Zubrin on NASA aversion to risk

[strangequark]

Many unmanned vehicles have much better reliability records than the Shuttle

Name one.

Vehicles that haven't flown often enough to be statistically comparable don't count.

The Delta II, Atlas II, and Tsyklon-2.

http://www.spacelaunchreport.com/log2009.html

Old data, so missing the last STS missions, and includes an STS mission with a low orbit. If you only include LOC/LOM cases (or ones that would be), then STS does not vary from the Delta II and Tsyklon-2. Atlas II only had about 60 launches, so not enough to compare.

[deltaV]

And STS-107 was not strictly a launch failure.

I disagree. The orbiter thermal protection system was not designed to withstand debris strikes because the external tank was required to not shed debris. The only failure to meet specifications that contributed to the STS-107 accident was the external tank foam shedding. The external tank was used only during launch and discarded afterward, so it's hard to call STS-107 anything but a launch failure.

It is true that the STS had stricter payload protection requirements than most launchers do, and the foam shedding event would have been a non-event if the payload were located at the top of the rocket. However the unusual strictness of the requirements doesn't excuse the launch system from failure to meet those known requirements.

Edit: maybe I exagerated a bit in STS-107 could also be seen as a failure to set reasonable requirements.

Edit: this is getting a bit off-topic.

[mars.is.wet]

I thought NASA missed a golden opportunity to re-brand "astronauts" as "explorers" and enter into a meaningful discussion of risk with the American people. 

I also think Mike Griffin missed an opportunity to innoculate NASA to an exploration failure by ending his tenure with a variant of the speech "people will die, and it will be worth it".

Pretty much no matter how you slice it, a trip to the surface of the moon and back will be about as risky as recent Shuttle flights (by similar analysis).

For once, Zubrin is right on the mark.

[DarkenedOne]

Many unmanned vehicles have much better reliability records than the Shuttle

Name one.

Vehicles that haven't flown often enough to be statistically comparable don't count.

The Delta II, Atlas II, and Tsyklon-2.

http://www.spacelaunchreport.com/log2009.html

I'd say those reliability numbers are a bit better than shuttle, not "much better" as you stated.

The shuttle has historically had a pretty low LOM rate. Its fatal flaw as a manned launch system was how 2 out of 3 loss of mission events also caused loss of crew. Shuttle had one partial failure with no injuries (abort to orbit) and two loss of crew events.

The Shuttle's fatal flaw was that there were only 5 of them. 

If an expendable rocket blows up the only real loss is the payload since the rocket was going to be lost anyway. 

When a Shuttle is lost you lose the payload, a 7 man crew, and an irreplaceable $1+ billion launch vehicle.  Even if you did not care for the astronauts.  For such a vehicle worth that much and is not replaceable a LOC of .03 was unacceptable.  Imagine if you car had a similar rate of failure. 

[Robotbeat]

I thought NASA missed a golden opportunity to re-brand "astronauts" as "explorers" and enter into a meaningful discussion of risk with the American people. 

I also think Mike Griffin missed an opportunity to innoculate NASA to an exploration failure by ending his tenure with a variant of the speech "people will die, and it will be worth it".

Pretty much no matter how you slice it, a trip to the surface of the moon and back will be about as risky as recent Shuttle flights (by similar analysis).

For once, Zubrin is right on the mark.

I'm pretty sure that going to the Moon's surface and back will be significantly riskier than one of the latest Shuttle flights (with enough testing, a flight to the surface of the Moon and back may be only as risky as one of the first Shuttle flights... 1/15-1/20 LOC chance). But I agree in general that we need to accept this risk.

[mars.is.wet]

I'm pretty sure that going to the Moon's surface and back will be significantly riskier than one of the latest Shuttle flights (with enough testing, a flight to the surface of the Moon and back may be only as risky as one of the first Shuttle flights... 1/15-1/20 LOC chance). But I agree in general that we need to accept this risk.
[/quote]

I saw a 1960's analysis once that showed LOC for the end-to-end Apollo missions calculated at 1/18.

I agree that a factor of 6 improvement (if we went back today) would be phenomenal. That is what Cx was shooting for.

But realism doesn't sell.

[Warren Platts]

@ Darkened One: That's a good point about RLV's in general.

I'm pretty sure that going to the Moon's surface and back will be significantly riskier than one of the latest Shuttle flights (with enough testing, a flight to the surface of the Moon and back may be only as risky as one of the first Shuttle flights... 1/15-1/20 LOC chance). But I agree in general that we need to accept this risk.

Not necessarily. Atlas has a good track record. Put a capsule on that to get them to orbit, that's half the battle right there. Getting back to Earth is also pretty safe; there were only a couple of Russian mishaps a long time ago. As someone said, the Columbia accident should really be considered a launch failure. And it could have been avoided if they had a LON capability and inspected the tiles before coming back.

Similarly for the Moon: hopefully, they will use the ULA DTAL landers. These are an inherently safe design: with 4 RL-10's they can have 3 engine-outs and still land safely. Since these would be evolutionary descendents of ACES, then there will already be a long track record before the first landing, thus upping the reliability, reducing risk. Set up a separate lander at the EML Gateway that would always be ready to mount a rescue mission if necessary.

[Robotbeat]

There's no abort option except for a long trip to Earth, ala Apollo 13. If the problem happens at the Moon's surface, you're out of luck.

Shuttle's last missions were pretty reliable and low risk (even without LAS). It's pretty unlikely for a Moon trip to get that low risk anytime soon. Shuttle made over 100 flights and had the real possibility of abort-to-orbit (and abort to Station) and rescue missions which wouldn't be an option for a lunar mission once it leaves LEO.

[deltaV]

The Shuttle's fatal flaw was that there were only 5 of them. 

If an expendable rocket blows up the only real loss is the payload since the rocket was going to be lost anyway. 

When a Shuttle is lost you lose the payload, a 7 man crew, and an irreplaceable $1+ billion launch vehicle.  Even if you did not care for the astronauts.  For such a vehicle worth that much and is not replaceable a LOC of .03 was unacceptable.  Imagine if you car had a similar rate of failure.

The amortized cost of loss of shuttles was around 0.02 * $3 billion = $60 million per flight. That's annoying but a small part of the overall launch costs.

[DarkenedOne]

The Delta II, Atlas II, and Tsyklon-2.

That's your idea of "much better"?

Keep in mind that Shuttle's failure rate of slightly less than 1.5% is on the margin of statistical significance as it is.  You cannot reasonably maintain that Tsyklon-2's 1% is "much better" when you're talking about one failure versus two - if the Shuttle program had been ended after the same number of flights it would have had the same failure rate.  Certainly the Delta II's 1.3% failure rate is indistinguishable from Shuttle's.  The successful run between STS-51L and STS-107 was 40% longer than the Atlas II's entire career.

And STS-107 was not strictly a launch failure.  I find the Shuttle's performance quite impressive considering the unusual number of failure modes...

Your statement was wrong.

First of all there is a 30% chance of the Shuttles actual success rate is equal to or greater than the Tsyklon-2 based on the experimental data.  Therefore it is not statistically insignificant. 

Secondly I would like to point of the flaw in this method.  Judging the launch vehicles reliability by this data is not accurate.  The Shuttle has been modified many times since its beginning, so there is no doubt that it is significantly more reliable than these launch figures depict.  Then again the same if true for the other vehicles as well.

In any case I did not make that point in say that the Shuttle is inferior in any way with data that for the reasons I stated above are irrelevant anyway. 

The point I am trying to make is that there is a learning curve, and that the best way to improve reliability and reduce risk is through experience. 

[mars.is.wet]

The amortized cost of loss of shuttles was around 0.02 * $3 billion = $60 million per flight. That's annoying but a small part of the overall launch costs.

Not when marginal launch costs were < $300M (up to the flight rate limit)

That is a real hit.

[Warren Platts]

There's no abort option except for a long trip to Earth, ala Apollo 13. If the problem happens at the Moon's surface, you're out of luck.

Shuttle's last missions were pretty reliable and low risk (even without LAS). It's pretty unlikely for a Moon trip to get that low risk anytime soon. Shuttle made over 100 flights and had the real possibility of abort-to-orbit (and abort to Station) and rescue missions which wouldn't be an option for a lunar mission once it leaves LEO.

Did you read what I just wrote? If you had an EML space station (it wouldn't have to be nearly as elaborate as ISS, a Bigelow module and a single launch would do) then you could have an abort-to-station option in the vicinity of the Moon. Also trajectories to EL1 don't typically require a big injection burn, so even though a free-return trajectory wouldn't be an option, there would still be a safe place to go that wouldn't require a lot of delta v to get to. And if you had a separate lander at the EML station, then rescue from the Moon's surface would be an option as well.

The place where you would be really SOL without a paddle would be the surface of Mars!

[Archibald]

If you only include LOC/LOM cases (or ones that would be), then STS does not vary from the Delta II and Tsyklon-2. Atlas II only had about 60 launches, so not enough to compare.

The plain old Ariane 1 / 2 / 3 / 4 flew 144 times, of which 116 were Ariane 4.
Of the 144, seven failed, but five of these failures were traced to the third, cryogenic stage (ESA had hard times with the HM-7).
So the lower composite only failed two times over 144 flights.
Of the two failures, one in 1980 was pogo during a test flight, it was solved, and never happened again.
The other was a human mistake: Flight 36, February 24, 1990. A worker forgot a piece of cloth in a Viking coolant tube.   
So the failure rate sounds very much like the shuttle. The bitter irony is that the old Arianes were never considered man-rated, and neither was the Viking.
Ariane 5 and the Vulcain were supposed to be man-rated, but when one looks at Ariane 5 early history (1996 - 2004) it is not exactly re-assuring. Ariane 5 beginnings were rather catastrophic.

Looking at Ariane history makes me think that man-rating is a rather subjective notion.

[Rhyshaelkan]

Zubrin is getting older and Mars does not seem any closer. Unfulfilled dreams can make a person bitter.

[aquanaut99]

Looking at Ariane history makes me think that man-rating is a rather subjective notion.

Man-rating is entirely subjective. The old Mercury-Atlas and Gemini-Titan would never pass current man-rating requirements, and I suspect Saturn IB and Saturn V wouldn't, either.

And STS-1 would be a total no-go, putting astronauts on an untested rocket with new and untested engines on the very first flight, for Pete's sake!