Yeah and that sucks, let’s delete requirements!
As Matt pointed out, if you print the tank it’s 100% weld. Holding over the
requirement to inspect all welds means you’re going over the entire tank
surface with the same scrutiny usually used for <1% of the tank’s surface. It’s
a hopeless exercise unless you can get way smarter about how you do inspections
or remove them altogether.
On Mar 10, 2023, at 9:40 PM, George Herbert <george.herbert@xxxxxxxxx> wrote:
You have to inspect the welds in normal metallic tanks…
-george
Sent from my iPhone
On Mar 10, 2023, at 3:44 PM, eric.pillai01@xxxxxxxxx wrote:
I question the role of NDE for Relativity overall. The main reason you
inspect composites is because they’re workmanship and damage sensitive.
Metallic crack inspection is driven by fracture control requirements, which
can be satisfied through test. NASA 5009 doesn’t currently have a provision
for statistical fracture control but that’s the kind of thing that
Relativity will live and die around finding a solution for.
If you need to dye pen an entire tank surface you’ve already lost.
Ultrasonic for metallics is somewhat lacking as an inspection method, and
supposing you find something, doing artisan rework defeats the purpose of an
automated 3D print workflow.
Relativity’s structures need to be both qualified and accepted through test,
then have the process for building them locked down and controlled
statistically. Trying to get to analytical closure on minimum detectable
flaw sizes is an exercise in pain and futility unless you torpedo the entire
business case with ridonkulous NDE.
For reference on how insane these assumptions need to be: a typical dye pen
test will force you to assume the presence of a 90% thru crack in a weld
even if you pass with no indications. The scary thing is that it can
actually happen, but a more efficient workflow figures out how to have a
repeatable process such that you don’t need to do the dye pen inspection in
the first place and you have a test campaign that effectively screens the
finished parts for flaws physically rather than relying on analytical
margins.
On Mar 10, 2023, at 6:34 PM, David McMillan <skyefire@xxxxxxxxxxxx> wrote:
On 3/9/2023 10:16 PM, Matthew JL wrote:I imagine the process is largely automated. SpaceX has (or had, it's been
I just can’t imagine what the cost of inspecting 3D printed tanks must be
- one continuous weld…
a while) a similar setup that used a robot to run an ultrasonic inspection
device over the F9 fairings at the Hawthorne plant. The inspection pattern
was very thorough.
