From Device to Grid: Using Simulink with FMI for HVDC Cosimulation

As the transition to renewable energy accelerates, demand is increasing for high-voltage direct current (HVDC) transmission systems to interconnect AC grids and efficiently transfer power with lower losses over long distances. The HVDC-WISE project, a European Union–funded collaboration involving 14 partners from 10 countries, aims to overcome key challenges in the areas of system reliability and resilience for hybrid AC/DC transmission architectures. Engineers and researchers working on the project are focused on identifying HVDC-based grid configurations that address these challenges while facilitating the integration of renewable energy sources.

A persistent challenge in HVDC system development and deployment lies in the disparate toolchains used across the two principal engineering domains involved. Power electronics engineers at equipment manufacturers, as well as power systems engineers at grid operators and research institutions, increasingly rely on MATLAB^® and Simulink^® to model, simulate, and validate a wide range of components and systems, from converters and energy storage to complete grid-connected configurations.

However, interoperability challenges often arise when subsystem models developed in one environment must be used within another toolchain, for example, during certification or integration studies where the reference grid or benchmark network is modeled in a non-MATLAB and Simulink environment. This need to translate or replicate models across tools is typically time-consuming and error-prone.

In parallel, intellectual property (IP) protection remains a key concern. Manufacturers are frequently reluctant to share detailed implementation models when collaborating with grid operators or certification bodies. To address both the interoperability and IP issues, workflows based on a Functional Mock-up Interface (FMI) and parametrizable model exchange are increasingly adopted. These workflows enable users to export Simulink models as Functional Mock-up Units (FMUs)—a standardized package containing the model’s compiled code but not the original viewable and editable model. An FMU preserves fidelity while allowing secure integration into external simulation environments or, when required, provides black-box models that maintain confidentiality without sacrificing interoperability.

For the project needs, our teams at the University of Strathclyde and SuperGrid Institute, both HVDC-WISE partners, have developed a new workflow that addresses both problems using the FMI standard to unify cross-domain cosimulation and model exchange. In the workflow, power electronics engineers can model, simulate, and validate their control system designs in Simulink at the level of fidelity the application requires—whether an average value model, a full-switching model, or something in between. They then export the model as an FMU power conditioner that can be imported by systems engineers at grid operators who rely on power system simulation environments such as DIgSILENT PowerFactory, which are adopted in grid-planning and operational studies (Figure 1). These environments are often selected because existing grid models, data, and validation workflows have long been built around them, making their continued use a practical requirement for interoperability and certification purposes.

The following sections illustrate an example use case that we implemented using this workflow: the design and system-level analysis of an energy storage system connected directly to HVDC terminals.

Modeling and Simulating the Energy Storage System

Energy storage systems (ESSs) provide essential flexibility and stability in networks with a high penetration of renewable energy. The ESS we designed for this application is connected directly to HVDC terminals through a branch of energy storage submodules (ES-SMs) connected in series, along with a branch inductor. Each ES-SM combines a half-bridge submodule with a storage element—either a battery or supercapacitor—interfaced through a DC/DC converter (Figure 2). This modular architecture enables the ESS to inject or absorb power from the HVDC system across the full voltage range of the storage.

For this use case, we were primarily interested in grid-level analyses rather than the detailed behavior of the ESS itself. So, we decided to implement an average value model that captures ESS behavior without simulating individual power electronics switch states. Rather than modeling each submodule explicitly, the average value model represents the combined dynamics using an equivalent capacitor and equivalent storage elements. To create a plant model for the ES-SM branch, we translated the circuit relationships into equations—for example, computing capacitor voltage as the integral of net current according to Kirchhoff’s current law—and implemented these equations as MATLAB Function blocks.

We then implemented two different cascaded control strategies for the ESS: One uses an inner current loop with an outer power control loop to regulate DC power injection, while the other uses an inner current loop with an outer energy control loop to manage the equivalent capacitor energy (Figure 3).

We modeled both control strategies in Simulink (Figure 4) and ran closed-loop simulations with the controllers and plant to tune control parameters and evaluate and validate the design’s performance.

This average value modeling approach provides the computational efficiency needed for grid-level simulations while maintaining sufficient accuracy for power flow and energy dynamics.

Exporting the Subsystem Model as an FMU for Grid-Level Cosimulation

Once we validated the ESS model in Simulink, we exported it as an FMU for grid-level cosimulation, which in this example was carried out using PowerFactory. To do this, we used Simulink Compiler™ and the FMU Builder for Simulink support package to generate a standalone FMU that contains a binary shared library compiled from the ESS Simulink model. The resulting FMU file acts as a standardized, platform-independent, and self-contained black box. It contains executable code that implements the model’s behavior without exposing the underlying Simulink design.

In PowerFactory, we created a simple test model consisting of a branch inductor and a controlled voltage source and connected this model with the FMU—containing both the controllers and ES-SM plant—that we had generated from Simulink (Figure 5).

To validate our FMU-based workflow, we ran identical simulation scenarios in both Simulink and PowerFactory. The simulation starts with no power absorption and with a voltage on the equivalent capacitor equal to \(700\) \( kV \). At t=0.5 s, the equivalent capacitor voltage reference jumps to \(710\) \(  kV \), while at t=1 s, the power reference jumps to \(10\) \(  MW \), and at t=1.5 s, it goes up to \(100\) \(  MW \) (in both cases, the power flow direction is toward the storage system). The Simulink results are presented in Figure 6, whereas the DIgSILENT PowerFactory simulation results are shown in Figure 7.

The simulation results showed close agreement (Figures 6 and 7), with both the power and voltage waveforms virtually identical after the initial transients. This result confirmed that the workflow we used enables effective cross-platform cosimulation of HVDC systems and subsystems.

Realized Benefits and Next Steps

The FMI- and FMU-based workflow we have established delivers significant advantages for HVDC system development. By separating the device design activities from the grid-level certification activities, power electronics engineers can focus on subsystem design and validation in Simulink, while power systems engineers can perform certification and stability analyses in specialized tools like PowerFactory—without requiring manual model re-implementation. This capability is particularly valuable because rebuilding complex control models in PowerFactory can be time-consuming, requiring specialized knowledge of DIgSILENT Simulation Language. Our workflow bridges these two environments, using the FMI standard to enable model exchange and cosimulation. Beyond efficiency, our workflow provides three key benefits: interoperability across simulation platforms, reusability of verified models (saving development time and ensuring consistency), and IP protection through black-box encapsulation. The dual advantages of interoperability and IP protection have proven particularly valuable within the HVDC-WISE consortium, enabling partners to securely share models and enhance collaborative development efforts.

We have already used this workflow for a number of models in addition to the ESS, including a dynamic braking system and a grid-forming MMC converter. Future efforts will focus on refining and expanding the existing model library of FMU-based HVDC components, conducting interoperability testing across additional simulation platforms, and performing further cross-validation studies to enhance model accuracy and usability. By providing standardized, interoperable models and extending the application of our workflow to more realistic, large-scale power system scenarios, we aim to promote this approach throughout the power engineering community in support of the HVDC-WISE project’s broader goal of accelerating the deployment of reliable and resilient hybrid AC/DC transmission architectures.

Acknowledgments

HVDC-WISE is supported by the European Union’s Horizon Europe program under agreement 101075424.

UK Research and Innovation funding for HVDC-WISE is provided under the UK government’s Horizon Europe funding guarantee [grant no. 10041877 & 10051113].