The intent is to specify a pump with a stable curve, not a pump with an unstable curve. A stable head / capacity curve is defined as one where head rises continuously as flow decreases and reaches its highest level at shut off. This type of curve is typical of most of centrifugal pumps. There are times, however, when a pump design may push the efficiency envelope and end up with an unstable curve
Although pumps with unstable curves can work well in many applications, they are not suited for operation in parallel. Why? Pretty simple, if the primary pump is operating at a head that is higher than shut off, the secondary pump may not be able to produce enough head to come on line. This is especially true for larger pumps that may be started against a closed valve. Several agencies recommend that all pumps should have stable H/Q curves. API goes even further and mandates them for parallel operation. They also recommend a minimum head rise from rated capacity to shut off of 10 – 20%.
Stable, or continuously rising, curve is a different animal than head rise to shutoff. Some curves will still meet 10% head rise with an unstable curve.
Centrifugal pumps are controlled based on discharge pressure. If your head rise is too low, a very small change in pressure can mean a very large change in flowrate. Even more important for pumps in parallel operation, due to allowable test tolerances, one pump could always be operating at a higher flow than the other. Or flow can surge as one pump overtakes the other, and vice versa. These effects are minimized with a steeper (more head rise to shutoff) curve.
When you have a problem (vibration, premature wear) you will ask the pump vendor for assistance. They will ask for operating conditions. You will provide pressure data and it will not be easy to determine the flowrate. Even if you have flow data (not as common as having pressure) the accuracy of the flowmeter may be called into question. 9 times out of 10, you will have inadvertently operated the pumps outside of the allowable operating region and repair may not be covered under warranty.
Having said that, an unstable curve (where a given pressure can have two potential flowrates) makes it impossible to determine the flow from pressure data alone even for a single pump. With two or more pumps in parallel, forget it.
A 25 to 30% rise to shutoff is pretty common, of which 30% gives an average control sensitivity of 1% of BEP flow per each Percent change in BEP head.
% of BEP
Head % BEP
Rise Flow
to Change
ShutOff
5 8%
10 4%
15 3%
20 2% 30 1%
A pump with BEP Q = 1000 gpm,
and a 250 psig BEP discharge pressure,
30% rise to shutoff = 325 psig,
has a variance of only 10 gpm per psi.
10% Rise to shutoff gives 40 gpm/psi
If it was a 5% rise pump and it goes off just a couple psi, you're looking at a flow variation of as much as 15%. If you're trying to hold BEP flow, maybe you'd be at only 85% on one pump and 115% on the other, then, if your wires arn't getting hot and your process still keeps going, power inefficiency starts creeping up the bill and pump maintenance can start becoming troublesome.
If it ain't broke, don't fix it. If it's not safe ... make it that way.
This is a mandatory requirements of API 610. Also if we have a flat curve as mention above, small changes on head will have a big delta on flow.On the other case, if we have a flat curve, possible eratic flow will be experience by the pump because of the possibility of having one head with two flow.
My experience has been that a great number of garden-variety single stage small centrifugal pumps in the 1 - 50 bhp range have flat characteristics without anything near 20% rise from BEP to shut-off. Putting them in parallel makes this situation even worse.
I guess I have seen a lot of "bad" pumps in my time.
"I guess I have seen a lot of "bad" pumps in my time."
Yes, there are some "bad" pumps round but not as many as "bad" engineers who can't apply a basic pump to a simple operation.
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.)
On the other hand, there are places where a flat curve can be useful - shower systems, for example, where a varying number of nozzles may be on, but you'd like the pressure to be relatively stable over a range of flow rates.
SNORGY,
Nothing implied with my response - just an observation after years of trouble shooting "bad" pumps.
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.)
