Maximum Pump Inlet Vacuum to Prevent Cavitation

Mikerho is quite right, the vapour pressure of the fluid is key. But it also goes beyond this, and pump design, geometry, speed, etc are also factors that will determine the onset of cavitation. As you probably know, the cavitation process begins in the low pressure zone behind the impeller vane (or gear tooth, or whatever) as it attempts to pull the fluid through, so it is at this point that you would really want to measure the pressure vs. the vapour pressure, but in practice this is usually impossible.

In short, I don't think there is a simple formula that will account for all the variables that will allow you to determine a specific figure. You could calculate NPSHa at different suction pressures, and compare with the pump vendors NPSHr curves, but this is not always that precise if you don't know what safety margins are built in.

Your best bet might be just to induce cavitation (if you have a valve in the suction line) and measure the suction pressure at which it occurs, and thus at what pressure the vacuum switch should be set.

Or you could use Lynlynch's accelerometer method: http://cr4.globalspec.com/comment/626955/Re-Centrifugal-pump-in-series-or-flooded-suction-operation

"induce cavitation (if you have a valve in the suction line) and measure the suction pressure at which it occurs, and thus at what pressure the vacuum switch should be set."

It is a good idea but has to be exercised with enough care. The joints with gaskets, O-Ring are mostly designed for pressure applications and might give way on subjecting to negative pressure, unless they are designed for both the services. Also shaft seal is another vulnerable element which might cave in on negative pressure and let air into the system.

Yes, I agree except that you will end up with many false stoppages, caused by fluctuating temps, viscosities and so forth. I tried this years ago on a vane pump and molasses. As some of you may know, molasses viscosity changes drastically with temperature, and thius had the techs running annecessarily many times. In my case a decent pressure transducer, mounted on a 50mm stainless diapraghm ( Din fitting) did the trick. Any vacuum at suction will equate to lower pressure at delivery. All you have to do is set the transducer, after simulating a vacuum at intake. Viscous fluids, from cream to any oil, will allow a certain time of "dry/cavitational" running without damage. The moment cavitation occurs, the pressure shows a distinct drop, and if you add a timer, if the drop persists, the pump will stop or, in your case switch over. My system is now, 17 years later, still funtional, and is now even included in the process PLC.

I firmly believ in not reinventing the wheel.

I am sorry to say that the experience with molasses is not totally transferable to hydraulic oils. Even modern hydraulic pumps are very sensitive to cavitation and even short starvations can be very harmful. Pumps work at quite high contact pressures and have quite high sliding velocities. Lack of oil leads to local heating and rapid surface destruction.

I beg to differ. Molasses come in various viscosities and degrees of impurity. Most pumps for oil are also used for mols. If you go to infinit clearances, then obviously you will use diluted, heated and filtered mols. However, I repeat, ANY pump will tolerate some some cavitation caused by vacuum or air in the system, and the result will be loss of pressure. BUT your term "short starvations" and mine are relative. How long is a piece of string? The trick here is to apply the methods advised correctly, not send a fool into where engineers fear to tread. That is exactly the reason why the OP asked this question. There are NO fixed and hard results for his scenario. Any amount of hypothetical writing and supposition wont change what has been trialed and tested.

Joints with gaskets will handle differential pressure in either direction, as will O-Ring seals if they are trapped on both sides, which is usually the case. If the pump/system is rated for, say 10-16bar, then it will certainly handle the max dP under vacuum conditions (say 800mbar). So I wouldn't be concerned about that.

Mech seals is another issue, but bear in mind that you are only looking at a short period while the point of onset of cavitation is located, and since the stuffing box pressure is usually suction pressure plus about 10-20% of discharge pressure, it will probably remain pressurized during the test anyway. Single mech seals are used on self-priming centrifs that have to pull a partial vacuum during priming, usually without problems.

The trickiest part might actually be identifying the point of onset of cavitation (rather than full-on cavitation, which is usually quite obvious). Different pump types behave differently under different conditions. Noting that, under certain conditions, a gear pump can actually go quieter before it gets noisier, measuring noise or vibration might not be the best method. jvrj's method of measuring the point at which performance starts to drop might be better. As long as the OP can set his controls to switch pumps with a safety margin, before the onset of cavitation.

Thank you Holzfeller, pleasing to note that some posters do read and assess another posters post before blundering on, and completely miss any merits. And you read it correctly, any pump will react very quickly to inlet press/vac changes. The OP decides at what point he switches and thats that. If the oil is a good lubricant he can exyend the switch by a secon or two and still no damage. OF course, to run a pump, any pump, "dry" will guarantee failure. Also remember, you need the volute/casing volume to be fully replenised at every rev to maintain the rated pressure, if that volume changes for whatever reason, the outlet pressure will change negatively. And thats actually all you want!

I'm a bit pushed for time, so I'll read this later and report back.

I don't doubt what you say, but it might be down to the terminology you use. "Cavitation is due to the presence of air which separates and builds up the bubbles." makes it sound like you are talking about mixed phase, with entrained air. It is more accurate to use the term 'gas' or 'vapour' rather than 'air', because this is what you get when the liquid vapourises and moves from liquid phase to gas phase.

My point is that all fluids have a vapour pressure, and if they can be pumped they can cause cavitation where the pressure drops below vapour pressure. The vp of oil might be very different to the vp of water, at a given temperature, but the principle is the same.

The oil is in contact with air and it builds up a solution of air in oil. The amount of air dissolved in oil is characterized by the Bunsen coefficient whose variations are shown in a previous comment.

When pressure decreases and due to shear in the boundary layer temperature increases the 2 combined changes lead to a separation of the air from the solution and build a 2 phases mixture. This is NOT a generation of vapour as you insist to write.

The volume of bubbles grows first and then collapses. This can be even seen in experimental fixtures. There is as a plume from the edge which grows and then decreases and disappear as long as pressure stays over a threshold, if it goes under bubbles cannot any more collapse all of them and a layer partition occurs, this is very heavy cavitation which fortunately does not occur in most cases.

The higher the pressure drop (thus the shear in the boundary layer) the more air is leaving the oil and generates bubbles. I am sorry but I cannot write of "vapour" when the oil did not vaporize since pressure did not decrease under the 0.01 mm Hg which would be required for it. As for gas as far as I know air is a gas (or better said a mixture of gases).

You are right that all liquids have a "vp" and they will vaporize IF this pressure is reached.

There is a "but" in the case of hydraulic systems the separation of the dissolved air occurs at HIGHER pressures than the "pv" for oil so that cavitation is in THIS case not generated by the oil "pv" but by another physical phenomenon.

I am sincerely surprised that you still do not want to understand the differences.

Are you always so convinced to detain the whole truth ?

Being superficial is for people like me; there is no need for you to descend to the same level.

I appreciate you ironic comment but I will take care of what you wrote and will not reduce my level.

Thanks for your rewarding comment.

Your self-underestimation is not correct, you are not superficial and you have a VERY good logical analysis. Even if I do not share sometimes your opinions this does not mean that I do not give them the real value they have.

As far as I understand your problem is a hydraulic system using oil as fluid. In this case the vapour pressure of oil is not the cavitation generator at least at usual temperatures. In oil systems there is a total difference in comparison with water systems since water has a high vapour pressure and oil a very low one.

Cavitation is due to the presence of air which separates and builds up the bubbles.

It is quite difficult to give a 100% valid guide line since the presence of air is very variable function of system design and how different parameters (as joints) do change their behaviour over time. As example for design errors: if the suction is too near to surface in the reservoir tiny turbulence lead air to the entry.

As temperature grows less air will build a solution in oil so that this risk will be less.

In general it is recommended that from the ambient pressure no more than 10% will be lost in the filters or strainers at entry and be sure that the distance to oil free surface in the reservoirs is sufficient. Here is a compromise to be found since if you place to low there more dirt coming in so that...

I think that you should do for your own conditions some measurements (pressure drop versus time) in order to optimize your MTBM.

You may use also noise measurements since cavitation (even small amounts) generates a special noise pattern and this can be used a detector even better than pressure.

Hope it will help.

."#"In oil systems there is a total difference in comparison with water systems since water has a high vapour pressure and oil a very low one.

Cavitation is due to the presence of air which separates and builds up the bubbles."

You make a number of good points, but I would question others.

The fact that water and oil have very different vapour pressures does not mean that it applies to one, and not the other. It's like saying that cavitation happens in centrifs, because they pump water, but not in gear pumps, because they pump oil. (I have lost at least one project due to NPSH and the risk of cavitation!)

Cavitation can occur not only in all types of pumps, on all types of fluids, but it can also occur in other parts of the system. You can get cavitation, depending on design, around a bypass poppet, and I have no doubt there are many other places within hydraulic systems where you can see cavitation.

Cavitation is not due to the "presence of air", it is due to the 'gassing off' (or boiling) of a fluid when it's pressure drops below vapour pressure (the pressure at which it will effectively boil). The destructive part of cavitation occurs when the fluid moves from the low-pressure zone to a high-pressure zone, and the gas bubbles implode. This can happen with any fluid, in many circumstances, the critical parameters being vapour pressure and pressure change.

Furthermore, noise measurement might not necessarily be the best way to determine the onset of, or make measurements of cavitation. The nature and effects of cavitation is different in different pump types and hydraulic systems.

Ultimately, you only have to determine what conditions are acceptable to what you want from the pump and system, and what suction conditions will permit this.

Thank you for the "good points" but I also have to make some comments on the rest:

1- I was involved in research about cavitation in hydraulic systems so that I know what I write about.

2- I NEVER said that cavitation occurs ONLY in water using systems. I am sorry that you lost the project if I could give you a consulting you could have avoided such an unpleasant situation. I only said and repeat that in hydraulic systems cavitation is generated by the separation of dissolved air from the oil.

3- Vapour pressure for water at 37.8 °C is 49.2 mm Hg at same temperature the vapour pressure for an industrial oil (hydraulics) is <0.01 mm Hg. Don't you consider that as a big difference ? If it would be as you claim there is no reason to have cavitation in hydraulic systems!

4- Cavitation occurs at all restrictors (not only poppet valves) where the ratio pressure drop/pressure before restrictor goes over a threshold value which is characteristic to the type of restrictor. This is for instance the reason why if an important pressure drop has to be generated several restrictors are put in series. This is the reason why in several regulating industrial valves there are several edges in series in the control zone.

5- Have you heard of the "diesel effect" in hydraulic systems ? It is the implosion of the air bubbles generated by the shear strain in the oil following the small boundary layer in the restrictor. I saw it it is impressive. The high shear strain generates locally a higher temperature which helps to separate air from oil.

6- Have you made noise analysis of systems before and after occurring of cavitation ? In any "hydraulic cavitation" bubbles have due to the oil properties about same dimensions thus same behaviour at implosion and same noise profile (frequency-amplitude) making the apparition easy to notice. Have you ever heard a cavitating valve ? I did and even measured the noise and made an FFT analysis.

7- Your knowledge is from books, mine from the lab! It is a pity that instead of accepting that there more than you know, you stick to the limits and defend what is not defensible.

In our profession we do not know all and we should be able -if we are professionals - to learn every day from what other want to share with us.

I must recognize that during the time I spend with CR4 I learned a lot and it is in fact the only reason to stay since the level went tremendously down.

If you need more explanations you may ask when you want I shall be glad to give them.

My approach to this issue would be to monitor pressure drop across the inlet strainers, rather than worry about exactly predicting the pressure at which cavitation will occur. As noted by others, process parameters can vary significantly over time, and these can result in a significant difference in the "cavitation point". You can set your differential pressure across the inlet strainers to set off an alarm at some pressure that gives you sufficient margin to accommodate anticipated system variables. Of course, this won't tell you that an operator inadevertently shut the suction valve when firing up the pump...

Rather than worry about DeltaP on a SUCTION strainer, why not just strain the oil as it comes back to the reservoir and monitor that strainer's DeltaP? A parallel, isolated strainer would allow quick switch over when "full" and always give you reasonable inlet pressure for the power side.

We cannot have any restriction whatsoever on the line coming into the reservoir. This is a fairly unique system. Oil is gravity fed into my reservoir from another process in the plant, If the line entering my system were ever to plug, this can cause the entire plant to shut down... (thus my redundancy). My unit will be underground, and acting as a scavenge system for dirty, recovered oil. The system will then have set-points using float switches to turn the pump/s on and off. The oil will be pumped through a filter, and back into the main system. The pumps in this system are cast iron gear pumps, and were spec'd in and designed to chew through dirt, as gear pumps die a slow and painful death, and can generally handle the most contamination of all the hydraulic pumps. In the end, I ran some rough calculations, and figured my normal vacuum to be about 3.5 in-hg when full and about 5 in-hg when nearly empty. during our initial startup, these figures were found to be spot on the money. I have set the kill triggers to be at 7-8in-hg, which seems to be just out of the "red" zone for cavitation on these pumps. All in all, the system works well. Thank you all very much for all the ideas, recommendations, and vast amounts of cumulative knowledge. -R

"The reason being, is I'm trying to spec in a Vacuum switch to automatically switch pumping operation over to the redundant pump while setting off an alarm to notify the operator that the suction strainer in a system is clogged sufficiently to cause cavitation in the pump."

For similar requirement in the compressor - Lube oil and Seal oil applications, the 'differential pressure switch/transmitter' does the same job of alarming and starting the stand-by pump. Such differential pressure switch/transmitter takes it signals from inlet and outlet of filter and watches the pressure drop across the filter.

Well, well, here we have the CR4 general community analyzing and instructing and convincing the OP of their particular pump prowess, while the OP have actually already done one (in a way) of my tried and tested methods i.e.

"Such differential pressure switch/transmitter takes it signals from inlet and outlet of filter and watches the pressure drop across the filter".

He just do not recognize it yet, but he spelled out what i said to NICKN, any problem at inlet will provide drop in pressure ( pressure drop at filter will equate to pressure drop at pump inlet). So, close the eyes, and keyboard, fit the pressure transducer and set it, then come back here and call me a liar. A lot of time is being wasted again while posters analyze each others posts. No doubt a lot of them are already confused if you judge by some of the posts.

To make it very simple any time you are pumping faster than you can feed the pump you are creating conditions that can cause cavitation. Any cavitation is deadly to your pump and will cause failure over time. You can hear cavitation with ultra sound detection equipment such as UE and SDT. I have detected cavitation in gas pumps with pressure feed simply due to fact pump was pumping faster than it was being fed. If you listen with ultra sound equipment you will hear popping sounds which are the bubbles bursting and he empty space slapping the gears or the vanes etc. You may see foaming in your reservoir as a result and it is recommended to discard inlet strainers and use return filters to take care of particle contaminants. You will end up with less maintenance but make sure what ever is being pumped is clean when installed also filtered at this point.

A possibility for misunderstanding seems to be occurring in the concerns over presence of vapor pressure of oil vs. water or other liquids. Cavitation does not need any vapor pressure at all - the most destructive cavitation occurs by simply pulling a vacuum in the system, since there is nothing to cushion the collapse of the "bubble" of nothing. Cavitation can occur in open systems as well as pumps, as in boat propellers, etc. An earlier (no. 2) input pointed out that cavitation is a problem requiring analysis of the whole system. The sound sensors sound like a good idea . . .

Hi woodpower,

True. I remember many years ago I was taught that by far the most aggressive form of cavitation is caused by "holes in a liquid" (not air bubbles), the cooler the liquid the more aggressive the cavitation, as there is no vapour to cushion the high impact velocities of the small particles of liquid, as they implode under an increase in pressure.

Just as a matter of interest, although here, in this discussion cavitation is spoken of in a negative sense, it can be put to very good use.

Indeed, there are a number of applications involving "constructive cavitation" currently in use or in development, including heat generation, hydrogen generation, accelerating transesterification processes, etc.

Woodpower - I'm not sure what you mean by "Cavitation does not need any vapor pressure at all "?

You mention gear pump for the sake of conversation. I too work in the industry and getting generalized rules for how much vacuum is acceptable is highly problematic. You should get manufacturers data on the pump you are trying to protect and the speed it is being driven at. If you can not get it for that specific manufacturer and model try to get specifications from a competitive brand with similar design, and size.

Quick check of a Parker Gear Pump Catalogue. This is a hydraulic gear pump, aluminum housing, pressure balance bushings. Displacement range for this model is 1.4 cubic inches to 5.3 cubic inches. Rated at 10in of Hg (0.35Bar) at 1800psi. Half this amount 5in Hg at rated speed of the pump which varies with pump displacement. 2000rpm up to 3600rpm for smaller displacements.

These specifications will be totally out the window for piston pumps, vane pumps and potentially even the gear pump your customer is using. If you have the pump specifications might be able to be more specific.

Oops - I meant 10in-Hg at 1800rpm not psi.

Every pump curve I've ever seen includes a curve showing the net positive suction head (NPSH) over vapor pressure required at a given GPM. As I understand it's a value determined by testing each pump model, size and speed rather than calculated. Your pump vendor should be able to nail it down for you. Good luck.