Frontier supercomputer helps Boeing tackle aviation’s toughest physics challenges

Designing a safer, faster, and more fuel-efficient aircraft depends on understanding one of physics’ most difficult problems: turbulence — chaotic airflow. ​To tackle that challenge, Boeing researchers used Frontier, the world’s first exascale supercomputer at the U.S. Department of Energy’s Oak Ridge National Laboratory (ORNL), to simulate how ice buildup and microscopic turbulent airflow affect aircraft performance during critical phases of flight such as landing. ​The simulations are expected to help engineers improve aircraft designs, explain results that have puzzled wind tunnel experiments, and accelerate the development of more fuel-efficient airplanes. Even small aerodynamic improvements can make a measurable difference. According to the airline’s research, reducing aerodynamic drag by just 1 percent during cruise flight can cut fuel consumption by 0.75 percent, lowering airline operating costs and potentially reducing ticket prices. Turbulence still remains physics’ hardest problems Airflow around an aircraft constantly shifts between smooth, orderly motion and chaotic turbulence. Predicting exactly when and how that transition occurs remains one of the biggest unsolved challenges in fluid mechanics. ​The difficulty lies in the enormous number of air molecules interacting near the aircraft’s surface. Researchers have long focused on the boundary layer—the thin layer of air flowing closest to a wing—where tiny changes in airflow determine how much drag the aircraft experiences. ​Capturing those interactions requires simulations that examine airflow at an almost microscopic scale. The more detail engineers include, however, the more computationally demanding the calculations become. Simulating billions of airflow calculations Aircraft aerodynamics are typically studied using wind tunnels, flight tests, and computational fluid dynamics (CFD) simulations. ​Wind tunnel testing and flight experiments provide valuable data but are expensive and time-consuming. CFD offers a faster alternative by dividing an aircraft into a digital mesh containing hundreds of millions of tiny cells, some only millimeters across. ​The software calculates airflow properties—including pressure, velocity, and temperature—for every cell, then combines them into a complete picture of airflow around the aircraft. ​Most engineers rely on Reynolds-averaged Navier-Stokes (RANS) simulations, which estimate turbulence instead of calculating every swirl of air. The method is fast enough for full aircraft simulations but cannot accurately capture rapidly changing turbulent flows. ​Higher-fidelity approaches provide much greater accuracy but demand vastly more computing power. Exascale computing cuts months of work to days That is where Frontier made the difference. ​Capable of performing 2 quintillion calculations per second, the exascale supercomputer enabled Boeing to run simulations that would otherwise take months on conventional industrial systems. ​The researchers used the system to study how different ice shapes and textures influence wing performance and to model turbulence in far greater detail than previously possible. The simulations also helped evaluate aircraft anti-icing systems and explained previously unexplained wind tunnel observations. ​According to Boeing, the higher-fidelity models predicted an aircraft’s lift coefficient to within 1 percent, giving engineers greater confidence to move promising designs toward production. ​The team believes the findings could benefit the wider aerospace industry, as the simulation methods are based on a widely studied aircraft model that other researchers can build upon to improve future aircraft designs.Get the latest in engineering, tech, space & science - delivered daily to your inbox.