Distillation towers in VMGSim

Posted in hacks and kludges on Monday, February 18 2013

I've been spending my days noodling around in VMGSim recently, and I figured now was the time to talk about how I get my towers to converge. Some of my peers have a hell of a time getting their towers to converge, and I think it boils down to a few basic problems in how they specify the tower.

Firstly, you need to know something about your tower first, specifically the mass balance, number of stages, and operating pressure. This requires some thought and a little design before you sit down at VMG and try to simulate.

I have a little tool I like to use to figure out the number of stages, if I am designing the tower from scratch, which is just a simple flow-sheet for doing the Fenske equation.

Fenske flow

In this case you specify the feed, the split, etc in the balance op unit (you could also use a component splitter). In the process calculator you link up the light and heavy key components and it calculates the Nmin. The nice part of this is that once the light and heavy key components are specified the feed and splits can be manipulated however you want and everything else re-calculates.

Fenske calc

A good note for VMG: If there is some calculation you want to do, or some simple control thing you want to do and you can't think of another way, try the process calculator. I usually go into the settings tab and turn off the signal lines because it can look like a crazy mess otherwise.

You could go further and use the Gilliland correlation to come up with a guess for the actual number of stages, or you could just use a rule of thumb and multiply by 2 or something. This gives you a starting point around which you fiddle to optimize your tower.

When setting up the tower you need to specify something. For design I like to specify the mole recovery of the light key in the distillate and the mole recovery of the heavy key in the bottoms. This way VMG figures out the temperatures and reflux ratios needed to make it happen.

Things are trickier with a partial condenser since you need one more spec for the tower. If the product can leave as a vapour, however, you have an easy one since you can set the mole flow of the liquid leaving the top to zero.


I find with specs like this I don't tend to have much trouble getting the towers to converge. Once the tower is converged the first time then it is easy to fiddle with things like the reflux ratio and have it still converge.