Is there a reason SpaceX does not spiral weld Starship rocket bodies?

Question

Spiral welding is a standard manufacturing technique for forming tubular metal structures.

Being a continuous process, spiral welding is ideal for automated welders. Quality control and re-welding are also incorporated into the continuous process.

Spiral welding can produce sections of optimum length, not just rings which happen to be the same axial length as the material roll. Current Starship ring width was not chosen to optimize design. Ring width was limited to the maximum roll width available from steel mills.

Spiral welding can reduce total weld length. The total weld length is equal to the perimeter of the roll. Any other design strategy which involves cutting the roll (like rings) increases the perimeter and therefor increases weld length.

Spiral welding eliminates the vulnerable "skelp weld" which closes the ring sections. In a pressurized cylinder, wall tension is greater circumferentially than axially. This maximum tension is carried by the skelp weld. Spiral welding eliminates the skelp weld. If this is the most vulnerable location in the ring, spiral welding could potentially reduce material thickness and rocket mass.

Because spiral welding can occur on a horizontal surface (as opposed to vertical assembly of rings), Submerged Arc Welding techniques can be used. This technique can be up to 10X faster than the Gas Shielded Metal Arc welding used by SpaceX on its vertical welds. https://en.wikipedia.org/wiki/Submerged_arc_weldingm. Spiral welding would also be an ideal application for Friction Stir welding.

This technique would seem to have advantages for mass production of rocket bodies.

https://www.techtimes.com/articles/288269/20230227/worlds-first-spiral-welded-wind-turbine-tower-now-up-running.htm

According to Elon's "Idiot Index" Is Elon’s “Idiot Index” an over-simplification? , his mantra is:

1) Question: why are tubular rockets not manufactured using the technique which has been the industry standard for fabricating tubes for over half a century?

2) Delete: get rid of the axial "skelp weld" in every ring section

3) Simplify: go to a continuous, rather than batch process

4) Accelerate: submerged arc welding is many times faster than gas shielded arc welding

5) Automate: an automated spiral weld process (like that used for making thousands of miles of pipeline) would allow automation of the entire process including post weld inspection, corrective welding, and cut-off into pre-selected optimum lengths

Elon has envisioned mass production of thousands of starships. This would justify an on-site rolling mill to produce continuous sheets (optimum width, no rolls, no skelps) feeding into a continuous spiral welder. The flying cut-off would produce sections of lengths that were optimized to suit the component they were intended for.

This proposed on-site rolling process could vary material thickness as desired, producing variable wall thickness including tapering thickness. Varying the sheet width could produce tapered sections including transitions between cylinders of different diameters and "bullet" noses.

Answer 1

I'll take a stab at this from an industrial engineering perspective because that's what I'm most familiar with:

Efficiency, Effectiveness, and Cycle Time

Basically, the idea of using spiral welding to make the body of the rocket has similar vibes to 3d printing buildings: it is very cool in theory or even in a prototype, but in the end it is mostly pointless because you are fundamentally optimizing the wrong thing (efficient but ineffective).

When you're constructing residential houses, putting up the walls and framing is basically the fastest and easiest step (the Amish famously build a barn in a day). What takes time is doing all the "fiddly bits" like HVAC, electrical, etc. Now, if you have your building 3d-printer, you are spending a lot of money on a highly complex machine that... doesn't really have any advantages. Yes, it theoretically it might save time, but from a cycle time perspective, that doesn't matter because there are other work steps that are the bottleneck.

In industrial engineering this cycle time is often shown with a chart like this one:

(click for source)

Here, the different sequential manufacturing process steps are shown on the bottom (A, B, C, ...) and the time it takes to complete each manufacturing step before it is handed on to the step is shown by the cycle time (CT, usually in seconds, minutes, or hours).

On this chart you can draw a horizontal line (such as "Takt time" in the chart) and if all the process times are below this line, you have a (theoretically) functional assembly line. Helpfully, the image also has a bottleneck illustrated. This is a process step that takes so long, that essentially all other steps in the chain are waiting on it to complete, and thus it determines the pace at which the product can be manufactured.

In this example, switching to continuous spiral welding would be like lowering the time it takes to complete process C. Welding is very likely not a bottlneck that the whole Starship production line is waiting on, and since engineers only have a limited amount of work hours they can spend on optimizing process, they should logically choose to invest their optimization effort in another step where gains will impact the entire production chain.

Simultaneous manufacturing

Another point is that assembly through attaching rings has advantages for parallelizing manufacturing: If you have a bunch of ring sections, workers can do work on these sections simultaneously and then combine them later. Starting out with one long tube might be practical in one way, but then they need to install a bunch of electronics, plumbing, valves, etc and you can only have so many people working on one thing at a time.

If you break it up into ring sections, not only can people work on it indoors at reasonable heights and angles, but you also have a modular system where you can have some manufacturing sub-chain entirely focused on cranking out "ring section 18" or whatever and then you end up with a universal part that can be combined into any stack or swapped out before being welded on.

Eventually though?

Of course, there does come a point where welding of the rocket body will become a bottleneck if you are trying to push manufacturing truly into the mass-manufacturing scale. Like, if you're cranking out >10 rockets per workday and you need tube sections as fast as possible. Then, it might be worth investigating if creating the body tube in this way can shave a couple work hours off of the production process.

Doubt this will be soon though.

Answer 2

It may be that it is just easier to manufacture a number of sub assemblies and fit them with bulkheads, thrust structures, domes, flap fittings, access ports, stringers, heat shield mounting points, header tanks, COPV's, cold thrusters etc individually while they are in small pieces and can be turned upside down if required for easier processing rather than try to do it all on some large enclosed monolithic structure.

It might also be easier to make changes. A change to the dome design could mean just scraping one section rather than the whole ship.

Answer 3

I can think of a few reasons:

1. A spiral welding jig large enough to fabricate a Superheavy booster would be enormous and expensive, and it would lock in aspects of the construction: it would give an upper bound to the length and diameter of the booster, for instance.

   SpaceX is still at the prototyping stage, which benefits from having a more flexible assembly process.

2. When you use spiral welding, you end up with an empty tube 71 meters long, which would have to be fitted with its contents from the inside, unless you make the jig even more elaborate to allow fitting internals to the section being currently welded together (place a tank bulkhead and wrap the skin around it). That long tube also has no structural strength yet, and would deform under its own weight, so you have to build another giant jig to support the tube while it's being outfitted.

Answer 4

How do you know that you've completed the correct length of spiral? What if it's 2cm longer, or shorter? Or if the temperature has varied between this morning and this afternoon, and so the metal has shrunk or expanded or...

Where was with rings, it's very clear you've finished - you can count them easily. And welds on a single ring section are going to be always at a fairly consistent temperature.

What if it matters if a fitting goes in at a specific spot in the spiral? How will that complicate things? Rings are really easy to plan around.

Spirals... there's a reason we talk about spiraling into depression!