[gas_treater]

I saw so many centrifugal compressor equipped with KO drum which is as simple as vertical vessel with a demister pad in the vessel's internal. I learned that the demister pad's typical limitation for liquid droplet removal is 10 micron or larger. So does it mean that the centrifugal compressor can handle gas with some liquid droplet as long as the droplet size are not larger than the size limited by the demister pad?

Thank you for any helpful information

[eugenedakin]

Hi gas_treater,

I believe that the particle/droplet size limit is determined by the pump manufacturer and then an appropriate Knock-out drum (with associated demister pad) is sized accordingly.

The main reasons for the knock-out drum are to: 1) catch and retain larger liquid and solid particles and 2) to prevent 'slugs' of liquid from entering the compressor.

When many large particles/droplets enter the compressor there is an increased possibility of fouling.  Lets say there are asphalenes and paraffins at any concentration in your feed.  These materials will eventually coat and plug-off portions of the compressor.  This inturn will increase the workload and decrease the efficiency of the compressor, which may cause it to overheat and lower overall part lifetime expectancy of your expensive compressor. 

If a surge of liquid were to pass through the compressor, mechanical damage will occur do to the signifacently lower compressibility of a liquid compared with a gas.

I hope this helps answer your question

Sincerely,

Eugene

[Montemayor]

Gas Treater:

The answer to your basic question is: Yes.

However, the terse answer deserves and requires an explanation.

We are discussing centrifugal compressors and, as such, this type of compressor has a more liberal tolerance for liquid inclusion in the feed gas stream than its counterpart, the reciprocating, positive displacement compressor.  Nevertheless, the basic requirement for both types of compressors is that NO liquid particles should be allowed to enter the suction stream.  However, practical and real-life situations dictate that some minute and small droplets will somehow always be present in most process streams due to entrainment or other physical effects.  Some droplets will, undoubtedly, enter both types of compressors.  Our goal then, is to limit the size and quantity of the particles.  The reasons that liquid particles are not supposed to enter compressors are:

1. Liquids are incompressible in nature; compressors are designed to compress gases, not liquids.  A compressor that attempts to compress a liquid will be acting much as an automobile trying to drive through a concrete column – the concrete column is going to win.

2. Liquids have a propensity for stripping lubrication from bearing and sealing surfaces.  When this happens in a compressor, the compressor loses again.

3. Liquids are a high concentration of mass as compared to gases.  When liquids hit the impeller of a centrifugal compressor they can not only do a lot of mechanical damage but they also cause an imbalance in a high rpm machine.  This imbalance will damage the bearings and other critical components in the same machine due to vibrations being set up.

Compressor manufacturers have determined - through experiment and empirical field knowledge - that liquid particles of a certain diameter (approx. 10 microns) will not cause any harm to their machines.  And due to this finding, they insist on suction separators on their machines to ensure that only liquid particles of a maximum size will not enter their machine.  That's why your statement “the demister pad's typical limitation for liquid droplet removal is 10 micron or larger” is wrong.  It should say “……10 micron or smaller”.

[Borek]

That's why your statement “the demister pad's typical limitation for liquid droplet removal is 10 micron or larger” is wrong.  It should say “……10 micron or smaller”.

Perhaps my English fails me here, but for me it is not obvious whether quoted phrase tells about droplets filtered out or droplets allowed to pass. At least at first read I thought it means that droplets not smaller then 10 microns are filtered out - and that'll fit compressor requirements.

[Donaldson Tan]

At least I am sure as long as the droplet's diameter does not exceed 10 microns, the compressor would not have any problem functioning.

[gas_treater]

Hi, everyone. Hi, Borek.

Forgive me for my english, what I meaned is droplet filtered. So the demister pad would allowed droplets smaller than 10 microns to go to the compressor suction. I

 have another experience with another type of compressor, which is an Oil Flooded Screw Compressor. In this kind of compressor, the compressed gas will be mixed with lube oil after they can be separated in a lube oil separator. So the consequence for every liquid entrainment - no matter small or big its droplet size - is mixed with the lube oil and reduce the lubrication performance. Does it mean that this kind of compressor is not appropriate to be applied in hydrocarbon gas services which contains some amount of quite heavy HC (C5-C6)?

[Montemayor]

Gas-Treater:

I think we all agree on the idea of not allowing any liquid particle larger than 10 microns into the compressor.  That, I believe, is the basis for separating most compressor KO pots or vapor-liquid separators.  That’s also the idea I’ve always employed in designing and building my own separators.  However, bear in mind that the idea of limiting the droplet size is a bit ingenuous when one considers the practicality of measuring the droplet sizes that do get through.  How is one to measure the droplets diameter?  --with a “droplet meter”?  And once measured, how do you conserve the droplet to prove to others (and yourself) that larger sizes haven’t gotten through?  Of course, there is no such meter.     The answer is that this part of engineering design is done on a theoretical basis and one uses as much contingency factors in order to safeguard the compressor.  It’s a type of “tongue-in-cheek” design.   

You are absolutely correct in asserting that any contaminant in the feed stream to a flooded screw compressor is a potential bad result.  However, your statement “In this kind of compressor, the compressed gas will be mixed with lube oil after they can be separated in a lube oil separator” is not correct.  Flooded screws do not employ circulated oil in the compression chamber in order to lubricate the screws.  The screws are never designed to be lubricated in the compression chamber.  In fact, they should never, never touch each other.  The oil is circulated in order to keep the screws sealed (or “flooded”) from the chamber walls and themselves to avoid slippage and leakage due to inherent mechanical clearances and tolerances.  This maximizes the compression efficiency of the machine.

Screw type compressors (flooded) are quite appropriate for any gas compression operation – as long as you thoroughly separate the feed gas from any liquid contaminant.  As you have noted, any light hydrocarbon in the feed gas will dilute and breakdown the seal oil in the compressor.  This will create compression problems in the pumping of the seal oil and it’s cooling and recirculation.  If the pentanes and hexanes continue to accumulate, they will be a process hazard, plus they will defeat the process of sealing the clearances within the machine.   I hope this experience is of some help.

[gas_treater]

I  read your opinion in cheresources.com about the same issue just right before I write this message. Your opinion about the accuracy of droplet measurement and the suggestion for using Souders-Brown equation and not using droplet diameter as a variable for knock-out drum sizing is very interesting.

I have an interesting story about actual condition in one of our screw compressor's KO drum. After I read the discussion about that issue in cheresources.com, I try to re-calculate the sizing of the KO drum using Souders-Brown equation. And for designed flow rate, the KO drum should be at least have 45 cm diameter, and what we already have is a KO drum with 75 cm diameter and 3.35 m T/T (so theoritically the KO drum's size is adequate). The KO drum is equipped with a mesh pad in its top. The compressor normally operated at only 55-70% from the designed flow. Last week we conduct an ISO 8573-2 test, and at the top of the KO drum we collected 21 ml of liquid per Standard cubic feet of treated gas (gas exit the coalescer). So I'm started to think that the theory which interrelated liquid droplet diameter with separation efficiency is correct. Please correct me if you think I'm wrong, but that was what I've saw.

And about the oil-flooded screw-compressor, this is a statement I've quoted from our oil-flooded-screw-compressor's manual: "The RWB II  is an oil flooded screw compressor. Most of the oil discharged by the compressor separates from the gas flow in the oil charge reservoir. Some oil, however, is discharged as a mist which does not separate readily from the gas flow and is carried past the oil charge reservoir. One or more coalescer filter elements then coalesce the oil mist into droplets.

Nice to know you, Mr Montemayor. If you have any other opinion, please let me know.

Best Regards 

[Montemayor]

GasTreater:

I have the impression that factors involved in the calculation and the operation of a vapor-liquid separator are getting a little mixed up or we are not communicating accurately.

I thoroughly agree with your statement that “interrelated liquid droplet diameter with separation efficiency is correct”.  That is basically a true statement.  What isn’t true is that the “efficiency” of a separator is measured by its ability to separate ALL particle sizes.  I believe you are treating the term “efficiency” in a generic manner when it must be identified as to WHAT PARTICLE SIZE we are talking about.

I have not opined on “not using droplet diameter as a variable for knock-out drum sizing”.  In fact, I believe I clearly stated “Our goal then, is to limit the size and quantity of the particles”.  What I also believe is that I failed to explain that it is practically impossible to measure the particle sizes that are either separated (retained upstream of the vapor outlet of a separator) or allowed to pass on through with the overhead vapor.  The latter particle size is what we have focused on and identified as a nominal 10 microns in size.  And this is, co-incidentally, the approximate size assumed by the Brown-Souders equation to be the size of droplet going out with the vapor.  That is why the Brown-Souders relationship does not take any particle size into account.  It’s already figured in.  Some particles of liquid will always avoid separation in a separator.  There is no 100% efficient separator for all particle sizes.

The important point here is to accept the fact that the efficiency of a separator is based on separating down to a certain size of particle.  The remaining, smaller particles will escape separation and do not figure in the calculation of the efficiency.  This can explain the fact that some liquid can exist downstream of a separator – if one takes into consideration the ability of  some liquids to agglomerate together and form larger particles in those places where there is stable residence time and no turbulence.  I believe your quotation from your screw compressor’s operation manual repeats exactly what I have expounded repeatedly in the past.

I believe your claim that “the consequence for every liquid entrainment - no matter small or big its droplet size - is mixed with the lube oil and reduce the lubrication performance” is a wrong statement and does not represent what is taking place in a screw compressor.  The compressor is flooded with oil not for lubrication purposes, but for sealing purposes.  Therefore, the entrainment of the oil out of the machine is expected and is handled accordingly by trying to effect as - efficiently as possible – the subsequent removal of the same oil and its return to the machine.  That is what I have tried to explain.

I hope I have succeeded in explaining myself better.

[gas_treater]

Mr Montemayor,

Thank you very much for your helpful information. You have made so clear all the point I want to know. Sorry for all the understanding that have took place in our discussion.

Best regards.