According to the Center for Disease Control (CDC), coughing is a symptom of COVID-19. Coughing is also a transmission mechanism for the disease and understanding the fluid mechanics of cough events can help establish guidelines for the public. Computational Fluid Dynamics (CFD) simulations with SIMULIA PowerFLOW can provide valuable insight into the physics of human coughs which can help establish guidelines for the public.
The starting point for this effort was a literature survey of coughing. We identified peer-reviewed journal articles that experimentally measured mass flow as a function of time and particle size and distributions of different human subjects which served as boundary conditions for our high fidelity simulations. There was also information on the angle at which coughs exit the mouth of the subject.
Our initial simulations involved setting an inlet condition at the subject’s mouth and enforcing the particle distribution[i]" target="_blank" rel="nofollow">[i] and transient flow profile[ii]" target="_blank" rel="nofollow">[ii]. Unfortunately, the resulting jet looked nothing like the experimental results we had. The jet was far to uniform and smooth lacking the large scale structures expected from a naturally turbulent flow.
Our initial solution was to overcome this by introducing synthetic turbulence at the inlet boundary condition. This drastically improved the quality of the results but created a lot of unknowns about what turbulence qualities to seed the flow with. Since there were no papers available about the characteristic turbulent length scales and turbulence intensities that we needed as inputs we felt a bit stuck. Then inspiration struck, where does the turbulence come from? Pondering this question we realized the turbulent flow likely developed because of the geometries within the body the flow interacts with. For example, we expected that the teeth would act as a trip for the turbulence.
Running with this idea we switched to a manikin that had geometry for the mouth and moved the inlet inside of its mouth. This again improved the results but we were still not getting the right exit angles for the flow and the turbulence structures were still too small. Just as with many other fluid flow simulations we had done in the past, improving the realism of the geometry had improved the accuracy of the flow simulation but we had not yet gone far enough. We needed to deeper into the throat of the subject.
Moving the inlet back past the uvula into the oropharynx. Once we did this we found that synthetic turbulence was no longer needed. The turbulent structures developed naturally as did the exit angle of the cough. With the new position of the inlet, we had eliminated a lot of the unknown parameters we were feeding into the simulation such as initial cough angle, initial particle speed, and inlet turbulence properties. Now the particle diameter distribution and the transient flow profile were enough to describe out boundary conditions.
Centerline slice of velocity showing the generation of turbulence through interaction with the mouth geometry.
Along with this discovery, we also realized that we have the ability to now look at the flow within the oral cavity within the mouth, something that would be very difficult to do experimentally. What we saw was the interaction with the soft pallet and the teeth seem to be very important to the cough flow development and those small discrepancies between our manikins oral cavity and that of the experimental subject. While we are not sure how this discovery will help in the fight against COVID-19 and other communicable diseases, we hope that the improved realism in our simulations will.
[i]" target="_blank" rel="nofollow">[i] Zayas et al. BMC Pulmonary Medicine 2012, 12:11 http://www.biomedcentral.com/1471-2466/12/11
[ii]" target="_blank" rel="nofollow">[ii] Gupta, J., Lin, C.‐H., and Chen, Q. 2009. “Flow dynamics and characterization of a cough,” Indoor Air, 19, 517-525
