Experimental study of the heater size effect on subcooled pool boiling heat transfer of FC-72 in microgravity

Experiments of highly subcooled nucleate pool boiling of FC-72 with dissolved air were conducted on a large scale silicon chip (2 × 2 × 0.05 cm³, denoted as chip S 2 × 2) in short-term microgravity condition and normal gravity condition by utilizing the drop tower in Beijing respectively. The results were compared with published results of a smaller scale chip (1 × 1 × 0.05 cm³, denoted as chip S 1 × 1) both in normal gravity condition and microgravity condition, to study the heater size effect on boiling heat transfer. It is indicated that in microgravity, the input heat flux range in which bubble departure took place for chip S 2 × 2 is wider than that for chip S 1 × 1, and an interesting phenomenon observed is that coalesced bubbles departed at a continuously smaller radii at a given heat flux. Initially, the average bubble departure radii for chip S 2 × 2 increased linearly with heat flux while later remained constant. At a same heat flux, chip S 2 × 2 showed a bubble departure radius larger than that of chip S 1 × 1. By comparison, it is found that nucleate boiling performance deteriorates with increase in heater size in both earth gravity condition and microgravity condition. However, in microgravity the q_CHF of chip S 2 × 2 is 20% greater than that of chip S 1 × 1, contrary to the CHF characteristic in earth ground condition. Moreover, boiling patterns in microgravity are different in high heat flux region: a smooth hemispherical bubble was generated on chip S 1 × 1, while on chip S 2 × 2 an oblate vapor blanket was formed, which indicates that different dominated boiling heat transfer mechanisms exist in both cases. It is found that in microgravity, the boiling was dominated by buoyancy for chip S 2 × 2, but it was in surface tension dominated boiling regime for chip S 1 × 1, which proved the above speculation. Moreover, the range of transition heater size criteria is 1.45<LhLc<2.89 in the present work. By using the updated model developed by Raj and Kim, it is discovered that predicted values are generally lower than experimental values in both experiments. Moreover, differences were more evident for chip S 2 × 2.