Data-Driven Framework for Quantification of Reactive Power Using Aggregated Load Data

Introduction

The development of the upcoming decentralized power grid, which will include small-scale power systems made up of distributed generation, flexible loads, and storage systems, has been supported by the surge in grid-connected renewable energy sources. However, dynamic changes in the load and random nature of renewable energy sources result either in excess energy production or energy deficit. Either way, the utility loses revenue besides facing other risks. Further, active and reactive powers need to be managed separately to maintain the voltage and frequency stabilities of the grid. In other words, real-time management of modern smart grids is a complex task. To raise energy production levels, system operators need an accurate estimation of the load, which is generally obtained from the meters from various parts of the power grid. For managing the energy pool over the next day or over the next week, load needs to be predicted or forecasted. The tasks of load estimation (for real-time) and forecasting (for immediate future) is typically done on aggregated load data. Modern grids do employ well-trained professionals to attend the task of aggregation. The presence of an aggregator serving as a middleman between the prosumers, distribution and transmission network operators is the key to manage the supply-demand mismatch. In the case of larger networks, an independent market operator is employed who controls the aggregated resources to achieve a specific load demand modification, defines ongoing demand response programs (Musti, 2020a; Öztürk et al., 2022). Thus, managing the supply-demand mismatch requires aggregation of both loads and generation sources (typically in the intervals of 15 minutes) that are connected to the grid in different geographical locations. It is clear that a lot of data is being collected and monitored for aggregation purposes. In other words, management of smart grids is data intense activity as several decisions need to be taken based on the dynamically varying data (Musti, 2020b). Though there are several components, stakeholders and influencing factors associated; four major aspects – energy markets, load data obtained from different locations, computational processes and the key decisions required be to made – are the important ingredients in managing the supply-demand mismatch. This chapter deals with these major aspects in order to explain the importance of data, computations and the process of making the decisions in operating the modern smart grids.

Reactive power is a vital component in managing the inductive and capacitive loads. Without adequate reactive power, transformers (responsible for energy transformation) cannot function. Voltage levels are essentially managed either by increasing or decreasing the reactive power supply (Georgios et. al, 2022). In most cases, reactive power needs to be supplied as a resource to keep the voltage stability within the limits. The task of controlling the reactive power supply needs aggregated data from various load points. Since this is a complex task, reactive power can be provided as an ancillary service in power markets. Ancillary services are additional services beyond the supply of energy that are required to maintain the reliability and security of the power system. Reactive power as an ancillary service can be provided by generators, synchronous condensers, or static VAR compensators. These resources can be dispatched by the system operator to provide reactive power support as needed to maintain system stability. Reactive power control needs to be done in real-time and thus the process assumes high significance. Control extents are determined through complex computations that require a lot of data. If the reactive power is not managed well in real-time, the system voltage fluctuates and this in-turn can result in equipment damage (Falabretti et al., 2022).