The amount of power generation plants based on renewable energy sources has increased in the last couple of years. Uncertain fluctuating generation becomes more and more important for the supply of electrical energy. Besides big power plants such as windparks many small, distributed plants such as photovoltaic installations will feed-in their power at distribution grid level. These kinds of generation plants are hard to forecast and schedule. Consequently, the amount of balancing power has to be increased in order to fulfill the demand for power. 

The future Smart Grid will allow to control electric loads at the demand side. As a consequence, the load curve will be smoothened out. However, a simultaneous switching-on of loads may lead to a capacity overload. The detection of possible violations must happen in due time before the system reaches an instable state. 

OFFIS has long-term expertise in the development of distributed energy management systems that adhere to certain non-functional requirements relevant to this domain (real-time capability, stability, optimality) as well as their implementation in distributed agent-based systems. Methods for network calculations and evaluations have been successfully employed and developed in related projects. Furthermore, at the Carl von Ossietzky University in Oldenburg those topics are taught in the specialization subject energy informatics. 

Using knowledge of the R&D domain Transportation for the analysis of hybrid systems, i.e. systems that show both continuous and discrete dynamics, methods are developed to assess the security and reliability of distributed control and optimization methods in the domain of future energy grids. 


In the project SafeGrid methods are being developed for the estimation of operational states of an electrical network. The aim of this estimation is not only to check whether operational constraints are violated but to identify minimal mitigation actions for stabilizing an infeasible operational state. One the one hand, control actions on the demand side can be initialized in order to avoid violations that are about to occur. Hence, stable grid operations may be guaranteed. On the other hand, residual capacities of the grid can be detected due to the knowledge of feasible operation margins. As a consequence, the existing grid can be used more efficiently and capacities for the integration of additional power suppliers and consumers can be located. Here the variability of the system must be taken into account. 

In cooperation with the technology cluster of the OFFIS R&D division Transportation, a model of the electrical grid is being examined with regard to security and reliability of operating control. The most important properties to be considered here are expected loads and generation, control of loads and generation, deviations in the detection of an operational state as well as grid characteristics. 


The latter, of course, can be defined precisely. However, in order to provide robust operating control, only assumptions are allowed that are fulfilled by all grid components regarding loads and generation. If it is possible to proof secure and reliable operating control under these assumptions it is robust. If this is not the case, secure operating control in practice can fail due to incorrect load and generation configurations. 

Support Vector Machines for an efficient Representation of Voltage Band Constraints

Blank, Marita and Gerwinn, Sebastian and Krause, Olav and Lehnhoff, Sebastian; 11 / 2011


Start: 31.03.2011
End: 30.12.2011