Determining the critical properties for the selection and optimisation of systems to evaluate filter integrity.

This thesis examines two filter integrity tests: Forward Flow for hydrophilic membranes, FF, and Water Intrusion Testing for hydrophobic membranes, WIT. Liquid filters are high area and diffusive flow through fully water-wetted pores masks bulk flow from the smallest defects during FF, so an alternative fluid was sought. Solutions were modelled and properties influential to diffusive flow identified using multi-linear regression. These were used for alternative fluid selection. Maltitol was identified as the fluid that would theoretically demonstrate most improvement. Experimental confirmation revealed a flow reduction of 23% at 10% concentration, compared with water. WIT measures ‘flow’ through an un-wetted filter under pressure. If there is a defect present, this pressure will be sufficient to achieve breakthrough and the test will report a failure. Questions about WIT remained unanswered: what is the mechanism of ‘flow’, and why is pressure the driver? The latter remains unanswered but conclusively demonstrated, and the former was proven as evaporative flow of water vapour. Salt/water was used, and conductivity reduced from 17.23 µS upstream to ~0 µS downstream, proving that the flow is water vapour evaporation. Both tests have been improved by filter design. FF requires every pore be fully wetted. An incompatibility between polypropylene and polyethersulphone means sometimes this cannot occur. The surface energy of the polypropylene is the property that provides the problem so it was altered by the incorporation of POSS, bringing the contact angle from 99.7° to 85.8°. Altering air filter design to accommodate increased flow during WIT would be advantageous as customers relate higher flows to higher reliability. Space downstream was introduced by using a thicker support layer to increase flow successfully. Further improvement was demonstrated by modifying the core to a ‘star’ design, increasing flow by 3.3x for a commercially-available product and by 2.8x for an R&D prototype