General Mechanical

Topics relate to Mechanical Enterprise, Motion, Additive Print and more

How to handle stress concentration

• Pyry Salmi
Subscriber

Hello,

I'm doing static structural analysis on a machine frame and I'm getting results that I don't think are quite right. I'm very much new to Ansys and thats why I ended up asking around here for some help.

In the picture below is a part of the frame from which I have removed some parts for easier visualization. The rectangular block-like part, a cylinder frame, is attached to a lug which in turn is welded on a RHS-pipe. The lug surface has the maximum stress in the model. The cylinder frame is removed from the right side picture.

I have set the contact between the lug hole and cylinder pin to be frictional. There is also a frictional contact between the lug face and cylinder frame face in case they come in contact. I have tried creating a contact sizing for the pin and the lug hole and the last thing I tried was applying a face sizing for both the cylinder pin and lug hole. I have used element sizes of 3, 2 and 1mm both for the case of contact sizing and for the face sizing methods. Diameters of the hole and cylinder pin are 36mm (~1 3/8") and lug thickness is 10mm (~3/8") to give a little perspective. Picture below shows the similar stress concentration I keep getting.

The reason I thought this is not correct is that the averaged maximum seems to occur on a single node instead of a larger area. The pin is in contact with the hole on the whole cylindrical face. Also the unaveraged stress plot doesn't look very good to me. Total deformation along the whole structure is less than a millimeter. Material for every part is structural steel- Element order is program controlled and seems that Tet10-elements are the most common ones in the mesh. The model is made with Solidworks and imported as a step-model into Ansys. Thats something I unfortunately cant change.

What could be wrong and how could I try to solve this? The model mesh with a 1mm sizing on the problem area has a total of 2,65 million nodes and 1,58 million elements (as i pointed out, the model is a bit larger than whats just shown on the pictures) and with the computer that I have in use, it takes a little less than 4 hours to solve so I cant really go and try things because it takes quite a long time ? Thats another reason for asking here and not just trying different things. The radius on the corner of the hole is 1mm and the maximum seems to occur on the edge where the hole geometry turns into the radiused corner. Could it be that even the 1mm element size is not small enough? Or could this result form penetration in the contact areas, and if so, how should this be fixed? The auto-scaled deformation animation shows movement of the pin in the lug hole. I probably could try automatic mesh refinement with convergence but I have gotten some error messages in the refinement module and a couple of times the computer memory hasnt been adequate to handle such a large model.

• Daniel Shaw
Ansys Employee

I recommend reviewing this video.