Viscosity and thermal conductivity optimization of ZnO-MWCNT/water hybrid nanofluid: A combined SVR-PSO modeling and NSGA-II approach for heat-transfer applications

This study developed and trained two optimized Support Vector Regression (SVR) models, with hyperparameters tuned by Particle Swarm Optimization (PSO), to accurately predict the viscosity and thermal conductivity of ZnO-MWCNT/deionized water (DIW) hybrid nanofluids. These nanofluids were prepared at a fixed total volume concentration of ϕ =0.1%, with varying ZnO: MWCNT mass ratios (20:80 to 80:20), and tested over eight temperatures (20–55°C). The SVR models demonstrated excellent performance, with mean relative errors of 0.4160% and 0.2178% for viscosity and thermal conductivity, respectively, on test data. Subsequently, a multi-objective optimization problem was formulated to simultaneously minimize viscosity and maximize thermal conductivity, and solved using the Non-dominated Sorting Genetic Algorithm II (NSGA-II). This optimization yielded optimal operating conditions (temperature and ZnO: MWCNT mass ratio) within the tested domain ( ϕ =0.1 % fixed total concentration, 20–55°C, mass ratio 20:80 to 80:20) and the corresponding Pareto front, illustrating the trade-offs between the two objectives.