Data-driven multi-objective optimization of surfactant-modified nano-organic working fluids for enhanced heat transfer in ORC phase-change processes

Surfactant-mediated interfacial interactions in nano-organic working fluids effectively modulate colloidal thermophysical properties, leading to marked improvements in suspension stability and heat transfer performance within organic Rankine cycle (ORC) systems. Based on experimental data for nano-organic working fluids with three surfactants (CTAB, SDBS, and Span80), this study employs machine learning framework for prediction and optimization of the heat transfer and flow characteristics in ORC phase-change processes. A back propagation artificial neural network (BPANN) model is developed using 350 sets of transient steady-state experimental data, with the first 280 sets as training samples and the last 70 sets for testing, configured with 13 hidden nodes, a learning rate of 0.4, and the trainlm training function. The effects of six key operation parameters on evaporation and condensation heat transfer coefficients are investigated. A further bi-objective optimization considering maximum heat transfer coefficient and minimum pressure drop is addressed, while the Pareto-optimal solutions for surfactant-modified nano-organic working fluid is obtained. Results indicate that the evaporation heat transfer coefficient presents a parabolic trend with outlet dryness, while the condensation heat transfer coefficient increases with the mass flow rate. Increasing the evaporation heat transfer coefficient will deteriorate the condensation heat transfer coefficient. The optimal evaporation heat transfer coefficients and condensation heat transfer coefficients for 0.4 %SDBS + 0.1 %TiO₂/R123, 0.3 %CTAB + 0.1 %TiO₂/R123, and 0.3 %Span80 + 0.1 %TiO₂/R123 are 3986.03 W/(m²·K) and 851.23 W/(m²·K), 4139.71 W/(m²·K) and 825.25 W/(m²·K), and 4180.3 W/(m²·K) and 724.32 W/(m²·K), which are 23.76 % higher and 38.01 % higher, 28.53 % higher and 33.79 % higher, 29.79 % higher and 17.44 % higher than that of 0.1 %TiO₂/R123 of 3220.88 W/(m²·K) and 616.78 W/(m²·K), respectively. The nano-organic working fluid with Span80 demonstrates superior performance due to its relatively high comprehensive heat transfer coefficients for both evaporation and condensation.