Open Science – freely available and open source scientific results
Freedom and openness are essential characteristics of science. Without an open exchange of ideas and unrestricted access to preliminary scientific work, no sound scientific progress - and thus no innovations - can emerge. And free access to results in the form of software artefacts, data and publications, among other things, determines the quality and connectivity of scientific work in the long term. Access to the underlying data and the software tools used or developed for data generation and analysis is indispensable for the traceability of scientific results.
Especially when scientific results are generated in projects with public funding, we are convinced that they should be able to be used collaboratively and without hurdles in order to improve the quality of research and contribute to progress in society as a whole. An essential aspect of this is also the possibility of reproducing published research results in the sense of the DFG's guidelines on good scientific practice.
In the following, we present some of our published results. Our software tools, frameworks, data, models and other artefacts are usually published in our Gitlab group.
mango – Modular Python Agent Framework
mango is a Python framework for the development of multi-agent systems. It provides a basic framework for building a single intelligent software agent, offers simple interfaces for communication between agents and enables the modularisation of complex agents. A container mechanism is used to accelerate the exchange of messages for agents located within a dedicated process.
Using mango, large-scale simulations can be carried out, but it can also be used to develop agents for the field. The software has been under development at OFFIS and the University of Oldenburg since 2019, is available as open source and extensively documented. Furthermore, there is a public "mango library", in which various reference implementations of mango agents are available.
cosima – Communication Simulation for Agents in Energy Systems
In smart energy systems, co-simulation is used to connect heterogeneous simulators such as generation units, control algorithms and analysis tools to complex scenarios and to investigate these scenarios for different objectives. In these scenarios, the communication necessary for the exchange of information between simulators has so far been assumed to be undisturbed and undelayed. The interaction between the energy and communication systems was therefore neglected.
In cosima, the communication simulator OMNeT++ is therefore integrated into the co-simulation framework mosaik to enable a combined view of energy and communication systems and to investigate ICT-based monitoring and control applications (such as multi-agent systems) in the energy system. Through the OMNeT++ integration, concrete communication technologies (LTE, 5G) can be simulated and the effect of topology changes and disturbed and undisturbed message traffic on e.g. latency, bandwidth and packet losses can be recorded.
Further information can be found at https://cosima.offis.de/.
Amplify – Abstract Multi-Purpose-Limited Flexibility Model
To increase the yield generated by battery storage systems, it makes sense to use them for more than one application. This is possible because it has been found that they are often not fully utilised only through e.g. peak load capping or self-consumption maximisation.
With Amplify, excess flexibility from battery storage can be described and dispatched compactly after a local application has already been fulfilled and no further flexibility is needed for it. The model is characterised by the fact that it can be calculated very quickly and has integrated problem detection. This allows it to detect when the battery storage contributions for the different applications conflict. In addition, an aggregator does not need any further knowledge about the battery storage and can still size contributions for e.g. electricity market transactions.
Amplify is published on Pypi as "amplify-model" and as open source on Gitlab.com.
In order to increase the yield generated with battery storage systems, it makes sense to use them for more than one application. This is possible because it has been found that they are often not fully utilised only through e.g. peak load capping or self-consumption maximisation.