Kolmetz.com Thermo Paper Page 9
optimum feed stage was chosen as stage 30, which also corresponded to the expected
efficiency for the column. All of the simulation results presented with the pseudo
McCabe-Thiele diagram used stage 30 as the feed stage.
For each of the simulation runs a pseudo McCabe-Thiele diagram was developed. The
pseudo McCabe-Thiele diagram was developed by using the mole fraction data for the
light key and all lighter components calculated for each stage by the simulations. The
equilibrium data and the operating lines were determined from these data. The pseudo
McCabe-Thiele diagram helps one identify pinched regions, mislocated feed points, over
refluxing or reboiling, or where feed or intermediate heat exchangers are needed. (6)
Test Case Results
The results from the simulation runs were compared to the plant operational data. The
feed pressure was specified as 38.6 psia and the feed vapor fraction was specified as
0.4544. The pressure profile was held constant in all simulation runs. In addition, the
simulation runs held constant the feed rate at 67,995 lbs/hr, the reflux rate at 174736
lbs/hr, and the ethylbenzene (EB) composition in the bottoms at 0.671 weight percent.
Theoretical stages were used and the number was based on what is believed to be the
overall efficiency of the tower. Figures 2, 3, and 4 show the comparison of the
simulation results of the commercial ethylbenzene recovery column.
Peng Robinson Simulation Runs
Figure 2 shows a comparison of equilibrium data from simulator A, B and C using the
Peng Robinson model. The equilibrium data for simulators A and C are almost identical.
The equilibrium curves as well as the operating lines lie on top of each other. This
indicates that the results from the two simulation packages will be very close. The
equilibrium curve for simulator B is consistently higher than simulators A and B. This
shift in the equilibrium curve indicates that separation is easier and that the internal
traffic in the column is lower. The operating line from simulator B is identical to the
operating lines from simulators A and C.
A comparison of the simulated results using the Peng Robinson model is given in Table
1. The data in Table 1 show good agreement between the plant data and the simulated
results. There is only a one-degree difference between the overhead temperature
measured in the plant and the temperature calculated in the simulation runs. The feed
temperature calculated for simulator A and B is between one and two degrees less than
plant data. The feed temperature calculated by simulator C is slightly more than two
degrees cooler than plant data. There is some slight variation in the calculated results for
the EB distillate composition and heavies' composition in the bottom. However, all of
the results are considered to reasonably match plant data.
9
