Energy systems are often socio-technical complex systems that are facing new challenges regarding technological and environmental changes. Because of their complex nature, they cannot be approached solely through analytical modeling, hence the inefficiency of most classical modeling approaches. In this article, a Hybrid Approach based on both systemic and analytical modeling is presented and applied to a case study. From this novel approach, a model—the Multi-Institution Building Energy System—is presented. It allowed us to highlight and detail the need for greater governance of energy systems. The socio-technical solutions identified to answer the issues of governance (Accuracy, Reliability and Fairness) were DevOps methodology and the use of Distributed Microservices Architecture. Based on this framework, the design of a Decision Support System assuring and exploiting state-of-the-art scalable tools for data management and machine learning factories is described in this article. Moreover, we wish to set up the conceptual basis necessary for the design of a generic theoretical framework of optimization applicable to complex socio-technical systems in the context of the management of a shared resource.
1 January 2012 in Journal of Air Transport Studies
The dynamical behaviour of groups of airspace sectors, e.g. Functional Airspace Blocks, is not trivial to be analysed without appropriate theoretical tools. In this paper, we suggest a discrete model based on cellular automata and multi agent systems to express the congestion dynamics and complexity in the controlled airspace. Discrete time simulations have been performed with random selected scenarios of traffic and with independent sector parameters to investigate the impact of availability of local sectors on the whole state of the airspace. Obtained results show the existence of a traffic threshold that leads to a theoretical saturation of airspace. The test scenario showed a phase transition phenomenon towards the congestion of the European airspace at the resulting traffic threshold circa 50 000 flights. Validation using real data shows the predictive abilities of the model.