The European transmission system has more than 3,400 power plants, 10,000 substations, and 25,000 pieces of equipment, which include transformers and transmission lines. Because it can take years to set up a new transmission line, grid operators must make the most of the equipment they have. They continuously monitor changes in demand as well as changes in production from wind and solar throughout the day and then respond accordingly, increasing production or curtailing production to reduce congestion.

To develop automated systems for grid optimization, the OVGU researchers first needed an accurate model of the entire grid, including all power plants, substations, and lines. Because their goal was to use this model in advanced research on optimization strategies as well as in teaching graduate-level courses on power network planning and operation, they wanted to build their own model rather than use prepackaged commercial software. Finally, they needed to integrate both the grid model and the optimization algorithms they developed with the industry-standard control software used by all of Germany’s largest transmission system operators.

Working in MATLAB, OVGU researchers modeled the power flow behavior of transformers and other individual pieces of equipment using systems of nonlinear equations.

To create a model of the entire grid, they merged all the individual systems of equations. They automated this process by writing MATLAB code that reads files provided by grid operators in UCTE and other data exchange formats that define the grid’s network topology and parameters for individual elements.

The team developed a Java interface for the MATLAB grid model. Based on the IEC 60870-5‑101 protocol, this interface enables data exchange between the model and the commercial control software used by grid operators.

Using a combination of control software and the grid model, the team ran simulations in which they evaluated the performance of the grid during normal operations and under failure scenarios. During these simulations, they accelerated the power flow calculations for the grid model by running them in parallel on a multicore workstation using Parallel Computing Toolbox™.

The team used Mapping Toolbox™ to create visualizations of the simulation results that show, for example, lines, substations, and congested areas of the grid on a map display.

Turning their focus to optimization, the researchers began automating the previously manual process of responding to the contingency list produced by the grid control software each time it runs. This list details current problems with the grid, including voltages or frequencies that are outside normal ranges.

Using Optimization Toolbox™ and Global Optimization Toolbox, the team developed optimization algorithms that generate a set of changes to be implemented in the grid—for example, reducing power generation at specific plants—to remediate the identified problems while minimizing costs and outage risks.

The optimization algorithms incorporate linear and quadratic programming solvers, as well as more advanced techniques based on genetic algorithms and particle swarm optimization.

The research team’s work is also used in a two-semester graduate level course at OVGU. In the first semester, students use MATLAB to develop their own models of grid elements, which they use to perform grid calculations and stability simulations. In the second semester, they use MATLAB and Optimization Toolbox to optimize switching states, grid losses, and remedial actions to maximize grid performance.

A power grid simulation serves as an important training tool for students as part of the OGVU curriculum and climate-change initiative. The course “Power Network Planning and Operation,” for example, includes a virtual exercise in which instructors demonstrate and students experience the operation and function of a control room. The trainer-student system simulates network faults to which students have to react. With this training system, OVGU instructors can operate 25 virtual workstations simultaneously from a computer lab, where exercises and lectures are held.

• Renewable energy curtailments reduced by more than 50%. “It’s always our goal to use as much renewable energy as possible,” says Professor Wolter. “With MATLAB, we were able to optimize an emergency procedure that reduced the amount of renewable energy that had to be curtailed by more than half—from 1.5 gigawatts to just 0.6 gigawatts.”
• Calculations of entire grid completed in less than one second. “Because MATLAB is so efficient at the sparse matrix calculations we need in computing power flows across the grid, we can calculate the entire European grid in under a second,” notes Professor Wolter. “That’s faster than most commercially available software designed for that task.”
• Student learning curves shortened. “As an instructor, I see how much easier it is for students to use MATLAB for modeling, optimization, and visualization compared to other languages,” says Professor Wolter. “In fact, one of the main advantages of MATLAB is that it makes it easy for students to start programming.”