NPSH calculation question
NPSH calculation question
(OP)
Hello,
Mcnally institute gives a bunch of examples regarding calculating NPSH.
http://www.mcnallyinstitute.com/11-html/11-12.html
The example I had a question regards the vacuum tank example.
The example states the tank is under -20 inches of mercury vacuum (-22.7 ft head).
Thus the absolute pressure in the tank is 34-22.7= 11.3 ft
The fluid in the tank (water) is given as 180 deg F with a vapor pressure of 16.7 ft of head.
The vapor pressure of the fluid is higher than that of the pressure in the tank. Shouldn't the fluid be boiling/flashing in this case? If so why doesn't the fluid vapor pressure and tank pressure reach an equillibrium? With either the fluid cooling down as it flashes and/or the vacuum in the tank decreasing?
Mcnally institute gives a bunch of examples regarding calculating NPSH.
http://www.mcnallyinstitute.com/11-html/11-12.html
The example I had a question regards the vacuum tank example.
The example states the tank is under -20 inches of mercury vacuum (-22.7 ft head).
Thus the absolute pressure in the tank is 34-22.7= 11.3 ft
The fluid in the tank (water) is given as 180 deg F with a vapor pressure of 16.7 ft of head.
The vapor pressure of the fluid is higher than that of the pressure in the tank. Shouldn't the fluid be boiling/flashing in this case? If so why doesn't the fluid vapor pressure and tank pressure reach an equillibrium? With either the fluid cooling down as it flashes and/or the vacuum in the tank decreasing?





RE: NPSH calculation question
Good catch!!!
rmw
RE: NPSH calculation question
In this case, pressure and vapor pressure would cancel out and the NPSHA for the pump would be the vertical height between the level in the vacuum tank and the pump centerline minus line losses between the tank and the pump.
RE: NPSH calculation question
RE: NPSH calculation question
rmw
RE: NPSH calculation question
Barring those on atmospheric storage tanks, almost every pump in the refinery is pumping from a vessel where the liquid is at its vapor pressure, which allows you to shortcut the NPSHa calculation as stated. (fractionators, strippers, separators, etc.) Keeping this in mind also helps stop people when they start talking about 'raising tower pressure' to gain NPSHa.
RE: NPSH calculation question
To rzrbk, pumps can tolerate a minimal amount of flashing with no significant adverse cavitation effects.
Thus, I'm inclined to believe that when people suggest raising the tower pressure they probably don't refer to the available NPSH or the liquid reaching its bubble point at suction, but to enable the pump to keep working with a diminished vapor bubble volume at the new higher pressures.
Do you agree ?
RE: NPSH calculation question
It has been my experience throughout many years of dealing with the pumping of saturated fluids that people who suggest raising the vessel pressure don't really have a good thorough technical grasp of what NPSH really is.
They may think they do and say they do, but they really dont.
I for one, as a young engineer did not grasp it immediately and until it finally clicked (we manufactured a seawater distillation product that always had 3 saturated fluid streams to pump) I greatly frustrated the chief engineer.
Until one completely grasps what TD2K has stated above that the saturation pressure and the vessel pressure cancel each other out, and all there is to deal with is the height and the piping losses, one cannot get a grip on the problem.
I took the knowledge I learned about pumping saturated fluids at that firm and used it in later employment to sell billable hours in a lot of situations where severe NPSH problems existed. I had to deal with that approach (just raise the pressure) a lot when the real problem was that someone had not physically built the vessel high enough above the pump suction. In all too many of those situations there was someone, supposedly technical to some degree or the other, telling management, who generally knew nothing about nothing technical except that there was a horrendous cost associated with their pump maintenance, and productivity that all that had to be done was "raise the **(name your process) pressure.
Actually, this can work in the short term, during the transit conditions while the vessel pressure exceeds the saturation pressure of the bulk of the liquid in the vessel, but by the time the genius who suggested that walks off to tell the boss that he solved the problem, the liquid saturates, and the pump goes into distress again, while he is still singing his acolades to the boss. Seen it done.
Just what I have seen.
rmw
RE: NPSH calculation question
The fundamentals of cavitation are explained brilliantly on pages 403-406 (Section 9.8) of Streeter and Wylie, "Fluid Mechanics" First Metric Edition (McGraw-Hill Ryerson, 1983). The first three sentences say it all:
"When a liquid flows into a region where its pressure is reduced to vapor pressure, it boils and vapor pockets develop in it. The vapor bubbles are carried along with the liquid until a region of higher pressure is reached, where they suddenly collapse. This process is called cavitation".
With a boiling fluid in a drum, the absolute pressure in the drum equals the vapor pressure. Therefore, the only way for the fluid to enter the pump without bubble formation is for the height of the liquid level in the drum above the pump centerline minus the friction losses (the actual NPSH) to exceed the minimum required by the pump design (the minimum NPSH).
The MINIMUM required NPSH increases with flow through the pump. Therefore, one must consult the manufacturer's pump curves to determine it. Then, the safe value of AVAILABLE NPSH for pump design at maximum flow can be found after calculating the friction loss.
The confusion cited in this thread by rmw results from a failure to realize that, for a boiling fluid, you CANNOT "raise the pressure" in the drum; the drum pressure is set by an intrinsic property of the fluid, namely the vapor pressure. Therefore, you simply must raise the drum sufficiently to overcome the friction losses AND the minimum NPSH required.
Pump cavitation can sometimes be caused by unexpected factors. Sometimes, the suction line diameter is too small, leading to excesive friction loss above some flow value. On other occasions, an isolation valve in the suction line may have accidentally been left partially closed. In still other cases, there may be debris or other obstruction in the pipe. I once saw a hard hat lodged against the suction strainer. Go figure.
RE: NPSH calculation question
To rmw, those facts are indisputable.
However, (NPSH)A as estimated for "boiling" liquids may be too optimistic when the liquid is "saturated" with dissolved gases.
In these cases the "adjusted" (NPSH)A must be determined to cover for the case of the release of more than, say, 2% by volume of gases at the pump's eye, to avoid cavitation (formation and collapse of vapor bubbles). 2-3% flashed gas being considered -by experts- tolerable and safe from that viewpoint.
Take the case of a pump lifting water from a well exposed to the atmosphere (head=34 ft), with a static head difference of -13 ft, a suction pipe friction loss of 1 ft, at 30oC (VP=0.62 psia= 1.4 ft).
The (NPSH)A would generally be estimated as
It is true that the pressure should never go below the liquid vapor pressure in any application, but as the pressure is reduced the released gas volume increases.
Thus a different "vapor pressure" of about 7.5 psia (17.4 ft) should be selected to keep the flashed gas volume below 2%.
The new (NPSH)A should be
In this extreme example I tried to emphasize the importance of suction vesel pressure when dealing with gas-saturated liquids.
Please correct me if I'm wrong.
RE: NPSH calculation question
Eppur si muove. The effect of higher boiling temperatures, and the corresponding pressures, appears to go more in the direction of those who advise using higher pressures to suppress cavitation.
I'll quote from the Centrifugal Pumps Users' Guidebook by Sam Yedidiah (Chapman and Hall):
Besides, if I interpreted correctly fig. 37 of the Pump Handbook by Karassik et al., the NPSH values required by a given pump for liquid hydrocarbons at a higher temperature are lower than the NPSH values required at lower temperatures.
It seems that the developed vapor volumes have an impact on NPSH at cavitation, after all.
RE: NPSH calculation question
At the risk of sounding audacious, I must say that the discussion of situations where the 'liquid is "saturated" with dissolved gases' is peripheral to the central theme that was in effect until rmw's post of 23 Jul 05 13:53.
Normally, process fluids that need to be pumped are at their bubble point, and therefore saturated by definition. Any reduction in pressure below saturation immediately leads to flashing, an undesirable situation even if it doesn't cause pump cavitation.
Prudent design calls for ensuring an adequate available NPSH under all foreseeable conditions to prevent flashing of inlet liquid as it enters the pump. This is an unalterable principle of good pumping system design.
On this basis, I would venture the following comments:
(1) To alleviate a low NPSH problem for a boiling fluid, it is simply not feasible to change operating pressure in the suction vessel, as the liquid pool comes - soon enough - to its new saturation temperature. As noted correctly by rmw, this HIGHLY TRANSIENT condition persists only for a few minutes in most situations and is related only to process dynamics.
(2) The effect of suction temperature on pump NPSH is hardly a criterion to be considered seriously when selecting the process operating pressure. In a distillation column, for example, pressure is selected based on available coolant and reboiler heating medium supply temperatures, and on the effect of operating pressure on separation efficiency (key component relative volatility). Once set, it is utterly unacceptable from a process point of view to attempt to fix a low NPSH pumping problem by changing operating pressure. Besides, it won't work for boiling liquids.
(3) For those situations where the fluid is not boiling but is merely saturated with dissolved gases, it is worth noting that (at low to moderate temperatures) dissolved gas concentrations are generally very low, based on the Henry's constants. That the evolution of a tiny amount of dissolved gas does not damage pumps should never encourage us to allow liquid flashing into pumps as a normal practice.
(4) I am not familiar with the Centrifugal Pumps Users' Guidebook by Sam Yedidiah (Chapman and Hall). However, several of the quotes provided make very little sense to me. For example, the entire paragraph that begins with the sentence "It is necessary to evaporate..." is about as devoid of chemical engineering sense, confusing, and poorly written as any I have ever seen. If the author is simply trying to say that cavitation at a higher suction temperature for the same boiling fluid may be less severe than at a lower temperature, that observation is quite unrelated to the issue we are discussing.
Please understand that the comments I have made above are based solely on my own experience and in the spirit that a healthy discussion, or even a mild hint of disagreement, can be beneficial in enhancing understanding by all parties.
RE: NPSH calculation question
The subject of NPSH has been dealt with quite extensively in the Pump Engineering forum under the heading of Mechanical Engineers; examples: thread407-51468, thread407-59805, thread407-103568, thread407-120848, thread124-115178, thread798-106798, thread798-121564.
Anyway, it still looks like a pons asinorum for some of us.
It seems to me that in order to give this issue the attention it merits by more specialists, it would be advisable for Chamquark (the originator) to transfer this thread in its entirety to that forum. Do you agree ?
RE: NPSH calculation question
So here are my comments for your scrutiny, perusal and autopsy etc.
The theory of the liquid vapor pressure and tank pressure equilibrium at various pressures seems to be not complete or atleast it is not instantaneous. As far as I think, in the absence of external heat, when the fluid is exposed to a pressure lower than its saturation pressure at the liquid temperature, evaporation starts at the cost of sensible heat of the liquid(which can't be instantaneous). Cancelling out the tank pressure with vapor pressure and calculating NPSH by deducting frictional losses from the static head of the fluid may not give a correct result. Further, as the tank gauge pressure is negative, this will not cancel the vapor pressure.
What 25362 alluded to about Yedidiah seems ok to me and I don't find any problem with it. The specific volume of water vapor at 200F is 33.6093cu.ft/lb and at 50F it is 1702.81cu.ft/lb. So, for any constant volume of vapor, we should evaporate about 50times more water when compared to water at 50F. The latent heat of steam at 200F is 977.58btu/lb and at 50F it is 1064.99btu/lb. So, effectively it is 50x0.9 = 45times more water at 200F.
32nd question in the link below speaks about what 25362 said earlier and the replier promises further explanation upon request.
PumpMagazine
RE: NPSH calculation question
If my mind showed bruises like my body does, I would look pretty beat up from all the fights (with others and with myself) that I have had about NPSH over the years. Plus, I would still carry razor strap scars from the old Chief Engineer (rest his soul) that finally pounded it into me.
One problem clouding the issue at hand is that the properties of water/vapor at 50F are not very real world, so they cloud the logic (to me, anyway.)
I would much rather have the example shown have compared 250F and 150F which are real world temperatures that I deal with commonly and can get my mind around.
I don't get into any situations where saturated water exists at 0.363" HG. My clients are usually struggling to get down to vacuums 10 times that amount. I do deal with chemicals with absolute pressures that low, however.
And, while I have certainly pumped hydrocarbons and a variety of chemicals, by far most my experience with saturated fluids is with water solutions, which may be different with respect to "gas" entrainment vs "vapor" entrainment/formation in the pump suction.
So, like you I continue to debate myself. And like UmeshMathur, I value, and learn by spirited discussions.
After all, I soundly lost the debate with my old Chief Engineer many years ago, and ultimately billed a lot of profitable engineering hours using what he taught me to solve people's real world NPSH problems.
rmw
RE: NPSH calculation question
I am not against the practical and plausible solutions given by you and others to overcome low NPSH but just to get clarification about some quick comments made on this issue.
Though 50F is too low a temperature, it is nearer to 80F which is considered for NPSHr testing.
Keeping all these aside, the discussion gets killed if all participants take it in the right spirit
RE: NPSH calculation question
Did you leave the word "dont" out from between the words 'participants' and 'take' in your last sentence??
rmw
PS, and yes, pumping out of 80F is more of a real world possibility fo me.
I'm still cogitating on the issue and will post some thoughts soon.
As an additional thought (on topic,) one of the tasks that fell to me (assigned by the aforementioned Chief Engineer) was the actual NPSHr testing of the pumps we used in our process. Since the cost (physical height) and successful application of our evaporator products depended upon the pumps working as designed, we tested all pumps in a lab under real conditions. Interesting our results, and the conclusions we ultimately drew as to the accuracy, veracity, and ultimately the believability of various pump manufacturers NPSHr curves.
We did not make our designs from their curves, we did it from testing.
And yes, that old Chief Engineer did wear a belt along with his suspenders.
RE: NPSH calculation question
A little late with the reply, it looks like most of this has been discussed already.
I agree that the suggestion 'raising the tower pressure' could be offered, as you stated, to reduce vapor volume and therefore the effects (noise and damage) of cavitation; but would say that 99.9% of the time this suggestion is offered in a misguided attempt to 'get more suction head'. I would agree with some of the others above, that most people don't understand the concept of vapor pressure and NPSH enough.
Although I haven't gone thru the calcs:
I would guess that in practice, the ability or capacity in to raise pressure in existing installations enough to significantly affect cavitation effects rarely occurs.
RE: NPSH calculation question
I wanted to make a couple of clarifying remarks about your earlier post of 25 Jul 05 12:40 and am sorry I was late getting back from another assignment.
It seems that some of the discussion in this thread is becoming clouded by consideration of dynamic (i.e., transient) phenomena versus the steady-state analysis that was in effect in earlier portions of this thread. The following comments are in order:
(1) LIQUIDS BELOW BUBBLE POINT TEMPERATURE
In the situation where a fluid BELOW its bubble point temperature is being pumped, the reservoir pressure is, by definition, higher than the vapor pressure and the vapor space must contain light gases whose partial pressure makes up the difference between vessel pressure and fluid vapor pressure. This difference provides a positive NPSHa. The net NPSHa at pump suction is somewhat lower, however, because of friction losses. In this situation, increasing reservoir pressure will increase NPSHa, by an equivalent amount, for a fixed system geometry. Doubtless, it is this experience that leads novices to suggest the same solution for saturated liquids.
(2) SATURATED LIQUIDS
Let us now discuss the "normal" situation where a saturated liquid (i.e., one at its bubble point) is being pumped. A temporary worsening of the NPSHa would result from a sudden REDUCTION in reservoir pressure, and would likely result in bubble formation at suction and, possibly, pump cavitation as long as the flashing liquid has not come to the new equilibrium temperature corresponding to the lower reservoir pressure.
However, a low NPSH problem CANNOT be corrected permanently simply by INCREASING reservoir pressure. Anyone who thinks this is just plain wrong. This is because there would be only a temporary increase in NPSHa caused by any increase in the vessel pressure without a corresponding instantaneous increase in fluid vapor pressure. This dynamic transient would disappear, pretty quickly in most cases, as the fluid temperature reached the new (higher) equilibrium value. At equilibrium, the vessel pressure and the fluid vapor pressure would again be equal - since we are talking about saturated liquids - and the NPSHa would be back to its old, lower, value. The ONLY exception to this general statement would be when fluid friction is a strong function of fluid transport properties (mainly viscosity). Normally, in fully developed turbulent flow, while reduced viscosity increases the Reynolds number, this does NOT reduce friction factors significantly.
Therefore, for boiling liquids, I think we can all agree that (a) decreasing vessel pressure is counter-productive, and (b) increasing pressure provides only a very short-lived improvement if you are suffering from inadequate NPSHa. The latter "solution" is advocated mostly by those who do not understand these fundamentals.
(3) GASES DISSOLVED IN LIQUIDS
For liquids that contain dissolved gases, the cavitation problem depends on how much gas will be evolved as pressure is lowered as a result of (a) friction losses, and / or (b) from hydrostatic head reduction, in case the pump suction is above the level in the reservoir. As noted by me earlier (24 Jul 05 20:07), the Henry's constants for gas solubility in most solvents are very small. For example, for air in water, the Henry's constant varies between 5.49E-4 and 10.8E-4 atm/mole fraction between 20-90 deg. C (Page 2-125, Perry et al, editors, "Chemical Engineers Handbook", 7th ed., McGraw-Hill 1997). Unless the air-water mixture is at hundreds of atmospheres pressure and so contains a large amount of dissolved air, it is hard to see a problem with cavitation unless the difference in elevation between pump centerline and reservoir level is inadequate or (hard to imagine in such a case) the pump is located much above the reservoir.
Generally, it takes very high pressures to dissolve significant amounts of gases in liquids. An example from petroleum refining might be hydrogen in hydrocarbons at 3000 psia or more. Another might be a concentrated ammonia solution, where much gas dissolves thanks to electrolytic ionic equilibrium. However, such situations are unusual and require very careful engineering.
Again, these thoughts are contributed with the hope that they will help clarify the fundamentals of the phase equilibrium problem and the issues raised by consideration of transient phenomena.
I wish to extend my thanks to all for their reflections and shared experiences.
RE: NPSH calculation question
There is one footnote I can add to this valued discussion: It is obvious, per mathematical relationships and the expert experience donated by tested and recognized contributors such as rmw, TD2K, 25362, quark, and UmeshMathur that increasing the vapor pressure of a saturated liquid isn’t going to help a steady-state situation. Once the system re-adjusts itself to the new vapor pressure, as UmeshMathur notes, it’s back to the old rules – again. The liquid is still saturated! However my footnote is this old “trick” that I’m sure my colleagues will recognize when I mention it:
I can increase the NPSHa of a saturated (& stored, static) liquid by simply converting it into a pseudo-subcooled liquid. I’ve (& many others) have done this numerous times. It simply means imposing an inert (& hopefully not too dissolvable) gas pressure on the vapor space of the saturated liquid. This works and it works on a steady-state basis. However, there are caveats:
1) The solubility of the inert gas must be very low in the parent saturated liquid;
2) The saturated liquid must be static; i.e., it must be in a storage or holding vessel and not in a process vessel, since the inert might not be suitable for process requirements.
3) When the time comes to refill the tank, you have to overcome the N2 partial pressure.
I’ve done this with LPG and liquid Propane quite a bit in the field. I used Nitrogen, which also lent a flavor of inertness to a combustible vapor space. The saturated liquid remains a liquid and still exhibits its corresponding vapor pressure contribution according to Mr. Dalton. The addition of the N2 partial pressure in the vapor space turns the parent liquid into a “pseudo-subcooled” liquid that exhibits an increased NPSHa at the pump out pump piped to the storage vessel.
I understand perfectly well what UmeshMathur means by stating: “However, a low NPSH problem CANNOT be corrected permanently simply by INCREASING reservoir pressure.” And he is right, of course. I believe he is referring to PROCESS, equilibrium conditions and not to external, engineered conditions as I’ve imposed. I wanted to make sure that the rookies that inevitably (and hopefully) will read this thread, don’t get an erroneous idea from the excellent comments that have been written.
My compliments to all who participated.
RE: NPSH calculation question
However, Montemayor's method is certainly a GREAT way to take a saturated liquid to an unsaturated condition, when it is physically feasible to do so, by pressurizing the vessel with an inert gas.
In reviewing this thread, I think the discussion has been very pointed and well reasoned. It should be valuable to all practitioners who solve tough problems, even those that seem to be "intractable".
RE: NPSH calculation question
I want to point out for anyone who might be struggling to follow the concept (as I have already mentioned that I once did) that in a steady state process (Montemayor, I am leaving your bag of tricks out of this one) that the saturated fluid only exists at the surface of the fluid/vapor interface in the reservoir.
If one were to take pressure and temperature measurements at increments down through the fluid between the surface of the reservoir to the pump suction we see that the fluid begins to subcool with depth by the amount of the static head of the fluid at the depth at which you are making the measurement.
In other words, if we do a "freeze frame" and stop the process for a snapshot, you will have a fluid that is saturated at the surface, and if that surface is located, lets say, 10 ft. above the pump suction, then the fluid is subcooled by a measurable amount which is the vessel pressure at the surface plus 10 ft of static head of the fluid at the pump suction. So, at this instant in this "freeze frame" the pump is pumping a fluid subcooled by a 10 foot margin.
So what is the problem?
Process is a dynamic situation, so that one has to take into account the head losses in getting the fluid to the suction of the pump, and if those losses exceed the amount of static head that is acting to subcool the fluid, then the fluid becomes saturated again, and the bubble point is reached which is a terrible thing to have happen in a pump suction.
So, for pumping situations where the NPSHa is only a matter of a foot or two from the NPSHr, then the fluid is only subcooled by that foot or two as it enters the pump. Or, if the flow of the pump changes so that it runs way out on it's curve and "eats up" all the margin built into the design, then the pump cavitates, or worse, flashes.
Which, in a lot of words brings us back to the statement made by TD2K on 18 July.
So, the moral of the discussion is; don't design saturated fluid pumping systems with close margins in the NPSH.
rmw
RE: NPSH calculation question
Based on Montemayor's caveat no. (3), as long as you can get the boiling liquid into the reservoir at the higher pressure, presumably by pumping, you are OK. For storage situations, it is very likely that you already have a transfer pump. The small increase in discharge pressure (caused by higher downstream reservoir pressure) should not present a problem for the upstream transfer pump.
rmw's last sentence of 28 Jul 05 23:56 should be required reading for all pump designers! Can we all agree that violators should promptly be dispatched to chemical engineering purgatory (repeat the undergraduate fluid mechanics class?).
Cheers.
RE: NPSH calculation question
Good luck,
Latexman
RE: NPSH calculation question
Latexman:
You're right. Although it wasn't specifically noted, I believe the above threads dealt with pulsation-free, irrotational flow of Newtonian fluids.
RE: NPSH calculation question
By chance I came upon an article by T. Henshaw on the Hydrocarbon Processing issue of October 2004, titled NPSHA- how much is enough ?, under the pump/reliability heading, where it is stated that, at least for water services, when the NPSHA is two to three times the NPSHR, impellers may be damaged by erosion, and gives a formula to estimate the safe NPSHA for continued operation of 40,000 hrs for stainless steel impellers.