Field data collection for quantification of reliability and availability for photovoltaic systems

Predicting reliability and availability is a data driven capability of interest to the entire photovoltaic community, from material suppliers to system owners. Sandia National Laboratories is developing a predictive model and a host of methodologies needed for creating accurate predictions. Operational and maintenance (O&M) data from operating systems is only one piece of a broader data set required to make accurate predictions. Estimating reliability and availability of fielded photovoltaic systems requires times-to-failure or times-to-suspension and downtime data for each of the major components of each system. This paper addresses the collection (data set) and organization (standardized format) of data necessary for reliability and availability analyses. Typically, for a large photovoltaic system the data are censored. The data sets are composed of a mixture of components with failure and components without failure. The data sets must be organized into times to failure, or suspension time in use without failure, for each component that is being analyzed. To accurately estimate availability the various contributors to system downtime, such as corrective or preventative maintenance and grid perturbations, must also be identified and modeled. Preparation of the data for analysis usually consumes a significant percentage of the time required to generate a system reliability or availability estimate. A case study with data from a five year period of a fielded photovoltaic system is used to illustrate how a commercially available software tool for failure reporting and corrective action, XFRACAS™, was adapted to efficiently organize field data and transfer data into a suite of software tools. The software tool Weibull++™ was used to fit life distributions or growth models and to estimate parameters of the distributions. Another software tool, BlockSim 7™ was used for Reliability Block Diagram (RBD) development and simulation of system relia- - bility and availability. XFRACAS™ is a web-based application that provides the capability for point of source data entry into a centralized data base. With some slight modifications, XFRACAS™ is capable of exporting data from the database that is properly organized and formatted for analysis by the life data analysis or reliability growth analysis tools.