Characterization and Diagnostics for Photovoltaic Modules and Arrays

Research output: Book/ReportPh.D. thesis

Abstract

Photovoltaic energy has become one of the most important renewable energy technologies, reaching a global cumulative installed capacity of about 140 GW at the end of 2013, and at least 174 GW forecasted by the end of 2014 by EPIA. This exponential growth was possible due to favorable feed-in tariff policies, as well as decreasing cost of PV modules and balance-of-system components. As these main market drivers reach their limit, the system operating costs and long-term reliability of the PV modules becomes more relevant in reducing the total lifetime cost of the PV system. In this context, characterization and diagnostic methods are increasingly important in identifying and understanding the failures and degradation modes affecting PV modules and arrays, as well as developing relevant tools and tests for assessing the reliability and lifetime of PV modules.

This thesis investigates diagnostic methods for characterizing and detecting degradation modes in crystalline silicon photovoltaic modules and arrays, and is structured into two parts. The first part of this work is focused on developing PV module characterization and diagnostic methods for use in module diagnostics and failure identification, accelerated stress testing, and degradation studies. The second part of this work investigates diagnostic methods for PV arrays that are suitable for implementation in the solar inverter or incorporated in a condition monitoring system.

The PV module diagnostic methods investigated in the first part of this work were developed based on two well-known module characterization techniques, namely current-voltage (I-V) characterization, and electroluminescence imaging.

he I-V based module diagnostic methods were developed by combining the strengths of light I-V and dark I-V characterization, for the purpose of identifying degradation modes such as: 1) optical losses; 2) cell cracks and breaks; 3) degradation of the external circuit of the PV module; 4) potential-induced degradation (PID). This method, which is machine analysis friendly, can identify incipient degradation in modules, which would otherwise be difficult to detect from light I-V measurements alone. The method can be used as a laboratory diagnostic tool, and can be also implemented in future field applications, for example in: 1) in I-V tracers for PV degradation studies; 2) or as a diagnostic function in module-integrated converters.

A similar approach was used in developing an in-situ power loss estimation method for modules undergoing PID stress testing. This method implies semi-continuous dark I-V characterization of the PV modules under test, at elevated stress temperature, without interrupting the test. The dark I-V curves are then used to determine the degradation of the module’s maximum power, at standard testing conditions (STC 1000 W/m2 25°C), as a function of time and stress level. This leads to reduced test duration and cost, avoids stress transients while ramping to and from the stress temperature, eliminates flash testing except at the initial and final data points, and enables significantly faster and more detailed acquisition of statistical data for future application of various statistical reliability models.

The second module diagnostic technique investigated is electroluminescence (EL) imaging. EL Imaging is an excellent tool for visually (qualitatively) identifying different types of failures and degradation in PV modules, however, currently there is no standard methodology to quantify them. In this study a method was developed for quantifying the extent of different failures and degradation modes present in crystalline silicon PV modules from EL images. The method relies on identifying specific failure signatures in EL images, by analyzing the luminescence maps and distributions, as well as applying the fundamental diode model to analyze the distributed solar cell parameters. The method is suitable for automatic analysis and assessment of module quality from EL images, for the purpose of PV module diagnostics and module quality inter-comparison.
The second part of this work was focused on developing diagnostic methods for detecting failures and degradation occurring in PV arrays. This research resulted in two types of diagnostic methods that can be implemented in a PV system, depending on the hardware available.

The first PV array diagnostic method proposed uses a model-based approach, and is suitable for PV systems where irradiance and module temperature sensors are available. The method is based on estimating the power output of the PV generator from in-plane irradiance and module temperature measurements, and using a performance model of the generator.

The novelty of the method comes from its ability to (automatically) self-parameterize the performance model from measurements acquired by the PV inverter after the PV system has been commissioned, and does not require a dedicated system modelling effort.

The second type of PV array diagnostic method investigated can operate without ambient sensors, and is based on measuring and analyzing parameters of the light I-V curve of the array. In comparison with yield measurements, I-V curves can provide a much more information regarding the condition and electrical properties of the PV generator, such as: short-circuit current, open-circuit voltage, fill factor, series and shunt resistance, ideality factor, as well as indicate the presence of shading and soiling. Considering this, the diagnostic method proposes to calculate certain diagnostic parameters from the I-V curve that can indicate the presence of shading, increased series-resistance losses, or PID. These parameters were then used to design a fault detection algorithm, based on fuzzy inference systems that can be implemented in the power electronic converter. Experimental results show that the I-V based diagnostic method performs well in identifying both shading and increased series losses affecting PV arrays. The method requires low computation resources and it uses hardware capabilities that are already present in most PV inverters.
Close

Details

Photovoltaic energy has become one of the most important renewable energy technologies, reaching a global cumulative installed capacity of about 140 GW at the end of 2013, and at least 174 GW forecasted by the end of 2014 by EPIA. This exponential growth was possible due to favorable feed-in tariff policies, as well as decreasing cost of PV modules and balance-of-system components. As these main market drivers reach their limit, the system operating costs and long-term reliability of the PV modules becomes more relevant in reducing the total lifetime cost of the PV system. In this context, characterization and diagnostic methods are increasingly important in identifying and understanding the failures and degradation modes affecting PV modules and arrays, as well as developing relevant tools and tests for assessing the reliability and lifetime of PV modules.

This thesis investigates diagnostic methods for characterizing and detecting degradation modes in crystalline silicon photovoltaic modules and arrays, and is structured into two parts. The first part of this work is focused on developing PV module characterization and diagnostic methods for use in module diagnostics and failure identification, accelerated stress testing, and degradation studies. The second part of this work investigates diagnostic methods for PV arrays that are suitable for implementation in the solar inverter or incorporated in a condition monitoring system.

The PV module diagnostic methods investigated in the first part of this work were developed based on two well-known module characterization techniques, namely current-voltage (I-V) characterization, and electroluminescence imaging.

he I-V based module diagnostic methods were developed by combining the strengths of light I-V and dark I-V characterization, for the purpose of identifying degradation modes such as: 1) optical losses; 2) cell cracks and breaks; 3) degradation of the external circuit of the PV module; 4) potential-induced degradation (PID). This method, which is machine analysis friendly, can identify incipient degradation in modules, which would otherwise be difficult to detect from light I-V measurements alone. The method can be used as a laboratory diagnostic tool, and can be also implemented in future field applications, for example in: 1) in I-V tracers for PV degradation studies; 2) or as a diagnostic function in module-integrated converters.

A similar approach was used in developing an in-situ power loss estimation method for modules undergoing PID stress testing. This method implies semi-continuous dark I-V characterization of the PV modules under test, at elevated stress temperature, without interrupting the test. The dark I-V curves are then used to determine the degradation of the module’s maximum power, at standard testing conditions (STC 1000 W/m2 25°C), as a function of time and stress level. This leads to reduced test duration and cost, avoids stress transients while ramping to and from the stress temperature, eliminates flash testing except at the initial and final data points, and enables significantly faster and more detailed acquisition of statistical data for future application of various statistical reliability models.

The second module diagnostic technique investigated is electroluminescence (EL) imaging. EL Imaging is an excellent tool for visually (qualitatively) identifying different types of failures and degradation in PV modules, however, currently there is no standard methodology to quantify them. In this study a method was developed for quantifying the extent of different failures and degradation modes present in crystalline silicon PV modules from EL images. The method relies on identifying specific failure signatures in EL images, by analyzing the luminescence maps and distributions, as well as applying the fundamental diode model to analyze the distributed solar cell parameters. The method is suitable for automatic analysis and assessment of module quality from EL images, for the purpose of PV module diagnostics and module quality inter-comparison.
The second part of this work was focused on developing diagnostic methods for detecting failures and degradation occurring in PV arrays. This research resulted in two types of diagnostic methods that can be implemented in a PV system, depending on the hardware available.

The first PV array diagnostic method proposed uses a model-based approach, and is suitable for PV systems where irradiance and module temperature sensors are available. The method is based on estimating the power output of the PV generator from in-plane irradiance and module temperature measurements, and using a performance model of the generator.

The novelty of the method comes from its ability to (automatically) self-parameterize the performance model from measurements acquired by the PV inverter after the PV system has been commissioned, and does not require a dedicated system modelling effort.

The second type of PV array diagnostic method investigated can operate without ambient sensors, and is based on measuring and analyzing parameters of the light I-V curve of the array. In comparison with yield measurements, I-V curves can provide a much more information regarding the condition and electrical properties of the PV generator, such as: short-circuit current, open-circuit voltage, fill factor, series and shunt resistance, ideality factor, as well as indicate the presence of shading and soiling. Considering this, the diagnostic method proposes to calculate certain diagnostic parameters from the I-V curve that can indicate the presence of shading, increased series-resistance losses, or PID. These parameters were then used to design a fault detection algorithm, based on fuzzy inference systems that can be implemented in the power electronic converter. Experimental results show that the I-V based diagnostic method performs well in identifying both shading and increased series losses affecting PV arrays. The method requires low computation resources and it uses hardware capabilities that are already present in most PV inverters.
Original languageEnglish
PublisherDepartment of Energy Technology, Aalborg University
Number of pages141
StatePublished - 2015
Publication categoryResearch

Projects

Download statistics

No data available
ID: 209620414