Optimization of the refinery production is essential for the economic success of a petroleum refinery. Classical refinery production planning models employ input output relationships for different refinery units. However, these input output relationships usually do not include decision making variables related to the units' operation (pressure, temperature, etc.). In addition, they fail to capture the nonlinear nature of the refinery units, and the inaccuracy caused by these models may reduce the overall profitability or compromise product quality. The corresponding nonlinear refinery problem is nonconvex due to the presence of bilinear or quadratic, exponential or logarithmic terms in some of the mass balance, product yield, and quality constraints, and hence the standard methods for solving this refinery-wide optimization problem may fail to converge to a solution or lead to suboptimal solutions. In this work, a planning model for refinery-wide production is proposed which employs two different nonlinear models for the distillation unit and empirical nonlinear models for the remaining refinery units. The effectiveness of the nonlinear models in terms of improving the overall refinery profit margin over a linear input output model has been investigated. In addition, the global optimal solution of refinery profit substantiates the huge advantage of nonlinear model utilizing distillation correlations based on Geddes fractionation index over the data-based nonlinear and linear input output models. From a managerial perspective, this study provides an effective tool to make advanced decisions on the amount of the crude oil to purchase and to obtain higher profits by adjusting process severity which results in maximum yield and the desired product property.