CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Centrifugal Pump simulation (

b.eshghi November 10, 2010 06:40

Centrifugal Pump simulation
I`m trying to simulate a centrifugal pump in fluent, I created the geometry in CATIA and meshed it with ICEM,
The mesh type is tetrahedral.
Can I use the following boundary conditions:
1- Inlet---> Velocity Inlet
2- Outlet---> Out flow
3- Impeller---> Rotational wall with specific rpm
4- volute---> Stationary wall

Note that I am trying to calculate the head of the pump so i dont have the outlet pressure.

siri November 10, 2010 08:04

R U using MFR ? I don't think so and u din't mention ..

is it steady case ? FRM approach is good if so.

Then of course, assuming you want to optimize the pump performance., it may be good to specify mass flow rate at inlet and pressure outlet if corresponding data is available. By doing this, I think you will be able to control the outlet to inlet pressure ratio (p02/p01) by varying the back pressure at outlet.

But if you are interested in just determining the head, do you really need a CFD for it ? why spend time on it (building geometry, meshing etc) where normal hand calculations might still save the day.

Well, this is no expert recommendation, i just happened to work on similar prob earlier. Let's see what other forum users have to say


b.eshghi November 10, 2010 08:53

To Mr.Siri
Dear Siri thanks for your attention,
Yes my case is steady, and i have not try MRF yet,
Does it really work?
How can I learn this approach? Do you have any tutorial?

I want to simulate it becouse the geometry of the impeller is more complicated than common pumps, and on the other hand I want to see the effect of impeller curves on the pumps head so is the "Outflow" BC appropriate for the end of the volute?

b.eshghi November 10, 2010 09:10

I found this topic in fluent guide :
Using Multiple Rotating Reference Frames
i will try it

Hamidzoka November 11, 2010 01:08

Dear Eshghi;
when you are going to set boundary conditions for a CFD problem, boundary conditions should contain both pressure and velocity. in this problem, your inlet and out conditions both empasize on velocity.
another point is that: if you do not know pump head, it means you do not know the pump characteristics. so what can CFD do for you? of course it can not determine pump head.


b.eshghi November 11, 2010 02:32

To Hamid Zoka
Dear Hamid,
I have made a test set up and tested the pump,
I have the "outlet pressure-volume rate" curves,
For example at a specific volume rate I Know the outlet pressure,
But I want to compare the CFD, H-Q diagram with experimental H-Q diagram,
If I use all experimental data for boundary conditions, So what does CFD do for me?
In other word My question is that : Can CFD predict the behavior of the pump just by knowing the geometry and inlet velocity?

Hamidzoka November 11, 2010 07:44

Dear Eshghi;
Geometry and inlet velocity is not enough to predict the bevaior of the pump. you need to set a pressure condition as well.
but there is still a solution. if you know pump inlet statirc pressure (it must be around 1 bar normally), you can calculate inlet total pressure (using inlet velocity) and set it as inlet condition.
for the outlet you can apply mass flow rate condition.
theoritically, this can solve your problem and you can compare calculated curves with experimental ones.
nevertheless, please note that outlet mass flow rate condition can impose instability to your solution and it may diverge. therefore it needs a serious numerical work.
which software are you going to use for CFD simulations?


b.eshghi November 11, 2010 07:59

To himdzoka
Dear hamid,
As I said in the first post my software is FLUENT and it`s virsion is 6.3.26.
come to talk about the boundary conditions later,
I am trying to use MRF method, have you ever used this method?

siri November 13, 2010 03:06


It is unclear from ur later posts what ur model is.

From original posting, i figured that you might be trying to simulate the flow for one/more rotor-stator stage(s). Hence the recommendation to model impeller & volute using a Moving reference frame (mrf). In which case I would further suggest you to look up the Frozen rotor method (frm) which does good job for steady flow simulations.

However, as MRF is computationally intensive and demanding parallel computing resources, it may not become practical to use it for the whole pump.

Your later posts indicate that you are probably trying to simulate the whole pump to validate against expt test data. That is by no means a simple task, u will need atleast a hi-end multi-core workstation with parallel cfd license.

In view of above, you may want to resort to other approximate but more simplified approaches such as Mixing Plane etc to still work it for you.

To sum up, you will need to research a lot of issues not necessarily just the above mentioned - before you jump into setting up cfd.

In fact, if you haven't done this yet, you are way far off to fix the BCs. BCs are chosen as consequence of physical data available, analysis requirements and modeling options.

An intelligent and learne effort cannot be substituted with outside advice if you don't want to go in loops.

I may be sounding hard here but you will definitely enjoy working as you go along if you followed the right steps.

Good luck .


All times are GMT -4. The time now is 09:51.