Learning Forum

[timothy.lewis]

I am very well aware that this type of geometry pushes the limits of Discovery Live, but since we are evaluating its usefulness in our processes, we are looking to see what conditions it can and can't work with. After some geometry modification, we were able to get the results shown below: However, in order to get these results, we had to remove the bends from the sheet metal. When this modification is not done, we get the results shown below: I have three questions related to this:

1. Why does removing filets make the simulation in Discovery Live feasible? Isn’t Discovery Live supposed to work regardless of the number of filets on the geometry? Clearly, the thickness is a factor, but if it works okay elsewhere in the sheet metal, why can't it work just as well near the bends?
2. Are simulations with sheet/surface bodies possible? In other words, is it possible to define a thickness for sheet/surface bodies that can be used in a Discovery Live simulation? I know it is possible to define a thickness for a sheet/surface body in SpaceClaim, but is it possible to do a similar thing in Discovery Live?
3. Since Discovery Live is known to have issues with thin geometry, is it possible to add a feature that automatically detects thin parts and switches to using shell elements for those parts instead of using solid elements?

[Brian Bueno]

Tim Lewis

In response to your questions:

1. A distinction has to be made between feasible and accurate.  In both cases pictured above, the simulation is running.  However, as you've pointed out, the bends are not being taken into account by the simulation.  

Discover Live is intended to be a tool for design exploration, not design validation.  It will give you very fast results that are fairly accurate, but that speed comes at the cost of fidelity.  Large, thin bodies can be challenging to simulate.  In your particular scenario, it might be beneficial to increase the fidelity of the simulation to see if the geometry around the bends is picked up.

2. As I mentioned above, bodies that are relatively thin compared to their overall size will need to be monitored closely when evaluating the results of a simulation.  They won't always work well, but that depends on the specific geometry being used.

3. At the current point in time, Discovery Live only works with solids.  I am not aware of any plans to support shell elements in the simulations.

[timothy.lewis]

Brian Bueno, thanks so much for your response. My only comment at this point is related to point 3. We do a significant amount of work with sheet metal, and obviously there are other large companies that work with sheet metal far more than we do. Therefore, this tool would be far more useful if it could handle thin parts more robustly. Obviously, if the only option is to use solid elements in the solver, then getting a robust solution on thin parts would be extremely difficult. However, it seems like adding the capability to use shell elements should be a high-priority task. To say the least, I don't know what percentage of the industry uses sheet metal, but it certainly seems like it would be a problem if the entire sheet metal industry does not have a robust solution.

[Brian Bueno]

Tim Lewis 

This type of geometry requires higher accuracy to simulate.

For the most robust results, I'd recommend Discovery AIM or the other ANSYS simulation tools, which can handle this type of geometry with no problem.  

[timothy.lewis]

Brian Bueno I understand that. I'm just saying that there is undoubtedly an enormous number of people who would find it beneficial to be able to do the same type of things with sheet metal parts and assemblies that are currently possible with thicker parts. Specifically, it would be nice to be able to do design exploration with sheet metal parts.

I should also add that I understand very well why thin parts would be difficult to handle in the current Discovery Live product. Obviously, if everything has to be modeled using solid elements, then thin parts are certainly going to be difficult to handle. However, if shell elements could be used to mesh the parts that are too thin for solid elements to be used, then a robust solution could be found even when there are thin parts.

[Brian Bueno]

Tim Lewis 

For the record, I don't disagree with you.  There are a lot of instances where being able to simulate thin bodies would be hugely beneficial, and this limitation affects each type of physics simulated by Discovery Live. 

It's worth noting that the simulation results are highly hardware dependent, so the ceiling for maximum fidelity can vary.  What I mean to say is, while thin bodies will always pose a challenge, exactly how thin you can go before encountering inaccurate results can be different from one PC to the next.

This verification and benchmarking document goes over the solvers and methods and compares various hardware configurations: https://discoveryforum.ansys.com/t/m2rw27/discovery-verification-and-benchmark-cases

I'll pass your suggestion along to the product development team.  Our goal is always to keep improving.