Liquid Separation From Plunger In Reciprocating Pump (Water) 3

[SNORGY]

Sorry to ask a rather odd question...

In the PUMP HANDBOOK - Karassik, Krutsch, Fraser, Messina...

In Chapter 3 (Page 3-11, 3-12 in Second Edition) is offered an equation to estimate the pump rotational speed at which liquid will separate from the plunger. My question is (questions are), what values apply for "l", "hf", and "As" when you have a suction piping arrangement consisting of several sections of different diameters and lengths of pipe leading into the pump? Is each section to be looked at individually or is there some kind of blended rationalization / normalization that would need to be applied? In particular, "l" is defined as:

"l = length of pipe where resistance of flow is to be measured"

This wording kind of leads me to think that the intent is to look at each section / segment of pipe (each with one single applicable diameter and length) individually for "l" and "As", but then would "hf" be the friction loss specific to that segment or to the whole piping system?

Maybe this equation cannot be relied upon in situations like this. Any insight on this is appreciated, as I am not a pump expert.

Regards,

SNORGY.

 

[btrueblood]

Snorgy,

You should be able to derive the head loss term for each segment of the suction pipe, and then sum them into a single loss coefficient. e.g.

dP = (K1 + K2 + K3)*rho*V^2/2

where K1, K2, K3... are the loss coefficients (or equivalent fL/D) for each segment in the string.

 

[MikeHalloran]

Clearly, hl applies to the whole suction system. As should the remainder, as the equation is trying to evaluate conditions at the piston face.

So if the suction system is at all complicated, I'd evaluate its head loss at a representative but arbitrary flow, back-calculate a single suction pipe with equal head loss, and use the appropriate parameters of that equivalent pipe in the pump equation.

In any reasonable system, the pipe immediately adjacent the pump should dominate the numbers anyway, so after going through all the machinations suggested once, you _may_ find it's safe to ignore the more distant, larger diameter stuff in subsequent iterations. But don't skip evaluating the whole system at least once.







Mike Halloran
Pembroke Pines, FL, USA

 

[JJPellin]

I know this is not relevant to the questions you have asked. But, a very interesting paper was presented on this topic at the International Pump Users Symposium this year in Houston. The title of the paper was "Cavitation in Reciprocating Positive Displacement Pumps." The main conclusion of the paper was that come amount of cavitation is almost unavoidable. But more interesting was the conclusion that cavitation in a reciprocating pump does not have any detrimental affect unless the amount of cavitation is quite extreme (full cavitation). For lesser levels of cavitation, they found no damage to the pump and no detrimental affect on pump performance or capacity.

Johnny Pellin

 

[SNORGY]

btrueblood:

I have in fact done as you have suggested, evaluating all of the individual segment friction losses together, and then doing something similar for the corresponding acceleration head components.

JJPellin:

Your response is directly relevant to my question, and I thank you for it. This exercise for me has come as a direct result of my involvement a while back in trying to determine and explain how a 300 hp triplex pump in water service failed, shattered the plungers, and expelled the crankshaft from the pump onto the floor in the building. It was a pretty spectacular failure. Is there a link to somewhere I can get a copy of that paper?

MikeHalloran:

I think your approach aligns with mine. I am thinking that what I need to do is quantify the total system head loss component, the total system acceleration head component, and then "normalize" that into a hypothetical length of pipe having equivalent diameter to that segment of pipe that is connected directly to the pump suction nozzle. I do not think that the friction (and elevation) head losses will exactly match simultaneously if I do this, so I am inclined to make the acceleration heads match and then make up the difference between the friction / elevation losses in terms of more or fewer fittings. That's probably reasonable since the acceleration head will probably be found to dominate anyway for a system to which this analysis needs to be applied.

The above said, I am struggling a bit with the visualization of an effect that I am not sure whether it even occurs. That is, suppose you have a segment of reduced diameter piping somewhere in the system significantly upstream of the pump suction nozzle and pulsation dampener (if present). Due to the nature of the reciprocating pump (pulsation source) and with the assumption that the suction system is completely liquid-filled, is it possible to have an effect *similar to* "separation of liquid from the plunger face" occurring in that segment of pipe, i.e., can you have column separation occurring upstream in the piping system and upstream of compensating effects afforded by the suction pulsation dampener (if present)? If so, then you might not have a pump issue but still have at least a vibration issue somewhere else. That led me to think that maybe an equation similar to that in Karassik might be applicable in each individual segment of piping. That is what led to my question. I think the landing that I have come to is that by following MikeHalloran's approach I can quantify the effects on the pump and predict the potentially problematic operating speed (i.e., onset of plunger face fluid separation), but to look at similar effects elsewhere in the system probably involves either software or some other approach that, at the moment, I don't have.



Regards,

SNORGY.

 

[MikeHalloran]

Considering the crank rpm and the rod geometry, if the instantaneus flowrate going into the pump exceeds what you can get through the upstream pipe at that flowrate with available pressure difference, the pump will be starved, and the fluid will separate, or boil, or something. The pump won't like it much either.







Mike Halloran
Pembroke Pines, FL, USA

 

[JJPellin]

You should be able to view the paper at this site:

http://turbolab.tamu.edu/articles/27th_international_pump_users_symposium_proceedings

Johnny Pellin

 

[SNORGY]

Thanks all for your help and suggestions on this.

Regards,

SNORGY.

 

[BigInch]

Separation occurs whenever and wherever the pressure in the fluid at any point goes below vapor pressure. The trick will be accurately evaluating the pressure at any given point with local accelerations increasing and decreasing at all points throughout the system.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso