GPU-based power converter transient simulation with matrix exponential integration and memory management

Wu, Wei; Li, Peng; Fu, Xiaopeng; Wang, Zhiying; Wu, Jianzhong; Wang, Chengshan

doi:10.1016/j.ijepes.2020.106186

Cited by 10 publications

(4 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As illustrated in Figure 9, the test system has two configurations, which correspond to different stages of the project development, one with a PV unit and a permanent magnet synchronous generator (PMSG)‐based wind energy conversion system (WECS), and another with two PV units and a PMSG‐based WECS. Detailed parameters of the IEEE 123‐node network can be found in [27], and the structure and parameters of the PV unit and PMSG‐based WECS are provided in [19] and [28], respectively. We considered different transient events in the two test cases to evaluate the performance of the proposed AMR method.…”

Section: Numerical Test and Validationmentioning

confidence: 99%

Asynchronous multi‐rate method of real‐time simulation for active distribution networks

Wang

et al. 2022

IET Renewable Power Gen

Self Cite

View full text Add to dashboard Cite

The real-time simulation of active distribution networks (ADNs) can provide an accurate insight into transient behaviours, but faces challenges in simulation efficiency and flexibility brought by larger system scales and wider time-scale ranges. This paper presents an asynchronous multi-rate (AMR) method and design for the real-time simulation of large-scale ADNs. In the proposed method, the entire ADN was decoupled into different subsystems according to accuracy requirements, and optimized time-steps were allocated to each subsystem to realize a fully distributed simulation. This not only alleviated the time-step coordination problem existing in multi-rate real-time simulations, but also enhanced the flexible expansion capabilities of the real-time simulator. To realize the AMR real-time simulation, a multi-rate interfacing method, synchronization mechanism, and data communication strategy are proposed in this paper, and their hardware design is also presented in detail. A modified IEEE 123-node system with photovoltaics and wind turbine generators was simulated on a 3 field-programmable gate arrays (FPGAs)-based AMR realtime simulator. The real-time results were captured by the oscilloscope and verified with PSCAD/EMTDC, which demonstrated the superiority in simulation flexibility and accuracy compared with the synchronous multi-rate (SMR) method.

show abstract

Section: Numerical Test and Validationmentioning

confidence: 99%

Asynchronous multi‐rate method of real‐time simulation for active distribution networks

Wang

et al. 2022

IET Renewable Power Gen

Self Cite

View full text Add to dashboard Cite

show abstract

“…Systems with identical circuitry, like multilevel converters, are suitable to be executed on GPUs [115]- [117]. For instance, [118] suggests an exponential integration technique for parallel computing to improve the efficiency and accuracy of transient simulation of power converters. To further accelerate EMT simulations, massive-thread parallelization of device-level simulation implemented on GPUs was presented in [119].…”

Section: Hardware-based Simulation Toolsmentioning

confidence: 99%

Review of Methods to Accelerate Electromagnetic Transient Simulation of Power Systems

et al. 2021

View full text Add to dashboard Cite

Converter-based generators are increasingly replacing classical synchronous generation, resulting in significant challenges to the operation and planning of modern power systems. Power electronics (PE)-based equipment, along with non-linear PE-driven loads, introduce time-varying characteristics and fast switching behavior that increases the complexity of the power system model. Faster control actions are needed to overcome the fast switching dynamics to ensure the reliability and stability of future power systems. Thus, this requires advanced and detailed simulation methods and tools with highly accurate equivalent models to embody the relatively slower electromechanical to faster electromagnetic transient (EMT) phenomena. Conventional transient stability analysis using positive-sequence simulators has become inadequate for representing converter-dominated power systems, while EMT simulators suffer from the high computational burden. This review paper presents accelerated EMT simulation methods and tools that are categorized and discussed in three topics: system equivalents, simulation methods, and accelerating tools. Dynamic system equivalent techniques are discussed to model small to large interconnected external systems of the grid network. Moreover, a systematic review is made for existing EMT simulation methods, along with advanced co-simulation methods, for addressing simulation speed and accuracy issues in large power system networks. Emerging hardware-based simulation tools are reviewed that reduce the computational burden and increase the simulation efficiency of the power system model. Challenges and trends in EMT simulation are also presented and concluded by providing perspectives on this research topic.

show abstract

“…Following the pioneering work [14] in 2011, many studies have been carried out to advance the acceleration of devicelevel or system-level EMT simulations by incorporating graphics processing unit (GPU) technology [15]- [18]. The single instruction multiple thread execution mode facilitated by GPUs allows for an effortless implementation of threadlevel parallelization in EMT simulations, leading to a speedup factor up to 6 [17] and 40 [18] compared to conventional central processing unit (CPU)-based simulations.…”

Section: Introductionmentioning

confidence: 99%

ParaEMT: An Open Source, Parallelizable, and HPC-Compatible EMT Simulator for Large-Scale IBR-Rich Power Grids

Xiong,

Wang,

Vaidhynathan

et al. 2024

IEEE Trans. Power Delivery

View full text Add to dashboard Cite

The electromagnetic transient (EMT) simulation is an essential tool for studying power grids dominated by inverterbased resources (IBRs). However, due to small simulation time steps and increasing problem sizes, performing EMT simulations for large-scale power grids becomes computational-intensive, and often impractical. To address this challenge, we developed Pa-raEMT, an open-source Python-based EMT simulator that is parallelizable and compatible with high-performance computing (HPC) systems for simulating large-scale power grids with a significant presence of IBRs. Its key features include: 1) utilizing parallel computation for network solution by decomposing the network conductance matrix into the bordered block diagonal form; 2) enabling parallel updates of device states and network historical currents; 3) leveraging HPC to further accelerate simulation through a developed generic interface. The accuracy of ParaEMT has been validated on the reduced 240-bus (720-node) Western Electricity Coordinating Council system by benchmarking the EMT dynamics against PSCAD. Furthermore, ParaEMT achieves a notable speedup of approximately 25 to 36 times on a synthetic 10,080-bus (30240-node) system by leveraging the HPC resource named Eagle at the National Renewable Energy Laboratory. A regional 100% renewable case of the reduced 240-bus system has been developed for simulating system-wide IBRs' interactions in large-scale power grids using ParaEMT.

show abstract

GPU-based power converter transient simulation with matrix exponential integration and memory management

Cited by 10 publications

References 31 publications

Asynchronous multi‐rate method of real‐time simulation for active distribution networks

Asynchronous multi‐rate method of real‐time simulation for active distribution networks

Review of Methods to Accelerate Electromagnetic Transient Simulation of Power Systems

ParaEMT: An Open Source, Parallelizable, and HPC-Compatible EMT Simulator for Large-Scale IBR-Rich Power Grids

Contact Info

Product

Resources

About