Study of models for the nominal power characterization of a photovoltaic generator and the power estimation of different photovoltaic technologies in Lima, Peru

This work investigates two main aspects related to photovoltaic: systems and module characterization and performance modeling. The first part aims to characterize a PV generator located in Spain with a nominal power of 109.44 kW under standard test conditions according to the datasheet. An operational photovoltaic system's nominal power is a valid parameter for determining its current operational state. The applicability of a standard procedure to estimate the nominal power of an operating generator, proposed by Martínez-Moreno and based on Osterwald's model, is investigated. However, the standard procedure does not specify how to deal with experimental data when unexpected behavior impedes the nominal power estimation under operating conditions. During the 6-month study, the power-irradiance relation showed a hysteresis effect with varying amplitudes throughout the campaign. Adding a data filter that removes the non-linear part of the data proves necessary to estimate the nominal power, complementing Martinez-Moreno's procedure to enable the generators' characterization. The second part contributes to closing a knowledge gap in the performance behavior and predictability of multiple PV technologies in Peru. The quality of two simple analytical models for estimating the outdoor performance of three different photovoltaic module technologies in Lima was investigated. Osterwald's and the Constant Fill Factor models were applied to estimate the maximum power delivered by an Aluminum Back Surface Field, a Heterojunction with Intrinsic Thin-layer, and an amorphous/microcrystalline thin-film tandem PV module. The results point that both models overestimate the expected power compared to the measured one. Implementing a correction factor adjusts the estimated maximum power by both models. This correction factor allows us to estimate losses, calculate an adequate nominal power and minimize the estimated power error. The normalized root mean square error and mean bias error determine the implemented methodology's quality. The two crystalline silicon-based technologies present a similar behavior throughout the year. However, both differ considerably from the tandem one during different months, implying that the ambient variables have other seasonal impacts on their performance