Tubular Chassis Force Simulation

[Anonymous]

Hi there

Im busy building a motorcyle powered kart for my kids. Ive started out with a simple ladder frame, which I will now be adding structural elements (diagonal members, shear plates etc) to create beaming and torsional stiffness.

The issue Im running into is how to approach a tube structure when modelling. > As you can see, Im failing to join the seperate pieces of tubing.

Chassis v6: Study 1 - Static Stress: _____ 2016-04-09 12:32:43 _____
Chassis v6: Study 1 - Static Stress: Warning: Body 'Chassis v6:1/Component1:1/Component2:1' uses voxel mesh. Adaptive mesh refinements are not supported for this body.
Chassis v6: Study 1 - Static Stress: Base mesh: 38295 nodes, 20791 elements
Chassis v6: Study 1 - Static Stress: Solver mesh: 38295 nodes, 20791 elements
Chassis v6: Study 1 - Static Stress: Warning: Model separates into 4 independent groups of bodies
Chassis v6: Study 1 - Static Stress: Warning: Soft spring was added
Chassis v6: Study 1 - Static Stress: Number of equations solved: 114885
Chassis v6: Study 1 - Static Stress: Solution time: 1.3s

[Anonymous]

OK, figured out  a workaround in the pipe flow. But Im not convinced this is optimal. So Im hoping theres a better way to do this.

• Instead of sketching one large rectangle with inner struts, (which is what I initially did), I sketched one rectangle, then applied PIPE.
• Then sketched a 3 point rectangle to snap to the first rectangle sketch.
• Applied PIPE to this new SKETCH selection, with JOIN active.
• Repeated this process another two times.
• This gave me a properly joined up tube model. You can see by the wireframe that the connections are correct.

[Anonymous]

In the simulation environment, you need to define Contacts. The contacts between frame members will "Bonded". When you initiate the command, you will be asked to pick the surfaces that are bonded together, so you will pick the outer diameter of the long rails and the end face of the cross members. If you haven't change the default setting, you could also run "Automatic Contacts" which will create Bonded contacts between every face that is modeled within a millimeter or so of another face.

[Anonymous]

BTW, if you're not already aware of it, look up Formula SAE. It's a college design/racing competition where most of the cars end up being motorcyle powered. Might be a good place to "borrow" ideas.

[Anonymous]

Yes! I am aware of Formula SAE! The development on the cars are really good. 🙂

Apolgies for my late response. Your suggestion on the [BONDED] parts setting makes absolute sense, BUT in fact I did have this checked.

The issue is with the actual modelling in the PIPE / SKECTH environment. So, in retrospect, a better title for this thread wouldhave been "PIPE MODELLING INFORMATION" or something.

------------------

Both the example below have [BONDED] active. But look at the wireframes where the pipes are joined, and the results: Left side is correct, right sides are incorrect.

BUT, if the wireframe geometry is CORRECT - as below, then the resuts are accurate.

So I guess, a proper PIPE modelling tutorial would be really helpful.

I havent found anything - Anyone have any info?

[josef.kucera]

Hello @Anonymous,

Automatic contacts are surface/surface contacts, and not only are the planar end faces and cylinder surfaces not close enough, we also rule out certain cases like penetrations. You can adjust the contact distance (tolerance) under Manage, Settings, General but we still won’t detect a contact in this case.

You can run the Bonded contact command and manually place the contact between the circular end face of one tube and the cylindrical face of the other tube. To select the disc face that has penetrated into the other pipe, hold down the left mouse button until the geometry selector comes up then you can cycle through the faces to select the hidden one.

However, since you have a several physical penetrations this really isn’t a valid model, and the simulation results may not be as accurate as if you modeled the pipes with the proper physical (mitered) connections.

Josef

[Anonymous]

Hi Josef.

Thank you for the reply.

I seem to be making progress on this issue, but still have some (probably) basic questions on things I'm not getting right.

Here is the step by step way I've done a rough test, in which I'm getting expected results:

Are you seeing something in this process that I'm doing wrong? I seem to be getting the results I would expect.

Only thing I don't understand is the Voxel Mesh Error I'm getting?

------------------------------------------------------------------------------------------

A secondary issue, and again, probably should be another thread, is the process of building tubular objects.

The acute angled sections, highlighted in the red circle go beyond the limit that I would have expected them to go into. IS there a setting that defines this? 

What would the correct method be for creating the model below without the angles protruding as far as they do?

Thanks again! 

[josef.kucera]

Hi @Anonymous,

warning about voxel mesh couldn't be neccessary an error, it just informs you that for some part of your design voxel mesh was used instead of traditional one with tetrahedral elements...

Briefly about voxel mesh:

Voxel meshing first subdivides the space into small blocks.  It then decides which blocks are fully inside the model, and which partly overlap the model.

For the ones that partly overlap, it tries to recover the boundary of the model inside that cube, but sometimes it needs to make approximations to the boundary to succeed.  That is why the boundary of the mesh and the model may be different in some places.

Voxel mesh is usable if it's not in a critical area then it may save some time with meshing process.

But if it is in the critical area you can suppress it, replace it with a point mass or replace it with a simpler shape (fix it).

Regards,

Josef