Energy Yield Maximization of Photovoltaic Module under Dynamic Irradiance

Photovoltaic (PV) energy is an attractive source of electricity: its cost has been dropping significantly and its market is growing rapidly. Both these reasons have come to mean that PV contributes majorly to energy transition to renewable energy, for which there is a dire need to limit the global increase in temperature. As the global demand for electricity increases due to electrification, economical growth and population growth, increased energy efficiency becomes necessary. The goal of this thesis has been to conceive various ways of boosting the energy efficiency of the PV module under realistic operating conditions. Common PV modules have been designed to achieve a high energy yield in ideal or near-ideal environmental conditions. They typically have simple electrical interconnections only guaranteeing a good energy yield if the entire module is illuminated uniformly. When there is non-uniform irradiance or partial shading across a system, the yield is impacted negatively. A typical PV module is also controlled by a basic maximum power point tracker (MPPT) that only achieves a high energy yield under static or quasi-static operating conditions. These systems degrade considerably in performance under dynamic irradiance conditions. To improve the efficiency of partially shaded PV modules, we develop the idea of optimized static configurations. Since shades on PV modules are often caused by stationary objects, they are often repetitive in nature and thus predictable. Based on this observation, we optimize the electrical interconnections of a PV module for a given shade scenario. The resulting optimized static configurations achieve a significantly higher energy yield for several considered shade scenarios than the conventional PV module and have a much smaller complexity than dynamic reconfiguration of the interconnections. For operation under dynamically varying irradiance conditions, we propose a novel MPPT algorithm based on probabilistic reasoning. Commonly used trackers suffer from oscillations, noisy measurement conditions, and divergence during dynamic conditions. To overcome these limitations, we develop a tracking algorithm based on Bayesian inference, which is an efficient method for performing probabilistic reasoning, and derive two variants: one for uniform irradiance conditions and one for partial shading conditions. These Bayesian inference based MPPT (BI-MPPT) algorithms outperform the commonly used MPPTs under a wide variety of operating conditions. We, finally, perform a tracker speed analysis to determine the hardware requirements for different static configurations and MPPT algorithms. In this analysis we consider the impact of the tracker sampling interval, which affects the hardware complexity, and the slowness of the tracker algorithm, which describes how well a tracker can follow the maximum power point, for a broad range of operating conditions. The analysis reveals that the choice of employed configuration and MPPT algorithm not only affects a module's efficiency, but also the sampling interval and thus hardware complexity that it requires.

Jaar van publicatie:2021

Toegankelijkheid:Open