Enhanced Methods for Performance Evaluation and Fault Detection of Concentrating Solar Process Heat Plants

Accurate measurements of direct irradiance are required for precise performance analysis and fault detection of concentrating solar process heat plants for industrial processes (CSHIP). For this reason, a cost-effective and low-maintenance pyranometer SPN1 has been investigated by comparing four years minute mean direct normal irradiance measurements of a SPN1 to a reference pyrheliometer at the Institute for Solar Technology SPF in Rapperswil (Switzerland). A significant overestimation of the direct normal irradiance which has already been stated in relevant literature could be confirmed. The most promising recalibration method to reduce the systematic and random error was based on a quadratic correction function, which reduced the normalized root mean square error NRMSE from 11.5% to 6.9% and completely eliminated the systematic error of 7.8% (referred to mean direct irradiance of 426W/m²). It was also shown that even recalibrations based on only one month of data already reduced the NRMSE to 6.9 - 7.8% if conducted between April and September. More sophisticated recalibrations considering the influence of radiation intensity and solar elevation did not lead to a significant improvement in accuracy. It was further demonstrated that the recalibration of the SPN1 is essential for CSHIP plant evaluation: An uncalibrated SPN1 pyranometer at a CSHIP plant in Switzerland would underestimate the annual plant effciency by 15% using experimental data of 2014. Since a transfer of the determined correction function onto another device did not lead to a significant reduction in the RMSE, a one month recalibration is recommended for each single device. Furthermore, a simple method for the experimental quantification of capacitive thermal losses has been developed which balances the change in internal energy during heating-up periods and which only requires commonly used measurement equipment for plant control. Integrating this method into the experimental plant performance analysis of two investigated CSHIP plants in Switzerland showed that these losses play an important role for the design and performance analysis of CSHIP plants. For the investigated plants these losses were in the range of 14% to 24% compared to annual useful solar gains and about three times higher compared to thermal pipe losses during operation (5 - 8 %). About half of the capacitive thermal losses of the CSHIP plant using water-glycol as heat transfer medium can be reduced by the implementation of a nightly thermal storage saving the heat contained within the hot fluid over night, so that heating-up periods in the mornings would be reduced from 1-2.5 hours (summer and winter) to less than 20 min on a clear sky day. The quantification of capacitive thermal losses also served for the enhancement of an algorithm based fault detection method which compares the experimental and theoretical daily solar gains. The already existing I/O-Controller by ISFH Hameln has been improved and adapted for the application at CSHIP plants. This was realised by using a more accurate method for the determination of the direct normal irradiance, considering heating-up periods within the analysis and determine the plant specific minimally required daily yield loss by error analysis. In case of the investigated plants, a minimal daily fault-related yield loss of 0.30 - 0.75 kWh/m² is required for a 100% detection rate depending on daily net direct irradiation.