Hexyl Acetate Sensitivity Analysis

1-hexanol - - acetic acid hexyl acetate + water. 

A scheme of an RD process to produce hexyl acetate is shown in Fig. 4.4. There are two feeds below and above the reactive section of the column for acetic acid and 1-hexanol respectively. Owing to a low-boiling heterogeneous azeotrope, two liquid phases form upon condensation, so that a phase separator is used. 

For process simulation, the thermodynamic properties of the quaternary mixture 1-hexanol + acetic add + hexyl acetate + water have to be known. The vapor-Uquid equilibrium model of that quaternary mixture is, as usual, parameterized based on information from the binary systems alone. There are six binary mixtures for a quaternary mixture, two of which are reactive in our example. The special challenges in the experimental determination of phase equilibria in reactive systems will be discussed later. Using typical G models, at least 12 parameters have to be adjusted to experimental binary data or estimated. Additionally, at least the 

In principle, it would be desirable to have information on vapor-liquid equilibria of all binary systems in the temperature range in which the RD is carried out, which is about 100-150 °C in the case studied here. Furthermore, it would be desirable to have at least some data points for ternary systems (all of which are reactive) and for the quaternary system to be able to check the predictive power of the phase equilibrium model. That ideal situation is almost never encountered in reality. In many cases, even reliable experimental data on the binary systems is missing. In the present study, no data was available for the binary systems acetic acid + hexyl acetate and 1-hexanol - - hexyl acetate. Estimations of missing data using group contribution methods such as UNIFAC are possible, but their quality is often hard to assess. 

Furthermore, in the present case not only vapor-liquid equilibria have to be known, but also liquid-liquid equilibria. It is well known that the simultaneous description of both vapor-liquid and liquid-liquid equilibria often fails with standard models, if quantitative agreement is needed. Therefore, in practice, often different sets of model parameters are used for the distillation columns and the decanters. 

---

A helpful but rarely used tool for deciding whether accurate information on a given binary system is needed for the design of a certain separation is provided by sensitivity analysis. An example for such an analysis of the hexyl acetate RD process shown in Fig. 4.4 is given in Fig. 4.5. The basis of the sensitivity analysis is a given operating point of the RD column and a fully parameterized process model. In our case the NRTL model with two adjustable parameters per binary system was used to describe liquid-phase non-idealities in that model. Parameters for systems, for which no experimental data was available were estimated with UNIFAC in a first step. 

Fig. 4.S Sensitivity analysis Influence of errors in binary data on predictions of hexyl acetate RD (example for a given operating point)...

Such sensitivity analysis is a valuable tool for planning the experimental program for the phase equilibrium measurements, which are often needed to complement the literature data. Another field of application of such sensitivity analysis is the assessment of possible errors in process simulation results due to uncertainties of the thermophysical data. In the case studied here, the sensitivity analysis showed that the two binary systems, where data was missing (acetic add + hexyl acetate and 1-hexanol + hexyl acetate) have an important influence on the process model predictions. 

---