Loading summary
Transcript1 lines
- [00:02]
A
The 787 is a fascinating machine. Conceived for a changing, rapidly digitizing world, it leveraged new materials and emerging computational methods to produce unprecedented gains. In today's video, a techno economic exploration of what brought about the 787 and its challenging design and how the computer helped it soar. William Boeing was born on October 1881 in Detroit, Michigan. Ah, you thought I was going to do all that, weren't you? Well, you and I both can breathe a sigh of relief. But the economic origins of the 787 and how those affect its technologies are fascinating. And let's talk about that. Throughout the 1970s and 1980s, Boeing dominated the long haul high capacity plane market with their 747. Back then, if you wanted a plane that can carry over 350 people, your only choice was the jumbo jet. For decades, 747s carried hundreds of passengers on trunk routes between key major airports like Narita, JFK and Heathrow. Passengers then take secondary flights to their final destination. This is the famous hub and spoke model. By contrast, we have the point to point model where passengers travel straight to the final destination without any layovers. Hub and spoke became the dominant route system in the wake of the airline Deregulation act of 1978. This is because it lets the airline service the most final destinations with the fewest routes. If you have five destinations, including the hub itself, hub and spoke lets you connect to all five with just four additional routes. And doing this with a point to point system, on the other hand, would require offering 10 routes. And the more routes you offer, the more planes you need, which means laying out more money. So Hubbin spoke profits thrown off by the 747, estimated by Airbus to be as high as $40 million per plane at one point subsidized Boeing's other models. So Airbus built the A380 plane at an estimated cost of $12 billion to challenge the 747 and and slay the cash cow. Boeing struggled to find a proper response to this new superjumbo challenge. First, they announced the 747X, which would extend the 747 literally. The plane looked like a stretched out heftier 747 and was capable of carrying up to 500 passengers on trunk routes to key hubs. But the late 1990s was a time of change for the airliner industry, exposing some of the weaknesses of the old hub and spoke model. While hub and spoke might let airlines cover more destinations and fly less planes, it is also operationally complex. Luggage will be lost, and passengers often have to endure layovers. Moreover, hub airports in major urban areas like New York City and Chicago started to get congested, leading to missed flights and delays. At the same time, the emergence of the Internet and online travel agencies like Expedia in the late 1990s made it easier for passengers to compare flights and choose what they preferred. And they preferred shorter flights and less layovers. Boeing's airline customers made it clear that they did not need so much capacity. Instead, they wanted something with a bit more range and efficiency. Airlines had already started to scale back use of the 747 in favor of smaller Boeing 767s and Airbus A340s. So in March 2001, Boeing canceled the 747X and at the same time announced an exotic looking plane called the Sonic Cruiser. I once knew a dog named Cruiser. The cruiser would carry 200 to 300 passengers at high altitude at just under the speed of sound, 15 times faster than other planes. Boeing's thinking was that additional speed would cut down travel times for business travelers who would want to pay more for that extra speed. But environment advocates and Airbus criticized the Cruiser's environmental impact. In June 2001, European Environmental Commissioner Margaret Wahlstrom criticized Boeing, saying that an hour shaved off a transatlantic flight was not worth a 35% increase in carbon emissions. For their part, Boeing's airline customers were concerned about the operating costs. And that was before 911 smashed worldwide air ridership numbers and inflicted $7 billion of losses on the global industry. Shortly after that, geopolitical instability in the Middle east caused oil prices to skyrocket. One of the airline's highest costs is fuel, Counting for anywhere from 20 to 30% of total operating costs, depending on hedges, the Cruiser would have been dazzling. But with the airlines all in survival mode, they made it clear that they valued efficiency far more than the speed. So in December 2002, Boeing, not without a bit of wistfulness, cancelled the Cruiser and announced that its technology would be rolled into the 7E7. The 7E7, later named to the 787, would be a plane with longer range and 20% better fuel efficiency. The value proposition immediately resonated with the airlines, and they ordered the new plane in record breaking numbers. To achieve those goals. Boeing implemented innovations like a new engine. But perhaps the single biggest change was the extensive use of composite materials. Special engineered materials made from two or more constituents. An example being the carbon fiber reinforced plastic, or cfrp, provided by Toray Industries in Japan. These composites are not only 70% lighter than steel and 40% lighter than titanium, but also offer superior resistance to fatigue and corrosion, Two major things that affect airplanes. Composites have been used in airplanes since the 1940s, but in limited amounts for specific parts like the plane's fairing, nose and cockpit. There are still metals in the 787. Aluminium for the wing and tail's leading edges, Titanium for the engine, and so on. But composites take the lead. The fuselage, tail, doors, interior and wings. No commercial plane has before used so much composite material. 50% of weight and 80% of volume. To compare, composites make up just 9 to 12% of the Boeing 777 by weight. There are 32,000 kg of composites in the 787, saving over 18,000 kg of weight as compared to a similarly sized metal plane. The upsides are great, but of course, having all these new composites in such volume presented massive challenges. A major one was in assembly. You produce this carbon fiber by first wrapping it like a tape or wallpaper around massive molds of the plane's fuselage sections. Then we cure it under heat and pressure. An early problem discovered during assembly was the presence of air bubbles due to faulty molds or inconsistent application of the tape. These bubbles inherently weaken the overall structure and can also collect water. At high altitudes, the water would freeze and expand the possibly cracking the fuselage. But another major challenge of using these composite materials involved their design. A challenge most pressing in the 787's most important components, its wings. The wing must generate enough lift force to keep the plane in the sky. At the same time, we want to minimize drag. There are several types of drag, but they are all forces created by the air that resists the plane's forward thrust. Reducing drag saves fuel, and it adds up. One Boeing airline customer estimated that for a Boeing 777, reducing drag by 1% saves over 300k per plane each year. So the wing's design is a complex three dimensional structure that must trade off between factors like size, weight, the aircraft's use, desired speeds, the performance metrics, the lift, drag, and so on. At its core is the wingbox, which starts at about roughly the plane's middle and extends to about two thirds of the wingspan. It provides structural integrity and also contains a fuel tank. Inside the wingbox are ribs and spars. The ribs run the wing's width from its leading to trailing edges and help maintain its aerodynamic shape. The spars run through the wing's span or length. And the are shaped kind of like an I beam, they carry the wing's bending loads. Wrapped around all of this is the wing's skin. The skin helps distribute pressure to the ribs and the spars. Stiffeners might be added to help keep the skin from bending. Over the years, wing structure and composition have evolved with technology. The first planes had wingboxes of wood and skin of fabric before evolving to metals. The metals have their limits, so the industry continues to try new material candidates. The 787 was the first commercial aircraft to ever have a wing made from composites. Not only that, 787's composite wing box is larger than any previous composite component. At 15 meters long and 5 meters wide, it is bigger than the F2 fighter's wing box and the Boeing 777's vertical fin. Now, making the wing out of composites offers compelling upsides. A composite wing can be made lighter and thinner than a metal one, producing the same amount of lift, while also cutting down on fuel burning drag. By contrast, a thinner metal wing suffers strength issues, so designers have to add stronger structures, thicker ribs or spars, thicker skin, what have you, all of which adds weight. So a composite wing offers the potential of thinness without compromising on strength. But achieving those gains is no easy task. Composites are different. Metals are uniform all throughout, with the same strength and stiffness in all directions. We call this behavior isotropic. Composites, on the other hand, are heterogeneous. They're made up of thousands of carbon fiber layers, or piles, stacked on top of one another and embedded within an epoxy. Their properties are anisotropic. In other words, their strength and stiffness depend on how we orient these piles in relation to the wing's axis. 45, 90 degrees, 20 degrees, so on. Changing the angle will change strength and stiffness. While these composites can support the same structural loads as metal, they also flex. This flexibility can cause the wing to change its shape as it flies, altering the plane's flight dynamics and making it less predictable and harder to control. And if that was not enough, the design has to be manufacturable and affordable. Composites can produce geometries that metals cannot. But they too have their limits. And they cost way more than metals. 100 to $200 per kg back then, compared to just a few dollars for metals. So to sum up the problem, creating the composite wing requires the team to craft both a design and a brand new material at the same time. Having so many variables presents a massive exploration space to search through for many years. New plane Designs were tested in wind tunnels, and wind tunnels still have their uses. You're dealing with actual air, which is a very good real world grounding. And if you want to test a whole lot of variables at the same time, a wind tunnel is quite cost effective at the same time. Wind tunnels aren't perfect. They do not always replicate the conditions of real world flight. For example, they have walls, and users need to mount the model. Both the walls and mounts can influence the final results, so designers have to apply corrections. And with a physical item in the real world, your insight is inherently limited. Think of a flowing river and trying to measure the speed and direction of the water in that river, you stick a sensor into it and read the water flow there. But what about the other parts of the river? To get readings there, you add more sensors and so on and so on. But you can only add so many. In the case of a wind tunnel model, in order to obtain detailed surface pressure measurements, the designers need to drill many holes called taps. This is not only time consuming, but also only gives you a measurement at that particular tap. And there are only so many you can do. This ties into the larger issue between building the actual thing and setting it up for the test. Wind tunnels cost a lot of time and money. If you want to run a smaller test just to see what happens, then building a whole model and putting it into a wind tunnel is a little bit of an overkill. But what can you do? So Starting in the 1970s, a new tool joined the aerodynamic design process. Computational fluid dynamics. I mentioned these programs in a few prior videos relating to cfd for chip design or weather prediction. A computer runs a set of mathematical equations describing the motion of fluid flow and solves them for complex geometries. CFD's greatest strength is that it can quickly and cheaply run a number of simulations on a small part of the design. It's not perfect. A lot depends on the equations and the input variables. But it saves you from having to go to the wind tunnel. Over the years, computers have gotten faster. Mathematicians have developed governing equations like the Reynolds average Navier Stokes equations. And by the late 1980s, CFD emerged as a real tool that designers can use to complement the wind tunnel. The ideal flow thus became to design as much of the plane and its wing inside the computer first, and then only use the wind tunnel as a real world sanity check. The goal is to get as much right as possible on the first try. Back in 1986, Boeing designed 77 wings for wind tunnel tests for the Boeing 700. 67 for the 787. 22 years later, Boeing built just 11 wings. At the same time, the amount of CFD compute soared. By 2006, the company racked up over 800,000 hours of compute time on Cray supercomputers. The design used 60 times more CFD runs than the 777. Interestingly, the 7077 and the 787 saw differences in how the two teams used the CFD tool. The Boeing 777. Am I saying enough? 7s or too many? I don't know. Designed in the late 1980s and early 1990s, was the first commercial plane whose design included some form of computational fluid dynamic technology. For that plane, the Boeing team at first started with a traditional cut and try methodology where you first cut out the shape and try to analyze its performance in the air. But the team quickly found that they were not able to hit their aggressive goals. So the team turned to something called inverse design, where they work backwards from a goal, like an ideal pressure distribution over a wing surface. The computer then runs CFD algorithms to calculate the best wing geometry to meet those ideals. Such a method required trust in the program's ability to accurately predict things like the distribution of pressure across the wing. In 1990, the Boeing 777 team did over 19,000 runs of CFD production code. The cancelled Sonic Cruiser would have further pushed the limits of design with cfd. This was due to wind tunnel limitations and air dynamics as models approach Mach 1 speeds. CFD was what made the Sonic Cruiser technically feasible. And the tool proved itself for the 787 too, but not without a little turbulence. First, I refer to a fascinating interview conducted by Ben Thompson of Stratecheri with Joris Port, one of the former lead engineers on the Boeing 787 wing. Port is now the CEO of Rescale, a high performance computing cloud platform startup. Anyway, in the interview, Port related how the first version of the 787 wing design used composites like as if they were metals, simplifying their full complexity. The internal phrase was something like black aluminium, referring to how they essentially assumed metal like behavior in the carbon fiber for simplicity's sake. Plenty of manufacturers, like those making carbon fiber bike frames, for instance, do it this way. But the downside of this approach is that it does not take full advantage of the material's strength and weight benefits. So the first 787 wing came out way too heavy. Extra weight costs airlines extra fuel, so the contracts have heavy weight penalties. So The Boeing design team had no choice but to abandon the black aluminum approach and find some way to wrangle the material's complex exploration space. This leads us to how the 787 took the next logical step of computational direct optimization. With direct optimization, the computer iteratively adjusts the many variables of a wing's design that we discussed. The material, structural design, aerodynamics. What have you to maximize for a desired performance metric, like the lift to drag ratio, drag itself, weight or fuel efficiency. Such a process was practical not only due to improvements in computational power, but also mathematical breakthroughs in the late 1980s, like Anthony Jameson's adjoint method. The adjoint method lets us calculate how sensitive those performance metrics are to changes in the design variables by deriving a second equation called an adjoint equation. This saves us from having to run full CFD simulations for every possible variable per mutation. Port recalls stitching together compute power across supercomputers in multiple Boeing divisions to run what he called optimization algorithms. This had to be done over the weekend because nobody believed that the effort would actually deliver. In the end, the 787 wing team was able to deliver a thinner, lighter wing that saved Boeing $1 billion in weight penalties. A noticeably flexible wing too. During takeoff, the 787's wingtips can deflect up to 12% of its semi span up to 3 meters. And it was more than just the wing. Direct optimization was applied across the whole design, optimizing the wing, body and engine housing shapes all at the same time to achieve the best possible compromise. Before we conclude, I want to thank Ben Thompson for inspiring this video idea. Today we remember the 787 for a few things, most of which are pretty negative, like the plane's complicated and controversial supply chain strategy, where Boeing outsourced 70 to 80% of the plane to companies around the world. Many people remember the plane for its long delays, roughly three to four years from 2007 to 2011. Not quite as long as intel delayed its 10 nanometer process, but a long time. And many people also remember the plane for its electrical troubles after the first deliveries. And many, many people can only think right now of the company's current struggles. Boeing, yet another former American technology giant, now trying to turn itself around. All of this is true, but I don't think it should forever define this plane. Boeing, in the end, delivered a beautiful thing. Exotic materials. Moore compute new algorithms. The 787 demonstrates how far modern manufacturing technology has gone and is a monument to the continued awesome power of Moore's Law. Alright, everyone, that's it for tonight. Thanks for watching. Subscribe to the channel. Sign up for the Patreon and I'll see you guys next time.