Flame-based synthesis of nanoparticles is an important chemical process used for the manufacturing of metal oxide particles. In this aerosol process, nanoparticle precursors are injected into a high-temperature flame that causes precursor oxidation, nucleation, and subsequent growth of solid particles through a variety of processes. To aid computational design of the aerosol process, a large-eddy simulation (LES) based computational framework is developed here. A flamelet-based model is used to describe both combustion and precursor oxidation. The solid phase nanoparticle evolution is described using a bivariate number density function (NDF) approach. The high-dimensional NDF transport equation is solved using a novel conditional quadrature method of moments (CQMOM) approach. Particle phase processes such as collision-based aggregation, and temperature-induced sintering are included in this description. This LES framework is used to study an experimental methane/air flame that used titanium tetrachloride to generate titania particles. The simulation results show that the evolution process of titania nanoparticles is largely determined by the competition between particle aggregation and sintering at downstream locations in the reactor. It is shown that the bivariate description improves the prediction of particle size characteristics, although the large uncertainty in inflow and operating conditions prevent a full scale validation. (c) 2013 American Institute of Chemical Engineers AIChE J 60: 459-472, 2014