Why the Grid Needs Co-Simulation: The Time-Scale Problem

Co-simulation of power electronics and grid models is necessary because the switching dynamics of modern power converters operate at time scales orders of magnitude faster than the electromechanical transients of the bulk transmission network — and no single simulator can faithfully represent both at once. As the density of converter-interfaced renewable sources increases, this incompatibility becomes the central engineering constraint for any credible stability study.

Research from Qingdao University of Science and Technology (2021) places model segmentation at the core of cross-platform co-simulation, proposing an improved joint-simulation strategy for large-scale flexible direct-power grids fed by wind farms using RT-lab and Hypersim platforms. The study demonstrates that multiple interface algorithms are required because a single interface cannot address all data interaction issues when large- and small-step subsystems must exchange data simultaneously.

The convergence behavior of co-simulation methods for combined transmission and distribution systems has been rigorously characterized by Iowa State University (2019), which found that parallel and series computation modes exhibit distinct numerical stability behaviors and that the integration time step exerts a decisive influence on convergence. This is particularly consequential for renewable-rich systems, where converter-interfaced distributed generation couples the two subsystems across orders-of-magnitude-different time scales.

The Korea Institute of Energy Research (2016) formalised this architecture into a general co-simulation framework that couples an electromagnetic transient (EMT) simulator — representing the detailed study area with converter models — with an equivalent external system, and introduces a delay compensation algorithm to handle the latency introduced by inter-simulator communication. According to IEEE standards for power system simulation, managing inter-simulator latency is one of the defining engineering challenges in real-time co-simulation platforms.

For hybrid thermal-electrical distribution grids, Vienna University of Technology (2020) established that co-simulation is the only design approach capable of simultaneously capturing control dynamics and physical network behaviour, enabling design optimisation with explicit control consideration — a capability unavailable in conventional single-domain design tools.

Eigenvalue, Impedance, and Dynamic Phasor Methods for Converter-Rich Grids

Three complementary analytical methods have emerged as the backbone of stability assessment for grids with high converter penetration: linearised eigenvalue analysis for small-signal modal interactions, impedance-based methods for resonance identification in multi-infeed systems, and dynamic phasor calculus as a replacement for quasi-static assumptions that break down at high renewable penetration.

Small-Signal and Eigenvalue Analysis

The Hong Kong Polytechnic University Shenzhen Research Institute (2021) established equivalent dynamic models for full converter-based wind and photovoltaic generation and applied linearised state-space modelling to investigate modal interaction and converter-driven stability issues in an IEEE 16-machine 68-bus system. The study reveals that modal resonances under specific operating conditions can deteriorate dynamic performance or threaten overall system stability.

State Grid Jiangsu Electric Power (2021) extended eigenvalue methods to grids with energy storage devices by constructing the state-space model using a discrete time domain model (DTDM), which linearises nonlinear component characteristics and reduces computational burden in large-scale converter-dense power grids. This offers a practical pathway for incorporating power electronics dynamics into grid-wide stability assessments without prohibitive simulation cost.

Technische Universität Berlin (2021) directly challenged the adequacy of classical quasi-static phasor calculus — which underpins most legacy stability tools — demonstrating that as converter penetration increases, the assumption that fast transients decay rapidly becomes invalid. The paper proposes dynamic phasor calculus as an alternative and develops a methodology including frequency response, modal, and sensitivity analyses to compare the two frameworks systematically. According to guidance from IEC, the transition from quasi-static to dynamic phasor representations is now a recognised requirement for grid code compliance studies in high-penetration renewable markets.

“As converter penetration increases, the assumption that fast transients decay rapidly becomes invalid — making dynamic phasor calculus a necessary replacement for quasi-static tools in stability assessment of integrated power electric and electronic systems.”

Impedance-Based Stability Assessment

Universitat Politécnica de Catalunya (2021) addressed the challenge of applying traditional stability criteria to large, complex power systems with high converter penetration, proposing a simplified admittance matrix-based criterion that remains tractable for realistic system sizes. For HVDC-connected grids, the University of Kassel (2022) extended nodal admittance matrix resonance mode analysis to include both AC and DC dynamics via a hybrid AC-DC admittance matrix — modelling HVDC interfacing converters as three-port admittance networks — achieving evaluation accuracy that single-side approaches cannot match.

Hardware-in-the-Loop and Digital-Physical Hybrid Simulation

Hardware-in-the-loop (HIL) co-simulation bridges pure software simulation and physical testing, enabling validation of control algorithms and power electronics components under realistic grid conditions that software models alone cannot replicate. AIT Austrian Institute of Technology (2018) introduced a HIL co-simulation validation framework combining large-scale power network simulation with real-world control components, demonstrating — through a voltage control example — that HIL co-simulation can validate advanced control concepts for smart grids that cannot be adequately tested through purely software-based methods.

For MMC-HVDC systems — critical for long-distance offshore renewable energy delivery — Northeast Electric Power University (2021) proposed a power interface algorithm based on damping impedance to improve the stability of DC power grid hybrid platforms. The paper derives matching principles between damping impedance at the power interface and equivalent impedance of the physical simulation system, and addresses interface delay through DC voltage compensation. Wuhan University (2018) complemented this by improving the ideal transformer method (ITM) for digital-physical hybrid simulation of MMC-HVDC, adding a virtual resistance current to the controlled current source signal — with the virtual resistance value optimised to balance stability and simulation precision.

Distributed real-time simulation (D-RTS) extends these concepts to large-scale systems. Hatch Ltd. (2022) proposed a general methodology for developing power system models suitable for distributed real-time simulations based on topology, simulator interfaces, and data exchange, testing it on the IEEE Australian Benchmark model and the IEEE 300-Bus system. The work demonstrates that splitting monolithic models into distributed sub-models introduces synchronisation and accuracy challenges that must be systematically characterised — an important finding for any organisation deploying co-simulation at national grid scale. Bodies such as ENTSO-E have highlighted distributed real-time simulation as a priority capability for pan-European grid planning with high renewable penetration.

From Offshore Wind to Multi-Energy Systems: Co-Simulation in Practice

Co-simulation frameworks have been deployed across a widening range of application domains — from offshore HVDC-connected wind farms and large-scale transmission planning to building-integrated distributed energy resources and multi-energy carrier networks. Each domain introduces distinct stability challenges that motivate specific co-simulation architectures.

Offshore Wind and HVDC Integration

Cardiff University (2017) used impedance-based representation of an offshore wind power plant to analyse the interaction between offshore HVDC converter control and electrical resonances, adapting the positive-net-damping criterion for this purpose. At the planning scale, Case Western Reserve University (2020) applied statistical steady-state stability analysis to a 1,000-MW offshore wind power plant integrated into a 63,000-bus model of the U.S. Eastern Interconnection — one of the largest-scale stability studies in the reviewed corpus. This scale of analysis is only tractable through co-simulation approaches that decouple the detailed offshore plant model from the bulk network representation.

Delft University of Technology (2021) applied co-simulation directly to transient stability evaluation using the MOSAIK framework and the Functional Mock-up Interface (FMI) standard, evaluating grid-forming converters modulating wind power plants. The study highlights that grid-forming converters are a promising solution for weak-grid conditions and islanded operation, and that co-simulation is essential for a wide range of transient stability studies required for grid code compliance verification. Standards bodies including IEC and IEEE have both published technical reports on grid-forming converter requirements that reference co-simulation as the recommended validation methodology.

Smart Grids, Buildings, and Multi-Energy Co-Simulation

The University of Amsterdam (2022) provided a scoping review of co-simulation research in smart grids, identifying that coupling of simulators is the standard approach for large-scale smart grid testing and mapping research areas, simulator coupling patterns, and future directions. At the distribution and building level, the University of Kentucky (2023) demonstrated a co-simulation testbed for electric power distribution systems and distributed energy resources — including behind-the-meter solar PV and HVAC systems — with machine learning-enabled ultra-fast HVAC models accelerating simulation by up to 133 times.

Multi-Time-Scale Simulation for Distribution Circuits

Chulalongkorn University (2017) developed a distribution circuit simulation tool in MATLAB providing three separate models for each equipment type — steady-state, electromechanical transient, and electromagnetic transient — enabling long-term simulations that cover wind speed and solar irradiance ramps as well as fault events. The coordination of multi-type generation sources — wind, solar, hydro, and thermal — within a unified simulation framework is addressed by State Grid Corporation of China Northwest Branch (2018, active patent), which defines objective functions and relationship matrices between multiple generation base clusters and identifies grid security thresholds for dispatching control under fluctuating clean energy penetration.

Key Institutions and the Shift Toward Data-Driven Co-Simulation

The research landscape for co-simulation of power electronics and grid models is dominated by a set of institutions whose contributions span the full methodological spectrum — from cross-platform interface algorithms to industrial model-based design workflows — and a clear directional shift toward data-driven and machine-learning-augmented approaches is evident across all of them.

Chinese research institutions and grid operators — including State Grid Corporation affiliates, Northeast Electric Power University, Wuhan University, Tsinghua University, and North China Electric Power University — form the most numerically dominant cluster in the dataset, driving innovation in hybrid simulation platforms, MMC-HVDC co-simulation, eigenvalue-based stability methods, and coordinated multi-energy simulation. Tsinghua University (2023) exemplifies the frontier direction with a feasibility study of Neural ODE and DAE modules for data-driven power system dynamic component modelling, which reduces the computational burden associated with large-scale converter-dense power grids.

Spanish polytechnic universities — particularly Universitat Politécnica de Catalunya and Universidad Politécnica de Cartagena — are prolific contributors to impedance-based stability analysis for VSC-HVDC and multi-infeed converter systems. Dutch and German institutions, including Delft University of Technology, Karlsruhe Institute of Technology, Fraunhofer IWES, and Leibniz Universität Hannover, are particularly active in co-simulation framework development and renewable energy system integration modelling. As tracked by WIPO‘s Global Innovation Index, Europe and China together account for the majority of power electronics patent filings in grid integration technologies.

Vestas Wind Systems (Denmark) represents the industrial perspective, appearing in the dataset for work on large-scale grid integration with a double synchronous controller (2019) and a model-based design approach for stability assessment, control tuning, and verification in off-grid hybrid power plants (2019). US national laboratories and universities — NREL, Iowa State, Case Western Reserve, University of Kentucky, Carnegie Mellon, and Virginia Tech — contribute frameworks for combined transmission and distribution co-simulation, statistical stability analysis for large-scale offshore wind integration, and co-simulation of buildings with grid systems.

The overarching innovation trend across all these players is the movement from single-domain, single-time-scale simulation toward integrated co-simulation platforms that couple EMT-level converter models with phasor-domain grid models, and increasingly incorporate data-driven components — neural ODE/DAE modules and machine learning-based surrogate models — to manage computational burden at scale. This trajectory is consistent with the broader push in power systems research toward digital twin architectures that can support real-time grid operation decisions, not just offline planning studies. The IEA‘s Electricity Security report identifies advanced simulation and digital twin capability as a critical enabler for grids targeting 80% or higher renewable penetration by 2035.