Controlling both top and bottoms quality in distillation 1

[mjpetrag]

Looking for direction on a column we have been having problems with. Basically the overheads & bottoms product need to be a certain quality, but increasing reflux to get this quality will cause an undesirable light component to go out the bottom.

We have a set point bottom temperature that controls the steam flow to the reboiler, and a top temperature set point that controls the reflux flow. However, the problem is when we get an upset and need more steam in the bottom, we end up over-refluxing the column, which builds up a positive feedback loop and the bottoms product goes off-spec.

We can control the column in manual, but it is tedious and takes up operators' time to react to other issues at the board.

Has anyone had success controlling a column like this in auto?

-Mike

 

[ash9144]

Most distillation towers operate in automatic modes of some sort. Very inefficient if you rely on ops to manually control. And bottoms and tops quality are specified and maintained. Is this a new tower? or new service for this tower? Have you had these problems before?

 

[katmar]

You cannot have independent control over the top and bottom compositions. The quantity of each component in the feed is presumably fixed, so if any one of them does not go out with the tops then it has to go out in the bottoms. This is simply a constraint of the overall mass balance.

It is not a good idea to try to control two temperatures in a column. And it is especially not a good idea to try to control a temperature near the bottom of the column (below the inflection point). The bottom temperature is usually more affected by the pressure drop through the column and this can cause the controller to falsely think that the composition has changed. It would be preferable to put the steam control on a fixed flow, possibly cascaded to the feed flow rate.

It is very usual to have the bottom (or a couple of trays up) temperature indicated as it can show if there is a breakthrough of lights, but not to use it as the control parameter.

Katmar Software - Engineering & Risk Analysis Software

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[mjpetrag]

Katmar,

That's kinda what we have set up right now. We have the steam flow controlling of a pressure-compensated temperature a few trays from the bottom. We also have the reflux flow controlling off a pressure-compensated temperature near the top of the column, but they end up fighting each other

However, what you're saying is to have the steam run either in manual or on a ratio control with the feed, and then control the top temperature or another control temp off the reflux?

The one problem I see with running the steam cascaded with the feed is any change in the feed composition. The feed can be highly variable. So we'll have to have better control of the feed coming into the column. Is that the only way to get this column to behave properly?

-Mike

 

[katmar]

That is quite a sophisticated setup that you have.

You cannot produce two constant composition products from a feed that is varying. The best you can do is control the composition of one product (usually the overheads) and allow the other product to float to satisfy the mass balance. You cannot spin straw into gold (just guessing you aren't Rumpelstiltskin!).

Katmar Software - Engineering & Risk Analysis Software

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[PaoloPemi]

I agree with katmar, controlling a parameter as top (or bottom) operating conditions or composition (from analysis) works in most cases, the alternative could be a "predictive" control, that could be based on an accurate model of the column able to redefine optimal parameters as one changes and force the column operating to the right condition, there are companies which do that but it has a cost.

 

[maddocks]

Hate to say this, but it may be time for a chromatograph and some predictive modelling - in theory you could sample the feed conditions, plug into a simulator and then output the control parameters to achieve the necessary products. Some plants look at product value and try and adjust system paramters to achieve maximum value of the towers products. Maybe that's an option? Changing feed is a tough problem.