That was exciting and a proud moment in time to be repeated on Saturday I’ll be watching! 
In the meantime who’s up for a spin on the Vomatron at Screamer’s Park Daytona Beach? Bob& Doug? 
Yes. Eastern Time.
There is plenty of data (135 flights with 2 catastrophic failiures and 1 Abort-To-Orbit) to make an informed posterior assessment of estimated reliability of the Space Transportation System (STS) launch vehicle of 0.971. I would argue that the Abort-To-Orbit of STS-51-F as a ‘success’ in terms of returning the crew unharmed, which bumps it up to 0.978 which is a rounding error of 0.98, but a failure is a failure. Falcon 9v1.2 (which should be considered separately in statistics from previous Falcon 9 family because of significant design changes) has 63 flights and no failures for an estimated reliability of 0.985. However, were it to experience a failure on this launch (or a hypothetical failure on any previous launch) the estimated reliability would be 0.967. Given that the Falcon 9 has less than half the launches of the STS, it is fair to question whether the confidence in the Falcon 9v1.2 vehicle is the same as the STS and look at the Adjusted Wald 95% confidence interval, which is 93% to 100% for both vehicles. The “1 failure in 10,000 launches” is an anecdote that came from Richard Feynman’s solicitation of that question during the Rogers Commission investigation of the Challenger failure, and it was provided by one manager. Other NASA engineerings hazarded guesses of 1 (failure) in 500 (launches) to 1 in 1000 (and one reportedly wrote “99 44/100% pure”, so I guess he was an Ivory soap fan). The actual system engineering estimates predicted a failure rate (not necessarily catastrophic) of between 1 in 50 and 1 in 100, which is actually just about right.
All of this is true, and is a general argument for capsule-type vehicles for personnel rather that “spaceplanes”. However, the odds of surviving any catastrophic loss of vehicle unharmed are not really great regardless of the abort system configuration. If the launch vehicle fails on the pad, the likelihood of a successful ‘pad abort’ maneuver is not great, and there are any number of ways that launch abort systems can fail, even partially, resulting in loss of crew. I would not be sanguine about declaring the Dragon capsule so much safer without seeing actual evidence, even though I do think the design of the launch abort system is preferred to the abort tower ‘tractor motor’ to be used on the CST-100. The criticisms of the STS abort modes are on point, but again, that is a limitation of that design choice that I hope no one will make in the future.
At that azimuth they don’t have much of a choice. However, I’d far rather launch in a snowstorm–which aside from the effects of the ambient cold air on systems doesn’t really pose much of a threat–rather than lightning which can disable electronics or even damage the vehicle structure. The reality is that weather (and wind) is a factor about rocket vehicle launch is that there is no control over and can pose a significant hazard to flight. Scrubbing was the correct decision (and mandated by the range) even though the odds were that the launch would not be impacted. In launch operations, you can scrub and launch another day, but you cannot recover and relaunch a crashed vehicle or put a damaged payload back together.
Stranger
That’s the most likely value, but there’s a wide range of reasonable values. For a real chance of failure of 0.28%, there are 5% odds at seeing 2 failures out of 135 (binomial distribution). And for a real value of 4.9%, there are also 5% odds at seeing 2 failures out of 135.
I don’t think it’s useful to quibble about the exact numbers here, but going purely from the observed failure rate, there’s a pretty broad range of possible real failure rates that would lead to that outcome.
You can do better with a more detailed analysis, but the Shuttle is a very complicated system. I think that a 1-2% failure rate is about the right range, and tightening those bounds is going to be very tough. And until we see a launcher with hundreds of consecutive success, I think we’ll be hard pressed to say that *any *launch system has less than about a 1% failure rate.
As an aside, lightning was the cause of the famous try SCE to aux incident that earned John Aaron the title of steely-eyed missile man. It also caused NASA to consider the possibility of lightning strike in their launch criteria. It’s by no means a theoretical problem.
From an architecture standpoint, the STS isn’t really more complicated than a nominal stacked cylindrical launch vehicle; you have the main propulsion system consisting of three RS-25 Space Shuttle Main Engines (SSME), and then the two Reusable Solid Rocket Motors (RSRM) as equivalent to side cores or (very large) strap on boosters and they have actually been the most reliable part of the ascent system from a propulsion standpoint. What makes it complicated are the variety of complex interactions that can or do occur because of the specific configuration between the Orbital Vehicle (OV), External Tank (ET), and the RSRMs, which is what led to the STS-51-L failure; had the leak not been at that particular orientation on the field joint, the vehicle would have made it into orbit just fine. Ditto for the buildup of ice that caused damage to the reinforced carbon-carbon wing caps on the OV. A stacked vehicle doesn’t have these kinds of interactions but it also cannot do the things the Shuttle could (hypothetically) do, like a once-around polar orbit return to launch site trajectory or returning a large payload from orbit. That it rarely or never did these things really indicates that the Shuttle program was manipulated in favor of political interests rather than rigorous top level mission or system requirements, but I think that was well understood even before the STS started developing problems.
The Falcon 9v1.2 is also a “very complicated system”, albeit more from the standpoint of needing to use cyrogenically densified propellants to get the necessary performance. I’m still not sure how NASA got comfortable with that but I worked on a study for that on a different launch system and we found that the liabilities in terms of increase in hazards was not worth the benefit. SpaceX obviously had a different conclusion, and aside from their one on-pad blowup (which is not considered as a launch failure and therefore don’t get counted against SpaceX even though it should be factored into reliability) they’ve made it work although it narrows their launch window considerably. However, it was done as a way to extract a few percent more total impulse rather than increasing the vehicle diameter, which from a mass ratio standpoint would actually be preferable but introduce logistical complexities into their established launch infrastructure.
I think it is possible to develop a much more reliable and robust system, albeit by trading specific performance for simplicity and utilizing some technologies that have been discounted in the past for being either too esoteric or just too far away from the norm of launch vehicle design and operations. However, I would agree that getting above 99% reliability in the field of space launch is a challenging threshold just because of the sheer number of things that all have to go right for success.
Stranger
The meeting notes where the crucial decision was made (by the Aerospace Safety Advisory Panel) are available here. The relevant part:
As of yet, I have been unable to find the NESC report they are referring to (not yet finding anything via the NASA Technical Reports Server).
Question: Why exactly would poor weather scrub a launch? Doesn’t the rocket basically blast through the clouds and reach the weather-free high atmosphere within seconds?
There is the aforementioned lightning. Not just existing lightning strikes, but the exhaust of the rocket can itself induce a lightning strike, so they have to consider the ambient voltage gradient. Lightning can damage electronics and structure.
The Falcon 9 in particular is a slender rocket, and has tighter wind shear criteria than typical. Passing through a sharp difference in wind velocity causes the rocket to bend, and too much of that could lead to structural failure.
And obviously, there’s going to be some limit to ground-level winds. You don’t want winds blowing the rocket into the tower.
The rocket does not reach above the atmosphere in seconds. It takes over a minute to reach “max Q”, which is the point where the rocket experiences the maximum aerodynamic forces. It takes a while to reach the max because while the air thins as the rocket rises, the speed increases, and so there is some part of the flight where these two effect combine to reach a maximum.
The jettison of the fairing (the protective shroud that covers the satellite) happens at around 3.5 minutes. The atmosphere is basically negligible at this point. But it’s certainly not seconds.
Manned launches also have to consider the splashdown point in case of abort. And the Falcon 9’s booster lands (either on a barge or a landing pad), and the weather there is a consideration as well.
No. First of all, it takes several seconds for the vehicle to even lift above the launch gantry, and the “high atmosphere” of the stratosphere is about 60 kft (~10 miles) up. Until it gets to there, it is subject to phenomena like lightning, dense hail, and wind shear, all of which can damage the vehicle or render it uncontrollable. Dr. Strangelove noted the Apollo 12 anomaly that shut down the flight computer, and other problems have been seen (the Challenger failure was in part attributed to the beyond 3 sigma wind shear seen on launch day although I’m personally not convinced the burn thru on the O-ring wouldn’t have occurred anyway). You certainly don’t want a vehicle in flight to be struck by lightning and unintentionally activate the flight termination system as unlikely as that may be.
There are launch vehicles designed to fly through inclement conditions; they’re called ICBMs and they have particular provisions to harden them against such conditions albeit more to fly through a close proximity nuclear blast than mere precipitation, and because they don’t have to carry a delicate human payload they can be designed to accelerate fast and fly through any threat more quickly than a crewed launch system. Designing this kind of protection with high assurance is contrary to making a vehicle optimized for performance and weight, and of course ICBMs are stored in hardened subterranean silos until launch whereas a space launch vehicle is propelled from an elevated stand and protected only by a rollaway cover and a few lightning protection towers up to launch.
Stranger
Although structural failure of the fuselage is always possible because it is designed to only be as strong as necessary to minimize inert weight, the larger concern with high wind loads is just inducing so much flexure that the flight computer cannot maintain stability. A certain amount of flexure is anticipated as part of the inevitable natural modal vibration of the vehicle and the aeroelastic contribution of wind is factored into this, but once the wind shear (the difference between the force of wind at different stations on the vehicle) becomes too pronounced the center of mass of the vehicle can become displaced too far from the axis of thrust and the control system can induce oscillations trying to recover control. This is particularly a problem once a lot of the mass of the propellant is consumed and the modal frequencies of the vehicle can couple to vortex shedding around the fuselage with little internal damping, so it is more a problem when it gets at a certain point in flight, and especially around max-Qα, where the combination of unequal forces from the angle of attack combined with the dynamic pressure can cause a seemingly stable vehicle to suddenly fly out of control. If you ever see GN&C briefings for flight readiness review, you’ll note that they do a bunch of charts for general stability of flight but everyone is always focused on max-Qα and Stage 1 separation/Stage 2 ignition because those are the two events where the vehicle is most likely to suddenly go out of control. (There is also an issue with upper/insertion states referred to as–and I’m not kidding–“tail wags dog” where the mass drops so low and the center of mass is so far aft that the torque applied to the nozzle can cause the rest of the vehicle to rotate in pronounced fashion, and the more the TVC system tries to correct for it the worse the oscillation gets in an analogue of pilot-induced oscillation in aircraft. However, this is in what is almost always an exoatmospheric condition and has nothing to do with wind shear.)
In theory a ground wind could blow the vehicle into the gantry tower, I suppose, but I have never heard of this happening. Even once it has lifted off and is in an inertial state (with respect to lateral movement, anyway) it is so heavy that wind isn’t really going to push the entire vehicle very far. Unlike an aircraft that is large aerodynamic surfaces and mostly hollow space inside, the rocket vehicle at launch is a pretty narrow lateral profile and in the case of a liquid propellant vehicle is literally bulging with propellants, so it would take hurricane level winds to really push the whole thing very far. However, there are placard limits for when the shelter can be rolled back because a high enough ground wind acting along the moment of the entire vehicle could cause the restraints to fail or put enough lateral load on the vehicle to cause it buckle while constrained to the pad. I don’t know offhand what the placard limits would be for a vehicle like the Falcon 9 but for something like an Athena or Delta II they’re usually somewhere around 25 mph or less depending on the ground systems.
Stranger
Thanks; stability issues make sense as a larger concern here. I seem to recall Musk himself comparing rocket stability to balancing a garden hose on end in a hurricane. Perhaps a bit of exaggeration but difficult in any case.
I was only half-serious with my comment about colliding with the tower. Velocity wasn’t sure why weather would play a part at all; Florida does in fact encounter hurricanes; hence one might want to pay *some *attention to the weather.
The F9 Crew Dragon Launch Weather Criteria mentions a 30 mph limit at 162 ft (50 m, I guess). I’m not sure if this is due to the placard limits you mention or something else.
I believe they mentioned 3 criteria that were violated (Field mill, one of the anvil cloud ones, and a lightning one)
Initially the WX officer mentioned 10 more minutes and it might have been good, but later I think they mentioned it was later than that
(not that it makes a difference, as orbital mechanics dictated an instantaneous launch)
Brian
Dumb question time: It seems like they are ALWAYS fretting about the weather during a launch, and it rains and storms constantly in Florida. Can’t they find somewhere more ideal to launch? Would a desert be better? Tundra? Middle of the ocean?
Cape Canaveral in Florida is an almost ideal site–and for some trajectories, mandatory–to launch from North America because at 28 degrees North it is as close to the equator as you can get and still be able to launch pretty close to due east without overflying something important. Northern Brazil or French Guiana would be better (Arianespace has their main spaceport) but logistically and politically that would be problematic. The US also flies space launch vehicles from other facilities on the east and west costs (Vandenberg AFB in Central California, Wallops Flight Facility in Virginia, Kodiak Launch Complex in Alaska) but all of these have significant constraints (VAFB and KLC can pretty much only launch polar or retrograde trajectories, WFF has a narrower range of eastward trajectories and is at a higher azimuth which is less efficient) so the Cape ends up being the default for a lot of missions including ISS. We do have launch facilities in New Mexico (Ft. Wingate, White Sands Missile Range) but these are used only for short range testing, and facilities in the Pacific on Kauai, Wake Island, the Kwajalein Atoll, but these are also logistically complex to support as SpaceX discovered during the Falcon 1 program where they flew out of Omelek.
Mid-ocean is possible but the only real attempts at that have been the Sea Launch venture and the Russian Shtil (an SLBM converted as a small satellite launcher). These have some pretty significant constraints as well operationally and so they can’t really support many types of payloads. Bob Traux pushed for the Sea Dragon vehicle which would be towed to position and launch directly from the ocean without a support platform, and while feasible no one has found a need for a 550+ ton payload vehicle and it never got past proof of concept development and a paper design. There are also air launch ventures like the Orbital Sciences (now Northrop Grumman) Pegasus and the Virgin Orbit LauncherOne but these are only suitable for small payloads; the much larger Stratolaunch Systems never got off the ground, so to speak, despite building and flying the world’s largest carrier aircraft.
Stranger
Stranger ninja’d almost my entire post but I’ll add one thing. If you simply don’t care about what happens downrange, then there’s no need to have an ocean to the east for typical low-inclination launches.
China’s Xichang Satellite Launch Center, for example, is far inland. But it also once dropped a rocket on a village, killing probably a few hundred people. Developed nations don’t generally think of that kind of thing as acceptable.
The Baikonur Cosmodrome is also far inland, though at least their downrange is a bit less inhabited.
they were supposed to launch shuttles from CA coast but after the Challenger explosion they dropped that plan. NASA launches smaller rockets from the Virginia coast now.
Launches from Vandenberg AFB Space Launch Complex 6 (SLC-6) were exclusively US Air Force ‘Blue Shuttle” polar orbit missions. The Air Force didn’t really want to use the Shuttle anyway (despite driving the crossrange requirements, funding a program to develop and advanced fiber-wound composite solid rocket motor, and spending billions of dollars outfitting SLC-6 for STS operations) so they were happy to find a reason to reject Shuttle and go back to expendable launch vehicles.
Stranger
All systems go for tomorrow, however “strong possibility” weather won’t allow launch.