Operating Assistance System for Transitioning between Operating Modes of Molten Salt Receivers

Abstract

Molten salt solar tower (MST) power plants represent a concentrating solar power (CSP) technology already in commercial use. Nevertheless, they still have great potential for efficiency improvements and cost reductions. In contrast to conventional power plants, CSP plants are subject to highly volatile boundary conditions facing components with a wide range of inertias. Hence, to realistically predict and increase the actual yields of an MST, complex simulation models and algorithms are required to pay particular attention to the operation of the receiver system. Within the scope of this work, it was fundamentally investigated how an operating assistance system can support the receiver operation concerning availability and net yield. An assistance function, which provides model-prediction-based decision proposals for transitioning from normal receiver operation to drained standby, was developed and tested. As a basis, a detailed dynamic process model of an MST receiver system, including a threedimensionally discretized receiver, less detailed other components, a distributed control system for start-up and shutdown procedures as well as control loops, was developed and implemented in Modelica. For this purpose, a thermo-hydraulic two-phase model for molten salt and air as well as specific component models, were implemented, which enabled complying with local limits in the receiver during and after flood or drainage of the receiver. To reduce the computing time of the yield prediction, simplified models and a heuristic decision algorithm were developed and implemented in Modelica and Python, respectively. In this context, a virtual net power approach for the receiver system was developed, which allows realistically predicting the net output while considering the complex dynamic behavior and quickly finding the optimal timing without iterations. This operating assistance function quantifies the yield gain/loss achieved by a temporary receiver standby and serves as the basis for the decision proposal. The test results show that one single maneuver proposed by this operating assistance function can increase the net yield of a utility scale plant by several megawatt-hours, depending on the cloud situation. Uncertainty analyses show a particular sensitivity for forecasting errors, which is why better forecasting accuracy is needed, especially for larger prediction horizons, compared to the examined data. Furthermore, there is still a need for research on the accurate modeling of the local convective heat loss at such a receiver to predict the thermal losses during start-up and shutdown and thus its duration reliably.