Like the Boeing 787 and Airbus A350, Bombardier’s in-development CSeries airliner has a composite wing. Made at the manufacturer’s new factory in Belfast using the company-developed resin transfer infusion process, it helps reduce weight and so contributes to a fuel burn 20% lower than those of competing aircraft.
Automated manufacture and corrosion-free materials mean the wing should also be less labor-intensive, both to build and to maintain than a metal wing. But the testing required for airworthiness certification is far more onerous, says Colin Elliott, vice president of engineering at Bombardier Belfast.
Over the last 20 years, he says, the certification authorities have developed a five-level building-block approach to certification of composite airframe structures. It starts with testing coupons – small samples of the materials used – to characterize their basic mechanical and physical properties.
Level 3 testing looks at elements. “You’re testing things like mechanical joints between different components within the structure,” Elliott says. Level 4 examines subcomponents: “These would be bench tests using fairly large specimens up to 3m or 4m in length representing, for example, spars in the wing. And eventually you get up to Level 5 testing of the complete structure, which includes the complete wing.”
This all adds up to “a very laborious, very expensive, very time-consuming way of doing things”, Elliott comments. “The industry has been doing metallics for very much longer. There’s a much greater database available, so unless there’s something very peculiar or very unique in the design, we would design the metallic structure purely by analysis and finite element modeling, and then go straight to static and durability testing.”
The first four levels represent an additional burden for composite structures, although Elliott expects that to diminish: “As time goes on and the manufacturers and certification authorities become more comfortable and more confident with the materials and with our methodologies for finite element modeling and analysis, we should be able to start reducing the amount of testing at those sub-levels.”
For now, though, the additional testing is a significant burden in terms of cost and time: “You have to start that testing from day one of the development program, build up your database, and keep the certification authorities involved and informed of the test program and the correlation between your predicted results and the actual results,” Elliott says. “It’s vital to keep communication going throughout the development cycle so the authorities know what we’re doing.”
The authorities, he says, will want to view all the test plans, and the entire strategy for demonstrating compliance with the regulations. “As you put more detail around that and determine exactly what you’re going to test – and we’re talking about thousands of test specimens here – they will review those test plans and decide what is their level of involvement.” That may involve viewing the test plans and drawings, watching the specimen being manufactured or assembled, and/or being present for the actual testing. The US FAA, Europe’s EASA and Transport Canada are all involved in the CSeries certification process. “They all have different interests,” says Elliott. “Sometimes they demand to witness the same tests, other times they have particular areas of interest that they want to come and see.”
There are other tests, such as the ability to withstand bird strikes and impacts from runway or tire-burst debris. Fuel tank ignition prevention is another major concern, and one
with added complications for non-conductive composite structures: “In a metallic structure, a lightning strike attaches to wherever it happens to hit, flows through the structure and departs the aircraft, usually from the trailing edge or the wingtip. So the intent is to make sure that there’s nothing in there that’s going to cause the current to jump across from one component to another component and cause a spark that could ignite the fuel.”
All aircraft have to be certified with inert gas systems to ensure that the level of fuel vapor in the free space within the wing is low enough not to ignite, even if there is a spark. Bombardier also uses isolators to create an electrical barrier between system components, such as electrical harnesses and fuel or hydraulic pipes, and the wing structure, to prevent sparking or arcing. “That’s all the result of new regulations that have come into force within the last three or four years,” says Elliott. “Again, that entails an awful lot of testing.” Samples from small flat panels, right up through subsystems representative of the structure and installed systems, are sent to specialist test facilities such as Cobham Technical Services to have lightning strikes induced onto them. And there is further testing to demonstrate the structure’s ability to withstand a post-crash fire.
The five levels of testing involve the manufacture and assembly of more than 6,000 individual test pieces. Bombardier has around 20 test machines at its Belfast plant to test them both statically and in fatigue, Elliott says, “You’re putting them through literally hundreds of thousands of fatigue cycles to make sure that they retain their mechanical properties over the life of the project.”
The test specimens incorporate imperfections of the kind that may arise during manufacture and the sort of damage that might go undetected in service to ensure that worst-case scenarios are covered. And even with so many machines of its own, Bombardier still needs to subcontract about 20% of the testing to specialist test houses, laboratories and universities.
For full-scale testing, Bombardier Belfast has a team of people in its ground test and experimental department who design and manufacture the test rigs and set up the hydraulic and controlled loading systems. “We have three separate full-scale test articles to make sure that we understand the behavior of the airframe, including the wing, and that we can meet all the regulatory requirements,” Elliott says.
In Montreal, there is the complete aircraft static test article (CAST), an entire airframe that is put through hundreds of load cases to mimic all the conditions that the aircraft is going to encounter in service, such as high-speed maneuver, up-gust and down-gust wing bending, landing, ground maneuver and take-off.
For each condition there is a limit load representing the maximum load that the aircraft will see once in its life: “For up-gust cases, for example, we take the extreme, the absolutely worst case for an up-gust of wind which is going to bend the wings upward and load not just the wings but the fuselage attachments as well,” Elliott says. “Then to meet the certification requirements, we apply a 50% factor on top of that, so we test at 1.5 times that limit load for every one of those static cases.”
Then there is a full-scale test article for each of the major structural components. “We have a full-scale wing test article here in Belfast,” says Elliott. “On that article, we’re applying fatigue and damage-tolerance cycles. If the aircraft is designed to meet, for example, 60,000 cycles during its life, then we would test it to 60,000 times three, and again we go through all the different regimes from take-off, climb, cruise, with all the incidents or up-gust, down-gust, maneuver, whatever else may happen during the cruise, then coming in to land and landing. We take a complete spectrum of all the loads from that complete flight from take-off to landing and we come up with a fatigue spectrum.”
The fatigue spectrum is a statistics-based model. “Obviously you don’t want to assume that the aircraft will see all of those worst-case events on every single flight,” Elliott explains. “So we go through a statistical process to come up with a model that says the life of that aircraft over those 60,000 cycles is going to see so many of these cycle types or load patterns induced.” They are then compiled into a computer program, and over time the test installation replicates the 180,000 flight cycles with all the bending and twisting, and all the different types of loads.
“It’s a very dynamic case,” says Elliott. “You can literally stand back and watch the wing being bent upward and downward and twisted and put through its paces, replicating all those different elements of the typical and non-typical flights, and other flights with these extraordinary events in them from time to time.”
There is a complication, however. “The wing or any other composite structure is not entirely composite,” Elliott points out. “You will have metallic components in there. The environmental knock-down factor means the metallic components within that structure are actually seeing loads that they were never designed for. We don’t design for environmental degradation in the metal components because they don’t degrade over time. We’re actually penalizing the metallic elements of the structure by putting them through that test program.”
Accordingly, there is another full-scale test article at IABG in Germany: “They put it through metallic fatigue and damage-tolerance testing,” Elliott says. “They take basically the same load envelope, the same fatigue spectrum, but they don’t apply the same factors to it. So they’re testing all the metallic elements of the airframe in Germany and we’re specifically testing the composite elements of the wing here in Belfast.”
For the metallic fatigue testing, Elliott says a number of cycles agreed with the authorities, possibly a complete lifetime of 60,000 cycles, is carried out with pristine structure with little or no deliberately induced damage: “After we’ve completed 60,000 cycles, we’d go in and start adding local impact damage, literally firing metallic ball bearings at the structure to induce delaminations in the structure, and we test for another lifetime. And just to make sure, we would go in then and scan the structure on a regular basis, to ensure that the delaminations that we have induced with the initial impact are not growing to the point where they might reduce the load capability of the structure.”
After the second lifetime of testing, there is a thorough inspection of the test article to see if any cracks have started in the metallic structure. “If there are none, then we’ll start inducing cracks by putting little saw cuts into the structure in what we believe are the most highly loaded areas and then continue the testing by a few thousand more cycles and watch each of those crack sites to see whether the crack is growing at all, and if it is growing, how fast is it growing and therefore what the in-service inspection intervals would need to be.”
All this testing represents a significant additional burden, Elliott says, “Typically we talk about the weight advantage for composites. In some cases there can be a recurring cost advantage, because you’re making bigger components and therefore you can reduce the number of man-hours required to assemble those larger components. But you also have to take into account all this testing, which can cost tens of millions of dollars. That’s all part of the equation that determines whether or not going with a composite structure is a good idea or you just want to stick with the metallic structure.”
Metals and composites have dissimilar failure modes. Metallic structures tend to buckle and stretch. Elliott says, “You get a certain amount of yielding of the metallic materials. In tension, they stretch a certain amount before they break, or on the upper surfaces of the wing with the wing bending up you can see the metallic structure beginning to buckle.” But they can continue to carry load and remain safe despite this buckling and yielding.
Composites behave differently: “It’s just an instantaneous failure, it absorbs so much energy, so much load, and then it will just go, usually with a very loud bang. You don’t get any of that yielding or pre-failure activity; it’s either working or it’s not working.”
The test burden should ease as understanding of composites grows, Elliott adds. “The industry is getting much better at predicting the behavior of composite structures. On the full-scale static article, obviously the specimens are covered with hundreds of strain gauges, so we can measure exactly what load and what strain is going on in every element of the structure. So we do a correlation between what the finite model is telling us and what we’re actually seeing in the structure.”
The industry as a whole is getting very good at being able to predict those strain levels, he says. “That’s all part of building up confidence with the industry and the certification authorities so that over time we can reduce the number of specimens that we have to test and the number of different specific tests that have to be done.”
Bernard Fitzsimons is a journalist specializing in air transport business, technology and operations