Learning Forum

[Jpret99]

Hi all,

Let me explain my question better. I have a turbine blade with a thin thermal coating on its surface. I have run some simulations using two different methods which give very different results. The basic load conditions are body temperatures for the blade and coating imported from CFX.

The fist method is modelling the coating using a thin layer in design modeller, and then forming a new part with the blade and coating together, which allows the mesh of each component to connect properly. (i believe this works the same as using shared topology for individual components)

The second method is by only modelling the blade, and then using the mesh pull (extrude) option to model the coating in Mechanical. I then used bonded contacts between the blade and coating.

Using the second method produced much lower stress, but takes much longer to solve. So my question is what is the difference between these two methods in terms of how stress is generally affected.

Thanks !

[peteroznewman]

Yes, Form New Part in DM is creating Shared Topology, which causes elements touching coincident faces to share the same nodes. This is the best way to obtain accurate stress output.
Another way to get elements to share the same nodes is to use the Node Merge feature. When you use the mesh pull (extrude) option to pull a layer of elements from the face of a solid, the nodes will match and Node Merge will delete coincident nodes and cause elements on each side to share the same node. This is identical to shared topology, so the stress results will be the most accurate.
Using contact to join a layer of elements is not as accurate, especially if the nodes do not line up, because some interpolation has to take place, but in your case, the nodes should line up.
Another reason contact elements are not as accurate is because some contact algorithms have a spring rate that permits some very small motion between the parts to occur. This has the effect of relieving some stress that would otherwise be higher. You should be more specific when you say you used contact because there are many formulations of contact algorithms and they have different characteristics. For example, the MPC formulation does not permit the very small motion that some other formulations allow. If you have nodes that line up, MPC formulation might give the same result as a Node Merge, except the solution will take longer to compute because of the extra elements you introduce.

[Jpret99]

thanks for the response.
I should have been more clear with my problem in the question. For the model with extruded mesh for the coating, i initially received an error indicating that the model was not restrained properly. I had solved this issue using the manual bonded contacts with default settings between the coating and blade. From your response, im thinking that i shouldnt have to use bonded contacts if the coating is extruded from the surface mesh of the blade ?

[peteroznewman]

I assume extruded meshes are not connected to the nodes on the faces of the elements they were extruded from. Rather than add contact to connect the extruded mesh for the coating to the blade elements, you should try Node Merge. That is the best option. Make sure the tolerance is small enough not to collapse elements, but large enough to capture all the coincident nodes.

[Jpret99]

It seems that using the mesh extrude options causes the node merge option to become greyed out. Is there any suggestion around this ?
Thanks !

[peteroznewman]

I did some testing and found that the mesh Pull feature does indeed create a mesh that is not connected to the underlying face it was pulled from and as you found, Node Merge is no longer an option.
The good news is that when you use Bonded Contact, and set the Formulation to MPC, you will get node-to-node bonding, which should give the same result as node merge.