[jgross]

Hi everyone,


I have the following impeller domain.
Impeller.jpg
ImpellerCross.jpg

In order to import this impeller into TurboGrid, I had to define the following flow path. The problem with this is that in order to get a domain which TG likes, I had to cut off a section of the impeller eye.
Simplifications.jpg
Surely cutting off this portion of the domain will mess up with any mass flow rate BC. In particular, if I fully rotate the suction inlet of the new domain, the new inlet is only 91% of the original suction inlet. However, if I use the same mass flow rate BC, there will clearly be a discrepancy in the results.


To see this, I created a mesh of this domain (including the suction pipe) using TG.
ImportedPump.jpg
ImportedPump2.jpg

When I run the CFD using this mesh (with periodic boundary conditions), I get a massive overprediction of head (12% more than when I run the CFD of the original domain using ICEM). The total pressure at the outlet is nearly the same in both cases. However, the total pressure at the inlet is 75% less in the case of the periodic pump simulations.

I am certain this is due to the removal of an annular portion of the suction pipe. I have seen such annular suction pipe domains throughout CFD analysis found online, but in this case, the original domain does not have such a shape.

I have the following questions:

1) How should I go about modelling this domain with a mesh from TurboGrid?
2) Should I change the mass flow rate to correspond to the difference of the inlet area?
3) Should I find an alternative method of modelling this domain (e.g. by only modelling the impeller domain with the annular 'cut off' with TG)?

I appreciate any help/advice anyone could give on the best way for me to proceed.

Thanks!

[islamOraby]

Originally Posted by jgross

Hi everyone,


I have the following impeller domain.
Attachment 75377
Attachment 75380

In order to import this impeller into TurboGrid, I had to define the following flow path. The problem with this is that in order to get a domain which TG likes, I had to cut off a section of the impeller eye.
Attachment 75381
Surely cutting off this portion of the domain will mess up with any mass flow rate BC. In particular, if I fully rotate the suction inlet of the new domain, the new inlet is only 91% of the original suction inlet. However, if I use the same mass flow rate BC, there will clearly be a discrepancy in the results.


To see this, I created a mesh of this domain (including the suction pipe) using TG.
Attachment 75378
Attachment 75379

When I run the CFD using this mesh (with periodic boundary conditions), I get a massive overprediction of head (12% more than when I run the CFD of the original domain using ICEM). The total pressure at the outlet is nearly the same in both cases. However, the total pressure at the inlet is 75% less in the case of the periodic pump simulations.

I am certain this is due to the removal of an annular portion of the suction pipe. I have seen such annular suction pipe domains throughout CFD analysis found online, but in this case, the original domain does not have such a shape.

I have the following questions:

1) How should I go about modelling this domain with a mesh from TurboGrid?
2) Should I change the mass flow rate to correspond to the difference of the inlet area?
3) Should I find an alternative method of modelling this domain (e.g. by only modelling the impeller domain with the annular 'cut off' with TG)?

I appreciate any help/advice anyone could give on the best way for me to proceed.

Thanks!

I run a case like yours.I made an axis at the inlet to be able to revolve the hub and get the true inlet shape, I defined it as free slip wall in B.C IN CFX.
Can you tell me more details about your model.