"A Geometry-Driven Neural Operator for Real-Time Digital Twins in Seawater Reverse Osmosis Systems"

"This dataset is prepared for the reproduction of the GDNO framework and includes seawater reverse osmosis (SWRO) membrane parameters under multiple geometric configurations. Each subset corresponds to a distinct operating condition, where both geometric and physical inputs vary, leading to different flow field outputs. The variations include inlet velocity, spacer height, spacer width, and internal crossing angle of the spacer structure, which together influence the resulting velocity distribution and pressure drop characteristics. Due to the large scale of the full dataset, only 10 representative subsets are randomly selected and provided to ensure efficient data reproduction and manageable upload size. Additional data covering a broader range of configurations can be made available upon request."