Bjorn’s Corner: New aircraft technologies. Part 42. Improving the learning curve

By Bjorn Fehrm

December 15, 2023, ©. Leeham News: We are discussing the different phases of a new airliner program. After covering Conceptual, Preliminary, and Detailed design, the manufacturing of prototypes, and their roles in flight tests, we now look at the production phase.

Last week, we discussed why production costs vary over time and why they can be up to 500% higher for the first units than for units past 400 to 500 aircraft produced. Now we go deeper into the reasons behind this and what can be done to improve the situation.

Figure 1. The program plan for a new airliner. Source: Leeham Co.

How to lower initial production cost

We explained last week that the initial fitting together of manufactured parts is one root cause of why early production has a higher cost than when such problems have been corrected, and the assembly runs smoother. This is part of the easier problem areas to address.

The easy part

Modern 3D design and simulation tools allow designers, production engineers, and shop floor mechanics to visualize the parts and their assembly on large screens or through VR 3D headsets. Cross-functional teaming in the design reviews can catch problems before they appear on the shop floor.

By keeping all models and data (BOMs, instructions, 3D drawings, etc.) digital and using interactive screens in production where mechanics find all the needed information on how to assemble items (Figures 2 and 3), the learning process goes faster.

Figure 2. A pair of mechanics retrieve the information for fitting Electronic Warfare (EW) units to a Gripen E fighter. Source: SAAB.

In Figure 2, two mechanics on the SAAB Final Assembly Line (FAL) for the Gripen E fighter learn how to do the next work steps, reading the information directly from the 3D design database system and not from paper information.

It creates a tighter feedback loop to design/production engineering as problems can be fed back immediately as the assembly is done, Figure 3.

Figure 3. The mechanics fit the EW units and can provide feedback directly on any fit or procedure problems into the design/production engineering database system without searching for responsible persons. Source: SAAB.

The hard part

I’ve deliberately used examples from the SAAB Gripen E production, as it’s one of the first instances (if not the first) in aeronautical production where an end-to-end digital chain was introduced all the way to the assembly mechanic, from design to shop floor.

This know-how was the reason Boeing teamed with SAAB for the USAF training project T-X/T-7A Red Hawk and learnings from this project will now be used in other parts of Boeing.

I took the SAAB example for more reasons than the Digital Chain/Digital Twin. The example also addresses the more difficult part of improving the effectiveness of initial production, the mental part.

I have worked in Swedish, US, German, and French work cultures, and they are all different. The largest difference is between the Swedish soft, cooperative culture, where workforce and management cooperate to achieve results and the top-down culture of US companies.

Swedes learn to work and function in groups from Kindergarten, and how to tone down their personality in favor of group function. In the US, on the other hand, you learn how to articulate and portray yourself so you stand out and get recognized. The focus is on the individual.

It’s no accident that so many useful production improvement methods come from Japan, where the person and work culture has similarities with Sweden, and Swedish SAAB is picked up by Boeing as the catalyst for its design-to-production change program.

Production of a product of the complexity of an airliner is INCREDIBLY complex. It’s why top-down initiatives such as aggressive targets and the reliance on KPIs (simple metrics that define success or not) as part of hot-shot manager’s promises: “I will fix the production issues within XX months by applying my XYZ method I’ve used in the past” leaves a management team that cash in their bonuses and move on and a shop floor that has not really changed although the KPIs says it has.

The reality is that only a change process that combines modern technology that can deliver instant, easy-to-understand information on what’s the state of things to learn-ready cross-functional teams can deliver real and lasting results.

It also means that the change process is a marathon that progresses in small but ever-present steps in the minds of real, hands-on individuals, not on the KPI screens of managers.

The improvement of the learning curve requires long-time commitments from the teams and their management. There are no shortcuts and no “quick fixes.”

The role of automation

Part of productivity gains in production comes from the automation of work-hour-intensive processes. The building of an airliner is work-hour intensive, especially the assembly of structural parts. It’s known as the “drill and fill” problem.

Present aircraft materials cannot be welded together (thermoplastic composites are introduced to change this); it’s why the parts are either riveted or bolted (often called fastened) together. The drilling, separation of parts for deburring, assembly, and inserting and tightening of the bolt+nut/rivet require several labor steps.

As we have 60,000 rivets in a Boeing 777, the manual drilling and filling of these 60,000 holes is a target for automation. But aeronautical automation is hard.

I have visited more than 20 aeronautical production lines and seen few semiautomated assembly stations and no fully automated ones, including the brand new Airbus single-aisle Final Assembly Line (FAL) in Toulouse.

The complexity of achieving improvement and failure to do so can be exemplified by the FAUB (Fuselage Automated Upright Build) project to “drill and fill” Boeing 777 sections with Robots at Boeing’s Everett site, Figure 4.

Figure 4. Screen capture from Boeing YouTube on 777 FAUB. Source: Boeing via YouTube.

The FAUB project was eventually abandoned as a team of mechanics with hand tools could do a better job, where they easily adapted to any fit or other issues.

But I could equally well have put the picture of Airbus’ automation of the A320/321 FAL number four in Hamburg, which also failed to deliver what was expected. It’s probably why the newer A320/321 FAL in Toulouse only had limited automation (fuselage crawling drillers but nothing more exciting).

What has been learned in all these projects is the incredible flexibility and efficiency of a trained and experienced mechanic. If he/she is, in addition, motivated and well supported by real-time information systems like in Figures 2 and 3, it’s a proven way to increase efficiency and gain an improved learning curve.

Sensible step-wise automation adds to these improvements but requires a well-functioning team around it so as not to spoil the initial learning curve when the complicated automation shall be trimmed in.

31 Comments on “Bjorn’s Corner: New aircraft technologies. Part 42. Improving the learning curve

  1. It is a difference trying to robotize and old design not designed for robotic build, like the 777 and A320 trials. The car industry has been designing cars for robots since the 1980’s and very few aircrafts has yet been designed that way. It will come. Also in jet engine module builds to get expensive engine parts assembled to the QEC’d engine to reach FAL’s and paying customers quicker.

    • As kind of a support to that, I trained in the Cessna 150/172 back in the 70s. I think you could buy a 172 new for $25,000.

      I like to somewhat keep up with that area, recently one of my favorite guys (Juan Brown) did one on the current Cessna 172. $500,000 for an airframe that dates back to the 50s. Cessna claims its not making money on those (I doubt that). They compared it to a modern efficient new build with all the bells and whistles included and it was the same price.

      Its mind boggling.

      • This is case in point. Yes sirree, what is now a dog performance wise (both speed and fuel efficiency) we can turn into a modern whiz bang setup with pretty interiors!

        Throw in a glass cockpit and we have leaped into the 23rd century with Hyper Sonic like performance.

      • In 1979 Cessna was the number one planemaker in the world and sold over 8,000 aircraft.

        In 1988 they sold 161.

        How/why that happened is the real reason their aircraft cost so much today.

          • The very poor safety record for General Aviation gives those lawyers plenty to do.

          • Vincent:

            I would say the very poor safety record of Pilots gives the lawyers something to do.

            I believe about 80% of losses are due to a pilot decision, some truly awful.

            Juan recently had one of a lady who could barely fly, bought a high performance single and had an obsession with the auto pilot (sort of auto pilot, it gave directions)

            It was clear she did not have any good basic pilot skills and the auto pilot looked to be a crutch.

            She also had all sorts of filming angles for her flying. As a great example of how not to do it, that was it.

            Yes you sometimes have an equipment failure, Mike Petey had one of those where his Turbine just let go. Properly maintained and from P&W (used).

            But those are far less common than just a bad or stupid decision. The only time I had issues was due to that same dumb decision end.

    • auto mobiles are not generally made of aluminum nor are they built to .001’s of an inch – not an easy task for an en effector on the end of a roboit cantilever

  2. Bjorn’s message about productivity has 2 aspects. One is that the steep learning curve is driven by workers who are motivated and empowered. The second aspect is that you can see and understand this by the experience of working and observing how that happens. If you don’t spend time in the workplace, you wont see it, and you won’t understand it.

    Bjorn makes a second key point – that some leadership cultures encourage productivity improvement, and other leadership cultures discourage it.

    That can be obvious to anyone with manufacturing experience. On the other hand, too many leaders identify with finance and not production. So, we have Upton Sinclair’s pithy realization that “It is difficult to get a man to understand something, when his salary depends on his not understanding it.”

    • Stan S:

      Well put, I am familiar with Sinclair but not that one. Thank you.

      I can add its not just mfg its everywhere. I see a Personal Dept type (aka Human Exploitation) who write a column and its all about her telling her clients you are abusing your people and you need to correct that or your mess will continue.

      What is truly funny as I was looking at being able to quit work I started to reeduce my self imposed work load. My company manager did not notice nor did the client who I worked for basically.

      My manager did not care (more actively wanted to stab me in the back) and the client was all on board with the toxicity as long as I was around to trot out when he needed a tech review or information for a contract bid.

      I literately had cut my production to 25% of former (which was still 25% better than all but one young guy who was an image of me at that young age and loved to work).

      The Union Electrical guys did work, they had to be supervised, but if they did not work then they got laid off and the Electrical Company would hire on others and determine if they were worth keeping. Yea at time they had to lay off but they could request people and they had their list and used it as soon as they could.

      When you have companies that go into a bossiness knowing its crowded and then its dog eat dog and a race to the bottom.

    • Often in the old days the engineers choose tolerance requirements between 0.001″, 0.003″ or 0.005″ for machined parts. Later detailed engineering analysis when quality drowns them in deviation reports from parts masurement protocols the tolerances get revised and documented. That opens up new manufacturing methods and tools (still with measurement protocols). The analysis is important and hopefully be free of top management business decisions, like the PWA powder disk inclusions.

  3. It was interesting to read, among other things, of “the incredible flexibility and efficiency of a trained and experienced mechanic”. It seems automation
    has quite a ways to go, yet.

    • Vincent:

      Mechanics I know are not considered high position, but for those that like that kind of work (and I did) it takes brains to be successful.

      They are not suited to office work but capable as anyone.

      Aircraft assembly mechanics clearly are good at getting things to fit. Boeing mechanics proved that when they put together the poor fitting 787s.

      Charleston fell apart due to lack of that experience as well as no union to protect them if they said anything.

      It was the Everett group that put out the alert on the failing quality control issues.

      I don’t think it was a coincidence that the Shim fit became an issue until Charleston became the sole production line for 787.

      Ironic in that they are still providing work for the Everett group.

      As a mechanic you never want something to fail, its like an obsession. But there was a saying that things needed to fail to get fixed. Glad I am out of it.

  4. Bjorn did a great job at opening this discussion. He uses the Gripen E as an example of the future, and its a solid bet that someday we will get there. The elephant in the room is Hardware Variability. There are 2 major causes for hardware to be different one line number to the next. First is the accumulation of tolerances in all the components that requires the shifting of individual part locations to make the assembly function per drawing. Second is the differences in the configuration itself where the customer requirements mandate alternate installation sequences using a different bill of materials. When you overlay tolerance accumulation with a difference in parts one aircraft to the next, it gets very difficult to practice hardware variability control. When building low line rate products with limited opportunities to make improvements, the personnel with magic fingers must be relied upon to make the structures work. The loss of skills has been terribly disruptive to building aircraft, and unfortunately, we will be forced to rely on the artistry of the assemblers to get us to the point where redesigns and DFMA can actually happen. Too much is being made of theoretical production process improvements by managers that need to show some results instead of them actually accomplishing something. For now, until aircraft are designed to be assembled net, we will be stuck with the need for a bunch of artists/assemblers with magic fingers that can shim the airplane together correctly. This is one of the huge unspoken problems with rate increases, we need far more magic fingers than are available.

  5. Bjorns comments on American culture stings because it has a huge aspect of the truth.

    But I would like to put in a caveat. Workers can be made to shift (or quit) with management just supporting the foreman.

    But the support is not their from the very top down to the foreman level and when you get to that point, all the cooperation in the world fails.

    I worked in a pretty unsupervised area of repair and maint (far more fixing things than maint).

    We had one guy who bright and useless. He would get his tools togehter making a big deal of it and spending hours then go over to the hangar ande come back and, well I cna’t do the PM there, there is an airplane in it.

    Uhh dude, call aircraft maint and fine out or you can drive over there and focus on something else.

    He finally decided he wanted to go into the Electrical Trade. I thought, yea this should be fun, have to work his butt off as its all supervised and no slack not to mention 4 years of low pay.

    He zipped through it and got his Journeyman’s. He was motivated by two things, supervision and money (those fellers made good money).

    I would not have trusted him to walk myi pet rock.

    My take was a lot of people were not self motivated, but put into the right situation they could works and work well.

    But on top of that was poor management and bad management and that leaves the lower levels to seek some kind of low balance.

    • Thanks for the pertinent link. I liked Aboulafia’s final, apt sentence..
      Boeing “leadership” do seem intent on driving that company into the
      ground (again).

    • “Consider the strategic picture (ironically) from the top. Boeing’s share price hasn’t budged over the past three years, despite a remarkable unexpectedly strong jetliner market recovery, and despite the strongest global defense spending environment in over three decades. Boeing’s peers, by comparison, have done great. But for Boeing, that shareholder focus hasn’t worked very well. This is because the two primary business units are not in great shape on a fundamental level:

      […] we’ll know in the next few years if they’ll get anything from the NGAD, F/A-XX, and CCA decisions. If they don’t, there’s no real future for BDS.

      […] The absence of a new jetliner also means they’re at risk of losing their engineering core; they haven’t launched a clean-sheet design since 2004.

    • Airbus was more lucky on the neo’s than skilled. The analysis after Sully’s Hudson river landing showed that it will statistically happen again (and did in Russia), hence the need to replace the CFM56-5B with its its narrow chord fan blades. V2500 fan was ok. CFMI had the LEAP-1C in works for China and PWA the geared fan. Initially Airbus wanted the same fan diameter as the CFM56-5B/V2500 and CFMI did the intial LEAP-1 that way, then Airbus allowed PWA a much bigger fan on the PW1100G that sold well (LH) then CFMI reacted and did the LEAP-1A with bigger fan and Airbus A320neo with bigger lading gears could fit them both. Airbus already having the A321ceo with winglets and its neo became a huge success. So Sully and a french mathematical statician are the “fathers” of the neo’s. (I cannot guarantee the full truth of above story)

      • I am twitching pretty hard here.

        The Dump into the Hudson was a good injestion caused.

        The one in Russian I know about was some kind of operation mess up.

        CFM-56 has fantastic reliability, not perfect but very very very good.

  6. ….perpetuator, that is…. Darn cell phone doesn’t want to write in English.

  7. If the 777 has 60,000 rivets, then it must have 120,000 rivet holes. Or slightly more..!

    • What with all those rivets and holes, imagine how much money Boing could
      save by not building those pesky planes!


      and some here say their competition is “just lucky”.. over, and over, and over, and over..

    • Steve….
      1) 60 thousand rivets is WAY understating the rivet count.

      2) 1 rivet gos into 1 hole. The stackup does not count each part element as a seperate hole. When you do a hole check, all elements of the hole are measured coaxially, and if found to be discrepant, you count it as 1 hole. And yes, we hole check rivets using SPC survielance unless of 4 thou diameter windows unless they are Close Tolerance rivets found in fatigue critical locations such as the fuselage skin panel lap joints where a 3 thou diameter window exists

  8. Why 2023 was a record-breaker for the A320 Family

    ‘By the end of November 2023, Airbus’ highly successful A320 Family had already logged 1,395 orders, including the single largest order for 500 aircraft, from Indian airline Indigo.’

    Add in the Dec orders from Avalon & Turkish (another 250) and they’re at ~1,650 for the year.

    ‘To keep up with such strong demand, Airbus is growing its global industrial footprint with the goal of bringing A320 Family production up to 75 aircraft per month in 2026. At the same time, the company’s single-aisle final assembly lines (FALs) are evolving to produce a mix of aircraft models with new cabin and comfort standards.

    Airbus is doubling the number of FALs capable of producing the A321neo, compared to 2019. By 2026, ten A320 Family FALs around the world will be capable of producing the A321neo.’

    If you have a consistent and successful method of doing things, you can export it all over the place, I guess. Now we just need pilots to fly all these aircraft…

    • FedEx is transferring its spare pilots over to the regionals so there you go.

      Airlines finally got serious about pilots and there are a lot of tracks now to get a license and fly for a regional and then an airline.

      500k for a C-172 is not helping though.

      • Not transferring:

        ‘American Airlines offering $250K bonus to help attract pilots from FedEx and UPS, reports say’

        ‘According to multiple reports, the airline is offering a $250,000 signing bonus to any captain who leaves FedEx or UPS. They will get $175,000 in their first check and an additional $75,000 after one year.

        PSA Airlines, a regional carrier owned by American, is putting the six-figure bonus out there to pilots to help fill a gap in service to smaller U.S. cities, according to The Wall Street Journal.

        The gap in service has reportedly worsened over the past year as smaller airports have been forced to cancel flights due to a lack of pilots.’

        They’re trying to head hunt them at $250k each. Whether they go to PSA to fly puddle jumpers and start at the bottom of the seniority list, is an another matter.

        No one has reported on how this is working out. Does anyone have any news?

Leave a Reply

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