Primary Causes of Rocket Failures?

[Exosphere]

I've read that the failure rate for rocket launches is in the low single-digits percentage -- a very high rate compared to other "transportation" systems (i.e. commercial aircraft). Early rockets in particular had a high failure rate:

While watching that video, I was reminded of a statistic I had heard in a Project Mercury documentary stating that, at the time of John Glen's orbital flight, the Mercury-Atlas booster had a failure rate of 50% -- something like two out of four launches had failed catastrophically.

What kinds of failures occur that cause that high of a failure rate? I know hard starts and combustion instability are two major problems, but what are the other most common rocket failure modes, and what kinds of steps do today's rocket designers use to prevent them from occurring? In particular, what causes rockets to have a drastically higher failure rate than, say commercial aircraft? I'm assuming it's the high energy requirement for achieving orbit vs just flying and the fact that most rockets have a production rate similar to test aircraft, but I'm not sure. Is there any way to further reduce the failure rate of orbital rockets to bring it closer to that of most standard aircraft?

[SargeRho]

Rockets release insane ammounts of energy. That is bound to break something every once in a while.

[Streetwind]

Take a metal tube ten meters tall. Build it as flimsy as it possibly can be. Fill it to the top with high explosives. Make the whole contraption fly in a manner that keeps it from spinning, tipping, falling or spontaneously detonating. Do all of this without any prior knowledge. Then report your success and reliability ratings

You can't really compare early experimental rocketry with modern car and aircraft manufacture. How about comparing it to early attempts to build an aircraft? There are videos that show whole series of silly contraptions meant to fly that at best shook themselves apart a few feet above ground, if they even managed to move at all. Success and reliability ratings were awful. And even long after aircraft flew regularly, they were still prone to failure.

For instance, take the Orteig Prize. Offered 1919 for the first pilot to cross the atlantic ocean by plane in a single, non-stop flight (from New York to Paris, specifically). For the first five years, nobody even tried it because nobody had any confidence in being able to build a plane that wouldn't kill them. And even then, it took another three years during which five people lost their lives and more were injured before Charles Lindbergh managed the feat. He flew 33 hours straight, solo without sleep, in a single-motor craft that had been tested only for less than 30 hours before the attempt. It was a super high risk gamble, he could easily have crashed for both technical and human failures. But such was the state of aviation in the 1920's, and similarly, such was the state of rocketry in the 1960's.

Another reason is the measuring method used to calculate rocket launch reliability. After its first successful flight, the rocket is awarded not a 100% rating, but rather a 50% rating. Every subsequent successful flight then pushes the rating upwards, a failure downwards. Thus, no rocket can ever get to 100%, at least not without tens of thousands of successful flights. And rockets only fly a couple dozen times, usually. This means that, even if the rocket never fails, mathematically the rating can only go so high over so many flights. Thus many rockets, especially newer ones, are still rated at a failure rate of a small few percent.

Edited August 13, 2014 by Streetwind
words, how do they work

[sgt_flyer]

Well, combustion instabilities etc can cause severe vibrations in a rocket. (Which generally use welded plumbing) if there's an undetected fault in one of the tubes / welds (welds quality can be controlled through x-rays) then the vibrations can make this fault crack - and you get spilled fuel or oxidizer outside the plumbing (which is bad ^^) (those vibrations can become pogo oscillations - engine failure detection can trigger an early engine cutoff if they are too extreme - resulting in the need to cancel the mission / safely self destruct the booster.

The gimbal / vernier systems can also break - (which could cause assymetric thrust - although the systems are designed to mitigate those risks, ie in case of the control systems malfunction, the gimbal / vernier can return to neutral position)

You can also have a turbopump failure, because of many things (imagine something was forgotten inside one of the tanks, it can lead to damage to the pumps - with no pumps, no thrust )

You can also have human errors / guidance system failure. like when they put Ariane IV guidance systems in ariane V without reconfiguring the system for the Ariane V capabilities. (Hint, the guidane system crashed, and the rocket became uncontrollable )

[magnemoe]

Take a metal tube ten meters tall. Build it as flimsy as it possibly can be. Fill it to the top with high explosives. Make the whole contraption fly in a manner that keeps it from spinning, tipping, falling or spontaneously detonating. Do all of this without any prior knowledge. Then report your success and reliability ratings

You can't really compare early experimental rocketry with modern car and aircraft manufacture. How about comparing it to early attempts to build an aircraft? There are videos that show whole series of silly contraptions meant to fly that at best shook themselves apart a few feet above ground, if they even managed to move at all. Success and reliability ratings were awful. And even long after aircraft flew regularly, they were still prone to failure.

For instance, take the Orteig Prize. Offered 1919 for the first pilot to cross the atlantic ocean by plane in a single, non-stop flight (from New York to Paris, specifically). For the first five years, nobody even tried it because nobody had any confidence in being able to build a plane that wouldn't kill them. And even then, it took another three years during which five people lost their lives and more were injured before Charles Lindbergh managed the feat. He flew 33 hours straight, solo without sleep, in a single-motor craft that had been tested only for less than 30 hours before the attempt. It was a super high risk gamble, he could easily have crashed for both technical and human failures. But such was the state of aviation in the 1920's, and similarly, such was the state of rocketry in the 1960's.

Another reason is the measuring method used to calculate rocket launch reliability. After its first successful flight, the rocket is awarded not a 100% rating, but rather a 50% rating. Every subsequent successful flight then pushes the rating upwards, a failure downwards. Thus, no rocket can ever get to 100%, at least not without tens of thousands of successful flights. And rockets only fly a couple dozen times, usually. This means that, even if the rocket never fails, mathematically the rating can only go so high over so many flights. Thus many rockets, especially newer ones, are still rated at a failure rate of a small few percent.

This, add that plane launches has multiple abort options if something goes so wrong they can not reach target. Rockets has none, the escape system is to save the life of the crew just as an ejection seat in a fighter.

[cantab]

Expected lifespan vs performance factors in too. Modern road cars are very reliable, their engines, gearboxes, and so on can do thousands of miles before they even need any maintenance, and hundreds of thousands if looked after. Formula One cars, on the other hand, regularly suffer gearbox failures and engine failures, because their gearboxes and engines only need to last a few race weekends (and not so many seasons ago they only needed to last one race) and they need to be as light as possible. The inevitable random factors though mean that sometimes one will break in a race.

It's the same for (non-reusable) rockets. They only need to last one flight and they need to be as light as possible to maximise payload, so they aren't going to be built for thousands of flights. The upshot is that sometimes they'll fail in that one flight.

[lajoswinkler]

They mostly tilt and then the pressure of the air rushing becomes uneven. They aren't built to withstand much lateral forces, twisting, rotating and stuff like that.

Early rockets were less sturdy so they would break up at low speeds. These days when they break up, it's usually either because of the overpowering rush of air or the solid rocket fuel has cracks.

[TythosEternal]

The single highest-likelihood, highest-consequence failure mode in launch vehicle (liquid or solid) is combustion instability.

Inconsistent combustion of oxider feeds in particular are very, very dangerous. This can result from a large number of causes, from pressure variations in your turbopump feeds to injector plate inconsistencies to unusual vibration modes in your combustion chamber. Poorly-controlled variations in your steady-state conditions and manufacturing defects are also frequent factors, as are operating conditions outside the design space (a la Challenger).

The result of combustion instability is, simply, things blow up. Unconsumed oxidizer, at the temperatures and pressures that exist inside your combustion chamber, can actually react and combust with metallic components, induce a fatigue or rapid crack failure, or produce uncontrollable variations in thrust. Uncontrolled combustion products induce material failures in surrounding components, creating a chain reaction--and it doesn't take much for a delicate, high-energy piece of equipment operating under extreme conditions to catastrophically fail.

In short, you will not be going into space today.

[Kryten]

Explosion due to combustion instability isn't going to happen to any rocket that has already got past the ground testing stage and isn't being shaken to pieces already. Most failures are due to quality control errors; as an example, tens of rockets from multiple countries have fallen to cleaning rags being left in pipework.

[Moar Boosters]

Personally I think it's usually down to a lack of boosters.

[TythosEternal]

Explosion due to combustion instability isn't going to happen to any rocket that has already got past the ground testing stage

Ground testing is one way to mitigate this particular risk factor. (Risk factors, as I pointed out in the beginning of the response, are functions of their likelihood and their consequence--not testing and mitigation efforts, which are made in response to the assessment of the event.) Mitigation does not mean irrelevance, however, and many vehicles can still encounter disastrous combustion issues. (The Delta II launch of GPS IIR-1, in which a high-reliability GEM-40 developed a crack and subsequently experienced a 'structural failure', is one of my perennial favorites.)

and isn't being shaken to pieces already.

"Being shaken to pieces" is close to being the very definition of combustion conditions (even during normal steady-state and transient sequences) and effects. Considerable effort goes into mitigating vibrational modes excited by vehicle engines--including for the engines themselves. Especially for main engine sequences, there's a controlled explosion taking place at one end of your rocket, and a whole lot of shaking going on!

Most failures are due to quality control errors; as an example, tens of rockets from multiple countries have fallen to cleaning rags being left in pipework.

I think you are confusing quality control, which ensures the manufacture of components within sufficient tolerances, with process and inventory control, which ensures High Bay Bob keeps track of all his wrenches. I'd be interested to see a source on the 'tens' figure, but it's irrelevant anyways: Do you know what failure mode is induced by FOD (foreign object debris) in plumbing? That's right! Combustion instabilities. (Catastrophic turbopump failures are another result, but it's the uncontrolled combustion that does the greatest and fastest damage.)

I would encourage you to look at an AIAA paper written by a colleague of mine on launch vehicle failures. Only the front page is readable for non-members (well, unless you want to pay the $1065 cover charge... AIAA is ridiculous like that), but even that much is interesting and very accessible to non-engineers, actually.

Lastly, I would end with a note on mitigation. As long as there have been rocket engines, engineers have been trying to control the intricate instabilities of the combustion process and its sensitivities. Even today, though, when something goes wrong after ignition anywhere in the fuel train, the typical response is to shut down the effected systems BEFORE an instability occurs. Sometimes this means you continue on other engines, sometimes it means the RSO hits the big red button and you're out the better part of a billion dollars. This is all to say that, looking at a history of launch failures, many cases in which rockets deliberately shut down were the result of efforts to avoid instabilities, trading them for a more controlled (and de-escalatable) failure mode instead.