Bob,
Agreed. Though for getting to a level of nines that's significantly better
(and enables price points dramatically cheaper) than current expendables,
it's probably more than sufficient.
~Jon
On Sun, Jun 9, 2019 at 4:49 PM Robert Steinke <robert.steinke@xxxxxxxxx>
wrote:
I think this is a matter of how many nines of reliability you want. For
people who want rockets to eventually get to 6+ nines like commercial
aviation, the strategy of, "We have no propulsive margin, but don't worry,
the control algorithm will get it perfect every time." Seems like a bad
idea.
On Sun, Jun 9, 2019 at 11:16 AM Craig Fink <webegood@xxxxxxxxx> wrote:
When I worked on the Space Shuttle GN&C, I wrote a closed loop first
stage guidance algorithm that was small enough and fast enough to fit on
the Orbiter's tiny computers. Just upload the measured wind on the day of
launch and it would do the rest. Really nice algorithm that was never
implemented named BIG, Boost Iterative Guidance. When I was done, it flew
beautiful trajectory right down the center of the loads corridor and
hitting the SRB separation targets every time. At the end of the project I
decided to see if I need to add anything for the Engine Out Case, so I ran
various engine out cases without telling the Flight Software that an engine
had fail. The closed loop first stage guidance handled every engine out
better than the actual flight software. Instantly pitching up, within the
load/heating constraints and hitting the staging targets better than the
actual FSW. That's what a closed loop guidance is supposed to do.
There is nothing stopping SpaceX, if they haven't already done so, from
doing something similar with it's landing guidance. If the center engine
fails during the landing burn, I can easily imagine making a safe landing
with two side engines running with the right guidance algorithm. Or, even
landing with one side engine running, that would be a fun one.
--
Craig Fink
WeBeGood@xxxxxxxxx
On Sat, Jun 8, 2019 at 8:01 PM Henry Spencer <hspencer@xxxxxxxxxxxxx>
wrote:
On Thu, 6 Jun 2019, Craig Fink wrote:
... It's more efficient to make the corrections early than wait until
the last second.
Assuming you have the necessary information early. And that it's right,
and stays right. And that no unexpected complications prevent early
corrections. The Apollo 11 lunar landing wasn't *supposed* to need all
that last-minute maneuvering -- the pilot was supposed to see where he
was
going, do the corrections early, and then just let the computer fly the
LM
down the efficient nominal descent path -- but it didn't work out that
way...
"It's more efficient", so long as nothing goes wrong; if something does,
well, it can rather spectacularly fail to be efficient.
Henry
