Analysis of variables affecting mixed liquor volatile suspended solids and prediction of effluent quality parameters in a real wastewater treatment plant. Issue 45 (25th September 2016)

• Record Type: Journal Article
• Authors: Bagheri, Majid; Mirbagheri, Sayed Ahmad; Ehteshami, Majid; Bagheri, Zahra; Kamarkhani, Ali Morad
• Is Part Of: Desalination and water treatment. Volume 57:Issue 45(2016)
• Page Start: 21377
• Page End: 21390
• Publication Date: 2016-09-25
• Subjects: Artificial neural network -- Genetic algorithm -- Multi-layer perceptron -- Radial basis function -- Wastewater treatment plant; Saline water conversion -- Periodicals; Saline water conversion; Water -- Purification; Periodicals; 628.167
• DOI: 10.1080/19443994.2015.1125796
• Languages: English
• ISSNs: 1944-3994

Abstract

Abstract: This article was an effort to predict effluent quality parameters and analyze variables affecting mixed liquor volatile suspended solids (MLVSS) for Ekbatan wastewater treatment plant in Tehran, Iran. These parameters were predicted and analyzed using two of the most common classes of artificial neural networks (MLP and RBF) coupled with genetic algorithm. Temperature, pH, influent concentration of the parameters, sludge volume index (SVI), and sludge volume after 30 min of settling (V30) were inputs of the neural networks. These inputs were used to predict biochemical oxygen demand (COD), total nitrogen (TN), and total suspended solids (TSS) concentrations as well as MLVSS concentration in the aeration tank. The introduced models for training and testing data sets indicated an almost perfect match between the experimental and the predicted values of COD, TN, TSS, and MLVSS. The models were verified by evaluating their performance in propitiously simulating the statistical features of the observed data. Furthermore, another criterion applied for judging the validity of the models was the assessment of the goodness of fit according to available criteria. The mean average error in prediction of all parameters for the train and test models did not exceed 6 and 4%, respectively. The results of sensitivity analyses for the models indicated that the variation of the MLVSS concentration in the aeration tank is influenced by V30, influent TSS, T (°C), SVI and pH, respectively. It was observed that the V30 and influent TSS significantly affect the MLVSS concentration in the aeration tank.