HLHengyu Li

Hengyu Li

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Hengyu Li has a total of 8 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • A Decoupled Detailed Equivalent Model for Accelerated EMT-type Offline and Real-time Simulation of Three-Port DAB-MMC with BESS

    by Walid Hatahet, Hengyu Li, Jared Paull, Liwei Wang, Juan Páez Alvarez

    Three-port dual-active-bridge modular multilevel converter (DAB-MMC) with battery energy storage system (BESS) enables efficient integration of high voltage DC (HVDC) grids and medium voltage DC (MVDC) networks. Electromagnetic transient (EMT)-type simulation is critical for the design, control, and prototyping of the DAB-MMC with BESS. This paper proposes a decoupled switching function-based detailed equivalent model (DEM) with combined implicit and explicit (ImEx) multi-step Backward Differentiation Formula (BDF) solver for accurate EMT simulation of DAB-MMC with BESS. The proposed DEM employs 2nd- or 3rd-order ImEx-BDF to enable circuit decoupling, thereby reducing node count, and producing a constant network nodal conductance matrix. The proposed model can accurately represent the dynamics of the converter system under deblocking and blocking modes. Furthermore, a switching interpolation technique is employed to ensure accurate compensation of intra-time-step switching events. To further accelerate the simulation of DAB-MMC, a hybrid-parallel EMT solver is implemented using central and graphic processing units (CPU-GPU). The accuracy of the proposed decoupled DEM is validated against a detailed model (DM) developed in Simulink/Simscape Electrical toolbox with a 1μs time step-size. The proposed hybrid CPU-GPU solver achieves 49-fold speedup compared to a CPU-only serial implementation, and a 59,229-fold speedup compared to the DM reference solution.

  • A Numerically Efficient and Accurate Model for Real-Time Simulation of Solid-State Transformer Using Implicit-Explicit Integration Methods

    by Hengyu Li, Walid Hatahet, Jared J. Paull, Fei Zhang, Yuanshi Zhang, Liwei Wang, Wei Li

    Electromagnetic transient (EMT) models of power electronic converters are essential for converter design, control, and fault analysis. This article proposes a switching-function-based detailed equivalent model (SFB-DEM) using combined implicit and explicit (ImEx) multistep Gear’s integration methods for numerically efficient and accurate EMT simulation. The proposed SFB-DEM integrates the benefits of ImEx solvers, featuring converter circuit decoupling, node number reduction, and constant nodal-network conductance(G)-matrix in the EMT model. The SFB-DEMs employing the ImEx 2nd and 3rd order Gear’s (i.e., ImEx-G2O and ImEx-G3O) methods are implemented for solid-state transformer (SST) simulation. In addition, a switching interpolation technique is proposed and integrated with the ImEx-G2O and ImEx-G3O solvers to account for intra-time-step switching events. The proposed SST SFB-DEMs greatly accelerate the EMT simulation, compared to the conventional detailed model (DM) and variable conductance(G)-matrix DEM (VG-DEM). For the SST with 60 SMs, the proposed SFB-DEM with ImEx-G3O method achieves the EMT simulation speedup by 171 and 7.5 folds, compared to the DM and VG-DEM, respectively.


  • Decoupled Detailed Equivalent Model for Parallel and Multi-Rate EMT-Type Simulation of Modular Multilevel Converter With Battery Energy Storage

    by Walid Hatahet, Hengyu Li, Liwei Wang, Wei Li

    Modular multilevel converters (MMCs) integrated with battery energy storage systems (BESS) enable efficient utilization of renewable energy resources such as wind and photovoltaic, while enhancing reliability and scalability of high-voltage direct current systems. This paper proposes a decoupled detailed equivalent model (D-DEM) for BESS-integrated MMC for electromagnetic transient (EMT) simulation. The proposed model can accurately represent the dynamics of the converter under deblocking and blocking modes. To efficiently utilize available hardware resources, a multi-rate simulation technique is adopted to simulate the MMC subsystems with different time steps. Additionally, switching interpolation technique is proposed to accurately compensate for the intra-time-step switching events of the MMC multi-valves. To further accelerate the EMT simulation of MMC, a hybrid parallel computing EMT solver is implemented, using both central and graphical processing units (CPU-GPU). The accuracy of the proposed D-DEM is validated against a detailed model (DM) using Simulink/Simscape Electrical toolbox with 1 µs time step. The simulation efficiency of the proposed D-DEM is faster than the conventional DEM model by a factor of 2.81 with CPU-only implementation for the MMC with 400 submodules per arm. Furthermore, the proposed hybrid CPU-GPU solver achieves 79-fold faster than the CPU-only sequential implementation.


  • Universal Decoupled Equivalent Circuit Models of Solid-State Transformer for Accelerated EMT‐Type Simulation

    by Hengyu Li, Walid Hatahet, Jared J. Paull, Jintao Han, Liwei Wang, Wei Li

    Multilevel Multimodule Solid-State Transformer (SST) is emerging as a key technology interfacing MVAC and LVAC systems via chainlink AC-DC converter and Dual Active Bridge (DAB) DC-DC converters. The SST has the advantages of high modularity, bidirectional power transfer, galvanic isolation, and high-frequency power conversion. Fast control prototyping of the SST requires numerically efficient and accurate equivalent circuit models for Electromagnetic Transient (EMT) simulation. This paper proposes universal decoupled equivalent circuit models using switching function to simplify the power converter circuits. Various types of power converters including full-bridge converter, DAB DC-DC converter, and three-phase 3-level converters can be universally modeled by the proposed equivalent circuits in both deblocking and blocking modes. The use of switching function and DC-link decoupling strategy in equivalent circuit models achieves constant G-matrix and significantly reduced node number in the EMT solution. A switching interpolation technique is proposed for the SST to accurately represent switching events when a large simulation time step is used. Case studies have demonstrated that the proposed equivalent circuit models significantly improve the numerical efficiency, compared to the conventional detailed model and variable G-matrix detailed equivalent model.

  • High-Order Exponential Integrator Algorithm for Real-Time Simulation of Power Electronic Systems

    by Jared Paull, Hengyu Li, Liwei Wang, Wei Li

    Real-Time (RT) Electromagnetic transient (EMT) simulation is growing in popularity for hardware-in-the-loop (HIL) simulation of physical hardware devices. This paper proposes a variable-step high-order RT exponential integrator (RTEI) algorithm that is well suited for HIL studies, particularly for controller HIL (CHIL) studies with digital controllers. The RTEI algorithm relies on precomputation of matrix exponentials to offload runtime complexity. The RTEI algorithm works at each intra-control cycle and aims to only recalculate system states at switching events. The proposed solver flexibly adapts the number of forcing function terms to achieve high accuracy with minimum number of computed points. The proposed algorithm is L-stable, making it generally applicable to power electronic systems. A case study validates the accuracy of the proposed algorithm by comparing with hardware experimental results. Further case studies benchmark computational efficiency of the proposed solver with fixed-step Trapezoidal rule-backward Euler solver (TR-BE), ART5, and an existing discrete hybrid time-step (DHT) algorithm. It is shown that the proposed algorithm achieves over 10-fold and 2-fold efficiency increase to TR-BE and DHT, respectively. The advantages of the proposed algorithm lie in both efficiency increases for general circuit topologies and the stability of the algorithm in the presence of high network stiffness.