Groove Weld - Discussion

[Anonymous]

I am looking at adding:

- bevel

- v

- u

- j

groove weld types to the bead types supported as
modeled solids. This would mean you'd be able to do mass property analyses,
etc... not supported when these welds are modeled using the "cosmetic"
representation.

My question is this:

If I add only the above four bead types and I
do not add support for the forward creation of backing
beads or melt through (on the opposite side) will these be used?

I'd request that anyone who choses to reply to this
- please limit your comments to the question. I'm not looking to open a general
"weld bashing" thread.

Thanks!

G

[Anonymous]

Thanks for the update Gary.  (especially on a
Sunday)<g>

Cory McConnell

[Anonymous]

If Autodesk could solve the problem of modeling self intersecting sweeps
and fillets, I suspect it would make your weld issue easier to deal
with. The other thing this would enable is the ability to make cuts
which are trivial to do on a 2D mill or with a router. This is
something we have asked for since MDT / R1. It would also provide a
little advantage for Autodesk in the market place.

As an example of what I mean, draw a 2x2x2 cube. On one of the edges,
place a .5 fillet. On a face which is away from the fillet and parallel
to it, place a sketch and draw a 1 radius arc. Now, try to sweep this
arc as a cut down the straight edge and around the .5 corner. The
preview will show correctly but the sweep will fail. To construct the
correct geometry, we must perform a number of operations using sweeps,
extrudes and revolves. In the real world, the mill / router just runs a
1 radius cutter down the path.

Gary R Smith (Autodesk) wrote:

> I realize that we have quite a large opportunity in making the welding
> functionality more useful for those of you who use it a lot. One area
> that I want to focus on is the welding of structural frames with
> tubing, angle and channel. Often in these welds, you'll have a fillet
> down one edge and a groove along the perpendicular edge with a really
> "ugly" bit of transitional topology that wraps around the corner. For
> the welder, this is just business as usual. For the mathmaticians we
> have working on the building the surfaces and solids that represent
> the weld bead - this gets ugly.
>
> The reason the groove welds I asked about are on the top of the "short
> list" is that they were actually in the earliest builds we were doing
> for what shipped as R6. We ended up having to disable the modeling
> functionality because we were unable to get the symbol and
> caterpillering to correctly recover (with repeatable stability) within
> the drawings.
>
> I'm hopeful that we'll get all of the capabilities that most of you need.
>
> A very large percentage of the design work that I used to do (back
> when I wore your shoes) involved a base frame weldment which was
> subsequently machined that then had parts bolted on.
>
> Please be patient - the weldment functionality will improve.
>
> Thanks,
>
>
> Gary

[Anonymous]

Charles noted:

> If Autodesk could solve the problem of modeling self intersecting sweeps
> and fillets,

The issue of self-intersection comes up in the welding environment quite a
bit.

Take a simple triangular fillet weld bead and wrap it around a piece of
square tubing where it would be welded to a piece of plate to form a pad.
The solution we have used in running the fillet involes actually running a
quad profile rather than a triangular profile - this has to do with how the
bead topology would run along a non-planar surface. As it turns out the quad
profile was self- intersecting on this simple case (the fillet weld around a
square tube) and failing. We have to detect these cases and handle them.

The quad profile approach is part of the solution that will allow us to
bridge "gaps" - allowing two pieces to be welded together that don't
physically touch.

(oh, and by the way, I can actually spell mathematician... 😉

G

[Anonymous]

I debated whether to leave it misspelled or not. My spell checker
caught it ( I can't spell squat without a spell checker).

Is there a chance that the self intersection fix will be used in other
places? (you don't have to answer 😉 )

Gary R Smith (Autodesk) wrote:

>Charles noted:
>
>
>
>>If Autodesk could solve the problem of modeling self intersecting sweeps
>>and fillets,
>>
>>
>
>
>(oh, and by the way, I can actually spell mathematician... 😉
>
>