On Simulation of Power Systems and Microgrid Components with SystemC-AMS
Rahul Bhadani, Satyaki Banik, Hao Tu, Srdjan Lukic, Gabor Karsai
TL;DR
This work tackles the slow and costly electromagnetic transient (EMT) simulation of microgrids by introducing a SystemC-AMS framework that enables fast, mixed-signal EMT modeling with reusable component libraries. It demonstrates a model-based design workflow using COSIDE to build grid components (e.g., three-phase voltage sources, transmission lines, abc-dq0 converters, and PLLs) and develops PV-based grid-following inverter variants with PLL-driven synchronization in the $dq$ frame. The authors report roughly a 3× speedup over Simulink while preserving EMT fidelity, and they present a real-time, hardware-in-the-loop capable DC microgrid using ZeroMQ for plant-controller communication. Collectively, the results validate SystemC-AMS as a scalable, real-time-capable digital twin platform for microgrid study and design, with clear pathways toward broader component libraries and middleware integration for large-scale grids.
Abstract
Cyber-physical systems such as microgrids consist of interconnected components, localized power systems, and distributed energy resources with clearly defined electrical boundaries. They can function independently but can also work in tandem with the main grid. Power system converters and their control loops play an essential role in stabilizing grids and interfacing a microgrid with the main grid. The optimal selection of microgrid components for installation is expensive. Simulation of microgrids provides a cost-effective solution. However, when studying the electromagnetic transient response, their simulation is slow. Furthermore, software packages facilitating electromagnetic transient response may be prohibitively expensive. This paper presents a faster method for simulating the electromagnetic transient response of microgrid components using SystemC-AMS. We present a use case of a photovoltaic grid-following inverter with a phase-locked loop to track reference active and reactive power. Our results demonstrate that the simulation performed using SystemC-AMS is roughly three times faster than the benchmark simulation conducted using Simulink. Our implementation of a photovoltaic grid-following inverter equipped with a phase-locked loop for monitoring reference active and reactive power reveals that the simulation executed using SystemC-AMS is approximately three times faster than the benchmark simulation carried out using Simulink. Our implementation adopts a model-based design and produces a library of components that can be used to construct increasingly complex grid architectures. Additionally, the C-based nature allows for the integration of external libraries for added real-time capability and optimization functionality. We also present a use case for real-time simulation using a DC microgrid with a constant resistive load.
