HLRB Project pr32ne
Coupled Problems in Computational Modeling of the Respiratory System
Lehrstuhl für Numerische Mechanik {bisher: Lehrstuhl für Werkstoffe im Maschinenbau}
Proposing Institution
Lehrstuhl für Numerische Mechanik {bisher: Lehrstuhl für Werkstoffe im Maschinenbau}
Project Manager
Dr.-Ing. Michael Gee
Boltzmannstr. 15
85747 Garching
Abstract
Mechanical ventilation is a vital supportive therapy for critical care patients suffering from respiratory insufficiencies in view of oxygen supply. However, a number of associated complications often occur, which are collectively termed ventilator induced lung injuries (VILI). Understanding the reason why the lung still often becomes damaged or inflamed during mechanical ventilation is the key question sought within the German Research Foundation's interdisciplinary priority program "Protective Artificial Respiration". We are interested in the influence of air flow patterns in the conducting zone on the development of VILI. Despite a reduced modeling of the unresolved lower parts, we currently need to restrict the number of finite elements in a CT-based 10-generation airway model to 2.5 million corresponding to computing times of approximately 11000 CPU hours. Considerably finer meshes are actually needed for modeling smaller airways accurately and enabling the simulation of regional oxygen distribution in the future, suggesting computing times of about 50000 CPU hours. VILI is known to occur mainly in the respiratory zone of the lungs, i. e. on the alveolar level. In order to investigate the mechanical stresses developing in lung tissue during mechanical ventilation, a comprehensive model of pulmonary alveoli was developed. Due to the geometric complexity, a simulation of a small cube of tissue with an edge length of 300 microns already involves the solution of equation systems with 5 million unknowns. Straining this parenchymal cube by 10% hence requires computing times of several weeks on 30 processors.Since already the separate models of parts of the lung are computationally very intensive, simulating the respiratory system on the whole resolving all 23 generations of airways ending in 300 million alveoli is impossible from the outset. Therefore, we are working on novel multi-field and multi-scale approaches bridging the gap between the resolved parts of the lung. Computations are performed with the massive parallel multi-purpose finite element C++-code "Baci" developed at the Institute for Computational Mechanics.References1. Wiechert L. Rabczuk T. Gee M.W. Metzke R. Wall W.A.; Coupled Problems in Computational Modeling of the Respiratory System. High performance computing in vector systems; Springer, Berlin, Germany; 20072. Wiechert L. Metzke R. Wall W.A.; Modeling the mechanical behaviour of lung tissue at the micro-level; Journal of Engineering Mechanics;accepted; 2009
