Friday, October 12, 2018

What are the major causes of rocket launch failures?

What are the major causes of rocket launch failures?
Casey Handmer
Originally answered on Quora, July 28 2014

Let's look at the launch failures since 2000. And there have been a few! (Let me know if I missed any...)

Two failures prior to 2000 of Ariane 5, one from guidance software, one from anomalous upper stage torque.
Ariane 5 flight 10: Partial failure due to upper stage anomalously low thrust.
Ariane 5 flight 14: Upper stage anomaly, self destruct.

GSLV has had five failures.
D1, F02 and F04 due to guidance issues.
D3 due to upper stage booster pump failure.
F06 due to loss of control of liquid boosters (guidance or control issue).

H-IIA F6 due to stage separation failure.

Russia launches a LOT of rockets. Some of their rockets are the most reliable in the world. Still, they have had some failures.

Proton-M has had 10 failures. 
1 due to overfueling of the upper stage (human error) in 2010. 
In 2002, optional Blok-DM fourth stage shut down prematurely.
5 due to problems with optional Briz-M fourth stage. 
In 2006, Briz-M stage shut down prematurely due to oxygen supply line burning through during second firing.
In 2008, Briz-M stage shut down due to failure of gas duct between gas generator and turbine.
In 2011, Briz-M stage lost attitude control due to software error.
In 2012, Briz-M stage failed after 7 seconds.
In 2012, Briz-M stage failed 4 minutes early.
3 due to Proton-M lower stage. 
In 2007, a damaged cable prevented stage separation. 
In 2013, yaw sensors were installed incorrectly, resulting in failure shortly after liftoff. 
In 2014, third stage engine failure. 

Soyuz 2 rocket has had 2 failures.
In 2009, due to under performing third stage leaving payload in lower orbit.
In 2011, due to failure of combustion chamber wall in third stage RD-0124 engine.

Rokot (a converted ICBM) has had 2 failures.
In 2005, due to a software error.
In 2011, due to upper stage malfunction.

Volna (a converted SLBM) has had 3 failures.
In 2001, due to payload separation failure.
In 2002, due to a payload/launcher interface issue.
In 2005, due to the failure of the first stage turbopump.

Soyuz U has had 21 failures out of an incredible 745 launches.
In 2002, due to engine failure on one of the boosters.
In 2011, due to an upper stage problem.

Not every Chinese launch failure is public knowledge. Two prominent fatal accidents of the Long March 3B in the mid 1990s were due to guidance problems shortly after launch.
Long March 2C (unknown)

Dnepr-1 (converted ICBM) in 2006, due to hydraulic failure on one of the first stage combustion chambers. 

Zenit-3SL (launched in the US by Sea Launch). Originally developed as a booster.
In 2000, due to a software error.
In 2004, due to a wiring fault in the upper stage.
In 2007, due to debris in the first stage turbopump.
In 2013, due to premature engine shutdown and guidance faults.

Atlas V in 2007, due to a fuel leak from a faulty valve.

Delta III in 2000, due to a guidance issue. Two previous Delta III launches in the late 1990s failed due to a software issue and an upper stage issue respectively.

Delta IV in 2004, due to premature engine cutoff. 

VLS-1 in 2003, due to exploding on the launch pad.

Shavit 1 in 2004, due to unknown reasons.

Safir 1 in 2008, due to unknown reasons.
Three further failures in 2012 and 2013, reasons unknown.

North Korea
Unha-2 in 2009, due to third stage malfunction.
Unha-3 in 2012 shortly after liftoff.

Falcon 1 in 2006, due to first stage engine failure - control electronics burned.
In 2007, due to second stage oscillation (guidance and control issue).
In 2008, due to software error in staging.

Rockets are very powerful machines with lots of parts and very little margin for error. In fact, if you assume some tiny probability for any given part to fail, then look at the overall probability of success, it is 0. All rocket parts need to be tested. Today, with computers, we are able to analyse some parts of the problem (like guidance software) in the loop very thoroughly. The hardware is correspondingly admittance tested to ensure that it will work.

Nevertheless, the above list of launch failures has a few trends. Some common problems are:
  • Inexperienced designers and builders, be that on new rockets, new companies or, in the case of several Russian launch failures, a lack of expertise caused by retirement and death of elderly program engineers.
  • Third/upper/vacuum stage problems. The vacuum stage is difficult to test on Earth, often runs on cryogenic fuel, is the last thing to go wrong, and has to operate in a bizarre environment. Some organizations have jealously guarded expertise in this area, others obviously do not.
  • Old systems or old parts.
  • Software errors. The cheapest part to change is often changed wrongly, and has poorly understood failure modes.
  • Guidance and sensor problems. This may seem obvious, but rockets need dynamic control to work. The systems are complicated and poorly understood.
  • A culture that doesn't prioritize getting it right. This is not as obvious from the above list, but many failures, particularly in the Japanese and Indian space programs, have been attributed to bad error handling. Invariably, someone knew something was wrong, but was unable or unwilling to fix it. Sometimes this is due to the junior engineer fearing or respecting the senior engineer, sometimes it is due to organizational issues. Either way, it's a dumb reason to crash a rocket.
  • Probably sabotage has contributed to North Korea's and Iran's ongoing difficulties.
  • Prior to the Long March 3B crash in 1996, China's launch success rate was around 80%. After that crash, which compromised and destroyed a classified Boeing payload, Boeing helped institute changes in project management which has subsequently greatly increased their success rate to about 98%.

No comments:

Post a Comment

If you want to comment without a google or open-ID account, sign enough of your name that at least I know who you are, or leave an email address or bank account details or something.

Note: Only a member of this blog may post a comment.