Hybrid and intelligent energy storage systems in standalone photovoltaic applications

Resumen

Remote systems such as communication relays or irrigation control installations cannot usually be powered by the electrical grid. One of the alternatives is to power these systems through solar panels, in what is known as standalone photovoltaic applications. Most of these systems need a continuous operation, but a standalone photovoltaic installation cannot be powered during the night. For this reason, they use batteries to store excess energy during the day. These storage systems have been traditionally based on Valve Regulated Lead Acid (VRLA) batteries, but some effects can alter their performance in terms of reliability, operation cost and maintenance. One of the key issues that alter the energy behavior of the photovoltaic off-grid systems is the Partial State of Charge (PSoC) effect: Batteries cannot be completely charged as manufacturers indicate due to the day-night cycle. This gets the battery into an intermediate state of charge that effectively reduces its capacity, even halving it in some cases. To mitigate the impact of these effects on the installation, batteries tend to be oversized with some security margins. These oversizing factors can be incredibly high and have a huge impact on the deployment and maintenance cost of the facility. The first part of this thesis highlights some of these key concepts, analyzing which of them are critical in specific design cases, modeling them into a simulation tool, and as an outcome, establishing optimal sizing regions for the installations. After the analysis, different ways of improving the performance of the installations are proposed. One idea to mitigate PSoC is to combine different storage technologies in a Hybrid Energy Storage Systems (HESS). HESSs have traditionally combined high energy density elements as batteries with high power density elements as ultracapacitors. An iteration of this idea is carried out throughout this thesis, where different types of batteries are combined. Each of them is best fitted to different power patterns in the application, such as daily cycles or emergency periods. It is possible to further increase the performance by using intelligent algorithms to improve the functionalities of the Battery Management Systems embedded in these applications. To this end, failure prediction and health estimation algorithms are proposed as contributions of this work. These new algorithms endow the HESS with tools to predict possible energy disruption events and to anticipate aging, and thus, act accordingly.