100% Success Rate for Reflown Boosters

[john smith 19]

Since the first re-launch was on parity (0.01% higher) with new, this could be lost in the noise of insurers and their internal competition -- which is uninteresting.

As I understand it, Falcon rates are on par with Ariane et al whose reliability is higher than Falcon, so absolute reliability is not a clean discriminator for insurance rates (or vise versa).  Ed's statistics will be better.

Let's say that sounds quite doubtful given that A5 has now launched 70+ launches in a row without a mishap and SX is on it's 2nd F9 RTF.

I'd be very surprised if that didn't have quite a lot of bearing on insurance premiums.

Now as flight data on re-flown booster stages accumulates I would also expect that to have a strong bearing on insurance rates, provided there are no US mishaps.

Ariane 5 had several failures in its early years, both partial and complete failures.

As you never cease to remind us.

Except that A5 is not in it's "early years" any more. A 70+ successful launch history is a significant selling point for a vehicle that only flies once.

Let's hope F9 is starting to come out of it early years and first stage reuse accelerates its maturing process. Fortunately stage reuse should be able to raise the launch rate quite considerably, if there are sufficient payloads to launch at the new price.

[john smith 19]

Ed Kyle, do you have any breakdown for launch failures, across the stage of flight during which they occurred; roughly 'boost' vs. 'upper'?

For Spacex, 100% upper stage.

You crack me up, Jim!  😂

But he does have a point. This is a step change in how we think of TSTO stage reliability.

Historically stage failure typically meant mission failure (unless stage propellant loads were up to the limit of the payload and the target trajectory parameters in case they had to compensate).

How do we measure this increased reliability wrt to expendable stages, given that historically it's been all or nothing? 

[dglow]

You have a point, John Smith, though Jim did not.

How will we measure reliability of expendable stages? At a high level, exactly as we do today. What will change is the tracking of 'lifetime' data for boosters, which boosters are active, which are retired, and the variations between them (block-5, etc.).

Imagine the stats that will result: 'median missions flown for active block-X boosters in the fleet'.

If any first stage failure occurs, expect a lot of attention will be paid, right or wrong, to the stats of that particular booster.

[meekGee]

How long before insurance rates on flight-proven boosters are lower than insurance rates on unflown boosters?

Meaningless point, since the second stage is expendable and operates 3 times longer than the first stage.

What's meaningless?

"How long before insurance rates on flight-proven boosters are lower than insurance rates on unflown boosters?"

The insurers's opinion is what matters, and apparently even on the first re-use, the rates were pretty much the same.

[Kaputnik]

You have a point, John Smith, though Jim did not.

How will we measure reliability of expendable stages? At a high level, exactly as we do today. What will change is the tracking of 'lifetime' data for boosters, which boosters are active, which are retired, and the variations between them (block-5, etc.).

Imagine the stats that will result: 'median missions flown for active block-X boosters in the fleet'.

If any first stage failure occurs, expect a lot of attention will be paid, right or wrong, to the stats of that particular booster.

We didn't find this difficult for STS, which was also a mixture of expendable and reflown components.

[john smith 19]

You have a point, John Smith, though Jim did not.

How will we measure reliability of expendable stages? At a high level, exactly as we do today. What will change is the tracking of 'lifetime' data for boosters, which boosters are active, which are retired, and the variations between them (block-5, etc.).

Imagine the stats that will result: 'median missions flown for active block-X boosters in the fleet'.

If any first stage failure occurs, expect a lot of attention will be paid, right or wrong, to the stats of that particular booster.

Indeed. I expect special attention to wheather the boosters have had regular servicing (which Musk said can give them a 100+ launches) or minimal necessary, in which case Musk thought maybe 10 would be OK.

Obviously a booster with minimal servicing will be cheaper for SX but with statistical variation in things like TPS thickness and quality I'd suggest anyone after say the 8th would start to be pushing the edges of probable failure.

We didn't find this difficult for STS, which was also a mixture of expendable and reflown components.

With a modern ERP system I wouldn't expect collecting and tracking the data will be any problem. I strongly doubt SX has the 100+ individual data bases (some still manual) that Boeing found when it studied Shuttle servicing in the mid 80's

It's the idea of a "full service history" coming to TSTO VTO rockets. I think some who are used to expendables will find it a bit odd, although I'm sure they will get used to it.

[ChrisWilson68]

How long before insurance rates on flight-proven boosters are lower than insurance rates on unflown boosters?

Meaningless point, since the second stage is expendable and operates 3 times longer than the first stage.

It's only meaningless if the risk for the first stage is zero.  Even if the risk from the second stage is higher, if there's any contribution to overall mission risk from the first stage, insurance rates should be different if that risk is higher or lower.

[meekGee]

You have a point, John Smith, though Jim did not.

How will we measure reliability of expendable stages? At a high level, exactly as we do today. What will change is the tracking of 'lifetime' data for boosters, which boosters are active, which are retired, and the variations between them (block-5, etc.).

Imagine the stats that will result: 'median missions flown for active block-X boosters in the fleet'.

If any first stage failure occurs, expect a lot of attention will be paid, right or wrong, to the stats of that particular booster.

Indeed. I expect special attention to wheather the boosters have had regular servicing (which Musk said can give them a 100+ launches) or minimal necessary, in which case Musk thought maybe 10 would be OK.

Obviously a booster with minimal servicing will be cheaper for SX but with statistical variation in things like TPS thickness and quality I'd suggest anyone after say the 8th would start to be pushing the edges of probable failure.

We didn't find this difficult for STS, which was also a mixture of expendable and reflown components.

With a modern ERP system I wouldn't expect collecting and tracking the data will be any problem. I strongly doubt SX has the 100+ individual data bases (some still manual) that Boeing found when it studied Shuttle servicing in the mid 80's

It's the idea of a "full service history" coming to TSTO VTO rockets. I think some who are used to expendables will find it a bit odd, although I'm sure they will get used to it.

How do you know that the "10 flights" doesn't already include your consideration?

So that the actual number is maybe 15, and only flying 10 takes care of statistical variations with sufficient confidence?

[edkyle99]

What was the highest single year for any launcher in history?
Would topping that (in a non Cold War environment) be a more significant benchmark?

The busiest single year by an orbital class launch vehicle would likely be 63 launch attempts (including two failures) by the R-7 family in 1980.  This number does not include the March 18, 1980 pad explosion of 8A92M Vostok-2M at Plestesk, the result of a fueling accident prior to launch, that killed 48.  The breakdown was 45 Soyuz U, six Vostok 2M, and 12 Molniya M (the two failures were Molniya M launches).

I don't think we will see a number like that again for the foreseeable future.

 - Ed Kyle

[meekGee]

What was the highest single year for any launcher in history?
Would topping that (in a non Cold War environment) be a more significant benchmark?

The busiest single year by an orbital class launch vehicle would likely be 63 launch attempts (including two failures) by the R-7 family in 1980.  This number does not include the March 18, 1980 pad explosion of 8A92M Vostok-2M at Plestesk, the result of a fueling accident prior to launch, that killed 48.  The breakdown was 45 Soyuz U, six Vostok 2M, and 12 Molniya M (the two failures were Molniya M launches).

I don't think we will see a number like that again for the foreseeable future.

 - Ed Kyle

Of course we will.  Unless by "foreseeable" you mean "2018".

By all accounts, the constellations will require multiple launches per week.

Which brings up a new disparity:

The time to recover from a launch failure remains more or less constant.  But with a launch tempo like this, the impact of an offline pad will be huge.  This argues for more pads.  I bet we'll hear about those in the next year.

[JamesH65]

What was the highest single year for any launcher in history?
Would topping that (in a non Cold War environment) be a more significant benchmark?

The busiest single year by an orbital class launch vehicle would likely be 63 launch attempts (including two failures) by the R-7 family in 1980.  This number does not include the March 18, 1980 pad explosion of 8A92M Vostok-2M at Plestesk, the result of a fueling accident prior to launch, that killed 48.  The breakdown was 45 Soyuz U, six Vostok 2M, and 12 Molniya M (the two failures were Molniya M launches).

I don't think we will see a number like that again for the foreseeable future.

 - Ed Kyle

Basically just over one per week. Although there were at least three different configurations, not sure how that compres with the F9 family of, basically, one. Still a great cadence from, I think, 4 different pads.

I'd expect SpaceX to start to get close to that in a year or two, if they speed up the US ranges reset time, and have the payloads ready.

[Robotbeat]

I expect 13 launches this year, 18 next. 25, 30, 35, 40 by 2022. Barring a failure, which will almost certainly happen in the next 100 launches, so within the next 3 or 4 years. 2025 is earliest they'll equal R7 rate, and actually they may have moved on to Raptor/ITS-based rockets by then, so would reset the clock on reliability (and to some extent, launch rate).

[AncientU]

I expect 13 launches this year, 18 next. 25, 30, 35, 40 by 2022. Barring a failure, which will almost certainly happen in the next 100 launches, so within the next 3 or 4 years. 2025 is earliest they'll equal R7 rate, and actually they may have moved on to Raptor/ITS-based rockets by then, so would reset the clock on reliability (and to some extent, launch rate).

Quite the high resolution crystal ball you've got there!!! 
(not even one launch per pad per month -- by 2022 -- hummm)

EM thinks the 4,425 satellite constellation (alone) will take 50 flights per year.  New application brings total sats to 270% of that figure.
GS expects to launch every two weeks from each launch pad.
Both plan for a 24hr turn around of boosters.

Payloads, launch facilities, and boosters will not be limiting it seems, so R-7 just may get challenged.
At least EM and GS think it will.

[Robotbeat]

I expect 13 launches this year, 18 next. 25, 30, 35, 40 by 2022. Barring a failure, which will almost certainly happen in the next 100 launches, so within the next 3 or 4 years. 2025 is earliest they'll equal R7 rate, and actually they may have moved on to Raptor/ITS-based rockets by then, so would reset the clock on reliability (and to some extent, launch rate).

Quite the high resolution crystal ball you've got there!!!

Just because I say "18" doesn't mean I think it's exactly that. I'm not intending to imply high accuracy. Maybe +/- 50%.

(not even one launch per pad per month -- by 2022 -- hummm)

Trying to be realistic. It's crazily ambitious no matter which way you slice it.

EM thinks the 4,425 satellite constellation (alone) will take 50 flights per year.  New application brings total sats to 270% of that figure.

Yeah, and the initial-operation constellation will be smaller. And I kind of think ITS will be used for the Constellation perhaps before other non-test payloads, and I was referring primarily to Falcon 9, intentionally excluding ITS. Also, stuff always takes longer than I, you, or Elon expect (which doesn't mean it won't happen, just a bit later).

GS expects to launch every two weeks from each launch pad.

2022 is only a few years away. Also, I don't think all pads will be equally busy. Texas isn't even built yet. It's primarily a big dirt pyramid. Maybe 2 years before it's ready for launches, and then a few years after that before it has ramped up. Meanwhile, LC-40 will need to be rebuilt by sometime this summer, and LC-39a will need upgrades for Falcon Heavy and then Commercial Crew (and perhaps later ITS) that will slow down launch rate. West Coast may not get that busy.

Both plan for a 24hr turn around of boosters.

Could easily take more than 5 years to get there.

Payloads, launch facilities, and boosters will not be limiting it seems, so R-7 just may get challenged.
At least EM and GS think it will.

Eventually it will. And I hope SpaceX will be the ones to do it.

I'm trying to do an accurate launch count projection. Last two years I've guessed slightly too optimistically, and before that I was almost spot on.

[AncientU]

Thanks for the details of your thinking.  Can't argue against any of them -- the future is just so damned empirically deficient.  I hope your estimates are the low end of the range because time isn't on the side of anyone wanting to make sweeping changes.

[FutureSpaceTourist]

Bumping this thread, after another 5 years …, to say still 100% success rate for reflown boosters and now past the 100 mark!

[AmigaClone]

I expect 13 launches this year, 18 next. 25, 30, 35, 40 by 2022. Barring a failure, which will almost certainly happen in the next 100 launches, so within the next 3 or 4 years. 2025 is earliest they'll equal R7 rate, and actually they may have moved on to Raptor/ITS-based rockets by then, so would reset the clock on reliability (and to some extent, launch rate).

Comparing some of your predictions vs what happened.
Launches.
2017 - Prediction 13, Actual 18
2018 - Prediction 18, Actual 21
2019 - Prediction 25, Actual 13
2020 - Prediction 30, Actual 28
2021 - Prediction 35, Actual 31
2017 t0 2021 - Prediction 121, Actual 111 overall within 10%

2022 - Prediction 40, Actual 34 (as of 8 August 2022). Potentially could challenge the R-7 launch rate in 1980.

Failures - 0 mission failures. While there have been 6 landing failures and 1 successful landing where the booster was destroyed on the way back to port, that has not affected the launch rate.

SpaceX has demonstrated  two launches in 14 days from all three F9 pads, (LC 4, LC 39A, and LC40), but appears to have decided not to pursue trying to refurbish a booster in 24 hours.

[Surfdaddy]

Still looks valid to my eyes.  Those experiments were not free.  $1 billion, according to Elon on March 30 after the SES 10 flight.  That is an extra $10 million per-flight surcharge over 100 flights to recover.   Not free.

 - Ed Kyle

This is the same short term, stock oriented thinking that allowed ULA to be so stagnant.  It is not about the 1 billion now.  It is about a sustainable 20-50 launches per year that leads to billions in the future.  When you have 20 stages sitting around in warehouses, production rate is not the determining factor.

You counter with the same wild-eyed optimistic projections that appeared during the early STS years, during the go-go "Little-LEO" times, etc. 

In all history, only two launch vehicles have ever crossed the 20 per year barrier (R-7 and Thor/Delta) over an extended number of years.  Both did so as Cold War machines launching now-obsolete film-return reconnaissance satellites during the 1960s-1980s.  No others have come close since the end of the Cold War.   China's DF-5 based CZ family currently leads the way, flying about 16.9 times per year (annual average since 2010, inclusive).  R-7 is currently logging about 16.1 per year.  Proton is flying 8.6 times per year average.  Atlas 5 is at 6.1 flights per year.  Ariane 5 is at 5.9 annually.

 - Ed Kyle

20 per year for several years sounds like a benchmark that will persuade you of Falcon's potential.
We'll come back to this in a few years...

I've bolded a couple of key phrases from this 2017 exchange in post #67.
It's now been a few years. It seems that AncientU was correct, and Ed Kyle was mistaken

[FutureSpaceTourist]

Interesting discussion at about the 25m 25s mark (of Crew-6 briefing below), about some unexpected fire seen in an F9 engine bay on a Starlink flight.

It was a 12th flight and there was evidence of combustion seen in post flight inspection. Speculation is may be a slight O2 leak, perhaps due to number of flights. Benji Reed said something like this has also been seen before.

So interesting data on potential effects of repeated reuse.

For the record NASA, are fine with crew mission, given analysis and testing performed.  Also crewed flights must be on at most the 5th flight of a booster.



Edit to add: the issue seen on the 12th flight did not cause any problems with the flight. The engine did not shutdown early.

[JamesH65]

What was the highest single year for any launcher in history?
Would topping that (in a non Cold War environment) be a more significant benchmark?

The busiest single year by an orbital class launch vehicle would likely be 63 launch attempts (including two failures) by the R-7 family in 1980.  This number does not include the March 18, 1980 pad explosion of 8A92M Vostok-2M at Plestesk, the result of a fueling accident prior to launch, that killed 48.  The breakdown was 45 Soyuz U, six Vostok 2M, and 12 Molniya M (the two failures were Molniya M launches).

I don't think we will see a number like that again for the foreseeable future.

 - Ed Kyle

This aged well.