Question on a U-tube bundle arrangement and more 1

[lastone]

I have a question about how one of the condensers in my unit is arranged. Every U-tube or two-pass exchanger in condensing service, with cooling water in the tubes, that I have seen is arranged with the water inlet on the bottom and the tubeside inlet and outlet nozzles are directly in the center of the channel head. Also, the baffle in the channel head runs in the horizontal plane.

The condenser I looked at today is a U-tube, but the baffle in the channel head is in the vertical position, and the inlet nozzle is on the bottom right side with the outlet nozzle opposite on the top left side. I believe the TEMA designation is BEU. The question I have, is if this is a common configuration for a condenser bundle. I know that AEU and BEU exchangers are very common, but the front head baffle in the vertical? The overhead vapor from the distillation column enters the top of the shell at the opposite end and the condensed liquid exits the bottom of the shell at the end near the water inlet.

The exchanger operates as a flooded condenser. The condensed process liquid outlet is throttled by a control valve, which manipulates the level of condensed liquid in the shell. This is the pressure control for the column. There is also a pressure equalization line running from the column overhead line to the overhead accumulator, which ensures the liquid in the condenser shell will gravity drain to the accumulator.

Wouldn't rotating the bundle 90 deg from the horizontal give a less than optimal temperature approach? Or does it not matter since the condenser is flooded. Would the vertical orientation be chosen so that the shell fluid sees an average of the cool and warm water? Would it be possible that with the tubes mounted in this configuration that some of the tubes in the inlet pass do not run liquid full?

I am asking these questions because I have been experiencing a very strange and difficult problem which I have not been able to solve. Every 12 minutes, and you can set a clock by it, the column overhead pressure takes an initial dip (0.3 psi), and then simultaneously my reflux flow cuts in half, the overhead temperature and pressure spikes, my overhead accumulator temperature increases from 95 to 150 deg, the overhead accumulator level increases 15-20%, and the suction pressure on the distillate/reflux pump drops to 0. Actually all of the temperatures in the column spike a small amount, the column pressure drop jumps up a little, (0.5 delta psi increase), and then just 30 seconds later everything returns to normal.

I have zeroed all of the flowmeters on the column and checked all the valves. They are accurate and operating correctly. I have checked in the column, everything is in good shape - all the distributors are level, packing looks good, nothing stuck where it shouldn't be. The reflux, feed, and reboiler nozzles are all clear and functioning properly. The reboiler and overhead condenser are clean as a whistle and no leaks. We have run every controller associated with the column in manual, and it still happens. We have checked the seal on the tops pump and it is in good shape. We have checked the piping from the condenser to the accumulator and its control valve, and it is clear. The accumulator is clear and clean. The equalization line is clear. I have modeled the column in ASPEN using the number of design stages and current operating conditions and the model matches and predicts the steady state conditions. The lab data also matches with the model. I also have used the pack rating feature in ASPEN, which uses the specific packing vendor pressure drop and HETP correlations. At the current operating conditions it says the column is operating at 63-67% of maximum fractional capacity in all of the packed sections. The predicted pressure drops are pretty close to the actual as well. But I don't know how to model for the blip. The column is separating a fairly small amount of C2 and C3 alcohol from a water stream. The distillate rate is about 1250 lb/hr and the bottoms rate is around 25000 lb/hr. The overheads contain about 15% water. I'm not trying to make a perfect separation. Actually I can't.

I know this post is extremely long, and my main question is still concerning the condenser, but if anybody has any ideas of other things we should look at I would really appreciate it. I hate not being able to figure this out.

 

[25362]

I still haven't digested the information you provided, but it seems you haven't looked into the reboiler's performance, and the upstream feed quality (composition- and temperature-wise). Have you ?

 

[Leclerc]

I assume that reflux flow to the column is a controlled flow from the Accumulator. How is this control effected?

 

[lastone]

The feed to the column does swing some. Its temperature is very consistent, but the feed swings +/- 500 lb/hr on about the same frequency. The reboiler performance on this column is steady. The chest pressure on the reboiler is consistent, (or at least as consistent as I can tell from a gauge), and the steam flow is constant. I beleive the trap is working and not backing up in the exchanger. The reflux to the column is on straight flow control. The accumulator level is controlled by the distillate flow. The steam flow to the reboiler is reset by the temperature in the feed distributor tray.

Also, I forgot to include in the previous post, I do not believe we have a problem with inert buildup in the condenser or accumulator. We have run with the inert bleeds open and closed and still see the same blip.

 

[DickRussell]

Does the feed contain any higher alcohols at all? C4 and higher alcohols are increasingly insoluble in water, and thus have a high activity coefficient when present in small concentrations in a lot of water. This makes them easily stripped from the water-rich end of the column (actually more so than the lower alcohols). However, at the top end of the column, as water is depleted and the liquid becomes alcohol-rich, the lighter alcohols will fractionate the heavier alcohols downward. Depending on actual concentrations in the feed and purity overhead, a small amount of heavier alcohol can make it out with the distillate without bottling up too much between the feed and the top, but under certain circumstances the bottling up of the heavier stuff can reach a point where a small second liquid phase appears in the column. This heavier phase will run down the column until it is vaporized by the hotter (mainly steam) flow coming up from below. It is conceivable that at some point instability results in a sudden release of heavier stuff out one end of the column. If this recycles upstream in the process and comes back at you, that could account for regularity in the frequency of your problem. This is a long shot, I know, but the question always must be asked when separating alcohols from water. With Aspen you can include in your model any trace amounts of other stuff and ask the column to check for second liquid phase on stages near the top (say, the top six).

HTH

 

[sshep]

I think this is probably a hydraulic problem. Check your nozzle velocities out of the condenser- there are some rho*V^2 guidelines.

The typical mechanism is that vapor (either vortexed; or formed by flashing of liquid at bubblept pres + small static head) restricts the outlet flow when the level gets low. The level then rises until normal flow is established. When this happens the level starts to draw down again and the cycle repeats.

If the problem is flashing in the nozzle then the orientation of the bundle could be an aggrevating factor. Normally I think that bundle orientation would not be a big consideration in the original design, so you may have encountered something that the designer didn't think of.

Do you have any way to check the level in the condenser to see if it is cycling? You will obviously need to break some control cascades and/or run some loops in manual to weed out the higher level responses from the root cause. Your operators may have already done this for stability reasons.

just a thought, sshep

 

[sshep]

Your post was long and I didn't read the whole thing thoroughly with respect to configuration of the condenser accumulator, or the controllers. I was focusing on the reflux flow and level- if the controllers are in manual, why would the reflux flow ever drop? That is why I am still thinking a hydraulic problem. Please check for some choking of the pump suction line as well by the vapor mechanism which I suggested. thanks, sshep

 

[25362]

Since the starting effect is a dip in pressure, and the presure is controlled by the LLCV, it appears the problem is indeed located around the condenser.

Can it be that air being released from the warm cooling water exiting the condenser creates a bubble and flow resistance every so many minutes until it is re-absorbed or otherwise pushed along ?

 

[lastone]

First, thank you all for your responses and suggestions. I'll answer some of your questions first.

There are no higher alcohols in this stream. C3 is the heaviest of the light components. I wish I had a way to tell the level in the condenser, and there is a magnetic level gauge on the side of it, but it is broken. We are fixing it.

I spoke with one of our heat-transfer specialists this morning and I think he had a pretty good idea of what is going on. He said what is occurring is a symptom of the vapor equalization line being undersized. I'll try and explain what I envision is occurring.

Each time we get an oscillation in the feed to the column, (it goes up 500 LB/HR), we get a slight decrease in column pressure. The flooded condenser pressure controller quickly responds and pinches back to cover more of the exchanger tubes to raise the pressure back up in the column. This takes a little while to occur and while the level is building in the condenser the level in the accumulator is still dropping. This lowers the pressure in the accumulator below the pressure in the condenser momentarily. A significant amount of the hot gas in the overhead line rushes down into the accumulator and heats it up. This hot gas in the accumulator then condenses rapidly and lowers the pressure in the accumulator significantly. It actually causes a vacuum in the accumulator. This vacuum in the vapor space of the accumulator rapidly lowers the NPSH of the overhead pump. This is why I see the suction pressure at the pump go to zero. The pump doesn't stop pumping entirely though. Its now pumping the hot 150 degree liquid that just condensed and not pumping as much. This hot liquid gets fed back into the column as reflux which falls back down the column and the lower flow condenses less of the vapor that is coming up in the column. All of the temps in the column spike and the overhead pressure spikes. The flooded condenser controller sees this spike in pressure and throws open the pressure control valve, which quickly dumps lots of the cool liquid in the condenser down into the accumulator. Now we are back to normal, but then we get another small increase in column feed, and here we go again.

Obviously the immediate answer is to keep the column feed steady and maybe slow down the action of the pressure controller. We will keep trying to do this, but the feed is about as steady as it can get, which is a whole other story that I am sure by now y'all do not want to hear about.
I'm not entirely sure if I understand how increasing the size of the equalization line will improve this, maybe by being larger it is able to transmit changes in the overhead pressure quicker to the accumulator? Anyway, thank you all again for your help.

 

[LuizSouza]

Lastone,

Concerning to vertiacal baffles. I've seen it, here, in Brazil and it's supposed be very commom.

Regards

Luiz Souza
Santos Barbosa Company
A Wood Group Brazil Company

http://www.woodgroup.com

 

[lastone]

Yes Luiz, I have learned that by orienting the tubesheet baffle vertically and having the shell side baffles horizontal, you limit the ability for the shell side fluid to flow in the middle of the shell where there are no tubes. Thanks.

 

[25362]

To lastone. I may be wrong, but could you explain how can a flooded condenser operate with horizontal baffle cuts ? When the alternating "windows" become flooded, there wouldn't be a normal open path for the condensing vapors, except at the clearances around tubes through the baffle holes, and through the peripheral shell-to-bundle clearance. Or are the window cuts "unusual" in this respect ?

 

[rmw]

Not speaking for this particular condenser, but some Hx's where sub cooling is desired have horizontal baffle cuts with the assumption that the "damming effect" caused by the alternating windows will keep the bundle flooded. I am not saying I agree with the approach, but have seen it promoted by certain manufacturers, while their competitors extoll the virtues of vertical cut baffles in the same application.

My preference is for the vertical cut baffles in the application I am referring to.

In this condenser in question, I suspect intermittent flooding, and/or condensing/non-condensable separation zone baffle leakage, but can't quantify my suspicions.

rmw

 

[25362]

How many packing sections are installed in the tower ? Every redistributing plate may introduce a dynamic delay of, say, 20 seconds depending on the volume of liquid on the tray. A similar delay would take place in the reboiler's boil-up.
Can you estimate whether all these delays could explain the 12-minutes span you have noticed between consecutive pressure dips ?

 

[nakagawa]

You wrote "There is also a pressure equalization line running from the column overhead line to the overhead accumulator" Is this true? aren't the overhead balance line installed on the top of the condensate shell to the accumulator? If this is the case, I experienced the similar phenomena and it was caused by siphon effect through pressure balance line. That is
1. The shell side pressure drop is high and is able to flood the condensate up to the balance line with shell inlet portion free of liquid and condensate exit side full of liquid. The liquid gradient is created by the pressure drop.
2. When the condensed liquid is pushed up to the highest point of the pressure balance line, the condensate starts flowing down to the accumulator establishing siphon system.
3. Once siphon system is formed, flow rate is very high and increase the accumulator level quickly. This flow continue untill a point is reached where the siphon breaks.
4. The liquid throu the siphon is the upper part of the liquid in the condenser which is much hotter than the subcooled liquid flowing? throu condensate line.
5. The sudden increase of the effective tube area in the condenser liquefy a vast amount of vapor even pulling vapor from the accumulator after siphon break and bring down the accumulator pressure.
6. The rest of the story is explained by your post.
By the way the vertical pass partition is probably intended for providing smooth effective tube area when flooding condenser for column press control at clean condition and/or low load. I've designed a 45 degree staggered tube bundle for this purpose. Thank you.