Abstract
Synchronous generator systems are complex dynamical systems essential for energy transmission in modern industrial production. This article proposes predictor-corrector methods and structure-preserving methods for a generator system based on the first benchmark model of subsynchronous resonance. The structure-preserving methods preserve a Dirac structure associated with port-Hamiltonian descriptor systems. To demonstrate this, a simplified generator system in the form of index-1 differential-algebraic equations (DAEs) is derived. Our analysis provides global error estimates for Gauss methods, a special class of structure-preserving methods, which guarantee superior performance over industry-standard PSCAD/EMTDC and predictor-corrector methods in terms of computational stability. Numerical simulations verify the effectiveness and advantages of these proposed methods.
The Challenge of Modular Simulation
Simulating the electromechanical transients of power systems is a cornerstone of power engineering, yet it remains notoriously difficult. Traditional "Electro-Magnetic Transient Program" (EMTP) software typically decomposes the simulation into three independent modules: the electrical circuit, the generator, and the mechanical shaft.
While this modularity simplifies the design, it introduces a "one-time-step delay" as data is exchanged between these disparate modules. This delay leads to a cumulative decline in numerical accuracy. Furthermore, standard models often treat network equations as algebraic constraints, meaning node voltages are not treated as "state variables" within the differential equations. This mathematical formulation makes it difficult to apply sophisticated, high-order numerical solvers. This research bridges this gap by integrating these components into a single, mathematically consistent framework.
Methodology: Unified Port-Hamiltonian Modeling
The researchers propose a unified model that reformulates the generator equations into a port-Hamiltonian descriptor system. This framework is powerful because it explicitly accounts for energy dissipation—such as friction and resistance—while preserving a fundamental geometric property known as a Dirac structure.
Two main categories of numerical integrators were developed and tested:
- Predictor-Corrector (P-C) Methods: These are generalizations of the Adams-Bashforth-Moulton method. They utilize a fast explicit "predictor" step (similar to the Euler method) followed by an implicit "corrector" step (trapezoidal rule). While computationally efficient, these methods can struggle with long-term stability in complex systems.
- Structure-Preserving Gauss Methods: This represents the core innovation of the paper. By applying s-stage Gauss collocation methods—a specific class of implicit Runge-Kutta methods—the researchers can maintain the system's underlying Dirac structure during discretization.
Key Insights: Geometric Preservation and Stability
The primary finding is that geometric preservation is more critical for long-term simulation stability than simply increasing the order of the solver. By respecting the physical structure of the system, the Gauss methods avoid the numerical "drift" that plagues traditional approaches.
Additionally, the authors successfully reformulated the flux linkage equations to eliminate mathematical singularities. This allowed the system to be expressed as index-1 differential-algebraic equations (DAEs), which are significantly more tractable for numerical analysis. The P-C methods also found success by exploiting the physical reality that mechanical rotary speeds change much more slowly than electromagnetic fluxes, allowing for a hybrid approach that balances speed and accuracy.
Performance and Results
The effectiveness of these methods was tested using the IEEE First Benchmark Model for Subsynchronous Resonance. The results demonstrated clear advantages over existing standards:
- Short-Term Accuracy: In simulations spanning 0–10 seconds, the proposed methods converged rapidly to equilibrium. In contrast, the professional software PSCAD/EMTDC exhibited significantly larger fluctuations.
- Long-Term Stability: In extended tests (up to 1500 seconds), the "Predictor-Corrector I" method eventually diverged (blew up). However, the Structure-Preserving Gauss method remained perfectly stable throughout the duration.
- Error Reduction: The structure-preserving method maintained an error level approximately half that of the best Predictor-Corrector method throughout the 1500-second simulation.
Conclusion and Future Work
This research provides a rigorous theoretical foundation for the use of Gauss methods in power system simulations, including proven global error estimates. While these implicit methods are more computationally intensive than simple explicit schemes, their superior stability makes them ideal for high-fidelity transients.
Future work will focus on scaling these structure-preserving methods to large-scale, multi-machine power systems. Testing whether these stability benefits persist as system dimensionality increases will be crucial for implementing these methods in real-world, massive-scale power grids.
Summary updated