Strange top pressure issue in Distillation Column

[mp662]

We installed a new like-for-like condenser on a small distillation column. After startup, we couldn’t pull down vacuum like we used to on the column (500 mmHg now vs 350 mmHg before). After 2 air leaks were found, the top pressure improved slightly, however the top pressure still remained high (450 mmHg.) The reflux is gravity fed out of the bottom of the condenser back to the column, and the overheads is also gravity flow to a storage tank. We have no reflux flowmeter on this column, however the overheads flow is steady.

The top of the overheads condenser goes to a jet ejector and then a jet condenser. Interesting thing is that the overheads condenser makes a “popping” sound like popcorn right at the vapor inlet to the condenser (however the old condenser did this too). It goes away when we block the overheads valve in and force reflux on the column. Also, the reflux line is constantly hammering, which it did not do before. There is no water in this system, all oils.

Things we have tried:
• Swung jets, installed a new jet
• Throttled cooling water on the ohds condenser
• Bled out the condensate of the jet condenser
• Pressure tested all lines and the entire column (we removed all insulation on the feed, reflux, and ohds lines)
• Bled out air on the water side of the condensers
• Opened up the vent to atmosphere on the jet condenser. No puffs on non-condensibles were observed coming out the vent.
• Blocked in overheads and observed the top pressure fall dramatically, however this can be due to cooler reflux going back into the column and decreasing vapor upflow.
• Grabbed temperatures of the cooling water in and out, with a 25 deg C delta. We have not checked the flow yet, as we are waiting for a portable flowmeter to come in.
• Manually blocked in vacuum breaker


The column has 45 bubble cap trays and a 3’ diameter. The other strange thing is that we cannot get a good dP per tray on the column. It used to run at 3.5 mmHg per tray, and now it hangs around 1.5 mmHg. We thought it might be since the pressure is higher and the volumetric flow through the trays is lower, resulting in a lower dry pressure drop, however, the column was run in an Aspen model and we should have 3-3.5 mmHg per tray still.

We ran a simulation of the overheads condenser and found that it has a very high pressure drop just at the vapor inlet to the condenser since the tubes are so close to the inlet nozzle. However, it is the exact same design as before.

It seems like the hammering is due to vapor from the column back flowing and condensing the in the reflux line. Not sure though. What bothers me is that the pressure is high as well as the dP per tray being low.
Any ideas as to what to look for?

 

[katmar]

Why was the old condenser replaced?
Have you checked that the old condenser was built to drawing? Is it possible that some tubes near the vapor inlet nozzle were removed if the same problem was experienced at the original start up?
Is the pressure between the condenser and vacuum jet still the same?
Were any changes made to any piping? Sizes and routes.
Are all nozzles on the new condenser in exactly the same position and orientation as the old one? Same size?
Are baffles in the condenser the same as before? Number and orientation.
Has the steam flow to the column changed?
25 C is a large temperature rise for cooling water. Was it always this high? What is the design rise?
You need to provide an accurate sketch of all the piping showing high and low points and relative heights of the equipment items and valves.
Have you checked the calculated pressure drops in the piping?


Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[katmar]

A follow up to the above:
The most likely culprit, and certainly the first thing to fix before you try anything else, is the hammering in the reflux line. I have seen this before and it has come up in these forums in a similar case - see

http://www.eng-tips.com/viewthread.cfm?qid=137464

The entire relux line should be designed for self venting flow, and there should be a goose neck seal as close to the column as possible. You sketch should include details of this line, and also the detail of how the split is made between the o/h product and the reflux.

Katmar Software - AioFlo Pipe Hydraulics

http://katmarsoftware.com

"An undefined problem has an infinite number of solutions"

 

[sshep]

Hey mp662,

The reflux hammers are most likely due to the colder reflux from the new (clean U) exchanger. The diagnosis of why (collapsing vapor which backflows from the tower) is correct. It depends on the reflux entry design and flow, but subcooled reflux (>20F subcooling) is likely to do this.

How are you collecting your pressure survey data? If this data is just from instruments on the column rather than a direct mounted gage from a field survey, then I would make such a survey. The pressure transmitter sensing lines can pocket liquid (or not depending on inerts) and give a bad reading. The general rule for troubleshooting is to get two completely independent bits of data which agree before making an assumption that something is true. Ultimately troubleshooting is a cycle hypothesis testing and new data collection until the Eureka moment when the answer comes, and then it all finally makes sense.

best wishes,
sshep

 

[mp662]

katmar:

Why was the old condenser replaced? - due to baffle thinning. It was built 60 years ago and in rough shape.
Have you checked that the old condenser was built to drawing? Is it possible that some tubes near the vapor inlet nozzle were removed if the same problem was experienced at the original start up? - Same as original
Is the pressure between the condenser and vacuum jet still the same? - I don't have past data unfortunately.
Were any changes made to any piping? Sizes and routes. No changes
Are all nozzles on the new condenser in exactly the same position and orientation as the old one? Same size? Yes
Are baffles in the condenser the same as before? Number and orientation. - Yes, however the old ones have thinned over the years
Has the steam flow to the column changed? - No changes
25 C is a large temperature rise for cooling water. Was it always this high? What is the design rise? - I will check into that.
You need to provide an accurate sketch of all the piping showing high and low points and relative heights of the equipment items and valves.
Have you checked the calculated pressure drops in the piping? - Yes, no issues

 

[mp662]

sshep:

The pressure instruments have their outputs to the DCS. We put about 5 separate pressure gauges on the column and they all confirm what the instruments read so I don't believe we have a measurement issue.

 

[25362]

It seems to me all the "new" effects are the direct and indirect results of less cooling water flow rate. Look for water flow restrictions inside and outside the condenser.

 

[mp662]

One interesting thing to note is that the reflux is about 85 degrees subcooled. The inlet vapor is 150 C and the outlet liquid is 45 C. Right before the column (about 3') the reflux line is about 130 C. The reflux line drops down about 10' vertically in a goose neck. The temperature at the bottom of the gooseneck is 45 C. So in 10' of pipe there is a 85 degree delta. I think this is the heat from the column pushing back on the reflux line. It's almost like the reflux flow is stalled out.

 

[mp662]

sorry 105 degrees subcooled

 

[sshep]

Hey mp662,

The hammers are almost certainly due to vapor backing into the reflux line right where it enters the tower. You have nailed that.

The next step for you is to verify the flows by energy balance (steam, cooling water, and reflux should all match), you can check the cooling water flow via ultra sonic measurement. A low reflux flow would explain the low tower dP, and aggravate the hammers. In the short term, try raising the reflux to see the impact on hammers and dP. You will figure this out.

Best wishes,
Sshep

 

[sshep]

Any chance of a PFD with controls. Your top pressure problem sounds like an further air leak. The blocking in of overheads (distillate?) was an interesting effect of fixing the vacuum.

 

[hacksaw]

what is going on with the reboiler? and how are you running it, controls etc.

 

[mp662]

So we got the ultrasonic meter working on the cooling water and did a heat balance of the condenser. It is only doing about 1.2 MMBTU/hr cooling, while our heat input to the column is 1.6 MMBTU/hr. The design flow is 50000 lb/hr cooling water, and we are getting 35000. The pressure drop design value is 6 psig, and we’re at 11-12 psig. The design cooling load on the condenser was 2.8 MMBTU/hr. I think we are fouled up on the cooling water side, maybe by a slug of junk that stuck to the tubesheet after startup or something left inside the condenser

I think what happens is that the vapor through the condenser doesn’t get cooled off, which chokes out the jets and raises the top pressure. When the top pressure raises up, we are able to condense more overheads since the delta T is greater between the cooling water and vapor on the shell. This has the effect of lessening the vapor load to the jets, and this drops the top pressure of the column. When this happens, the cycle starts over again. If we let it run long enough, it tends to dampen this effect out.

We also tried putting the column on total reflux, which dropped the top pressure significantly. This makes sense, since the reflux will basically reduce the vapor upflow on the column, and lighten the vapor load to the jets. Once the overheads were opened up again, the top pressure went high and the condenser started its familiar crackle.

We tested this theory by lowering the heat input to the column to about 0.9 MMBTU/hr, just below the maximum condenser duty we could get. The column pulled down nicely and the hammering went away.

We also grabbed sample of the jet condensate. It was about 1/3 full of oil, and the rest water. This shows that we are not condensing the Vapor enough in the overheads condenser and stripping it to the jets

I will follow up with what we find when we open up the condenser heads

 

[ash9144]

A similar problem I saw once was after an extended outage in a unit another completely unrelated HEX was lined up with the CW valves full open. It was normally throttled about half, this robbed the CW flow from other exchangers leading to similar issues.