I am using cfx to design a centrifugal pump. It is a steady case with only the rotor. But the computational resutls are far from the experimental results. For example, the experimental best efficiency point is 200gpm, but the calculated one is arount 400gpm. The efficiency values are also different. I also tried different grid resolutions and they did not help. The finest grid has 700000 elements. Anyone got any idea?

Does your geometry fully represent the pump and any geometry that would influence the flowfield (i.e. leakage, passes,etc)? I have found that sometimes it is not always what you are modeling, but what you are not...

Good luck,

Erich

- What are your Boundary conditions?

- Did you run Frozen rotor? Why not use a transient calculation with rotor-stator-interactions?

- 7000000 elements. What kind of elements? Hex?

- Is it turbulent? What model did you use?

Gert-Jan

www.bunova.nl

I currently have a centrifical pump/motor combo in a water pumping application. No head pressure as the distribution is level with the pumps outlet. Current unit is 230 vac 3ph, 3600 rpm, 30hp. I am replacing with 230 vac 3ph, 3600 rpm, but a 20hp. What will be my approx volume loss?

I am a new user to CFX. i want to know how to solve turbomachines problems for incompressible and compressible fluid.

any refence literature or paper if you have please send me.

problem of Centrifuga pump i think is the best problem to understand. Please do guide me.

thanking you,

Try cfd-online WIKI.

- What are your Boundary conditions?

I use mass flow inlet and pressure outlet.

- Did you run Frozen rotor? Why not use a transient calculation with rotor-stator-interactions?

I did not run Frozen rotor. I just run the simplest models without the stator for the first step. I could understand there should be some differences, but this result is not what I was expecting. I just tried the stage model for a steady calculation with rotor and stator and the results do not change much.

- 7000000 elements. What kind of elements? Hex?

700,000 elements is for the finest grid. And I use tetrahedrons.

- Is it turbulent? What model did you use? yes, k-e model.

Thank you for the suggestion. But will the leakage make such a big difference?

Do you have prisms on the wall? If you only use tets, your velocity near the wall will not be resolved sufficiently.

I do not understand your setup completely. Do freeze everthing and just calculate the pressuredrop over your geometry with a velocity set on the inlet?

Gert-Jan

I used prisms on the wall and y+ is less than 100.

I make the rotor as a rotating part in the setup. Since there is no stator, I guess no frame change model (frozen rotor, stage...) is needed. A velocity is set on the inlet.

centrifugal pump calculaton i need