Convincing the aerospace industry to push the “I believe“ button on large-scale software simulations
On Dec. 17, 1903, the Wright brothers made the world’s first successful powered airplane flight on a beach in Kill Devil Hills, N.C. Their 12-hp, chain-driven Flyer I biplane flew for 12 seconds and traveled 120 feet (37 m). Forty-four years later, U.S. Air Force pilot Chuck Yeager broke the sound barrier in a Bell X-1 research aircraft. Knowing that his wife might soon be a widow, Yeager named the plane “Glamorous Glennis” in honor of her. And on July 20, 1969—just 63 years after two bicycle mechanics changed the world—Neil Armstrong and Buzz Aldrin piloted the Apollo 11’s lunar module Eagle to the Moon’s Sea of Tranquility. Armstrong set foot on the lunar surface six hours later.
Those who designed these historic air and spacecraft did their work on drafting boards. There was no CAD software back then, no finite element analysis (FEA), computational fluid dynamics, multibody dynamics, or any of the other advanced simulation tools—just big rolls of paper, plenty of sharp pencils, and lots and lots of math.
No offense to these and other aerospace pioneers, but the mathematical calculations required to build and deploy modern aircraft is far greater than that needed in Chuck Yeager’s day. It’s a good thing, then, that equally modern computer systems and software make such number crunching both manageable and cost-effective.
“Other pilots died while attempting to break the sound barrier,” noted Todd Tuthill, vice president of aerospace and defense for Siemens Digital Industries Software, Plano, Texas. “Yeager was successful because he figured out his X-1 needed far more authority from its flight control system. The X-1 engineers redesigned the horizontal stabilizers so that the whole surface would pivot in response to stick commands, rather than just a small flap on the surface’s trailing edge. Fortunately, we can now simulate most of what Yeager and others had to learn the hard way, eliminating the need for dangerous test flights.”
The “Right Stuff” notwithstanding, that’s welcome news for test pilots and their families, and it’s also good for airplane manufacturers. Thanks to advanced simulation tools, designers understand the physics of flight in ways that were never before possible. They’re able to develop military aircraft fast and agile enough to have had Yeager looking for the ejection handle, while those in the far more conservative commercial flight industry have made similarly huge strides in fuel economy, passenger comfort, sound mitigation, and a host of other factors that can make or break an airline.
Tuthill noted that manufacturability ranks high on the latter list. In addition, there are understandably generous safety margins on all components, along with strict weight, durability, and cost requirements. “The target for most commercial aircraft components is 80,000 hours of flight,” he said. “Go much longer than that and the part could weigh too much, cost too much to build, or both. It’s really about designing for a predetermined lifespan and an optimized manufacturing process.”
Accomplishing this significant feat requires simulating for every foreseeable circumstance, what’s known in the engineering community as a Monte Carlo analysis. “I could put 1,000 engineers with 10,000 spreadsheets in a room for a year and they wouldn’t be able to solve what large-scale simulation can do in a matter of a couple of days,” Tuthill said. “A Monte Carlo simulation allows me to optimize my design for any given cost or performance parameters in ways that were impossible just a few years ago.”
One of the reasons these simulations were impossible is computing power, or rather the lack thereof. Massive server farms and cloud-based computing have changed all that, which is why many in this space have turned to providers such as Amazon Web Services. Here, a design team can leverage as much hardware as needed to meet project analysis and scheduling requirements, and do so with minimal financial penalties.
“Think about the complexity of a modern commercial aircraft,” said Tuthill.
“Just looking at the electronics systems, you probably have hundreds of boxes containing many thousands of components, all talking to other aircraft systems. It’s only through software simulation that systems this complex can be visualized and then optimized.”
Steve Bleymaier has plenty to add on this subject. The chief technology officer for aerospace and defense at Ansys Inc., Canonsburg, Pa., Bleymaier served in the U.S. Air Force for nearly as long as Yeager. He’ll tell you that today’s simulation environment is multi-dimensional, able to analyze everything from the performance characteristics of a single microchip on a backup system to how the aircraft will interact with others in its flight space.
Bleymaier refers to the latter of these capabilities as mission engineering. “Let’s say you’re flying from New York to Chicago,” Bleymaier explained. “An engineer can simulate the complete aircraft along its entire route and understand what each subsystem will experience along the way, whether that pertains to fluid temperatures, structural dynamics, air flow, electrical interference … everything.”
What’s more, they perform all of this analysis within a virtual environment, he added, long before finalizing the design or beginning any manufacturing. It then becomes the authoritative source of truth for the aircraft—its digital twin, in other words—from concept design to production, in-service sustainment, and, ultimately, the aircraft’s end-of-life.
Interestingly, Ansys didn’t begin its mission engineering story with aircraft. The year after Neil Armstrong made his giant leap for mankind, Westinghouse Astronuclear Laboratory (WANL) engineer John Swanson proposed using a mainframe computer and software written on punch cards to simulate the inner workings of a nuclear reactor. When his employer told him that the current manual approach to FEA analysis was just fine, thank you, Swanson left to form his own company.
His first customer? Westinghouse. “Back then, Ansys was known as Swanson Analysis Systems,” Bleymaier said. “We’ve since expanded well beyond structural and mechanical engineering into other physical domains, including optical, thermal, electrical, and fluid simulation.”
Bleymaier ticked off the numerous benefits of these comprehensive simulation capabilities, all of which relate to the digital twin. By having a digital twin, for instance, manufacturers can connect it in real time to the physical asset. They can capture data from onboard sensors and external systems to gain visibility into real-world performance. It can be used to understand anomalies and predict failures, for product improvement, operational sustainment, or maintenance purposes, and the list goes on.
“You have ready access to everything about the aircraft or racecar or naval vessel all the way back to its concept and design,” Bleymaier said. “The knowledge isn’t sitting on some engineer’s desktop, or worse, in his brain. Everything is in the twin. It’s all connected, and you never have to recreate anything or speculate about what happened (or was supposed to happen) at some point in the product’s lifespan. That’s the ideal situation for any manufacturer.”
Jennifer Peeples, technical director for aerospace and defense at Hexagon Manufacturing Intelligence, North Kingstown, R.I., agrees that the world of engineering simulation and analysis has become much larger over recent years. She stated that manufacturers in the aerospace and defense industries, as well as various government entities, utilize the company’s “large family of FEA and multibody dynamics solutions, not to mention other systems engineering products.”
To those yearning to understand the behavior of complex assemblies, the latter of these—multibody dynamics, or MBD—yields greater insight. Peeples explained that FEA allows manufacturers to discretize an individual structure into small elements and then apply boundary conditions to analyze external forces such as mechanical stress or loading. MBD, on the other hand, is concerned with the gross motion of a complete electromechanical system.
“Instead of evaluating a single strut or other landing gear component as you would with FEA, multibody dynamics analysis looks at the landing gear as a whole,” she explained. “For example, if I apply a certain force through the hydraulic system, will that be enough to lower the landing gear in the allotted time given the expected air resistance? What impact will that air resistance have on fuel consumption? And then, once the landing gear is fully deployed, how well will it handle the impact with the runway? Are the stresses within allowable limits? What about fatigue over many thousands of landings? Multibody dynamics allows us to solve pretty much any problem related to structural motion or load, interaction between systems and controls, and so on.”
As Peeples is quick to point out, the answers to all these questions probably won’t come from a single simulation or even the same software. This is why Hexagon and other solution providers have developed (or, in some cases, acquired) large numbers of individual engineering tools, each focused on specific types of analyses. For example, the company’s Adams product does most of the heavy lifting where MBD is concerned, while its CAEfatigue handles fatigue analyses, Cradle CFD solves for computational fluid dynamics, and MSC Nastran answers questions about structural integrity.
These tools can share information with others in the platform, allowing users to determine how a component or assembly will perform under a wide range of conditions.
Sound complex? It is. The landing gear analyses just described is but one small (though critical) example of a commercial airliner’s hundreds of subsystems, each of which is managed by an engineering team striving for optimal design performance, and each of which may impact related—and often dependent—systems. The question then becomes: How do the many pieces of this massive, hugely complex jigsaw puzzle fit together?
“One of the recent developments we’re really excited about on the multidisciplinary structural analysis front is the use of modules,” Peeples beamed. “With large-scale simulations, modules allow you to break a large model into smaller sections, assign them to different people or teams, and then bring everything together periodically for a macro view. This not only provides greater visibility and efficiency, but eliminates challenges like the elaborate node numbering schemes that have long been used in such complex engineering analyses.”
Aside from the classical domains such as fluid dynamics and mechanical structures, there are also significant calls for other types of large-scale simulation, according to Swen Noelting, director of SIMULIA aerospace and defense industry enablement at Dassault Systèmes.
“Consider electromagnetics and the need for high-speed internet connectivity.” he continued. “You have more and more antennas in the airspace, all of which need to perform correctly and not interact with each other in unplanned ways. There is also increased concern over sound mitigation, and of course, fuel economy is always high on the list. These are just a few examples of the industry’s many challenges.”
Noelting reiterated Peeples’ insight about multibody dynamics. “We and others are looking at how different aircraft components work together, like what happens during deployment of the landing gear or lowering flaps, and what effect these events have on the rest of the airplane, the pilot, and even the passengers.”
Simulation brings far more to the aerospace table than a robust understanding of electromechanical events and avoidance of system failures. Because simulation reduces (and sometimes eliminates) the need for wind tunnels, flight tests, and other types of expensive, time-consuming physical validation, it allows aircraft designers to reduce the risk of discovering failures late in the design process, relying on simulation to do what once took months or even years to achieve.
Time savings aside, this doesn’t come without cost. “These kinds of multiphysics optimizations are something that we see more and more, but they require huge amounts of compute power,” Noelting asserted. “That’s not just a factor of the large model sizes but also the number of simulations and the different disciplines that engineers need to evaluate in each one.
Regardless, designers now have the ability to optimize airplanes for lightweighting purposes, drag, and lift performance throughout the entire flight envelope, structural integrity, component fatigue, and much more, and they can execute all these different simulations on a single platform—such as Dassault Systèmes’ 3DEXPERIENCE Platform—which provides a single source of truth about the geometry and layout of the aircraft as it changes throughout the development process.”
Advanced, large-scale simulation is opening other doors as well. Noelting suggested that several well-known aircraft manufacturers have “taken a step back” of late, and are exploring what the next generation of commercial airliners might look like, given the industry’s recent advancements in manufacturing and design capabilities. This might lead to the elimination of traditional, nearly a century-old tube-and-wing architectures in favor of novel aircraft concepts, such as blended-wing aircraft; and new propulsion methods, such as electric or open rotor engines, an effort that—here again—will depend heavily on simulation.
“All aircraft that have been developed over the past 50-60 years look much like the previous one and the previous one before that,” said Noelting. “Moving to a completely new design will mean setting aside many of the simulation methodologies that have been templated around these traditional designs.”
Noelting explained that everything about aircraft design in this future world will require revalidation using a more or less whiteboard approach. So not only will the simulations change significantly, but they’ll also play a much more significant role, he said, noting that manufacturers simply can’t build an aircraft from scratch using physical testing and certification methods. The reasons are simple: It’s too expensive, and it takes too much time.
“That’s another huge factor, even without throwing novel aircraft designs into the equation,” added Noelting. Although few manufacturers have met their time and cost milestones over the past two decades, everyone is trying to shorten their development cycles through simulation. That’s why the current industry buzzword is certification by analysis. Of course, large OEMs will still have to do flight tests—nobody wants to step into an airplane that hasn’t been flown before. But these tests will be more of a final confirmation rather than an integral part of the development process.
Noelting concluded by saying that simulating all performance aspects of the complete aircraft before it ever leaves the ground for the first time will allow manufacturers to reduce the expensive and time-consuming flight test phase from more than a year to a few months—shortening the overall development cycle and saving hundreds of millions of dollars. “I think that, by the end of this decade, we will certainly come close to that, and it won’t be too long after that this grand vision becomes a reality.”
Connect With Us