Implications for the 787-9’s first flight

The first flight of the next member of the Boeing 787 family, the -9, is coming soon and this has greater implications than usual for what would be called a minor model variant.


Boeing has been making “minor model variants” since the dawn of the jet age with the 707: the 707-120/320/420, the 727-100/200, the 737-100/200, the 300/400/500 family and so on right through today’s 787 variants.


The first flight of the 787-9 ordinarily would be a little consequence. But because of the painful birth of the lead variant, the 787-8, and its troubled early service life that included a 3 ½ month grounding, the 787-9 will have greater scrutiny to see if Boeing has the program troubles behind it.


Boeing made significant changes to the 787-9 based on lessons learned from the -8. Changes amount to a reported 30% of the design and include lessons learned from the design and production of the tail plane, the side-of-body wing join, elements of the wing’s internal structure, some electronics and, of course, the lithium ion batteries that led to the grounding of the 50 -8s in service.


Also the engines. The GEnx and Rolls-Royce Trent 1000 engines fell short of specification for the -8. It remains to be seen whether the -9 will meet the specs with the engine upgrades, or whether more work will need to be done.


Dispatch reliability remains a challenge for the -8, hovering around 98%; Boeing wants to see it closer to 99.7%, the standard set by the 737NG and the 777. This may not sound like much difference, but it means a lot to the airlines, and Boeing is quick to compare the 99.7% for the 737NG with the 99.4% for the rival Airbus A320. This fractional difference amounts to more than 60 flights a year Boeing claims as an advantage for the 737NG over the A320.

A nearly two point difference between today’s 787 and tomorrow’s is significant. The lessons learned, incorporated into the 787-9, will be watched as a path toward this reliability target.


Expectations are high for the -9. We hear that the first plane emerged from the factory pretty “clean,” that is, without the troubles that bedeviled the -8. Let’s hope the flight test program comes off without a hitch.


Nearly 40% of the more than 900 787 orders are for the -9. We expect the -9 and its larger sibling, the -10, to eventually account for more than 50% of the orders for the family.


This was originally sent to our email list on September 3.

69 Comments on “Implications for the 787-9’s first flight

  1. “Changes amount to a reported 30% of the design and include lessons learned from the design and production of the tail plane, the side-of-body wing join, elements of the wing’s internal structure, some electronics and, of course, the lithium ion batteries that led to the grounding of the 50 -8s in service.”

    What about the electrical system? Is anything changed from the -8 to the -9? Is there in fact a need to change anything, or is Boeing right in its claim that there is nothing wrong with the system? In other words are we imagining things or is the system fundamentally flawed?

    In the meantime I have no doubt that the -9 will be a much improved aircraft and will probably eclipse its innovative predecessor because it has higher capacity. But on the other hand the asking price is much stiffer and it should take a while before the total sales of the -9 surpass the record setting number of the initial model.

    Anyway, any sale of the -9 will help absorb the losses on the -8 and make the programme more viable.

  2. The 98% availability is an interesting number which I take with a grain of salt. The problem is that you don’t know how much effort went into achieving this number. Two examples:
    1) SQ had a special team to ensure readiness of the A380 during its introduction. SQ claimed it was the best introduction of a new plane ever, but I know people who quit working for SQ because they had to be available 24/7 and meeting the needs of availability killed their private life. Getting a call at 4 AM in the morning to go to Changi airport to address a fault to ensure that the next flight at 7:30 AM is on time is no fun.
    2) In my previous job we have a complex piece of equipment which had to be running 24/7/365. Contractual uptime under warranty was 97%. I wanted to initiate a project to improve reliability but we had no business case because this 97% was met for all contracts. But what people forgot to mention is that because 97% was in the contract, it had to be met by hook or by crook. The actual maintenance to meet 97% was actually far higher than the forecasted number, but because the 97% KPI was met, no action was deemed to be needed because the amount of maintenance required to meet this number was not measured.

  3. Interesting that you think the -8 will hold 50% of the mkt. I thought I had heard of speculation that it would be less. One of the reasons, IMHO, that the 787’s conception was good, even brilliant (as opposed B’s botch of it’s production) was that it was designed to replace both the 767 and the 330. It looks like the -8 is doing that, and very well, at least in terms of efficiency, even if they may not be up to contract spec. Notwithstanding, the “low” dispatch reliability, it appears that airlines are opening long/thin routes with the -8 (United to China and Nigeria, BA to Austin, Norwegian to the US West Coast), thus confirming the wisdom of B’s estimate in the ’90s that the future was in twin-engined wide bodies for point to point routes. In a recent interview, Tim Clark emphasized that when explained that the reason Emirates is not using flight rights from Asia over the Pacific to the US is that the African mkt was getting so good that they are focusing on it.

  4. NdB agree, the window dressing using “dispatch reliability” is getting annoying, even deceptive if targeting the bigger public.

    You can “easily” improve dispatch reliability of an aircraft type by throwing a tonne of mechanics, spares and reserve aircraft at a fleet. Apparently it is what Boeing did (source WSJ Jon O) to prevent it dispatch reliability collapsing for the 787. E.g. the grounding had nothing to do with dispatch reliability, more so with inflight reliability.

    Everyone has high hopes the -9 will make us forget the Dreamliner development saga. No doubt Boeing put in a lot of effort and the specs look very good. It seems to become a very balanced, right sized long haul machine. The fact Boeing had to change, also results in a reality nobody can hope for a cheap, simple stretch.
    The testing and certification effort will be accordingly. And the risk of new hick ups from a basicly new aircraft will be there.

  5. Boeing claims 98% dispatch reliability. This is only achieved by having hot spares on stand-by. DR only says whether a flight number was on time or not, it is not related to a specifc aricraft serial.

    Statistics seem to prove that 787 in UAL service ‘break down’ (turnback, AOG) 4 times more often than the average of all other airplanes in their fleet. Service availability of 787s seems to be much lower than 98%.

    • Just look at hours flown per day.
      With the numbers made available at the time of the grounding I didn’t get
      beyond 3.5..4 hours per day.
      ( and note EIS is 2 years in the past now )

      • Grounding was 8 months ago and EIS was actually less than 2 years ago, on Oct 26 2011. Your utilization numbers are a bit dated now.

        • 40 days on two years is a delta of ~5%. Nothing against the creative margins used by others.
          If you read around nothing much seems to have changed in respect to frame useage per day. Regular notices on multiple days on the ground due to technical reasons for individual Dreamliners.
          ~18..20 month after EIS A380 utilisation seems to have gone to 10..12 hours and per day and beyond :

          If data massage and judicious weeding of undesirables would provide for a similar report on the 787 we would have seen it by now ( even if Randy had to do it on his own 😉 then I haven’t seen any further “positive” sfc numbers posted either after this iniitial piece on “much better than our 767”.

      • Point is you quote data from before the grounding that is 8 months old, so your delta is more like 35%. Cherry picking worst case numbers and extrapolating those to the entire fleet just to make a point is not how things should be done. Makes me doubt your sparkling A380 numbers. Also, in my view, spin or lack thereof is not data either way.

        Both RR and GE are correcting the SFC issue. This has been widely reported, and if you believe it, the PIP2 for the GE will bring it essentially to target, while the package C for the RR will still fall slightly short.

      • I’ve been watching this particular stat with interest, although not as scientifically as I would like. Based purely on what I see by viewing the flightaware ‘view all aircraft by type’ screen once a day, I am somewhat inclined to agree that the 787 is not getting nearly as many hours of utilization per frame day as the A380.

        I just checked and right now 27/83 (32%) of delivered 787s are in flight with an average expected flight duration of 8 hours. 55/109 (50%) of delivered A380s are shown as in flight with an average scheduled flight duration of nine and a half hours. The counts are consistent with what I see every day at this time — I never bothered to calculate the average flight duration before.

        I have occasionally checked late in the evening EST, and it appears the 787 makes out a little better then (I’ve seen over 40 in flight at once). I supposethis is due to the heavy Japanese concentration of the deliveries so far.

        Obviously, the flightaware data is fallible, and my once a day sampling is a poor view into its meaning, at any rate. However, I had hoped to see the 787 closing the utilization gap to the 380 by now.

        One additional factor that may be working against the 787 at this point is that a much higher percentage of the 787 fleet are recent deliveries which may just not be fully worked into their permanent route schedules yet. Similarly, a higher percentage of 787s are in the hands of operators where it is a relatively new type.

        • Hi Matt, didn’t think about going at it your way.
          I’ve set up a reaper script to periodically retrieve data from flightaware.
          Expect some plots in a couple of days.

      • Happy Hunting.

        It will be interesting to see how things measure up under a more thorough analysis. Bear in mind that the 787 type code can appear as either ‘B788’ or ‘787’. Don’t know if there is a similar split for the A380.

      • That was fast.

        The percent utilization graph is very interesting, especially with regard to the other models you added to consideration. I will be curious to see if regional concentrations cause substantial changes in the relative utilization percentage positioning over the course of a day.

        The 787 designation mostly shows up “overnight” Eastern time. It seems to be used for Asian regional flights only.

        • Wait for a couple of 24h rollovers ( time displayed is UTC, the images are regularly updated. There is an abdata.tgz under the same path )

      • Nice work Uwe! Did you automate your data extraction or did you do it by hand (cut and paste)?

        I also have noticed, over the last month or so, that the number of 787-8’s in the air seems to peak during the overnight hours (East Coast). Also, there are day to day variations throughout the week for any given daily time. I’m sure these types of variations exist for the A380 as well. To capture a true average, I think data would have to be collected over an entire week, but what you have done is pretty good.

        The next step would be to use the flight length in hours that FlightAware provides to figure out utilization in hours per day. However, one would have to do a weighted average, meaning that care would have to be taken not to over count individual flights that would be recorded in successive data extractions.

        I agree with Matt that the 787 will have a higher fraction of frames that are in the process of being introduced into service relative to the A380. This is due to both the higher monthly delivery rate and the introduction of the type to new airlines, and will skew the 787 utilization downward.

        • cut and paste ? naw, never, much too lazy for that.
          A set of scripts scrape flightaware, massage the data into individual files
          and a gnuplot script then generates the png images. currently run as a cronjob every 15 minutes. i.e. my work is done. reload the images in a couple of days and the cyclic behaviour will probably be quite visible.
          Getting at flight durations for individual flights would present significantly more effort. ( what I’ve got now represents less than 30 minutes of creative effort 😉

        • Prduction lot ( as the graphics title says )
          i.e. the number of produced frames ( taken from WP:EN )
          note 748 is 747-8i and 747-8F munged together. flightaware does
          not differentiate the subtypes.
          This potentially is a bit unfair to older frames ( A320 ( 97% active), A340 ( 80% active but number only given across all subtypes -200 .. -600), A330 ) active numbers taken from )
          83. “B787-8”
          108. “A380-800”
          54. “B747-8”
          418. “B777-300ER”
          508. “A330-200”
          476. “A330-300”
          218. “A340-300”
          2929. “B737-800”
          3420. “A320”
          If someone has the “in use” numbers per type I’ll substitute them.

      • Uwe, the early results from your FlightAware data mining seem to show that about 50% of the A380 frames and about 38% of the 787 frames are airborne on average at any given time. Assuming that the average flight length is 12hrs, which it is not for either type, the average hourly utilization would be around 6hrs/day for the A380 and 4,56hrs for the 787.

        An interesting snapshot from FlightAware at 8:30AM (Eastern Daylight) showed that the average flight length for the A380’s in the air at that time was 7.52hrs. On the other hand, the average flight length for the 787’s in the air at that same time was 9.51hrs, longer than the A380! Since there were 53 A380’s and 27 787’s in the air at the time (0.491 and 0.325 respectively), the utilizations work out to 3.69hrs/day for the A380 and 3.09hrs/day fro the 787.

        Now, I understand that this is only a snapshot in time and if I looked at another time the average flight lengths would be different, however, I think this illustrates that the 787 utilization is not much different than the A380 utilization. The A380 utilization might currently still be a little better than the 787 utilization, but nowhere near the 10-12hrs/day that you claimed earlier.

      • Mike and Uwe,

        I think that with Uwe’s data, you do not need to take duration into account. With 15 minute sampling, you just divide your total observed flights for the day by four times the type fleet size, and you have your average fleet-wide daily utilization for each type. So, by my estimate, the A380 does look better than the 787, but it is a tough peer to go against since it and the 777-300 are the two standout types by utilization.

        In the first day of this data, the 787 looks much better than I expected overall and stacks up well against its size peer the A330. It looks to me based on this data (bearing in mind its limitations) that as the negative factors we discussed yesterday fade over the coming years, the 787 should be on course to demonstrate very high relative utilization rates indeed, even if no improvements are made to reliability, which I think is an unlikely worst case scenario.

        Uwe – I did not check how you are getting your totals for each type, but I did notice that the 787 is not the only type that has an alternate designation on flightaware. There are also sometimes some ‘330’ and ‘777’ listings that I have seen. Not sure which subtypes these should be applied to, but they should slightly improve particularly overall A330 performance if they have not already been taken into account.

  6. Scott,
    Thank you for posting this! I knew that Boeing was incorporating -8 lessons learned into the -9, but I had no idea it amounted to 30% changes.

    A few questions for you.
    1) Are the Li-ion main/APU battery changes for the -9 totally new or are they just the same, widely reported changes that were made to lift the grounding of the -8? If they are totally new, do you have any details?

    2) Are the -9 electronics changes you mention the same as those that were made on the -8 during the grounding, or are they in addition to those -8 changes?

    3) To what extent are the -9 structural changes being incorporated back into the -8? It has been reported that the -8 has already benefited from weight reduction provided by the incorporation of -9 parts into the latest block, and because of this, the -8 is down to contractual OEW as of LN-103. Wouldn’t significant changes this significant require certification activities?

    4) Any idea when the -8 will start sporting the Trent 1000 package C’s and the GEnx-1B PIP2’s? I know these will be on the -9, but they will also make a big difference on -8 performance, especially with the weight now significantly reduced.

    Thanks again.

    • 1) Same as -8 batteries with the boom box mod
      2) There were changes to the electrics all along the program, not just during the grounding. Rumored figure is 3.000 and counting
      3) Good question. Redesigned wing root join as evidenced by flightblogger’s photos, new stabilizer, new main landing gear truck and larger wheel well. Remains to be seen how much of that goes into the -8 and if so, when.
      4) The weight is reduced, but not as significantly as “now meeting specification” may suggest – the specification MEW has been raised with the increase in MTOW implemented with LN20.

  7. Airbus claims a 99.7% dispatch reliability for the A320, too. But as noted, DR numbers an be twisted. And Airbus is traditionally weaker on DR due to the fact that everything out of Germany is slightly over-engineered.

    98% means 20 out of 1000 flights. 99.7% means 3 out of 1000. That is more or less a cut by one magnitude. Airbus never managed to get the A340-600 beyond the 99%, and added another nail in the coffin for this beautiful yet doomed aircraft.

    Many changes in the B787-9 will recover weight, and some will probably recover manufacturing cost. Each time I read about the B787-8 I realize how hard it fails in every aspect (performance, reliability, cost of production, even safety). I think many customers will change to the -9, especially when the B787-8’s CASM are beaten by A350-9.

    CASM: Cost per Available Seat Mile.

    • DIspatch has little do do with the aircraft.. If you replace it with another aircraft shortly before departure. Dipatch reliability is 100%.

      Lowest costs per seatmile (with different unrealistic seatcounts), available revenue volume (ignoring payload range restrictions), range (without any cargo), dispatch reliability (suggesting it has to do with aircraft reliability),

      In my opinion The Boeing company should stop confusing the public and their stock holders with half truths and purposely incomplete storylines. And no, Airbus, Bombardier and Embraer aren’t acting the same as far as I can see. Sometimes they get irritated and react.

    • Schorsch,
      How is the 787-8 failing hard in performance? Not a challenge, but I would just like to know because I’m not getting the same impression from what I read (just so you know, I ignore all the fluff about passenger experience and all the marketing speak about changing games).

      • Early models are overweight, short on range and engines ~2% short of fuel specs. Engine PIPs fixing the latter and models now being delivered should be on weight and range performance. Dispatch reliability is still below goal, according to our information as of two weeks ago from inside Boeing.

        • Are there perchance any numbers around on maintainance hours chalked up per flight hour?
          The preventive maintainance done by the japanese carriers to achieve good dispatch numbers must rack up quite a bit of extra cost.

      • It is getting to the performance, but it was – and compared to previous programs – quite a trip. What always surprises me is that original EIS was 2008, and even today the aircraft is far away from maturity. Reminds of Russian aircraft programs.

      • The very plain truth is that these last years, Boeing has constantly over promised and under delivered.
        Weight, fuel burn, delivery schedules….
        748, 788, and very probably , the MAX program are all following the same path.
        If the 789 actually delivers what was promised, that is a true revolution.
        Airbus seems to be rather different.
        Nobody is perfect, there are weight problems, but they are more limited and targets are reached a lot quicker.
        Same for fuel burn.
        According to Emirates, the 380 reached targets from day one.
        350 XWB and 320NEO seem to be on the same path…. Under promising and over delivering

    • Apples vs Oranges. Airbus OR figures are usually lower because they are calculated to a different formula than Boeing’s. Makes for a difference of roughly 1% in favor of Boeing.

  8. Concerning dispatch reliability percentage claim & counter claim, it’s perhaps worth considering that Airbus operators are more likely to be smaller, embryonic or virgins operators frequently after transferring loyalties away from Boeing.

    Boeing’s market share specifically within the single aisle sector is becoming increasingly dependant on larger operator orders, these operators by their very nature have tight type familiarisation & well prepared support infrastructures.

    Noticeably Boeing single aisle business is becoming increasingly dependant on domestic sales, which is somewhat reminiscent of the UK Aviation industry.

  9. Perhaps one reason for the 30% new items for the 787-9 vs the -8 is that it is much more than just a 20-ft stretch:
    – Higher operating weight limitations
    [553,000/425,000/400,000 lbs MTOW/MLW/MZFW for the -9
    502,500/380,000/355,000 for the -8],
    – engines ratings [71,000 vs 64,000 lbs]
    – Fuel capacity increase [33,640 vs 30,640 usg]
    – Slightly more range with 100% passengers
    [8,500 vs 8,200 nm, 280 vs 242 passengers]

    analogous to the original 707-120 vs the 707-320B [bigger wing]
    or the original transcon 767-200 vs the 767-300ER [same wing]

  10. IMO Boeing can thank ANA for driving up the 788’s worldwide average DR with their large fleet (which also operate a substantial number of short domestic hops) and their overall stellar on-time performance for all aircraft types they operate. I estimate the 787s DR at UA to have been in the low 90s through August – and that’s not counting any tag-on or cascading delays and cancellations, etc.

    I won’t be holding my breath for Smisek to reveal any concrete figures.

  11. I always find discussions about dispatch reliability very entertaining because they are so emotional. The definition of “acceptable” dispatch reliability varies airplane type as well as by airline. For the airframe manufacturer it is mainly about bragging rights. For smaller, shorter range airplanes such as the 737 and A320 families high dispatch reliability is of great significance because even short delays can have significant impact on downstream flight and crew scheduling. For the larger, long range airplanes, short delays are of little significance because they can easily be made up and therefore have little or no impact on downstream schedules. Long delays, on the other hand, tend to be extremely disruptive. So for those fleets, it’s more about minimizing the length of the delays than it is about needing to avoid the delays.
    Boeing has had dispatch reliability improvement programs for all of its major models. Not all of them were welcomed by the all of the airlines. Northwest, for example, refused to cooperate with Boeing on the 757 reliability improvement efforts because they were satisfied with their then current reliability. Similarly, during the early days of the 747-400, Lufthansa did not want to participate because based on their analysis, 97.5% was the optimum reliability and efforts to achieve the Boeing goal of 98.7% were not cost effective. (It took the retirement of LH VP to turn that around.) Qantas, on the other hand was not happy with their 98.8% because most of their delays were of two or more hours, which raised havoc with downstream flight and crew schedules.
    Some airlines went so far as to suggest the entire ‘dispatch reliability” concept be thrown out and replaced with the more meaningful, though harder to define, tracking of “airplane availability”.
    Regarding how dispatch reliability numbers are arrived at and whether or not manufactures “work the numbers” to their advantage, the definitions of what is counted and what is not are controlled not by the airframe manufactures themselves but by ATA Spec 100.

    • Well, laughable is the MBA derived reduction to an oversimplyfied single number and then seeming to decide on that. ( Einstein: make your model as simple as possible _but not simpler_ )

      I wonder what costeffectiveness JAL and ANA see with their 787 DR effort ?

    • Timm, I guess you are aware Dispatch Reliability tells you about operations, not so much about an aircraft type? Its about a FLIGHT being delayed. Replace the broken 787 with a reserve DC10 under the same flight number & DR is 100% !

      The ratio of the number of flights delayed because of technical faults to the total number of flights, expressed as a percentage. Delays caused by other reasons are not to be taken into account for this calculation.

      Read more:

      Marketing people understand the bigger public doesn’t go into these “details” and assumes the aircraft type mentioned in the same paragraph is pretty reliable +/- a few percent..

      • Keesje, I don’t know about the 787, but during the 747-400 EIS (which I am intimately familiar with and which was probably the worst in Boeing history) the number of equipment substitutions was so miniscule, substitutions had no significant impact on the numbers. I think you are trying to make a mountain out of a mole hill.

  12. Gosh, what a lot of negative, back-seat-driver comments!
    O.K., The 787 was 3 years late and accumulated Billions in cost-overruns, BUT…Boeing made record profits in the same time period, did NOT go
    bankrupt on it and the 787 is now the first and only “all-carbon-fiber”
    a/p in service world-wide! It is not only very popular with flying public,
    it is also the only a/p to save 20% in fuel-burn per seat-mile, which is,
    or will be passed on to the flying public, one way or another!
    A winner by any measure, thank you, especially the 787-10!

    • Right on Rudy!

      You and I as Boeing alumni used to see each model’s monthly reliability reports, Their data relied on Boeing analyzing all customer inputs. Reliability was tracked by customer, by model, by ATA, by individual airplane and system – every which way there was. Nothing was held back. Nothing was fudged. To what purpose? Boeing and its airline customers were all in it together. If a system had issues, maybe Boeing had a fix, maybe the supplier, or maybe the operator. If one airline wasn’t doing as well as its peers, they could find out why and remedy their situation.

      Airbus probably does the same. Can any Airbus readers out there confirm this?

      • toyuths, I participated in numerous industry meetings where dispatch reliability methodology was discussed and I can assure that as recently as five years ago, Boeing and Airbus methodologies were the same, both aligned with ATA Sec 100.

    • Rudy, making a profit these is about deferring/ignoring costs.. Anyway hopefully the 787 will mature. I think the 787-9 will be the 737-800 of the 787 family. Ignoring the media hype, the 787-10 is compromised . Or “optimized” depending which glasses you put on 😉

  13. @Mike Bohnet from September 17, 2013 @ 4:53 am

    Actually the %airborne avrg directly translates into %utilisation per day
    40% airborne on avrg ~= 24h * 0.4 = 9.6h
    current A388 avrg -} 54.0%
    current B788 avrg -} 31.5%
    If you go to Boeing’s new airplane site you will find as of sept. 9th:
    7m passengers : 45m miles distance flown : 34,000 flights
    I get around 1400 miles flown per flight
    ( equivalent to ~2.5 hours flight duration.)
    next step would be to compute available fleet days.
    first order guestimate:
    600 days(2years – grounding) * 86 (delivered frames) * 0.5(triangular area)
    –} 619200.0 hours available flown 88500 –} 3.4h/d

    • If %airborne does translate directly into %utilization per day, then:
      A380: 50% ave., 12h/d
      787: 38% ave., 9.12h/d

      Yet, I keep seeing this 3.4h/d 787 number.

      • I agree – the data from flightaware does not match the 3.4h/d assessment, so whatever calculation produces it is not very useful for assessing current utilization. Maybe the Boeing website data is just wrong.

      • Or the calculation using the Boeing website data is flawed somehow in it’s assumptions. I personally suspect the triangular area assumption that is supposed to account for the utilization over time. A triangular distribution probably puts too much weight on the past utilization. The first 6 months (at least), showed disproportionally low utilization numbers for the 787.

      • Also, there is a utilization ramp-up for each new airline that needs to be considered, especially the first few, ANA, JAL, Ethiopian, LAN, and United.

      • Sigh,
        scraping the nyc787 deliveries list I get 407,500 available aircraft hours.
        ( taking 93 days of grounding into account for frames delivered pregrounding, 514656 hours without that. So the first order estimate (612khours) was exactly that a good 1st order estimate )
        87,500 hours flown then is 21% or ~5 hours / day .

      • Uwe,
        I really do appreciate you going down this route to see if the calculation works, however, I think that even knowing the delivery dates is not enough and will over predict the amount of available aircraft hours. Available aircraft hours really starts accumulating at EIS for each individual aircraft. For example, the EIS of ANA’s first 787 was about 1 month after it was delivered. As we both know, such a list would take much more time to put together. I’m not quite sure I would know where to look for data on the EIS of later frames to an airline.

        Also, each airline takes it’s time ramping up the service of a new aircraft type, so some of the lower utilization at the beginning is intentional and not just due to reliability. What you are calculating is the fleet utilization averaged over the entire time since EIS.

        For the above reasons, I really think the FlightAware data gives the best utilization snapshot of the current state of affairs.

  14. @ Matt B from September 17, 2013 @ 5:08 am

    Data source is
    with all its limitations. ( it is a quick and dirty hack after all )
    Does not show any munged type entries. No idea on how to
    check if the mistypes count ( anywhere or not )
    How do you arrive at the “4 times fleet size” divisor value ?

    • Just because you are taking four samples per hour.

      I’m aiming for the same calculation as your percentage * 24, but I think your expression of it is more intuitive.

      • The sample rate as such does not go into the computation.
        For validity the requirement is that the sampling rate is at least a magnitude shorter than flight duration. ( actually it should be less than 1/2 the time of the shortest flightduration to avoid aliasing errors )

    • Wanted to wait until there was 24 hours of data to illustrate the calculation I was suggesting, but I’ve got my lunch now, so I’ll just extrapolate to a day by multiplying my result by 24/22 .. I’m using the data from your .dat files, but ignoring your first sample which was not on the quarter hour.

      For just B788, I have total of all samples = 2951
      For 787, I have total of all samples = 290, so B788 + 787 total = 3241

      So, if you ignore the ‘787’ samples, average daily utilization for the 787 would be 2951/332 * (24/22) = 9.7 hours

      If you add in the ‘787’ samples, average daily utilization for the 787 would be 3241/332 * (24/22) = 10.65

      10.65 h/d is a far cry from 3.4, and I suspect not too shabby at all.

      This data will be distorted in the 787’s favor by two factors, but I doubt either of them is great enough to even offset the inherent downward skew of recently delivered frames.
      – Boeing test flights will be captured in this data, amounting to maybe 10 hours in a day.
      – The flightaware type designation probably reflects the scheduled type for the flight, so if a 787 flight has to be flown by another type as a standby, I doubt this data will reflect that.

      • merging data files is a major hassle as flightaware gives sparse information.
        i.e. 0 airborne does not happen, the item is dropped from the listing.

      • if you retrieve the data pack there now is a 787M.{dat,avrg} file
        that is the merge of B788.dat and 787.dat
        I’ll probably do the same for all the A340 types to be able to apply the
        active numbers from

      • Good deal. Your efforts are appreciated and ingenuity noted.

        I did see after I posted that there were a few off-cycle samples in the data which skewed my calculations but probably would not effect yours.

        I recalculated both the 787 (merging ‘B788’ and ‘787’) and the A330 (merging ‘330’, ‘A330’, ‘A332’, and ‘A333’) and removing the off-cycle samples, and came up with 10.03 hours average utilization for the 787, and 9.81 hours for the A330.

        Overall, I would say you are right that automation is the only way to go.

      • Uwe, your efforts are appreciated indeed!

        What are those crazy spikes in the combined B778 and 787 data sets? Did ANA schedule their entire fleet to depart at once or something?

        • No idea.
          interesting to note is that both subtypes show the same number ( afaics )
          I haven’t watched flightaware for any reasonable time so I really don’t have
          much opinion on how they generate their data, what directed errors they introduce and how reliable it could be. Nonetheless I would expect some funnies in their data. They do it just for the money after all.

    • For a follow up on this:
      I’ve added data for average “around the clock” utilisation and day averages for trend visualisation.
      Unfortunately the numbers for Dreamliners in use appear to be worthless as
      the planned type and not the actual type used seem to go into flight aware statistics. ( known substitutions with A340 done by LOT and DY show as B788 🙁
      I have similar doubts over the validity of merging the two Dreamliner data files.
      Any ideas on how to overcome this ?

  15. Pingback: Boeing 787-9 departs on first flight | Leeham News and Comment

Leave a Reply

Your email address will not be published. Required fields are marked *