Abstract:
To accurately predict flow characteristics of steam control valve with filter, the model constant vectors in
k-ω SST turbulence model are recalibrated using ensemble Kalman filter against valve’s flow rate experiment data at several valve openings. In the procedure, Latin hypercubic sampling method is used to perturbate the model constants and generate samples of predicted valve flow rates. Then the samples are assimilated with independent experiment data to update model constant vectors iteratively until convergence is achieved so as to complete the recalibration. In terms of pressure ratio and flow rate measured in independent experiments, the comparison and analysis shows that the recalibrated model constants can be well applied to simulation of steam flow at approximate valve opening but different pressure ratios; this greatly reduces the deviation of valve inlet flow rate between model predictions and experiment measurements, while the configuration with significantly different valve openings is still exposed to notable errors in predicted flow rate. The predicted flow patterns and eddy viscosity determined from simulations with and without the recalibrated model constants vary significantly at different valve openings. Given the fact that optimizing valve flow characteristic prediction via model constant recalibration is made with the optimized eddy viscosity estimation, such methodology provides good reference for related engineering design.