Determınıng the effıcıency of membrane technologıes ın the treatment of drınkıng water resources by statıstıcal methods

Water is vital for the survival of all living things. Increasing population, changes in climate and rapidly developing industry affect the quantity and quality of water resources. This situation brings the issue of water treatment to an important point day by day. Water treatment includes many different techniques and when the treatment methods of water quality parameters are examined, it is seen that it is inevitable to use advanced technology treatment methods for some parameters. In recent years, membrane filtration processes have been widely applied in drinking water treatment in our country. In this study, it is aimed to determine the efficiency, removal efficiency and statistical methods of membrane technologies during seasonal monitoring of the quality of a surface water source and treatment in a full-scale treatment plant. Within the scope of the study, a reverse osmosis membrane-based filtration plant fed from the dam, which is a hard water source, was examined. Approximately 1250 water quality analyzes covering 44 parameters were carried out annually on the inlet and outlet water of ultrafiltration and reverse osmosis processes. Important parameters were determined for the Artificial Neural Network (ANN) model with Principal Component Analysis (PCA). It has been investigated whether ultrafiltration and reverse osmosis processes show efficient performance in the purification of the monitored parameters. Correlation between all ions was tested and conductivity was found to be the highest correlation representing other parameters. After that, TBA was applied to reduce the number of inlet water quality parameters. The suitability of the ANN model was determined by evaluating the mean square errors (0.00293 <0.01). The created model was useful for evaluating the efficiency of the treatment plant using conductivity. With the monitoring studies, it was concluded that the pretreatment processes are not suitable for the separation of mono/polyvalent ions. It has been concluded that UF membranes are effective for microbial and nitrogen parameters, and RO membranes are efficient for separation of anions and metals, which are defined as priority hazardous materials.