Hacker News Reader

Comment by Aeolun | Hacker News Reader

It seems to me that cutting out, and welding or glueing in a new part had a serious effect on the structural integrity of the vehicle.

View on HN

They are not going to weld castings except maybe for very small cracks and even then it's very unlikely. The casting is not in a place that could be repaired if it were steel, though. It's the main structural member, located behind the rear seats. If you needed to weld there you'd need to take the car practically down to its frame.

Even if it were in a place that was accessible and subject to repairable damage, you would not see it welded or cut. First off a lot of it isn't physically possible to weld; it's in very deep (6") pockets and you'd have a huge amount of difficulty getting penetration. Second, welding aluminum is not super easy. Anyone can weld steel, aluminum isn't the worst but it is much trickier. Third, aluminum welds create a weakness in the aluminum around the weld, because it gets hot but isn't properly tempered as it cools. It's still strong, but the loss is significant- temper is responsible for ~70% of the strength of common aluminum alloys. Fourth, the weld and the surrounding area are very weak (<40% strength) for several days after the repair. It's a huge pain in the ass and you probably can't just let somebody drive off on it.

I'm sure there's some cases where that might be true or it's not practical for other reasons (I'm not an automotive engineer or anything), but remember, the 3 has ~70 parts welded together for this same area of the car, so welds themselves are not fundamentally bad.

Just as the BMW i3 the Alfa Romeo 4C also has a carbon fiber monocoque: https://www.toraytac.com/media/story/XkKY/Alfa-Romeo-4C-Mono.... And Alfa Romeo has published a 162 page manual for bodyshops on how to do repairs. The PDF is out there in the internet for people that want to know the technical details.

I dont think the main concern here is that the new part would be welded in.

When the car is in design process, each weld is carefully calculated to provide the exact binding force at the precise location. In case of accident, the car is modeled to crumble just right to absorb maximum energy without compromising cabin space, thus protecting occupants. The process of cutting out and welding in parts almost always means altering the stiffness of such bonds (and/or moving the bonds to different locations), which can be very dangerous in the event of accident

From the internet: "assuming your joint is designed properly and you have an experienced welder performing the work, your welded joint will be as strong as the base materials it is joining."

Welds are key to most bicycle frames and they are not known as the part that's likely to fail.

The "weld is stronger than the base metal" argument is often repeated on the internet, but it's overly simplistic. Welded joints don't usually fail at the seam (a.k.a the weld). Instead, they fail near the seam where the base metal has been stressed due to thermal expansion/contraction and heat treating. It is correct that when done properly, everything should be fine, but I wouldn't trust the vast majority of body shops to have access to welders who would do it "properly."

This doesn't make it any more clear whether weld repair poses a practical danger for most people in most circumstances. If it's the case that metal stress near the point of welding is the where it's more likely to fail, that still doesn't tell me how much more likely to fail. For all I know it could be the weakest part but still such a trivial difference as to pose no concern.

And the point about welders leaves a dangling, open ended implication that invites anyone reading the sentence to just go ahead and assume that poor weld jobs leading to immenent danger are happening all the time. But are they? Or is that just being noted as a hypothetical possibility? I'm having a hard time telling whether a possible or probable danger is being asserted here.

Welding is like programming. Everybody with half a brain and some skills can weld. But proper welding is an art form, and the difference between the two levels is many years of work and study.

So for most people welding like that is not in the cards, but a trained and meticulous professional can do miracles (or close to miracles). It starts to get really interesting when you start hanging stuff from your welds, for instance when building an overhead crane.

Welding is just practice, practice, practice. Like everything else.

Welding lifting rings onto stuff is not really a high bar. High quality welds are easy if the materials you're dealing with are easy. Stuff that is exotic, expensive and hard to work with is the stuff that takes real skill because it's really hard to get a ton of practice at those things (there's only so much magnesium alloy that needs welded) so you need a lot of skill and experience in varied circumstances to do a high quality job from the get go.

I can weld and I can dick around in a bash shell but I'm much better at the one that is my current day job that I do for ~40hr a week.

You only know how good your welds are once you've seen them destructively tested. What looks good may not be all that good. A really good welder will be able lay bead after bead after bead with very consistent quality and material characteristics.

It's hard to look at someone else's weld and know if it's good because lord knows what they did under there (if it's externally bad there's some pretty obvious signs). It's easy to lay a weld and know that it's good from a combination of how you did it and how it came out. If you know what you're doing then a bend test just confirms what you already knew. X-rays keep honest welders honest.

For "easy" material combinations you can get away with a surprising amount of corner cutting and still pass a bend test. Good welds aren't hard. A community college will have you passing a bend test on an overhead weld in a semester and doing the same for pipe in two but it will all be in steel and a limited variety of consumables and processes. Those kids might be masters of 7018 but hand them some aluminum electrodes and they'll be lost. Good welds the first or nearly the first time on material + consumable combo on which you only burn say 10lb a year is what really takes practice.

The vast majority of welders do not possess the skills to properly repair an automobile frame. There's no way I'd trust a body shop to repair my car's cast frame with weld on replacement pieces.

Yep. Thermal hardening is a thing and if a weld isn't properly annealed post work then there is a good chance that it will fail right next to the weld itself. This is taken as proof that the weld is stronger than the material that's been welded but actually it is more often than not proof that the weld wasn't properly thermally treated.

Once you do that though, it is actually true, and even then you may see stress cracks next to the weld if the change in thickness is too abrupt. You need to gradually thin down or you'll get a focal point for stresses.

They are the part of the frame that fails, but they’re also at joints, which is where stress collects. I’d be interested to see failure modes for lugged (silver soldered) frames vs TIG welded vs pressed/glued or other bonded joins of tubed (no monocoque) frames.

Frames do fail at or around welds though, especially for even mildly exotic metals beyond plain low-carbon steel. I've had them break before, and it makes for a long walk home.

That is true in steel, but not aluminum.

Because of how thermally conductive it is it doesn't work the same way.

The single-cast is only the aluminum. Several common vehicle glues are stronger than aluminum (e.g. Steel Reinforced Epoxy), and aluminum itself is not very strong.

The safety critical structural elements of their cars are a mixture of mild and high strength steel[0].

[0] https://www.teslarati.com/tesla-model-3-body-structure-steel...

A 2017 article is very out of date. The Model 3 rear subframe is cast aluminum: https://insideevs.com/news/408645/tesla-model-y-mega-casting...

Which is certainly a safety critical structural element.

Additionally, glues are not stronger than aluminum. I don't know if you're thinking of a composite resin (eg glass fiber), but normal adhesives are weaker even than pure aluminum, which is ~5x weaker than all the commonly used (copper age-hardening) alloys of aluminum.

I have no interest in a pointless debate about the semantic difference between the passenger compartment Vs rear crumble zone and what "safety critical" means, no do I over the word "common" in "common vehicle glues."

If you're going to read my post with the worst possible interpretation then try to correct that uncharitable reading, I just have no interest in a discussion.

So basically you didn't look at the link, because the castings -which, again, are new- are part of the safety cell. They are not the rear extruded rails.

If you had specified an adhesive when you first wrote the comment, sorry- I managed to not see it. I don't know of any "steel reinforced epoxy" besides JB weld (which generally should not be used on anything bigger than a lawnmower), but it and actual 3M vehicle adhesives[1] are a full 10x weaker in shear and 15-20x weaker in tension that plain boring 6061.

[1]: https://multimedia.3m.com/mws/media/1467016O/07333-irsa-tech...

The Model 3 rear subframe is not the same as the Model Y which you link too.

my thoughts as well, otherwise what was the point of casting it whole in the first place?

The single casting for the Y is for cost an NVH reduction per the article & Elon. My early production Y has 4 pieces for the rear instead of the single one this new machine will produce, but it shouldn't be any structurally less sound

All the remaining advantages that are mentioned in the article: cost, weight reduction -> increased autonomy, better handling and driving comfort.

The point is: "We're moving to an aluminum casting instead of a series of stamped pieces. We'll go from 70 parts to 4, then 1 with a reduction in weight, improvement in [NVH], reduction in cost, and a significant drop in capital expenditure for all the robots that used to put 70 parts together."

Reply on news.ycombinator.com