Electrocoagulation process modelling and optimization using RSM and ANN-GA for simultaneous removal of arsenic and fluoride

The current study explored the modelling and optimisation of the spiral electrode-based electrocoagulation process for the simultaneous removal of arsenic and fluoride. The process modelling for pollutant removal and energy efficiency is done through response surface methodology (RSM) and artificial neural network (ANN). Process optimisation explored numerical optimisation for RSM models and genetic algorithm (GA) for ANN model. The experimental design was performed using Box-Behnken Design (BBD). Results indicate that ANN outperformed RSM quadratic modelling, with higher R-squared values for fluoride removal (RSM: 0.8529, ANN: 0.9443), arsenic removal (RSM: 0.8934, ANN: 0.9339), and treatment cost (RSM: 0.9468, ANN: 0.944). Under the optimized conditions from RSM-BBD, with a current density of 1.8 mA/cm(2), pH 6.5, a treatment time of 41.5 min, and initial concentrations of 11.5 mg/L fluoride and 253 mu g/L arsenic, the technology achieved reductions of 82.45% for fluoride and 86% for arsenic, with a remediation cost of 0.94 USD/m(3). Conversely, the optimized conditions from ANN-GA, with a current density of 4.8 mA/cm(2), pH 6, a treatment time of 30.5 min, and initial concentrations of 13 mg/L fluoride and 254 mu g/L arsenic, achieved higher reductions of 93.25% for fluoride and 95.52% for arsenic, with a treatment cost of 1.90 USD/m(3). The optimized use of spiral electrodes successfully reduced contaminant levels to meet World Health Organization (WHO) standards for drinking water, demonstrating the technology's potential for effective and economical remediation of carcinogenic arsenic and fluoride in drinking water applications.