Correlating and Predicting Nonideal VLE

The most reliable predictive equations are so complex that they are used only in computer programs. In this chapter we consider several predictive equations that are simple enough for hand (or spreadsheet) calculations, showing their logical basis and application, and indicate their relation to the more complex equations used in large computer programs. We also consider briefly how such programs estimate the effect of nonideal behavior in the vapor phase and discuss solubility parameter and gas-Uquid equilibrium. This chapter is all about nonideal behavior in one phase, mostly in the liquid, occasionally in the gas. For that reason the superscripts that identify phases will not appear in this chapter. 

---

For the most low-pressure VLE (up to several hundred psia) we don t worry much about nonideality in the gas phase if we use the L-R rule we will have reasonable confidence in our estimates in the gas phase. All of the examples before Example 8.8 assumed that the gas phase was practically an ideal gas. However, those examples showed that the liquid phase was often quite nonideal. Our attempts to correlate and predict the VLE have mostly been attempts to correlate and predict liquid-phase activity coefficients. Many mixtures with widely different chemical structures and widely different vapor pressures can be represented reasonably well by Eq. 8.5 and fairly simple equations or prediction methods for liquid-phase activity coefficients. The next chapter shows how that is done. Before we begin that, we will borrow a result from that chapter, and show how the results are used. 

This is a hybrid model and lacks a firm theoretical basis because it combines of the HVO model, which is evaluated at infinite pressure, and G of the Michelsen model, which is evaluated at zero pressure. Nevertheless the authors have shown that this model can be used to obtain successful correlation and prediction of the VLE of various nonideal systems. As we will discuss in Section 5.1, part of this success is due to a cancellation of errors. This model is also included in the computer programs on the disk accompanying this monograph and is tested below in this section for VLE correlations and predictions. 

---