REAL-TIME SIMULATION OF A PARABOLIC TROUGH SOLAR FIELD

Abstract

Concentrating Solar Power (CSP) plants are a crucial component of today’s renewable energy technologies. CSP plants offer economies of scale as they increase in size, and they can incorporate thermal energy storage to improve dispatch capabilities as well as extend operation into the night. However, achieving and maintaining high levels of performance in large-scale and complex CSP plants is challenging. Enhancing operator training has been identified as a key area for growth and improvement of the performance of these plants. This work presents a high-fidelity parabolic trough solar field model capable of emu- lating CSP plant dynamics for use in an operator training simulator. This flexible, accurate, and computationally efficient model uses a novel neural network methodology to calculate the heat absorbed by the heat transfer fluid (HTF) under various receiver conditions. The complete solar field model presented includes 808 solar loops and can simulate responses to operator inputs like pump speeds and valve positions all while solving 20 times faster than real-time. This detailed yet computationally efficient model is not only ideal for training simulators but also offers potential for optimizing future solar field designs and operations. Additionally, this research presents an artificial intelligence (AI) model that learns the complex dynamics of CSP plants. This AI model can support operators during training simulations by offering real-time feedback and assisting in the resolution of challenging scenarios. Combining advanced systems modeling with AI will hopefully enhance operator training and has the potential to improve overall plant performance.