Evaluation and optimization of anammox baffled reactor (AnBR) by artificial neural network modeling and economic analysis

Anammox baffled reactor (AnBR) had a moderate start-up period of 53 days. Interestingly, tangled relationships between key parameters affecting anammox performance were observed, i.e., polynomial function for nitrogen loading rate (NLR) with extracellular polymeric substances (EPS), linear relationships between EPS with granules diameter, granules diameter with settling velocity, and settling velocity with biomass concentration. The correlation coefficients (R2) were 0.97, 0.84, 0.86, and 0.88, respectively. Furthermore, a multi-layered feed forward artificial neural network (ANN) was utilized for simulating and predicting the performance of AnBR. An ANN structure of two hidden layers with four neurons at 1st layer and eight neurons at 2nd layer achieved the best goodness of fit with the minimum mean squared error (MSE) and maximum R ² of 0.002 and 0.99, respectively. Additionally, economic assessment stated that using AnBR at NLR of 4.04 ± 0.10 kg-N/m ³ /day achieved the maximum net present value of $48100.9.