Distributed real-time simulation of modular bidirectional dc-dc converters : for control-hardware-in-the-loop

• Verteilte Echtzeitsimulation modularer bidirektionaler DC-DC-Wandler für Control-Hardware-in-the-Loop

Joebges, Philipp; de Doncker, Rik W. (Thesis advisor); Monti, Antonello (Thesis advisor)

1. Auflage. - Aachen : E.ON Energy Research Center, RWTH Aachen University (2021)
Book, Dissertation / PhD Thesis

In: E.ON Energy Research Center ; PGS, Power Generation and Storage Systems 94
Page(s)/Article-Nr.: 1 Online-Ressource : Illustrationen, Diagramme

Dissertation, RWTH Aachen University, 2021

Abstract

The electricity supply system is undergoing constant change, which has accelerated in recent years as part of the transformation towards a sustainable, decentralized supply system that is accompanied by increased level of digitalization. This development will gain further relevance in the coming decades as the targets for energy transformation are aiming at zero net emissions.Modern power electronics provides a key enabling technology for the sustainable integration of renewables, generators, prosumers, and storage systems as well as for the efficient distribution of electrical energy. Flexible cellular direct current (dc) grids allow for intelligent energy routing with increased efficiency, along with a strong reduction of the required resources. Hence, dc technology utilizes the installed infrastructure in an optimal way. Highly efficient, reliable, and fault-tolerant modular scalable, galvanically isolated dc-dc converters form the core building blocks. Such converters must operate reliably under all operating conditions within the entire system. At the same time, they are continually optimized according to advancements in technology. The requirements that are imposed on these dc-dc converters can be ensured using control-hardware-in-the-loop (CHiL) as a methodology that, on the one hand, verifies the control functionality of systems and on the other hand concurrently allows for the development of hard- and software. However, CHiL requires accurate real-time plant models and high-performance simulation hardware that meets the high demands on processing speed. The accuracy and stability of the real-time simulation is directly dependent on the discretization step size that arises from the combination of model complexity and the execution times of the simulator. For complex systems, a subdivision into coupled subsystems, which can be simulated in parallel in real-time, is advantageous. In this thesis, a distributed real-time-capable modeling methodology is developed through the example of modular, high-power, dc-dc converters. The results show that significant performance gains can be achieved by optimizing the execution of the model. Furthermore, fast analytical models are developed to represent non-linear component behavior; these models offer the opportunity to evaluate the dynamic behavior of the control algorithms in a CHiL environment. In particular, this thesis focuses on decoupled parallel modeling and its influence on numerical stability and accuracy. Numerical stability is significantly affected by the decoupling of complex simulation models. Apart from the discretization step size, the selection of relevant state variables (of the energy storage components) directly influences the stability boundaries. Although the methodology is developed for an exemplary application, this thesis finds that it can be generalized and applied to common converter-based electrical systems. In this study, the developed control of a modular series-parallel-connected dc-dc converter is evaluated in a distributed CHiL setup. For this purpose, a scaling methodology is also developed and demonstrated to represent models with small time constants while maintaining high accuracy. Based on the findings, the proposed control approach is suitable and works both in offline simulations and in a real-time CHiL that uses the proposed scaling methodology. Finally, the effects considered in the real-time modeling are discussed on the basis of the measurement results of a high-power 5 MW, 5 kV dc-dc converter. In addition, the developed modular control system is implemented and verified in a multi-converter hardware setup of a 5 kV to ± 380 V converter consisting of eight dc-dc modules. The use of distributed real-time setups reveals that the partitioning of complex systems into coupled parallel-executed subsystems is a promising approach for CHiL. The selection of discretization step sizes in accordance with the relevant system time constants greatly influences the numerical stability of distributed simulations. As such, this thesis shows that the scaling of critical time constants, together with the corresponding artificial adaptation of the control parameters, can significantly improve the results for a real-time simulation while maintaining the dynamic performance of the corresponding initial control design. Moreover, the implementation of analytical system models improves the calculation performance and the reproduction of non-linear component behavior.