Punch Press Operation

I have a simple hollow tube that has a 18mm OD, 1mm wall thickness with a 20 degree bend. I need to simulate it being deformed by a 80 ton press using a 15mm round die that has a 2mm radius around it’s edge. The photo shows this process on a real steel tube.

I’ve spent over a couple hours trying different methods and all are unacceptable.

Please tell me how YOU would do this.

Thanks C

Hi C,

On my phone so bound to words only but fwiw:

1 model the tube before the punch

2 cut out the piece where deformation starts

3 create the cross section through the center of the punch but do not model the circular punch, just a section as if the tube was squished. Sort of like a rounded rectangle.

4 create a loft from 1 open edge through that section to the other edge. Keeping tangency or curvature from the edges.

5 if you are not happy with the result change the sections, positions of the cuts or add maybe a sections or 2.

6 create a cilinder with fillited edges to form the inside of the punch dent

7 do a boolean difference, subtracting the cylinder from the loft to form the dent.

8 join dent and loft and create a blendedge along the hard edge

Does this make sense in words only?


1 Like

Following the @Willem instructions:

1 Like

Thanks Willem and Diego.

Diego, you are missing the fact that from the side it is not symetrical. The top flattens down to the bottom. Not the top and bottom flattening toward the middle. When I follow Willem’s instructions the shape isn’t correct. The deformation is too thick along the axis.

If you’ve modeled it, you might try Impact.


It’s might not even be updated. The interface is so engineering-ghetto ugly. It’s java, so you know it’s going slow, but it was easy for me to make some stuff and destroy it.

I’ve also used Code Aster, but it seems better suited for either single or assembled contrivances with well-fitted parts, notsomuch, for smashing things like tired child with a bad temper. Cases are often prepared in Salome Mecha. The .stp files from Rhino 3D import well. The results are accurate enough for France’s nuclear containment. Though, the messages are in French.

There’s also scan and solve, which I really wish I had a copy, but once again, it may be more for stress than a bit of the old press ultraviolence.

There is Elemer, but I think it’s going to pigeon-holed to be a university program.

OpenFoam has some FEA. I have not seen any results. There are rumors that someone might have --ehem encouraged them to support Windows fully.

If you just wanted to draw that, I would do it from axial cross sections. Some of these sections would have a flat on them, the resultant curves could be filleted/matched/joined to the rest of an oval shape. The pipe could then be lofted.

Some profile curves, NetworkSrf, Blend the punch, and perhaps some point editing, should do the trick.

Profile curves for NetworksSrf

Not perfect but it can be done.

// Rolf


BlendSrf, F 10, Revolve and OrientOnSrf do the homework.
I think you already have several ideas
Good luck!https://aws1.discourse-cdn.com/mcneel/uploads/default/original/3X/d/1/d18892e4fcb8ae182ad3f25e594e2f2f74756e32.mp4


@Willem and I were pointing an example or an option if you want, about the details is your own job to tweak and try different methods.

@aerofiles Are you trying to determine the resulting shape from the press operation, or are you trying to create an approximate model of the deformed shape using knowledge of what the shape looks like?