The simulation of dilute emulsions in a model ACIP2 Sonolator is investigated using computational fluid dynamics and population balance methods. Two breakage frequency models are used that differ in the expression of the drop breakage time. Drop breakage modeling based on homogenous isotropic turbulence (HIT) shows poor agreement of the Sauter mean diameter when compared to the experiments; simulations with the empirical model from Alopaeus et al. yield better agreement. By perturbing the classical HIT spectrum, it is shown that the breakage time in the empirical model corresponds to a non-isotropic energy spectrum. Such spectra have been observed in the non-isotropic near field in a model A Sonolator, which provides a plausible explanation of why the empirical model performs better than the HIT-based model.