When to Model with Thin vs. 3D Elements

As many simulation experts will know, thin or shell elements are a great way to simplify a large model. Generally, these tools are used when the length to thickness ratio is large.

3D elements would require 3-4 nodes throughout the thickness for accurate analysis. But with a shell element, this can be done with a lot less nodes.

“Typically, shell elements are used when modeling plates like we see in a cranes, planes and tankers” said ANSYS Sr. Technical Evangelist Pierre Thieffry.

Challenges associated with Thin Elements Models

Though thin elements do have their advantages, they are also associated with a few challenges. “The big drawbacks to thin shell elements is that they have less detail than the original CAD model and require a lot of elements that will need connecting,” said Thieffry.

This means that users working with shell elements will need to defeature the model. The process will require extracting mid-surfaces, main lines, and cross sections of the model. “SpaceClaim is an efficient way to transform your model into thin elements. It cleans the geometry of imperfections, is CAD agnostic, and will quickly extract mid-surfaces,” said Thieffry.

This reduction of detail, however, will create another issue with your model. It could oversimplify the system so details like fillets and slivers are lost. To solve this issue, Thieffry suggests users produce local 3D simulations of individual parts to keep the processing time low.

As Thieffry mentioned, another challenge when working with thin elements is connecting all the elements after they have been meshed. To simplify this connection issue, a tool has been introduced in ANSYS 16.0.

Connecting Thin Elements in Ansys 16.0

There are a few options simulation users can use to connect thin elements such as: coincident nodes, welds, topology sharing, and constraint equations.

“In earlier versions of ANSYS, it may have been easier to use constraint equations to bond your thin elements. This way you don’t have to worry about aligning nodes. Unfortunately, these equations can result in over constrained models and unreal results if the simulation expert isn’t careful. Topology sharing is another option, it consists of merging bodies at the geometry level. This is computationally expensive especially with large models and when frequent design changes are needed. Users can also use welds to model the system connections. However, welds are difficult to model unless you assume standardized welding practices. As a result, people tend to prefer coincident nodes,” said Thieffry.

He added, “You could have 2000 faces or more to connect. ANSYS has created a tool that will connect these components by moving some nodes around to force node coincidence. The preceding meshing will take some time to complete but this connection process will take a matter of minutes.”

Essentially, the tool looks for elements that are within a distance tolerance and connects them via the nodes on their edges. The connection tool will refine the mesh of an element to match the nodal edge of the connecting element. Therefore, each element can still have different mesh density, the tool will just refine the mesh near the edge to fit the elements together.

“The connection tool can save hours or even days depending on how big the model is,” said Thieffry. “Especially since design changes can be done locally instead of globally.”

Since the tool is based on a distance tolerance, users should keep an eye out for gaps and holes they wish to remain in the final model. It is possible the tool might mistake the parts as needing a connection.

Additionally, Thieffry has found that the tool works best with T-junctions. Users can use imprints to ensure similar edges and geometry will make contact with minor mesh distortions.

Finally, once the elements are connected, the system can be treated as a subassembly. This subassembly can then be added to a larger system. Users can call up the assembly in a larger model using ANSYS workbench. The connection tool can then be used again to connect the various subassemblies together as needed.

ANSYS 16.0 Thin Element Parallel Meshing

Since the model is meshed before the thin elements are connected, parallel meshing can be used to produce the mesh faster.

Parallel meshing allows each face to be sent to an individual CPU. Therefore, the more faces in the model, and the more CPU’s available, the faster the mesh will be completed.

“An added benefit to parallel meshing is that each individual face can be updated independently of the system. Instead of having to re-mesh the whole system, users can simply re-mesh the faces that have been changed. At this point, the faces can then be reconnected for the next simulation. To save further time, all of these processes can be automated,” said Thieffry.

Unfortunately, with this automation users will lose some of their control over the mesh. ANSYS understands that some engineers like to have a lot of control on their mesh. ANSYS 16.0 has not forgotten this. Thieffry said, “Engineers can have a lot of control on the mesh. They can even move an individual node to improve the overall quality.”

To learn more about ANSYS’ thin element structures, follow this link.

Ansys has sponsored this post. They have no editorial input to this post - all opinions are mine. Shawn Wasserman