A Bidirectional Coupling Procedure Applied to Multiscale Respiratory Modeling | JoVE Visualize

Area of Science:

Computational Biology and Physiology
Biomedical Engineering
Fluid Dynamics

Background:

Accurate modeling of pulmonary mechanics and airflow is crucial for understanding respiratory diseases.
Integrating lower-dimensional models of distal lung mechanics with 3D computational fluid dynamics (CFD) of upper airways presents significant computational challenges.
Existing methods often lack efficiency and flexibility in coupling these disparate modeling approaches.

Purpose of the Study:

To develop and validate a novel multiscale computational framework for efficiently linking lower-dimensional lung mechanics models with 3D CFD models of pulmonary airways.
To incorporate physiologically appropriate outlet boundary conditions into integrated pulmonary models.
To enable flexible and computationally efficient coupling of different model types for biomedical research.

Main Methods:

Developed a multiscale computational framework extending the Modified Newton's Method with a nonlinear Krylov accelerator.
Implemented extensions for retaining subspace information and a special correction for efficient timestep acceptance (average of one residual evaluation per timestep).
Introduced a 'pressure-drop' residual for stable coupling between 3D incompressible CFD and lower-dimensional fluid systems, applicable to respiratory and non-respiratory flows.

Main Results:

The novel framework demonstrated efficient coupling of ordinary differential equations (ODEs) representing distal lung mechanics with imaging-based 3D CFD models.
Performance was comparable to monolithic schemes, achieving results with minimal computational overhead (often a single CFD evaluation per time step).
Validated the hybrid CFD-ODE models against ODE-only models in both idealized and imaging-based human lung geometries.

Conclusions:

The developed multiscale framework allows for the efficient and flexible integration of pulmonary CFD models with lower-dimensional lung mechanics models.
The method significantly reduces computational cost compared to traditional approaches, facilitating complex respiratory system modeling.
The framework's modular design offers broad applicability for biomedical researchers to focus on specific model components and design.