• Welcome to engineeringclicks.com
• # Pumps Curves?

Discussion in 'The main mechanical design forum' started by Archimedes, Apr 15, 2012.

1. ### ArchimedesActive Member

Joined:
Apr 2012
Posts:
25
0
Hello,

I am doing a project for which I need to select a pump. It is a closed circuit application. Does anybody know where I could learn how to read pump curves? I thought I knew how to do this, but I had a guesstimate for a pump, installed it then measured the flow rate (not enough), calculated the resistive head of the circuit re selected a pump using the pump curves, once I had installed the new pump the flow rate was nothing like what was expected (still not enough), not sure how many more pumps my boss will let me buy.

Any guidance with pumps most welcome, no pump manufacturers seem to give help with this.

Thanks

2.
3. ### VirguleActive Member

Joined:
Nov 2011
Posts:
31
0
The pump curves are really easy to read.

On the Y axis, you have the total head in feet or meters which is the resistance of the circuit to flow. This can be calculated using the Darcy-Weisbach equation(s).
On the X axis, you have the flow rate in gallons per minute (gpm), Liters per minute, ....

For a given pump (given curve), the more total head the pump faces, the less the flow rate will be. Indeed, if you choose a point on that curve and look for a point with a higher total head on the same curve, you will look to the left, at a lesser flow rate.

You usually see a bunch of parallel curves. These are pumps that usually differ in the diameter of the impeller, but have a similar construction. Indeed, for similar pumps, a bigger impeller usually means the capacity to overcome more total head or providing more flow for the same head.

Also, there are "C" shaped curves that intersect these parallel pump curves. These are efficiency curves. If you travel along a pump cuirve from 0 flow rate onwards, you cross the efficiency curves and progressively enter zones of better efficiency. You then arrive at a best efficiency point or BEP. This is the point where you're at the best efficiency, or enclosed in the most compact efficiency curve.

Finally, at the bottom of the graph, there usually is another single curve going up. This is the NPSH curve. At a given flow rate, the pump has a required NPSH for its use. The NPSH required by the pump must be lesser than your circuit's NPSH (available). NPSHa is the ability of the circuit to help feed the pump its liquid. For instance, a circuit where a feed tank is higher than the pump will have a bigger NPSHa than one where the pump pumps in a deep pit. Some pumps have a lesser NPSHr than others, meaning that they are better at "sucking".

The goal of pump selection is to choose a pump that functions at its BEP and that is fit for the NPSHa (of the circuit).

To calculate the NPSHa and to learn more about pump curves, try a quick google search.

4. ### chaos_zzyNew Member

Joined:
Apr 2012
Posts:
1
0

One suggestion to estimate or test the system flow resistance curve, which is helpful for you to select pump and determine the real flow.

The system flow resistance is a quadratic equation (Delta P vs Flow rate).

Joined:
Nov 2011
Posts:
78
1
6. ### msmaxwellNew Member

Joined:
Jun 2011
Posts:
1
0
That Grundfos manual is fantastic.

7. ### ArchimedesActive Member

Joined:
Apr 2012
Posts:
25
0
Thanks for everyone’s help with this. Reading everyone’s posts here it sounds like I did use the correct procedure to size the pump however I have tried this again and still the theory does not match reality. I have now installed a new pump that gives the required flow rate (sized by bigger guesstimate).

I am still going to carry on with the theory until I can get this to match the figures I am seeing in the lab.

Thank Grundfos manual is brilliant.

Joined:
Nov 2011
Posts:
78
1
I try

9. ### spiceageMember

Joined:
Apr 2010
Posts:
19
0
pump curves - learn from failure

Hi Archimedes
If you know the unsatisfactory flow rate that you achieved with the inadequate pump and you have the pump curves for the same unsatisfactory pump, provided there were no faults (e.g. are you certain your system was full and that there was no trapped air or vapour?), you now have achieved an important data point that you can use to calculate the true head requirements for your system (rather than your original calculated estimates.) If the head loss through your circuit is purely resistive, the solution is simple. However if part of the head loss is due to your needing to lift the liquid, e.g. from a borehole or to a header tank, then extra considerations are needed.

For the simple purely resistive case, examine the unsatisfactory flow rate that was achieved. Taking this value, the pump curve will show you the head gain through the pump. Letâ€™s say these flow rate and head gain values were Q1 and H1. If the flow rate that you really want is Q2, then the head rise that your re-specified pump needs to develop is

H2 = H1 *( Q2/Q1)^2

i.e. the head gain you need is proportional to the square of the flow ratio. In reality this is a bit of a simplification but unless you are pumping a viscous liquid slowly (laminar flow), it is likely to deliver the result you need. You are now on the right road - just find a pump whose curve passes through the point Q2, H2.

If part of the head requirement for the system is to do with a net lift of liquid, then you must subtract this from the value H1 above. In other words, we need to split the head gain H1 through the pump into two parts, Hg + Hf where Hg is the head from the pump which is needed to deal with gravity and Hf is the other part that is needed to deal with friction. The Hf term will change with the square of the flow ratio but the gravity term is unaffected by the flow rate. Thus the head rise that your re-specified pump needs to develop is

H2 = Hf *(Q2/Q1)^2 + Hg

and again you need to find a pump whose curve passes through Q2, H2.

The gradient of the pump curve at the point Q2, H2 is interesting. No system head requirement estimates are perfect and indeed there may be changing head requirements as a result of valves being turned on or off. The gradient illustrates the flow rate consequences of the system head loss being higher or lower than expected. If the pump curve is steep at Q2, H2, the flow rate will not change as much as if the curve were shallow. The bad news is that if Q2 is still too low, a steep characteristic means you will have to work extra hard on the frictional losses to change the flow rate.

A good source of information on pump characteristics and their integration into systems can be found in ESDU Data Item 80031.
Good luck
SpiceAge.

Joined:
Apr 2012
Posts:
25