Predictive Control Concepts for Power Grids of the Future

The current focus of the energy transition is mainly on power generation, transportation networks and electric mobility. However, in terms of creating a CO2-neutral energy supply, perspectives must be broadened because the energy cycle includes generation, conversion, transport, storage, and consumption in electricity, gas, and thermal grids.

Regardless of the energy medium, information technology and mathematics encounter a number of basic recurring problems in the modeling, simulation, and control of hierarchical energy grids with stochastic production and usage.


Project MathEnergy

In Project MathEnergy, sponsored by the Federal Economics Ministry, the solution to this problem is to be found in the development of mathematical methods, collected in a software library, and demonstrated in the fields of gas and electricity and even in a coupling of the two.

The project is divided in the following segments:

  • overall grid modeling
  • model reduction
  • scenario analysis
  • state estimation
  • control, overall integration and demonstrators

Our department, in particular, is working on a grid-independent, model-based monitoring and control concept for the planning and operation of the electricity transport and distribution grid.


Model-based optimal control 

The determination of the current system status based on the measured data underlying the mathematical models is the starting point for model-based, optimal control of the input and consumption of electricity or gas. Taking into account the observability of the model and an error analysis of the forecasted input to the power grid from alternative energy sources, methods can be developed for the optimal positioning of additional sensors for the dynamic state estimation.

The technical (sampling rates, signal propagation times, errors, etc.) and economic limiting conditions must also be considered. Kalman filtering and analyses using particle filter methods developed in the department are used for dynamic state estimation. The latter can be used for state estimates for systems with stochastic behavior, with physical limitations, and at the same time, unevenly distributed measurements.

The real-time tools developed for estimating the state and the scenario analysis methods are then used in a control module for multi-grid coordination by means of predictive model control. Hierarchical or distributed MPC methods with reduced dynamic models are required to enable the data exchange among the different controls.