Multi-response optimization and machine learning modeling algorithms of electrocoagulation parameters for cost-effective micro-algae harvesting

This study aimed to investigate the use of Electrocoagulation (EC) as an energy-efficient and cost-effective method for primary microalgae harvesting. A simple laboratory setup with aluminum electrodes and a direct current (DC) power source was used to harvest the microalgae species Chlorella sp . The effects of key process parameters—time, pH, voltage, and Sodium Chloride (NaCl) concentration—on the EC process were optimized. The Response Surface Methodology (RSM) models presented correlation coefficients (R 2 ) of 0.92, 0.69 and 0.56 for recovery efficiency, Aluminum and energy consumption respectively, while the Artificial Neural Network (ANN) models presented correlation coefficients (R 2 ) of 0.99, 0.93 and 0.95 respectively. Optimal conditions were identified at pH = 5, 15 min electrolysis time, 4 g/L NaCl and 5.5V, achieving a 93 % microalgae recovery with current density, energy and metal requirements at 3 mA/cm 2 , 0.95 kWh/kg and 39.69 g/m 3 respectively. The microalgae concentration increased from an initial biomass density of 0.76 g/l to 22.86 g/l, and metal concentration analysis revealed a low residual content of Al with 1.1 % Al in the harvested microalgae and 0.003 g/L Al in the recovered spent medium. The EC process also proved cost-effective with a total operating cost of USD 0.19/m 3 , based on laboratory-scale benchmark factors at the identified optimum conditions. This research demonstrated that at lower process times and current densities, the EC process can efficiently recover microalgae with minimal energy and metal consumption at a low cost. • Electrocoagulation (EC) was evaluated as an energy-efficient, low-cost method for primary microalgae harvesting. • Lower times and current densities enabled efficient recovery at minimal energy and metal inputs. • Optimal EC conditions (pH 5, 15 min, 4 g/L NaCl, 5.5 V) achieved 93 % microalgae recovery. • A total operating cost of $0.19/m 3 was estimated at optimal conditions. • ANN models outperformed RSM in fitting but exhibited nuanced prediction ability.