air bound pumps

[Windward]

Why does entrained air cause problems in a centrifugal pump? I am supposing that the air/liquid mixture separates under acceleration in the impeller, allowing the liquid fraction to flow backward through the air region - internal recirculation. How close am I?

 

[ccfowler]

If the pump is truly air bound, the problem is not so much a matter of recirculation as it is of separation, both centrifugal and quasi-static. Air pockets forming in the suction piping and impeller eye are the real problem. If the proportion of air is very small, then only some slight diminished capacity will be noticed. If much air is present, then the pump can become air bound and the flow can stop completely. Continued operation while air bound can quickly result in substantial overheating and damage to shaft seals.

The source of the air is very important. If it is from leakage into the suction piping, the amount can vary significantly even appearing to cease at times. Repair or modification of the suction piping can fix this problem. If the air is coming out of solution, then you may have to resort to some sort of vacuum pump system to provide continuous removal of the air from the suction piping system.

 

[Windward]

Suppose the air/water mixture got past the impeller eye and was proceeding through the pump in a steady flow about as usual. As you mentioned, this can happen if the proportion of air is low, about 5% or less by volume according to the Hydraulic Institute. No doubt the inlet conditions must also be good, to avoid the formation of air pockets.

If the pump stops working when the air fraction goes above this nominal 5%, is it because the air pockets inevitably begin to form at the inlet, or is there some other possible explanation? That is, no air pockets yet, but some change in the impeller flow?

 

[ccfowler]

The effective net density (specific gravity) of the flow decreases as the concentration of the "well mixed" air increases, so the effective discharge pressure drops too low for the pump to deliver the fluid into the existing pressure of the system at the pump discharge.

 

[Windward]

My speculation that the air would separate from the water in the impeller, leading to internal recirculation and stopping of the flow, must be wrong then. From your comments, I conclude that as long as there are no inlet or outlet problems as you have described, the pump will be able to handle air/water mixtures even higher than 5% air by volume. I wouldn't try to pump pure air, but how high do you think the percentage could go?

 

[ccfowler]

The internal recirculation would be the air finding its way into the eye due to what amounts to centrifugal separation of the lighter phase (air) from the heavier phase (water or other liquid).

Once the net flow stops, friction will cause the liquid to warm within the pump thereby further reducing its density. This further reduces the ability of the pump to re-establish flow into the discharge piping.

I don't really know how great a proportion of air in water can be tolerated. I'm sure that the details of the particular pump and system configuration would be important.

What is the nature of your application that requires handling much entrained air?

 

[Artisi]

The question is how does entrained air cause problems:

The impeller eye is usually the lowest area of pressure in a pumping system, so as entrained air enteries this region is starts to expand to the lower pressure, this in turn takes the place of the liquor being pumped and if sufficient air in coming in it replaces the pumped liquor entirely. Once the impeller eye is full of air, the pump cannot develop sufficient pressure on the air to discharge it against the pressure of the liquor on the pump discharge.
Remember, a pump is capable of pumping air but only the the equivelent head that the pump has been selected for when pumping water, ie, 100 ft head of water is approx. 43psi where as 100 feet of air (I can't recall what 1 foot of air is in pressure) but miniscule in comparision to water.

Also the problem of air "binding" is related to many varied conditions, ie, the inlet pressure on the pump either + or -, the pump type, impeller design etc etc.
But it is accepted that for standard enclosed impeller pump anything over 5% of entrained air is likely to be a problem. However, open impeller pumps or especially designed pumps can handle upward of 25% and under some circumstances more.

Naresuan University
Phitsanulok
Thailand

 

[Windward]

ccfowler, I will be glad to explain my application, but first I ask Artisi to tell me where I can get one of those 25%+ pumps.

 

[Artisi]

Windward (Mechanical)
An outline of your application would be helpful, from here I might be able give you some further advice as how to approach the problem.

Naresuan University
Phitsanulok
Thailand

 

[Windward]

Those who are concerned about the energy problem should investigate the subject of this thread. If a pump can handle a substantial percentage of air in an air/water mixture, then it should be able to compress that air isothermally. The water would be recirculated after the compressed air has been separated for use. If this process works, then the most important question is its efficiency.

Isothermal air compression can greatly increase the efficiency of power generation. At present, the only practical isothermal compressor is the hydraulic compressor. These have been proven effective, but they are too large and expensive for widespread use. I can provide references on this subject if there is any interest.

 

[ccfowler]

Windward,

Liquid ring vacuum pumps have been available for decades. (I believe that Nash was the original commercial manufacturer.) They are rugged, durable, and reliable machines. Although their compression process is closely approximates an isothermal process, the losses associated with fluid friction (of the circulating liquid ring) and the inherent pumping of some of the cooling/sealing water (or other liquid) result in power consumption somewhat greater than that of the comparable theoretical isothermal compression process. In spite of these inherent burdens, it is difficult to imagine circumstances where the efficiency of a liquid ring pump would not be substantially better than the gas compression efficiency of a "conventional" pump applied as you are apparently proposing. Please, bear in mind that your proposed arrangement involves pumping a relatively large amount of liquid to compress a relatively small amount of gas.

Although the liquid ring pumps are most commonly used as vacuum pumps, they can be used to compress gasses to pressures well above atmospheric pressure. Usually, the combination of practical considerations and capital costs do not favor this arrangement.

If relatively high gas compression efficiency is your goal, you would do well to consider multiple stage reciprocating or centrifugal compressors with suitable inter-stage cooling. These machines can compare very favorably with isothermal compression efficiency. As usual, the balance of capital and operation & maintenance costs can be expected to clarify the most favorable choice for a particular application.

 

[Windward]

ccfowler, thanks for hanging in there. In the process I propose, the air would be mixed in small bubbles throughout the water as the mixture passes through the pump. The total surface area of direct contact between air and water would be enormous. Without such a large surface area, the heat transfer would not be fast enough to obtain isothermal compression, since the mixture passes through the pump very quickly.

The liquid ring vacuum pump operates differently. It does not mix the air with the water. Although there is direct contact between the two phases, it occurs only at the inside surface of the water seal. This surface area would be almost negligible compared to that obtained by mixing the air in small bubbles throughout the water. In other words, there will not be nearly enough direct contact between air and water to obtain isothermal compression in the LRVP.

If the LRVP seems to approximate isothermal compression, I believe there are at least three reasons. First, decompression of the air on the inlet side of the pump will reduce its temperature, which will depress the temperature of the air leaving the pump.

Second, the pump is comprised of a substantial mass of metal and water relative to the amount of air it is going to pump. Furthermore, it will not take very long for the pump to establish the vacuum. During this short time, the large cold mass of the pump will absorb some of the heat generated during compression, further depressing the temperature of the air leaving.

Third, once the pump has established the vacuum, there will usually not be any significant air flow afterward. The pump maintains the vacuum but no air is moving through it. If no air is being compressed, then there is no air to experience a temperature rise during compression.

These observations are based on several years of practical experience with the LRVP during service in the US Navy and US Merchant Marine. I did not know that the liquid ring principle has been used in compressors. I googled the application but did not find anything. Can you provide some references on it? In any case, I doubt that they would mix the air in small bubbles throughout the water, and I do not see how else they could achieve isothermal compression.

I recognize that my method would require a large mass flow of water relative to the mass flow of air. This is not necessarily a reason to reject it. I would like to comment further on that and your other suggestions, but this post is probably long enough for now. I will discuss these other points later if there is continued interest.

 

[Artisi]

What started as a simple question seems to have escalated into some sort of academic discusion, although the direction or outcome haven't yet been defined.

The question was asked earlier about the pump (centrifugal)that was capable of handling large quantities of air:

Years back, Allis-Chalmers Pumps, Cincinnatti, Ohio designed and patented an impeller designed to handle large quantities of entrained solids, typically found in sewage pumping in conjunction with a lot of entrained air, this could be fitted to their PW/PWO and CW pump range for a number of different units, it was also available for the Allis-Chalmers Canadian PWO pump,which was also manufactured in Australia.
This was given the name of "Shearpeller" and was used in those applications where a normal centrifugal pump was not capable of performing properly. It also found a lot of use in the paper industry pumping from highly air charged stock chests.

It has a hydraulic efficiency of 55 - 57% and tends to be more efficient than recessed impeller (torque flow) pumps. But bear in mind, if it does the job by operating without any probelms it becomes 100% efficient compared to a unit that keeps air binding or chocking up.

An other option for handling a lot of entrained air is to use a "Froth-pump", these are found in the floatation section of the mineral dressing industry- Manufactures are /were Sala International of Sweden and Warman International Australia and probably other manufactures.

Air handling using the right know-how and equipment is not that difficult - but application experience is the secret to a sucessful equipment choice.

Naresuan University
Phitsanulok
Thailand

 

[jet1749]

Some designs of self-priming centrifugal pumps are also marketed as being capable of dealing with "large amounts of intrained air" particularly those designs with integral separator chambers on the discharge side. Some can run happily under snore conditions with only the fluid within the pump case until more fluid becomes available (subject to pumped product and heat development of course)Seems perfectly feasible, as they can often pull a vacuum in excess of 25inches. Could be worth a look

 

[Windward]

Thanks for all of the comments and information on pumps. Artisi, I am trying to determine whether a centrifugal pump can be used to compress air isothermally. For now, call it a PHAC, for "pumped hydraulic air compressor."

To be useful, the PHAC must have a respectable pressure ratio, say at least 4, but up to 15 depending on the application. Much above 15, absorption of air by the water begins to be serious. It must be at least 70% efficient compared to the ideal isothermal compressor. Although this goes without saying, the cost must be justified for a given application.

Supposing all of these requirements have been met, the most important application for the PHAC would be in power generation. A PHAC/recuperated gas turbine plant would be as efficient as a combined cycle plant, but simpler and less expensive to build and operate.

The PHAC could also be used for air conditioning, operating mostly or even entirely on solar energy (because the low temperature compressed air can easily absorb a lot of solar energy for conversion into power, although this only begins to explain the process). This might be as important as the power generation application, because air conditioning load can be as much as two-thirds of the total electrical load on the grid.

There would be many other uses for a practical PHAC. If anyone wants to test the idea, I have some suggestions.

First, the air should be slightly compressed before it enters the pump. A high static blower would do the job. This will reduce the volume of water that must be pumped, allow good control of the air flow and provide a means to distribute the air evenly in the water flow, without affecting the overall efficiency very much. To maintain approximately isothermal compression in the blower, wet compression could be used.

Second, the slightly compressed air should be injected into the water a little way downstream of the impeller eye. This will prevent the impeller eye from becoming air bound, and there are some other reasons you might think of.

Third, after the air has been separated from the water at the pump discharge, the dissolved air in the water must be removed before it is sent back to the pump suction. Otherwise the pump will become air bound.

Fourth, the bigger the impeller diameter, the better. A bigger impeller will turn slower than a smaller impeller for a given pressure ratio, and it will take longer for the air/water mixture to pass through the pump. The longer the residence time, the greater the heat transfer from air to water, hence the closer the approach to isothermal compression.

A PHAC would no doubt seem to be very large for a given purpose, but if it is efficient that should be no obstacle. It would soon pay for itself.

 

[Artisi]

It seems that the direction I was heading was different to your direction - but no problem - maybe a few of the comments have some validity somewhere in your research.

I’ve spent my whole life in pump applications rather than in academic pursuits therefore, I tend to leave these academic discussions to others more qualified than myself, however, a question on the injection of the slightly compressed air downstream of the impeller eye; How will you achieve this?
If you were to use an open impeller I see that as being easy provided the injected air is of a pressure slightly higher than that at the point of injection– but if using a closed impeller I don’t see this as being easy - and remember, whatever impeller type you use, any air finding its way to the impeller eye will rapidly expand to equal the pressure at the impeller eye - and you are back to where we started - the commencement of an air bound pump.(unless of course you have a high inlet condition as discussed in the last paragraph)
However, if the injected air pressure is equal to or slightly less than the impeller eye pressure you might have a chance of pumping the air with the impeller.

This can be seem in applications where a pump with a low inlet or even negative pressure is very sensitive to entrained air whereas a pump with a high inlet pressure ie., positive inlet head is less sensitive to entrained air.

For -- jet1749 (Mechanical)
The use of "Vacuum primed" or "wet-primed" air-handling self-priming pumps would form another discussion, which I am happy to discuss if required.

Naresuan University
Phitsanulok
Thailand

 

[Windward]

Artisi, I was thinking that it would be best to keep the air away from the impeller eye. With 100% water entering the pump, the impeller will always be able to get a grip and send the water on through. By injecting the air a little downstream of the eye, the air must go along with the water flow, away from the eye, so it can’t cause the pump to become air bound.

However, it would be good to know whether the air can be injected into the water just upstream of the pump. At present, I am not able to run even this simple experiment. If you would like to try it, here are some suggestions.

I would use 100% new water in the pump. That is, don’t recirculate the water for this experiment. Instead, use a constant source of new water. The dissolved air in recirculated water would cause a problem. I have a way to deal with it, but it is best to avoid it for now.

The water entering the pump should be at a slight pressure, say 1 or 2 psig. This can be achieved with a head tank, or maybe a booster pump to supply the main pump. Of course, the inlet piping should run straight into the pump suction for at least a few feet, no bends near the pump suction.

The water should be given a swirl before it enters the pump, in the direction of the impeller rotation. Install an air injector in the pump inlet piping, just upstream of the pump suction. I would make it like a reverse pitot tube, with the outlet of the injector at the center of the water stream, pointing in the direction of flow.

The important point here is to get the injected air into the center of the water stream. If the air were evenly distributed in the water before the mixture enters the pump, the pump would probably become air bound. The air should be divided into small bubbles as it leaves the injector and enters the water.

The impeller can be open or closed, but the bigger the pump, the better. Use the biggest pump you can find and run it at the lowest speed you can manage, although it probably should be at least 600 rpm to start with, or whatever the lowest speed is on your pump curve, so you can determine the flow at that speed. This gets complicated though, because the injection of air will affect the flow to some extent.

For every 100 gallons per minute of water flow, inject about 3 cubic feet of free air per minute. You will need a blower for this, which will handle that much air at the required static pressure. This static pressure will be the sum of two pressures: the water pressure at the outlet of the pitot tube injector, and the pressure drop from the outlet of the blower to the outlet of the injector.

I hope this is enough to go on. The main question is whether the pump will move the air/water mixture in a steady flow without any problems. If it does, then the mixture leaving the pump will be at a higher pressure, and the compressed air can be separated for use.

In that case, the next important question is the temperature of the compressed air. It should be close to that of the water. If it is very much higher, then something is wrong with the concept, or maybe the test setup.

If these conditions have been met, then what is the efficiency of this compressor? But that can be determined later. Please let us know if you get any results.

 

[Artisi]

Am I missing something here, you started by asking why does air effect centrifugal pumps and now you now seem to be asking for someone to undertake some experimantal work for you.

"If these conditions have been met, then what is the efficiency of this compressor? But that can be determined later. Please let us know if you get any results."



Naresuan University
Phitsanulok
Thailand