SpaceX Starship : Texas Prototype(s) Thread 8 : Discussion

[xvel]

tiling non zero gaussian curvature (which nosecone is) with hexagons IS impossible

Only if you limit yourself to regular hexagons.
Of course, as you approach the nose they will sooner or later have to become non-flat, but the 2D projection could still be very close to hexagonal. Guess it depends on the gap-filling approach. (I assume we do not know whether the test articles are flat or follow the 9 m diameter curve.)

and you got million different types of tiles (like on space shuttle), the point is to make it with only a few types of tiles

[xvel]

[zubenelgenubi: xvel, please remember to attach images.  Do not embed them.  Thank you.]

Center ring is definitely different, maybe different alloy

[Lotick]

Center ring is definitely different, maybe different alloy

Center ring is a part of the flipping rig

[xvel]

Center ring is definitely different, maybe different alloy

Center ring is a part of the flipping rig

[zubenelgenubi: xvel, please remember to attach images.  Do not embed them.  Thank you.]
No, it isn't.

[Lotick]

No, it isn't.

Yep, you are right. I got confused by this picture

[zubenelgenubi: Please remember to attach images.  Do not embed them.  Thank you.]

[zodiacchris]

The different look of the middle ring could be due to a number of reasons, it might be from a different roll than the upper and lower rings or it might have been sitting in storage for a while loosing its initial lustre and sheen. Highly unlikely they‘d use a different type of alloy for just on one ring in a stack as that could open a can of worms with dissimilar properties and coefficients of the two alloys...

My 2c worth...

[danneely]

Right now we’re seeing a build rate of ~17/year.

One caution: flappies and heat shield have yet to be integrated into builds, which will affect build rates.

Assuming SpaceX continues expanding the workforce and adding additional work tents/buildings, adding more steps to the process will increase the time individual starships take to complete; but not the overall rate.

[Coastal Ron]

When you are dealing with material that is has known dimensions and positioning, it makes it a lot easier to automate the tasks associated with it.

For instance, wire harnesses are very floppy, and the connectors are not always easy to connect (sometimes because they are put in hard to reach locations). So wire harness installation is likely the last task you would try to automate.

Cutting, bending and welding stainless steel is much easier to automate, though for Starship it will take large fixtures due to the size of the work pieces. But that is also where most of the material is on a Starship, the stainless steel tanks and skins. And we see that already with the machines that make the ring sections, where those machines cut and weld the rings very precisely.

I think they will continue to automate stainless steel part production, even as they are doing testing. I think they will learn enough to understand what they know is a design that won't change, so automation will be worth the investment.

[dnavas]

Right now we’re seeing a build rate of ~17/year.

One caution: flappies and heat shield have yet to be integrated into builds, which will affect build rates.

Assuming SpaceX continues expanding the workforce and adding additional work tents/buildings, adding more steps to the process will increase the time individual starships take to complete; but not the overall rate.

Flappy construction can be parallelized, installation of them and the heat shield requires deeper pipelining.  Probably need more high bays.  That said, not sure why nosecone isn't being handled in the other wind break.  Also, final construction process might be to pre-mount heat shield material and then stack (as cone and tank look to be).  Will still need time to fill in the gaps, but that reduces high bay use.  Unclear.

Regardless, my point was that it's a stretch to claim a current build rate for SS when what we actually have is a build rate of tanks & nosecones (the latter appears to be happening at roughly the rate of tanks).  Still work required for the rest of it.  Thus "caution."  I'm personally more optimistic than OTV's statements, but I try not to be optimistic without counting the hurdles.  YMMV.

[tommy099431]

It seems like SpaceX could've upped production or I am going crazy but looking closely at some sections SpaceX is putting more writing on them, One some of the pictures they have clear writing on them and one in particular has caught my eye.

On the top ring, in the picture we can see this writing
T-9008.60 (not sure what it means top Diameter in milimeters)
STX-00130 (Seems to be section/ring SNs, STX = Space Technologies Exploration or SpaceX, 00130 = Section SN)
B - 9008.65 (not sure, something to do with T,bottom Diameter in milimeters)
WO# 1155037 (not sure, but I believe It could be Worker #/Who built the ring?)

The middle ring is marked, although the T & B are different anything in () I am unsure of
TP-9008(5)5(B)249C
STX-00129
HM-9008.44(B)247C
WO#1155036

BUT the one below that is STX-00137 which I am not sure why it would skip 7 numbers, especially if this is one section
T-9008.37(C or @)20C
STX-00137
B-9008.46(C or @)20C
WO#1155044

SN4 Top Ring (lowest STX # I've found)
T-9005.36 02(0)
STX-00087
B-9008.58
I have seen these numbers on a couple more rings, the Lowest STX I can find is STX-0087 on SN4 Top Ring

[MichaelBlackbourn]

T and b are top and bottom circumferences in milimeters

[brendan290803]

[OTV Booster]

I definitely think we'll see a lot of automation happen for the Starlink and Raptor assembly lines, which definitely could end up reaching production volumes comparable to the car industry.

For the Starship hull assembly, probably not.

Starlink for sure. Raptors... over a 10 year period, 1000SS w/6 and 100 SH with what is it, the number keeps changing. 42? That’s less than 7000 over a ten year period. Call it 10,000. That would be somewhere between a not popular car and a Cessna 172.


I’d guess 3D printing for as much as possible in the turbos and bell cooling channels, automated tube bending, semi automated casting, CNC machining, semi automated bell and CC fabrication and manual assembly. Moving parts from one machine to another (casting to CNC for example) a mix n match. Really only a guess and I’ve probably missed quite a bit.

Phil

[meekGee]

Automation is brought in for reasons other than cost..

Repeatability, controlled variance, auditability...

You don't want your 60 cent solution to bring down a $60M system..

[OTV Booster]

tiling non zero gaussian curvature (which nosecone is) with hexagons IS impossible

Only if you limit yourself to regular hexagons.
Of course, as you approach the nose they will sooner or later have to become non-flat, but the 2D projection could still be very close to hexagonal. Guess it depends on the gap-filling approach. (I assume we do not know whether the test articles are flat or follow the 9 m diameter curve.)

Working with my hunch generator which sports a stellar .003 batting average, I predict that the nose, which shows strong promise of having the highest EDL temps, will have a monolithic one piece cap that will extend a short ways down the leeward side an a bit further down the windward side with some sort of floating attachment scheme to allow differential thermal expansion.


This will allow a bit of relaxation on the need for unique tiles in the area where the radius has the fastest rate of change. This also does away with the problem of stagnation gasses (not really sure of the proper term) scrubbing dead through the seams up on the tippy tip.


An alternative hunch that was generated is a transition from hex to a mix of smaller pentagonal and sliver like triangular tiles allowing a more ‘soccer ball’ like pattern that is a bit more curve friendly.


Any topologists out there?


Phil

[SkyRate]

T and b are top and bottom circumferences in milimeters

Diameter. (Which almost certainly is computed as circumference/Pi.)

[niwax]

tiling non zero gaussian curvature (which nosecone is) with hexagons IS impossible

Only if you limit yourself to regular hexagons.
Of course, as you approach the nose they will sooner or later have to become non-flat, but the 2D projection could still be very close to hexagonal. Guess it depends on the gap-filling approach. (I assume we do not know whether the test articles are flat or follow the 9 m diameter curve.)

Working with my hunch generator which sports a stellar .003 batting average, I predict that the nose, which shows strong promise of having the highest EDL temps, will have a monolithic one piece cap that will extend a short ways down the leeward side an a bit further down the windward side with some sort of floating attachment scheme to allow differential thermal expansion.


This will allow a bit of relaxation on the need for unique tiles in the area where the radius has the fastest rate of change. This also does away with the problem of stagnation gasses (not really sure of the proper term) scrubbing dead through the seams up on the tippy tip.


An alternative hunch that was generated is a transition from hex to a mix of smaller pentagonal and sliver like triangular tiles allowing a more ‘soccer ball’ like pattern that is a bit more curve friendly.


Any topologists out there?


Phil

Do the tiles actually need to be aligned that carefully? On a traditional capsule or the shuttle, the tiles make up pretty much all of the heatshield and the underlying body is entirely seperate. So the shield is responsible for forming the shock wave, soaking up the heat and shielding the strucure. In some ways, they don't care much what is put behind them.

With the hot steel structure, what specific purpose does the heat shield serve? Keeping the shock wave further from the body and splitting any contact into much smaller and more complex shapes would go a long way in smoothing out heat spikes. This might well be possible with gaps beween tiles, even slighty irregular ones.

[sdub]

Mary's pictures this afternoon provide a nice view of the welded shut hatches to the fuel tanks, quite the contrast from the Mk1 design.  I assume at some point that these would move back to a more accessible design for inspections, but the current design is not much more than a plate welded in.  Certainly faster to build and makes sense for a prototype when there is a decent chance you'll never have to open it again.

[rakaydos]

You cant tile a cone with hexagons... but do you have to?

Assume the belly centerline has a line of tiles that follows the curve of the nose, up to a custom nosecap.

Then lay a line of hexes on either side of that centerline. Near the nose, the alignment is distorted, but the custom nosecap can be shaped to accomidate it.
Then another line of hexes to either side. Again, more distorted, but you just leave off the tiles that are wrapping to the lee-side of the nose. Repeat, until you have a full heat shield.

[Coastal Ron]

Mary's pictures this afternoon provide a nice view of the welded shut hatches to the fuel tanks...

Were those hatches into the tanks, or the areas in between the tanks?