Hans-on Reliability (ft. Hans Königsmann, former VP at SpaceX)

Before anything launches into space, someone has to make sure it won't implode onto itself. This week, we sit down with reliability legend Hans Königsmann - one of the earliest SpaceX hires and former VP of Reliability - to talk about what it really means to build dependable systems for space. From launch failure postmortems to risk-taking with rigor, Hans shares war stories, lessons learned, and how those experiences shape the way AstroForge builds today.

Episode Transcript

Speaker 03 (00:00):
Today is going to be a pretty

Speaker 02 (00:02):
special episode of Roid Rage.
We're going to talk aboutreliability.
I'm not going to define it yetbecause I have someone in the
room who can give a way betterdefinition than I can.
But it's special because

Speaker 03 (00:12):
Robyn's not

Speaker 02 (00:13):
here, right?
I mean, it is special.
I mean, it's unfortunate youdon't get to dox anybody, and we
avoid probably 95%

Speaker 00 (00:18):
of the dad jokes.
Although, I don't know, maybewe've got dad jokes over here.
But without further ado, wehave Hans Konigsmann.
Did I pronounce that correctly?
Pretty much, yeah.
Okay.
Pretty much.
It means I got it wrong, but I

Speaker 02 (00:30):
will get better.
He is, for those of you whodon't know, he's a former VP of
Build and Flight Reliability atSpaceX.
He was there basically from thebeginning and

Speaker 00 (00:40):
saw all the good, the bad, and the ugly.
And he's here to talk aboutreliability with us.
So Hans Konigsmann.
Can you introduce yourself toour listeners?
Who are you and why are youhere?
Yeah, I mean, Hans Königsmann,I'm...
So I went to school in Germanyand to college there too.

(01:02):
I had my first two jobsbasically there.
Built satellites actually, asmall satellite at the
University of Bremen.
And...
Basically, I was interestedalways in attitude control and
satellites and so on.
And then I got picked up by acompany here in LA.
They basically imported me anddid this whole immigration thing

(01:25):
that now everybody talks about.
And then I worked seven yearsthere working on different
spacecraft, but also on smallerrockets.
What company was this?
It was Microcosm.
Okay.
Okay.
I worked on suborbital rockets,basically.
Kind of got me interested inrocketry a little bit.

(01:45):
And then Elon Musk basicallypicked me up more or less by
accident.
I just met him, and he was inthe process of founding SpaceX.
Tom Mueller and Chris Thompsonwere already on board, and so
they just asked me if I couldjoin, primarily as an avionics
person.
So it covered more thanelectronics, but it's

(02:08):
electronics, it was guidance andcontrol, it was software, it
was flight safety.
Did I miss anything?
It was like the whole thingthat neither Chris Thompson on
the structures covered and TomMueller on the propulsion
covered.
They were just the rest.
They were basically on my side.
Where did you meet him at?
I met him at a...
It was a rocket competition,like a weekend up in Mojave.

(02:31):
And I actually went there withGwen Shotwell because I used to
work with her before SpaceX.
Wait, so Gwen was at Microcosm?
Yeah.
Oh, really?
At the time, yeah.
And so we drove up there, spentthe weekend there.
Elon was there too, probablylooking around, probably
scouting for talent and just ingeneral trying to figure out how

(02:53):
this whole rocket thing works.
Yeah.
Yeah, that's the rest of thehistory a little bit.
So what years were you atSpaceX?
I started in 2002.
SpaceX itself, I think it wasformally founded on Pi Day 3-14

(03:13):
in 2002.
And then I joined in May, 28thof May.
Basically right after Thomsonand Mueller, basically the
third...
or fourth, depending on yourcurrent technical employer.
And then we started basicallyhiring people like crazy.
Yeah.

(03:34):
And when did you leave?
I left 21.
Okay, got it.
So 19, 19 and a half years.
Wow, so you were there almost20 years.
Yeah, almost.
Yeah.
Wow.
So I saw a lot of change.
Just a little bit, right?
It grew a little bit over thosetimes.
We had a couple of launches.
Yeah.

(03:54):
I think actually since I left,there's going to be like way
more launches than until Istayed there.
Whereas the rate got so muchhigher.
And the first launches, it tookus like half a year to get the
rocket off the ground.
So it's like, and now I look atthem and it takes like a half
day.
It's crazy, the speed.
What

Speaker 03 (04:13):
did you think about when you were first approached
by Elon to potentially join thecompany?

Speaker 00 (04:17):
I actually thought, this is great.
So for a couple of reasons, Ithink...
Somebody gave me his namebefore I did some research, and
I knew that he wanted to spendsome serious money on space
exploration and Mars.
So I thought...
To me as an engineer, somebodyputting their money on the table

(04:42):
and saying, go ahead and buildit, it's just perfect.
Because then you don't have todeal with trying to find the
money and going from one littlestudy to the next little study
and so on and so forth.
Because that was the mode I wasin before.
I was either working for othercompanies, consulting basically,

(05:02):
Or we had little studies underSPRS and other vehicles that
basically just paid for mysalary.
And so this was basically apurpose-driven thing, which I
really thought was awesome.
And then there was also a blanksheet of paper.
Elon basically had no historyin aerospace in that sense.

(05:26):
He wanted to have the bestsolution that we can come up
with.
And that meant...
you could actually try to builda really clean vehicle from the
get-go.

Speaker 03 (05:36):
What did you guys think?
So on that note, when you builta vehicle from the ground up,
right, from a blank sheet ofpaper, what were the things on
avionics you changed over the

Speaker 00 (05:45):
way?
We changed a lot.
I mean, the one constraint wehad was time at the end of the
day.
Time and to some extent alsomoney, right?
It sounded like a big numberwhen we started, but we knew
it's going to go away and we gotto spend it.
And then Elon said he's in forthree launches.
or three failures, basically.
We needed all three of them,and he was still in.
So money and time were the twoconstraints.

(06:12):
And so time being theconstraint, we started to build
everything at the same time inparallel.
And so we made compromises onboth computer, battery, and so
on and so forth.
And so we looked at it laterand go, well, yeah, it worked,
but it wasn't a great solution.
So we got like, it's a PC-104system.

(06:32):
I don't think it still exists.
Maybe it does.
It's like, I called it the ATMbecause it can't do a lot,
right?
But it was actually totallysufficient to fly a rocket.
We had to make some changesbecause there's so many valves
you have to flip on the rocket.
And so you have to add stuffbasically to the ones you can
buy.
And then we try to basicallypackage it so that it survives

(06:55):
the launch, which is, yeah.
Was Falcon 1 single string?
Yes, very much so.
We had no luxury on dual stringor redundancy and everything.
We basically worked everythingso that it should work during
flight.
It was all hand-worked andhand-soldered and everything
because we had no time to setanything up.

(07:17):
Any

Speaker 03 (07:18):
critical mistakes you made?
Obviously,

Speaker 00 (07:20):
Falcon 1 first three launches failed.
Were any of those your fault?
The second one, I think, wasmine.
The first one, we argued overthat.
The second one was basicallyspun out of control because of
sloshing baffles missing on thesecond stage.
And that was under me.

(07:40):
And I kind of presented that wehad unstable solutions, but we
also had stable solutions.
So we had analysis that wasinconsistent.
And one thing I learned is Ihad like, three different models
that were stable and then wehad two models that were
unstable, something like that.
Um, and so we, we looked atthat and then, then Elon decided
that we should fly it anyways.

(08:01):
And my lesson out of that isthat, um, uh, physics is
actually not democratic.
It's just, um, it doesn'tmatter how, how many models are,
are in favor and how many onesare not in favor.
It's the, if the model is rightor wrong, right?
I mean, that's what countsbasically.
And, and obviously models arenever completely right.
So, um, I mean, this vehiclewent pretty far.

(08:26):
The first stage worked actuallybeautifully on that flight.
The second stage spun out ofcontrol and got a couple
thousand kilometers downrange.
So it was pretty good comparedto the first one.
The first one was miserable.
The first one was going up for28 seconds and then it shut down
because we had a little fuelleak.
And the fuel leak tripped,basically ignited at liftoff and

(08:51):
then burned through somepneumatic lines that shut
everything down.
So it shut down because ofpropellant utilization saying
the tanks are dry?
No, no, no.
So basically the fuel leak wason, I think, on a deucer.
and they're stripped down onthe engine and they're just
ignited and burned through thepneumatic lines for the main

(09:12):
valves and stuff like that.
So basically what I always sayis fire on the wrong spot.
It should be all the way downthere and it's like up there and
that's not great.
But it was a combination, Iwould say, out of...
We concluded in the end thatthere was a B-nut.

(09:33):
I never actually know whathappened to the A-nut, but the
B-nuts are basically somethingthat connect lines, tubes.
And this was aluminum and ithad corrosion probably.
And so the stress corrosioncase was what was concluded.
I don't know.
I mean, it's kind of like hardto tell because we'd never found

(09:53):
the piece.
But yeah.
We left the rocket outside.
It was in salt spray.
And this particular B-nut wastalked a couple of times back
and forth.
And part of that was also thatwe had to make changes on the
engine controller.
So I do have a, there's afactor of me going in that, but

(10:13):
the B-nut is not my B-nut inthat sense.
So, yeah.
Yeah, it was partly, in myeyes, it was also careless
because we left it outside andit corroded.
We never did that after that.
How long was it outside for?
I think it was a couple ofweeks outside.
Wow, so it was quite a while.
Yeah, we left it outside over,we left the island for a couple

(10:36):
of days, came back and was leftoutside.
We would never do that again,frankly.
We learned how really harshcorrosion is there.

Speaker 03 (10:43):
This was in Kwajalein?

Speaker 00 (10:45):
That was Kwajalein, yeah, yeah, yeah.
I mean, people would, peoplewould tell us where to park the
bicycle.
Yeah.
If you park the bicycle upwindon the house here, it would
corrode basically by the timeyou got back.
And if you leave it on the, onthe other side, it would survive
another six weeks.

Speaker 03 (11:02):
So it was

Speaker 00 (11:03):
just super

Speaker 03 (11:05):
dense.
Yeah.
How hard was it to build anavionics system?
You know, today we havecommercial space.
We kind of understand it.
You guys led the way and we cango look at how other rockets
were built.
How hard was it back then whenyou had, you know, everything in
rocketry was pretty classified.
It was pretty under wraps.

Speaker 00 (11:21):
So we didn't really look for vendors.
We went out for vendors firstand tried to get like
components.
And if you build this together,it's millions.
And we just decided that's notthe way what we want to do.
So just to make it a little bitharder.
We build everything ourselves.
Although, actually, inhindsight, this is harder in the
beginning, but I think it's theonly good approach.

(11:42):
Because most vendors arealigned with expensive rockets.
And if you're trying to make itcheap...
I mean, if I go all the wayback, what SpaceX always tried
to solve is actually not thephysical problem, but the
monetary problem.
The financial problem withspaceflight.
What...
took what held Spaceflight downover decades is the cost, and

(12:05):
not anything else.
I mean, the physics we know,but we produced it at such a
high cost that we could just notdo it at scale.
And so in that sense, that wasthe overarching mantra, what we
tried to do.
And producing cheap meant, inmany cases, do it differently
than buying at an expensiveoutside vendor.

Speaker 03 (12:30):
How much cheaper was Falcon 1 compared to, obviously
different class rockets, right?
But compared to something likea Delta IV?
Oh.
What was the cost of Falcon 1,I guess?
What were

Speaker 00 (12:42):
you guys paying for?
It's a really good question.
I don't think I can answer thiseasily.
So the price we listed was $6million at the time.
And the cost of it...
The first one probably wasmore.
But in the end, if you wouldhave repeated it, we would have
probably been in like 5, 6million range on the rocket,

(13:04):
something like that.
And that

Speaker 03 (13:05):
was how much mass to orbit for Falcon

Speaker 00 (13:07):
1?
It was not as much as wewished.
Oh, man, that's been a while.
I thought actually the targetwas like 1,000 pounds or so.
Okay.
Yeah.
And I think we struggled with500, 600 probably in the end.
But this is not unusual.
You have a spec.

(13:28):
You're building the rocket tothe spec.
And then you find out everybodygets overweight during their
lifetime.
And then the propulsion is likethe ISP is not as good as it's
supposed to be.
So you obviously fall off onyour performance.
And then what you do is...
You iterate and design itbetter, and then you kind of
claw your way up to the originalperformance or even higher.

(13:50):
That's kind of what Falcon 9basically did over many, many
years.
So I'm pretty sure if he wouldhave continued Falcon 1, he
would have probably clawed backsome of the performance over the
long time.
But still, $6 million at that

Speaker 03 (14:03):
point in history compared to $450 for Delta 4 was
a big price cut.
True, but I mean, Delta IV alsocould launch Falcon 1.

Speaker 00 (14:13):
Fair

Speaker 03 (14:14):
enough.
Was there a small launchvehicle class that was
operational at the time?
There was Pegasus.
That was $55 million a launch.

Speaker 00 (14:21):
True, but Pegasus on the other side was also, was
Pegasus really the first truecommercial one?
I think yes.
But Pegasus was solidpropellant, and it's really hard
to change the cost on that.
It also

Speaker 03 (14:37):
dropped off the underbelly of an, what was it,
an LC-10?

Speaker 00 (14:40):
Yeah, whatever, the TriJet.
There were two, right?
There was a DC-10, there was aLockheed, and a Douglas.
And I think the orbital one wasthe Lockheed one.
Okay.
The one that had the S-turn inthe engine.
And it had the three engines,right?

(15:02):
Yeah, yeah, yeah.
Both have three engines.
That's why...
The DC-10 has, I think, thestraight line, straight engine
in the tail, and the Lockheed 1has the one with the S-tron,
right?
Neither one, I mean, peoplelike to fly them, but they
weren't really successfulbecause they did have a
reliability problem, some ofthem.
like doors falling off andhydraulic lines being cut and

(15:26):
everything.
There were a couple of horrificaccidents.
And

Speaker 03 (15:29):
also, we just got so much better at jet engines that
two jets were more efficient.
That was a regulatory change inthe 80s that just killed the
tri-jets.
That's why, because you couldgo to two jets for more than,
what was it, 120 miles

Speaker 00 (15:40):
aircraft range.
It's the intercontinentaloperation, basically, with two
twin engines, yeah.

Speaker 03 (15:46):
Yeah, that's probably why they used them,
because they became real cheapto operate to fly rockets off
of.

Speaker 00 (15:50):
True.
I mean, what airplanes learnedis something that rockets should
probably mimic to some extent,or satellites in that sense.
And that was basically, theyhad so many flights and so many
data on what fails and doesn'tfail, that they can basically
improve very quickly.
Targetly.

(16:10):
I mean, improve certainspecific things like, well, the
jet engines, yeah.
They know the maintenanceintervals.
They don't know that very well.
And so they can also drive thecost down without risking lives
and accidents, basically.

Speaker 03 (16:26):
Was there anybody else?
I don't know how much aPegasus, sorry, not a Pegasus,
the Russians charged at thattime for a rocket launch.
Were you guys alone on anisland as far as class of rocket
at that point in time otherthan Pegasus?

Speaker 00 (16:35):
Oh.
I think...
So the Russians were...
Earlier than that, you couldbuy cheap Russian rockets,
right?
And this was basically theimbalance between salary and
Russia versus salary in the restof the world.
So it took advantage of that tosome extent.
But then also the Russianslearned that they can just sell

(16:56):
the rocket for $1 less and youwould still buy it, right?
Because it is $1 less and notnecessarily $10 million less,
yeah?
So they learned how to do that.
That's actually what I think...
motivated Elon to found SpaceXbecause he tried to buy a rocket
from the Russians and theRussians basically just fooled
around and tried to jack theprice up as much as possible and

(17:19):
yeah so building a rocket anddeveloping this was the logical
conclusion in that case butthere were other rockets too
let's see there was Orbitalactually built Taurus at at a
later time.
So there was that rocket.

(17:39):
There was also a Minotaur,which is the, actually literally
was half Minuteman and half, Ithink, Taurus.
Oh, is that what it was?
Yeah.
I think that's what it was,yeah.
Was PSLV

Speaker 03 (17:56):
operational when you guys started?
The Indian one?
The

Speaker 00 (17:59):
Indian one, yeah.
They had some early versions ofthat, but I don't think they
were very successful on thecommercial market.
The most successful on thecommercial market was actually
always Ariane.
Oh, yeah.
So Ariane was the Europeanproject, basically.
And it took four iterations toget to something that actually
would fly to orbit.
Wow.

(18:20):
I think they jumped even overcertain...
So, I mean, this is how youactually not build rockets,
right?
You build the upper stage inGermany, you build the first
stage in France, and then youbuild peace in Italy, and then
you try to actually get throughthe interfaces in all different
languages, right?
I mean, that's just like alogistical nightmare.

(18:40):
But the goal in Ariane wasalways to bring Europe together
and not necessarily to build thecheapest rocket.
But Ariane 4 was actually agood rocket, It flew
commercially a lot ofgeostationary satellites.
It was also almost optimizedfor that particular trajectory.
It launched from Karoo at fivedegrees southern latitude, I

(19:03):
think.
So you basically go straightinto the equatorial plane where
you actually want to be for ageostationary satellite.
So that's a couple of thingsthat worked really well in
Ariane.
They were relatively fast.
cheap at the time, although itwas still expensive.
And they had a really good runwith Ariane 4.
Then they replaced it withAriane 5.

(19:24):
And then they made, in myopinion, the mistake to replace
it with another vehicle withoutany significant benefits.
And by that time, SpaceX wasalready up and running on Falcon
9 and basically killed themarket for the rest of the team.

Speaker 03 (19:42):
You guys built...
First three Falcon 1s fail.
I'm assuming the third one wasa pretty depressing day at the
company.

Speaker 00 (19:50):
Not to the first.
The first that failed waspretty depressing.
Really?
That was the most depressing?
That's the one that shocked methe most.
Really.
I mean, it's just like the wayit was.
We were pretty convinced thiswill work, at least mostly.
And seeing it fail like within30 seconds and then also picking
up the debris.
is not motivating.

(20:10):
That's actually not...
The other two went out ofsight, so that's kind of like
great.
So it's like, I mean,depressing too, but it's not
anywhere as depressing asactually picking up a whole
rocket in pieces.
So that depressed me at leastthe most.

(20:33):
And...
So the third one did a reallystupid thing.
We changed the engine and theengine shutdown was different,
but the stage timing was thesame.
So we basically re-ended thesecond stage after separation
because the engine just took awhile to turn off.

(20:55):
Oh, interesting.
So it was just a timingmismatch.
Nothing necessarily went wrong.
No.
And all we changed then is likewe changed the timing.
Yeah.
Yeah.
No, it's a new config file.
Wow.
And so by that time, we builttwo, almost two.
One was a little bit lagging,but so we had another rocket and

(21:17):
we launched within six weeks.
I think the goal, Elon told usto launch within four weeks, and
I think it took us seven orsomething like that, yeah.
And that was actually, and thenthe other thing that was super
ironic is like on all the othersatellites, early rockets, we
had satellites.
I mean, not the most advancedscience satellites from NASA,

(21:38):
but we had satellites thatpeople worked on for a long
time.
The Air Force Academy had oneon the first one that also was
aptly named FalconSat.
And that got like a mile up andthen down again.
And then we had some otherexperiments on number two, and
then we had another satellite, Ithink, on number three, which I

(22:00):
forgot.
I think it was taxed.
But anyways, on number four, wedecided no more satellites.
We're done with that.
We put a big mass on there, andthen that thing is over.
It's still up there.

Speaker 02 (22:19):
That's awesome.

Speaker 03 (22:19):
How do you think about risk, though?
When you go through all these,every one of these failed for a
different reason.
As you went through, you're incharge of risk at the company.
How do you go to Elon at thatpoint and say like, oh, I
promise you, we solved thisengine timing.
We solved the timing.
We got configs aligned.
We got this.
We're not going to screw up thefourth one.

Speaker 00 (22:38):
So it was different.
I actually, I worked a littlebit on the anomaly
investigation.
And being avionics, you're kindof in the center of this thing.
So I was interested in theresolution avionics.
So the failure investigationthing.
And I kind of ended up doingmanaging, we started doing a

(23:01):
list with risks at the time.
So I wasn't in charge of therisk, but we just started doing
a list.
And so I kind of kept onworking on that list too.
It was, I mean, it sounds likea...
a great thing, but it was justan Excel spreadsheet, of course.
This sounds so much moreimportant, though.
Exactly, yeah.
So we always had a list ofconcerns early on, and that

(23:25):
turned into a risk list,basically.
And then we went more and moreformal as time went on.
But we didn't really have arisk review or anything before
launch formally.
We had everybody try to aroundbasically what what can possibly
go wrong after a while after welearned that you know that

(23:46):
could be a good thing like canwe do anything else to make it
safer and more reliably yeahyeah um and we kind of learned
that basically the hard way thatthis is a good thing to do um
because then um after falconfalcon one fundamentally um was
too small for the market or themarket wants bigger rockets yeah
um and we we were We werelooking at cargo to the station,

(24:10):
which needed a dragon,basically.
And then you're now in 15,20-ton payloads with a rocket.
And so that's how Falcon 9 gotdeveloped.
And Falcon 9 went orbital onthe first flight.
I was really shocked by that,too.

Speaker 03 (24:29):
How long was the iteration from Falcon 1 Flight 4
to Falcon 9 Flight 1?

Speaker 00 (24:35):
Um...
I'm trying to remember.
I think it was 2010 when weflew Falcon 9.
It was June, I think.
Well, I need to look it up inWikipedia.
Anyways, I think it was.
And then the last flight onFalcon 1 was 2008.
We were successful.

(24:56):
2009 was the last flight.
So that was a year in between.
That's partly the reason why Iwasn't that much involved on
Falcon 9 because I finishedFalcon 1.
And I lost a little bit likethe train.
So that's part of the reasonwhy I didn't continue in
avionics for a long time,because I just missed the next
project, basically.

Speaker 03 (25:16):
Gotcha.
Now, I remember these comingout a long time ago, and I don't
think it was ever developed.
Whatever happened to Falcon 5?
Or

Speaker 00 (25:23):
was that what it was called?
It was just on paper.

Speaker 03 (25:25):
And no development was ever really spent on that

Speaker 00 (25:27):
rocket class?
Correct.
I mean, the diameter betweenFalcon 5 and Falcon 9 is the
same.
It's just the number of enginesis different.
Falcon 5, I think, was inspiredby Saturn 5.
Yeah, you can see, oh, we cando that, yeah.
And then along the way, we fellshort in performance, right?
And then we realized we needmore engine thrust, and then we

(25:48):
just filled the pattern, and nowyou had 9.

Speaker 02 (25:52):
Yeah.
Falcon 9 is almost 15 years agoto the day, June 4th, 2010.
Yeah.

Speaker 00 (25:58):
Oh, 2010, yeah.

Speaker 03 (26:00):
Wow, okay.
So you are finishing out Falcon1 program.
Falcon 9's already started.
And then you transitioned overto, obviously, Falcon 9 and Crew
Dragon are the only thingsgoing on at SpaceX at the time.
How

Speaker 00 (26:11):
did your role shift with the company?
Falcon 9 and Cargo Dragon, notCrew Dragon.
Cargo Dragon.
Very different.
Okay.
Because, I mean, yes, they lookthe same to some extent, yeah,
but Cargo Dragon attaches tothe...
It births...
Births...
It doesn't dock, it berths.

(26:32):
And what the difference is, youbasically fly next to the
station, the station grapplesyou with the grapple, and then
moves you to the door.
And the door is not the dockingport, it's the berthing port.
And it's also square, whichallows you to get like a
mid-sized piano to the station,versus the docking port is round

(26:54):
and the piano has to stayoutside.
So it's bigger.
Yes, exactly.
And you can actually– it'simportant for certain, I guess,
equipment that they can getthrough the building door and
not through the docking door.
The docking door is more likefor humans.
I want to say it's like twofeet and a little bit, two and a
half maybe.
Okay.

(27:14):
Anyways, it requires a lot lessfrom Dragon.
And the action is mostly on thestation to grapple.
And I was always– I mean– Thisis history to some extent
because we don't do thatanymore.
We now use the docking port foreverything.
And Crew Dragon and CargoDragons are basically the same

(27:35):
vehicle, just somewhat modified.
But it was the easier way todock or berth with the station
originally.
So that's why we picked thatway.

Speaker 03 (27:47):
Were you working reliability on...
Cargo Dragon?

Speaker 00 (27:50):
No, it was actually Guidance and Control and Flight
Safety at the time.
So I gave up on the electronic.
And that's actually fairbecause I'm not in EE.
I do a little bit.
I can probably get something torun.
But I'm not a designer.
I don't know the morecomplicated digital designs.
I'm not good at that.

(28:11):
I'm more like a systems person,actually.
I can put stuff together.
I can duct tape it together andzip tie it together and make it
work, which is exactly what youneeded for Falcon 1, right?
But for Falcon 9, you neededmore.
This is interesting in a waybecause companies change, yeah?
And then SpaceX, which had alot of universally talented

(28:32):
people, now moved into peoplethat were specialized.
They had their...
The Ws were specialized.
They had...
different software departmentsfor the code that runs the
guidance and control and thecode that runs on the radio.
And so you have suddenly youbuild like silos and niche and

(28:54):
have specialized people andcommunication gets harder and
it's all part of the growth onthe company, right?
Yeah.
And my talent was actually morelike on a broader scale.
I know how most of the stuffworks, but I'm not a specialist
in that sense.
And so I moved away a littlebit.

(29:14):
I continued to work on flightsafety because that was critical
for Falcon 9.
And we actually, I don't wantto say we messed it up, but we
delayed the launch by two weeks,something like that.
Mostly because this is likework with the regulatory
agencies.
And you have two, we have theFAA and the range.
And both the range is tough.

(29:36):
The FAA...
It's confused often.
What range?
It was the 45th?
It was the 45th Space Wing atthe time.
Okay.
I mean, they're still calledthe 45th, but there's Space
Force now.
Yeah.
Oh, gotcha.
Yeah.
And honestly, over the years,it was really good to work with
the 45th Space Wing and therange because this is all about

(29:57):
trust to some extent, right?
You're the new person on therange.
They don't trust you.
And so you had to work on thetrust that you actually do all
the tests and that you knowactually your stuff.
And so over the years, wegained the trust and it became
easier and easier to work withthe range.
So there

Speaker 03 (30:16):
was a Falcon 9 that blew up on the launch pad,
right?
That

Speaker 00 (30:19):
was way later, right?
There was one anomaly beforethat, actually, talking about
reliability.
There was one on flight 20 or19, depending on how you count,
that...
There's a COPV in the secondstage, and one of them actually
ripped off the wall and thendestroyed the second stage dome

(30:44):
and then destroyed the wholevehicle in flight.
This was a cargo flight, CR7,or I think we call it still
flight 28.
The thing was, flight 19 wasalready given to another rocket.
So it technically is 19, Ibelieve, or something like that.
Rocket numbering is difficult.
So anyways, it blew upbasically 140 seconds into

(31:10):
flight.
And I was actually on console.
So I looked down for a second.
I looked up and I see just likeglitter in the air and go,
where's the rocket?
And it's just like...
It takes a while to sink in,and it takes more while to say,
I think we lost the rocket, andto say this loud on the comm.

(31:32):
And it takes a while.
And I kicked off a prettyintensive anomaly investigation
We actually rented submarines,tried to find the debris.
We found a lot of debris, butnot from us.
I mean, this was thetraditional path that shuttle
and also Apollo took, right?

(31:55):
So there was some hardware.
There we go.
Let's move on.

Speaker 03 (31:58):
Could have sold some of those Falcon 1 engines or
the F1 engine to Bezos.

Speaker 00 (32:05):
I mean, yeah, actually, the first Falcon 9s,
We entered the ocean.
If you go there, you probablyfind them, similar to like
Southern Five.
And they're all probably withinthe vicinity of where usually
staging happens.
I mean, the plus trajectory.
Yeah.
So that was actually a prettymiserable time, too.

(32:26):
We found what it was.
It was a rod end that broke.
There's more to it.
I mean, wide breaks and so onand so forth.
But it was fixable.
And then we moved on.
Along the way, we alsoironically lost one of the
docking ports through thestation.
There were two payloads onthat, on CS-7.

(32:48):
One was a docking port, kind ofimportant for us, actually.
And then a spacesuit.
It was kind of important, too.
And one of the last, actually.
And so it was a really, reallyimportant cargo that we lost
with that flight, unfortunately.

Unknown (33:05):
Yeah.

Speaker 00 (33:06):
And it was also the first anomaly on Falcon 9.
The first, sorry, completeanomaly.
There was an anomaly beforewhere we lost an engine.
The engines were built througha process at the time.
It's different now, but at thetime it was electroplating and

(33:26):
it's sensitive to...
to contamination so the platingcan actually break open.
That's actually how many peoplebuild rocket engines.
It's a touchy process, and ittakes time.
Nowadays, it's been completelyovertaken by 3D printing.
This was the old

Speaker 03 (33:46):
process where you mill out the cooling chambers
and then put the wax in and thenelectroplate over them

Speaker 00 (33:50):
to get the walls so you have the channels.
Exactly.
You grow it from the outside.
And if you contaminate thatstuff where you grow it from,
then you end up with fractures.
And then over time, you crackthe outer hole.
And then you spray kerosene tothe side and bad things happen.

Speaker 03 (34:06):
What iteration of Merlin was this at the time?

Speaker 00 (34:11):
At least one, two, three, I think, number three,
something like that.
So the first one was like onFalcon 1, right?
And then Falcon 1 had a newerversion, which is the one that
we messed up the stageseparation with.
And then I think we had a newerversion for Falcon 9, more

(34:32):
thrust.
And then there were at leasttwo more major versions coming
after that.
Yeah.
Yeah.
Anyways, so having the firstanomaly in Flight 20 showed me
this is different.
Totally different problem.
It's like a manufacturingproblem, a testing problem along

(34:57):
the way as you build it and notnecessarily doing something
wrong.
On Falcon 1, we had what I callthe stupid mistakes.
You did something wrong and,yeah, okay, you fix it and then
you move on.
Yeah.
But then later on, as you havea vehicle that's been produced
regularly, you run into thisproduction kind of things.

(35:17):
And I think by that time, Itook over flat reliability and
built reliability.
It was not my intention, to behonest.
Flat reliability, I...
Why did I take that?
I think...
My predecessor left SpaceX andI was the next logical person in

(35:41):
that.
And then flat reliability alsois a...
Different word for missionassurance.
And I always struggled withthat because mission assurance,
I can't quite describe what itactually does.
It's difficult.
I mean, yeah, there's like adouble second look on the design

(36:01):
and whatever, the whole processand so on.
But I always wanted to do moreactive things.
So what I also did is I did thelaunch chief engineering, which
is actually...
Being there when you launch it,checking the vehicle, going
through everything that's notright on the vehicle, and then
actually working with theoperations to make sure that we

(36:21):
don't make mistakes in theoperations.

Speaker 03 (36:24):
But Hans, you guys did actually something really
impressive here that I think alot of other companies have
failed at.
It's really easy to say, oh, wehad a bad strut manufactured.
Let's add a whole bunch ofprocess and checks and just
continuously escalate this tillyou end up with unlimited
process and checks.
And now the vehicle costs halfa billion dollars.
How did you think about notscrewing that up, but also not

(36:45):
making the vehicle cost 10x asmuch to not have a mistake
there?

Speaker 00 (36:51):
It's a good question.
So when you do something likethat, you find out, oh, okay,
there was this one test that wedid wrong.
But then obviously you don'tstop there.
But you go, where else on thisvehicle do we do the same thing,
right?
And now you end up with like,whatever, 10 more places.
And that's where you escalatethe whole situation.
I think we were just...
reasonable I don't think therewas any magic to be honest I

(37:16):
think we just were reasonableand to be fair the reason Falcon
9 is so cheap now has nothingto do with the way we build it
but the fact that it reuses thefirst stage and the fairings
over and over that's justfundamentally the key thing that
makes Falcon 9 cheap these daysyou can use the first stage 28

(37:39):
times so far Maybe it's 29.
I didn't pay attentionyesterday.
Unlimited.
I don't think it's unlimited.
I think you do end up with likecertain fatigue issues and like
engines get old and stuff likethat.
Turbo farms get old.
So I think there's definitelysome runtime issues.
But anyways, I mean, even 30times is like you build the

(38:03):
first stage for a 30th of theprice, right?
And instead of building 150 forthe year, you end up with like
A handful.
But

Speaker 03 (38:11):
that's not entirely true because there's this
spacecraft before or rocketbefore called the Space Shuttle.
And it's like, it still becamemore and more expensive after
every flight.
So how did you, you know, onthat note, how do you hold
reliability but not make itcost, you know, Space Shuttle's
last flight was $1.5 billion torefurbish.

Speaker 00 (38:27):
So part of that was also the standing army and the
raid, right?
And And then the refurbishmenton Space Shuttle was just very
elaborate.
It took apart the engines, tookapart the turbo pumps, versus
SpaceX basically just inspectskey points and makes sure you
don't push it too hard.

(38:48):
So I think I try to answer thatfor myself.
Why did the shuttle fail onthat, and why did SpaceX succeed
on reusability?
And I think part of the...
The reason is that Falcon 9 wasnot always human rated.
So you could actually makechanges.

(39:08):
So there's two versions,basically.
The human rated one and there'sthe other one.
And you can make changes on theother one and then phase it
into the human rated one.
After you say, okay, this hasflown five times.
We know this works.
We're going to now fly thiswith astronauts too.
And the shuttle could not dothat.
So in all fairness, the shuttlecould never risk something.
I mean, they took a lot ofrisk.

(39:29):
The first shuttle flight, Imean, when you read about that,
it just feels like, wow, thefirst flight, there's two people
on there.
They basically have very littlechances to get out.
That's pretty gutsy.
I mean, kudos to NASA to dothat, actually.
And then they

Speaker 03 (39:46):
went with stolen electronics.

Speaker 00 (39:50):
Right.
And so all of this was prettygutsy.
And then I think they justcould not get...
bring the nerve to continuethis risk over the long time.
So how

Speaker 03 (40:02):
did you guys do it?
Because, you know, Dragon, Imean, not Dragon, I'm sorry,
Falcon has flown a crewedvehicle before, right?
You have flown humans on it

Speaker 00 (40:08):
before.
Yeah, but like I said, we justfly the tests all on Starlink.
And if it goes wrong, then youdon't make that change.
So there was more to it too,right?
I mean, we had...
um, we knew how to makechanges.
We had a change process.
We were very careful in puttingchanges into the, um, flight
line.
Um, but, but yeah,fundamentally we, um, the

(40:29):
Starlings were basically, Idon't want to say they're goody
pigs, but yeah.

Speaker 03 (40:35):
Were you nervous when you launched it, when you
launched the crewed vehiclethough?
I mean, you were flightreliability.
You knew all the gremlins onevery spacecraft, right?

Speaker 00 (40:41):
I wasn't, I wasn't.
So I was there in the monthbefore.
Um, I, I, um, I actually wasnot chief engineer for that
launch.
I was chief engineer for thefirst commercial crew.
I gave this to a good friend,and actually Bala Mamoudi was
the chief engineer for that.

(41:01):
And he worked very much withthat particular mission for a
long time, so I didn't want tointerfere with that, because I
worked on multiple missions, andhe only worked on this one, so
clearly he should be the chiefengineer.
But I was still there.
Somewhat nervous because thingsdidn't quite go as they should
have been.
But, you know, there's alwaysstuff that's leaking here,

(41:23):
leaking there.
Also, it was not 100% healthyrocket?
No launch is 100%.
There's always something onlaunches usually.
I mean, at least by the time Ifinish.
Maybe now they're perfect.
But there's always like littlethings.
Always like oxygen leaks andkerosene leaks, whatever.

(41:43):
But where was I?
All right, exactly.
I actually found the picturesbecause this was also like five
years, a couple of days ago,right?
On March, on May 30.
And I took a picture of myheart rate on that and it's 63.
So I figured, okay, I guesswhen you launch 100 rockets, you

(42:06):
kind of get used to it.
I don't

Speaker 03 (42:09):
know.
How was, so there was anotherbig development at SpaceX,
right?
in your tenure there.
And that was obviously thelanding of Falcon.

Speaker 00 (42:16):
Correct.

Speaker 03 (42:17):
How did you, was this closer to a Falcon one
development or was this closer?
Cause you guys took a lot ofrisk in the early days of
landing.
I mean, I saw that thing comein sideways and hit the drone
ship and it was, it was reallycool to see very early on.
How'd you think about it?

Speaker 00 (42:29):
Um, so it was not, I think there was risk for the
vehicle itself.
And the biggest concern for mewas that we sink the barge.
I mean, sorry, the drone ship,um, and basically punch a hole
in it.
We had to explain this tothe...
Those were rented ships.

(42:49):
You didn't own the ships?
You rented them?
By that time, yes.
Sorry, we put a hole in it.
We did punch a hole in there, acouple, but we never punched it
through.
Okay.
As far as I remember, at least.
So we had Grasshopper as aprogram to develop this on the

(43:12):
site.
Grasshopper was just awesome.
It was like a dozen people ledby Shana Diaz, and they did a
great job just hopping with it.
And there's still othercountries trying to mimic this,
right?
There's one company in Chinathat just recently did that
successfully, actually.

(43:32):
But Grasshopper flew...
Many times.
I'm not totally sure how manytimes.
So they were basically ourskunkworks within SpaceX to
solve the problems, how to land,and so on and so forth.
And then we transported thattechnology into the normal first
stages.
And it didn't quite go as well,partly also because we tried

(43:54):
this out in the ocean.
We first tried to land on thewater.
That worked a couple of times,but it's undecided at the end
whether this was good or bad,because they tend to fall over
and then they blow up,underwater mostly.
And then we put the barges anddrone ships there, and that

(44:21):
didn't go well the first coupleof times.
partly because the crosswindwas too strong or partly because
the sensor didn't quite workwith it and so on and so forth.
There's multiple issues that wehad to solve.
So I think we crash landed, Iwant to say a handful.
Then we also blew up F9R, whichwas the successor of
Grasshopper, now with threeengines.

(44:43):
And we just got a little bitcocky on the flight profile and
that make it tilt over and thenbad things happened.
But so the interesting thingwas the first one that landed
was actually the first flightafter the Flight 20 anomaly.
So here we are.

(45:03):
We did an anomaly, which, bythe way, happened on Elon's
birthday on June 28.
So not a great day at all.
Boss has birthday and we blowup the first Falcon 9.
And so then we're trying torepair this, basically, trying
to figure out what it is.
And then on that next flight,we actually decide to land on

(45:25):
land.
And landing on land soundsrelatively easy, but there's
some subtleties.
The rocket basically flies nowagainst the Cape.
This is the first rocket thatactually comes at the Cape and
not from the Cape.
And there's a cruise shipharbor behind the Cape.
So you've got to make sure thatit does not overstep the cape.

(45:46):
It's got to end on the landingside and not go any further.
If it goes any further, it'sgoing to be blown up.
So there's some added risk whenyou land on land.
And I thought it was reallygutsy to do that.
And then, as always, we end uppushed against Christmas
because...

(46:06):
Rockets are holiday-seekingmissiles, right?
Yes.
At the end of the day, right?
They tend to go onThanksgiving, and this one, the
schedule was such, it was likeone day left at the range where
they have closed for theholidays, basically.
And so, of course, like thehuge pressure, and then we

(46:27):
launched it, and it actuallyworks out great.
I mean, both works.
The mission itself was OpCom,worked great upcoming at all the
satellites now in orbit andthen coming back on the first
landing were great um it lookedperfect actually from what i
recall there was really nothingnothing wrong with that we um we

(46:47):
couldn't wait to get out tocheck the booster i know we were
just pacing up and down untilit was safe and then we just
drove out and um and checked thebooster There's a lot of happy
people around that boosterduring the rocket dance.

Speaker 03 (47:01):
Why do you think none of the other...
So obviously, there's a cost tolanding a rocket, right?
You take a fuel hit.
You have to hold more

Speaker 00 (47:07):
fuel on the vehicle to come back and land.
Especially when you come backto land, you're going the wrong
way.
You've got to twist that vectoraround.
And then you've got to...
There's a braking burn on theentry, and then there's a
landing burn.

Speaker 03 (47:20):
How much of the fuel on stage one are you reserving
for the landing?

Speaker 00 (47:22):
I don't know that number, actually.
Okay.
That sounds really terrible,but I...
I could probably find it out,but it varies, actually.
What works for you is the factthat you're pretty light.
The first stage is basically95% of 90-something percent
fuel.

(47:43):
So you're now down to reallylow fuel.
And that's why you can barelyland on one engine.
The engine is throttled, butit's still too much thrust.
Yeah, I got you.

Speaker 03 (47:54):
Why do you think nobody...
previous had tried this.
It wasn't some technologyadvancement, as far as I'm
concerned, that allowed us to dothis.
Why do you think Delta or Atlasor the Russians never looked at
re-landing?

Speaker 00 (48:05):
I don't know, actually.
To me, this...
Okay, you could speculate thereare a couple of reasons, right?
I mean, because the businessmodel works pretty well with the
reusable rockets.
Why mess with that?
I mean...
you get paid 200 million.
Why would you endanger that?
And then suddenly get paid onlyfor the second stage or

(48:25):
something.
So you think it's just

Speaker 03 (48:27):
incentive alignment was off for

Speaker 00 (48:29):
the government contractors?
Correct.
Incentive alignment was off,obviously.
And this works.
I mean, they don't thinkcommercially.
They think, in that case, likea government contractor that
works government-specificthings.
Another reason is if you don'tneed so many rockets, it's like,
what do you do with yourpeople?
I heard that argument.
Build Starship.

(48:50):
Exactly.
Go to the next rocket, buildmore rockets, whatever.
I mean, there's many, manythings you can do.
But I heard that argumentbefore.
And then I heard also it'stechnically not feasible.
There's many people that haveopinions on that.
It's actually on videos, some,and they didn't age well.
I mean, those videos, right?

(49:11):
They're pretty hilarious.
They're also with thisundertone of arrogance.
All these guys, I mean, yeah,they're new.
They're nice.

Speaker 03 (49:21):
That was very much the aerospace environment pre, I
would say, 2010, pre-Falcon 9was that.
I mean, SpaceX was not lookedat like it is today.
Let's be honest about it.
It's a very different company.
What do you think about,obviously, you're not at SpaceX
anymore.
How much of Starship programdid you work on?
And what do you think about,obviously it hasn't gone as well
as we would have liked to seeso far.

(49:41):
What do you think we need to doon Starship?

Speaker 00 (49:44):
So I worked very little on Starship to be fair.
I did a little bit on theregulatory work, which didn't go
great in the beginning.
And on the location, I workedon location actually.
And yeah, I personally feel thelast launch should have been
better.
Honestly, I think I know it's anew thing or rather an old

(50:08):
thing.
The other two problems, the twoflights before and flight seven
and eight, they got fixed.
But I think they need to getbeyond just fixing what failed
on the last one and actuallymaking the whole vehicle more
reliable.
It's a complex vehicle on theother side.
And it's methane is a finickygas.

(50:31):
If it leaks, it blows up.
So it's definitely a difficultdevelopment.
But I don't know.
I feel like the next launchshould really go well.
I hope so.
I think we're all hoping

Speaker 03 (50:46):
so.
It's such a big, cool vehicleto see.
It's almost mind-numbing to seethe size of that thing when it
goes.
That's true.
Hans, how do you take whatyou've learned over your tenure
at SpaceX on all these differentprograms and How do you help us
at Astroforge make sure that wedon't repeat the mistakes of
the past, but we also, you know,do it in a speedy way, at a

(51:08):
cost-effective way that can makesense?

Speaker 00 (51:10):
Yeah, I can.
So I think there's adifferent...
So I can definitely beparanoid.
I can help you with that.
And I think you have to beparanoid to some extent, right?
The...
Most mistakes we made wasbecause somebody thought this is
okay.
They thought about it brieflyand then kind of thought on

(51:32):
other things instead of justreally thinking a little bit
deeper on that.
So I can bring that flavor.
And my gut feeling is alsoreally simplifying it.
Rockets and interplanetaryspacecraft are not that
different, actually.
So satellites in general have alittle bit more, they're more

(51:52):
benign in a way.
They don't tend to blow up ifyou blink for a second.
They tend to, things fail, butthen you find workarounds in
flight often.
And I've always felt like asatellite is easier to make
reliable than a rocket.
I mean...
I think I can stand up for thatstatement.

(52:13):
Interplanetary, however, ismore difficult in a way because
you go away from the Earth andwhatever problem you have, your
time is running out.
You have to fix it and the oddsof making contact are getting
less and less and less.
So there's an element here thatkind of works against you.
It has to work from the get-goand then if you have a reliable

(52:34):
attitude, reliablecommunication, reliable power,
then you can work otherproblems, but there's not a lot
that can fail them.
These three things have towork.
And you don't have like an omniantenna pattern where you can
tumble and get the signal,right?
No, you must remain stable.

Speaker 01 (52:52):
Reliability

Speaker 02 (52:54):
is pretty simple from a concept.
It has to work.
How do you, especially at acompany like ours, how do you
make reliability a discipline ora practice here?

Speaker 00 (53:04):
There's certain analysis you can do and I don't
really believe a lot in aprobabilistic approach on that,
like trying to figure out fairrates.
Because the fair rates you haveare typically because something
went wrong or something gotdone wrong.
So you can't put that in anintegration really, right?

(53:24):
Electronics is really superreliable if you get everything
right.
And then it might actually...
Finally, when you produce likea million laptops, you might
actually get some theories thatyou can use analytically.
But that's not what we havehere.
We have a thing that we buildtogether once or twice.
And so the key thing for mealways is testing it, right?

(53:46):
Testing it as long, as hard,and as creatively as you can
possibly do.
And then really try to findwhat's going on.
That's a different mindset.
It's hard to...
So you're approaching now yourdesign with the mindset of what
can possibly go wrong, right?
And most people are notdesigned for that, right?

(54:09):
They're designed, they're proudof it, and they think this can
never fail, right?
And it does take a little bitof a personality change.
Maybe it helps if you haveanother person work on that,
then what can possibly go wrong,yeah?
Or somebody who is more like asystems knowledge or something
like that.
But, yeah, or just somebody whois like, I'm easily suspicious

(54:35):
and I have a good bullshitmeter.
And so somebody who is a littlebit suspicious, basically,
right?
But it's not necessarily whatpeople's personality goes the
other way in many cases, yeah.

Unknown (54:51):
Yeah.

Speaker 03 (54:52):
Awesome.
Well, thanks for spending timewith us, Hans.
I appreciate it.
It's amazing how much betterthese are without Robyn.

Speaker 00 (54:58):
So much better.
I guess one last question.
You spent 20 years buildingReliability.
It will probably be one of themost important space companies
that's ever existed.
Why'd you come hang out withus?
Good question.
No.
No, okay.
So, okay.
I think actually the satellitemarket did not keep up with the

(55:19):
rocket market.
Yeah, that's just the way itis.
And I feel like it can be moreaudacious.
It can be more braver projects.
And I think that's what you do,right, at the end of the day.
That's why I thought, okay, Ican dare something here.
I want to do that because ifyou don't dare anything on the

(55:40):
satellite market like we did onthe rocket market, the rockets
need daring satellites and notjust daring rockets.
So that's part of it,obviously.
Yeah.
And then I think you have apretty cool company, and it's a
great place to work.
I like it here.

Speaker 03 (55:55):
Awesome.
Well, thanks again for spendingtime with us, Hans.
I appreciate it.
You're welcome.

Speaker 00 (55:59):
Yeah.
Hans, this is great.
Thank you very much for joiningus for this.
And that's the end of thisepisode.

Speaker 03 (56:03):
Cool.
What episode was that, by theway?

Speaker 02 (56:05):
This is

Speaker 03 (56:06):
15.
We've made it 15 episodes.
Oh, my God.
Awesome.
Thank

Speaker 02 (56:12):
you, Hans.
Thank you, Hans.
This was awesome.
Bye.