Experimental and theoretical investigation of nanofiltration membranes for liquid desiccant regeneration in air conditioning applications

Multi-stage nanofiltration (NF) has been proposed to decrease the energy consumption of conventional reverse osmosis (RO) desalination for potable water production. But multi-stage NF can also be used for liquid desiccant (LD) regeneration. Existing theoretical studies of multi-stage NF desalination do not account for the influence of high LD concentration and applied pressure on the NF membrane behaviour. In this study, we experimentally investigate four commercial NF membranes using LD solutions of LiCl, LiBr and MgCl2 to show how the salt rejection changes with LD concentration (investigated range of 10-43 wt%) and applied pressure (of 30-39 bar). We present equations and an iterative solution to calculate the reflection coefficients of the membranes. These methods are integrated into existing models to simulate more accurately the multi-stage NF regenerator. The experiments showed that LiCl gives slightly lower rejection than LiBr and MgCl2, but permeate flux is 1.3 to 5 times higher. Although it increases with applied pressure, rejection is mostly too low. The first membrane of the multi-stage LD regenerator would require rejections of 10-30%, compared to values of only 6% achieved in this study when using concentrated LDs typical of LDAC applications. To achieve higher rejections, fabrication of membranes able to withstand higher operating pressures is needed. Using current commercial membranes (applied pressure <41 bar), the multi-stage NF regenerator coupled with a LD cooling system achieves a coef-ficient of performance (COP) of 3.37. But using future membranes with an operating pressure of 80 bar, a COP of 15 could be achieved.