Assessment of Machine Learning Algorithms for Predicting Air Entrainment Rates in a Confined Plunging Liquid Jet Reactor

A confined plunging liquid jet reactor (CPLJR) is an unconventional efficient and feasible aerator, mixer brine dispenser that operates under many operating conditions. Such conditions could be challenging, hence, utilizing prediction models built on machine learning (ML) approaches very helpful in giving reliable tools to manage highly non-linear problems related experimental hydrodynamics such as CPLJRs. CPLJRs are vital protecting the environment through preserving sustaining quality of water resources. In current study, effects main parameters air entrainment rate, Qa, were investigated experimentally a (CPLJR). Various downcomer diameters (Dc), lengths (Lj), volumetric flow rates (Qj), nozzle (dn), velocities (Vj) used measure Qa. The relationship between ratio suggests applying regression algorithms predict appropriate. addition work, applied determine parameter predicts Qa best. results obtained from ML showed K-Nearest Neighbour (KNN) gave best abilities, proportion variance can explained by CPLJR was 90%, root mean square error (RMSE) = 0.069, absolute (MAE) 0.052. Sensitivity analysis most effective predictor predicting Qj Vj influential among all input variables. sensitivity shows lasso algorithm create rate model with just two crucial variables, Vj. coefficient determination (R2) 82%. present findings support using accurately forecast system’s results.