Technical Articles

Developing a Real-Time Simulation Environment for Electrical Distribution Grids

By Jan Petznik, Alejandro Rubio, Moiz Ahmed, and Frank Schuldt, DLR e.V.

Generating and distributing electric power has traditionally involved large, centralized plants that burn fossil fuels. However, a new model characterized by decentralized infrastructure and renewable sources of energy is emerging in many parts of the world. As part of this fundamental transformation, engineers and researchers are looking for ways to design stable, efficient energy systems based on localized power generated by wind, solar, and other cleaner, weather-dependent sources of energy.

Our team at the Institute of Networked Energy Systems—one of 55 institutes at the German Aerospace Center (DLR)—is actively researching the robust operation of power grids that can facilitate this radical change by integrating technologies from electrical, heat, and transportation (or mobility) domains. Our research requires a combination of digital simulation and field testing on actual power hardware. However, while simulation enables us to consistently explore scenarios that would be too costly or dangerous to test on real equipment, it cannot fully account for latency, clock synchronization, and other real-world effects.

To close the gap between power grid simulation and field tests, we have built the Networked Energy Systems Emulation Centre (Grid Lab; Figure 1). The main feature of the Grid Lab is the integration of experimental grid setups with simulated environments and models. The most important role in this integration is played by the 18 power amplifiers, which can deliver or consume up to 50 kVA each. These can represent network participants within the experimental grids—such as entire buildings, apartments, photovoltaic inverters, charging stations for electric vehicles, and heat pumps, to name just a few. In addition, there are more than 16 bidirectional DC source/sink systems, each capable of delivering up to 15 kilowatts at a maximum voltage of 1,500 DCV. It also includes a 30 kVA synchronous generator, which we use to emulate the functionality of a classical power plant and reproduce the effects of varying rotational kinetic energy for inertial frequency support.

The Grid Lab’s infrastructure, including terminals for routing cables to emulate grid networks.

Figure 1. The Grid Lab (above) and terminals for routing cables to emulate grid networks (below). (Image Credit: DLR)

A core component of the laboratory’s operations is the real-time simulation and testing environment that we created using Simulink®, Simulink Real-Time™, and Speedgoat® target machines. This environment enables us to model control strategies as well as entire grids, and then run simulations in real time that incorporate real-world power hardware. We can use it to push processes outside their nominal operating conditions and safely evaluate the outcomes. In addition, the testing environment lets us rapidly validate new ideas and control algorithms that we develop in house with real-time tests using commercial equipment from partners and vendors.

Real-Time Testing in DLR’s Grid Lab

A few years ago, our team characterized the cables used in real low-voltage distribution grids. We used the data from this exercise to create cables with similar impedances that, in turn, enabled us to construct various grid topologies linking together a variety of components in the Grid Lab’s architecture (Figure 2). These components include actual or simulated inverters and converters, as well as a commercial grid control system, interfaces to other labs, and our real-time testing setup.

Architectural diagram of the different grid topologies linking together a variety of components in the Grid Lab’s architecture.

Figure 2. An architectural diagram of the Networked Energy Systems Emulation Centre. (Image Credit: DLR)

Within this architecture, our team tests and optimizes control strategies that we model in Simulink and deploy to Speedgoat target computer using Simulink Coder™ and Simulink Real-Time. The controllers access current and voltage signal measurements from grid nodes via the Speedgoat machine’s FPGA I/O modules. The controllers also transmit analog signals via those same modules to control various components in the networks we evaluate.

Cosimulations Across Geographically Distributed Labs

Our team collaborates with groups at several other labs, both within DLR and at other institutions, which are equipped with specialized power hardware. Rather than move that hardware to the Emulation Centre, we have built a cosimulation environment that enables us to link two or more laboratories via a UDP connection.

We are using this approach of linking labs together on the research project Zukunftslabor Energie (ZLE) with a team at the University of Applied Sciences Emden/Leer. Together, our teams are exploring strategies for employing flexible provisioning to handle load peaks and infeed-peaks in low-voltage grids. We have modeled roughly one-third of a low-voltage distribution grid for this project in Simulink with line, node, and load components from Simscape Electrical™. Another third of the network is emulated via hardware in our Grid Lab with Simulink Real-Time and Speedgoat, and the final third is emulated via hardware in the Laboratory of Renewable Energies of our colleagues at University of Applied Sciences Emden/Leer (Figure 3).

Model showing the components of a low-voltage distribution grid that is cosimulated across two locations.

Figure 3. A low-voltage distribution grid cosimulated across two locations. (Image Credit: DLR, Hochschule Emden/Leer, Forschungsstelle für Energiewirtschaft e.V.)

The three subsystems, running concurrently and communicating via UDP, can be used to run power hardware-in-the-loop (PHIL) tests in which control strategies automatically compensate for sudden drops or surges in power generation from a wind or solar power plant in Emden, Germany.

Grid-in-the-Loop Simulations

Looking beyond PHIL simulations, we are also planning grid-in-the-loop simulations in which an entire low- or medium-voltage grid is modeled in Simulink and then simulated in real time on a Speedgoat target computer. The real-time simulation is linked via a power interface and line emulators replicating the topology of a representative low-voltage grid to a variety of real grid participants that produce or consume power (Figure 4).

Model showing the architecture for a grid-in-the-loop simulation.

Figure 4. An architecture for a grid-in-the-loop simulation. (Image credit: DLR)

Internet of Things and Energy Management

We are using a similar approach to study Quarter Energy Management Systems (QEMS). For this project, an Internet of Things device is connected to the inverter or other device to be controlled while control and monitoring signals are received from the QEMS via the internet (Figure 5). An amplifier recreates the voltage at a node within a grid model that is simulated with Simulink and Speedgoat.

Model showing the architecture of a Quarter Energy Management System.

Figure 5. Real-time simulation for an energy management system. (Image credit: DLR, Speedgoat)

Ongoing Research

Our team is actively using real-time testing and simulation with Simulink and Speedgoat within the Grid Lab to pursue a wide range of research initiatives, including intelligent inverter systems that employ machine learning algorithms; DC, hybrid, and inverter-dominated grids; e-mobility technologies including bidirectional charging stations; decentralized supply structures; and other high-power applications.

Published 2022

View Articles for Related Industries