Biomass Leachate Treatment and Nutrient Recovery Using Reverse Osmosis: Experimental Study and Hybrid Artificial Neural Network Modeling

The application of reverse osmosis (RO) to recover nutrients from biomass leachate, with special reference to permeate flux behavior during the filtration of the leachate, is investigated in this paper through a comprehensive laboratory and modeling approach. Various sources of biomass, including soybean straw, switchgrass, and miscanthus, were industrially leached using distilled water with agitation during the extraction experiments. The leachates were filtered using a RO flat sheet membrane module to recover the nutrients and water. On the basis of inductively coupled plasma (ICP) analytical results, calcium, magnesium, phosphorus, and silica for all types of biomass leachate samples had rejection efficiencies of >80%. Permeate flux decreased sharply at the beginning of the filtration, followed by a slight decline during the filtration process. A hybrid intelligent model based on a feed-forward artificial neural network (ANN) was also developed to estimate the permeate flux during the filtration in terms of the filtration time and total solid concentration in the leachate. A Levenberg–Marquardt optimization algorithm was chosen to perform the training phase for the network. An ANN with four neurons in one hidden layer was selected as the optimum structure, such that a maximum percent absolute error of 14% was attained while predicting the permeate flux. A reasonable agreement was observed between the ANN predictions and experimental data, which exhibits the potential usefulness of the hybrid ANN model to predict permeate flux during RO filtration of biomass leachate.