What’s an anatomy of an aerospace project? How do all the processes, design, construction and testing, pull together?
Phew. You want me to discuss this? In a single post?
It needs a book!
And there are books available. I recommend Prof. John Fielding’s book Introduction to Aircraft Design, simply because I studied under him, and I never met anybody else with such a comprehensive engineering knowledge of aircraft and the industry that produces them. You could ask him any question, and if he couldn’t answer it directly himself, he’d point you in the direction of a book or a person who could.
People like that are few and far between. And you almost never find one inside a company that actually makes aircraft. They’re usually academics who have deliberately removed themselves from the business in order to study it from a distance.
That’s the problem with aerospace. The process (really, the process of processes) is so damn complex that it’s almost impossible to see the whole picture.
The exceptions are the academics, and the few who focus on small experimental aircraft, where a small team of people conceive, design, build and fly a single aircraft. There are few such companies in existence today. (Scaled Composites is the most notable example.)
The easiest way is to describe the chronology of a typical large project, as seen by someone on the inside.
In the timeline I paint below, the year numbers are very approximate. Typically, the larger the aircraft, the longer the timeline. But conversely, the more established and experienced the airframer, the shorter the timeline. Commercial aircraft tend to have shorter timelines than military aircraft, and military combat aircraft tend to have the longest timelines of all.
The A380, for example, was a decade from concept to EIS. Most general aviation or small corporate jets will be half that. However, there have been several small jet projects that have been under development for well over a decade, usually due to funding problems, but sometimes due to unforeseen design or certification problems.
Years 1-3: Conceptual design.
This is a mixture of marketing, engineering, and art. Nobody except a hobbyist develops an aircraft unless they have some reasonable idea of who their customer(s) might be and how much they will be willing to pay for it.
There will be a lot of discussions with prospective customers, to elicit their requirements for the new aircraft. The whole process of requirements capture is an engineering discipline of its own, and the process is never really complete.
A lot of time will be spent trying to tease requirements out of customers, mainly because customers, being human beings, often can’t articulate what they want, only what they don’t want. Steve Jobs, of Apple fame, was once found screaming at one of his chief designers to come up with more ideas, there weren’t enough ideas. When the designer objected, “Tell me what you want!”, Jobs replied, “I don’t know what I want! I’ll tell you when I see it!”
That problem exists at aircraft concept level, right down to equipment specification level. People find it hard to spell out what they want, either because they forget, or they can’t articulate it, or they assume a requirement already exists because it hasn’t occurred to them that it WOULDN’T exist.
With an as-perfect-as-possible set of initial requirements, the conceptual designer start his/her work. I will not attempt to describe what Dan Raymer has already so adequately described on his website AircraftDesign.com and in his book. Suffice to say, conceptual design is a highly iterative process that is part art, part marketing, part psychology, and part engineering.
Once a concept, complete with CAD models and estimate of performance, aerodynamics, costs, weight, etc, has been agreed with senior management, and prospective customers, the design team will start to grow, and investment in preliminary design will begin.
Years 4-6: Preliminary design.
Whereas the majority of aircraft concepts never see the light of day, a concept that reaches this stage has a very real chance of getting launched, and eventually flying. The airframer will begin assigning a dedicated team of specialists for every major structural component and every onboard system. There will be people assigned who specialize in maintenance, airfield operations, finance, ground support . . . . . . All told, the team will be 100-500 people in size, so the company is now investing heavily in the project, without even having launched the project yet.
The goals of this stage are twofold:
(a) Collect all the necessary data to support the company board of directors when the time comes to make a launch decision;
(b) Assemble all the players necessary to guarantee a successful aircraft programme, from EIS right through to the 60 or so years from now when the final aircraft is removed from service.
At high-level, procurement people will seek expressions of interest from suppliers of aircraft systems. The earliest and most important ones to line up are the engines and landing gear. These are the heaviest and most expensive systems. The engines typically constitute 25% of the purchase cost of an aircraft, and their performance is critical. (It is not unknown for an aircraft design concept to be adjusted to suit the engine concept, rather than the other way around.)
By launch decision date, most system suppliers will have been lined up at least by letters of intent, and the few that haven’t will be those for the least significant systems.
All of this system-supplier-related activity creates work for the related systems engineers. They must devise the first set of requirements for the system, to support negotiations with the supplier candidates.
The aerodynamicists will start refining the aero data (which thus far will have been devised with empirical methods) using CFD and full/half-aircraft wind tunnel model testing.
Loads engineers will devise a first loads loop, i.e. a first iteration of structural and aerodynamic loads for all major aircraft components. They will use the latest aerodynamic, mass, and structural stiffness that is available, even though they know the aero and mass guys are already producing better estimates.
Structural/stress engineers will resolve those loads down into specific components, and “stress” the components, i.e. conduct hand stress calcs or FEM. They will create space allocation CAD models that provide worst-case illustration of the spatial envelope their components might occupy. For the major structural members (e.g. spars, skins, longerons) they will update the stiffness data that gets used by the loads and aeroelastics engineers for their loads and flutter estimates.
Stability & control people will use the aero data to make first-pass estimates of the aircrafts’ stability in short-period pitch, phugoid, Dutch roll, etc.
Performance people will be using the aero data, mass data, and engine performance data, to forecast all the most customer-relevant performance metrics: Payload-range, climb rate, take-off and landing field performance at various airports around the world.
Cost people will be using the same data to forecast the direct operating and maintenance costs.
Aeroelasticians will be using very similar data to the loads engineers, but with a few to forecasting the aircraft’s vibration modes and resistance to flutter.
System safety and reliability people will be active with various tools to identify significant failure modes, and failure rates. This work has a fundamental impact on the detailed design of structure as well as systems. Higher system reliabilities allow structural designers to use lower factors of safety on associated structure, thereby saving weight. However, higher reliabilities must be verifiable, and this can drive up the system cost and weight.
At some point, the GO/NO-GO decision to launch must be made. The decision is never made by anyone other than the company’s board of directors. An aircraft programme is too huge an investment, and it does not begin to show profit until typically several hundred aircraft have been delivered.
Sometimes the board decides not to decide, i.e. to postpone. Usually this is due to a lack of customer expression of interest. Sometimes it is due to what the directors deem as inadequate supporting work done. In essence, a postponement decision means NO, but continue the investment and do not dismantle the team.
It is not unknown for a board to kill a project at this stage. Boeing killed their Sonic Cruiser proposal when the price of fuel suddenly skyrocketed around 2000, and the aircraft suddenly became much more costly to operate. They would replace it shortly thereafter with the 787. Around the same time, Bombardier shelved their BRJ, which was their initial attempt at entering the 100+ seat airliner market. Several years later, they came up with a new, bolder proposal, the C-Series.
Years 6-9: Critical, or Detailed design.
The number of people working on the programme will grow by at least one, and sometimes two, orders of magnitude. It get very distributed, often on a global scale.
All of the work that was happening in the Preliminary stage continues, but in progressively more and more detail. And there are deadlines, not-to-exceed limits on key parameters, and pressure, lots of pressure.
Every specialist team is screaming at their internal suppliers to deliver on time, with all the latest concept changes captured and covered. Each specialist team’s internal customers are screaming at them for the same reason. Inevitably things go wrong, mistakes get made, people resign or fall ill, the data delivered is late, or incomplete, or inaccurate, or out of date . . . . . . or all four.
Each team wants guarantees from their internal suppliers, and when they don’t get them, can’t provide guarantees to their customers.
In some companies, senior management will periodically announce a design freeze, to allow the frenetic activity to settle down a bit, and to allow teams to get caught up.
There will be regular review meetings, at various levels, some purely internal, some between internal teams and suppliers. The frequency of these reviews will depend on their purpose and on the health of the activity in question. Some reviews are progress tracking reviews, typically monthly. Some are design or “milestone” reviews, which are planned long ahead of time, and are rarely more often than once a year.
And throughout these meetings, certain words will be heard and spoken again and again. Sometimes quietly, sometimes loudly, sometimes accompanied with smiles and kind words, sometimes accompanied with shaking fists and angry faces.
- NRC (non-recurring cost)
- DOC (direct operating cost)
- Fuel burn
- Space allocation
Did I mention weight?
Weight always seems to be the big one, and for good reason. Every pound, every kilogram of mass on board the aircraft . . . . . . must be manufactured, maintained, and is therefore expensive. Every pound, every kilogram saved . . . . . is a cost reduction to the airframer and/or the airline.
Some activities will be started that weren’t significant activities at the Preliminary stage.
There will be extensive conversation between the airframer and the relevant airworthiness authorities. The authorities, being public-safety-preoccupied, will be seeking opportunities to load new or beefed up requirements on the aircraft design. Every new requirement represents an additional cost, and therefore the airframer will be looking to argue its way out of those. Sometimes that argument works, sometimes not.
The airframer will be looking for ways to meet existing requirements more cheaply that before. Each of these means arguing their point with the authorities. Again, sometimes it works, sometimes not.
For the same reason, the airframer will be looking to reduce the work of design validation and verification. Every requirement, at every level, on every component, must be verified, i.e. you must prove that the aircraft satisfies it. Sometimes this can be done by analysis (i.e. by mathematical model), sometimes it must be done by testing. You try to verify as cheaply as possible. If you can verify by analysis, don’t test. If you can test on a component test rig, don’t test on a system test rig. If you can test on a system test rig, don’t test in a ground or flight test.
Each system will have its own set of dedicated test rigs. Each major piece of structure will undergo testing on a dedicated test rig. The first specimens of each major component will never be assembled on an aircraft, but instead go for dedicated static or fatigue testing
Flight testing is the most expensive, time-consuming, but also the sexiest testing of all. The input of the prospective flight test crews will be sought at this stage, particularly as regards the layout of the cockpit instruments and control, but not only those aspects. The flight test department is considered the first aircraft customer. They will be the only ones flying on it for at least the first year of the first aircraft. Indeed, for the first aircraft, they will likely be the ones to EVER fly it, as the first aircraft usually is never sold, but kept for flight testing purposes.
For the people working in it, the essential emotion throughout this stage is Stress. The work is simultaneously fun, and not fun. This is the stage at which it gets messy. This is the stage at which people start having heart attacks, and going off work with stress-related illness.
Inevitably, there is bad news. Something is delivered late. And then later. Some piece of data is wrong, and affects your work. Now you have to redo it, and it’s not what your internal customer is going to want to hear. They scream. “It’s not my fault!” you say. “Don’t shoot the messenger!” It makes no difference. Unfortunately, messengers get shot every day. They need someone to shoot, and you’re convenient. You are now tarred with the brush of causing Big Problems for Everyone Else.
You’ll have more than one of these experiences. Be prepared for them, and don’t take them to heart.
Years 9-10: Flight testing and certification.
Talk to any professional at an airframer, and you will probably find them to be fairly down-to-earth, practical, consciencious people. Not much gets them excited. They just enjoy beavering along in their work.
There are two exceptions.
Aircraft roll-out. And first flight.
At roll-out, they finally get a close-up view to what they’ve been sweating blood over for several years. Up to now, it’s just been drawings and CAD models. But it’s funny how different the aircraft looks up close and physical.
(The last Farnborough Airshow I was at permitted me to see the A380, B787, and A400M, all in one afternoon. The A380 I had already seen several times, but not the other two. And with the A400M, I was struck particularly with how downright mean the thing looked. Despite being a transport aircraft, it looked plain sinister.)
And then, at first flight, well, you get to see the thing fly. It’s like witnessing the birth of a baby. It’s magical. And if you miss out on it, you feel cheated.
But whether you get to see it or not, it’s quickly back to work.
This last stage of the aircraft development programme is hardest to predict. We do testing, and we do flight testing, “to cover (as one colleague put it) a multitude of sins!” When you design anything, you have to make assumptions and judgement calls every day. Aircraft are no different. And because human being fly one aircraft (and live on the ground underneath them), we flight test.
Inevitably, flight testing uncovers problems. Sometimes we suspected the problems, sometimes they are complete surprises.
Irrelevant. The problems have to be solved. And with EIS a year away, or less, any delay means a potentially disappointed (or angry) airline customer, and loss of face.
So you scramble.
There are also a ton (or tonne) of documents that must be prepared and delivered to the airworthiness authorities. These get prepared, at some risk, concurrently with flight testing.
So unpleasant surprises in flight testing can do more than make you scramble. They potentially invalidate a lot of work other people have already done.
But the day eventually comes when the authorities hand over a certificate to operate the aircraft to the first airline customer.
Champagne, hors d’oeuvres.
But just for a day.
Because, in fact, you’re still not finished. There’s still a lot of testing analysis work to be done, that has been put on hold. You’ve done enough work to clear the aircraft to start service. You, and the suppliers, haven’t done enough work to enable the aircraft to actually operate for 30 (or however many) years.
That work, often called qualification or endurance testing, may have already started, but it will be several years before it is complete.
In the next post, I consider a question posed by a reader: Why is there such an opportunity gap between Europe and the USA?