Field Demonstration of an Automated Heliostat Tracking Correction Method

Abstract

This study provides a demonstration of an eight-parameter heliostat tracking-error correction method applied to heliostats at the National Solar Thermal Test Facility at Sandia National Laboratories. This method models the non-random physical error sources that arise from imperfect mount fabrication and installation processes. Previous work demonstrated this method for a single heliostat over a one-month period [1]. This study extends the previous work by expanding the number of heliostats modeled and tested to over 200. The duration over which errors were collected and modeled spanned 7 months. Smaller subsets of heliostats were observed and tested over an even longer, 10 month period. Error data was obtained by bringing heliostat beams (one by one) to a surveyed target on the Solar Tower and recording the measured elevation and azimuthal offsets. Beam position was determined by the centroid of the intensity image. Using automated sequencing, we were able to limit the time to process a heliostat to ∼32seconds. This allowed us to process the entire 200 heliostat field in just 2 full days (weather permitting). These 2- day collections were performed about once every 2-3 months, to cover the full annual range of motion. All error data were automatically collected, time tagged and synchronized, in real-time. The error data were then converted, by a post-processor, to updated model parameters. The updated model data were subsequently uploaded to the heliostat field for model performance testing. In this way, almost 50,000 error observations were collected for 200 heliostats during the course of the study. In addition, 19,000 model performance observations were also recorded. Initial results indicate this method was able to halve the RMS average pointing error from 1.5 mrad to 0.78 mrad across the field of 200 heliostats.