Optimization and performance prediction of air-flotation cyclones for oily wastewater

Treating heavy oil-containing wastewater on offshore production platforms remains challenging because heavy oil has a density close to that of water and dispersed droplets are difficult to remove using conventional centrifugal separation alone. To address this limitation, this study proposes an integrated treatment process coupling centrifugal separation with dissolved air flotation and develops a modified Bloom-Heindel interaction model for air-flotation cyclones. The proposed model explicitly incorporates the complete bubble-droplet interaction sequence, including collision, attachment, and stabilization-detachment in a high-shear swirling centrifugal field. Compared with a conventional model that neglects bubble-droplet micro-interactions and may produce prediction errors of up to 15%, the modified model reduces the average deviation from field measurements to 1.2%. Based on the validated model, computational fluid dynamics simulations were performed to analyze velocity, pressure, turbulent kinetic energy, and oil-phase concentration distributions. To reduce the computational cost of repeated multiphase simulations, a backpropagation neural network was established to predict separation efficiency under different structural and operating parameters. The independent test results show a Pearson correlation coefficient greater than 0.99, a mean squared error of 0.00223785, and more than 95% of prediction errors within ±5%. Particle swarm optimization was further used to determine the optimal design under field-oriented conditions. The optimized configuration, including a 6.4 mm overflow pipe diameter and a 13° large cone angle, yields a theoretical maximum oil removal efficiency of 95.71%. This coupled physical modeling, CFD simulation, neural-network prediction, and intelligent optimization framework provides a quantitative basis for improving compact oily wastewater treatment equipment on offshore platforms.