Experimental and numerical characterization of the water flow in spacer-filled channels of spiral-wound membranes

Abstract

Micro-scale flow distribution in spacer-filled flow channels of spiral-wound membrane modules was determined with a particle image velocimetry system (PIV), aiming to elucidate the flow behaviour in spacer-filled flow channels. Two-dimensional water velocity fields were measured in a flow cell (representing the feed spacer-filled flow channel of a spiral wound reverse osmosis membrane module without permeate production) at several planes throughout the channel height. At linear flow velocities (volumetric flow rate per cross-section of the flow channel considering the channel porosity, also described as crossflow velocities) used in practice (0.074 and 0.163 m·s(-1)) the recorded flow was laminar with only slight unsteadiness in the upper velocity limit. At higher linear flow velocity (0.3 m·s(-1)) the flow was observed to be unsteady and with recirculation zones. Measurements made at different locations in the flow cell exhibited very similar flow patterns within all feed spacer mesh elements, thus revealing the same hydrodynamic conditions along the length of the flow channel. Three-dimensional (3-D) computational fluid dynamics simulations were performed using the same geometries and flow parameters as the experiments, based on steady laminar flow assumption. The numerical results were in good agreement (0.85-0.95 Bray-Curtis similarity) with the measured flow fields at linear velocities of 0.074 and 0.163 m·s(-1), thus supporting the use of model-based studies in the optimization of feed spacer geometries and operational conditions of spiral wound membrane systems.

Influence of Membrane Intrusion on Permeate‐Sided Pressure Drop During High‐Pressure Reverse Osmosis

“…For feed-sided spacer-filled channels, either a hydraulic diameter including the wetted spacer surface area, the spacer filament diameter or the channel height is suggested for the characteristic length [29,34,45]. However, since the microstructure of the feed-sided spacer channel as well as the geometry of the feed spacer itself is considerably simpler than the woven structure of the permeate spacer and the porous support layers of the membrane, the general recommendations of the above-mentioned assumptions for the flow through a porous medium are applied and developed in this study.…”

The membrane fouling simulator: development, application, and early-warning of biofouling in RO treatment

“…Fouling development can be studied by sensitive monitoring of the FCP and visual observations through the simulator sight window. A variety of MFSs have been developed [3] for varying purposes such as for studying (i) feed spacers geometry causing variation of flow channel height in MFSs [37], (ii) the hydraulic biofilm resistance with permeate production [107,108], (iii) biofilm processes and its impact on hydraulics [109], and (iv) the impact and sequence of biofouling development on membrane performance parameters including FCP, permeate flux and salt passage in a recently developed 1 m long MFS (long-channel membrane test cell) with five sections over the length, enabling the measurement of permeate flux and salt passage over the test cell length [31]. This 1 meter long MFS is characterized by a more rigid structure (4 cm wide membrane sheets, 24 bolts with metal support) compared with the earlier developed prototype cell, and proved to be representative for membrane modules for both flow profile and flow pressure drop relationship.…”

On some issues in the computational modelling of spacer-filled channels for membrane distillation

“…In regard to the phenomena modelled, some papers [12,15,19,21] simulated only the fluid flow, while others considered also heat transfer [2-4, 9, 11, 18] or mass transfer [16-17, 20, 22].…”

“…In regard to the variation of the Nusselt number with the attack angle γ, most of the experimental and 3-D computational studies presented so far consider only a few orientations, with the main flow parallel or orthogonal to the filaments, or bisecting the angle formed by these latter [3,4,12,[15][16][17][18][19][21][22] (an orientation cannot even be defined in the case of 2-D numerical simulations [8,11]). Yet, the few studies considering different attack angles [14], as well as physical intuition, suggest that orientations other than the basic ones mentioned above may offer advantages in terms of heat / mass transfer and frictional pressure drop, and thus are worth investigating.…”