pump and system curves
pump and system curves
(OP)
the performance of our chilled water pumps is attached.
• There are four pumps on duty (the pump selection and duty point performance is on page 1 of the attachment).
• We obtained readings of each pump with all the four pumps running. (Curve is on page 2 of the attachment).
• We also took the readings of each pump without operating the other pumps (Curve is on page 3 of the attachment).
• The pressure measured on the common discharge header with all 4 pumps running is 75psi.
• Are the attached pump curves drawn correctly? If yes, the pumps are not performing as they should both when run one at a time or when all 4 are run in parallel, and how does this compare with system head





RE: pump and system curves
How did you vary the flow rate for one pump and for all pumps to get the data?
see also the other recent posts on similar subjects which might help, e.g. http://www.eng-tips.com/viewthread.cfm?qid=343897
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
Looking a bit further at the figures on table 4, flow could well vary between individual pumps but discharge pressure has to the same on each pump if pumping into a common pipeline - that is unless there is any major difference in pipe / connection congfiguration between the pump discharge and the connection into the common pipework.
Without running any check on the performance figures you supplied, the pumps could well be operating within the pump performance test code in terms of 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.)
RE: pump and system curves
I'd still like to see the set-up to see what is going on, but artisi is right - what you really need are the actual test curves for each unit, but testing conditions in a factory are different to reality, e.g. have you calibrated your presusre transmitters / guages - test ones are cerified guages, ditto the flow measurement devices. A few percent out on each one and you're some way to making up the difference.
What temperature is the water in the pumps - is this before or after the chillers? Doesn't make a lot of difference but if temp is above 40 C can make a couple of percent difference in density.
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
The pressure gauges of all pumps are giving nearly the same readings, and the sonic flow meter readings tally as explained above.
The temperature of the water going in and out of the pump is 15 degrees celcius.
(Reply to Artisi):
Could you please review the accuracy of the curve (by running a check on the supplied performance figures). If the curves are drawn correctly from the pressure and flow readings, it would then indicate an over-trimmed impeller as being the reason of deviation from the pump characteristic curve.
RE: pump and system curves
I know this is basic, but are the pumps going round in the right direction? Who checked?
Do you have actual test results for each pump ?
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
There is no fitting between the pump discharge and the pressure gauges.
The voltages are the average of the three phases and it is a pretty stable power supply. The readings were taken from inside the Control Panel adjacent to the pumps.
The pump curve (Page 1 of the attachment) is the factory supplied curve.
The hand drawn curve superimposed on this curve was plotted from the field data of pressures and flows.
RE: pump and system curves
When running your pumps individually, Pumps 3, 2 and 1 have a progressively greater suction pressure drop (relative to pump 1) at low flows. That indicates an imballanced arrangement for suction piping. You should feed into the center of the suction header, but it looks like you are feeding all pumps from the far end of the suction line and each pump draws a lower suction pressure according to their position on the header and as head losses increase down the header. BAD.
Somehow they tend to equal at around 800 gpm. There seems to be some control action going on.
Your pumps running individually all have a discharge pressure of around 90 psig at a flowrate of around 800 gpm and yet you have about the same discharge pressures when running all four pumps together. You have not explained how your system curve stays at 75 psig at the common point on the header even though you supposedly have 4 times as much flow running through it. Some unexplained control action is going on, or you have 4 pumps each feeding through separate closed loops, or maybe you are running first one chiller, than 2 chillers, then 3 chillers, then 4 chillers.
Explain your system and include a proper system diagram before you go asking any more questions. I've wasted enough of my time with this guessing game, as have other kind people above.
Independent events are seldomly independent.
RE: pump and system curves
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.)
RE: pump and system curves
RE: pump and system curves
Have all of the pumps seen similar amounts of usage? Are some relatively new while others have been rebuilt? Are all of the motors nominally original, or have some been re-wound?
How clean is the piping system internally? Is it possible that there could be some "debris piles" at some places in the piping system?
Obviously, the list of such possible issues can be nearly endless without more direct knowledge of the specific system.
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.
RE: pump and system curves
RE: pump and system curves
All data should be posted like this, so we don't have to do it for you.
Sometimes we might need to do a few basic calculations, at least to check your work, before we waste time with details.
your dishcarge head into the system appears more or less consistant.
That's why I said it may be somthing with the suction, which varies a lot, esp at low flows.
Your system curve is constant even though flow is probably nearly 4 times more. We need to know why. You can't do anything with pump system design without knowing what is going on with the rest of the system. Pumps are only 1/2 the story. The other half, piping and controls as applicable, are as essential as the pump curves.
Independent events are seldomly independent.
RE: pump and system curves
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
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.)
RE: pump and system curves
I think the mistake he's making is not adding suction head to get the system discharge curve. The pump curve needs to be superimposed on the system curve. He's doing the reverse and wondering why his discharge head is low. But I refuse to play guessing games. Sit tight.
Independent events are seldomly independent.
RE: pump and system curves
The spreadsheet and the curves are super.
The sketch will soon be underway.
More later. Many thanks
RE: pump and system curves
I am curious to see what else you've got for us.
Independent events are seldomly independent.
RE: pump and system curves
Here it is finally. The attachment gives the general arrangement of pumps.
Waiting anxiously for feedback.
What intrigues me is that the regardless of installation deficiencies, are'nt the pumps constrained to operate according to their curves. The data suggests they are not following their curves, which I thought was either an incorrectly drawn curve of an over trimmed impellor.
Thanks
RE: pump and system curves
For info can you explain how the double valve works.
At the end though are you getting less flow than you should? Your pump discharge is 90 psi, whilst discharge header is 75psi so I guess the control valves are kicking in on some sort of signal. Do you actually have a problem or are you just curious about the readings?
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
1. I have attached another sketch that depicts the location of the pressure gauges.
2. The Double regulating valve is simply a balancing valve to set the flow somewhat like the globe valve that throttles the flow. However on this balancing valve, there is a setting that can be locked in place and recorded. From the pressure drop through the valve (picked up form the pressure sensing ports upstream and downstream on the valve body, the flow rate can be read from the Cv charts supplied by the valve manufacturers.
3. The pump is not riding its characteristic curve, particularly at its duty point as can be seen in the hand drawn curve that is superimposed on the pump selection curve. This means that with the fouling of pipes, the pump has no room to be able to handle the increased head requirements of the system with ageing and scaling.
4. By the way the 75 psi that was given as the pressure on the main header was an incorrect reading. The reading on the common header is also 95 psi, so there is nothing enigmatic in that part of the network. Sorry about that error.
5. What is enigmatic is that why are the pumps not following their curves. If the system head is more than anticipated it should simply give less flow, but still remain on its curve. So I just want to rule out the possibility of an over trimmed impeller.
Looking forward to your feedback and of any other interested readers in the forum.
RE: pump and system curves
1. are the pumps from a reputable manufacturer or are they el-cheapo copies from an unknown source?
2. were they factory tested for performance?
3. have the pressure readings been correct back to the pump centreline, have all velocity heads, height differences etc etc been accounted for in your H/Q curve generation?
comment:
if the impellers were over-trimmed then the CV head wouldn't be exceeded which you are currently reporting, if under trimmed, the H/Q curve would be down by an even amount from CV to end of curve.
Statement:
From the shape of the curve (assuming the first part of the question 1 is in the affirmative) I doubt very much it is a pump problem, appears more likely a system problem or (my belief) a measurements problem.
Recommendation:
Have 1 pump hydraulic performance tested in a certified test-rig, the amount of money spent so far and which looks to be ongoing probably exceeds the cost of a test.
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.)
RE: pump and system curves
First, this is NOT a simple arrangement of several nominally identical pumps operating in parallel. That is prevented by the individual "DRV" associated with each pump. Each pump will always operate on its unique characteristic curve, and it will operate at the intersection of its characteristic curve with the piping system's characteristic curve. The individual DRV's serve as independent variables that alter the piping system characteristic curve seen by the individual pumps.
Based on previous experiences, I suspect that piping elbows (and other flow disrupting items) are likely to be uncomfortably close to the pump suctions so that the pumps may not be seeing the well-developed suction flow upon which the published curves are based. It is also likely that the pressure taps may not be seeing the well-developed flow needed for their instruments to operate at their greatest accuracy.
The calibration accuracy of the instruments used must also be taken into account and properly adjusted into the performance evaluations.
It seems likely that the performance curves that you reference are actually just the curves that are published for the generic pump model, and as such, variations of as much as 10% can be anticipated for individual pumps due to manufacturing tolerances and variables. Your information shows a nominal slip speed for an induction motor, but is this the actual shaft speed upon which the characteristic curves are based? It is not unusual for the published curves to be based on a pump being driven at exactly synchronous speed, and it may be necessary to adjust the nominal pump curves to the actual operating speed provided by the motor under the varying head and flow conditions.
Time and again I have seen that what is claimed to be a pump problem will be found to actually be a system problem compounded by an incomplete understanding of the true meaning of the pump's performance curves. The pump (or pumps) provide a seemingly obvious focal point for a misbehaving system, but in reality, the pump(s) almost always will be found to actually be performing according to properly adjusted expectations.
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.
RE: pump and system curves
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.)
RE: pump and system curves
For some reason you get a better pressure loss at higher flows. That might be due to proportionally less resistance vs flowrate as you sum the effects of the number of chillers and loops you're turning on. In other words, at low increasing flows your distribution efficiency becomes better, lesser pressure loss/gpm being distributed each time you open a loop in the flow circuit. Nothing particularly unusual about that. For example, twice as much flow area gives you twice, or more than, the previous flowrate at the same, or slightly less, pressure drop, until you get to 750 gpm in each loop, where any further advantage in increasing flow area, without increasing diameter, is lost. That's the velocity where your pipe diameter is being used to their "economic" capacity.
Independent events are seldomly independent.
RE: pump and system curves
I guess down the road when the deposits in the piping increase the friction, at that time we could explore upgrading the impeller.
Thanks to all the respondents.
RE: pump and system curves
Independent events are seldomly independent.
RE: pump and system curves
I termed it verdict because I sincerely felt it was a learned opinion from someone who has a solid theoretical and field experience.
Thanks again.
RE: pump and system curves
If it's one vote, it's an opinion. Let's see how the rest of the jury votes.
Independent events are seldomly independent.
RE: pump and system curves
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.
RE: pump and system curves
RE: pump and system curves
It's definetly been quite interesting...
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
LittleInch: Your feedback after double checking the calcs is keenly awaited. Thanks again for your wholesome and constant support from the beginning of this problem till its conclusion.
RE: pump and system curves
If you have what is the SG at your pump temp?
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
THANKS
RE: pump and system curves
Assuming the data is correct, and there is a lot of it that all is similar, shows that the hydraulic power required, even based on 79% efficiency at 800 us gpm, is higher than the electrical power in ( v x amps x sqrt 3). so unless you've discovered perpetual motion something is not right here.
I find it difficult to believe that the control valve has no impact on pressure even if fully open.
However your system seems to work even if less than you planned, so take some credit for that.
What can you do now? Use some calibrated guages / transmitters to get some better data, have longer straight lengths before the suction flange (5D min), flow data via a CV is not very accurate (10% at best), check my power figures or take a pump out of line and test it on the bench. Ask the vendor for info and double check that your data pack from the vendor shows you have the impellor that you think you have.
The people posting here on this are top drawer so read them all carefully before deciding what, if anything, you do next.
Good luck and thanks for all the information.
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
LittleInch: I am impressed. You even figured out the brand! Thanks for the bundle of useful pieces of information.
Thanks to all who favored me with valuable feedback.
RE: pump and system curves
RE: pump and system curves
The design should have allowed for a collection sump with room for a couple of electric sump pumps setup to handle the inevitable problem.
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.)
RE: pump and system curves
My motto: Learn something new every day
Also: There's usually a good reason why everyone does it that way
RE: pump and system curves
RE: pump and system curves
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.)
RE: pump and system curves
Artisi:
Your proposal makes sense. Actually the problem of flooding from any stray leaks inside the pump room is compounded because connecting the overflow of decorative fountain's balancing tank to the nearest storm or sanitary drainage manhole was missed by the Contractor. Since the earthwork is done, he wishes to core into the balancing tank wall that is common to the wall of the pump room, and route the overflow into a sump pit in the pump room, from where it will be force discharged into the backwash line. I am reluctant about this proposal, because the submersible in the pump room will only cut in in the unlikely event of leaks, but allowing the overflow to come to the pump room is like inviting trouble that can be avoided by direct drain of the balancing tank. Am I being overcautious?
RE: pump and system curves
Independent events are seldomly independent.
RE: pump and system curves
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.)