Multi-objective optimizations of vapor-liquid adjustment evaporator and its machine-learning based operational control strategy

Abstract

During flow boiling, there exists highly efficient heat transfer peaked at the high vapor quality. The vapor-liquid adjustment evaporator employs the liquid drainage and liquid refilling to redistribute the vapor quality and mass flux. In this way, the efficient heat transfer could be repeated, leading to the improved heat transfer capacity and reduced pressure drop at the same time. However, the path arrangement and separation efficiencies have been not mutually coordinated to release the potential of the vapor-liquid adjustment evaporator at various conditions. In this study, a numerical model of this evaporator is developed and verified by experimental data. By implementing the multi-objective optimization algorithm, three optimal layouts, targeting to the lowest pressure drop, the highest heat transfer capacity and the compromised one, are obtained at the design conditions. Comparisons of their local characteristics reveals that the fifth path offers most benefits in terms of 50 % entire heat transfer capacity and up to 73 % reduced pressure drop. At various off-design conditions, the constant separation efficiencies in vapor-liquid adjustment evaporator could lead to the inferior performance to the conventional evaporator. By implementing the machine-learning based control strategy, it could have maximumly 5.2 %-10 % increased heat transfer capacity or 5.2 %-51 % reduced pressure drop.