Galrahn over at Information Dissemination has some 'interesting' thoughts on the second engine fire/explosion to occur to US Navy/Marine F/A-18C's within a month. Read his whole article at his place but these are the statements that caught my attention.
My second thought is how the Joint Strike Fighter would have been FUBAR once the engine goes out. It is hard to believe the US Navy is still going down the one engine path.
My final thought is what a good job the pilot did getting the Hornet back on deck. There really is lots of good stuff here. Well done to the Navy for releasing the video, even in what is ultimately bad news (an engine blows up on a flight deck) there is a lot of great stuff in this simple video.Wow.
An old airplane has suffered two engine mishaps in less than a month and we get into a debate on which is better...single or twin engines?
Galrahn knows better and has access to NAVAIR studies on the issue. As Derek stated on his blog...
Two engines are not any better than one. It is a myth that all twin jets can survive on one engine. Not all systems are redundantly powered by both engines. Also if the failure is catastrophic it could kill the other engine too.As far as the part about the pilot doing a good job...I agree.
If you understand probability if you have two engines with the same rate of failure then you are doubling the chance of an engine failure over a single engined aircraft.
The F-8, A-4, and A-7 had similar safety records as all the other twin engine aircraft of their generation......
As far as there being alot of good stuff on the video...totally disagree. He lauds the crash crew for getting quickly to the crash but to be honest (unless he's never been up close to flight ops)...crash crews are on alert for all landings at military sites. Additionally even if they weren't they'd be alerted after a pilot aborted his landing (especially during night ops)...but lastly...the main focus should be on a second engine going tits up. I'm really surprised we aren't seeing a safety stand down so that these engines can be inspected. Could ops tempo be interfering with fleet safety?
These last two accidents show the need for a new aircraft with new engines. The F-404 engines have been rebuilt and re-used for years. They, much like the F/A-18A, B, C, D have been very reliable, but are nearing the end of their useful life, a life that continually gets extended. The more the F-35 is delayed, the more often we are going to see complete engine or other vital parts failures on these legacy aircraft.
ReplyDeleteThis comment has been removed by the author.
ReplyDelete(irritating typo corrected)
ReplyDeleteDerek was being kind. He could have driven his point home with some elaboration on how the fuel and other system complexities (and failure modes/probabilities) increase with numbers of engines.
Glad you liked my earlier comment BTW, Thanks
I know we mostly talk about NAVAIR and stuff like that but i think one point that should be made is one of the most successful fighters in the history of aviation, the F16 has one engine, and has been a corner stone of airforce military aviation for decades. As much as i like to see new aircraft come online, i do have a great appreciation for the planes and designers that came before and the single engine F 16 has more than proven its worth and proud service for the history books.
ReplyDeletejust my view though.
"Two engines are not any better than one. It is a myth that all twin jets can survive on one engine. Not all systems are redundantly powered by both engines. Also if the failure is catastrophic it could kill the other engine too.
ReplyDeleteIf you understand probability if you have two engines with the same rate of failure then you are doubling the chance of an engine failure over a single engined aircraft."
That is an engineering problem, not a problem with twin engine aircraft in general. They should be designed to operate on one engine like civilian aircraft.
That takes a 15% rate of an aircraft losing an engine and crashing to a 2.3% rate. (Using the F-15's reported 15% engine failure rate).
And to point out the tie-in of the comments so far,one reason the F-16 HAS done so well is that the INSTALLED reliability of the early P&W F100 engines in the F-15 gave the AF impetus to fix the F100 before it got to the F-16, and allowed GE to get back into the game on what was essentially a do-over of the first F100 vs F101 competition. Note: I used the term 'installed reliability' because the F100 was really a tour-de-force in design and capability and met ALL requirements specified. Unfortunately, it was also so powerful and responsive it allowed F-15 drivers to take it where no one ever anticipated, and it was in the 'corners' of the expanded envelope was where new problems arose. As a result, the AF learned it needed to add more specifications to ensure high R&M at higher performance levels, and (sidetepping the political aspects) the 'Great Engine War' was on...and the evolved F100-200 and F110 designs emerged. I'm still trying to re-locate a quote from an AF General I once had a reference to that essentially said "gas turbine engine couldn't be considered mature technology until the F100/F101". If anyone has the source, I'm not too proud to take it, but in the meantime, I'm still looking.
ReplyDeleteI would be remiss if I did not also point out that you cannot use twin engine civilian airliner design considerations as a basis for a fighter design. Besides the fact that they are completely different engine designs producing thrust on completely different scales in different ways (airliner non-afterburning, very high bypass designs vs. fighter afterburning, low-bypass design) and are optimized for completely different design regimes (airliner high altitude steady-state cruising vs fighter medium-to-high altitude,maneuvering, with large throttle transients), the airliners larger size allows system design redundancy not available to smaller aircraft. Electrical power example: the B777 has one primary generator on each engine, a backup generator on each engine also contains two permanent magnet generators that power three flight control power supplies,there is an APU generator that can be run in flight and if all else fails, there is a ram-air turbine. Even with all the redundancies of all its systems, the B777 can only fly so many minutes at any given time from an alternate landing field in case a problem arises and then oly under special operatin rules. To allow overwater operations > than 1hr away from a landing field, an Extended Twin Engine Operations (ETOPS) aircraft has special maintenance and dispatch requirements. None of the above is a luxury that fighter operations or design can afford.
ReplyDeleteand that's why i love you guy...you can blast through the nonsense that would leave me befuddled!
ReplyDeleteSMSgt Mac...you're hero qualified!
apperantly some people dont understand probability theory, they say "If you understand probability if you have two engines with the same rate of failure then you are doubling the chance of an engine failure over a single engined aircraft."
ReplyDeletewell, ummm...no.
http://en.wikipedia.org/wiki/Independence_%28probability_theory%29
in probability theory there are events that are independent and not independent. An engine flaming out with a 15% failure rate would not double the chances of failure.
so if you have two dice, the chances of getting a number is 1 in 6, well rolling the second dice is independent of the other, and thus also 1 in 6. now with an aircraft one could say they are conditionally independent, as they aren't completely independent (same maintenance records if in the same plane for same amount of time, same flight time, etc). needless to say it doesn't double the chances.
Also one would have to asses whether the probability of the loss of power is worth the added abilities of the jet with its second engine.
RE: "If you understand probability if you have two engines with the same rate of failure then you are doubling the chance of an engine failure over a single engined aircraft."
ReplyDeleteThe statement is ‘True’ IF…
1.) they are all the exact same engine, 2.) The engines system support (fuel, oil, controls) operation on the twin-engine aircraft are completely independent and, 3.) The engines are operated exactly the same way under the same stresses and 4. Catastrophic loss of one engine will not 'take out' the other engine.
Critical engine reliability is expressed as Mean Time Between Critical Failure (MTBCF), measured in failures/flight hour. A MTBCF of .001 = 1 critical failure every 1000 flight hours on average (mean). If you have an airplane with an engine MTBCF of 1000 hrs, you will have on average 1 critical engine failure every 1000 hrs. If you have two of the same engines in the case described you will experience 2 critical failures on average every 1000 hrs (hopefully not at the same time). You are in effect rolling each engine’s die at the same time…every hour. What the loss of an engine on most twin-engined jet aircraft has historically meant is a reduction in the loss of aircraft (aka hull loss) due to an engine loss compared to single engine aircraft - i.e. the consequences are less for the twin in the case described. (On older general aviation light twins it is said that the remaining engine will always get you to the scene of the crash). Once critical engine reliability increases past a certain point, the reliability of the support and other systems can become the dominant factors in the total system (aircraft) reliability equation. The added complexity of the twin installation could actually be more unreliable in its totality than the total single engine installation. Add to the equation an increased ability to monitor engine condition ‘real-time’ and also predict/anticipate critical engine failures before it occurs and the advantage of twin engine system reliability is further diminished or eliminated. You can still claim lower aircraft losses due to and engine loss with the twin, but you can’t claim fewer aircraft losses overall.
Sorry if I rambled, this area is one of my ‘things’. I actually prefer to use Fault Tree Analyses (FTAs) over Failure Modes, Effects, and Criticality Analyses (FMECAs) as a basis for Failure Modes Effects Testing (FMET) and avoid dealing with MTBCFs. FMECAs depend upon the meticulous rollup of MTBCFs from the lower levels but tend to mask combined systemic factors IMHO. FTAs require one to focus on ONLY that which can cause the ‘very bad things’ to happen and the required Functional Analyses performed at the front-end as a prerequisite will reveal what is truly important within complex systems, promotes verification of graceful system degradation without loss, identifies design target areas to prevent catastrophic failure modes and generally prevents a lot of charging down inconsequential rabbit holes.
what is the variance around the mean? i am not asking to criticize but just wondering as i am doing applied stats in school and interested in it.
ReplyDeleteDrat. You replied before I could correct an error I realized I had committed in my eager rambling -it hit me after I logged off last night.
ReplyDeleteI should have typed:
Critical engine reliability is expressed as Mean Time Between Critical Failure (MTBCF), measured in failures/flight hour. A MTBCF of .001 = 1 critical failure every 1000 flight hours on average (mean). If you have an airplane with an engine MTBCF of 1000 hrs, you will have on average .5 critical engine failure every 1000 hrs. If you have two of the same engines in the case described you will experience 1 critical failure on average every 1000 hrs (hopefully not at the same time).
I'd expect the mature fleet SD to be much < 1, with a very short right 'tail' for non-induced failures, althouht the distribution for very short snapshots in time could not resemble anything near a normal distribution.
so if the hours keeps increasing so does the chances of it having a failure, as this seems to be almost exponential in its distribution? or would proper maintenance limit the exponential nature of that?
ReplyDeleteRE: exponential distribution.
ReplyDeleteYes…‘sorta’...If you are talking the TRENDLINE of cumulative % failures over time from 0 to 100%. In reality there will be waves, steps and notches reflecting discrete failure mode sub-trends. We are dealing with finite sets of 'things' that have a finite set of failure modes (even if some may be unknown until they actually happen) and failure mode sub-trends. In the case of engines, we have mechanical parts, heat and friction, which means there is always an upper limit of time for things to go wrong. The exponential curve is a good approximation of what the trend line cumulative % failed over time would look like if we are dealing with fielded failures--after the initial checkout/verification and test at the manufacturer (which would contain 'infanticide' failures). If you match the fielded failure curve with the ‘at birth’ failure curves, it would resemble what R&M engineers call the ‘Bathtub Curve’.
The management of engine condition prior to actual critical failures will stretch the trend line along the x axis (time) compared to a 'no maintenance/servicing’ paradigm and is all about heading off critical failures before they happen and ensuring indicators of impending doom are found. Good management of engine condition can stretch the useful life and the bathtub curve significantly, but then IMHO you are reducing the value of using statistics at this high level to describe failure distributions because if you change the engine at all, how much change (classic ‘Sorites Paradox’) makes the engine a ‘different’ one?
Using word pictures is hard work: I've usually gone to a whiteboard long before now ;-).
lol, i love the white board, its hard to teach statistics without it :D HAHA, but its good to talk to someone who knows this stuff :D.
ReplyDeletedo you use time failure models, sounds like this would use censored data if you was looking at time to failure, i took event history analysis, and we used like log normal, weibull and other distributions to model certain things, but this was in the social sciences and not engineering.
hope you dont mind my questions :)
"Two engines are not any better than one. It is a myth that all twin jets can survive on one engine. Not all systems are redundantly powered by both engines. Also if the failure is catastrophic it could kill the other engine too."
ReplyDeleteThere are several qualifiers in this thata make it misleading.
"Not all twin jets..." Okay, but what was the last twin engine American fighter aircraft that could not maintain flight on one engine?
"Could kill the other engine too...", yeah it COULD, but it may not and if a single engine a/c has an issue it definately will.
"Not all systems are redundantly powered...", maybe not but the point is that if you lose an engine you can safely RTB or at least get it down and save the pilot and the a/c, even if you cannot complete the original mission. This is why it is considered more wise for NAVAIR a/c to be twins instead of singles - if you are running blue water ops, there are no divert fields and if you have to punch out the ocean is not as safe as dry land. SMSgtMac - Do some research into why the Marine's Cobra's have been twin engine variety instead of single engine.
I just noticed Privateer’s throwaway line directed at me but snuck in at the end of his comment on someone else’s quote: “SMSgtMac - Do some research into why the Marine's Cobra's have been twin engine variety instead of single engine.”
ReplyDeletePfft. I got yer’ research right here....
Fortunately I don’t have to do too much research (mostly just verification of memories from my youth). Also fortunately, your suggested example serves to illustrate some of what I’ve already asserted. The Marines sold the twin-engine idea on ‘safety’ (it was about other things as well) but we’ll focus on the safety aspect. It looks good on paper to have two engines, and the Marines moved ahead with the idea in ’69. So did it work out as they expected?
PART 1
The Marines found out they couldn’t get the reliability they needed out of the box. The AH-1J used the same main rotor as on the single engine ‘G’, but had a new gearbox and (I believe) tail rotor systems come over from the Model 212. The net effect of all the changes between the models was that the sea-level horsepower % increased about three times as much as the drive train load rating % and the empty weight % increased about 1.5 times the load rating % increase (gross weight went up even a higher %). This combination is not good if you are seeking to increase reliability – and it wasn’t. Add to the mix the fact that the overwhelmingly and single worst and most frequent catastrophic failure mode of that era was found where the engines tied into the rotor drive system, (‘spur’ gear if I remember correctly) followed by the rotor gearbox. Add to this, the fact that in case of engine loss, the surviving engine would have to provide emergency power beyond what it was rated for in sustained operation for a limited time, and that could damage the remaining engine: so it wouldn’t usually buy you much time from a safety POV. Before the last ‘J’ model was delivered, the Marines were trying to improve the result.
The improvement came in the form of the AH-1T, which was basically adding the Model 214A’s bigger main and tail rotor assemblies and an upgraded P&W Twin-Pack engine setup. This meant they had to stretch the ‘T’ a bit and do a lot more structural modification to make it work. This had the effect of a more balanced drivetrain in the sense that the system could take all the power the engines could provide, which bode well for reliability, but this sucker also had a lot more weight to haul around. The Marines learned a lesson from the ‘J’ so they levied a new R&M program on top of the ‘T’ to make sure they’d get the reliability they needed. Unfortunately, now the Marines weren’t getting the performance they wanted out of the total package: they were either limited in altitude or payload.
PART 2
ReplyDeleteTo solve the ‘T’s problem, Bell ginned up a design using even bigger engines, this time from GE and that finally brought the Marines the AH-1W, which finally resembled what they were looking for in the first place: capability and reliability.
The ‘J’ was first flown in ’69, and the ‘W’ was fielded in ’86. 17 years to get from A to B, with a third or more of that time with a lower or equal reliability than what you started with. BTW. the reason the twin-engine huey/cobra drivetrain even exists is because the Canadians wanted a heavy lifting Huey that could carry a lot of people on nice cold low-density altitude days, and there were no likely single-engine options with the equivalent SHP. If P&W Canada had a one engine candidate instead of the twin PT6s, I wonder if the twin engine Huey/Cobra would have ever existed?
“Privateer’s” case illustrates a classic problem with redundancy and how new failure modes and rates can be introduced as well as the unintended consequences of adding redundancy (new failure modes, maintenance problems, performance shortfalls). All evidence points towards spending your money on safety training and procedures as a more effective way to lower catastrophic loss rates. Twin engines are nothing more than a security blanket if you ground personnel and your aircrew are careless. Finally, single engine historical data doesn’t support the ‘twin-engine=increased safety’ idea, but finds such an assertion faith-based. (Although in the interests of stirring another hornets’ nest (pun intended), there’s an old study that does find a high correlation for ‘two crew’ fighters have lower accident rates than single-pilot systems.)
jp: Sorry I got all wrapped up in the snark from the comment after yours and forgot to reply in answer to your question. RE: ”Do you use time failure models, sounds like this would use censored data if you was looking at time to failure, i took event history analysis, and we used like log normal, weibull and other distributions to model certain things, but this was in the social sciences and not engineering.”
ReplyDeleteThe answer is ‘it depends’. Designers I’ve been involved with tend to rely on ‘time failure’ models, esp. ‘accelerated failure time models’ but I’ve seen ‘proportional hazard’ as well. Both support Weibull Analysis, and as I’m not a designer I couldn’t tell you the true balance of use rate between the two. When I have my tester hat on, I’m concerned with testing the validity of the predictions. When I have my Ops Research hat on, I’m concerned with reliability impact on mission effectiveness and reliability.
In military Aerospace, we work very hard to get real (vs. predicted) data and performance as fast as possible. For hardware, a designer may start with MIL-HDBK-217 (with the suffix depending on topic/type of hardware), select a design approach based upon reliability requirements of the device, and predict the initial reliability of the design. The R&M engineer does a Failure Modes, Effects and Criticality Analysis on the released design to arrive at the official ‘prediction’ of reliability, usually using an appropriate approach(es) selected from the Reliability Toolkit, a handbook published by the Reliability Analysis Center (now Reliability Information Analysis Center). This handbook, if one doesn’t already have a copy has been superseded by the System Reliability Toolkit, which pays more attention to software reliability, but IMHO still has serious limitations from a ‘system’ POV, and good luck getting a software engineer to agree to a stringent definition of ‘reliability’ (one area the commercial world does not ‘get’). Usually there are one or two tweaks to the design required to achieve predicted reliability requirements. But it is still ‘predicted’-- until we subject the item/subsystem to forms of life-cycle testing to simulate the expected operating environment and then actually ‘fly’ it to validate the predictions. We collect operational reliability data on all components of the system as the fleet matures, and make changes as required to reach required mature system reliability (usually 100,000 total flight hours or more for combat aircraft).