Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
water - toluene azeotrope 1
- Thread starter gilden
- Start date
- Thread starter
- #3
- Thread starter
- #5
Dear Chris,
> thank you for your help. Your data match our predicted values, but the
> predicted values do however not match the experimental data. In our lab
> scale trial a 3 wt%aqueous solution of a organic molecule (with very high
> molecular weight) is continuously fed at +-140g/h into a reactor (filled
> with 500g toluene) together with a 2nd stream of toluene at +- 660g/h. At 70
> mbar, the distillate showed a composition of +- 7wt% H20 / 93 wt% toluene,
> instead of 14.3 wt% H20 / 85.7% toluene as predicted. That's why I started
> to doubt of my predicted values. Do you have any idea what might be the
> reason for this discrepancy?
> thank you for your help. Your data match our predicted values, but the
> predicted values do however not match the experimental data. In our lab
> scale trial a 3 wt%aqueous solution of a organic molecule (with very high
> molecular weight) is continuously fed at +-140g/h into a reactor (filled
> with 500g toluene) together with a 2nd stream of toluene at +- 660g/h. At 70
> mbar, the distillate showed a composition of +- 7wt% H20 / 93 wt% toluene,
> instead of 14.3 wt% H20 / 85.7% toluene as predicted. That's why I started
> to doubt of my predicted values. Do you have any idea what might be the
> reason for this discrepancy?
after ruling out causes like miscalculation and property model inaccuracy, i am pretty out of ideas.
for further information:
the theoretical azeotrope composition was calculated using NRTL-Renon and a modified SRK model. both predicted the above mentioned composition 14.3/85.7 wt% h2o/toluene.
anybody w/ any further ideas?
chris
for further information:
the theoretical azeotrope composition was calculated using NRTL-Renon and a modified SRK model. both predicted the above mentioned composition 14.3/85.7 wt% h2o/toluene.
anybody w/ any further ideas?
chris
Since neither phex nor gilden commented on my message I'd add that a salt added to an azeotrope-forming mixture lowers the vapour pressure of the component in which it is more soluble.
An example: dehydration of ethanol by addition of a potassium acetate solution. Only one column is needed to distill over ethanol.
Separate evaporation of water from the bottom dilute salt stream removes water, leaving the salt solution to be recirculated and added at the top of the dehydration column.
I'd like to hear whether this is the case with the issue in hand. Thanks.
An example: dehydration of ethanol by addition of a potassium acetate solution. Only one column is needed to distill over ethanol.
Separate evaporation of water from the bottom dilute salt stream removes water, leaving the salt solution to be recirculated and added at the top of the dehydration column.
I'd like to hear whether this is the case with the issue in hand. Thanks.
thx for your ongoing efforts, 25362. we also thought about contamination causing the deviation with lab experiments, but this doesn't seem to be the cause, as per gilden's statement.
i do not know the last stage of affairs, i hope gilden will elaborate on his findings. so, meanwhile, keep going with your ideas =)
chris
i do not know the last stage of affairs, i hope gilden will elaborate on his findings. so, meanwhile, keep going with your ideas =)
chris
-
1
- #10
As toluene and water are immiscible, the vapor pressures of toluene and water in the vapor phase above the two-phase mixture of liquid toluene and liquid water will establish themselfes independently. That is, at a given temperature, the total pressure will be the sum of the vapor pressures of pure water (p_H2O) and pure toluene (p_TOL). The azeotropic concentration will be given by p_H2O/(p_H2O + p_TOL). To be thermodynamically exact the solubilities of water in tolene and of toluene in water would have to be taken into account. However, this will not dramatically change the results.
In practice the azeotropic concentration will only be obtained by distillation, if both phases are extremely well mixed with either toluene drops finely dispersed in the aqueous phase or water drops finely dispersed in the organic phase. If this is not the case, the toluene will float on top of the two-phase system and exert an additional hydrostatic pressure on the aqueous phase. The hydrostatic pressure cannot be neglegted if operating at reduced pressures (say 70 mbar). The results will then depend on the geometry of your apparatus.
Another problem with not-well mixed heteroazeotropes is heat transfer. Does the aqueous phase or the organic phase wet the heat transfer surface? If the organic phase is in surplus, it is likely that the organic phase wets the heat transfer surface. In that case the heat of vaporization will be delivered to the organic phase and not to the aqueous phase because there is no aqueous phase present to be evaporated. This is a non-equilibrium situation and you should not expect the phase equilibrium to be established.
I do not know if it comports you, but you are not the first one not to obtain the azeotropic concentration in the distillation of heteroazeotropes.
In practice the azeotropic concentration will only be obtained by distillation, if both phases are extremely well mixed with either toluene drops finely dispersed in the aqueous phase or water drops finely dispersed in the organic phase. If this is not the case, the toluene will float on top of the two-phase system and exert an additional hydrostatic pressure on the aqueous phase. The hydrostatic pressure cannot be neglegted if operating at reduced pressures (say 70 mbar). The results will then depend on the geometry of your apparatus.
Another problem with not-well mixed heteroazeotropes is heat transfer. Does the aqueous phase or the organic phase wet the heat transfer surface? If the organic phase is in surplus, it is likely that the organic phase wets the heat transfer surface. In that case the heat of vaporization will be delivered to the organic phase and not to the aqueous phase because there is no aqueous phase present to be evaporated. This is a non-equilibrium situation and you should not expect the phase equilibrium to be established.
I do not know if it comports you, but you are not the first one not to obtain the azeotropic concentration in the distillation of heteroazeotropes.
For some reason I have found NRTL and SRK are not good models of toluene and water. Instead I have found using Unifac, yes I know water is not really compatable with Unifac, but the results match actual experiments much closer, at least in the ranges of (0-20F) and (.001-14.7psia). I don't have time right now to run a simulation, but give it a try you might be surprized.
good luck.
SCular
good luck.
SCular
- Thread starter
- #12
Dear all,
thank you very much for all your efforts.
Our mixture toluene - H20 is very rich in toluene. We use a glass reactor with a double jacket. The reactor is equiped with an agitator (speed 500RPM) and the reactor has a height of approximately 20 cm. In my opinion the mixture is well mixed. So I do not think we have a two layer system (toluene on top and water in the bottom)as suggested by mjh127. On the contrary, I feel a lot for the "heat transfer problem" (also suggested by mjh127). It is possible that the system is not in equilibrium. Up till now we worked with a R=0, that might also be a reason for the system to be in a "non equilibrium" state. We will try to reflux a litte bit.
In answer to "25362", the high molecular weight substance is soluble in water but not in toluene, so it might also lower the vapour pressure of water.
In answer to "scular", we will simulate the composition with UNIFAC model.
Many thanks, all.
thank you very much for all your efforts.
Our mixture toluene - H20 is very rich in toluene. We use a glass reactor with a double jacket. The reactor is equiped with an agitator (speed 500RPM) and the reactor has a height of approximately 20 cm. In my opinion the mixture is well mixed. So I do not think we have a two layer system (toluene on top and water in the bottom)as suggested by mjh127. On the contrary, I feel a lot for the "heat transfer problem" (also suggested by mjh127). It is possible that the system is not in equilibrium. Up till now we worked with a R=0, that might also be a reason for the system to be in a "non equilibrium" state. We will try to reflux a litte bit.
In answer to "25362", the high molecular weight substance is soluble in water but not in toluene, so it might also lower the vapour pressure of water.
In answer to "scular", we will simulate the composition with UNIFAC model.
Many thanks, all.
dear gilden, excuse me for repeating myself, lowering the vapour pressure of water (ie, rising of its BP) is indeed the effect I was speaking about from the very beginning. Since this is apparently a batch operation, when the boiling starts the effect is less pronounced because of the lower molality of the polymer, however as the concentration increases the effect becomes appreciable at the low prevailing absolute pressures. Anyway, that was only a theory. Good luck.
I do not generally trust the default interaction coefficients in commercial simulators, as you don't know the basis on which they were developed. In my opinion, the best procedure is as follows:
(1) Collect ALL of the liquid-liquid equilibrium data for toluene-water (e.g., from the DECHEMA series of books). Add your propritary data to this collection, assuming that it is for these two components only.
(2) Fit the liquid activity coefficient model (NRTL or UNIQUAC) parameters to this data, using non-linear regression.
(3) Use these regressed parameters for your simulation work.
While it is true that toluene and water are sparingly soluble in each other, they are not completely insoluble either and the system temperature has a big effect.
If you have VLE/LLE data for the other components of the mixture with water and toluene, they too must be regressed for each binary. If no such data are available, UNIFAC may be used to estimate these "less important" coefficients for the UNIQUAC or NRTL models, via the UNIFAC infinite dilution activity coefficient predictions.
Beware of using UNIFAC for such problems if you need accurate predictions. UNIFAC is a group contribution method that, while almost always correct directionally, is hardly ever very accurate. Errors of 30% or more in LLE flash compositions are common. UNIFAC is OK to use when you have no experimental VLE or LLE data, and the estimated binary mixture coefficients are non-critical relative to the main binaries. This is surely not the case with the dominant toluene-water binary for which I know LLE data are available. Also, the UNIFAC LLE parameters are entirely different from the UNIFAC VLE parameters. In general, predicting LLE using VLE parameters (and vice-versa) risks disaster.
Also, in my opinion, liquid phase hydrostatic pressure is unlikely to affect the LLE flash results significantly. If you mix the phases adequately, and adequate residence time is available for equilibrium to be reached, there is no reason to worry about non-equilibrium effects.
It is hard to say much more without knowing the identity and concentrations of the other heavy components and the details of the physical equipment.
(1) Collect ALL of the liquid-liquid equilibrium data for toluene-water (e.g., from the DECHEMA series of books). Add your propritary data to this collection, assuming that it is for these two components only.
(2) Fit the liquid activity coefficient model (NRTL or UNIQUAC) parameters to this data, using non-linear regression.
(3) Use these regressed parameters for your simulation work.
While it is true that toluene and water are sparingly soluble in each other, they are not completely insoluble either and the system temperature has a big effect.
If you have VLE/LLE data for the other components of the mixture with water and toluene, they too must be regressed for each binary. If no such data are available, UNIFAC may be used to estimate these "less important" coefficients for the UNIQUAC or NRTL models, via the UNIFAC infinite dilution activity coefficient predictions.
Beware of using UNIFAC for such problems if you need accurate predictions. UNIFAC is a group contribution method that, while almost always correct directionally, is hardly ever very accurate. Errors of 30% or more in LLE flash compositions are common. UNIFAC is OK to use when you have no experimental VLE or LLE data, and the estimated binary mixture coefficients are non-critical relative to the main binaries. This is surely not the case with the dominant toluene-water binary for which I know LLE data are available. Also, the UNIFAC LLE parameters are entirely different from the UNIFAC VLE parameters. In general, predicting LLE using VLE parameters (and vice-versa) risks disaster.
Also, in my opinion, liquid phase hydrostatic pressure is unlikely to affect the LLE flash results significantly. If you mix the phases adequately, and adequate residence time is available for equilibrium to be reached, there is no reason to worry about non-equilibrium effects.
It is hard to say much more without knowing the identity and concentrations of the other heavy components and the details of the physical equipment.
- Status
Similar threads
- Locked
- Question
