At the time of archiving, the student author of this thesis opted to temporarily restrict access to it. The full text of this thesis became available to the public after the expiration of the embargo on 2014-12-31.
Increasing demand for fresh water in arid and semi-arid areas, similar to the Middle East, pushed for the use of seawater desalination techniques to augment freshwater. Seawater Reverse Osmosis (SWRO) is one of the techniques that have been commonly used due to its cost effectiveness. Recently, the use of Ultrafiltration (UF) was recommended as an effective pretreatment for SWRO membranes, as opposed to conventional methods (i.e. sand filtration). During UF operation, intermittent cleaning is required to remove particles and contaminants from the membrane's surface and pores. The different cleaning steps consume chemicals and portion of the product water, resulting in a decrease in the overall effectiveness of the process and hence an increase in the production cost. This research focused on increasing the plant's efficiency through optimizing the cleaning protocol without jeopardizing the effectiveness of the cleaning process. For that purpose, the design of experiment (DOE) focused on testing different combinations of these cleaning steps while all other parameters (such as filtration flux or backwash flux) remained constant. The only chemical used was NaOCI during the end of each experiment to restore the trans-membrane pressure (TMP) to its original state. Two trains of Dow™ Ultrafiltration SFP-2880 were run in parallel for this study. The first train (named UF1) was kept at the manufacturer's recommended cleaning steps and frequencies, while the second train (named UF2) was varied according to the DOE. The normalized final TMP was compared to the normalized initial TMP to measure the fouling rate of the membrane at the end of each experiment. The research was supported by laboratory analysis to investigate the cause of the error in the data by analyzing water samples collected at different locations. Visual inspection on the results from the control unit showed that the data cannot be reproduced with the current feed water quality. Statistical analysis using SAS JMP® was performed on the data obtained from UF2 determined that the error in the data was too significant, accounting for 42%. Laboratory inspection on water samples concluded that the water quality feeding to the UF membranes was worse than that of the raw water. This led to a conclusion that severe contamination occurred within the main feed tank where the water was retained before arriving to the UF modules. The type of contamination present in the feed tank is yet to be investigated. Though, frequent cleaning or flushing of the feed tank is recommended on regular basis.