Pump unphysical efficiency
 

Hello,

I compute a low specific speed pump with splitter blades. Inlet + Full impeller + Volute. No casing.

Periodically, I have an hydraulic efficiency > 1, it happens when the main blade (sometimes the splitter also) passes in front of the cutwater. The average efficiency is somehow realistic (~0.77)

I changed the mesh - the turbulence model - the boundary conditions, but I always have the same results (exact same amplitude). So I believe there is 3 ways out of this :

1 - My CFD is still wrong
2 - CFD can't model this case (I'm really not experienced enough to pretend this, just putting it here)
3 - It's physical, and if I try to explain it, here I go :

Citing Gulich, chap 10.1 :
In my case, this whole pressure minimum is present on the whole channel (see image, in this image, the efficiency is > 1). So locally (at this time), this low pressure on the pressure side causes a low torque > a low shaft power which is locally lower than water power > an efficiency higher than 1.

► Note on the impeller: It's a low specific speed impeller (nq = 9, high diameter, lot of blades), but the outlet width is way higher than the "recommended" one. b2 recommended is 4mm, I have 10mm. So basically, even at the designed point, my impeller behaves like a "part-load" classic impeller.

Any comments is appreciated because I am completely stuck right now :D

If the hydraulic efficiency > 1 for just a tiny fraction of the rotation, is that significant? I suspect not. You say the average efficient is realistic at 0.77, so isn't that the important factor?

I agree with you (because I hope it's ok ^^). But if the prediction is unphysical "some times", how to trust the results the rest of the time?

I am wondering if something like that is measurable on a tests-rig, but I didn't see it on any papers obviously.

The picture I showed is a 5+5 bladed impeller. For a 4+4, the "shock" was more violent, and the efficiency was up to 1.5 .

It is difficult to judge your case, but I used to perform calculation on fish-friendly pumps with only 1 blade. There I see this happening all the time because at a certain moment in time the torque is low, while the flow is high, resulting in efficiencies above 1. On average, the efficiency matches experimental values very well.

@Gert-Jan, nice! As you say, it doesn't mean it is also true with my case, but if someone experienced it before with good comparison with experiment it's already really nice and gives me hope :)))). Thanks for your reply.

For info, what was your CFD efficiency peak? 1.1? 1.5?

On the experiments, you saw that efficiency peak also? Or you measured just the average?

The maximum is about 110%:
I get time averages back from the pump manufacturer. And the values are within 1% (absolute)

Thanks!

I did not understand if the value you sent were from the manufacturer, or are your CFD values?

I only received average numbers from the manufacturer and that is clearly not what you see in the graph......