Concentrating solar power plants use a large array of mirrors to focus the sun’s rays and capture their heat, which boils water for steam turbines to produce electricity. The technology and systems in these massive plants are complex—requiring skilled operators who can reliably make the correct control decisions to ensure the plants can achieve and sustain high performance levels.
Until now, operator training has lagged considerably behind the tech. Filling a critical need, University of Wisconsin-Madison mechanical engineers have developed a computationally efficient, high-fidelity model capable of simulating concentrating solar power plant dynamics faster than real time. These capabilities translate the model into a useful operator training simulator.
“We’ve developed a new simulation platform that allows plant operators to play around with the system and develop an intuition for how the plant is likely to respond to various control decisions, without the risk of causing damage to a real plant,” says Mike Wagner, an assistant professor of mechanical engineering who led the research. “This new tool could accelerate the upskilling of operations staff and improve productivity of the actual plant over time.”
The researchers’ new model is of a “parabolic trough” solar field, which is the most deployed type of concentrating solar power technology. These systems use long, curved mirrors that focus sunlight on tubes, heating a fluid flowing through the tubes. There are about 80 parabolic trough solar plants totaling 5.3 gigawatts operational or under construction worldwide.
Overall, a solar collection field consists of thousands of individual heat elements, mirrors and piping equipment, and this vast complexity has made it challenging to create fully detailed models. Rather than modeling individual solar loops and characterizing all their unique performance characteristics, existing models typically characterize just a small number of loops and then assume that all the other loops will have similar thermal-hydraulic performance.
Leveraging a neural network methodology, Wagner and recent graduate Matt Tuman developed a model that accurately captures every individual loop in the field and provides detailed performance information—while also being 100 times faster than previous models. In addition to training simulator applications, the new model can be used as a tool for optimizing solar field control strategies.
“This work advances the state-of-the-art in dynamic modeling for concentrating solar power while also introducing human factors considerations,” Wagner says.
The researchers reported their results in a paper published in the journal Solar Energy.
This work was supported by funding from the U.S. Department of Energy under grant No. DE-EE0009803.
Featured image caption: A parabolic trough solar field for a concentrating solar power plant. Credit: iStock.