Machine learning enabled condensation heat transfer measurement

Measuring condensation heat transfer and its associated coefficient is not trivial. Rigorous measurements require careful experimental design tradeoff studies to properly select sensor type, sample geometry size, coolant fluid flow rate, operating conditions, working purity, purge methodology, measurement protocol. Conventional tube-based quantify the change in enthalpy of a single-phase via inlet outlet bulk temperatures. The uncertainties with this classical well-established method are high. high uncertainty stems from characteristic or flux process, making thermal resistance on external tube side typically same order magnitude as internal convective resistance. Even when taking utmost care using extremely accurate sensors having low uncertainty, relative can be range 20% 100%. Here, we take advantage machine learning (ML) develop an optical visualization for dropwise characterization. Using state-of-the-art intelligent vision, demonstrate previously-unexplored characterizing condensate droplet shedding frequency, without need high-speed imaging. We verify our technique by conducting rigorous steam Parylene C coated smooth copper samples 500 nm, 1 ?m, 5 ?m thicknesses. validate ML predictions data obtained simultaneously enthalpy-change custom chamber. In contrast conventional methods, constant (?10%) does vary flux. finally key custom-made axially varying surface properties resulting differing local coefficient. Our enables fidelity characterization phase flux, reduction resolution effects, elimination temperature across samples.