Kolmetz.com Thermo Paper Page 8
specific applications. Some thermodynamic packages are good for highly non-ideal
systems while others are not. The thermodynamic model used might be outside its
recommended range. For example, the temperature or pressure range or both ranges are
exceeded. In many cases thermodynamic models are extrapolated to ranges beyond the
data they have for that model. However, any time a model is extrapolated outside its
range the results are questionable until evaluated for validity.
Physical properties are critical to the success and accuracy of a simulation model. Poor
physical property data may prevent your simulation model from converging or matching
operating data. In this paper, three different thermodynamic packages are used for
estimating the physical property data to match the data for a commercial hydrocarbon
tower. In addition, three simulation packages are used to demonstrate how the results
from the same thermodynamic packages may vary from simulator to simulator. The
variations from simulator to simulator represent slightly different property estimations or
values for the components in the system. The purpose of using different simulators is to
demonstrate that there are variations and one should always analyze the results from a
simulator for accuracy and reasonableness. (7)
TEST CASE
The results from a test case, which is a commercial ethylbenzene recovery tower, will be
examined. Plant data from this tower were compared with results from three process
simulation packages using three thermodynamic packages. For each of the
thermodynamic packages, it was desired to match the external material and composition
data reasonably well. Then the rest of the data, including the internal vapor and liquid
flow rates, equilibrium data, and column diameter estimations were compared for
similarities and differences.
The ethylbenzene recovery tower separates ethylbenzene from heavier materials. The
feed was a 45% vapor mixture at 38.6 psia pressure. Since operational data were
available the simulations were run to match the operational data. The desired products
contained 99.4 percent ethylbenzene in the distillate and 85 percent heavy materials in
the bottoms. Several thermodynamic models were chosen for use in the simulation runs
on this tower. These models included Peng Robinson, Non-Random Two-Liquid
(NTRL) and Van Laar.
Feed Stage Optimization and McCabe-Thiele Diagram Development
To determine the optimum feed stage, simulation runs were performed at several
different feed positions. In the simulation runs, the material balance, reflux ratio, and
total number of stages were kept constant. The feed tray was varied and the light key
component composition in the bottoms was determined. Then a concentration versus
feed stage plot was created and is given in Figure 1. The optimum feed stage is where
the minimum of the curve occurs. From Figure 1 it can be seen that there are several
stages in which the minimum value of the light key is approximately the same. Based on
these data, the
8
