Evaluation of Three Different Cleaners Recommended for Ultrafiltration Systems by Direct Observations of Commercial-Scale Spiral-Wound Ultrafiltration Membranes

Efficacy of cleaners designed for use with ultrafiltration systems was determined by microbiological evaluation and thmugh visual inspection using scanning elecmn microscopy. The ultrafiltration system containing two commercial-scale, polysulfone membranes was soiled with sweet whey (4OOC) then rinsed with water and membranes were removed. One half of each membrane was snaked for 2 h at 38OC in one of the following solutions: control (no soaking), acid cleaner (pH 2.51, enzymebased cleaner (pH 1 1.5) and chlorinated alkaline cleaner (pH 11 .5). The membranes were repositioned in the ulhafilhation unit, rinsed with water, then removed and unwound for analysis. Sections of membrane, retentate spacer and permeate mesh were aseptically removed for enumeration of microorganisms remaining and for examination by scanning elecmn microscopy. Membranes cleaned with chlorinated alkaline cleaner averaged 2 x 10' CFul50 cm', enzyme-hasedcleaner6x 106/CFU,acid anioniccleaner 1 x 10' CFUandthecontrol5 x 10'CFU. Scanningelectronmicroscopy found soil and microorganisms present on all membrane materials exposedto all threecleanem The dairy industry has found a range of applications for ultrafiltration (UF). The spiral-wound configuration is especially popular for processing whey (5). Unlike other dairy processing equipment, determination of adequate cleaning in UF systems has relied on indirect indicators such as permeate flux and final product quality (3S.12.16). These cleaning indices indicate only major problems from inadequate cleaning. Scanning electron microscopy (SEM) has been used to determine the possible structure of fouling material on UF membranes following circulation of whey and purified whey proteins in laboratory-bench scale units (8.9). This SEM technique also has been used to evaluate the ability of cleaning agents to remove biofilms from reverse osmosis membranes used in commercial water treatment facilities (20). In this study, SEM was used with SDS-PAGE and determinations of microbiological populations on UF mem' 'Research supported by the College of Agriculrurol and Life Sciences, University of Wisconsin. Madison, W153706. brane sections to directly evaluate the ability of three cleaners to remove both soil and microorganisms from a commercialscale UF system. MATERIALS AND METHODS Equipment The UF system consisted of a 380-L jacketed vat equipped with agitator and thermometer, prefilter, centrifugal pump, and valves and pressure gauges located before and after two membranes in parallel ( Tri-Clover, Inc., Kenosha, WI). The spiralwound, polysulfone UFmembranes (Filmtec, Minneapolis, MN) had a 10.000 to 15,000 dalton cut-off and had not been soiled previously. A feed rate of 95 Llminlmembrane with a pressure drop of 1.4 kg/cm2 was selected according to recommendations of the manufacturer. Soiling procedure Pasteurized, separated sweet whey was obtained immediately before use from a local cheese factory. The soiling procedure consisted of recycling 380 L of whey for 2.0 h and concentrating the whey for an additional03 h. Cooling watercirculated in the jacket of the vat was used to maintain whey temperature at 38-43°C. Determinations of pH, percentage total solids and ash(I) and numbersofbacteria, yeasts andmolds (13) weremade for whey in the tank, retentateand permeatestreams initially, and at the start and end of concentration. Cleaning procedure Following soiling, the UF system was flushed with water (209 L, 40mC) to remove loosely held soil and permeate flux was determined with water. The membranes were removed from the housings and placed in tanks containing cleaning solutions. One-half of each membrane was soaked for 2 hat 38OC in one of the following solutions: control (no soaking), phosphoric acid (pH 2.51, enzyme-hasedcleaner (pH 11.5) andchlorinated alkalinecleaner (pH 11 S). Thecleaners wereobtainedfromthreedifferent manufacturers with their exact composition proprietary; however, manufacturer's recommended usedilutions were used. The design of the cleaning method used for this study was based on several factors: (a) Because of membranes costs, a maximum amount of cleaning information was desired from -