Effects of hydraulic retention time and cultivation on nutrients removal and biomass production in wastewater by membrane photobioreactor: Modeling and optimization by machine learning and response surface methodology

Discharge of excess nutrients in wastewater can potentially cause eutrophication and poor water quality in the aquatic environment. A new approach of nutrients removal in wastewater is by utilizing microalgae which grow by absorbing CO2 from air. Furthermore, the use of membrane photo-bioreactor (MPBR) that combines membranes and photo-bioreactor has emerged as a novel wastewater treatment method. This research sought to model, forecast, and optimize the behavior of dry biomass, dissolved inorganic nitrogen (DIN), and dissolved inorganic phosphorus (DIP) in MPBR by response surface methodology (RSM) and artificial neural network (ANN) algorithms, which saved time and resources of experimental work. The independent variables that have been used for modeling were hydraulic retention time (HRT) and cultivation. For this purpose, the dry biomass of algal production, DIN and DIP behavior were modeled by RSM and ANN algorithms, to identify the optimum mode of processing. RSM modeling has shown good agreement with experimental data. According to RSM optimization, the optimum mode for DIN and DIP occurred at 1.15 days of HRT and 1.92 days of cultivation. The ANN showed better performance than the RSM model, with the margin of deviation being less than 10%. Furthermore, the ANN algorithm showed higher accuracy than RSM method in predicting the dry biomass, DIN and DIP behavior in MPBR.