[Rodrigo_Eng_Mect]

Hi friends , I am simulating a rotor and centrifugal pump diffuser, I have the experimental values, however, the power value of the shaft is not coherent. The equation I use is TORQUE_Z () @ BLADE + TORQUE_Z () @ HUB) * 6 * 183 [rad s ^ -1], where 6 is the number of blades and 183 is the angular velocity. The result is much lower than the experimental value, am I having any errors in the equation? I have already done simulations with 3 models of K-e turbulence, SST and RSM without success. Do I have any term to add with roughness of blades?

[turbo]

See below my old post related with your question : find how many are missing in your CFD to evaluate torque comparable to test values.


Let me clarify "Pump Efficiency in CFD" for you (and also for many people who have misunderstood it).

Below are what you defined,
"head
(massFlowAve(Total Pressure in Stn Frame )@R1 Outlet-massFlowAve(Total Pressure in Stn Frame )@R1 Inlet)/(ave(Density )@R1 Outlet*g)

shaft power
abs(sum(Velocity in Stn Frame u * Force X + Velocity in Stn Frame v * Force Y + Velocity in Stn Frame w * Force Z)@Calc Wall Region) * numComponents"

The bottom part, what you named "shaft power", is however not the real shaft power because your CFD domain excludes the followings :
- Disk friction on the front and back cavities of the impeller
- Leakage flow power loss through the front and rear seals
- Any power loss due to recirculation flows returning back to the impeller
- Bearing windage power losses

Therefore, any well-designed pump impeller will show over 90% efficiency from CFD if all of them are excluded in the analysis, and if the bottom is a pure calculation of (Torque x omega) from CFX-Post. No surprises ! The CFD pump efficiency cannot be compared to test efficiency unless all of external parasitic losses are included in CFD (but so much rare cases).

In other words, what CFX-Post displays "Pump hydraulic efficiency = (rho * g * Q * dH) / (Torque x omega)" is not a true one, only confusing users. Be careful.

[bparrelli]

A few questions:

1. How far off percentage-wise is your CFD torque from the experimental torque?
2. How is the torque being measured in experiment? Are you using a dynamometer?
3. Does your CFD model include a tip clearance?
4. Your torque equation looks basically correct to me...I'm sure it's a typo in your post, but did you include the blade and hub torques in the parentheses? I use: (torque_z@rotor + torque_z@hub) * (# of blades) * (RPM)
5. Can you also provide a picture of the impeller so I can see what is included in your model?

Adding onto turbo's post, there will definitely be other losses in the real hardware that the CFD is not capturing. However, losses through seals and due to windage and disk friction can be accounted for separately to correct the CFD efficiency. What I will typically do is bookkeep these losses separately using 1D relationships and then adjust the CFD efficiency accordingly.