TL;DR
This paper presents a novel simulation-based optimization framework to determine the optimal placement and controller parameters for Battery Energy Storage Systems (BESSs) to effectively damp power system oscillations. By co-optimizing these factors using a Mixed-Integer Particle Swarm Optimization (Mixed-PSO) algorithm, the approach aims to enhance grid stability while considering practical constraints like BESS limitations and seasonal load variations.
What Problem Does This Paper Address?
Interconnected power systems are susceptible to low-frequency oscillations (0.2-2.5 Hz), which can destabilize the grid, especially with weak tie-lines. Traditional methods like tuning Power System Stabilizers (PSS) on generators are complex due to coordination challenges across multiple entities and communication delays. While Flexible AC Transmission Systems (FACTS) devices offer some damping, they primarily control reactive power. Battery Energy Storage Systems (BESSs) offer a promising solution for active power control to improve damping. However, a significant challenge lies in determining where to place these BESSs and how to tune their controllers simultaneously to achieve optimal oscillation damping without negatively impacting other system modes. Existing research often focuses on either placement or controller design in isolation, or on economic aspects rather than stability enhancement.
Methods
The core of this paper is a simulation-based optimization framework that integrates a Mixed-Integer Particle Swarm Optimization (Mixed-PSO) algorithm with time-domain simulations.
- BESS Modeling: The BESS power converter interface is modeled with dynamic characteristics, including power limits, State of Charge (SOC) limits, and time constants. This captures the practical operational constraints of BESSs.
- Mixed-Integer Optimization: A mixed-integer optimization problem is formulated to co-optimize BESS locations (binary decision: place or not) and controller gains (continuous values).
- Mixed-PSO Algorithm: A specialized Mixed-PSO algorithm is employed to tackle this non-convex, nonlinear, mixed-integer problem. This algorithm uses a direct-integer encoding for locations and real-number encoding for controller gains. It also incorporates a mechanism to handle duplicate location assignments.
- Simulation-Based Evaluation: The optimization process is coupled with a power system simulator (DIgSILENT) via MATLAB. For each potential solution (particle in PSO), the BESS placement and controller parameters are applied to the power system model, and a time-domain simulation is run to evaluate the objective function (minimizing controller gains) and check constraints (e.g., target damping ratio improvement).
- Modal Analysis: The damping ratio of the target oscillation mode is extracted from simulation results using the TLS-ESPRIT method, which is known for its robustness against noise.
Datasets
The proposed method is evaluated on two standard power system test cases:
- New England 39-bus system: A well-known benchmark system for power system studies.
- Nordic test system: A larger, more complex system representing a part of the northern European power grid.
The paper also considers variations of these systems to account for seasonal load changes.
Key Insights and Ideas
- Co-optimization is Crucial: The paper strongly suggests that simultaneously optimizing BESS placement and controller parameters yields superior results compared to optimizing them independently. This is because the effectiveness of a BESS controller is highly dependent on its location within the grid.
- Simulation-Based Approach for Realism: Using time-domain simulations within the optimization loop allows for the accurate consideration of nonlinearities and practical constraints of BESSs and the power system, which is a significant advantage over purely analytical or linearized methods.
- Robustness to Operating Conditions: The optimal BESS placement strategy demonstrates robustness across different operating scenarios (e.g., varying load and generation levels), with many of the optimal locations remaining consistent. This is important for practical grid deployment.
- Cost-Benefit Analysis: The study provides a framework for determining the minimum number of BESS units required for adequate damping support, balancing the investment cost of converters and battery cells.
- Improved PSO for Efficiency: The paper introduces an enhancement to the PSO algorithm by using generator locations and nearby buses for initial population seeding, which speeds up convergence without sacrificing optimality.
Key Results
- Effective Damping Improvement: The proposed method successfully improved the damping ratio of a target inter-area oscillation mode to above 5% in both the NE 39-bus and Nordic test systems, with minimal impact on other modes.
- Optimal Placement Identified: For the NE 39-bus system, three BESS units were found to be optimal, located at buses 35, 36, and 38. For the Nordic system, three units at buses g19, g20, and 4063 were identified.
- Cost-Effective Solutions: The cost analysis indicated that for the NE 39-bus system, two BESS units might be sufficient, while the Nordic system requires at least three units for adequate damping.
- Controller Superiority: The proposed BESS controller demonstrated better performance than a standard PI controller, achieving higher damping ratios with less power overshoot, leading to potentially lower investment costs.
- Convergence Speed-up: The improved PSO algorithm converged earlier (e.g., 19 iterations vs. 23 for the NE system) compared to the standard Mixed-PSO.
Limitations and Future Work
- Computational Complexity: While the improved PSO offers faster convergence, simulation-based optimization can still be computationally intensive, especially for larger systems or more complex BESS models.
- Generalizability of Cost Model: The cost model used for BESS unit determination is based on specific assumptions and unit costs (e.g., from Tesla Powerwall). Future work could explore more comprehensive and adaptable cost models.
- Handling Multiple Target Modes: The current formulation focuses on damping a single target inter-area mode. Extending the optimization to simultaneously improve damping for multiple critical modes would be a valuable future direction.
- Integration with Other Control Objectives: Future research could explore integrating BESS placement and control for oscillation damping with other grid objectives, such as frequency regulation or voltage support.
- Real-world Validation: While verified by time-domain simulations, real-world implementation and validation on actual power grids would be the ultimate test of the proposed method's effectiveness.
Summary updated