CFD Online Discussion Forums

CFD Online Discussion Forums (
-   OpenFOAM (
-   -   release of the ERCOFTAC centrifugal pump - Fourth OpenFOAM Workshop (

olivier May 13, 2009 13:00

release of the ERCOFTAC centrifugal pump - Fourth OpenFOAM Workshop

The ERCOFTAC centrifugal pump Case Study has recently been updated.
The centrifugal pump with a vaned diffuser problem is a testcase of the ERCOFTAC Turbomachinery Special Interest Group.
It will be presented at the Fourth OpenFOAM Workshop in Montreal (1-4 June 2009). Prof. Marina Ubaldi, from the Università di Genova, allowed us to use the measurements data for distribution and set-up of the analysis within OpenFOAM.

You can find on the wiki the following informations:

* How to check out the files from the SourceForge svn.
* How to generate and simulate the cases.
* Different informations on some useful utilities (mergeMesh, stitchMesh, MRFSimpleFoam, GGi interface, and even more ...).
* Automatic post-processing and validation using sample and gnuplot.
* Automatic plotting of residuals using foamLog and gnuplot.

The case study can be found in the svn, in the following branch:

You can find the ERCOFTAC Centrifugal Pump Case Study at:

Please help us improve this Case Study by giving feedback or by contributing!

Best regards,
Olivier, Maryse, Håkan and Martin.

ivana May 28, 2009 04:26

ERCOFTAC centrifugal pump test case - problem
Hello everyone,

I have been running ERCOFTAC centrifugal pump test case (running with no changes and with Allrun) and I face the following problem:
In the Time = 10 after Solving Uy I get “Floating point exception”. Actually I have this problem since I have made update to revision 1266. I have returned back the revision 1241 and it is working fine. I compiled the both versions in the same way and I have not noticed any problems. I have tried on different hardware and OS platforms (but the same compiler). I always get the same error.

Anyone else experience the same, or I should further search the problem by us?
Thanks for your reply.


olivier May 28, 2009 08:31

Hello Ivana,

There is indeed a problem with the latest version of the svn, hopefully it will be solved soon.
What I would recommend is to stay for now at the working version (svn 1240) until this is solved. However, it is possible to make the test cases converge with the latest update of the dev version, but you have to play with some parameters a bit: the reason why the simulations are stopping is that k and epsilon are bounding above the acceptable limit, so the simulation stopps. A bounding k and epsilon is usually quite common at the begining of a simulation, as a lot happens at that time, but it should stay under a certain level. If it does not, as it is now, you can choose to increase your under-relaxation, so that you take only a small amount of the solution into account.

So here, if you want it to work, you can put in system/fvSolution a under-relaxing parameter of 0.5 for k and epsilon, and from what I have seen, it should work. However, the convergence is not great.

Ultimately, I recommend to stay at a working svn version.


david July 13, 2009 05:11

Hi all

Thanks a lot for this interesting test case. At the moment I'm testing the GGI with turbDyMFoam on a similar, but 3D case. This 2D case will be interesting for me as I can get results in a much shorter time.

I have two questions concerning the ECPGgi2D case:

1) checkMesh reports the following error:

***Number of edges not aligned with or perpendicular to non-empty directions: 24545
<<Writing 49090 points on non-aligned edges to set nonAlignedEdges

The simulation seems to work without problems but could it be that the non-aligned edges cause difficulties in 2D cases? Or is this error negligible and not critical?

2) I've seen that a new solver called simpleTurboMFRFoam was developed. What was the reason for this? Will it have any advantages over MRFSimpleFoam?

Best regards,

husker June 25, 2013 03:54


Thanks for building the case study and contributing to it.

Although I'm not allowed to employ OpenFOAM in my office, I'm deeply interested in a centrifugal pump case study.

Could anyone provide geometry and performance data for which to be used in commercial CFD codes such as FLUENT or Star-CCM. Following to my assessment, I will be glad to take your attention and discuss the results.


Lisandro Maders June 26, 2014 11:10

Reference pressure - CHALMERS study

I am currently performing the validation case of ERCOFTAC centrifugal pump with vane diffusers. Looking at a CHALMERS pdf, I found the way they calculated the Cp coefficient was using the standard equation:


The p0 is the static pressure at the suction pipe. They made an assumption of such value by trying to obtain similar levels of Cp as the experimental results of Ubaldi had. The value they found is 700 Pascal.

What I am wondering is if this value is in absolute or Gauge pressure. Also, if they are using an incompressible solver in OpenFOAM, shouldn't they be using the pressure in m^2/s^2 (pressure/roh) ?? Ok, maybe they multiplied the pressure value taken from the simulation by the density and then figured out the Cp coefficient.
Anyway, the absolute/gauge pressure issue keeps in my mind.

Really thankful for any help.

Regards, Lisandro Maders

huangxianbei October 6, 2015 04:48

How is ensemble average achieved?
Hi, olivier:
I read your paper "Numerical Investigations of Unsteady Flow in a Centrifugal
Pump with a Vaned Diffuser" and found that the calculation results are ensemble-averaged as I read from the paper "Fig-ures 3 and 4 show the ensemble-averaged experimental and
calculated radial and tangential velocities in the radial gap at ...". So did you use the same procedure in the experiment (phase-locked sampling and ensemble-average)?


All times are GMT -4. The time now is 15:29.