Discrete model to estimate nucleate boiling heat transfer coefficient with gasoline using a double-log-normal distribution function

Abstract

Heating fuels before injection into internal combustion engines provides benefits related to the reduction of pollutant emissions and the improvement in cold-start performance. Current fuel-heating solutions used in cold-start systems of ethanolfueled engines in Brazil involve nucleate boiling heat transfer and, in order to extend its use to fossil fuels, further studies are required on the heating process of multicomponent mixtures. Because fuels like gasoline, diesel and kerosene consist of several hundred chemical species, estimating their heat transfer coefficient is virtually impossible unless simpler mixtures, known as surrogates, are used to emulate their targeted behavior. Following this strategy,we present in this paper a predictive method to estimate the nucleate boiling heat transfer coefficient with gasoline. In order to formulate surrogates, probability distribution functions were discretized to ground the choice of components and their respective molar fractions. At first, a traditional gamma distribution function was used but it provided unsatisfying results, so we developed an original one named double-log-normal distribution. Heat transfer coefficients and bubble points estimated at different pressures were compared with experimental data. The best surrogate was a 6-component mixture formulated with the proposed distribution and its application resulted in a 2.3% overall absolute deviation for the heat transfer coefficient estimation at 102 kPa. The model was validated at pressures up to 403 kPa. Finally, a sensitivity analysis of the double-log normal distribution showed the estimated nucleate boiling heat transfer coefficient is nearly the same regardless of gasoline composition variations.