Computational Fluid Dynamics & the Lungs

Because of the complexity of this branching structure, physical scale-model research on lung airflows is limited to about three generations of branching. The small airways deep within the lung, furthermore, are inaccessible to most experimental techniques. The group's goal is to use CFD tools to simulate flows in these inaccessible regions. With accurate computer modeling, it should be possible to develop the detailed understanding of lung flow patterns and particle transport that can make new aerosol medications a reality.

"If we want to deliver medications to the spot where the lung doesn't want them to go," says Hammersley, "we have to be more sophisticated. We must understand the transport process and learn to redirect the particles -- by modifying their speed, density or their electrical charge. With computer modeling, we can provide information that will avoid inappropriate clinical trials and wasted effort."

Using a grid generation program, EAGLE, developed by the Department of Defense, the group has produced grids that mimic the complexly curved walls and branching of the small airways. This has been a daunting task in that EAGLE was designed to simulate flows over high speed aircraft, which have a much more regular geometry. "With computations at PSC, we've shown not only that we can accurately reproduce airway geometry, but that it's possible to extend these grids and flow solutions to virtually any tubular biological structure."

In lung airways, the region where a parent branch splits into two daughter passageways, called the "carina," is the most difficult region to model. It requires generating grids that accurately reproduce the "central flow splitter" and wall geometry.

The researchers used two different flow solver codes to simulate flows over this three-dimensional grid -- NASA's INS3D and a new code developed by collaborators at NSF's Engineering Research Center for Computational Fluid Simulations. Because each flow-splitting branch requires between 125,000 and 450,000 grid points, these are very demanding calculations. The results to date, carried out to three generations of branching, compare well with flow measurements made in scale models.

Future objectives are to include flexible airway walls and to simulate how particles of different sizes are deposited. "We want to be sure the simulations match up well with experiment," says Hammersley, "then we can take the next step, to give us answers we can't get experimentally."

These graphical results from 3-D lung airflow modeling on the CRAY C90 depict flow velocity at selected cross-sections in a single bifurcation. Velocity (in dimensionless units) ranges from zero (black) through dark blue to red to maximum (white).

go back to the main page