Solar Ceils, 27 (1989) 455 - 464 455 UNCERTAINTY ESTIMATES FOR GLOBAL SOLAR IRRADIANCE MEASUREMENTS USED TO EVALUATE PV DEVICE PERFORMANCE DARYL R. MYERS, K. A. EMERY and T. L. STOFFEL Solar Energy Research Institute, Golden, CO 80401 (U.S.A.) Summary Broadband (0.3- 3.0/~m) global solar irradiance measurements are used in the evaluation of solar energy conversion devices. The uncertainty at. tached to such measurements is important in evaluating whether conclusions associated with the measurements are statistically valid. A standardized un- certainty analysis method, developed over the past 15 years in the arena of consensus standards and professional society organizations, is described and applied. The results of the uncertainty analysis for the instrument calibration and field data measurement process indicate that the total measurement un- certainty in pyranometry (i.e. the measurement of global solar irradiance) can approach 5%. Thus comparisons of results between laboratories using different pyranometers can have a total uncertainty of up to 10%. Statisti- cally valid conclusions on a conversion device's performance may be drawn only if such results account for known bias errors or exceed the uncertainty limits derived using this methodology. 1. Introduction A pyranometer measures the combination of direct beam, scattered, and ground reflected solar radiation (if tilted from the horizontal) in a 180 ° field of view. The detector is often a thermopile (group of thermocouples) embedded in an absorbing disc which heats up, when exposed to radiation, and causes the thermopile to generate a voltage (of the order of mfllivolts). The calibration factor is the conversion factor for determining the incident power density, in watts per square meter, as a function of the generated out- put voltage. This calibration factor is often assumed to be a constant value, independent of the incident angle of the incoming radiation, ambient tem- perature, etc. However, many pyranometer characteristics have been studied which demonstrate that this is not the case. This paper evaluates the sources of error in the thermopile detector pyranometer in the process of calibration and field measurements, including applications which involve the determina- tion of photovoltaic conversion device efficiency. 037 9-67 87/89/$3.50 © Elsevier Sequoia/Printed in The Netherlands