Coordinated Fast Frequency Regulation in Dynamic Virtual Power Plants via Disturbance Estimation
Saif Ahmad, Seifeddine BEN ELGHALI, Hafiz Ahmed · 2025
Read Coordinated Fast Frequency Regulation in Dynamic Virtual Power Plants via Disturbance Estimation by Saif Ahmad, Seifeddine BEN ELGHALI, Hafiz Ahmed (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-06713.txt.
Introduction
Saif Ahmad, Seifeddine Ben Elghali, and Hafiz Ahmed join fast frequency control across linked mini grids, in a 2025 paper. Mini grids pool solar, wind, and stored charge to act as one plant. They must fix dips fast, but stored charge in one node can run dry.
Their plan fixes small dips at once in each node with a watch for outside shocks. When a node runs short, close nodes lend spare charge in a shared step. Tests on a small four bus grid with much green power check out. The layered plan can scale as grids grow more green.
Catalog record
| CREATOR | Saif Ahmad, Seifeddine BEN ELGHALI, Hafiz Ahmed |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 26 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-06713
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Coordinated Fast Frequency Regulation in Dynamic Virtual Power Plants via Disturbance Estimation Thanks: Saif Ahmad and Seifeddine BEN ELGHALI are with Aix Marseille Univ, CNRS, LIS, Marseille, France. Hafiz Ahmed is with Nuclear AMRC, Coventry, United Kingdom. email: [email protected], [email protected], [email protected] Saif Ahmad, Seifeddine BEN ELGHALI, Hafiz Ahmed Abstract In the context of dynamic virtual power plants (DVPPs), the integration of frequency containment reserve (FCR) and fast frequency control (FFC) enabled via local compensation of power imbalance represents a significant advancement in decentralized frequency regulation. However, they still have to cope with the limited power and energy capacities associated with commonly available storage solutions. This work combines a disturbance estimation based decentralized local control with distributed imbalance compensation in the event of local shortfall. The layered architecture facilitates fast local corrections in power setpoints while enabling coordination between neighbouring DVPP nodes to leverage the aggregated capacity, ensuring scalable and efficient operation suitable for renewable-heavy future grids. The proposed approach is validated on an illustrative 4-bus system with a high percentage of renewables. I Introduction Power systems are currently undergoing rapid transformation to address the limitations and adverse environmental impacts of conventional fossil fuel-based energy sources with distributed generators comprising renewable energy sources (RESs) such as solar and wind providing a promising solution to this problem. However, uncoordinated integration of renewables in the electricity grid results in issues such as large voltage/frequency swings, increased operational costs ( e.g. payouts to balancing market entities, spinning reserve costs etc.) to compensate for inaccurate generation/load forecasts and more frequent blackouts due to generator and/or transmission network overloads [ 1 ] . To ensure increased penetration of renewables in the electricity grid and to meet the increasing energy demand with intermittent and unpredictable RESs, controlled aggregation of distributed RESs (DRESs) in the form of microgrids (MGs) [ 2 ] and virtual power plants (VPPs) [ 3 ] , energy communities etc. have emerged as a powerful tool to balance the uncertainty in production and allow small to medium capacity generators as well as prosumers to monetize their flexibility by participating in the energy, capacity and flexibility markets thereby generating community interest [ 4 ] . Furthermore, it is expected that a bulk of the ancillary services in the future power grids will have to be shouldered by non-synchronous DERs, implying the significance of these aggregated systems in ensuring grid stability and reliability. In particular, the concept of dynamic virtual power plants (DVPP) has been proposed to pave the way for DERs to provision future ancillary services [ 5 ] . In this work, we focus on leveraging the potential of DVPPs for frequency regulation in renewable-heavy power grids. The absence of rotational kinetic energy in DRESs and heterogeneous inertia distribution in the network results in faster frequency dynamics, thereby making the task all the more challenging [ 6 ] . On the other hand, when controlled properly, VPPs can be leveraged to provide inertia support via grid-forming IBRs to reinforce the system response in the event of contingencies. Traditionally, frequency regulation was implemented in a centralised manner via automatic generation control (AGC) where a central authority generates regulatory signals. Despite being effective for decades, this approach is inherently limited in response speed, making it unsuitable for fast frequency regulation [ 7 ] . In recent years, several studies have investigated the impact of increased penetration of RESs into the grid infrastructure and proposed IBR-based frequency regulation schemes, some of which are briefly discussed here. Model-free learning based techniques, such as those based on reinforcement learning (RL) [ 8 , 9 ] , are usually without any stability certificates and work by imposing soft penalties for deviations outside a predefined range, which restricts their application on real-time systems. Furthermore, RL-based techniques require massive training datasets, which compounds implementation complexity. On the other hand, model-based techniques such as model predictive control can satisfy hard constraints during transients, however, this capability relies heavily on the accuracy of the system model and places significant communication overheads [ 10 , 11 ] . Another model-based approach leveraging disturbance estimator (DE) was introduced in [ 11 ] for fast frequency regulation, in combination with distributed optimization based on alternating direction method of multipliers (ADMM) to overcome the shortfall in local RESs by coordinating with neighbouring assets. However, the coordination of neighbouring local area controllers was implemented through a central coordinator which significantly increases the communication overhead and impacts scalability for large systems. In this work, we follow the general framework adopted in [ 11 ] , where the power system is partitioned into smaller areas (DVPP nodes in our case) having unmeasured active power and then a controller is designed to coordinate frequency regulation in the aggregated system. In contrast to [ 11 , 7 ] , the disturbance estimator in our work allows a more general internal model for the unmeasured active power while the coordinating layer is completely distributed, requiring communication only between neighbouring nodes. This, in turn, improves the system’s reliability and preserves data privacy. The remaining sections in this paper are organised as follows: Section II introduces the modeling of electrical network and different nodes using the swing equation, followed by the design of a disturbance estimation mechanism for local control and distributed coordination in Section III . Section IV presents a numerical study to evaluate the efficacy of the proposed approach and conclusions are drawn in Section V . I-A Notations: \bm{I_{n}} is an identity matrix of dimension n\times n , 0 is a zero vector or matrix of appropriate dimension, \textrm{diag}(a_{1},\dots,a_{n}) denotes a diagonal matrix with a_{1},\dots,a_{n} as diagonal elements. Given a symmetric matric M , \textbf{M}\prec\textbf{0} means that M is negative-definite. The notation for positive definiteness is analogous. II Mathematical Model We consider the electrical grid to be composed of a network of interconnected nodes, where the local frequency dynamics of each node is defined by a swing equation. Nodes can be of different types depending on the constituting elements and available regulation mechanisms. In this work, we distinguish between two types of nodes, which are discussed in the following subsections. The terms ‘nodes’ and ‘buses’ are used interchangeably throughout the paper, with nodes being used mostly in the modeling and algorithmic part and buses in the numerical validation part to link with the physical system. II-A Dynamic Virtual Power Plant Nodes The local frequency dynamics of the i^{\text{th}} DVPP node can be defined by a standard swing equation of the form [ 12 ] \begin{split}\dot{\theta}_{i}&=\omega_{i}^{\delta}\\ \dot{\omega}_{i}^{\delta}&=\dfrac{1}{2H_{i}}\delta_{i}^{P}\end{split} (1) where \omega_{i}^{\delta} is the frequency deviation (p.u.), H_{i} is inertia constant (seconds) and \delta_{i}^{P} is the net power imbalance (p.u.). The DVPP nodes are considered to be without synchronous generators (SGs), and are characterized by low inertia and damping provided mostly by IBRs and rotating machines such as large motors. The power imbalance can be expressed as \begin{split}\delta_{i}^{P}=-P_{i}^{e}+P^{?}_{i}+P^{RES}_{i}-P_{i}^{FCRm}-u^{m}_{i}-\sum_{j}P_{i\to j}\end{split} (2) where P_{i}^{e} is electrical load, P^{?}_{i} is the unmeasured active power injected, P^{RES}_{i} is the power injected by RESs including storage, P_{i}^{FCRm} is power injected by the assets participating in the frequency containment reserve (FCR) (discussed in Section II-A1 ), u^{m}_{i} is the power consumed by the battery energy storage system (BESS) and \sum_{j}P_{i\to j} is