Hexane Recycling at Polycorp
Report prepared by John Stephenson
Prepared for J.S. Parent and J. Dupont
Chemical Engineering, Chee 311
Friday November 30, 2001
Hexane Recycling at Polycorp
Polycorp’s polymer modification plant located in Kingston, Ontario in the basement of
Dupuis Hall is current running a refining process for polystyrene. Polystyrene is
dissolved in a non-polar solvent, hexane, and precipitated by a polar solvent, ethanol. The
precipitated polystyrene is filtered and the liquid effluent stream is passed into a flash
drum to separate enough ethanol from the hexane so that it can be further purified and
reused elsewhere in the plant.
The first stage of the separation process is a flash drum operating at 64C and at
atmospheric pressure. Currently the effluent stream is being fed to the flash drum at a rate
of 1 kmol/h and contains 67mol% ethanol and 33mol% hexane. The Txy diagram for
ethanol/hexane contained in the plant manuals were consulted. It was found that under
these operating conditions, and assuming the equilibrium condition is reached, the vapor
phase will exit the drum at a rate of 0.56 kmol/h and a composition of 50mol% ethanol
and 50mol% hexane. The liquid phase will exit the drum at a rate of 0.44 kmol/h and a
composition of 88mol% ethanol and 12mol% hexane. These percentages were confirmed
by making use of data in the plant manual that contained the Wilson’s equations to
account for non-ideality of liquid mixtures.
Mr. Makework, an engineer at Polycorp, has informed the company that if the hexane
content of the ethanol is less than 3mol% then the ethanol can be used elsewhere in the
plant. To reach just less than 3mol% hexane in the ethanol, a second flash drum would be
required to operate at 73C and at 1 atmosphere. It would receive the liquid stream
exiting from the first flash drum. The vapor would leave the flash drum at 0.23 kmol/h
and contain 80mol% ethanol and 20mol% hexane. The liquid would exit 0.21 kmol/h at
just over 97mol% ethanol and just less than 3mol% hexane.
Recently an on plant emergency as occurred. Mr. Goingtolosehisjobsoon has
accidentally contaminated the hexane feed for the polymer modification plant with a
significant quantity of carcinogenic benzene. The result was a liquid steam from the
precipitation unit that contained 40mol% ethanol, 20mol% hexane and 40mol% benzene.
The flash drum continued to operate at 64C and at atmospheric pressure. Using Wilson’s
equations to account for the non-ideality of the liquid mixture, the problem was modeled.
It was found that the dew point pressure of the mixture would be 0.956atm and the
bubble point pressure of the mixture would be 1.03atm, this meant that the flash drum
would continue to generate both a liquid and a vapor phase. Assuming enough time had
passed in the flash drum for equilibrium to form for the mixture, a vapor phase would
exit the flash drum at 0.44 kmol/h containing 37mol% ethanol, 27mol% hexane, 36mol%
benzene. A liquid phase would exit the flash drum at a rate of 0.56 kmol/h containing
42mol% ethanol, 15mol% hexane and 43mol% benzene. Unfortunately this meant that
two streams were contaminated with significant quantities of the carcinogen.
Appendix
Part A) Normal operation of Flash Drum #1
The Txy phase diagram for ethanol/hexane in the plant manual was consulted. Flash
calculations were drawn up using Antoine’s equations, Wilson’s equations and Modified
Raoult’s Law with the assumption of a perfect gas mixture. The phase diagram, found in
Figure 4, and the equations produced virtually the same results. This verified not only the
accuracy of the Wilson’s equations, and the phase diagram but demonstrated that at low
temperatures and atmospheric pressures that the vapor behaves as a nearly ideal mixture.
The values shown in Figure 1 are from the spreadsheet calculations.
Part B) Designing Flash Drum #2
Having confirmed the accuracy of the phase diagram there was no need to do any more
calculations for this question. The operating conditions to reach an exiting liquid stream
with an ethanol concentration of just above 97mol% was simply read off the Txy
diagram, Figure 4. Once the concentrations of the respective streams were known, the
remaining values in Figure 2 were filled in. The Lever rule on the liquid phase and the
mass balance for the vapor phase were used:
 x b 
L2  L1  1 1 
 a1  b1 
L2 = 2.1x102 mol/h
V2 = L1 – L2
V2 = 2.3x102 mol/h
Part C) Contamination of the liquid effluent stream of with benzene
Part C was modeled using Antoine’s equations, Wilson’s equations and modified
Raoult’s Law in Microsoft Excel 2000. The assumption that the gases acted as perfect
gases was made, this appears to have been a reasonable assumption since for Part A the
Txy diagram and the flash calculations had no significant deviation. This makes sense,
because after all, the pressures and temperatures are quite low and most gas mixtures
behave ideally under these conditions.
Bubble point pressure and dew point pressure calculations were made on the stream. Pbubl
was almost trivial calculation to make, and was found to be 1.034atm. (see “Bubble” for
more details, found in the attached spreadsheet)
The Pdew was harder to make as the liquid stream composition had to be guessed, Pdew
calculated, then the liquid streams calculated, plugged back into where the guesses were
made, and this iteration was continued until Pdew converged. Pdew was found to be
0.956atm. Since Pdew <P< Pbubl the system has two phases present. (see “Dew” in the
attached spreadsheet)
Flash point calculations were then made. Liquid compositions were guessed so that some
activity coefficients could be found, then the percentage of the overall vapor phase was
solved for using goal seek to find a non-trivial, real root for the flash equation. Starting
guesses of around 0.5kmol were used each time, or else the solution would tend to the
trivial solution (V=0). Then the values of the liquid phase composition would be
calculated and plugged back into the activity coefficients so the solution could be iterated
until the compositions of the phases and the overall percentage of the vapor phase both
converged.
Discussion of errors
The most significant errors in applying these calculations to a real world process is that
the flash drums may not come to equilibrium and the exit streams may have significant
deviations from the ones calculated using equilibrium conditions. The assumption that
=1 in the Modified Raoult’s Law is not unreasonable considering that the Txy diagram
(which accounts for non-ideal mixing of gases) was in agreement with the spreadsheet
which did not account for non-ideal mixing of gases (“Confirm simple flash” in the
spreadsheet).