On Measurement 559 Impact of different weather data sets on photovoltaic system performance evaluation Chanikarn Yimprayoon and Mojtaba Navvab University of Michigan, Ann Arbor, Michigan ABSTRACT: Building energy simulation plays an important role in decision makings involving energy conservation measures and choices of renewable energy systems in building designs. Traditional simulation tools rely on weather data sets called Typical Meteorological Year (TMY), representing a typical year of weather at ground weather stations throughout the United States. Tese data sets are constructed using an algorithm to select the “most typical” month of the many years in the database for each month. Some recent publications suggest that one-year TMY data is no longer sufcient to evaluate long-term performance of PV systems, because a typical year does not taken into account extreme weather, and thus does not address the meteorological uncertainty that might occur. Actual electricity outputs from photovoltaic systems vary from year to year. Having more accurate information about production performance should help facilitate system selections that match building designs and how to operate them. In this study, four sets of weather data, Detroit TMY2, Ann Arbor TMY3, Ann Arbor 15-year NSRDB, and Ann Arbor 13-year SolarAnywhere®, are used as inputs in PV system performance simulation. Teir impacts on the PV system electricity output availability, variability and uncertainty are analyzed and compared. Te magnitude and consequences of the analyses of diferent weather data sets are presented. CONFERENCE THEME: On Measurement: What is performance? Approaches to energy, occupation, consumption and reuse. KEYWORDS: Weather data, photovoltaic system output, availability, variability, uncertainty INTRODUCTION Te performance of a building is a result of complex processes. A better building design can reduce energy use by 30% compared to a conventional building design, while still provide an equal or better environment for its occupants. To reach a 50% reduction or more, renewable energy system integration is needed (USGBC Research Committee, 2008). Barriers to achieve this goal is usually not technology constraints, but poor data to make informed decisions (Clarke, 2001). Building simulation tools are created to help provide real world replication and predict how buildings and systems will perform once they are constructed and implemented, thus providing information for decision making. Building energy performance prediction tools are a series of complex mathematical models that address the dynamic interaction of building and system performances with building geometry, plan, components, system choices, climate conditions and occupant use patterns. Tese computer based simulation tools usually require local weather data as main inputs for outside conditions. Performance simulation of solar energy systems such as a photovoltaic (PV) system relies heavily on specifc time-location hourly weather data. Te availability of solar radiation at a specifc site varies according to location latitude, topography, time of day, time of year, cloud cover and atmospheric aerosol condition. Te amount of solar radiation and its temporal distribution at a specifc location are essential information for determining if a PV system is suitable for that site. Tis information can be used to select the solar energy system’s size and predict its performance and operation. Onsite measured solar radiation provides the most accurate information. However, measurement equipment and their maintenance are costly. Solar radiation data can also be obtained from the nearest ground weather stations which monitor weather data such as daylight hours, air temperature, humidity, pressure, wind direction, wind speed and other climate- related information. Protocols such as the International Daylight Measure Program (IDMP) have been developed as a guideline for these meteorological data measurements (CIE TC 3-07, 1994).