An efficient co-simulation and control approach to tackle complex multi-domain energetic systems: concepts and applications of the PEGASE platform

Abstract

In this paper, we present a novel research software, called PEGASE, suitable for the design, validation and deployment of advanced control strategies for complex multi-domain energy systems. PEGASE especially features a highly efficient cosimulation engine, together with integrated solutions for defining both rule-based control strategies and Model-Predictive Control (MPC). The main principle behind the PEGASE platform is divide-and-conquer. Indeed, rather than trying to solve a problem as a monolithic entity, which can be highly complex for multi-domain large-scale systems, it is often more efficient to decompose it into several domains or sub-problems, and to simulate them in a decoupled way. To provide its cosimulation capabilities, we based PEGASE on two main components. The first one is a framework for integrating simulation models, which can be either compatible with the FMI standard or interfaced through an Application Programming Interface (API). The second one is a multi-threaded sequencer enabling several simulation sequences with different time steps. To provide advanced control capabilities, we also equipped PEGASE with a framework for MPC combining a comprehensive management of predictions data and a modeler dedicated to the formulation of Mixed Integer Linear Programs. We implemented this framework in C++ providing low formulation and resolution times for typical applications. Connection to hardware is also available via standard industry protocols thereby allowing PEGASE to control real energy systems. In this paper, we show how these basic functionalities, combined with dedicated modeling tools, enable setting up simulation and control applications suitable for tackling the complexity of various kinds of energy systems. To illustrate this, we present four application examples from our recent research work. These examples cover several domains, from concentrated solar thermal plants to optimal control of district heating networks. The variety of examples demonstrates the robustness and genericity of the approach.