After today's events can we form any kind of view whether it is better to have multiple engines on the first stage of a modern rocket or if it is best to have just one or two?To me having multiple engines equals more things that could go wrong but also means on the other hand more redundancy if something does go wrong.
Quote from: Star One on 10/08/2012 07:01 pmAfter today's events can we form any kind of view whether it is better to have multiple engines on the first stage of a modern rocket or if it is best to have just one or two?To me having multiple engines equals more things that could go wrong but also means on the other hand more redundancy if something does go wrong.It seems to me that there are benefits and drawbacks to both approaches, from the point of view of performance, reliability, and cost. And the jury's still out.
Soyuz-U, the worlds most reliable booster, has multiple engines.
Aircrafts have multiple engines. Elon loves to compare his vision with the cost of a 747, so why not compare the number of engines and overall security increase?
<snip>Engine redundancy works well for aircraft because uncontained failures are very rare. Rockets are higher pressure systems with lower margins.
Quote from: savuporo on 10/08/2012 07:38 pmSoyuz-U, the worlds most reliable booster, has multiple engines.N-1, arguably the world's least reliable booster, also had many multiple engines. You cannot make a simple argument based on one vehicle (that only has five primary engines).
To the OP: Good idea.
Quote from: strangequark on 10/08/2012 08:11 pmQuote from: savuporo on 10/08/2012 07:38 pmSoyuz-U, the worlds most reliable booster, has multiple engines.N-1, arguably the world's least reliable booster, also had many multiple engines. You cannot make a simple argument based on one vehicle (that only has five primary engines).Yes, statistically insignificant number of attempts like N-1 dont really contribute to the argument, but a booster that has flown more than 700 flights, with the highest ever achieved launch rate actually sort of does.
Quote from: savuporo on 10/08/2012 08:17 pmQuote from: strangequark on 10/08/2012 08:11 pmQuote from: savuporo on 10/08/2012 07:38 pmSoyuz-U, the worlds most reliable booster, has multiple engines.N-1, arguably the world's least reliable booster, also had many multiple engines. You cannot make a simple argument based on one vehicle (that only has five primary engines).Yes, statistically insignificant number of attempts like N-1 dont really contribute to the argument, but a booster that has flown more than 700 flights, with the highest ever achieved launch rate actually sort of does.Really has no bearing on Falcon 9, and doesn't even inform the broader discussion. The success of one launch vehicle, with 5 engines tells you nothing about the general trend, especially if you don't delve into the particulars of those failures. Otherwise, you could equally argue that the Proton K, with 300+ launches (statistically significant, yes?), and a ~15% failure rate for a 6 engine vehicle argues against larger numbers of engines. Like I said before, what you really need to make a valid assessment is to look at how likely it is that failure of one engine will be such that it can result in fratricide of neighboring engines. So, in answer to the original poster:Multiple engines on the first stage, good or bad idea?Can be a good idea, but there is a tip over point where redundancy reduces reliability, and that is vehicle and engine dependent. That tip-over point is likely to be smaller, rather than larger, because first-stage rocket engines are high pressure, low margin devices, which increases the chances of an uncontained failure.
From a back seat point of view, it might be a good idea to run these engines to the breaking point to see just what that is, do they even know?
Quote from: JohnFornaro on 10/08/2012 08:05 pmTo the OP: Good idea.Agree, on balance. Also the multiple engine set up has turned out fortuitous for the proposed reuseable first stage design. But of course SpaceX had to use multiple engines to get the performance they needed and made a virtue out of a necessity.
4. Redundancy: More engines means less risks for the mission outcome. That was shown today.
V1.1 Falcon will have redundancy improvements because Merlin 1D can throttle down instead of switching off 2 engines as for now.What happened yesterday would have not bring consequences even to Orbcomm if it were a Falcon 1.1, because no engine were cut off and you could have brought (say) from 80% times 9 engines to 90% times 8 engines.
To me it seems as if the long shadow that the N1 and its multiple failures casts is one that has maybe set some against the idea of rockets with multiple engines on the first stage.
DIRECT had a point with engine out for RL-10, which is a low pressure expander cycle engine (simple, and benign cycle) that they were going to use for EDS.
Quote from: thydusk666 on 10/08/2012 10:46 pm4. Redundancy: More engines means less risks for the mission outcome. That was shown today.That wasn't shown. Ask Obcomm.
The alternate outcome to this launch wasn't a perfect orbital insertion for Orbcomm, it was a smoking hole in the ocean (ala Taurus last year).
Quote from: strangequark on 10/08/2012 08:11 pmDIRECT had a point with engine out for RL-10, which is a low pressure expander cycle engine (simple, and benign cycle) that they were going to use for EDS. We also had engine out capability with the core stage, able to loose a single engine after 31 seconds of flight and still make orbit.I was then, and remain today, an avid supporter of multi-engine MPS for launch vehicles - to a point. I don't know where that point is yet but imho a launch vehicle should be able to loose 1 or 2 main engines, depending on the number in the cluster, and still make orbit.
Correct if I'm wrong here, but I don't think any launch vehicle has ever been designed engine-out capability as a requirement; it just happens to be a by-product of clustering smaller engines as a means of achieving higher thrust. Case in point: the Soviet Union in the 60's decided it was more feasible to cluster NK-33 engines rather than dealing with combustion stability challenges associated with a F-1 sized thrust chamber. Fast forward 40 years later: the Falcon 9 was built as a scaled up version of Falcon 1, utilizing as much commonality as possible (including Merlin engines).
Indeed, and I certainly don't think that 1 engine is optimal for first stage either. However, there is a tipover point, and I'm biased toward thinking it's lower with first stages. Falcon 9 did demonstrate engine out this time, but there was a lot of debris flying around those other engines. Nine is fair number of engines (27 is huge if they ever fly it), and the proof will only come with flight rate. The positive is that they sure as hell produce a lot of data points every flight .
Quote from: strangequark on 10/09/2012 04:30 amIndeed, and I certainly don't think that 1 engine is optimal for first stage either. However, there is a tipover point, and I'm biased toward thinking it's lower with first stages. Falcon 9 did demonstrate engine out this time, but there was a lot of debris flying around those other engines. Nine is fair number of engines (27 is huge if they ever fly it), and the proof will only come with flight rate. The positive is that they sure as hell produce a lot of data points every flight .2 * 27 = 54 enginesE(engine failure every) = 54/40 = 1.35 flightsAt a 1 in 40 engine failure rate that means a statistical expectation of an engine failure in the Falcon Heavy on more than half the flights.
What happened yesterday would have not bring consequences even to Orbcomm if it were a Falcon 1.1, because no engine were cut off and you could have brought (say) from 80% times 9 engines to 90% times 8 engines.
To me it seems as if the long shadow that the N1 and its multiple failures casts is one that has maybe set some against the idea of rockets with multiple engines on the first stage. Though to me this is somewhat unfair considering the time and circumstances that contributed to the issues of the N1.
A point to remember though, there is a certain flight regime where ANY loss of engines will result in LOM and falling out of the sky, main engine numbers are not redundant during the first few seconds of flight.
The philosophical wrangling of whether a high-multiple engine design came about through one thought process or another is a pointless debate. All such decisions rest on multiple considerations.
Minimum # of engines 2 – gives you 1 engine out reliability
Quote from: FuseUpHereAlone on 10/09/2012 12:34 amCorrect if I'm wrong here, but I don't think any launch vehicle has ever been designed engine-out capability as a requirement; it just happens to be a by-product of clustering smaller engines as a means of achieving higher thrust. Case in point: the Soviet Union in the 60's decided it was more feasible to cluster NK-33 engines rather than dealing with combustion stability challenges associated with a F-1 sized thrust chamber. Fast forward 40 years later: the Falcon 9 was built as a scaled up version of Falcon 1, utilizing as much commonality as possible (including Merlin engines). Yeah, you're wrong. Falcon-9 was indeed designed to survive engine out - that's why the on-board computer was ready and able to recompute how to use the remaining engines to meet its principle orbit insertion requirements. The philosophical wrangling of whether a high-multiple engine design came about through one thought process or another is a pointless debate. All such decisions rest on multiple considerations.
True Engine out reliability.So Elon likes to compare F9 to commercial aircraft most often the 747, I’m going to use the 777 to make things simpler. It is my understand that with any twin engine jet 777, 737, 757 etc… each engine needs to be able to produce 2x the thrust that is needed for take off. This is done in order to ensure the ability to take off and conclude the flight should an engine out occur at any point during the flight, including the critical time beyond the point of no return on take off.
Quote from: Joffan on 10/09/2012 04:43 amQuote from: FuseUpHereAlone on 10/09/2012 12:34 amCorrect if I'm wrong here, but I don't think any launch vehicle has ever been designed engine-out capability as a requirement; it just happens to be a by-product of clustering smaller engines as a means of achieving higher thrust. Case in point: the Soviet Union in the 60's decided it was more feasible to cluster NK-33 engines rather than dealing with combustion stability challenges associated with a F-1 sized thrust chamber. Fast forward 40 years later: the Falcon 9 was built as a scaled up version of Falcon 1, utilizing as much commonality as possible (including Merlin engines). Yeah, you're wrong. Falcon-9 was indeed designed to survive engine out - that's why the on-board computer was ready and able to recompute how to use the remaining engines to meet its principle orbit insertion requirements. The philosophical wrangling of whether a high-multiple engine design came about through one thought process or another is a pointless debate. All such decisions rest on multiple considerations.Nope, you are wrong. It was not a design requirement but a fallout
Quote from: Jim on 10/09/2012 11:28 amQuote from: Joffan on 10/09/2012 04:43 amQuote from: FuseUpHereAlone on 10/09/2012 12:34 amCorrect if I'm wrong here, but I don't think any launch vehicle has ever been designed engine-out capability as a requirement; it just happens to be a by-product of clustering smaller engines as a means of achieving higher thrust. Case in point: the Soviet Union in the 60's decided it was more feasible to cluster NK-33 engines rather than dealing with combustion stability challenges associated with a F-1 sized thrust chamber. Fast forward 40 years later: the Falcon 9 was built as a scaled up version of Falcon 1, utilizing as much commonality as possible (including Merlin engines). Yeah, you're wrong. Falcon-9 was indeed designed to survive engine out - that's why the on-board computer was ready and able to recompute how to use the remaining engines to meet its principle orbit insertion requirements. The philosophical wrangling of whether a high-multiple engine design came about through one thought process or another is a pointless debate. All such decisions rest on multiple considerations.Nope, you are wrong. It was not a design requirement but a falloutOh, interesting. Didn't know that. Source please?
Quote from: beancounter on 10/09/2012 02:06 pmQuote from: Jim on 10/09/2012 11:28 amQuote from: Joffan on 10/09/2012 04:43 amQuote from: FuseUpHereAlone on 10/09/2012 12:34 amCorrect if I'm wrong here, but I don't think any launch vehicle has ever been designed engine-out capability as a requirement; it just happens to be a by-product of clustering smaller engines as a means of achieving higher thrust. Case in point: the Soviet Union in the 60's decided it was more feasible to cluster NK-33 engines rather than dealing with combustion stability challenges associated with a F-1 sized thrust chamber. Fast forward 40 years later: the Falcon 9 was built as a scaled up version of Falcon 1, utilizing as much commonality as possible (including Merlin engines). Yeah, you're wrong. Falcon-9 was indeed designed to survive engine out - that's why the on-board computer was ready and able to recompute how to use the remaining engines to meet its principle orbit insertion requirements. The philosophical wrangling of whether a high-multiple engine design came about through one thought process or another is a pointless debate. All such decisions rest on multiple considerations.Nope, you are wrong. It was not a design requirement but a falloutOh, interesting. Didn't know that. Source please?The size of the existing Merlin engine dictated that 9 would be needed to meet mission requirements. When you have multiple engines, you can have engine out capability. If Spacex went to Merlin 2 right away, it would not have the ability to have engine out capability.Spacex mantra for Falcon 1 was one engine for reliability. They had to modify it for Falcon 9.
Spacex mantra for Falcon 1 was one engine for reliability. They had to modify it for Falcon 9.
Definitely good idea.Imagine the falcon 9 had only one engine. The mission would have been a total failure, and perhaps more importantly: they would only have found out about the design or QA error in the engine after the ~40th flight, which might have been a flight with a much more valuable payload.Using many engines is a prerequisite to getting high reliability per engine. And with the current extremely low launch rates (compared to any other mode of transportation), having multiple identical engines per stage is a good way to increase the production rate, with all the benefits that brings.
I agree with you.There are other design and manufacturing advantages over those you said:- multiple engines means distributed loads, hence simpler thrust structure (SpaceX will probably use this advantage in v1.1)- mid size engines means cheaper machine tools being used- multiple smaller engines means shorter engine bay
Quote from: cambrianera on 10/09/2012 07:50 pmI agree with you.There are other design and manufacturing advantages over those you said:- multiple engines means distributed loads, hence simpler thrust structure (SpaceX will probably use this advantage in v1.1)- mid size engines means cheaper machine tools being used- multiple smaller engines means shorter engine bayThere are even more advantages. The merlin engines can be worked on by placing them on a simple cart, similar to what you would do with a large truck engine.They can be rolled around the factory floor. Compare that with something the size of an F-1 or RD-170 where every single fabrication step needs a specialized fixture or an oversized crane.All these are small improvements, but they do add up.Once they do something larger than falcon heavy, they might have to upsize the engines. But the falcon heavy has a lot of potential by improving the upper stage. So I hope they stick with the relatively small engine as long as possible.
The distinction between an original specification point and the choice of meeting that specification with redundancy and engine-out ability
Quote from: Joffan on 10/09/2012 09:06 pmThe distinction between an original specification point and the choice of meeting that specification with redundancy and engine-out abilityWrong again. Engine out capability is bug and not a designed in feature
My gut is telling me that they really need to ramp up these engines on the test bed and blow a few up.
Quote from: giggleherz on 10/09/2012 11:10 pmMy gut is telling me that they really need to ramp up these engines on the test bed and blow a few up.They did what you want long ago. They use their test stands quite often. They do single, two, three, five, all nine. They do shorter and longer burns. They do multiple burns of same engines in row (you know, reuseability kinda requires engine capable of being used again). They had also their share of RUDs.
Quote from: Jim on 10/09/2012 09:49 pmQuote from: Joffan on 10/09/2012 09:06 pmThe distinction between an original specification point and the choice of meeting that specification with redundancy and engine-out abilityWrong again. Engine out capability is bug and not a designed in featureIt may not have been their original reason to go with 9 engines - but that statement is simply wrong, and even a layman can see that. Once you have a need for multiple engines, you have a deliberate decision to make if you need or want to support engine out capability.You may see the need for more than one engine as a "bug" - but it is quite clear that a deliberate effort from SpaceX with regards to hardware and software design to support engine out capability paid off significantly in this last flight.(I suppose I could have simply written "wrong" like you would have, so you'll have to excuse me)
No I agree with Jim here. You don't design in an engine failure. No commercial aircraft is designed just in case they have an engine failure. That's an off-nominal event and SpaceX shouldn't have considered that as part of the design criteria.
No I agree with Jim here. You don't design in an engine failure. No commercial aircraft is designed just in case they have an engine failure. That's an off-nominal event and SpaceX shouldn't have considered that as part of the design criteria. They would have built their engines not to fail. The fact that they lv can still make orbit under 8 engines is, I think, a by-product, not a deliberate design criteria.
No I agree with Jim here. You don't design in an engine failure. No commercial aircraft is designed just in case they have an engine failure.
Well if it was actually designed with engine-out capability as a requirement, then the CRS-1 flight proves that Falcon 9 has not met this requirement.
Quote from: FuseUpHereAlone on 10/10/2012 02:01 amWell if it was actually designed with engine-out capability as a requirement, then the CRS-1 flight proves that Falcon 9 has not met this requirement.I'm choosing to believe that you wrote this as a joke, because the other alternative is too depressing.
Why do we believe Falcon is a high reliability design?The vast majority of launch vehicle failures in the past two decades can be attributed to three causes: engine, stage separation and, to a much lesser degree, avionics failures. An analysis of launch failure history between 1980 and 1999 by Aerospace Corporation showed that 91% of known failures can be attributed to those subsystems.Engine ReliabilityIt was with this in mind that we designed Falcon I to have the minimum number of engines. As a result, there is only one engine per stage and only one stage separation event – the minimum pragmatically possible number.In the case of Falcon V, there are five first stage engines, but the vehicle is capable of sustaining an engine failure at any point in flight and still successfully completing its mission. This actually results in an even higher level of reliability than a single engine stage. The SpaceX five engine architecture is an improved version of that employed by the Saturn V Moon rocket, which had a flawless flight record despite losing an engine on two of its missions.
You forgot the smiley-face...
But did they ever blow one up intentionally to see what would happen to the others?
how many tested when at supersonic speeds?
They would have built their engines not to fail.
If the engine is 100% reliable, then a single engine would be best.
I thought the Orbcomm boost was scrapped due to safety concerns regarding the ISS, rather than a technical limitation of the SpaceX.Statistically, the answer is clear: Multiple engines are more reliable as long as the failure of one engine does not cause the failure of others, or the loss of mission. A 9 engine vehicle is 9 times more likely to have an engine failure than a 1 engine vehicle (see note), so the probability that the failure will effect other engines must be lower than 1 in 9. Note: One would also assume that a production run 9 times greater will result in more reliable engines. Whether it is more cost effective remains to be seen. If the engine is 100% reliable, then a single engine would be best. What is the most reliable engine to date? Would it be the space shuttle SRBs with 1 failure in over 200 launches? However, solids have unpleasant failure modes, so what is the most reliable liquid fuelled engine?
The 2nd burn was automatically canceled due to the fuel check after the release of dragon. There was not enough fuel to get it to the approved orbit.
Apples-to-oranges, Taurus failed because of a lack of payload fairing separation not engine failure.
At a 1 in 40 engine failure rate that means a statistical expectation of an engine failure in the Falcon Heavy on more than half the flights.
It was not a design requirement but a fallout
The size of the existing Merlin engine dictated that 9 would be needed to meet mission requirements.
They designed for redundancy. They designed in engine separation barriers. They designed in in-flight recovery. Sure sounds like they designed for loss of an engine.The distinction between an original specification point and the choice of meeting that specification with redundancy and engine-out ability is precisely what I was referring to as "philosophical wrangling".
Engine out capability is bug and not a designed in feature
It would be pointless to test all engines to destruction - you would have no engines left to actually fly.
It really all comes down to whether or not fratricide is common with your typical failure modes, and the jury is out on that one.
Quote from: FuseUpHereAlone on 10/10/2012 02:01 amWell if it was actually designed with engine-out capability as a requirement, then the CRS-1 flight proves that Falcon 9 has not met this requirement. Depends on what you mean by "it". The CRS-1 mission appears OK, and thus F9 met its nominal goals; the rest is up to Dragon and is TBD. The ORBCOMM mission was not OK and thus F9 did not meet its nominal goals.Queue interminable argument about "launch" vs. "flight" vs. "mission" vs. etc...
Quote from: Lars_J on 10/10/2012 02:05 amQuote from: FuseUpHereAlone on 10/10/2012 02:01 amWell if it was actually designed with engine-out capability as a requirement, then the CRS-1 flight proves that Falcon 9 has not met this requirement.I'm choosing to believe that you wrote this as a joke, because the other alternative is too depressing. Alternatively, could you say that Falcon 9 has engine-out capability for one payload, but not two?
Quote from: Lars_J on 10/10/2012 02:05 amQuote from: FuseUpHereAlone on 10/10/2012 02:01 amWell if it was actually designed with engine-out capability as a requirement, then the CRS-1 flight proves that Falcon 9 has not met this requirement.I'm choosing to believe that you wrote this as a joke, because the other alternative is too depressing. This isn't a joke, just a simple exercise in determining design requirements, and validating them based on performance. Otherwise we’re just spinning the truth.
Well if it was actually designed with engine-out capability as a requirement, then the CRS-1 flight proves that Falcon 9 has not met this requirement. Proof: The CRS-1 flight profile called for Falcon 9 to separate from Dragon in one orbit, boost to another orbit, then separate from the Orbcomm satellite. The first stage suffered an engine-out event. Falcon 9 successfully releases Dragon, but lacks enough propellant to boost the Orbcomm satellite to its proper orbit (per the flight profile). If Falcon 9 could not execute the entire mission with an engine-out, then engine-out does not allow Falcon 9 to meet its rated performance. Therefore, Falcon 9’s rated performance cannot be attained with an engine-out.The point here is that you can’t just cluster a bunch of engines together and say that it has “engine-out” capability. If it does, then it better perform as planned when one or more of those engines cuts out. I suspect that a Falcon 9 with engine-out designed into it would really have performance numbers similar to a Falcon 8 (an imaginary Falcon 9 with one unused engine). Maybe a Falcon 7 if we wanted 2 engine-out capability.
And the main disadvantage is that the chance of an engine failure, especially during the first couple of launches, are much higher.
I've developed a new pet peeve since the launch. Am I the only that prefers the industry euphemism RUD not be used ever again? Has to be something better.
It isn't an industry euphemism, but a Spacex one
I totally agree with you FuseUpHereAlone, when the maneuver for the Orbcomm satellite wasn't performed because there wasn't enough propellant (margin) left to execute it. But it they didn't do it for ISS safety reasons, I disagree.
That said, I think using multiple engines on a stage has pros and cones. Others have given good points for this. I think the main benefit are cost savings, and increased reliability over time. And the main disadvantage is that the chance of an engine failure, especially during the first couple of launches, are much higher.
{snip}Quote from: A_M_Swallow on 10/09/2012 04:59 amAt a 1 in 40 engine failure rate that means a statistical expectation of an engine failure in the Falcon Heavy on more than half the flights. In other words, they cannot learn from their mistakes and improve the engine to reduce the failure rate?
Quote from: JohnFornaro on 10/10/2012 02:39 pm{snip}Quote from: A_M_Swallow on 10/09/2012 04:59 amAt a 1 in 40 engine failure rate that means a statistical expectation of an engine failure in the Falcon Heavy on more than half the flights. In other words, they cannot learn from their mistakes and improve the engine to reduce the failure rate?That is a different equation.My commentary. SpaceX (and FAA) had better learn from their mistakes.Engine rateLaunch anomaly Rate1:27 Every launch1:2701:101:27001:100
Quote from: JBF on 10/10/2012 12:06 pmThe 2nd burn was automatically canceled due to the fuel check after the release of dragon. There was not enough fuel to get it to the approved orbit.There was not enough fuel to get it to the approved orbit with sufficient margin.
Quote from: A_M_Swallow on 10/10/2012 09:48 pmQuote from: JohnFornaro on 10/10/2012 02:39 pm{snip}Quote from: A_M_Swallow on 10/09/2012 04:59 amAt a 1 in 40 engine failure rate that means a statistical expectation of an engine failure in the Falcon Heavy on more than half the flights. In other words, they cannot learn from their mistakes and improve the engine to reduce the failure rate?That is a different equation.My commentary. SpaceX (and FAA) had better learn from their mistakes.Engine rateLaunch anomaly Rate1:27 Every launch1:2701:101:27001:100All the above is simply jumping to conclusions on virtually no data, i.e. speculation. You all need to wait until SpaceX has examined their data and determined root cause. That will lead them to whatever fixes they need to make to prevent or reduce the possibility of such a situation in the future. There's no evidence yet to base any conclusions on whatsoever.
Quote from: peter-b on 10/10/2012 12:56 pmQuote from: JBF on 10/10/2012 12:06 pmThe 2nd burn was automatically canceled due to the fuel check after the release of dragon. There was not enough fuel to get it to the approved orbit.There was not enough fuel to get it to the approved orbit with sufficient margin.In which case, under ordinary circumstances, they would have tried it? They'd have a choice between try it and maybe fail, or not try it and definitely fail.
With enough data it becomes less like rocket science and more like complicated statistics.
Quote from: Jim on 10/10/2012 07:27 pmIt isn't an industry euphemism, but a Spacex oneA NewSpace euphemism I think, but it's intended as humour, just like "engine-rich exhaust".
In which case, under ordinary circumstances, they would have tried it? They'd have a choice between try it and maybe fail, or not try it and definitely fail.
It really all comes down to whether or not fratricide is common with your typical failure modes, and the jury is out on that one. Engine redundancy works well for aircraft because uncontained failures are very rare. Rockets are higher pressure systems with lower margins.
Would it be more economically beneficial for them, if they had a slightly more powerful engine and could still do engine-out?
Quote from: upjin on 10/11/2012 07:05 pmWould it be more economically beneficial for them, if they had a slightly more powerful engine and could still do engine-out?Not at the moment, since that engine is not on the shelf, ready to use. Remember too, that a different number of engines also means a different thrust structure, different plumbing, and different flight software. All of that would have to be designed, built, and tested.And Martijn: You owe me a cup of green tea. After reading "engine rich exhaust", too many giggles and snorts, while holding the cup.
The Company has filed a notice of claim under its launch insurance policy for a total loss of the OG2 prototype. The maximum amount covered by the policy is $10 million, which would largely offset the expected cost of the OG2 prototype and associated launch services and launch insurance.
Even if you have fratricide, it may only be one other engine that is damaged, which may mean you can still make orbit.Btw, it is obvious that in order for engine-out on falcon 9 to make sense, the risk of fratricide must be LESS THAN an order of magnitude less than typical failure. In your calculation, proponent, you arbitrarily made the assumption that fratricide was only an order of magnitude less likely than regular engine failure, which unsurprisingly leads to the conclusion that the reliability is no better than a single engine. (not taking into account increase in reliability due to higher production volume) if, say, fratricide only occurs for 5% of failures, then 9 engines wins.
4) The failure rate for 1 big engine is the same as the individual failure rates of small engines. (Not true, but close enough).
Well, at least they got some testing done on the prototype."Notwithstanding the shortened life of the OG2 prototype, the OG2 program engineering teams from ORBCOMM, Sierra Nevada Corporation and Boeing made significant strides in testing various hardware components. After telemetry and command capability was established, several criticalsystem verifications were performed. The solar array and communications payload antenna deployments were successful, along with verifying the performance of various components of both the OG2 satellite bus and the communications payload. The OG2 satellite bus systems includingpower, attitude control, thermal and data handling were also tested to verify proper operation. The unique communications payload, which incorporates a highly reprogrammable software radio with common hardware for both gateway and subscriber messaging, also functioned as expected.These verification successes achieved from the single prototype satellite validate that the innovative OG2 satellite technology operates as designed before launching the full constellation of OG2 satellites. With this verification data, ORBCOMM can focus on completing and launching the OG2satellites as the primary mission payloads on two planned Falcon 9 launches, the first in mid-2013 and the second in 2014, directly into their operational orbit."
We don't know yet what happened but if these engines take a beating and wear out fast, is it really worth it to spend so much time and money to make them reusable.
Quote from: upjin on 10/11/2012 08:26 pmWell, at least they got some testing done on the prototype."Notwithstanding the shortened life of the OG2 prototype, the OG2 program engineering teams from ORBCOMM, Sierra Nevada Corporation and Boeing made significant strides in testing various hardware components. After telemetry and command capability was established, several criticalsystem verifications were performed. The solar array and communications payload antenna deployments were successful, along with verifying the performance of various components of both the OG2 satellite bus and the communications payload. The OG2 satellite bus systems includingpower, attitude control, thermal and data handling were also tested to verify proper operation. The unique communications payload, which incorporates a highly reprogrammable software radio with common hardware for both gateway and subscriber messaging, also functioned as expected.These verification successes achieved from the single prototype satellite validate that the innovative OG2 satellite technology operates as designed before launching the full constellation of OG2 satellites. With this verification data, ORBCOMM can focus on completing and launching the OG2satellites as the primary mission payloads on two planned Falcon 9 launches, the first in mid-2013 and the second in 2014, directly into their operational orbit."why would an insurance policy pay out to the full amount, when they clearly got a lot (or enough) value out of the satellite to mean they don't need to fly another one?seems to me like Orbcomm should be wringing their hands at the moment, complaining bitterly about how this is going to affect their bottom line, and how they may need to launch another test satellite..
Quote from: KingAlbert on 10/11/2012 08:35 pm4) The failure rate for 1 big engine is the same as the individual failure rates of small engines. (Not true, but close enough).With the same budget and time, there is no way to design, develop, produce, test 1 big engine for first stage and 1 small engine for upper stage, yet have the same failure rates as merlin engine in 9+1 falcon 9 configuration.
vertical landing not only allow reusable rocket, also they can build mars lander for NASA if ever need one. kinda kill 2 birds with 1 stone. It's all about business strategy.
Quote from: cordor on 10/11/2012 09:24 pmQuote from: KingAlbert on 10/11/2012 08:35 pm4) The failure rate for 1 big engine is the same as the individual failure rates of small engines. (Not true, but close enough).With the same budget and time, there is no way to design, develop, produce, test 1 big engine for first stage and 1 small engine for upper stage, yet have the same failure rates as merlin engine in 9+1 falcon 9 configuration.You have no proof of that
{snip}However, if SpaceX makes any significant upgrades to engine thrust for the 1 series or gets the Merlin 2 out, they would have to rethink the 9 engine configuration and likely go to fewer engines.I'm also wondering about the 9 engine configuration from an economic stand point. Let's suppose they already had that 1E or Merlin 2 engine, how many engines would still give them engine-out capability AND be cheap to build? Seems like 5 or 6 is that magic number, in theory.
You are more than welcome to proof me wrong, explain to me how you can use the same amount of resources to get the job done. teach me how specialize can be more cost effective than modularize especially at design and development phase. In fact, i really want to know. Im begging you.To me, if i can reuse the same engine everything, that's perfect. Earlier, i even point out micro engine like this one. RCS use it, escape system use it, stage 1 stage 2 use it.
Using standard reliability analysis we can calculate what the reliability of a multi-cluster engine relative to one big engine.For simplicity’s sake let us assume the following:1) In a multi-engine design the individual engines can fail without causing an automatic failure of the entire stage.
2) Engine failures are independent of each other, ie the failure of any one engine has no effect on the failure probability of any other engine. This is probably not totally realistic as the failure of an engine would probably stress the remaining engine more, but it’s probably close enough for these calcs.
3) The timing of the failure is treated as irrelevant, failure in the 1st second of burn is as bad as failure during the last second of burn. (Not true, but close enough).
5) For a 9 engine cluster we will assume that the failure of any 1 engine won’t cause a launch failure and the failure of 2 engines might not if they are on opposite sides of the cluster. Assume the engines are laid out:
why would an insurance policy pay out to the full amount, when they clearly got a lot (or enough) value out of the satellite to mean they don't need to fly another one?
You need to give up this line of reasoning. "Not true" is your hint, particularly when making assumptions about one engine with thrust capability of the nine. "Close enough" is where your mistaken assumptions can be found, if you dig deeper. You know about GIGO; you can't use standard reliability analysis without appropriate assumptions.
Quote from: KingAlbert on 10/11/2012 08:35 pmUsing standard reliability analysis we can calculate what the reliability of a multi-cluster engine relative to one big engine.For simplicity’s sake let us assume the following:1) In a multi-engine design the individual engines can fail without causing an automatic failure of the entire stage.You can't "assume" that, you have to engineer that.
Quote from: KingAlbert on 10/11/2012 08:35 pm2) Engine failures are independent of each other, ie the failure of any one engine has no effect on the failure probability of any other engine. This is probably not totally realistic as the failure of an engine would probably stress the remaining engine more, but it’s probably close enough for these calcs.Another element of engineering, not assumption, which you sorta ackowledge with the squishy term "probably".
Quote from: KingAlbert on 10/11/2012 08:35 pm3) The timing of the failure is treated as irrelevant, failure in the 1st second of burn is as bad as failure during the last second of burn. (Not true, but close enough).Not anywhere near close enough without a more complete analysis.
Quote from: KingAlbert on 10/11/2012 08:35 pm4) The failure rate for 1 big engine is the same as the individual failure rates of small engines. (Not true, but close enough).You need to give up this line of reasoning.
Quote from: JohnFornaro on 10/12/2012 01:03 pmYou need to give up this line of reasoning. "Not true" is your hint, particularly when making assumptions about one engine with thrust capability of the nine. "Close enough" is where your mistaken assumptions can be found, if you dig deeper. You know about GIGO; you can't use standard reliability analysis without appropriate assumptions.It's called Fermi estimation, and I found his modelling interesting. I particularly liked that he stated his assumptions and where shortcomings might be found. Since your tone is that of someone who thinks they know what they are talking about, some constructive criticism (i.e. suggesting improved models and some possible sources of data) would perhaps be in order, rather than just saying "your assumptions are bad and you should feel bad". I wish there were more NSF forums who put as much thought into their posts as KingAlbert. We should encourage them, rather than driving them away with unnecessary negativity and confrontational attitude.
Since your tone is that of someone who thinks they know what they are talking about, some constructive criticism...
Quote from: Jim on 10/12/2012 01:15 amQuote from: cordor on 10/11/2012 09:24 pmQuote from: KingAlbert on 10/11/2012 08:35 pm4) The failure rate for 1 big engine is the same as the individual failure rates of small engines. (Not true, but close enough).With the same budget and time, there is no way to design, develop, produce, test 1 big engine for first stage and 1 small engine for upper stage, yet have the same failure rates as merlin engine in 9+1 falcon 9 configuration.You have no proof of thatYou are more than welcome to proof me wrong, explain to me how you can use the same amount of resources to get the job done. teach me how specialize can be more cost effective than modularize especially at design and development phase. In fact, i really want to know. Im begging you.To me, if i can reuse the same engine everything, that's perfect. Earlier, i even point out micro engine like this one. RCS use it, escape system use it, stage 1 stage 2 use it. With the quantity, we can do mass production and bring engine cost to dirt cheap. Of cause, plumbing becomes a problem, and john smith 19 pointed out it's not ready yet.
You seem to have a misunderstanding about what my model is for. It's to get a rough sense of the reliability implications of 9 ... vs 1 engine. I don't work for spacex so I don't know the exact details of its' design, I have to "assume" things just to get started. I'm the one making the assumptions, and being open about it, I'm can't be sure about what spacex has actually built. I don't even pretend to know the engine reliability numbers (an incredibly important number), which is why the calcs cover a range of reliability numbers for the engines.
In the case where the F9 cluster doesn't have engine-out capability then the calcs become pretty trivial and we can say for sure it's much less reliable to have more engines. Not very interesting.
With regards to F9 is particular, they've stated that have engine-out capability and they seem to have, at least partially, displayed it, so I decided to include that. If you're sure that's not true then refer to my previous paragraph.
I did post a 2nd set of calcs that also included the probability of a catasrophic failure from a single engine out with an assuming 10:1 ratio of benign:catastrophic failure.
Quote from: JF on 10/12/2012 01:03 pmQuote from: KingAlbert on 10/11/2012 08:35 pm2) Engine failures are independent of each other, ie the failure of any one engine has no effect on the failure probability of any other engine. This is probably not totally realistic ... but it’s probably close enough for these calcs.Another element of engineering, not assumption ...Given that the engines are designed to burn for much longer than they're actually used, I doubt that the small increase in burn time ... would significantly alter the failure rate of the remaining 8 engines. A small increase in failure rate for that case simply won't alter the final number much. If you have proof that there is a big change in failure rates, please provide it and I'd be happy to take it into account.
Quote from: KingAlbert on 10/11/2012 08:35 pm2) Engine failures are independent of each other, ie the failure of any one engine has no effect on the failure probability of any other engine. This is probably not totally realistic ... but it’s probably close enough for these calcs.Another element of engineering, not assumption ...
2) Engine failures are independent of each other, ie the failure of any one engine has no effect on the failure probability of any other engine. This is probably not totally realistic ... but it’s probably close enough for these calcs.
Quote from: JohnFornaro on 10/12/2012 01:03 pmQuote from: KingAlbert on 10/11/2012 08:35 pm3) The timing of the failure is treated as irrelevant, failure in the 1st second of burn is as bad as failure during the last second of burn. (Not true, but close enough).Not anywhere near close enough without a more complete analysis.I'm making a very pessimistic assumption. If timing did matter then that would mean that a 'late' failure might allow the 9-engine design to withstand 2 or 3 late engine failures, thus improving the 9-engine failure rate. As it is I just have the pessimistic case.
Quote from: JohnFornaro on 10/12/2012 01:03 pmQuote from: KingAlbert on 10/11/2012 08:35 pm4) The failure rate for 1 big engine is the same as the individual failure rates of small engines. (Not true, but close enough).You need to give up this line of reasoning. Given that the "one big engine" for a F9 is purely hypothetical, good reliability numbers be hard to come by. In any event, it's completely trivial to see the effect of the big engine having different reliability numbers so I don't even see the point of including that.
I predict that there will be a Merlin 1E, the 1D with reliability enhancements. Since the 1D already exists the new version number will be needed for public relations purposes.
...pinnacle...
I realize that you hope to derive a rough sense of reliability, which is fine in principle. Yes, you are "open" about your assumptions, but the assumptions are so crucial to the bottom line of your calculation, that I end up questioning the point of the exercise.As I concluded in my previous post, you don't have the most important number at all, which is not your fault by any stretch; it is proprietary knowledge. Without their reliabilty numbers, I continue to struggle with your results, and question them on the basis of "why bother?", when so much is unknown.
QuoteIn the case where the F9 cluster doesn't have engine-out capability then the calcs become pretty trivial and we can say for sure it's much less reliable to have more engines. Not very interesting.A case which is clearly, by design, not the case. I'm engaging on this to try and understand why calculate about that which is not the case?
Burn time is one part of the equation, and the engines are conservatively rated so that burn time does not influence the failure rate, but I didn't read your sentence the same way you did. You haven't accounted for fratricide. You will have to find out what that rate should be, and plug that it. I have no idea what would be a reasonable figure.
Quote from: KingAlbert on 10/11/2012 08:35 pmGiven that the "one big engine" for a F9 is purely hypothetical, good reliability numbers be hard to come by. In any event, it's completely trivial to see the effect of the big engine having different reliability numbers so I don't even see the point of including that. Then, what's the value of the comparison?
Given that the "one big engine" for a F9 is purely hypothetical, good reliability numbers be hard to come by. In any event, it's completely trivial to see the effect of the big engine having different reliability numbers so I don't even see the point of including that.
Let me try and help you out a bit here......maybe move these micro engines to another project? See this thread..... http://forum.nasaspaceflight.com/index.php?topic=28369.30
Quote from: JohnFornaro on 10/12/2012 03:12 pmI realize that you hope to derive a rough sense of reliability, which is fine in principle. Yes, you are "open" about your assumptions, but the assumptions are so crucial to the bottom line of your calculation, that I end up questioning the point of the exercise.... Without [SpaceX's] reliabilty numbers, I continue to struggle with your results, and question them on the basis of "why bother?", when so much is unknown.Spacex is never going to release detailed information about engine reliability, as you yourself said. Never. People on the outside will always be engaging in speculation and educated guesses. I still find it interesting, but if you have to have hard, detailed, and very reliable numbers, I think you're out of luck.
I realize that you hope to derive a rough sense of reliability, which is fine in principle. Yes, you are "open" about your assumptions, but the assumptions are so crucial to the bottom line of your calculation, that I end up questioning the point of the exercise.... Without [SpaceX's] reliabilty numbers, I continue to struggle with your results, and question them on the basis of "why bother?", when so much is unknown.
Quote from: JohnFornaro on 10/12/2012 01:03 pmQuote from: KingAlbert on 10/11/2012 08:35 pmGiven that the "one big engine" for a F9 is purely hypothetical, good reliability numbers be hard to come by. In any event, it's completely trivial to see the effect of the big engine having different reliability numbers so I don't even see the point of including that. Then, what's the value of the comparison?The value of the model is to see the reliability implications of of 9 engine/1 out design, which is interesting in itself. Other people might be able to add more information in the future to improve the analysis.
One engine out in 39, actually. The last three flights of FalcOn 1 used merlin1c and they didn't have an engine out.Edit: also, because of the small sample size, we can't say with certainty that the real reliability is 1/39. As far as we know, the real reliability of merlin 1c could be somewhere between 90% and 99.9%. So we don't have enough of a sample size to say whether theirs is a good design choice versus a single engine even assuming the rest of the analysis is correct.
.................... just exactly what is the definition of a rocket engine?
There are two crucial parameters in the stage failure rate calculation:1) the probability that an engine failure is catastrophic and destroys the whole stage2) the number of benign engine failures that the stage can tolerateIn order to have good stage failure probability, you need to have both1) probability of catastrophic engine failure much less then 1/#of engines(here, 1/20 vs 1/9, not significantly smaller)and2) the probability of too many benign failures (P9 in the code) should be much less than the probability of single engine failure. Here, it is also not the case - essentially because 40 is not much smaller than (9 choose 2) /9. This term would get much better if you can tolerate 2 failures out of 9 engines (which might be the case for late failures, although it is not the case for early ones).In any case, stage failure rate vs engine failure rate gets much better as the engine reliability improves .. 1/40 is not good enough.
It is in the equations in the code mentioned above.P9 is the probability of stage failure due to too many benign failures.If the stage tolerates only one bening failure, the leading (largest, most important) term is (nchoosek(9,2)-4).*Pf.^2.*Ps.^7, which is close to (9 choose 2)*1/(40^2)=0,0225, i.e. about 1/44, which is almost the same as single engine failure probability. In other words, even if we assume the are no catastrophic engine failures, stage reliability would be only cca 98%.On the other hand, if the stage tolerated 2 benign failures, the leading term in P9 would be approximately (9 choose 3)/40^3 ~ 0.0013, i.e. stage reliability (assuming no catastrophic failures) of cca 99.87%.Alternatively, with only 5 engines and one engine out tolerance, the stage failure rate (again assuming 1/40 engine failure rate) would be about (5 choose 2)/1600 ~ 0,00625, i.e. stage reliability of 99.375%.