologies (ICTs) emulation platforms used to evaluate the interaction between power systems and communication infrastructures [34,36].

3.1. Determinism in Substation Transport and Messaging Protocols

Ensuring temporal determinism in a DRTS platform depends not only on the available computational resources but also on the ability of the communication infrastructure to exchange information within strictly defined time constraints. As modern DRTS platforms increasingly rely on distributed execution, HIL, and collaborative simulation environments, communication protocols have evolved from simple data exchange mechanisms to fundamental enablers of synchronization, stability, and simulation fidelity.

At the transport layer, the selection of the communication protocol directly influences the timing and reliability of data exchange in DRTS environments. Reliability-oriented protocols such as TCP/IP provide connection management, packet verification, and retransmission mechanisms, which are useful for monitoring, data storage, and supervisory applications but may introduce variable communication delays. UDP, in contrast, reduces protocol overhead and avoids retransmission mechanisms, which can be advantageous for latency-sensitive HIL, PHIL, and distributed simulation applications. However, UDP itself does not provide deterministic packet delivery. Deterministic communication performance depends on the underlying network architecture, traffic management, synchronization mechanisms, redundancy, and other bounded-delay techniques [52].

At the Ethernet layer, Time-Sensitive Networking (TSN) provides mechanisms for bounded latency, traffic scheduling, and time synchronization that can improve the determinism of industrial communication systems. In DRTS environments, TSN can therefore complement transport and application-layer protocols by providing more predictable communication behavior at the network level [53].

In the field of substation automation, IEC 61850 has become the reference standard for interoperability in modern power systems [54]. Its architecture enables the integration of equipment from multiple manufacturers through standardized information models and communication mechanisms specifically designed for critical protection and control applications. Generic Object-Oriented Substation Event (GOOSE) messages are used to transmit binary states and trigger signals under stringent timing requirements, whereas Sampled Values (SVs) messages transport instantaneous current and voltage measurements at high update rates, facilitating the implementation of digital substations and advanced protection systems [55].

IEC 61850 Edition 2 introduced simulation functionalities specifically aimed at testing protection and automation devices [17]. The Simulation Flag allows physical devices to distinguish simulated GOOSE and Sampled Values messages from normal operational messages, enabling logical isolation of the device under test without modifying the physical communication connections [10].

The increasing integration of DERs has also driven the adoption of open protocols geared toward remote energy asset management. Standards such as IEEE 2030.5, IEEE 1815 (DNP3), and SunSpec Modbus have become increasingly important because they provide standardized mechanisms for the exchange of information between grid operators, EMS, and IBRs. In particular, IEEE 2030.5 has been widely adopted for DERMS applications due to its ability to support advanced distributed resource control, monitoring, and coordination functions using secure web service-based architectures [56,57,58].

The communication protocols discussed throughout this section fulfill different functional requirements within modern DRTS environments, ranging from deterministic real-time interaction to supervisory control and enterprise-level interoperability. Rather than representing competing alternatives, they are typically deployed in a complementary manner according to the temporal requirements, communication architecture, and operational objectives of each application.

Table 5. Comparison of Representative Communication Protocols Adopted in Modern DRTS Platforms.

Communication Protocol/Standard	Primary Application	Main Advantages	Principal Limitations
TCP/IP	Monitoring, supervisory control, databases, and asynchronous communication	Reliable data transmission through packet verification and retransmission	Variable latency and communication overhead make it unsuitable for strict real-time applications
UDP	HIL, PHIL, and distributed real-time simulation	Low protocol overhead and low communication latency	Does not guarantee packet delivery, retransmission, or bounded delay.
IEC 61850 (GOOSE/SV)	Digital substations, protection, and automation systems	High interoperability and deterministic communication for time-critical applications	Requires synchronized communication infrastructure and careful network configuration
IEEE 2030.5	Distributed Energy Resource Management Systems (DERMSs)	Standardized secure communication and advanced DER coordination	Greater implementation complexity and dependence on IP-based infrastructures
IEEE 1815 (DNP3)	SCADA systems and remote monitoring	Robust supervisory communication widely adopted by utilities	Limited suitability for fast real-time control applications.
SunSpec Modbus	Monitoring and control of inverter-based DERs	Broad industrial adoption and straightforward implementation	Limited semantic interoperability and scalability
OPC-UA	Integration with SCADA, cloud platforms, and enterprise systems	Semantic interoperability, native security, and service-oriented architecture	Higher communication overhead compared with lightweight real-time protocols

summarizes the principal communication protocols currently adopted in DRTS platforms, highlighting their representative applications together with their main advantages and operational limitations. The comparison also distinguishes transport-layer protocols from application-level communication standards, since deterministic real-time behavior depends on the interaction between these layers and the underlying network architecture.

Beyond power system communications, the convergence between electrical systems and data analysis platforms has driven the use of industrial protocols designed for the structured exchange of information, such as Open Platform Communications Unified Architecture (OPC-UA), which has emerged as one of the most widely used solutions for integrating DRTS simulators with supervision platforms, cloud services, and business management systems. Unlike traditional telemetry protocols, OPC-UA incorporates robust semantic modeling, service discovery, and native security, facilitating interoperability between different levels of a smart grid’s digital architecture [8,15,53].

However, the use of specialized protocols does not completely eliminate the constraints imposed by the communications infrastructure. Even under standards designed for critical applications, latency, congestion, and synchronization losses continue to affect the overall performance of distributed environments [51,52].

The reviewed literature indicates that no single communication protocol satisfies all the requirements imposed by modern DRTS platforms. Instead, protocol selection depends on the temporal requirements of the application, the acceptable communication latency, and the level of interoperability required between power system components and external devices. Achieving deterministic real-time communication therefore requires not only appropriate protocol selection but also communication architectures capable of preserving synchronization as the complexity of the system continues to increase. These communication constraints naturally motivate the discussion presented in the following section, which focuses on latency, jitter, and synchronization mechanisms in geographically distributed real-time simulation (GD-RTS).

3.2. Latency and Jitter Limits in Distributed Simulations (GD-RTS)

Despite having optimized communication protocols, the physical limitations of the communications network continue to govern the success of a distributed simulation. Communications problems such as latency and jitter emerge as determining factors for the stability and fidelity of the simulation [34,51].

Latency is defined as the time required for information to travel between two points within the communication infrastructure. Although small delays may be negligible in conventional monitoring applications, their impact becomes significant in HIL, PHIL, and distributed simulation environments, where interactions among subsystems must remain synchronized with the real-time simulation clock [14,47].

Communication delays introduce effects equivalent to additional dynamic elements within the simulated system. These delays may modify controller interactions, alter power references, and compromise voltage and frequency regulation, ultimately affecting numerical stability and the dynamic response of converter-dominated networks [34,52].

Although latency is an important phenomenon, its stochastic variability, known as jitter, generates greater concern. While constant delays can be corrected by techniques based on time shift or phase compensation, the random fluctuations introduced by jitter generate uncertainty in the exchange of information and progressively degrade the coherence between the different simulation subsystems. As a result, the accuracy of simulations can be compromised even when the average latency remains within acceptable limits [14,49].

The literature indicates that communication delay becomes a practical design constraint when distributed simulation and control loops operate across geographically separated infrastructures. In such environments, round-trip time (RTT), end-to-end latency, and their variability become relevant performance indicators because communication delays can directly influence the stability and temporal consistency of closed-loop experiments. Studies on geographically distributed real-time simulation and multi-site co-simulation have shown that communication performance must therefore be characterized together with the dynamic requirements of the application rather than treated as an independent networking parameter [42,54].

The growing dependence of DRTS platforms on distributed communication infrastructures has also increased the importance of cybersecurity within cyber–physical power systems. In addition to communication performance, distributed simulation environments must remain resilient against communication failures, false data injection attacks, and coordinated cyber–physical attacks capable of compromising the validity of experimental results [59]. Recent studies have highlighted the use of DRTS platforms as safe environments for evaluating resilient control strategies, cybersecurity mechanisms, and communication recovery procedures before field deployment. Hence, cybersecurity has become an emerging research direction that complements traditional assessments of communication performance in modern DRTS environments [59,60].

Network load also becomes an important factor in geographically distributed DRTS. As communication traffic increases, variations in transmission delay and data availability can become more pronounced, potentially affecting the performance of closed-loop control and DER coordination. These effects reinforce the need to evaluate communication performance under realistic traffic and operating conditions rather than relying only on nominal network latency [49,51].

The growing adoption of distributed architectures and collaborative laboratories has further increased the relevance of these challenges. The communication infrastructure in the so-called super-laboratories is no longer a simple data exchange channel but a fundamental component of the experimental platform itself [61]. Therefore, accurate characterization of latency, jitter, and packet losses is essential to ensure that the results obtained reflect the actual physical behavior of the system [35,36].

Overall, the reviewed literature demonstrates that communication performance has become a determining factor for the fidelity of geographically distributed DRTS platforms. Latency, jitter, synchronization accuracy, and network reliability directly influence the interaction between computational subsystems and physical devices, while the increasing integration of cyber–physical infrastructures introduce additional requirements related to communication resilience and cybersecurity. Accordingly, realistic assessment of modern active distribution networks requires co-simulation environments capable of representing both electrical and communication network dynamics under normal and disturbed operating conditions. This requirement naturally motivates the discussion of ICT co-simulation and network emulation platforms presented in the following section.

3.3. Network Co-Simulation and ICT Emulation Platforms

The digitalization of electricity grids has transformed communication systems into an integral component of the modern operation of distribution networks [62]. As a result, the evaluation of DRTS platforms can no longer be limited exclusively to the electrical behavior of the grid, but must simultaneously consider the impact of the Information and Communication Technology (ICT) infrastructure on the stability, control, and coordination of the DERs. This need has driven the development of co-simulation environments capable of jointly representing both electrical phenomena and the processes associated with the exchange of information.

Co-simulation integrates specialized tools within the same experimental environment, facilitating the interaction between power simulators, control systems, energy management platforms and communication models, where each simulation tool preserves its specialized capabilities while exchanging synchronized information with the remaining components of the simulation environment [18]. These environments have proven particularly useful in studies related to smart grids, storage systems, microgrids and DERs, where control decisions depend directly on the quality and availability of the information received [63,64,65].

To coordinate these environments, orchestration platforms such as HELICS, VILLASnode, and OpSim have emerged. These frameworks synchronize simulators developed by different vendors, enabling interoperability and distributed execution across multiple computing platforms [66,67,68]. These tools function as intermediate layers responsible for managing data exchange, time synchronization and translation between different communication protocols, allowing power simulators, physical devices and analysis platforms to operate together within the same simulation scenario [57,58,69].

In parallel, specialized network emulators such as OPNET, NS-3, and ExataCPS reproduce the behavior of communication infrastructures under realistic operating conditions. Unlike analytical approaches that assume constant latencies, these tools allow the reproduction of communication network phenomena such as traffic congestion, packet loss, temporal variations in delay and bandwidth limitations, thus providing a closer representation of the operating conditions found in real communication networks [70].

Network emulation platforms can further extend these approaches by introducing controlled impairments into the communication channel, such as latency variation, packet loss, and bandwidth constraints. Such capabilities allow researchers to quantify how communication disturbances propagate into electrical variables, controller responses, and DER operation under repeatable experimental conditions. KauNet is one example of this approach, providing mechanisms to reproduce controlled communication degradation scenarios [49,71].

The integration of these platforms with DRTS simulators such as OPAL-RT, RTDS or HYPERSIM has significantly expanded the experimental capabilities available to researchers and network operators [72]. Their integration enables comprehensive validation environments where electrical dynamics, communication behavior, and physical hardware can be evaluated simultaneously under controlled operating conditions [12,72,73].

Figure 4 illustrates the conceptual integration of the principal components involved in modern DRTS-based cyber–physical co-simulation environments. Rather than operating as isolated tools, power system simulators, ICT emulators, orchestration frameworks, and physical hardware are coordinated within a unified experimental platform that enables comprehensive validation of distributed energy systems under realistic operating conditions.

ICT co-simulation and emulation platforms have transformed DRTS environments from simple test tools for individual devices to true interactive cyber–physical ecosystems. This integration significantly extends the scope of DRTS from component validation toward system-level assessment of cyber–physical interactions. The ability to simultaneously represent electrical and communication phenomena enables a more realistic assessment of the challenges associated with the digitalization of distribution networks, providing a safe and controlled environment for the validation of new strategies for the operation, control and coordination of DER [35,42].

The reviewed literature demonstrates that ICT co-simulation and communication network emulation have become essential components of modern DRTS platforms rather than complementary validation tools. By integrating electrical models, communication infrastructures, orchestration frameworks, and physical hardware within a unified cyber–physical environment, these platforms enable comprehensive assessment of distributed energy systems under realistic operating conditions. DRTS has evolved beyond conventional hardware validation toward an experimental framework capable of supporting advanced applications such as digital twins, autonomous control, and intelligent energy management systems. This evolution naturally motivates the discussion presented in the following chapter, which focuses on advanced DRTS applications and their role in the next generation of active distribution networks [36,74].

4. Advanced RTS Applications: Beyond Digital Twin Representation to Phenomenological Analysis

The massive integration of DERs, storage systems, advanced electronic converters and distributed control architectures has driven the development of new DRTS-based applications. Originally conceived as platforms for the validation of equipment and the reproduction of power system dynamics in controlled environments, the evolution of computational capacity, co-simulation platforms and HIL systems has significantly expanded their scope toward increasingly sophisticated applications with direct strategic value for grid operators [15,75].

Among these emerging applications, digital twins have become one of the most representative technologies supporting the digitalization of modern power systems [76,77,78]. Their ability to maintain a synchronized virtual representation of physical assets allows for improved monitoring, diagnostics, and operational analysis. However, the practical implementation of these systems continues to face challenges associated with model conversion, interoperability between platforms, constraints derived from proprietary models, and the need to maintain continuous synchronization with dynamically evolving networks [16,79]. These limitations motivate much of the discussion presented in this chapter.

Beyond digital representation schemes, the DRTS infrastructure offers capabilities that exceed the conventional functions of a digital twin system. By executing high-fidelity models in real time, integrating physical equipment within the simulation loop, and representing complex electromagnetic phenomena, these platforms enable the exploration of extreme scenarios. In this context, real-time simulation is no longer limited to system replication but evolves into an experimental environment for phenomenological analysis, technology validation, and knowledge generation about the future behavior of power systems [8,80].

Building upon the computational and communication challenges discussed in the previous sections, this chapter examines the principal technological barriers limiting the deployment of advanced DRTS-based applications, including the challenges associated with model automation and conversion, the constraints imposed by proprietary “black-box” models, and the role of real-time simulation as a platform for exploring complex phenomena that extend beyond the conventional concept of digital twins. Rather than presenting DRTS solely as a validation technology, this chapter discusses how modern real-time simulation platforms are evolving into scientific environments for understanding, anticipating, and experimentally evaluating the emergent behavior of increasingly complex cyber–physical power systems [81,82].

4.1. The Model Conversion Bottleneck and Automation Challenges

The increasing integration of DER has driven the need to migrate from traditional offline planning and analysis models to DRTS platforms capable of dynamically representing the behavior of distribution networks. However, having a detailed electrical model does not automatically guarantee its implementation in real-time environments. In practice, converting, maintaining, and synchronizing models to preserve consistency between the physical network and its digital counterpart is one of the principal bottlenecks for the deployment of advanced applications [22].

This problem arises due to the coexistence of multiple analysis and planning tools that employ heterogeneous data formats to describe electrical infrastructure. Examples such as OpenDSS, PowerFactory, PSCAD, and various corporate systems use different structures to represent topology, electrical parameters, and control strategies. This fragmentation prevents seamless interoperability between planning, operation, and real-time simulation environments. As a consequence, network models frequently require repeated manual adaptation, increasing engineering effort and reducing the consistency of studies performed across different software platforms [30,53].

For DSOs, this challenge is particularly significant because network topology, distributed resources, and operating conditions continuously evolve. Digital models must be updated frequently to preserve consistency with the physical grid. Despite this, traditional conversion processes demand greater manual intervention, increasing implementation times while raising the risk of mathematical inconsistencies between models used for planning, operation, and real-time simulation [30,53].

In order to reduce this dependence on manual processes, the use of unified databases (UDBs) has gained considerable attention as central repositories capable of storing electrical, topological and operational attributes of the network. These platforms establish a single source of truth for multiple analysis tools, facilitating model synchronization and reducing inconsistencies between software environments. Similarly, the Common Information Model (CIM), defined by the IEC 61970 standard, has promoted standardization for the exchange of information between heterogeneous platforms, significantly improving interoperability between the different simulation environments [55,83].

In parallel, automation tools and standardized data models have been developed to reduce the manual effort required to exchange network information between planning, operation, and simulation environments. The CIM, standardized through IEC 61970, provides a common representation for exchanging electrical and operational information across heterogeneous software platforms [55]. Distributed application architectures such as GridAPPS-D further support the integration of grid models and applications through standardized interfaces [84]. However, these approaches primarily facilitate data exchange and software interoperability; they do not eliminate the need for model verification and adaptation when detailed dynamic models and control implementations are transferred to DRTS platforms.

Despite these advances, fully automated model conversion remains difficult to achieve. Translating complex dynamic models and custom controllers still requires manual verification and additional tuning before they can be executed reliably in DRTS platforms. In particular, automated conversion may not preserve all the dynamic characteristics of detailed controller implementations and IBR. As a result, engineers must often revise and validate the converted models to ensure that their dynamic response remains consistent with the original system representation [24,84].

4.2. The Black Box Barrier and Intellectual Property (IP) Restrictions

The development of high-fidelity digital representations is fundamentally constrained by the limited availability of technical information describing the internal operation of many devices deployed in modern distribution networks. This limitation is particularly evident for DERs, where manufacturers commonly restrict access to control algorithms, firmware, and detailed implementation parameters through intellectual property (IP) protection and commercial confidentiality. These restrictions are intended to safeguard proprietary technologies and prevent reverse engineering, but they also limit the information available for developing accurate simulation models. As a result, DRTS studies frequently rely on only a partial representation of the dynamic behavior of commercial devices [80,81].

These restrictions often require the use of generic or simplified models that approximate the observable behavior of commercial devices without reproducing their internal control strategies. For IBRs, such representations may rely on simplified electrical equivalents that reduce computational complexity and facilitate integration into simulation platforms, but can omit control-dependent dynamics and high-frequency interactions relevant to stability, network support, and rapid disturbances [25,38].

This limitation becomes increasingly significant in distribution networks with high penetration of IBRs, where system dynamics depend strongly on embedded digital control architectures. When key parameters governing synchronization, filtering, and control remain inaccessible, researchers and network operators have limited ability to calibrate simulation models, validate dynamic responses, or assess emerging technologies before field deployment. Consequently, uncertainty is often greatest under the operating conditions where accurate dynamic representation is most critical [28,85].

To overcome this limitation, DRTS has emerged as an alternative for the phenomenological validation of commercial devices. Unlike traditional analytical approaches, DRTS platforms allow the integration of real physical controllers using Controller Hardware-in-the-Loop (CHIL) configurations or entire equipment using PHIL schematics [86]. Under these architectures, the device interacts directly with a digital representation of the power grid, allowing its dynamic behavior to be observed without the need to access the control algorithms protected by the manufacturer [12,87,88,89].

Figure 5 illustrates how DRTS platforms overcome the practical limitations imposed by proprietary device models. Rather than requiring direct access to internal control algorithms, HIL- and PHIL-based experimental configurations enable the characterization of the observable dynamic behavior of commercial equipment under realistic operating conditions.

This approach is particularly valuable for the characterization of grid-forming technologies, storage systems, and advanced converters, where internal control algorithms often represent the main source of uncertainty. Through HIL and PHIL tests, equipment can be subjected to extreme stress conditions, frequency disturbances, voltage variations and fault scenarios, allowing the identification of functional limits, transient responses, and real grid support capabilities, thereby supporting systematic “what-if” analyses under controlled operating conditions. In this sense, the DRTS acts as an experimental tool capable of revealing behaviors that cannot be inferred from simplified models alone [6,90].

The reviewed HIL and PHIL studies differ significantly in their experimental objectives and interface requirements [91,92]. CHIL configurations are mainly used to validate control and protection algorithms with limited physical-power interaction, whereas PHIL configurations introduce additional interface, measurement, and power-amplifier dynamics that can directly affect closed-loop behavior [86,87,89]. Studies based on laboratory-scale platforms generally prioritize controller validation and interoperability, while larger experimental configurations place greater emphasis on interface stability, communication delay, and the representation of physical equipment [57,72,86,89]. These differences indicate that HIL/PHIL performance cannot be evaluated only by the simulation platform or nominal time step; the experimental objective, interface architecture, and dynamic characteristics of the hardware under test must also be considered [87,93,94].

Beyond the distinction between CHIL and PHIL, experimental fidelity is strongly influenced by the characteristics of the interface itself. In PHIL systems, the power interface, signal conditioning, sensors, analog-to-digital and digital-to-analog conversion, and communication paths introduce additional delays and bandwidth limitations that become part of the closed loop. These effects can modify the phase and gain characteristics of the interface and, under certain conditions, compromise closed-loop stability. Multirate partitioning can reduce the computational burden near the physical interface, but it may also introduce additional synchronization and delay effects that must be considered during validation [29,86,89]. Therefore, PHIL experiments require explicit assessment of interface delay, power-amplifier dynamics, measurement uncertainty, signal bandwidth, tracking error, and closed-loop stability. CHIL configurations generally avoid power-amplifier and physical-power interface dynamics, making them particularly suitable for controller and protection validation, whereas PHIL provides a more complete representation of the interaction between simulated networks and physical equipment at the cost of additional uncertainty and stability constraints.

The importance of experimentally characterizing commercial devices has been increasingly recognized by recent grid integration standards. For example, IEEE Std. 2800 emphasizes the need to validate the dynamic performance of IBR before their large-scale integration into the power system, particularly when system stability depends on advanced control functions implemented by equipment manufacturers [95]. Within this context, DRTS platforms provide one of the most suitable experimental frameworks for assessing the behavior of commercial devices under realistic operating conditions without requiring direct access to their proprietary control algorithms [85,96].

The persistence of proprietary control models suggests that the black-box barrier should not be interpreted only as a limitation to model development. Instead, it also motivates a shift toward experimental characterization, in which DRTS is used to evaluate the observable dynamic behavior of commercial equipment without requiring access to its internal control architecture. This perspective reinforces the evolution of DRTS from a model-based validation tool toward a phenomenological experimentation platform [16,75,79].

Ultimately, the widespread adoption of proprietary black-box models illustrates that the principal barriers to industrial DRTS deployment are no longer exclusively computational. Instead, they increasingly arise from the tension between intellectual property protection, interoperability, scientific reproducibility, and long-term model sustainability.

Table 6 summarizes representative studies that exemplify the current constraints in simulation fidelity, communication latency, and digital twin maturity levels.

Table 6. Comparative Analysis of Representative Digital Twin Implementations in DRTS-Based Studies.

Review	Application Objective	DRTS Platform	Model Scale Domain	Time Step (µs)	HIL/PHIL Type	Comm. Protocol & Latency	Maturity Level	Validation Metric	Deployment Setting
Han [14]	Prototyping for DERs	RTDS	Distribution (EMT)	50 μs	CHIL	TCP/IP-Modbus	R2 (Monitoring)	Tracking accuracy	Laboratory CHIL testbed
Nguyen [82]	Renewable Resource assessment	OPAL-RT	Microgrid (EMT)	100 μs	PHIL	IEC 61850	R2 (Monitoring)	DRES response system agreement	Laboratory PHIL testbed
Hueros-Barrios [76]	PV-PEM-BESS	OPAL-RT	Multi-domain	100 μs	PHIL	Modbus/UDP	R2 (Monitoring)	Anomaly detection	Laboratory PHIL/HIL
Zeynivand et al. [96]	Anomalies Industrial Failure Analysis	Not Specified	Industrial (RMS)	>1000 μs	None (Digital)	OPC-UA	R2 (Monitoring)	Prediction accuracy	Industrial case study/real machine data
Menga et al. [42]	Insolated Microgrid Control	Geographically Distributed	Microgrid (EMT)	50 μs	CHIL	RTT < 5 ms	R2 (Monitoring)	Control performance	Distributed laboratory CHIL testbed
Hoke et al. [39]	Islanding Detection	Not Specified	Multi-inverter (EMT)	<100 μs	PHIL	Not Specified	R1 (Representation)	Islanding detection	Laboratory PHIL testbed
Chang & Vanfretti [89]	Smart Inverter/DERMS	Not Specified	Grid-scale (EMT)	50 μs	PHIL	IEC 61850	R2 (Monitoring)	Inverter response	Laboratory PHIL
Archetti [93]	Microgrid Storage control	RTDS	Microgrid (EMT)	100 μs	CHIL	Not Specified	R2 (Monitoring)	Control decision performance and computational time	Laboratory CHIL testbed

Table 6. Comparative Analysis of Representative Digital Twin Implementations in DRTS-Based Studies.

Review	Application Objective	DRTS Platform	Model Scale Domain	Time Step (µs)	HIL/PHIL Type	Comm. Protocol & Latency	Maturity Level	Validation Metric	Deployment Setting
Han [14]	Prototyping for DERs	RTDS	Distribution (EMT)	50 μs	CHIL	TCP/IP-Modbus	R2 (Monitoring)	Tracking accuracy	Laboratory CHIL testbed
Nguyen [82]	Renewable Resource assessment	OPAL-RT	Microgrid (EMT)	100 μs	PHIL	IEC 61850	R2 (Monitoring)	DRES response system agreement	Laboratory PHIL testbed
Hueros-Barrios [76]	PV-PEM-BESS	OPAL-RT	Multi-domain	100 μs	PHIL	Modbus/UDP	R2 (Monitoring)	Anomaly detection	Laboratory PHIL/HIL
Zeynivand et al. [96]	Anomalies Industrial Failure Analysis	Not Specified	Industrial (RMS)	>1000 μs	None (Digital)	OPC-UA	R2 (Monitoring)	Prediction accuracy	Industrial case study/real machine data
Menga et al. [42]	Insolated Microgrid Control	Geographically Distributed	Microgrid (EMT)	50 μs	CHIL	RTT < 5 ms	R2 (Monitoring)	Control performance	Distributed laboratory CHIL testbed
Hoke et al. [39]	Islanding Detection	Not Specified	Multi-inverter (EMT)	<100 μs	PHIL	Not Specified	R1 (Representation)	Islanding detection	Laboratory PHIL testbed
Chang & Vanfretti [89]	Smart Inverter/DERMS	Not Specified	Grid-scale (EMT)	50 μs	PHIL	IEC 61850	R2 (Monitoring)	Inverter response	Laboratory PHIL
Archetti [93]	Microgrid Storage control	RTDS	Microgrid (EMT)	100 μs	CHIL	Not Specified	R2 (Monitoring)	Control decision performance and computational time	Laboratory CHIL testbed

4.3. Beyond Digital Twin Replication: Toward Relational and Autonomous Systems

The adoption of digital twins has transformed the way power system operators interact with physical assets by facilitating monitoring, supervision, and diagnostic tasks. Nevertheless, limiting the role of DRTS to supporting digital twin implementations provides only a partial perspective of its actual capabilities. As DRTS platforms continue to evolve, their contribution increasingly extends beyond synchronized asset representation toward the experimental analysis of complex cyber–physical phenomena [8,75].

Although digital twins have attracted considerable attention within the power systems community, the representative studies examined in this review are concentrated at relatively early stages of maturity. In the evidence matrix, most of the selected implementations correspond to R2, characterized by synchronization and monitoring capabilities with limited autonomous intervention, while only a small number exhibit characteristics associated with higher levels of interaction or autonomy. This pattern is consistent with broader discussions in the digital-twin literature, which emphasize that the distinction between a synchronized digital representation and a fully operational digital twin remains important [75,79,97].

The limitations discussed throughout this review collectively explain why this transition remains difficult. Model conversion bottlenecks, interoperability constraints, proprietary (“black-box”) devices, computational requirements, and synchronization challenges collectively restrict the development of digital representations capable of evolving together with increasingly dynamic electrical networks. As a result, many digital twins continue to operate primarily as monitoring and visualization tools rather than as autonomous decision-support systems [80,82].

Within this context, DRTS provides capabilities that extend beyond the conventional concept of a digital twin. Unlike digital twins, which are intrinsically linked to continuous synchronization with a specific physical asset, DRTS platforms create independent experimental environments where complex electrical phenomena, advanced control strategies, and operating conditions can be investigated without affecting the physical network. This flexibility enables systematic exploration of scenarios that would be impractical, unsafe, or economically infeasible to reproduce under real operating conditions [15].

One of the principal strengths of this approach lies in the possibility of executing high-fidelity electromagnetic transient (EMT) simulations capable of capturing fast converter dynamics, low-inertia phenomena, and complex interactions between controllers. Rather than simply reproducing system behavior, these simulations allow researchers to investigate the mechanisms governing system response, identify emergent dynamic phenomena, and evaluate interactions that are difficult to observe experimentally. In this sense, DRTS evolves from a validation technology toward a platform for phenomenological analysis of modern cyber–physical power systems [6].

Another promising application involves the generation of high-fidelity synthetic datasets for AI. Since many critical operating conditions occur infrequently in real networks, historical operational records are often insufficient for training advanced machine learning algorithms [98]. DRTS enables the controlled reproduction of rare operating scenarios, providing representative datasets for applications such as fault diagnosis, predictive maintenance, autonomous control, and intelligent energy management. These capabilities may accelerate the evolution of digital twins toward higher levels of autonomy while supporting the development of next-generation intelligent power systems [16,99,100].

The future impact of DRTS will not depend exclusively on faster processors or lower communication latency. Its long-term impact will be determined by the ability to automate model management, overcome interoperability barriers, integrate AI, and exploit real-time simulation as a scientific framework for understanding increasingly complex cyber–physical phenomena. From this perspective, DRTS should no longer be viewed solely as a technology for digital twin implementation or hardware validation, but as an experimental platform where representation, learning, and scientific discovery converge to support the next generation of intelligent power systems. [8,79,82,101].

5. Discussion

5.1. The Paradox of Simplification

One of the most consistent observations emerging from the reviewed literature is the growing mismatch between the increasing complexity of modern distribution networks and the computational capabilities required to simulate them in real time [1,13]. While transmission system studies have traditionally relied on reduced-order representations without significantly compromising their objectives, active distribution networks demand high-fidelity EMT models capable of capturing the fast dynamics introduced by DERs, IBRs, and advanced control systems. This evolution places DRTS platforms in a position where model fidelity and real-time execution become competing objectives rather than complementary design goals [13,20,22,25].

The widespread use of Thevenin and Norton equivalents illustrates this paradox. Although these reductions remain essential for maintaining computational feasibility, they inevitably neglect sub-transient dynamics, converter interactions, and high-frequency electromagnetic phenomena that have become increasingly relevant in converter-dominated distribution systems [24,25,38]. In practice, the simplification techniques that make large-scale DRTS simulations computationally viable are often the same techniques that reduce their ability to reproduce the phenomena motivating such simulations. This trade-off emerges as one of the principal methodological challenges identified throughout this review [1,22,46,102].

5.2. The Grid-Forming Dilemma: Network Support Versus Converter Limits

Grid-Forming Inverters (GFM Inverters) are widely regarded as one of the most promising technologies to support the transition to electricity systems with a high penetration of IBR [85]. The literature usually presents these devices as an alternative capable of providing functions traditionally associated with synchronous machines, including frequency support, voltage regulation, grid-forming capability, and even mechanisms equivalent to rotational inertia. From this perspective, GFMs are projected as fundamental elements to maintain the stability of future electricity systems characterized by a progressive reduction in conventional generation and a decrease in the natural inertia available in the grid.

However, there is a fundamental difference between the physical nature of a synchronous machine and that of an electronic power converter [103]. While the synchronous machine stores kinetic energy inherently in its rotating masses and has a high capacity to withstand overcurrent during transient events, GFM inverters are completely dependent on the ability of their semiconductors, storage systems, and control strategies to emulate these behaviors. Thus, many of the functions attributed to GFMs are not intrinsic physical properties, but responses generated by control algorithms that operate within strictly defined energy and thermal limits [20,33].

This limitation becomes particularly evident during severe grid events such as faults, short circuits, or large frequency deviations, when grid-forming inverters may rapidly reach their current limits. Once these thresholds are exceeded, the converter protection mechanisms constrain the output current, reducing the capability of the inverter to sustain the grid voltage. Under these conditions, the converter progressively shifts from behaving as an ideal voltage source to operating more like a controlled current source. This transition alters the dynamic response expected by grid-forming control algorithms and may compromise system stability in ways that are not yet fully captured by many of the simplified models currently adopted in the literature [1,42,104,105,106].

This reveals the central dilemma surrounding grid-forming technology. On the one hand, grid operators expect GFM inverters to provide advanced support services, such as synthetic inertia, oscillation damping, frequency support, and short-circuit current contribution. On the other hand, each of these functions increases the energy and thermal demands placed on the converters, reducing the available margin for active power delivery and speeding up the approach to the physical limits of the equipment. From this perspective, the flexibility promised by GFMs is not unlimited, but depends directly on the physical limits of the converter, the available energy storage, and the implemented control strategy [36,85].

The discussion becomes even more relevant considering that many studies reported in the literature are conducted under ideal simulation conditions or rely on simplified models that do not adequately represent saturation, current limitation, thermal constraints, or storage system degradation. As a result, the actual capability of grid-forming inverters to replace functions historically provided by synchronous machines may be overestimated [6]. This suggests that performance assessments based exclusively on idealized simulation conditions may not fully reflect the operational limitations of future converter-dominated power systems [6,8,85].

The key question is no longer whether grid-forming inverters can emulate synchronous machines under nominal operating conditions, but rather under which physical, energetic, and operational limits this equivalence remains valid. Answering this question is essential for defining the role that GFMs will play in future low-inertia power systems and for determining whether they should be regarded as direct replacements for conventional synchronous generation or as complementary resources within hybrid grid-support architectures [20,28,33,107].

5.3. Stagnation in R2: Why Do Digital Twins Fail to Reach the Next Level?

For the purposes of this review, Digital Twin maturity is considered according to the degree of interaction established between the physical asset and its digital representation. This distinction is important because the term “Digital Twin” is not used consistently across the literature. A recent systematic analysis of 358 Digital Twin definitions found that 33.52% of the examined definitions described a digital model rather than a Digital Twin, while the definitions identified as conceptually complete incorporated a physical entity, its virtual representation, and a dynamic continuity of bidirectional data and information exchange [108]. Accordingly, this review distinguishes between lower and intermediate maturity levels, where the digital representation primarily supports monitoring, visualization, diagnostics, and synchronization, and higher maturity levels, which require prediction, bidirectional interaction, adaptive control, and autonomous decision-making [75,79].

One of the clearest findings emerging from the reviewed literature is the gap between the capabilities commonly attributed to digital twins and the maturity demonstrated by current implementations. Representative studies illustrate this progression differently. Han et al. [14] developed a real-time HIL platform for Digital Twins of DERs, while Nguyen et al. [82] integrated a digital twin with PHIL for the assessment of distributed renewable resources. More recent work has demonstrated real-time Digital Twin implementations with physical interaction, such as the PV–PEM–BESS prototype reported by Hueros-Barrios et al. [76] and industrial energy-efficiency and failure-analysis applications such as that presented by Zeynivand et al. [96]. These studies demonstrate increasingly close interaction between digital and physical systems, while their reported capabilities remain concentrated on monitoring, synchronization, validation, and experimental assessment [14,76,82,96].

This situation raises a relevant question: if the potential benefits of digital twins are widely recognized, why do most implementations fail to evolve towards higher levels of autonomy? A common explanation attributes this limitation to the black-box nature of many commercial devices and the associated intellectual property constraints that prevent access to inverter-based internal control algorithms for many resources, making it difficult to build high-fidelity models and limiting the digital twin’s ability to perform advanced dynamic analysis. However, reducing the problem exclusively to the lack of transparency of manufacturers simplifies a much more complex problem [80].

The evidence matrix provides a representative indication of this maturity gap. Among the eight implementations examined in the comparative table, seven are classified as R2, while one remains at R1; none of the selected cases reaches R3 or R4. Although this sample is not intended to represent the entire Digital Twin literature statistically, the concentration of the reviewed implementations at R1–R2 provides evidence that current applications are still predominantly oriented toward representation, monitoring, synchronization, and validation rather than sustained autonomous operation. This pattern supports the need to examine the underlying technical constraints that prevent progression toward higher maturity levels.

This distinction becomes more pronounced at higher maturity levels, where a Digital Twin is expected to maintain continuous bidirectional interaction with the physical system and support prediction, adaptive control, or autonomous decision-making. Achieving these capabilities requires the coordinated operation of model fidelity, data availability, communication reliability, computational capacity, and temporal synchronization. These requirements are tightly coupled, meaning that advances in one dimension cannot compensate for fundamental limitations in the others [19,71]. Achieving this level requires not only an accurate representation of the physical asset, but also the ability to understand complex dynamic phenomena, anticipate future operating conditions, and execute control actions in a reliable and secure manner. Under these conditions, model fidelity, information availability, communication robustness, computational capacity, and temporal synchronization become interdependent requirements that ultimately determine the feasibility of advanced digital twin architectures. In this context, DRTS provides capabilities that conventional offline simulation cannot provide by itself. Its deterministic execution enables the evaluation of models under strict temporal constraints, while HIL and PHIL configurations allow continuous interaction with physical controllers and equipment. In addition, DRTS can reproduce communication delays and control-loop dynamics under real-time conditions, making it possible to evaluate how computational, communication, and physical-interface constraints affect the behavior of a digital twin. These capabilities are particularly relevant when the digital twin must be validated under closed-loop operating conditions rather than used only as an offline analytical representation [11,12,13].

The evidence discussed throughout this review suggests that the principal obstacle preventing digital twins from evolving beyond the R2 maturity level is not the absence of more sophisticated AI algorithms, but the difficulty of sustaining an integrated cyber–physical infrastructure capable of simultaneously preserving model fidelity, temporal determinism, interoperability, continuous synchronization, and computational scalability [80,88]. Advancing toward higher levels of autonomy will depend less on incremental improvements in individual technologies than on the ability to integrate these capabilities within a unified and continuously evolving real-time ecosystem [19,81,82].

5.4. The Adoption Gap: Why Is DRTS Still a Primarily Academic Infrastructure?

One of the most striking observations emerging from this review is the contrast between the remarkable capabilities demonstrated by DRTS platforms and their still limited adoption in industrial practice. The ability to validate Grid-Forming resources, integrate physical devices through HIL and PHIL, support advanced Digital Twins, and generate synthetic datasets for AI demonstrates the breadth of applications that DRTS can support in future active distribution networks [71]. This reveals a clear discrepancy between the high number of applications proposed in the literature and the relatively limited level of adoption observed in industrial settings [76,77].

A first explanation lies in the inherent complexity of DRTS infrastructures. Practical implementations require specialized real-time computing hardware, parallel or multicore execution, low-latency communication interfaces, and personnel capable of developing, validating, and continuously maintaining the associated models [85]. Laboratory studies demonstrate that these infrastructures can support increasingly complex HIL and PHIL experiments, but they also show that interface configuration, synchronization, model preparation, and hardware integration become additional engineering tasks as the experimental scope increases [72,87,89,93]. These requirements become more demanding when DRTS is expected to support multiple devices, heterogeneous software environments, or continuously changing DER configurations. Consequently, the practical challenge is not only the initial acquisition of the simulator, but the long-term effort required to integrate, validate, update, and maintain the complete real-time experimentation environment.

In addition, the rapid evolution of DER creates a persistent gap between model development and the continuous introduction of new commercial devices. Network operators must not only update electrical models, but also validate proprietary control strategies, manage intellectual property constraints, and ensure compatibility among equipment from different manufacturers [16,81]. As a result, maintaining a functional DRTS infrastructure continues to require highly specialized multidisciplinary expertise. This dependence on specialized multidisciplinary expertise can limit large-scale industrial deployment, particularly in organizations without dedicated teams for advanced real-time simulation environments [19]. This dependence on specialized multidisciplinary expertise represents an important distinction between demonstrating a DRTS application in a controlled laboratory environment and sustaining it as an operational infrastructure within a utility or industrial organization.

Advanced applications such as autonomous digital twins, large-scale DER coordination, grid-forming validation, and AI-based energy management require the simultaneous integration of high-fidelity models, physical interfaces, real-time communications, and large volumes of operational data. The challenge is therefore no longer the availability of individual technologies, but their integration into economically sustainable and operationally scalable solutions [75,76].

On the other hand, the electricity industry has historically prioritized reliability and risk mitigation criteria over rapid technological adoption, where the incorporation of DRTS infrastructures not only competes against technical limitations, but also against consolidated organizational processes, conservative regulatory frameworks, and planning methodologies that have proven effective for decades [15,75]. As a result, the speed of industrial adoption is usually considerably lower than the speed of innovation observed in the scientific literature [80].

The evidence matrix reinforces this gap between experimental maturity and operational deployment. Most of the representative studies reviewed were validated in laboratory-based CHIL or PHIL environments, where communication behavior, controller performance, and hardware interaction could be evaluated under controlled conditions. Only a limited number of cases involved industrial equipment or data from real operational assets, and these studies generally remained within experimental or validation-oriented settings rather than constituting sustained deployment in utility operations. This distribution of evidence suggests that the main barrier is not the absence of technically mature DRTS applications, but the limited transition from controlled experimental environments toward operationally integrated infrastructures.

The evidence reviewed throughout this paper suggests that the principal barrier limiting the widespread adoption of DRTS is no longer computational performance or simulation fidelity alone, but the challenge of transforming a scientifically mature technology into an economically viable, interoperable, and operationally sustainable industrial infrastructure [80,82]. Closing this gap will require coordinated advances in standardization, automation, interoperability, workforce development, and regulatory acceptance. From this perspective, the future success of DRTS will be determined not only by continued technological innovation, but also by its ability to become part of the operational culture of modern power systems rather than remaining primarily a research infrastructure [71,76,109].

6. Conclusions

This review shows that the role of DRTS has evolved far beyond its original function as a platform for equipment validation. The increasing digitalization of distribution networks, together with the rapid integration of DERs, inverter-based resources, storage systems, and advanced control architectures, has transformed DRTS into a cyber–physical experimentation environment capable of supporting controller validation, communication assessment, interoperability studies, and the analysis of increasingly complex power system dynamics. As power systems continue to evolve toward converter-dominated architectures, maintaining both high-fidelity electromagnetic representation and real-time determinism becomes one of the defining challenges for future DRTS applications.

The reviewed literature consistently indicates that computational scalability remains one of the principal limitations of modern DRTS platforms. The need to represent detailed EMT models while preserving strict temporal determinism creates an inherent trade-off between model fidelity and computational feasibility. Rather than being solved exclusively through faster processors, this challenge increasingly depends on efficient partitioning strategies, multicore execution, multi-rate simulation, communication synchronization, and distributed computational architectures. Future advances will rely on the coordinated optimization of computational resources rather than isolated hardware improvements.

The growing dependence of modern distribution networks on power-electronics-based devices has exposed the limitations of simplified mathematical models and proprietary (“black-box”) commercial equipment. In this context, HIL and PHIL methodologies implemented on DRTS platforms have become essential experimental tools for validating Grid-Forming converters, storage systems, advanced controllers, and other inverter-based technologies under realistic operating conditions. The review also highlights that the future role of Grid-Forming inverters should not be evaluated solely based on their capability to emulate synchronous machines, but also considering the physical, thermal, and operational constraints that ultimately determine their performance in low-inertia power systems.

The representative Digital Twin implementations examined in this review are concentrated at intermediate maturity levels, with the comparative evidence matrix showing a predominance of R2 cases. The transition toward higher autonomy is constrained not only by AI capabilities, but by the combined requirements of model conversion, interoperability, temporal synchronization, computational scalability, communication robustness, and continuous maintenance of high-fidelity digital representations.

Model conversion automation and cross-platform interoperability constitute critical enablers for the evolution of advanced DRTS-based applications. The use of UDB, exchange standards such as CIM, and automatic translation tools can reduce errors, improve consistency between models, and keep the digital representation synchronized with the changing topology of the physical network.

The principal challenge identified throughout this review is no longer demonstrating the technical capabilities of DRTS, but enabling its widespread adoption within industrial environments. Despite the remarkable progress achieved in research laboratories, large-scale deployment continues to be constrained by implementation costs, interoperability issues, workforce specialization, and the complexity of maintaining integrated real-time simulation environments. Ultimately, the future impact of DRTS will depend less on incremental improvements in computational performance than on its successful integration into interoperable, automated, and economically sustainable operational environments. Under this perspective, DRTS is evolving beyond a validation platform or a digital twin enabler toward a scientific infrastructure where real-time simulation, cyber–physical experimentation, and phenomenological analysis converge to support the next generation of intelligent and resilient power systems.