Welded Uprights

[Horace]

Hi,

For those of you who construct your uprights by welding sheet metal or tubes, how do you FEA your uprights?

[Horace]

Hi,

For those of you who construct your uprights by welding sheet metal or tubes, how do you FEA your uprights?

[exFSAE]

Same way you'd FEA anything else? Shell mesh may work well depending on the construction.

The problem is the material properties. They will be hiiighly dependent on your fitup, welder quality, filler rod, and post-weld procedure.

This is why I stayed away from welded uprights and stuck with billet. Plus, I could make billet ones much faster.

Best way with a weldment like that is to do a one-off prototype and proof test it to load. Thats just my opinion though.

[flavorPacket]

Weldments are very easy to analyze. just connect the shell elements together. The problem comes when you have to correlate the FEA with the live part. The software's assumptions (isotropic etc.) can make it very challenging.

[MikeDutsa]

Model it as a single part instead of a sheet metal weldment assembly. Then run it in cosmos. Make sure you have a decent FOS, it will be lighter, and cheaper than a billet one, and will not break. Yeah the welds change the material properties, but you can take care of that with a little heat treating.

[Composites Guy]

One nice part about welding them is that you can close out those sections and reap your good torsional stiffness. We have welded them (4130 steel) in an aluminum jig and then had a material science student heat treat them for the past two years with no problems.

I would argue that the difficult part of the FE analysis is knowing what forces to apply and applying them correctly. We start with some lateral, bump, braking and accel assumptions, applied at the contact patch or wheel center as appropriate. We treat all supsension members as pin-ended members (force only along member, no bending moment) so a simple truss analysis can be constructed. We consider all possible combinations of these forces and pick some "worst case" loadings for each part. The proper application of these forces to the upright takes some thought. LU Bolton... care to weigh in???

[The AFX Master]

Since we use CosmosWorks to do all our fea (learning ansys too http://fsae.com/groupee_common/emoticons/icon_biggrin.gif), i'll talk in terms of it.

For an upright you'll have two zones to pay attention, the upright body that's made with sheet metal, the hub bearing supports and balljoint/rodend fixtures. Sheet metal will be modelled with shell elements and the bearing supports for sure will need to be modelled with solid elements (since those always have section changes and such), so a mixed mesh will do the job if you define properly all the bonded contacts between the sheet metal edges and the solid parts. The main challenge is to define properly you load cases and be SURE to don't overconstrain your model (ie if you do a free body diagram of your wishbones and then fix the upright from the spindle you'll overconstrain it). So the approach that Composites Guy said, using poin joints and the loads applied to the bearing races (and brake caliper jigs) is the way to go.

See the CosmosWorks tutorial on mixed mesh to learn how to do the whole thing. You'll need to do an assembly with all the solids as one part and all the shells as another, the fast way to do that is to pick the original model and copy it twice on a folder, then rename one of the two copies as name_of_the_model_here shells.sldprt Then you can open it and cut extrude all the solid features to left the shells alone. Do the same with the solids on the another copy.

Then you can assemble the shells and the solids, and run your analysis.

Post weld normalization help with stress relieving reduce brittleness on the welds

[Chris Lane]

I like your approach AFX, very simple and effective.

I'm gonna try that tonight!

[exFSAE]

Material properties are still going to throw you for a loop I think, even if you post-weld normalize.

I could be wrong, but from what I recall stuff like 70S-2 filler is designed to have 70ksi strength in the as-welded condition. Likewise 80S-2 is in the 80ksi range.

The idea is for something like 4130 you have a low carbon filler rod which 'dilutes' the weld pool of 4130 so when it quickly cools when you remove the heat, you have a more ductile weld zone. HAZ of the base metal is still brittle, obviously.

Depending on how much filler you use, and how good your welder is, you can have varying amounts of dilution in the weld area. And if you go ahead and normalize the whole weldment I'd think that while you make the base metal back into a more ductile microstructure, your low carbon weld area likewise is going to lose strength. How much? Good question.

That's the way its always been in my mind. I could be wrong. But that's why I stayed away from weldments and FEA on weldments. Plus.. the shape of the bead your welder puts down is going to change the geometry and stress concentration in the area.

Build it to a high stress-based FOS is only thing I can say. Will help keep the thing stiff at least, which is just as critical.

[Drew Price]

...But if you're going to heat treat a 4130 weldment, why would you use carbon steel filler? Yes, more ductile joint area, but weaker. With 4130 filler, yes more brittle, but the same strength, and closer to the same microstructure. Nice fillets at the joint should help alleviate a lot of the concern too, since the joint section will be greater thickness.

That was how I understood it all anyway.

Anything we made from 4130 and annealed was welded with ER70-S2 specifically to help keep the joints from being too brittle, even after the annealing process.

Best,
Drew

[The AFX Master]

Upright welding can be done sometimes without filler if you fit properly all your sheets so you can have butt joints to make your melt pool. You´ll have good weld penetration if you do your butt joints without clearances. Welding on a jig and post weld normalizing are our way to go.

About FEA, for sure all weldment modelling is always underdesigned due to the nature of welds, but as someone said above, use a conservative criterion about the factor of security over the welds. Minimize membrane stresses over your upright, that is, don´t clamp things on the middle of a sheet far from the welds (ie a push rod) then your welds will work on shear that´s way better than making them work in tension/bending due to membrane loads.

[Steve O]

The problem with no filler though is your weld will typically be concave which will put your weld in constant tension. Normalizing will help to alleviate this but its not a guarantee. I think you would be better off with a convex weld with filler than a concave with no filler.

Steve

[Drew Price]

It can be done correctly either way, you just have to make the decision ahead of time, and design the joint correctly for either case.

Whenever I joint alloy steels which will be annealed or normalized only I make sure to give the joint as much of a chance as I can, and increase the joint cross section appropriately with filler. The only really nice joints I have seen that don't utilize any filler are on 300 series stainless, which seems to lend itself to using less filler based on how nicely it flows when molten.

Using filler on alloy steels also has the added benefit of lowering the temperature of the weld pool, while allowing you to put in a little more heat to keep the pool size a little bigger, helping expand the joint a wider. Cooling the pool probably keeps the surrounding heat affected area a little cooler and does less damage to the steel farther away.

Obviously, there are many ways to do it. My $0.02.

Best,
Drew

[VFR750R]

I know FEA has been made famous based on it's stress calculations, but the biggy is deflection which is probably why your doing box steel uprights in the first place.

And luckily, the modulus doesn't change with heat treating, weld type, or weld quality (assuming it doesn't break). I would always recommend heat treating after assembly to get the strongest part with least amount of internal stress.

Once you use some FEA, the design juices really get flowing too. Opened my eyes alot when i first started using it.

Mechanica (pro-e) is very effective on solids (no experience with thin shells), but getting the constriants and loads right are critical otherwise results are worthless. Doing rockers for instance is much harder then you might imagine.

[RacingManiac]

Just thought I throw in my 2 cents here,

I did our 2006 and 2007 upright and the design served as the basis for their current design this year. I basically modeled the parts in Pro/Mech with the walls as shell element and all the housing/inserts and bolt-on clevice as solids. Tooks some minor convincing on Pro/mech's part to mesh the assembly properly. But once done I just took the same constrants and loads as our 2005's CNC billet design and compare number. I couldn't manage to get it down to the weight of the billet but I did get an order of magnitude more stiffness at ~40% more weight(finished upright is about 3lbs including all fastners, bearings and spacers used)in FEA. The uprights are welded in a pretty heavy duty steel jig designed to keep the upright constrain geometrically as they are sent for stress reliving. And yes, we use 4130 sheets with 4130 filler.

I did make a test piece to do the physical validation but it was kind of a rush job and it was not built to the same standard, so the results weren't that convincing. However with the uncertainty of the weld analysis in mind the upright is pretty overdesigned to anticipate issues, which we've had none in last iterations, so failure isn't an issue, the issue in the end comes down to how accurate the deflection numbers are. We did have our reasons to go away with the CNC design, as we simply don't have the access or machining hours available to us to make them CNCed, with weldment we can pretty much make everything inhouse and just use our heat treating sponsor which is readily available to us. Personally I like these better simply because they are more reliable than our last CNC effort, which was hard to work on and have some wear and tear issues that might be just due to poor consideration on the designer's part.

Our design choice was driven by practicality more than anything else, so may or maynot apply to you....