(691c) Understanding the Performance of Forward Osmosis Hollow Fiber Membranes at the Module Scale | AIChE

(691c) Understanding the Performance of Forward Osmosis Hollow Fiber Membranes at the Module Scale

Authors 

McCutcheon, J. - Presenter, University of Connecticut
Ren, J., University of Connecticut
This study evaluates an approach to make thin film composite (TFC) hollow fiber membranes for forward osmosis (FO) by simply employing commercial ultrafiltration (UF) membranes as support material. A thin polyamide film with excellent selective properties was synthesized on the inner surface of hollow fibers via interfacial polymerization. Besides demonstrating good performance as TFC FO membrane supports, the commercial hollow fiber UF membranes also provide a systematic platform with consistent properties to study structure-performance relationship. A series of commercial UF membranes were used to evaluate how support layer surface pore size (MWCO) impacted the properties of the selective layer and overall performance of the TFC membranes.

Aside from using commercial hollow fibers as FO membrane supports at bench scale, we also demonstrate that the TFC hollow fiber FO membranes can be made on existing modules at pilot scale. A series of commercial modules with different fiber size were used to make TFC hollow fiber modules. The resultant modules were evaluated under various operating conditions. While we evaluate how performance was impacted by module operating conditions, overall the modules demonstrate impressive FO results.

With the availability of reproducible membranes and modules, we developed a computational fluid dynamics (CFD) model with COMSOL Multiphysics to optimize module design for FO application. The FO process in a hollow fiber membrane was simulated in a 2D axis-symmetry geometry and described by fluid dynamics coupled with mass transfer. We verified the model accuracy by constructing modules with the same dimensions and fibers and testing under pertinent conditions. The agreement between the simulation and experimental results provided insight into how independent variables would impact module performance. To allow our colleagues to take the advantage of this model, we also built it into an application and will release it for the community to use for free.