Agent-Based Analysis of the Impact of Near Real-Time Data and Smart Balancing on the Frequency Stability of Power Systems

TL;DR

This work analyzes how near-real-time (NRT) imbalance data publication and smart balancing influence the frequency stability of power systems. It introduces a dynamic multi-agent framework where each BRP acts as an autonomous agent with a private probabilistic model to estimate the FRR demand $P_{\mathrm{demand}}$ and decide on smart balancing actions within a shared control-block environment, driven by a lookahead $\bm{l}_k$ and NRT feedback. Through Monte-Carlo simulations over thousands of scenarios and historical German control-block data, the study finds that smart balancing can reduce balancing energy and costs but generally increases frequency variability, with effects strongly dependent on data publication type (exact vs interval) and delay (60 s vs 120 s). The results emphasize that finer NRT data publication improves frequency stability and reduces adverse effects, guiding policy and market design on data transparency and balancing incentives. These insights are relevant for system operators and regulators aiming to balance market efficiency with grid reliability in the presence of fast, data-driven BRP responses.

Abstract

Single imbalance pricing provides an incentive to balance responsible parties (BRPs) to intentionally introduce power schedule deviations in order to reduce the control area imbalance and receive a remuneration through the imbalance settlement mechanism. This is called smart balancing or passive balancing and is actively encouraged in, e.g., the Netherlands and Belgium through the publication of near real-time (NRT) data on the control area imbalance by the transmission system operator. It is known that under certain conditions, smart balancing can deteriorate the frequency stability of the power system. This paper examines how the publication of different types of NRT data affects smart balancing and the frequency stability. A Monte-Carlo simulation of a dynamic multi-agent model is performed to analyse the effects of smart balancing with different parameters for the agents and the environment, using historical time series of the power imbalance of the German control block as a basis. It is found that smart balancing can significantly reduce the amount and cost of frequency restoration reserve activation, but leads to a general increase of the frequency variability. Depending on the type of NRT data and agent parameters, the frequency stability margins are also reduced. The negative effects on the frequency stability are stronger when NRT data is published using large bins and with long delays.

Figures (20)

• Figure 1: Fast oscillations in the open-loop in the Netherlands caused by overreactions of on signals ENTSOETransparencyPlatform.
• Figure 2: Overview of the multi-agent system.
• Figure 2: Historical FRR demand for the for selected periods in 2023.
• Figure 3: Single busbar model of the control block.
• Figure 4: Visualization of the publication methods for the NRT imbalance data.