5 replies to this topic

[Butterfly]

Hello

 

I have a steam stripping column were my inlet stream is water with some amount of furfural. I want to get a water stream with a furfural content spec (ppm). I was getting some trouble with column convergence. I don't get a solution if I use HYSIM Inside-Out solver method (in Parameters Tab). However, I do converge the column with Modified HYSIM Inside-Out method, as well as using Newton Raphson Inside-Out method. In all the cases I am working with Peng-Robinson property package.

 

The problem is that with HYSIM Modified, I need only 500 kg/h of steam (and I get rid of all the furfural, getting to a mass content of less than ppb wt), but with Newton I need at least 2500 kg/h of steam to get to a content of several ppm.

 

Which method shall I use? Utility consumption in my plant will be quite different depending on my solving method, as well as column sizing.

 

Thanks in advance.

[Bobby Strain]

Call your friends at AspenTech for help. Or, find experimental data on the binary.

 

Bobby

[Saml]

First: get at least one set of actual data about the furfural water system and check if Peng Robinson is close to that. Equations of state are normally not good for aqueous solutions, specially on systems with partial miscibility.

 

Apart from using an inadequate thermodynamic method, and focusing just on convergence I see two options:

- The solution is unique and your problem is the calculation tolerance.

- The problem has multiple steady state and the two methods converge to different ones.

 

I've personally never found an example of the last case  This was a field of theoretical study during the 90's. If you search on google for "multiple steady state on distillation columns" you will find several academic papers for that time. But I repeat: those are special cases and you will be lucky if you stumble upon one.

 

Regarding tolerances, you are looking into the ppm level of furfural in water. The calculation tolerance should be way below that concentration.

[Butterfly]

Thanks for your help. 

 

Yes, I got the tolerances below the ppm level. I also tried NRTL thermodinamic model, but with this one I am not able to get to my specification (which might also be the case)

[pm11]

well the numerical methods used to solve a problem definitely impacts your solution but there are things to consider:

1- experience: knowing what kind of method works better with your problem

2- knowing methods as well: in this case newton-raphson finds the local minimum ie : your solution may not be the exact one,

but modified hysim has a higher chance to find the global minimum ie: the solution might be the right one in this case.

[serra]

Butterfly,
I know two solvers, the double loop (the so called inside-out proposed by Boston and. Sullivan and then modified by Russel) and the std. Newton,
for many applications the double-loop solver is (at least in my experience with Prode software) quite reliable (compared with Newton which can require initial profiles) and with a proper selection of thermo models I never found examples of multiple solutions,
said that, I think that for a mixture with water, furfural... probably Peng Robinson (at least with std. alpha and vDW mixing rules) doesn't work well,
you may wish to evaluate different models for VLE and enthalpy... for example including complex mixing rules etc.