Piece-Wise Linear (PWL) Probabilistic Analysis of Power Grid with High Penetration PV Integration

Giambattista Gruosso, Luca Daniel, Paolo MaffezzoniEnergies 2022 · DOI

TL;DR

This paper introduces a novel and efficient method for analyzing the probabilistic behavior of power grids with high solar (PV) integration. By combining Gaussian Mixture Models for PV uncertainty with piece-wise linear approximations of power flow, the approach accurately predicts grid behavior and identifies critical operating conditions, offering significant speedups over traditional methods.

What Problem Does This Paper Address?

As solar power (PV) becomes more prevalent in electricity grids, it introduces significant uncertainty into power balance calculations. This uncertainty can jeopardize grid stability and operation, leading to issues like overvoltage or unacceptable voltage fluctuations. Traditional methods for analyzing these probabilistic effects, such as Monte Carlo simulations, are computationally intensive and slow to converge, especially when dealing with a large number of distributed PV sources. The paper aims to address this by developing a more efficient and accurate probabilistic analysis tool that can handle the complexities of high PV penetration.

Methods

The core of the proposed method involves two key components:

  1. Gaussian Mixture Model (GMM) for PV Uncertainty: The paper models the uncertainty of multiple, correlated PV sources using a GMM. This approach leverages historical PV power data to create a statistical model that captures both the individual PV source behavior and their interdependencies. The GMM provides a flexible way to approximate the joint probability distribution of PV power generation.
  2. Piece-Wise Linear (PWL) Approximation of Power Flow: Instead of relying on complex, non-linear power flow equations for every scenario, the method approximates the relationship between PV injection and grid voltages using local, piece-wise linear models. These linearizations are strategically placed around the most probable operating points identified by the GMM. This significantly simplifies the calculations while maintaining accuracy, especially in regions where the probability of PV generation is high. The method also leverages voltage sensitivity analysis to build these local linear models.

Datasets

The paper utilizes a publicly available dataset of PV measurements from the DKA Solar Centre in Alice Springs, Australia, to train and validate the GMM for PV sources. For evaluating the overall method, a modified three-phase version of the IEEE 69 bus test case is used, which is a standard benchmark in power systems analysis.

Key Insights and Ideas

  • PWL as a "Smart" Approximation: The key insight is that while the overall voltage-power relationship in a grid is non-linear, it can be well-approximated by linear segments in regions of high probability. This is because PV generation uncertainty is often concentrated around certain typical values, and the grid's response in these regions is more predictable.
  • GMM Centers as Approximation Points: The centers of the Gaussian components in the GMM naturally identify the most probable operating points for PV generation. These points become the ideal locations to build the piece-wise linear approximations, ensuring accuracy where it matters most.
  • Direct PDF Calculation: A significant advantage of this approach is its ability to directly derive a closed-form expression (a GMM) for the joint probability distribution of observable grid variables (like node voltages). This is a major improvement over methods that only provide statistical moments or require extensive sampling.
  • Curse of Dimensionality Mitigation: The PWL approach effectively circumvents the "curse of dimensionality" that plagues other surrogate modeling techniques (like Polynomial Chaos expansions) when dealing with a large number of uncertain parameters (i.e., many PV sources).

Key Results

  • High Accuracy: The proposed PWL method achieves excellent accuracy, with the calculated probability density functions (PDFs) of node voltages closely matching those obtained from reference Monte Carlo simulations (relative difference always < 0.5%).
  • Significant Speedup: The method demonstrates remarkable computational efficiency. It achieves a 200x speedup compared to Monte Carlo simulations and a 50x speedup compared to a Copula+gPC method for probabilistic load flow analysis. This is achieved with a drastically reduced number of load flow simulations (e.g., 18 vs. 10,000 for MC).
  • Identification of Critical Buses: The analysis effectively identifies critical buses in the grid that experience high voltage uncertainty, particularly under high PV penetration.
  • PV Hosting Capacity Assessment: The method provides a robust framework for assessing the probability of violating safe operating conditions (e.g., overvoltage) as a function of PV penetration, enabling better estimation of PV hosting capacity.

Limitations and Future Work

  • Generalization to Other Sources: While the paper demonstrates the method for PV sources, it notes that the methodology can be generalized to other renewable energy sources.
  • Three-Phase Unbalanced Grids: The numerical experiments were performed on a modified three-phase version of the IEEE 69 bus grid, which was initially a single-phase equivalent. While the method is presented as general, further validation on more complex, inherently unbalanced three-phase grids might be beneficial.
  • Future Work: The authors suggest that this efficient PWL analysis tool can be valuable for complex problems like estimating PV hosting capacity. Further exploration into generalizing the method to other renewable sources and more complex grid topologies would be a natural next step.

Summary updated

Abstract

This paper aims at presenting a novel effective approach to probabilistic analysis of distribution power grid with high penetration of PV sources. The novel method adopts a Gaussian Mixture Model for reproducing the uncertainty of correlated PV sources along with a piece-wise-linear approximation of the voltage-power relationship established by load flow problem. The method allows the handling of scenarios with a large number of uncertain PV sources in an efficient yet accurate way. A distinctive feature of the proposed probabilistic analysis is that of directly providing, in closed-form, the joint probability distribution of the set of observable variables of interest. From such a comprehensive statistical representation, remarkable information about grid uncertainty can be deduced. This includes the probability of violating the safe operation conditions as a function of PV penetration.
BibTeX
@inproceedings{gruosso2022piece,
  author = {Giambattista Gruosso and Luca Daniel and Paolo Maffezzoni},
  title = {{Piece-Wise Linear (PWL) Probabilistic Analysis of Power Grid with High Penetration PV Integration}},
  booktitle = {Energies},
  year = {2022},
  doi = {10.3390/en15134752},
}
Figure