Treatment of palm oil mill effluents using rambutan (Nephelium lappaceum) and its modeling using artificial neural networks

Abstract

Palm oil extraction is one of the important industries in Malaysia. The extraction of palm oil will not only result in the production of palm oil for human consumption but also generate palm oil mill effluent (POME), which contains highly polluting properties. Synthetic coagulants such as aluminum sulfate (alum), aluminum chloride, and sodium aluminate contribute toward environmental pollution and risk to human health. Thus, natural coagulants and flocculants have been introduced throughout the globe to slowly replace synthetic coagulants. In this study, rambutan (Nephelium lappaceum) seed was used as natural coagulants combining with aloe vera as natural flocculants. The standard jar test experiments were conducted to study the effects of pH, coagulant dosages, and rapid mixing speed on the removal efficiency of chemical oxygen demand (COD), total suspension solid (TSS), and turbidity of POME. The model of the coagulation–flocculation process was evaluated by using response surface methodology (RSM) and artificial neural network (ANN). RSM associated with Box–Behnken design was used to identify the optimum parameters (inputs), which consist of pH, coagulants dosages and rapid mixing speed for the optimum removal efficiency (outputs) of COD, TSS, and turbidity of POME. In terms of removal efficiency of COD, TSS, and turbidity of POME, rambutan seed has the optimum removal efficiency of 28.48%, 56.38%, and 60.36%, respectively. ANN was utilized with the aim of evaluating the performance of 12 training algorithms to predict the output results. Levenberg–Marquardt (LM) training algorithm was verified to be the best training algorithm for coagulation–flocculation process of rambutan seed based on the lowest mean squared error (MSE), mean absolute error (MAE), and mean absolute percentage error (MAPE) for validation data and the correlation coefficient R of.9347.