In the toluene-ethanol-water system, the mass fractions of toluene, ethanol and water are 99%, 0.3% and 0.7% respectively. The goal of distillation separation is to remove ethanol from the system.
I would go for NRTL, since the all parameters for the binary sub-systems are available in Aspen and the convergence behaviour of the model is quite nice. UNIQUAC should show a equal performance. Do not use UNIFAC, since this is a group contribution model. In addtition I suggest to also neglect Equations of State here, since the gE model are well suited for VLE calculations.