Who Cares About Airline Passenger Safety During the Pandemic? Siemens and Airbus, Apparently

It’s been a year since the COVID-19 virus spread throughout the world, with much of it spread by passengers in commercial aircraft. Although the spread almost never occurred from passenger to passenger on board a flight, the resulting fear of flying has nearly brought the airline industry to its knees. However, these studies by Siemens and Airbus arrive in time for what industry analysts think will be a surge of leisure air travel resulting from pent-up demand, although business travel will lag, according to the Wall Street Journal.

Other studies of cabin airflow (there have been many) have used mostly featureless mannequins, but Siemens used highly detailed human models from its NX Human Modeling application. Siemens provided STAR-CCM+, part of its Simcenter analysis portfolio, and multiphysics to account for flow, evaporation, discrete particle flow and heat transfer to model particles and airflow. The particles in question range from large, that is, gobs of spit, to invisible particles that are small enough to be affected by Brownian motion.

The full, 35-minute presentation expertly presented by Prashanth Shankara, product marketing manager of Siemens Simcenter STAR-CCM+, and Sèbastian Dalla Barba, head of integration and innovation for Airbus, is well worth watching in its entirety here, which is available for free save the cost of your email address and a few other contact details. However, if you are more into just the highlights (spolier alert), then keep reading.

The number of factors affecting particles released in airflow from human exhalation are many, making accurate prediction a considerable challenge. They include, but are not limited to:

• Indoor/outdoor setting
• Temperature and humidity
• Airflow
• Mouth opening, head position, and size of the person
• Passengers breathing asychronously (this may be a first)
• Heat plume from human

There are plenty of other factors, but some are impossible to consider. For example, the fibers in a mask are too thin and too numerous to be modeled distinctly. Therefore, the mask as a whole is considered to be a baffle with leakage (5% to 40%, depending on the study). Even then, the Airbus model entailed 5 million cells and required half a day using 120 cores on an HPC.

Modeling on the scale of molecules for a simulation of an entire aircraft cabin is simply out of the question. The number of cells required would overwhelm supercomputers. Even high performance computing (HPC) on the cloud, billed as “limitless computing,” is incapable of solving models of the size that would result from such an endeavor. There would be 10²⁴ cells of 10^-7 m size—the diameter of a SARS-CoV-2 virus particle—in each cubic meter of an aircraft cabin volume. The largest CFD model we know of (Aerodynamic Drag in Cycling Pelotons: New insights by CFD Simulation and Wind Tunnel Testing, Dr. Bert Blocken, et al., Journal of Wind Engineering and Industrial Aerodynamics, 2018) had 3 billion (10⁹) cells.

To make the model more manageable, analysts use a form of shorthand that homogenizes the particle/fluid mix, with the smallest of particles in volumes several orders of magnitude greater in dimension. Simulation vendors refer to this as “multiscaling.” Siemens calls it a “parcel approach” and is able to model a million particles with only a thousand parcels.

While common enough, we should be reminded that multiscaling is an approximation, a departure from physics borne of necessity that requires that we take one leap of faith after another. Our first leap of faith was accepting discretization of a continium—such as occurs in finite element modeling and CFD "as fair and accurate. Multiscaling is the second leap of faith.

The mask modeled as a baffle is shown to be highly effective. As one would expect, the larger droplets hit the fabric of the triple layer masks modeled. Siemens’ simulation must allow the larger droplets to be trapped by the mask’s fibers. Only the smaller droplets escape, mostly from the gap on the sides of the mask but also from the limited sealing at the top and bottom of the mask, and waft in the air.

Airbus followed Siemens with its own simulation of passengers breathing, talking and coughing on an airplane. Researchers assumed a particle to be 94 percent liquid and 6 percent solid, with no evaporation to change the mix. Motion of the particle is governed by inertia, gravity and drag. Passenger models are using surgical masks with gaps between the masks and the models’ faces allowing a higher portion (20-40%) of leakage than was assumed by Siemens.

Sèbastian Dalla Barba reminds us that in Airbus’ planes, cabin air is renewed every 2-3 minutes, comparing it to a hospital, where air is renewed every 10 minutes, and office space, where it is renewed every 30 minutes.

In what may be worst-case scenario consideration, wishful thinking, or both, Siemens and Airbus both assumed full flights with passengers occupying every seat. Airbus concluded that aerosol droplets diluted by cabin ventilation is an adequate subsitute for social distancing on land, and that trapping small particles the size of viruses in HEPA filters and coughing into a mask will drop the number of airborne particles by 99 percent in less than two minutes. Its study “confirms the low risk of virus transmission by aerosols and droplets” because “the cabin ventilation reproduces the social distancing by diluting micro-droplets and quickly reducing their number by high air exchange.”

Cough simulation as presented by Siemens relies on the Lagrangian Multiphase (LMP) theory, which attempts to model particles within fluid. For this presentation, the particles are droplets and the fluid is air, but any type of spray, like fuel injection in automobiles or hairspray, can be modeled with LMP. It can be used for flow that contains discrete particles, such as blood flow, where the particles are blood cells and the fluid is plasma. Or sand flowing down a chute, where the particles are sand and the fluid is air. Furthermore, LMP can account for phase change between particle and fluid, such as evaporation of the droplets into the airThe behavior of particles when they make contact with solids, such as the walls of the aircraft cabin or the person in the next seat, is also accounted for.

We can tell when the particles stick, bounce or splash, says Prashanth, discussing the merits of Siemens’ Simcenter STAR-CCM+.

Both studies make the point that the actual rate of infection is to be determined by medical researchers as a matter of biology rather than physics. Engineering simulation models particles as lifeless and inert, and therefore needs medical science to complete the picture.

Nevertheless, the simulations are a credit to Siemens and Airbus, who both stepped up and volunteered their expertise and software knowledge to help during a crisis. Siemens could have been content with just selling its software to Airbus to help design and simulate, but instead it chose to employ its sophisticated simulation tools and devote time and resources to provide the most detailed, realistic and rigorous simulation done to date. Airbus could have deferred to the airlines to issue a study that showed they cared about airline customers, but instead it chose to take the lead and issue a study itself.

Kudos to both companies for stepping up.