I am trying to perform static structural analysis of a turbine rotor which is rotating at a given angular speed, say 1000 rad/s. I also know the pressure distribution from CFD analysis. What are the boundary conditions for static structural analysis? I can use angular velocity as inertial load, pressure distribution from CFD as applied load. Do I have to restrain the translation/rotation of surface 1, shown in figure? If yes, why?

enter image description here


If your CFD analysis for the pressure loads was correct, the resultant of all the pressure loads should contain axial load and a torque on the complete assembly. If there is no torque, the turbine is useless! The axial load is not useful for the complete machine, but you can't avoid it.

Whether it is rotating or not, the turbine wheel can move in space as a rigid body. So you need to restrain it to prevent that, and create reaction forces that balance the applied axial load and torque.

The best way to do this is to restrain it at the location where it would be attached to the turbine shaft. You will then get the "correct" stress distribution in the disk. If you restrain it at some other arbitrary point or points, you will probably get an unrealistic high stress at those points.

(The turbine disk is probably not attached to the shaft at your "surface 1", unless it is a shrink fit on the shaft - for example there is no practical way that you could bolt it to the shaft through that surface.)

Actually, a much more efficient way to model this would be to model a sector of the turbine wheel containing just one blade, and derive the correct boundary conditions from the fact that the deflections of all the other sectors are identical (when measured in cylindrical polar coordinates) - but going into the details of how to do that is outside the scope of your actual question.

| improve this answer | |
  • $\begingroup$ Assuming that the turbine is attached to the shaft at "surface 1", should I take all three displacement and all three rotations to be zeros at surface 1, or just displacements? $\endgroup$ – Ali Baig Oct 2 '18 at 10:55
  • $\begingroup$ If your elements have rotation degrees of freedom, set them to zero as well as the translations - though in practice it probably won't make much difference, if you have a sensible mesh density in the model. $\endgroup$ – alephzero Oct 2 '18 at 20:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.