Space News
space history and artifacts articles

space history discussion forums

worldwide astronaut appearances

selected space history documents

related space history websites

  collectSPACE: Messages
  Space Shuttles - Space Station
  Quora: Risk analysis of flying on the shuttle

Post New Topic  Post A Reply
profile | register | preferences | faq | search

next newest topic | next oldest topic
Author Topic:   Quora: Risk analysis of flying on the shuttle

Posts: 1571
From: Atlanta, GA, USA
Registered: Feb 2007

posted 09-01-2012 09:52 AM     Click Here to See the Profile for garymilgrom   Click Here to Email garymilgrom     Edit/Delete Message   Reply w/Quote
There is an interesting analysis of the risk of flying on the space shuttle on Quora.

The writer notes the difference between safety and survivability (underlined below). Here's the gist of the analysis:
  • Number of individual space shuttle astronauts/cosmonauts: 355
  • Number of space shuttle fatalities: 14
  • Chance of dying on a shuttle if you were a shuttle astronaut: 1 in 25
For comparison -
  • Number of individual Soyuz astronauts/cosmonauts: ~173
  • Number of Soyuz fatalities: 4
  • Chance of dying on a Soyuz if you are a Soyuz astronaut/cosmonaut: 1 in ~50
GM here - I added the Saturns for comparison:
The number for the Saturn V would be 100% as there were no fatalities on that launch system, but for the Saturn 1B we have to include the Apollo 1 fire so with three people killed out of 18 "flown attempts" (Apollo 1, 7, ASTP and 3 Skylab crews) the chance of dying on a 1B was worst of all - about 1 in 6.

Back to the original answer:

More important than the difference in fatality rates Soyuz' fatalities occurred in the first 10% of the program — the 41 years and 103 nonlethal flights since then suggest that the most lethal bugs have been worked out. Fatal shuttle accidents were evenly spaced through the program. The Columbia Accident Investigation Board judged that the shuttle could not be made into a safe, operational vehicle, and that the right way to manage risk was to fly only essential missions.

The key lesson is the difference between safety and survivability. Rockets are unsafe. They "explode" about 1% of the time. The key is to build a system that allows people to survive a catastrophic failure of the launcher. This usually requires putting people on top of the rocket, not to one side, and this has the added benefit of protecting your heatshield from debris. shuttle had incredible accomplishments but was an unusually lethal experimental space vehicle. The future's safer operational human space exploration vehicles will not look like the shuttle.

Lou Chinal

Posts: 946
From: Staten Island, NY
Registered: Jun 2007

posted 09-01-2012 10:30 AM     Click Here to See the Profile for Lou Chinal   Click Here to Email Lou Chinal     Edit/Delete Message   Reply w/Quote
Gary, great post! The question still stands parachutes or wings?

Spaceflight will never be an everyday event if you plan to drop into the ocean. Flying to a runway is a much better way of coming home.

But, for right now getting ones feet wet may be the safe and sure way. A small shuttle on top of a rocket?

Maybe you could do a graph: weight of wings over the weight of the parachutes vs. number of astronauts.


Posts: 170
From: Ohio
Registered: Mar 2006

posted 09-01-2012 11:23 AM     Click Here to See the Profile for canyon42   Click Here to Email canyon42     Edit/Delete Message   Reply w/Quote
I think the statistic they're coming up with for a "chance of fatality" is a little misleading and/or meaningless. If you only flew a single mission, obviously your chances of dying were less than if you flew multiple missions. The notion of dividing the total number of astronauts (regardless of how many missions they flew) by the number of fatalities doesn't take that into account.

A more meaningful statistic might be to divide by the 14 fatalities the total number of crew spots launched (somewhere in the 700s, if I'm correct, although I don't know the exact number). That would give you the chance that any particular astronaut would be killed on a single mission. You could then multiply that by however many missions a person actually flew if you wanted to figure out that particular individual's "odds."

Just as an extreme example of what I mean, suppose NASA only maintained a pool of 30 astronauts and just kept recycling all of those folks over and over throughout the whole history of the shuttle program. You would still have the same number of fatalities, but now the "chances" of any single astronaut dying would be 14 out of 30, or almost 50 percent. This would not mean that the system was any more "dangerous" than it was in our real-world experience--just that fewer people overall flew on it.

Also, I'm not sure that the point of including the Apollo 1 fatalities in an analysis of the Saturn 1B is a valid one. The obvious point is that the fire did not occur during a flight, but beyond that it really had nothing to do with the booster itself. It could have just as easily occurred on top of the Saturn V (or any other booster for that matter).


Posts: 1571
From: Atlanta, GA, USA
Registered: Feb 2007

posted 09-01-2012 12:09 PM     Click Here to See the Profile for garymilgrom   Click Here to Email garymilgrom     Edit/Delete Message   Reply w/Quote
I'm not sure the fire should be included either.

I don't know what you mean by crew spots vs. crews flown. Can you explain please?


Posts: 400
From: Pampa, TX, US
Registered: May 2011

posted 09-01-2012 12:24 PM     Click Here to See the Profile for Spaceguy5   Click Here to Email Spaceguy5     Edit/Delete Message   Reply w/Quote
Crew spots meaning the sum of the number of crew members for each mission.

Remember that some people flew more than once.

  • Columbia: 160
  • Challenger: 60
  • Discovery: 252
  • Atlantis: 207
  • Endeavour: 148
Total: 827

I don't know if the Discovery/Atlantis/Endeavour statistics account for people who only landed on the space shuttle (but didn't launch on it).

Robert Pearlman

Posts: 27328
From: Houston, TX
Registered: Nov 1999

posted 09-01-2012 12:33 PM     Click Here to See the Profile for Robert Pearlman   Click Here to Email Robert Pearlman     Edit/Delete Message   Reply w/Quote
The total number of crew spots, or seats filled, for the shuttle program was 847 crew members launched; 833 crew members returned, per the STS-135 press kit.


Posts: 355
From: CA.
Registered: Jan 2011

posted 09-01-2012 03:04 PM     Click Here to See the Profile for SkyMan1958   Click Here to Email SkyMan1958     Edit/Delete Message   Reply w/Quote
14 dead out of 833 landed produces a chance of fatality of ~ 1.68%, e.g. a little less than a 1 in 50 chance in dieing. (Obviously 14 dead out of 847 would be a little less than that).


Posts: 1571
From: Atlanta, GA, USA
Registered: Feb 2007

posted 09-01-2012 05:20 PM     Click Here to See the Profile for garymilgrom   Click Here to Email garymilgrom     Edit/Delete Message   Reply w/Quote
OK thanks for everyone's help in understanding this. Using the 847 crew members launched gives a chance of dying of 1 in 60. Does anyone know if the Soyuz statistics used have a similar error?

After more thought on the Apollo 1 fire I do think it should be included somewhere in the statistics because the fatalities include all types, not just types caused by launch vehicles. Maybe it would be more appropriate to examine the Saturn series in total.

Jay Chladek

Posts: 2211
From: Bellevue, NE, USA
Registered: Aug 2007

posted 09-02-2012 02:21 AM     Click Here to See the Profile for Jay Chladek   Click Here to Email Jay Chladek     Edit/Delete Message   Reply w/Quote
The other thing misleading about these statistics as they don't necessarily take into account the improvements and refinements made to a spacecraft design. Those refinements can skew the numbers when a change is made resulting in the potential for improved safety.

Rocket and spacecraft design tends to be a somewhat reactionary thing. This means while testing takes place (usually) before an all up flight, sticking a man onboard becomes its own test and any major bugs encountered on the flight are addressed and fixed (hopefully) before the next one.

So, the risks faced by the Apollo 1 astronauts was much higher than it was for say Apollo 17 or ASTP that their spacecraft might kill them (although after nitrogen tetroxide fumes entered the Apollo on ASTP during descent, death of the crew was still a real possibility).

Soyuz is also the same way. Soyuz 1 had the parachute problem and it killed Komarov. But if Soyuz 2 had flown with its crew of three, chances of fatality while not being 100% necessarily were pretty high since apparently that craft had the same parachute flaw.

But by comparison, the death of the Soyuz 11 cosmonauts resulted in a better craft with the pressure suits and rapid re-pressurization capability for all crew that flew on Soyuz after. So death by asphyxiation in that particular manner (due to not wearing pressure suits) became an impossibility.

Granted so many other things exist that can still kill a cosmonaut or a NASA/ESA/JAXA astronaut on a mission. It is a form of Russian roulette in the manned spaceflight business. But instead of a six shot revolver with one bullet loaded, it is a revolver with millions of chambers and maybe hundreds of possible bullets loaded.

Each time a craft flies, the drum spins. But after it comes back, some of those bullets are taken out. But others can take their place over time as success leading to complacency can lead to its own magic bullet getting loaded (like ET foam from a bipod impacting an RCC panel on shuttle Columbia).

Besides, simple statistics like that are for those who try to pair things down to a very simple equation or explanation. While certain things I suppose could be paired down that way, I don't know if it can be done with the world's space programs since there are so many variables to account for.


Posts: 1571
From: Atlanta, GA, USA
Registered: Feb 2007

posted 09-02-2012 06:21 AM     Click Here to See the Profile for garymilgrom   Click Here to Email garymilgrom     Edit/Delete Message   Reply w/Quote
Jay makes some good points and seems to agree with the original post about how some rockets evolved and became safer over time.

And while spaceflight will always be complex sometimes simple stats are useful. Looking at Apollo/Saturn and including Apollo 1, we get a 1 in 16 chance of dying (three deaths in 48 crew slots) but per the OP all the deaths happened in the first 5% of launches — the first launch if you will — leading to zero deaths in 45 crew slots for the rest of this program.

Leaving the Apollo fire out of the statistics, isn't it interesting that of three major programs (Apollo, Soyuz and Shuttle) spanning two nations and different designs all achieved about a 1 in 50 chance of a fatality during the flight for the flyers?

As an aside to this topic I don't think the ET foam became a concern (magic bullet) late in the Shuttle program due to complacency, I think it was waiting in every launch. They never solved the problem of foam shedding.


Posts: 170
From: Ohio
Registered: Mar 2006

posted 09-02-2012 10:44 AM     Click Here to See the Profile for canyon42   Click Here to Email canyon42     Edit/Delete Message   Reply w/Quote
Of course, the 1 in 60 chance of dying on a shuttle mission that is derived from that method corresponds pretty closely to the 1 in 67.5 chance that you get by simply dividing the number of missions by the number of accidents resulting in the loss of the crew.

These two methods would give the exact same odds under two conditions: first, that every incident that kills any crew member kills ALL crew members (which has been the case every time so far, although there are undoubtedly some possible scenarios that could have resulted in the loss of some crew members with others surviving); and second, that the number of crew members is the same on every mission.

The second of those is obviously not true and results in the slight discrepancy between the two methods, simply because both flights that were lost had a higher number of crew (7) than the average. If both of the lost missions had been among the first four flights (with only 2 crew) then the numbers would have been skewed even further — then you would have 4 fatalities out of 847, which gives a much smaller percentage than 14, even though the same number of missions were lost.

Another thing that has to be considered is the sheer number of missions the shuttle flew compared to M-G-A. Only 10 Saturn Vs ever flew manned, for example, and because none of those missions lost a crew it can be said to have a 100 percent safety rate.

The corollary to that, though, is that the shuttle also had no losses in its first 10 missions — it took until 25 of them for anything to happen. (Although both the Saturn V and the shuttle certainly had "near misses" within their first 10.) How many missions could the Saturn V/Apollo combination have flown before suffering a crew loss? How many before suffering TWO crew losses? Pretty much impossible to say, really, but the fact that they came so close on Apollo 13 (which really had nothing to do with the booster, but still), along with additional potential losses on Apollo 12 and ASTP, would suggest that sooner or later it would have happened. (And yes, ASTP was a 1B, but you see the point.)

Even Mercury (with Grissom in the water, and discounting Glenn's heat shield indications and Carpenter's overshoot) and Gemini (Armstrong and Scott) each had a "near miss" in terms of a crew loss, even though they only flew 6 and 10 missions respectively.

Maybe a tangent to all this would be to suggest that (with the exception of the Apollo 1 fire, which I'm still not sure how to categorize) while M-G-A all had near misses, each time the mission/booster/craft/crew managed to survive in spite of the incident, while the shuttle was lost with all hands in both of its major incidents. Of course, that doesn't take into account all of the near misses that the shuttle experienced that did NOT result in the loss of a crew, either.

The final thing that always strikes me is that in the end the shuttle VEHICLE proved to be a ridiculously safe and reliable piece of equipment. It was the launch system, not the vehicle, that killed the crew in both instances. I think it's safe to say, for example, that an Apollo capsule would not have survived having a Saturn V disintegrate right alongside it, nor would it have survived reentry with a large chunk of its heat shield blasted away. The shuttle vehicle was a remarkable achievement — the notion of putting it on the side of its launch system, not so much.


Posts: 258
From: Bristol UK
Registered: Jan 2005

posted 09-02-2012 11:18 AM     Click Here to See the Profile for star61   Click Here to Email star61     Edit/Delete Message   Reply w/Quote
This is almost like looking at a poker hand knowing the implied and absolute odds at a point in the hand, as against after the cards are laid out . Would you have played the hand the same knowing what you now know. Feynman said it clearly in his critique of the pre Challenger risk assessments. The reality was if you stepped into a shuttle and launched up to and including STS-51l, you had one chance in 25 of dying. So at the end of the shuttle era we know for a fact that if you launched on that machine your odds of survival were approximately 1 in 68.

Probability theory is very subtle and not at all obvious when dealing with complex engineering but the raw figures are reality.

All times are CT (US)

next newest topic | next oldest topic

Administrative Options: Close Topic | Archive/Move | Delete Topic
Post New Topic  Post A Reply
Hop to:

Contact Us | The Source for Space History & Artifacts

Copyright 1999-2012 All rights reserved.

Ultimate Bulletin Board 5.47a