parallel pump problems 1

[hitchhiker]

Dear fellow members,

I'd like to ask your expert opinion on my current situation. I'm currently handling left-over projects that have 5 booster pumps in paralell delivering liquids from storage tank (produced water) to a pig launcher. 3 of the booster pumps are old existing pumps and another 2 are fairly recently bought with matching H/Q characteristic with the existing (not really identical). These pumps fail frequently and they took turns (imagine that!). so operation guys came to my department (project) to look for solutions and hopefully create something to solve their predicament.

the first thing when I look up when I step in is the calculation by FEED contractor. The old pump itself basically reused from another project, rated at 600 gpm @ 735 ft, water temp 180 F. the newer 2 pumps also have the same rating, 600 gpm @ 735 ft to suit the old one. So before they actually bought the 2 new pumps, FEED contractor have studied the system and made recommendation for 5 pumps in parallel. So the combined pump curve only intersect the system curve at 4 & 5 pumps operating. At single, double and triple pump operation the system is well below the pump curve (way to the right). So I wonder does this possible?

by right, operating 1,2 or 3 pumps will have cavitation problem, since it will push operation way to the right curve (it doesnt even intersect the system curve!)and while the pumps do have problems and regulary in-out service for maintenance, there were no report of cavitation damage on impellers. there are erosion damage on the casing, however (it turns out there were sand sediment in the fluid that was not considered). there are even report on casing leak and damage. I still wonder if this is because of the sand or other things.

So back to the my question, can parallel pumps designed this way? that the system curve only intersect the combined curve of 4 & 5 pumps?

 

[hitchhiker]

Here is the pump-system curve

correction, only 1 & 2 pumps operating in parallel dont intersect with the system. The 5 booster pumps operates 4x25%, with 1 pump standby. And here is the P&ID

Any comments?

 

[LittleInch]

Yes, you can design this way and in rality, two pumps will probably work, but clealry out of their operating zone. Three looks ideal.

The issue I can see is that no one has thought too much about how to start this system or operate with less than three running. Given you appear to have a flow meter, there should be a control valve with set point based on flow driven by how many pumps are operational, hence set at 600 for one, 1200 for 2 and then fully open / 1900 once the third starts.

Multiple paralell pumps are common, but you need a good control system / operator to make them work effectively AND a means on controlling flow on start up and when you only have one or two available. My guess is that when one of the old ones went off line or was reqaired, they kept working with two and hence operated them at >>100% of rated flow and power and contributed to their condition.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Artisi]

Congratulations to the OP for supplying good data and a clear description of the problem, it's a shame more don't follow this example.
What else would help is the individual pump curves with NPSHr and the NPSHa for the system - this will allow for discussions as to how far the pumps can run out on their curves before you need to apply artificial head into the system so that the pump curves / system curves can intersect.
However, I would say that operating 1 or 2 pumps is a big ask and unlikely to be a reasonable possibility, far too much head to dissipate to get them onto their curves.

As pointed out by LittleInch, 3 pumps appears reasonable -- but NPSHr for the pumps would be worth having for checking.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[hitchhiker]

Thanks for the reply guys.

@littleinch, when I first learned about the calculation I straightaway thought that this was really recipe for disaster, having a curve like that, and nobody notice that unfortunately. But since u said it's possible, and after hearing report that there were no cavitation damage on impeller perhaps I have to reconsider my stance. the system did work though with a lot of problems. Yes there is a control valve at the thick line, and there is also restriction orifice after control valve, just to drop the pressure upstream pig launcher. the pig max inlet is about 200 psig. the pump is well oversized for the system (it's a left over, just for the sake of using it so they dont need to buy new pump long time ago), reality on the field so we made adjustment on the system to make the pump works. Control valve is dictated by level switch on the water tank upstream of the pump, not by flowmeter (no flowmeter).

Beside the casing damage, pumps also experienced severe vibration problems (which i suspect that would happen if only 1 or 2 pumps running). but with such system, u have to run 1 pump first then 2nd come online, then the 3rd etc, right? when 1 and 2 running, can it sustained well runout condition? or u can (or have to) actually make simultaneous 3 pumps online with this kind of system? My guests so far is that operation actually throttled the gate valve upstream & downstream control valve to boost system resistance, but this I havent got any feedback from them.

I will try to post the casing pics showing the suspected sand erosion damage. It's quite perplexing to hear that casing damaged to the point it got holes (from maintenance team), while impeller is fine.

 

[LittleInch]

One thing we forgot was that when you get multiple pumps like this, and as you said there is no such thing as identical - you can get some inequality and the more the pumps the bigger it could be, hence at 1800 gpm, one pump could be doing 800 with the others 500 each if the head is only a few percent off. You need to check the motor amps of each of them to see, but a difference of anything less than 10% is probably the best you can do. Four pumps will even flow out a lot more and should be your default mode based on your info.

I can see that there is no flow control - there is no control valve shown on the PO & ID inless I'm much mistaken (looks like an on off isolation valve to me)- my point is that there should be but it is often overlooked and will result in your pumps not starting properly. Is that a re-cycle line on this system? Is that supposed to be used at start up to allow three pumps to come on line before you open the main valve??

If you start one pump with no more resistance than the mian line I can't see it doing anything other than tripping on excess amps if not vibration..

congrats on a decent post with most of the info - doesn't happen too often....

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[BigInch]

What do you mean that one and two pumps do not intersect the system curve. Extend the curves! One pump will run at about 925 gpm. Two pumps will run at 1550 gpm

The system curve does not appear to be plotted correctly. System curves are plotted in absolute head, but you appear to have plotted it on the pump curve, ie. pump curves are plotted using differential head, or there is no decrease in suction pressure as flow increases. Pump curves must be superimposed on system head diagrams, but it seems that you show no suction head decrease as additional pumps are turned on, as all pumps have the same absolute +/-925 discharge head. Are you holding suction pressure constant in some manner even though you are running from 1 to 4 pumps? Normally suction pressure decreases when flows increase by 4X.

Independent events are seldomly independent.

 

[hitchhiker]

BigInch,

I don't really get it. System curve plotted using absolute head? Normally we only use absolute head in calculating NPSHa. And I thought suction pressure should increase while flow increase? suction pressure increase, discharge decrease ---> flow increase? please do correct me if i'm wrong.

LittleInch,

Yes, pump is not identical, I've replot the pump curve and superimpose it below:

Below is enhanced section of previous snap-shot on the discharge line. It has control valve (pressure), and an orifice. And there is a recycle line back to the main suction tank. I was told that operation normally closed this line (on normal operation)

 

[hitchhiker]

Sorry I forgot to explain on the pump curve, the red dot line is the old existing pump and the solid black line is the new pump. and btw here are the foto of the current casing condition with suspected sand erosion.

let say when the casing di deteriorated as above, does that affect the overall pump performance, perhaps created a "new" performance curve?

 

[LittleInch]

~As ever new information makes a lot of difference.

First though I would like to take issue with a couple of BI comments - I have always looked on these graphs as being needed to be plotted from the same elevation point - normally pump discharge. Thus any static head is the start point of the system curve at 0 flow, with the pump curve based on pump differential plus whatever it's inlet head is. From your graph and description (pumping from storage tanks) there is probably only a low inlet head, but in other situations the inlet head will normally fall as more flow goes through the pumps hence greater friction on the inlet line leading to lower inlet head, and therefore lower outlet head as you add more identical pumps and increase flow..

The only common thing your pumps will see is the outlet header pressure. The difference between the pumps is quite a bit and the real issue is if the back pressure increases above 850 m differential head such that one of your new pumps will have very little or no flow and the older ones end up with all of it.

Still can't work out how your sytem is supposed to work or start, but I think you really need to start two pumps at the same time as a minimum as one pump, whilst it could end doing 900+ GPM is clearly well off the end of the curve and should not be operated like that. Given that you have a control valve you should be able to create a start mode where you fix the position of the valve that equates to 600gpm for one pump before starting another one and then release the control back to the pressure control.

Actually, giving this a bit more thought, this looks like a back presusre valve to me. Hence your system curve may be incorrect and in fact is a horizontal line at whatever the pressure set point is until it meets your system curve when it will then be fully open. This might already then allow you to start one pump if the pressure set point is above the equivalent head of about 625 feet. Find out what that set point is and convert it to feet head then draw a horizontal line on your system curve and it will also tell you a lot about how your system works

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[BigInch]

Take it off the test bench. Is discharge pressure held constant. If discharge pressure is held constant, then yes, the discharge head is constant in differential and absolute terms, if not, then no. Is suction pressure held constant? If not, then the same issue applies. Remember Bernoulli. Z + v2/2g + P/[γ]. If neither suction pressure is held constant, nor discharge pressure, then it's a free running pump and absolute discharge head varies, so you don't know where the static head and HGL are in relation to the system. You must first know how the system will operate before you superimpose pump and system curves. If not, you assume Z is 0 for all points in the system.

Where a pump should be operated is in relation to BEP and Power Rating. A pump sized for 600 could be operated off the end of the curve, and it will overdraw power in overspeed conditions, if not controlled to do otherwise, it will operate at the system intersection point, wherever it is, including if it is off the end of the curve. The pump doesn't know anything about end of curve limits.

Independent events are seldomly independent.

 

[hitchhiker]

Well, I've just got news from operation that they just got mechanical seal failure on one of the pumps. Come to think of it, is it possible that the pumps experience discharge cavitation? As I quote from one website:

"Discharge cavitation occurs when the pump discharge pressure is extremely high due to output restriction. A restricted discharge that reduces the pump efficiency to a low value (10 – 20%) forces the majority of the fluid to circulate inside the pump instead of being allowed to flow out the discharge. As the liquid flows around the impeller the increased fluid velocity and heat buildup within the pump housing causes a vacuum to develop at the housing wall that turns the liquid into a vapor. A pump that experiences discharge cavitation for a period of time shows premature wear of the impeller vane tips and the pump housing and can result in the pump shaft failure."

With the picture of the casing that i supplied, what do you think? I don't have impeller failure (at least not reported) but pump casing failure has already happened. If this is the case, is it possible that the failed pump goes into deadhead operation? Would that make sense? what do you think?

 

[BigInch]

Oh boy. "discharge cavitation with a vacuum turning liquid into a vapor", is it?? This is recirculation and failure to maintain minimum flow-through, due to pump not being to overcome static head (including valves) at minimum flow rates, heating of liquid to above the water's vapor pressure, causing boiling off of the water into water vapor. No liquid cooling causes overheating, your seals dry out and burn up, if not blow out due to higher and higher pressures as the vapor continues to heat into steam. Your already high suction temperature isn't helping. Do you know what the vapor pressure is of water at 180F? I doubt you meet NPSHR.

Independent events are seldomly independent.

 

[LittleInch]

Well I was a bit worried when you said that the re-circulaiton line, which is designed to prevent v low or no flow was
"I was told that operation normally closed this line (on normal operation)". Operators often have little or no comprehension about what pumps can and can't do or how to operate them efficiently and just develop certain ways of working which they find works, but can lead to problems in operation. These are quite decent sized units and if you dead head them, all that energy has to go somewhere - water can heat up very quickly and cause all sorts of damage.

As I said earlier, the difference between your pump curves becomes crucial as the pressure climbs / flow decreases as the "new" pumps can be severly affected even if the old ones are still flowing.

Big bro' is right though , you won't get a vacuum at the pump discharge, but all sorts of bad things start to go on when you dead head pumps.

My motto: Learn something new every day

Also: There's usually a good reason why everyone does it that way

 

[Artisi]

As stated way back, individual pump curves showing full pump peformance would help in answering the initial question " At single, double and triple pump operation the system is well below the pump curve (way to the right). So I wonder does this possible? " "So back to the my question, can parallel pumps designed this way? that the system curve only intersect the combined curve of 4 & 5 pumps?"


The photo of the pump casing, appears to be typical corrosion erosion, the result of pumping and maybe re-circulating sand round and round within the casing. The main consideration would be casing wear ring / impleller wear ring clearance to maintain performance..




It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[hitchhiker]

Artisi

what are the chances that the pump experience discharge recirculation? can it have the same erosion damage on the casing even if there were no sand content in the liquid? plus we just got seal failure too..

 

[Artisi]

if the pumps are throttled and operating left of BEP, then there is recirculation taking place and the further left you go the worse it gets, you will also very poor inlet conditions as well, inlet flow will not be entering the impeller eye correctly and not flowing onto the blades as designed. Both will result in noise, vibration, shaft deflection, and probably unbalanced axial loads etc etc.
I notice in your photo the inlet has a vortex breaker built in, if some pumps don't have this you could also be suffering pre-rotation of the inlet flow which will only worsen the problems.
As these are high head pumps they should be running at BEP +- maybe 10 - 15%, 20% should be the max.
Yes, if suffering from discharge recirc.you will / could have erosion damage.

It is a capital mistake to theorise before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts. (Sherlock Holmes - A Scandal in Bohemia.)

 

[hitchhiker]

Let say I have a minimum flow problem, how does a recirculation line back to the suction tank would help? because the suction tank itself have continuous incoming stream of fluid and adding back small portion from recirculation line should not make any different anyway, unless there are no more flow into the tank?

 

[clay87]

A dedicated miniflow line on each pump leg ensures that, if one of the pumps is being dominated by the other(s) with a stronger pump curve, the weak pump maintains its minimum flow. The diagram you posted is too blurry for me to see a miniflow line.

I've only quickly browsed this thread but I haven't seen any pressure readings posted, only the system curve at beginning. If the actual discharge pressure (does it vary?) is a little higher than what is shown on the curve, then the new pumps may need an auto recirc valve in 4 or 5 pump operation (e.g.

http://www.agivalves.com/prod01.htm).

Alternatively, you could put a restriction at each pump discharge to reduce strong/weak pump problems (for non-permanent situation try partially closing discharge valves).

 

[ccfowler]

hitchhiker,

As others have commented, you are to be complimented for providing substantial information from the outset!!!

The only thing certain about any given number of identical pumps is that no two will every be truly identical. With a mix of older and newer pumps with probably not well documented actual operating history (and likely unintentional operating abuse), their individual actual operating characteristics will likely vary much more than one may casually presume.

Presuming that your diagram fairly well represents the suction piping system configuration, the suction (OK, inlet, if one cares to be fussy) conditions at the several pumps will not all be equally influenced by variations in the total flow rate. This may or may not be an issue of concern for some of the pumps.

Having dealt with some astoundly troublesome systems of mis-matched pumps operating in parallel, the most useful characteristics that I found worth primary attention are the suction and discharge temperatures at each pump. The discharge temperature of a pump will be greater when the pump is doing proportionately more spinning than pumping thereby serving to become more of a mechanical water heater than a pump. If the suction temperature is greater at a pump, that is a good indication of troublesome recirculation. The greater the several temperatures vary from their counterparts at other pumps, the more attention their associated pump will deserve in your studies.

Because the temperature rise across a good-running pump is normally very small, the calibration of the instruments is important. Given the apparent problems of the pumps in your system, it would be no surprise to find some of the temperature problems to be quite obvious. Where I am trying to monitor good-running pumps, I prefer using thermocouples connected to indicate the temperature differential across the pump since that avoids the problem of calibration differentials between two separate instrument systems measuring the two temperatures.

Valuable advice from a professor many years ago: First, design for graceful failure. Everything we build will eventually fail, so we must strive to avoid injuries or secondary damage when that failure occurs. Only then can practicality and economics be properly considered.