Rotating disc-blade contact, centrifugal loading convergence problem

Hi all coleagues.
I am a new enter to your discussion. I have solved the rotating disc-blade contact interaction, 3D problem. CAD solids were created in UGNX and exported to .stp file, two solids, a quarter disc and like-turbine blade solid. Mesh was processed in Salome Meca platform, exported to .unv format. I have used CCX Launcher version CL 34. My main.inp govern file seems to be OK. And master.sur and slave.sur files as well. But there is a problem with convergence. The results (.frd) are like-expected, but not fully converged. CCX solver is interupted with ERR message. If somebody will have a time, please, have a look at it and give me an advice. Thanks a lot…

message

1 Like

Quick thing to check first - constrain all 6 degrees of freedom of the blade (eg. all 3 DOFs on each of 3 non-colinear nodes using *BOUNDARY) to make sure it’s not flying away. Doesn’t really look like it from the displacement plot but maybe that’s the final non-flown-away state it reached. If that solves OK, it might be underconstrained or the contact might be behaving weird and perhaps needs tweaking of the contact and/or friction stiffness.

You could also use a much smaller centrifugal force and it might solve but show where it’s starting to fail as you increase the speed.

I’m not clear on how the transformed DOFs work but is it missing an axial displacement constraint on the disk? Even though there’s no load in that direction, it can still fail with that kind of freedom. The displacement is a long way from zero everywhere, which is suspicious. Could the whole thing be moving along the axis?

Hello,
Thank you for your fast respond. For additional explanation, indeed, the blade-like solid is not constrained, it is only inserted into disc and centrifugally loaded. I will try to fix it with three nodes. There are some rules for the nodes choice? Advice would be helpful. All nodes under centrifugal load are displaced to the new position. If some node will be constrained to zero displacement, it causes a local additional stress…
Regarding the disc constrains, I have used *TRANSFORM card (Type=C), i.e. global CS transformation to cylindrical CS. Coordinates X, Y, Z has been locally changed to Radial, Tangential and Axial direction respectively. I used for rotation global X-axis with definition of two points (0,0,0,1,0,0). Boundaries uy,2 and uz,2 define zero displacement in Tangential direction and boundary base,3 define zero displacement in axial direction, see next picture.
Now, I will have to rebuild my task. I use your recommendations for constrains. If some result will be available (neg or pos), I post it here…
Best regards

1 Like

I see. Looks like the disk is properly constrained. I wouldn’t have used *TRANSFORM on this simple model though because all the constraints are parallel to the global rectangular coordinate system, but it should be fine either way.

You’re right that the extra constraints I suggested will make it wrong. They’re just to help diagnose the problem. Put them anywhere so the blade has no rigid body freedom. If that works, you can remove them and look for more realistic constraints such as modelling whatever holds the blade in the slot or adjusting the contact.

Hello,

I have a positive result. After introduction of four constrained nodes, positioned symmetrically on the blade corners, it reached full convergence. No matter the magnitude of angular velocity. The result look good. But, there is the problem with the elements, belongs to constrained nodes. Excessively high local stress in all four corners. I will proceed according to your recommendation. I am going to include the feature, that hold the blade in the slot.

Best regards