Resolution of RO flux decline problems at Englewood, Florida

Abstract

The Englewood Water District (EWD) commissioned the first of a series of reverse osmosis (RO)_trains in December 1982. Pretreatment of the well water feed included addition of scale inhibitor and acid prior to a cartridge filtration. The feedwater was not exposed to atmospheric oxygen or chlorine before the RO train, to preclude formation of colloidal sulfur from hydrogen sulfide present in the groundwater. From startup, this first 500,000 USGPD brackish RO train exhibited unacceptably high flux decline resulting from feedwater fouling. This occurred despite all feedwater quality parameters being well within normally specified requirements. The RO membrane elements lost approximately 30 percent of their normalized productivity in less than 18 months of operation. Subsequent pilot testing onsite and membrane autopsies have characterized the foulant or foulants sufficiently to resolve the problem; however, further investigation would be required to precisely identify the substance or substances causing the flux decline. The feedwater, with very low silt density indices (typically less than 0.5), caused a nominal 10,000 molecular weight cutoff (MWCO) polysulfone ultrafilter to lose one-third of its initially measured flux after one week of operation. However, pretreatment using an ultraviolet sterilizer followed by a nominal 100,000 MWCO polysulfone ultrafilter did not mitigate the flux decline of RO membrane elements in a pilot test unit. It is currently hypothesized that dissolved organics are the primary contributors to the unexpectedly high flux decline at Englewood. The pilot tests also showed that flux declines could be decreased to acceptable levels by operating at lower permeate flux rates. The testing also revealed that the flux declines of at least one type of CAB membrane were unrelated to applied feed pressure. Finally, 2,000 hours of further pilot testing demonstrated that several types of commercial RO membranes should have acceptable flux decline at lower flux rates than typically used. The situation at Englewood indicated the need for development of better RO feedwater quality criteria. The pilot testing showed that further pretreatment is not easily identified and it determined technical solutions to the flux decline problem using either cellulose acetate blend or polyamide membranes.