15 replies to this topic

[jprocess]

Dear All,
As you know for vertical 2-phase separator sizing, L and D values are calculated independent from each other.
D is a function of liquid and vapor densities and also droplet diameter and drag coefficient.
L should satisfy residence and surge times and some required distance at the top and bottom of demister.
Now if the calculated L and D values results in a value of L/D ratio near 1 is it necessary to modify it?
The answer may be No because with current calculated values the requirements are already met!
The answer can also be Yes because one of the purposes of keeping a reasonable value for L/D is good appearance and a vertical separator with L/D around 1 can not be called a vertical one! In this case what is the minimum required value for L/D?
What do you think about this simple question?

Best of Luck.

[Art Montemayor]

Mojtaba:

I’m glad you brought up this old, simple question again. I just happen to have had a quick, but serious meeting with some of our young process design engineers regarding their interpretation of what the Brown-Souders relationship is and how it should be applied on the actual, mechanically designed vessel. Some of our guys have been laboring under the wrong impression as to the application of the Brown-Souders equation.

So-called “good appearance” is not an engineering criterion in specifying an L/D. Resultant clean vapor with the specified liquid droplet size (as a maximum) should be the principal criterion and not aesthetics.

The height of a separator vessel should allow for the following vertical measurements:

1. A height of liquid inventory in the sump of the vessel;
2. A height of disengagement space between the liquid level surface and the entry of the 2-phase mixture;
3. A height of disengagement space between the entry of the 2-phase mixture and the bottom of the internal separation fixture (either vanes or demister pad);
4. A height between the internal separation fixture and the vessel’s outlet nozzle; this is normally used for maintenance and inspection of the internals and can accommodate a hand-hole or flanged access nozzle on the vessel’s wall. Sometimes, in the case of large vessels, the top nozzle can serve as maintenance and inspection entry.

With the above needs or requirements met, I have never designed or seen a vertical separator vessel that had an L/D close to 1.0. If that were the case, I would seriously consider a horizontal separator as a possible solution. Please submit your spreadsheet calculations for the presumed vessel you have calculated and where the L and D values result in an L/D ratio near 1.0. This is to allow all of us on the Forum to see the specific calculation methodology and the results for ourselves. I am interested as to how you have arrived at that result. The best way to discuss this subject is to have the evidence in hand and not just assume the evidence. If the question is based on reality, then it should be confronted. If this is an academic exercise, then we should also recognize it as such.

[jprocess]

Dear Mr.Montemayor,
Thanks a lot for your valuable comments.
Because the calculation sheet is a confidential document, I prefer not to upload it here. But I am interested of knowing your comments about it. I checked your personal page but could not find any e-mail address. Could you please give it to me here or through sending a private message to me to enable me to send you the calculation and process data sheet?
Many thanks in advance.
Cheers.

[Art Montemayor]

Mojtaba:

At least tell us:

The liquid density;
The vapor density;
The inlet flow rate;
The separated vapor flow rate;
The liquid flow rate;
The diameter of the vertical vessel;
The height of the liquid in the sump;
Height of disengagement space between the liquid level surface and the entry of the 2-phase mixture;
Height of disengagement space between the entry of the 2-phase mixture and the bottom of the internal separation fixture;
Height between the internal separation fixture and the vessel’s outlet nozzle;
The "K" value used in the calculation.

At least we can analyze that data.

[jprocess]

Dear Mr.Montemayor,

Thanks a lot for your kind attention.
Please check your mailbox(I mean your hotsheet.com mail bax). I sent the calculation note to you.
For K factor we use the value of 0.11 proposed by GPSA.

Many thanks in advance.
Best of Luck.

[Alawi]

Mr. Montemayor,

I don’t have any calculations on hand for a vertical separator, but I have actually dealt with a real one! It is rather an old design built by an Italian company used as a crude oil distillation column over head receiver, there is a little more to the design but it will be difficult to explain in just using text, anyway it does exist.

Regarding an old topic about I posted about book copy rights I apologies for not replying, I was out of earth.

Alawi

[eilpar]

I have seldom come across L/D ratio 1 for vertical vessel. While Vapor velocity determines diameter and hold-up alone does not determine height. There are minimum clearances required for LSHH-HLL , LLL-LSLL , for the inlet feed nozzle , top demister pad etc. Please see the attached sketch. Adding all these clearances you normally will end up with an L/D ratio of 2 and above.

Regards

PAR

Attached Files

•  Vertical_vessel.doc   34.5KB   489 downloads

[jprocess]

I have seldom come across L/D ratio 1 for vertical vessel. While Vapor velocity determines diameter and hold-up alone does not determine height. There are minimum clearances required for LSHH-HLL , LLL-LSLL , for the inlet feed nozzle , top demister pad etc. Please see the attached sketch. Adding all these clearances you normally will end up with an L/D ratio of 2 and above.

Regards

PAR

Dear,
What you have mentioned is the typical approach and is also the one that we did but at the end the resultant L/D value is between 1.1-1.3 !

[Art Montemayor]

Mojtaba:

Thank you very much for the information you sent. Now, with additional data in hand I can see what the real picture looks like. As is the case in almost all engineering, reason and logic always prevail.

What I have received from you are, what appear to be, HySys print outs and not calculations. Therefore the design logic and algorithms can’t be analyzed or reviewed. However, the data sheets are very revealing and resolve the issue once and for all.

What you have put forth as “Vertical, 2-phase separators” are not even separators in the strictest interpretation. They have been “designed” (and we don’t know by what method) as compressor suction drums – and rather large drums, at that. If we read through the documentation, we find the real truth behind the final dimensions of these 2.75 mt (diameter) x 3.10 mt (T-T length) vertical vessels.

The data sheets clearly state:

• The vessels are normally operated as dry drums;
• The gas has been dehydrated at the offshore source, resulting in no free water at the compressor inlet conditions;
• The design liquid rate for these vessels = 0.0 m3/hr

The above data clearly explain what I have stated. When you take into consideration that there is no liquid allowed for, then these vessels’ scope of work is to merely function as surge vessels or pulsation drums – not as 2-phase separators. I think my original premise holds true.

As is often the case, when we hear the rest of the story – the real story – the picture changes. This is why I always insist on seeing all of the data and the calculations, together with references.

Additionally, when I apply the height of the 2:1 ellipsoidal heads to the height of the vessels, I get:

• Medium Pressure Suction Drum, L/D = 1.62
• High Pressure Suction Drum, L/D = 1.73
• High Pressure Discharge Drum, L/D = 1.8

All of the above are distinctly different from the original L/D values we were given.

[jprocess]

Dear Mr Montemayor,
Thanks a lot for your valuable comments.

The above data clearly explain what I have stated. When you take into consideration that there is no liquid allowed for, then these vessels’ scope of work is to merely function as surge vessels or pulsation drums – not as 2-phase separators. I think my original premise holds true.

1. You mean that for surge vessel or pulsation drums there is no need to keep a reasonable L/D value?
2. The only difference for dry case when sizing a vessel is that the minimum required length will be used(for example between LLLL-LLL) but still it should be accurate enough to results in a reasonable length. It seems that our criteria is not accurate enough because the calculated L is near to D.

Additionally, when I apply the height of the 2:1 ellipsoidal heads to the height of the vessels, I get:

• Medium Pressure Suction Drum, L/D = 1.62
• High Pressure Suction Drum, L/D = 1.73
• High Pressure Discharge Drum, L/D = 1.8

All of the above are distinctly different from the original L/D values we were given.[/font][/size]

But the L value that should be used for L/D evaluation should be tangent-tangent length and not the total length including the heads, Am I wrong?

[Art Montemayor]

Mojtaba:

1. Yes, I believe that for GAS surge vessel or pulsation drums there is no dependency on the L/D value. I don't understand how it could be of any practical influence on the expected results. Does anyone disagree with this stance?

2. In your comment you are forgetting that in the case for a GAS vessel (your "dry" case) the use of a demister pad - or any other internals for that matter - is not a practical or logical option. There are no liquid droplets, particles, or Stokes Law to take into consideration. The Brown-Souders relationship is not applicable.

3. I have always referred to the OVERALL length (or height) when discussing a vessel's L/D ratio - particularly a vertical vessel (as in this case). The reason for this being that the over-turning moment and the wind stresses have to be considered when dealing with large L/D numbers. As you can see, in the case of rather large diameters, the effect of the height of a 2:1 ellipsoidal head starts to have an effect -- and I haven't even added the height of the obvius skirt that is required for the vertical vessels in this example.

My personal opinion is that - aside from aesthetics - there is far too much importance and discussion given to the L/D ratio for vessels. If a vessel meets the minimum process requirements for it, it matter little or nothing what the L/D ratio is.

My response to the basic question you posed: "....if the calculated L and D values results in a value of L/D ratio near 1 is it necessary to modify it?" is NO; if the vessel does the job, leave it alone. The fact that a vessel with L/D = 1.0 cannot be called a vertical vessel means absolutely nothing if the vessel is doing a proper job as calculated and expected.

[jprocess]

Dear Mr Montemayor,

2. In your comment you are forgetting that in the case for a GAS vessel (your "dry" case) the use of a demister pad - or any other internals for that matter - is not a practical or logical option. There are no liquid droplets, particles, or Stokes Law to take into consideration. The Brown-Souders relationship is not applicable.

I can not accept this rik to leave the compressor suction drum without any demister!

[sdarone]

Dear Mr Montemayor,

2. In your comment you are forgetting that in the case for a GAS vessel (your "dry" case) the use of a demister pad - or any other internals for that matter - is not a practical or logical option. There are no liquid droplets, particles, or Stokes Law to take into consideration. The Brown-Souders relationship is not applicable.

I can not accept this rik to leave the compressor suction drum without any demister!

Mojtaba,
My experience is limited, but i had the opportunity of being involved in the operation and the design of compressors suction drums.
regarding your statement of not accepting the risk of not having a demister in the compressor suction drum, i guess is based on the likelyhood of having some liquid carryover to the compressor, however from what Art mentions, in the design basis the person who provided the design assumed that such likelihood is really low to the extent of saying will never happen.
Therefore is not about the drum appearance what is into play but the design basis, which is the start point for every single design.
The design Basis are stablished based on the experience and knowledge of the process, and the environment. So i have a few questions myself :
1 - Is this an existing drum or new one
2 - Is this an existing unit, were you are replacing only the drum or is a complete new unit.
3 - Is there any existing unit in operation with similar configuration and operating conditions from which you can learn and compare?

If this is a new unit, and you have no similar units around to learn from them, then i will say that the assumption of having no liquid carryover on the inlet gas stream is too optimistic, and probably you should consider on revising this statement and use some typical industry values. The design for the drum shall be updated for the new design basis.

I Hope this helps.

Sergio

[jprocess]

Dear Sergio,
Here is a new drum for a new plant.
Anyway thanks a lot for your reply.
Cheers.

[djack77494]

Dear Sergio,
Here is a new drum for a new plant.
Anyway thanks a lot for your reply.
Cheers.

First a very general comment: We oughtn't get hung up or fixated on specific numbers for any engineering results. Applying rules of thumb and typical values or ranges is fine, but knowing the reasoning behind the numbers is far better. You'd be amazed a how often it is prudent or even necessary to (temporarily) discard these rules for various excellent reasons. Even published guidelines by well respected organizations are "guidelines" (duh), and even the authors understood that there are times when it is necessary to "step outside the guidelines".

For a high speed centrifugal compressor, I personally would never design (i.e. expect) an always dry Compressor Suction Drum. If I have a H&M balance or simulation output that suggests dry gas, I will keep in mind the possibility that my recycle anti-surge line may have sat in contact with saturated or near saturated gases for a long period of time. During that time, condensation may have occurred, and I sure don't want to chance introducing a slug of liquid into the compressor when the anti-surge valve opens. I think that is the reasoning behind Sergio's comment, and I absolutely concur with his opinion. There may be other reasons (e.g. startups, shutdowns, upsets) that generate wet gas. A demister is a small price to pay to safeguard a very expensive machine against unforeseen circumstances. Just do it.

As we design most vertical drums, we typically provide for a lower liquid surge and well as connections for instrument lead lines to sense the liquid and a separation over which to work. Sometimes we have two liquid phases, and we may even have some "dead space" in which to allow solids to settle out. Above the highest liquid level will be one or more feed nozzles, and further up will be the demister. (There may also be various baffles inside the drum.) I tend to allow 1/2* drum diameter from the top of the uppermost feed nozzle to the bottom of a demister (assuming we have one). The demister is typically 6"+, and above it I allow 1 ft of vapor space. More often than not, I'm working with a "K" of 0.35. (I do realize that these numbers are rather conservative, but I'm quite comfortable with them, and, in most cases, so are my clients.) Now, it's pretty hard to put these numbers together and not get to a vessel having an L/D ratio of 3:1 or higher.

I concur with Art that a dry gas surge vessel (what we used to like to call a "bump in the line") is not bound by any of these considerations and can have any L/D ratio as dictated by a variety of economic, space, weight, convenience, and other considerations. But even in this case, I understand that for pressure vessels the most economical construction implies that the L/D ratio will be in the 1.5 - 6.0 range; I think of 2 - 5 as "typical". Of course, as the design pressure increases, the tendency is for larger L/D ratios. (Please understand this and the reason for it.) Also, for really large vessels, which are typically not really high pressure, the tendency is for smaller L/D ratios. Ultimately you can have an atmospheric storage tank with a diameter several times larger than its height! (But that's not what we're talking about.)

Ultimately, I think Art again "hit the nail on the head" when he stated, "there is far too much importance and discussion given to the L/D ratio for vessels"

Doug

[jprocess]

Dear Doug,
Thanks a lot for your valuable comments.

For a high speed centrifugal compressor, I personally would never design (i.e. expect) an always dry Compressor Suction Drum. If I have a H&M balance or simulation output that suggests dry gas, I will keep in mind the possibility that my recycle anti-surge line may have sat in contact with saturated or near saturated gases for a long period of time. During that time, condensation may have occurred, and I sure don't want to chance introducing a slug of liquid into the compressor when the anti-surge valve opens. I think that is the reasoning behind Sergio's comment, and I absolutely concur with his opinion. There may be other reasons (e.g. startups, shutdowns, upsets) that generate wet gas. A demister is a small price to pay to safeguard a very expensive machine against unforeseen circumstances. Just do it.

This is exactly the reason that I wrote in my earlier post that can not accept the risk of ignoring a demister for a compressor suction drum even if this is a dry one!

As we design most vertical drums, we typically provide for a lower liquid surge and well as connections for instrument lead lines to sense the liquid and a separation over which to work. Sometimes we have two liquid phases, and we may even have some "dead space" in which to allow solids to settle out. Above the highest liquid level will be one or more feed nozzles, and further up will be the demister. (There may also be various baffles inside the drum.) I tend to allow 1/2* drum diameter from the top of the uppermost feed nozzle to the bottom of a demister (assuming we have one). The demister is typically 6"+, and above it I allow 1 ft of vapor space. More often than not, I'm working with a "K" of 0.35. (I do realize that these numbers are rather conservative, but I'm quite comfortable with them, and, in most cases, so are my clients.) Now, it's pretty hard to put these numbers together and not get to a vessel having an L/D ratio of 3:1 or higher.

I guess that our design criteria is not a good one especially for minimum distances that resulted in L/D ratio of near 1.

Thanks again.
Cheers.