No abstract

Virtualisation is a concept successfully applied to IT systems. In this work, we analyse how virtualisation approaches, such as edge computing, brokerage and software-defined networking, can be applied in the area of electricity grid management and control systems. Power system information and communications technology is currently subject to significant changes. Networked power grid components including renewable energy units, electric vehicles and heat pumps need to be integrated into grid management systems. We studied how virtualisation techniques can support system operators in increasing an energy and communication system’s dependability and situational awareness, and how manual (mostly field-level) configuration and engineering efforts can be reduced. Starting from a working hypothesis, three concrete use-cases were implemented and the performance enhancements were benchmarked to allow for well-informed answers to the questions above. We took a close look at application-protocol-independent redundancy, grid-based routing and online system integrity control. In these study cases, we found significant improvements could be achieved with virtualisation in terms of reduced engineering effort, better system management and simplification in high-level system architecture, since implementation details are hidden by the virtualisation approach.

Due to changed power consumption patterns, technological advance and deregulation, the appearance of the power grid in the low and medium voltage segment has changed. The spread of heating and cooling with electrical energy and an increase of electric vehicles as well as the broad rollout of photovoltaic systems has a major impact on the peak power demand of modern households and the volatility smart grids have to face. Thus, besides the load impact of the growing population of electric vehicles, modern households are not only consumers of electrical power, but also power producers, so called prosumers. The rising number of prosumers and the limitations of grid capacities lead to an increasingly distributed system of heterogeneous components, which have to be managed and operated with locality and scalability in mind. Virtualisation technologies, particularly known as state of the art in data centre computing, can lead to a paradigm shift needed to meet the growing demands of this evolution. A key issue here is to forward data to the correct data sinks, where data are required in order to keep the grid balanced. This routing process has to be able to react on grid changes in a timely manner, i.e., it must be based on the instantaneous state of the grid. In this paper, we propose a solution based on virtualising the communication infrastructure in the low and medium voltage grid. We evaluate two different approaches. The first approach is based on SDN; an ONOS SDN controller is used to change the behaviour of the communication infrastructure according to information provided by components of the power grid. The second approach uses Coaty and a Mosquitto MQTT broker to deliver messages to the desired endpoint, again based on information from the power grid.

Epitaxial integration of direct-bandgap III–V compound semiconductors with silicon requires overcoming a significant lattice mismatch. To this end, GaAsP step-graded buffer layers are commonly applied. The thickness and composition of the individual layers are decisive for the envisaged strain relaxation. We study GaAsP growth by metalorganic vapor phase epitaxy in situ with reflection anisotropy spectroscopy. We find that the growth surface exhibits optical fingerprints of atomically well-ordered surfaces. These allow for tuning the interface preparation between adjacent layers. The spectral position of the characteristic peaks in the RA spectra, which are related to surface-modified bulk transitions, behaves similarly upon an increased As content as does the E1 interband transition of GaAsP at the growth temperature. The impact of strain on this shift is negligible. We thus monitor a bulk property via the surface reconstruction. An empiric model enables quantification of the As content of individual layers directly in situ without growth interruptions and for various surface reconstructions. Our findings are suitable for a simplified optimization of the GaAsP buffer growth for high-efficiency devices.

No abstract