Split Curve Discretized - Solidworks and Ansys

[tube_connections]

I am having a problem with Solidworks discreetizing 3D curved lines which is giving me problems down the road when importing my model into Ansys DesignModeler. I'll illustrate the problem by way of example:

 

I have a number of bodies that I want to "cut up" for better meshing. In the instance I have attached, I used a slightly over-sized sketch of the chord profile and the split feature to divide the weld into two sections. As you can see in the first attachment, the resulting split *appears* to be a continuous line in Solidworks. This is want I want. In Ansys, however, the continuous line is broken up into discreet segments (see attachment two). I'm fairly certain this issue is on the Solidworks side, but if there are any Ansys experts who think otherwise, I'm happy to explore all options. I've sunk the better part of a few days into this issue and I can't seem to move forward with it.

 

[shift1313]

Welcome to the forum!

 

The curve in the SW screen shot looks to be below the visible surface? Maybe its just cause your sketch is visible throwing me off. Anyways you are splitting a complex curve, how are you exporting the model? Depending on the filetype there are options to export curves as splines or plines. The preview on Ansys looks like its been made into some polylines.

 

I see the Academic tag on your Ansys, why don't you use SW simulation? If you have a student version the full simulation is included.

 

It has been many years since i have used Ansys but you shouldn't have to model the weld between components. You should be able to have them as separate components and make a bonded connection between them. In addition to that you can control the mesh density in specific areas(like near that edge) i believe.

[tube_connections]

Hi Matt,

 

Thanks for the warm welcome!

 

The curve is indeed complex, which I'm sure is lending itself to my troubles. To answer your question, I'm importing the geometry in Ansys through a direct link with Solidworks, my aim being to use this feature to do a comprehensive parametric analysis. I'm also passing the model parameters between the programs. I agree that the preview on Ansys looks like it's been made into some polylines. I have read somewhere that Ansys has trouble dealing with higher order splines, which could be causing this, but I can't confirm.

 

I have thought about using Solidworks Simulation. My current version of Solidworks is actually Professional, but I may be able to get access to Simulation through our licensing server. Do you know if the same type of parametric study described above would be possible on Simulation? It is certainly a feasible solution that deserves looking into.

 

This project is actually about looking at the behavior of the weld, and thus modelling the weld for appropriate load transfer through it is somewhat of a necessity. I'm trying to keep the weld throat size constant around the perimeter of the tube also, which further increases the complexity of the splines in the model.

 

Thanks for your response, and I very much appreciate if you have anything else to add after reading the above!

 

K

[shift1313]

So you are trying to model the physical weld? Are you assuming that it is perfectly smooth? How do you factor the penetration?

 

The sim shouldn't be an issue in solidworks. Its not the typical approach for welds though so I haven't specifically done what I think you are doing.

 

First thing I would do(since im not versed in the transfer to ansys) is check any settings. Ive used something similar to go from mesh software to solidworks and there were some tolerance values.

 

Next I would try a simple model with a true arc to see if it behaves the same.

 

Its an interesting and complex problem. Weld joints, how the tubes fit, metal prep etc all would seem to affect things. Do you have a physical test part you are trying to match for a baseline?

[tube_connections]

Hi Matt,

 

Yes, there are many basic assumptions including that the weld has a flat face (neither convex nor concave), and penetration, as well as base metal inclusion, are being ignored. There is also a 1 mm gap between the two tubes so that the load is transferred entirely through the weld.

 

I've been through settings ad nauseum in both Solidworks and Ansys, but unfortunately I can't seem to solve the problem from there. A slightly simpler model including the weld ad the two tubes as separate bodies but no further splits seems to work just fine though! (I did as you suggested in one of my attempts over the last few days). What I'm thinking of doing, now, is importing this "simpler" model into Designmodeler and doing my splits for meshing using the Ansys platform. As long as I can parameterize those splits within Ansys to accommodate the different models I'm sending in from Solidwork, it should work. I need to invest some time in learning to use DesignModeler a little better first.

 

I have a total of 12 similar connections already built and about to be tested, so I will have data from these experimental tests to validate the model. It is indeed an interesting and complex problem; just when I solve one problem (like how to model it!) I run into another (like how to mesh it!).

 

Thanks again!

 

K

[tube_connections]

Hi Matt! I'm happy to report: Problem Solved.

 

Ansys Designmodeler has some geometry cleaning tools, in particular one called "Connect", which will connect edges in close proximity (a tolerance is specified and edges selected). I selected all of the edges in my model and used a 1 mm tolerance, and this solved the problem. Now my curves are continuous and my selection sets from Solidworks are consistently passed through to Ansys. I tried varying some model parameters too, and it worked, so this fix should be robust enough to take me through my parametric modelling study.

 

Thanks for all your help! It was certainly encouraging to have someone on my side trying to solve this!

[shift1313]

Great work! Glad there was a way and you found it! Now hopefully the rest is smooth sailing if the results are something you can share publicly I'd love to see an update!

[tube_connections]

Alas, my celebration seems to have been premature.

 

It did the trick initially, but after a few more passes back and forth between Solidworks and Ansys, and some minor changes to the geometry, the problem returned. With this level of complexity in the geometry, I'm wondering if perhaps the connection between Solidworks and Ansys is unstable. I'm considering trying to recreate the geometry in Ansys, and thinking if I go that route, I might consider using APDL (instead of workbench) since it offers some other capabilities that I would benefit from having.

 

Any other thoughts on a solution would still be appreciated to mitigate my going back two steps though!

[shift1313]

Could you use some intermediate format like .stp that might give you control over resolution?