Evaporation-rate controlled combustion in oxygen of a liquid fuel such as ethanol is modified by the fact that a major combustion product, here the less volatile H2O, readily dissolves in the fuel droplets. We demonstrate that this behavior, and many other features of EtOH spray combustion, can be economically described by adopting a population balance (PB-) approach in which the "state" of an individual droplet is defined by two variables----here the masses of contained EtOH and H2O, m1, and m2, respectively. This bivariate PB-formulation is illustrated by examining the theoretical performance of a well-mixed adiabatic combustor fed by sprays with prescribed droplet size distribution (DSD) and composition introduced at a temperature close to the relevant size-independent droplet "wet-bulb" temperature at the prevailing chamber temperature and pressure. As a historically interesting special case we consider an idealization of the V-2 chemical rocket at a nominal combustion chamber pressure of 15 atm---which was fuel-cooled using a 75 wt% EtOH + H2O solution, and used pure O2 as the oxidizer. Of special interest are the predicted dimensionless combustion intensity (chemical heat release rate per unit volume; GW/m^3), and the fraction of EtOH actually vaporized at each overall equivalence ratio and 'vaporization Damkohler number' (ratio of combustion chamber mean residence time to the QS-vaporization time of a droplet with the fuel-feed Sauter-mean diameter). We demonstrate that, in general, these performance characteristics, and their sensitivity to feed DSD-spread, are not well-represented using a simpler (univariate) description [which approximates such well-mixed thermodynamically non-ideal droplets as a pseudo-single component fuel with "effective" values of the heats of combustion and vaporization]. As an interesting byproduct of the present BV-PB analysis, we can predict/display the joint pdf(m1,m2) = n(m1,m2)/Np exiting such a chamber, the corresponding unconditional droplet size distributions, and jpdf(diam, temperature)---- perhaps amenable to direct measurement. We are currently examining the ability of approximate moment- and spectral/weighted residual methods to predict the shape of these bivariate distribution functions. Several generalizations of future interest to chemical reaction engineers are outlined in the light of this analysis.

_____________________________________