Research

Coronary artery disease is the leading cause of death amongst men and women in the United States. Kawasaki Disease is the leading cause of acquired heart disease amongst children. Both these diseases cause flow abnormalities in the coronary arteries, which perfuse the myocardium. While these flow abnormalities can lead to devastating cardiovascular complications, most clinical methods to assess abnormal flow are either invasive or inadequate. Our research focuses on developing new techniques to non-invasively diagnose and plan treatments for patients affected by these diseases using personalized computational models of blood flow in the coronary arteries.

The human cardiovascular system includes vessels that have diameters from the centimeter to the micron scale. Each of us has unique circulatory features and is uniquely affected by diseases. We develop computational models that account for this wide range of scales, and are personalized to each patient. Using routinely measured clinical data as well as novel imaging in conjunction with optimization and reduced-order modeling techniques, we develop multi-scale cardiovascular flow models that are personalized to each patient. The aim of this research is to develop personalized treatments, gain additional insights from routinely measured clinical data, and predict health outcomes at scales beyond routine measurements by combining clinical data with physics-based models.

The force partitioning method is a physics-based and data-enabled technique to disentangle the contribution of individual features (vortices, shear layers, etc.) and physical mechanisms (added mass, momentum diffusion) to the forces induced on surfaces in complex viscous flows. This technique has uncovered previously unknown physics in flows around oscillating wings and bluff-bodies. Combined with data-driven and reduced-order models, FPM opens promising avenues to dissect the physics of a wide variety of complex, viscous fluid structure interactions problems such as bio-inspired swimming/flying and renewable energy harvesting.

Fluid flows interact with flexible and moving bodies in a wide range of natural and engineering applications. These problems are often characterized by highly non-linear flow physics due to the generation and shedding of several vortices, their interactions, and the forces they induce on surfaces within the flow. We use computational modeling to understand the interaction of flexible surfaces with fluid flows, with the aim of understanding the underlying physics and controlling these oscillations. Flow-induced oscillations are relevant in a wide variety of applications, from aircraft wings to emerging renewable energy harvesters and the functioning of heart valves.

Uncertainty quantification is crucial for robust engineering predictions and design. This is especially true for applications that are informed by measurements, such as clinical data, which are potentially noisy. In such applications, we work with collaborators to incorporate Bayesian techniques to estimate model parameters and novel data-driven uncertainty quantification and model reduction methods to propagate uncertainties in our data and modeling techniques to computational predictions.