The lattice model gives a starting point for a theoretical description of the thermodynamic properties of polymer solution systems. Classical models, such as the Flory-Huggins model and the quasi-chemical model, present too narrow or parabolic coexistence curves when compared with experimental data. It is well known that failures of the lattice model are due to mathematical approximations for the effects of nonrandom mixing in order to gain an analytical solution. Moreover, the existing configurational energy of mixing, in which the residual terms are truncated, results in significant errors in the prediction of the coexistence curve calculations for polymer solution systems. The proposed model in this study improves the mathematical approximation defect and gives a new expression for the configurational energy of mixing. To correlate the energy of mixing term, including the effect of non-random mixing on the configurational thermodynamic properties of a binary mixture with simulation data, we use Monte-Carlo simulation data. Monte-Carlo simulation gives essentially exact results for the lattice model. The configurational Helmholtz energy is obtained upon combining the Gibbs-Helmholtz equation with Guggenheim’s athermal entropy of mixing as a boundary condition. The coexistence curves generated by the proposed model are compared with experimental data.