[Lor_enz]

Dear Community,

I trying to figure out the capabilities of SU2 for turbomachinery. As documentation is fairly good for that sector, i am hoping someone can give me some hints.

I am doing my phd in the development of radial turbines and need SU2 as part of my optimisation routine. So far, only using its direct solution capabilities and not the adjoint optimisation (this gonna come in the future)


A couple of things I have recognized so far:


Orientation of geometry:
For a turbomachine that is centrifugal-axial, the radial outlet/inlet must be located in the positive y axis, else the number of blades is not calculated properly. I think there is a y(max) value implemented, this should maybe adapt to look for r(max) instead to be more flexible.


Ramping rotational speed
For stability, i would like to ramp my rotational speed slowly up. However, when applying the ramp option it does not increase the velocity over the iterations. It keeps the 100 rpm that was defined in the initialisation.

ROTATION_RATE= 0.0 0.0 1885
RAMP_ROTATING_FRAME= YES
 RAMP_ROTATING_FRAME_COEFF= (100, 10, 500)

Surface monitoring

Surface monitoring seems to have some flaws, as the calculation is only done every 2nd iteration. Could that be the issue of having multiple config files in the multizone problem? Would it make a sense to declare surface monitoring in the sub config files?


Dry run turbomachinery
When running dry run on a turbomachinery to evaluate the output fields, the initialisation of the spans fails. This issue is already mentioned in https://github.com/su2code/SU2/issues/857 and is stated here only for completeness.



Meshing with Turbogrid
When meshing with turbogrid some things should be mentioned to be able to use the mesh in SU2:
> "Enable regions based on high and low blade geometry" has to be turned off, in order to have the right boundary markers
> Active "Conformal Tip Enabled" as mentioned already in Correct Tipgap Marker?


Best practice guidline?
It would be great, if people working on turbomachinery could start a best practice documentation helping newcomers get there first foot inside the wonderful world of turbomachinery simulation. I think it does not have to be yet something official, but the test cases are outdated and sparely documented.

Maybe it is possible to set up a thread in this forum, where people can share their experience with turbomachinery. Especially with a regard of the upcoming conference in October.




Best regards,

Lorenz

[Sakun]

Originally Posted by Lor_enz

Dear Community,

I trying to figure out the capabilities of SU2 for turbomachinery. As documentation is fairly good for that sector, i am hoping someone can give me some hints.

I am doing my phd in the development of radial turbines and need SU2 as part of my optimisation routine. So far, only using its direct solution capabilities and not the adjoint optimisation (this gonna come in the future)


A couple of things I have recognized so far:


Orientation of geometry:
For a turbomachine that is centrifugal-axial, the radial outlet/inlet must be located in the positive y axis, else the number of blades is not calculated properly. I think there is a y(max) value implemented, this should maybe adapt to look for r(max) instead to be more flexible.


Ramping rotational speed
For stability, i would like to ramp my rotational speed slowly up. However, when applying the ramp option it does not increase the velocity over the iterations. It keeps the 100 rpm that was defined in the initialisation.

ROTATION_RATE= 0.0 0.0 1885
RAMP_ROTATING_FRAME= YES
 RAMP_ROTATING_FRAME_COEFF= (100, 10, 500)

Surface monitoring

Surface monitoring seems to have some flaws, as the calculation is only done every 2nd iteration. Could that be the issue of having multiple config files in the multizone problem? Would it make a sense to declare surface monitoring in the sub config files?


Dry run turbomachinery
When running dry run on a turbomachinery to evaluate the output fields, the initialisation of the spans fails. This issue is already mentioned in https://github.com/su2code/SU2/issues/857 and is stated here only for completeness.



Meshing with Turbogrid
When meshing with turbogrid some things should be mentioned to be able to use the mesh in SU2:
> "Enable regions based on high and low blade geometry" has to be turned off, in order to have the right boundary markers
> Active "Conformal Tip Enabled" as mentioned already in Correct Tipgap Marker?


Best practice guidline?
It would be great, if people working on turbomachinery could start a best practice documentation helping newcomers get there first foot inside the wonderful world of turbomachinery simulation. I think it does not have to be yet something official, but the test cases are outdated and sparely documented.

Maybe it is possible to set up a thread in this forum, where people can share their experience with turbomachinery. Especially with a regard of the upcoming conference in October.




Best regards,

Lorenz

Hello Lorenz,


How is your experience with SU2 regarding turbomachinery simulations. I am thinking try and use the software.


Regards,

[Lor_enz]

Hey Sakrun,



I think SU2 has its value in the field of turbomachinery. There is a working group now active, that is improving the code with an annual meeting and some regular group meetings.



For me SU2 was not compatible with my problem, so i switched back to openfoam. Based on the work of the foam-extend working group, i have implemented a mixingplane approach with rothalpy as an energy equation into the current ESI version.



I think it is really up to the problem that you need to solve. Openfoam proberly gives you most flexibility and most documentation, but SU2 is newer but turbomachinery development has been on hold for while until recently.



hope that helps and reach out for any further information

[Sakun]

Originally Posted by Lor_enz

Hey Sakrun,



I think SU2 has its value in the field of turbomachinery. There is a working group now active, that is improving the code with an annual meeting and some regular group meetings.



For me SU2 was not compatible with my problem, so i switched back to openfoam. Based on the work of the foam-extend working group, i have implemented a mixingplane approach with rothalpy as an energy equation into the current ESI version.



I think it is really up to the problem that you need to solve. Openfoam proberly gives you most flexibility and most documentation, but SU2 is newer but turbomachinery development has been on hold for while until recently.



hope that helps and reach out for any further information

Hi Lorenz,

Thank you very much for the reply and you have done pretty amazing work too .

My PhD is related to turbomachinery as well, working on an axial compressor blade. At the moment I am stuck in a validation case for 2.5 years now and initially I used esi version of openFOAM 2112 for it. In order to conduct unsteady validation, I have tried to stabilise steady state simulations but couldn’t do it till now. So I have switched Ansys fluent and managed to converge the steady state simulation (for a mesh size of 5 million) but cannot do unsteady without having a mesh size of 25 million, which will take more than 6 months to go atleast 5 flow through time in the cluster. So I need 4th order or above and Riemann solver to carry on simulation for a 5 million mesh. Basically I have a inlet velocity of 175 m/s and 0.7 mach (Tot Pressure and temperature is 115775, 340.2 respectively). Also I need to set up outlet flow angle as well, in the outlet BC. I haven’t seen any outlet bc which capable of doing that ,apart from ansys fluent. So I am wondering if SU2 have all these conditions, so that I can put my time and sweat to it.
i have attached my Geometry and BC too,

[Lor_enz]

Well SU2 and openfoam (e.g. mixed bc) both give you the possibility to write your own BC for what you need. However I am not sure that setting the outlet angle is the right thing, as it is a result from the internal flow.



From the snippet you have provided, it is a classical Totalpressure/Total temperature inlet condition and a static pressure outlet condition. As from the total conditions, the solver calculates the velocity, you can also define the inlet angle of the flow. This works in ansys either in cartesian or cylindrical coordinates. Likewise can be done in openfoam. Not sure about SU2.



I had quite a struggle to get openfoam running as well. Especially my energy equation kept crashing. My solution was using the rothalpy which needed some adjustement in the solver and boundary conditions. I dont have yet to much expierence with SU2 to be honest. I think if you wanna discuss about using openfoam we can open up a new thread for that.



If you want to use SU2 i would recommend to reach out the SU2 workings group. You should find contacts on their website.

[Sakun]

Originally Posted by Lor_enz

Well SU2 and openfoam (e.g. mixed bc) both give you the possibility to write your own BC for what you need. However I am not sure that setting the outlet angle is the right thing, as it is a result from the internal flow.



From the snippet you have provided, it is a classical Totalpressure/Total temperature inlet condition and a static pressure outlet condition. As from the total conditions, the solver calculates the velocity, you can also define the inlet angle of the flow. This works in ansys either in cartesian or cylindrical coordinates. Likewise can be done in openfoam. Not sure about SU2.



I had quite a struggle to get openfoam running as well. Especially my energy equation kept crashing. My solution was using the rothalpy which needed some adjustement in the solver and boundary conditions. I dont have yet to much expierence with SU2 to be honest. I think if you wanna discuss about using openfoam we can open up a new thread for that.



If you want to use SU2 i would recommend to reach out the SU2 workings group. You should find contacts on their website.

Hi Lorenz,


I have found in a research paper that they have given a formulas to define both INLET flow and OUTLET flow angles, so that’s what I am looking for in opensource software as well. I have run some simulations in fluent by defining both INLET and OUTLET flow angles and I got results within an acceptable range (attached the formulas).


Well I am waiting for openfoam to be released their new version in the June to give it try once again tbh. Are you doing your work in foam-extend or in a different version of openfoam ?


At the moment I am doing their tutorials to learn SU2 , so hoping for the best.
I would like to discuss further wit you regarding openfoam related simulations, can I get your email address or something to get in contact? you can dm me if you’re ok with it.

Best regards,

[bigfootedrockmidget]

There is an SU2 turbo workshop in Delft next month:


SU2 Turbo Workshop July 11-12 Delft

[joshkellyjak]

As far as best practices go I would highly reccomend the workshop listed above. For the first point doesn't seem like too difficult of a problem for you to implement yourself - we will be hosting a introduction to SU2 turbo development at the workshop if you are looking for an introduction to the codebase. Dry run should be fixed, it works fine for me on the latest version. Meshing with turbogrid can be tricky, you are correct to use the Conformal Tip Enabled option, but you should export as CGNS through TurboGrid (Save Mesh as.), do not use CFX-pre as this adds in the tip interface.

[Sakun]

Originally Posted by joshkellyjak

As far as best practices go I would highly reccomend the workshop listed above. For the first point doesn't seem like too difficult of a problem for you to implement yourself - we will be hosting a introduction to SU2 turbo development at the workshop if you are looking for an introduction to the codebase. Dry run should be fixed, it works fine for me on the latest version. Meshing with turbogrid can be tricky, you are correct to use the Conformal Tip Enabled option, but you should export as CGNS through TurboGrid (Save Mesh as.), do not use CFX-pre as this adds in the tip interface.

Hello Josh,


Thank you very much for the valuable information regarding SU2 options and troubleshooting Josh,
I am using ICEM as for the meshing so I will import that as CGNS and try to make it run, hopefully.
I badly wanted to attend this workshop, even in online but I am managing resit examinations for the final year students. I don’t mind paying for the recording sessions if the sessions are being recorded.

Best regards,
Sakun