Enhancing the Modeling and Efficiency of Photovoltaic Systems
Solar energy is a strong contender among the sustainable alternatives that offer practical potential for replacing increasingly depleted fossil fuels and supplying the world’s growing energy demands. However, despite its sustainability, the spread of its use has been limited due to the high costs arising from its inadequate efficiency.
With this challenge as motivation, the goal of the research presented in this thesis was to contribute to the expansion of the utilization of photovoltaic (PV) systems. To achieve this goal, the work was approached from two perspectives: 1) facilitation of research into PV systems through the enhancement of existing PV models and simulation tools, which are highly complex and necessitate substantial computational effort, and 2) improvement of the efficiency of PV systems through the development of new techniques that mitigate power losses in PV systems.
With respect to the first perspective, two innovative modeling approaches are introduced. The first, a new circuit model for PV systems, features accuracy comparable to that of previous models but with a reduced computational requirement. The accurate PV models described in the literature are complex and involve considerable computational time due to their dependency on a transcendental implicit equation. Although this aspect poses no problem for examining PV systems that receive homogeneous irradiance, it presents a challenge for studies that include consideration of the impact of partial shading because such studies, in which each PV unit might be subjected to a different irradiance level, entail the simultaneous simulation of numerous connected PV models. This constraint unfortunately results in substantial computational time, which escalates exponentially as the size of the system increases. The proposed model mimics the accuracy of existing models without their dependency on a transcendental implicit equation, thus providing a shorter computational time without scarifying the accuracy. The accuracy of the new model has been validated based on a comparison of its performance with measurement data for a variety of commercially available PV technologies. The results of the case study conducted demonstrate the decrease in computational time and show that the percentage of the reduction improves with increases in the number of PV units in the simulated system, thus providing a clear advantage for the simulation of large PV systems.
The second modeling approach, which was developed for use in model-based online applications, involved the creation of a fast tool for estimating the power peaks of the power-voltage curves for partially shaded PV systems. Utilizing a PV circuit model for estimating the power peaks in large PV systems through the simulation of their entire power curve consumes extensive computational time, which is unacceptable for online applications even with the use of the proposed circuit model mentioned above. Rather than employing a PV circuit model to find the power peaks, the proposed tool relies on simple rules that govern the formation of power peaks in a partially shaded PV system as a means of establishing the power peaks directly, thus significantly reducing the time required. A MATLAB simulation verified the effectiveness of the proposed estimator for accurately determining the power peaks of PV systems, within a computational timespan suitable for model-based online applications.
The second perspective led to the development of three methods for reducing different types of power losses prevalent in PV systems. The first is an MPPT technique for use with partially shaded PV systems that exhibit multiple power peaks in their output power curves. In spite of the ability of existing MPPTs to accurately find the global maximum power point (GMPP) in such systems, they require periodic curve scanning, during which the system is forced to work on non-maximum power points that cause power losses. These losses are referred to as misleading power losses. The proposed MPPT is uniquely distinguishable because of its ability to eliminate these losses in PV systems. Rather than searching and scanning heuristically for the GMPP, it employs the fast modeling tool mentioned above to calculate the location of the GMPP deterministically, thus avoiding the need for curve scanning. The irradiance values required by the modeling tool are estimated innovatively using captured images of the PV modules obtained by an optical camera. Tests conducted with an experimental prototype verified the efficacy of the proposed MPPT for a variety of shading scenarios.
The objective of the second was to reduce the mismatch power losses common in partially shaded PV systems through the development of an improved PV reconfiguration method. Relocating the configuration of PV systems during their operation has been reported in the literature as a successful means of minimizing mismatch power losses. However, existing reconfiguration methods require long computational times in order to determine the optimal configuration, during which time the system continues to work based on a non-optimal configuration, and mismatch power losses are not minimized. In contrast, rather than requiring heavy optimization techniques, the reconfiguration proposed in this thesis is produced by a simple algorithm that establishes the optimal configuration and requires only negligible computational time for ensuring the minimization of mismatch power losses. The effectiveness of the proposed method for reducing mismatch power losses has been verified for a number of shading scenarios based on a comparison of its performance with that of existing methods.
The third is an enhanced maximum power point tracker (MPPT) for reducing tracking power losses in PV systems that operate under rapidly changing irradiance levels. The proposed method combines model-based and heuristic techniques in order to accelerate the tracking speed and thus decrease this type of loss. In the proposed MPPT, the temperature measurements typically necessary in any model-based PV application have been eliminated through reliance on a new set of equations capable of estimating the temperature through the utilization of current and voltage measurements. The effectiveness of the proposed MPPT with respect to reducing tracking power losses was verified with the use of both a MATLAB simulation and an experimental setup.