Membranes in the form of sheets are used to form spiral wound membrane modules for a wide variety of separations ranging from seawater desalination to carbon dioxide capture from coal-fired power plants. Spacers are a critical component of these modules. Spacers create and maintain uniform flow channels. Spacers also are used to mix fluid within the channel to enhance mass transfer. Unfortunately, spacers increase pressure drop. This increase depends strongly on spacer geometry.
Computational fluid dynamics (CFD) has been used extensively to investigate flows within spacer filled channels. This body of work has examined the dependence of velocity field and pressure drop on spacer geometry variables.
Simulation results typically are validated by comparison to experimental measurements of pressure drop. Limited point-wise, volumetric comparisons of velocity fields with experimental measurements are available.
Detailed comparisons of experimental velocity fields, obtained using Particle Image Velocimetry (PIV), with simulation results are reported here. PIV captures the movement of seed particles in the flowing fluid to determine the local fluid velocity. Measurement of the two velocity components in the interrogation plane are presented for three types of spacers: 1) symmetric, 2) asymmetric, and 3) static mixing. The results indicate CFD simulations are in good agreement with experimental results. The greatest experimental challenges are uniform particle distribution throughout the flow channel and precise orientation of the flow cell.
Stereographic measurements of all three velocity components in the interrogation plane also are reported. The out-of-plane velocity component is small for the symmetric and asymmetric spacers and values are near the resolution level of the experiment. However, good agreement is found for the static mixing spacer which possesses much larger out-of-plane velocities.
The results suggest CFD simulations provide good predictions of spacer performance. This further validates the use of CFD to optimize spacer geometry.