Projects
The followings are selected projects that I have been the main researcher.
Modal analysis of blood flows in saccular aneurysms
What've we done?
High-fidelity Computational Fluid Dynamics simulations were performed and systematically coarsened to generate 72 cases with varying spatial and temporal resolutions. The Hankel DMD method was then used to identify dominant flow modes and instabilities within the aneurysm sac.
What've we found?
The results demonstrate that DMD analysis reveals unique patient-specific 3D flow structures independent of aneurysm geometry, and the energy pseudo-spectrum effectively captures flow dynamics even at low resolutions comparable to clinical imaging (CT or MRI). Aneurysmal hemodynamics can be categorized into 'laminarized flows' and 'transient dynamics' based on cumulative energy curves, offering a new quantitative tool that could significantly improve clinical rupture risk assessment.
Dynamics of White Blood Cell in Curved Microchannels with Trapezoidal Cross-Section
What've we done?
The study employed a hybrid continuum-particle computational framework to analyze white blood cell behavior in a spiral trapezoidal microchannel designed for cell sorting applications. The channel geometry featured a symmetrical trapezoidal cross-section created using cloud-based CAD software, with the WBC membrane modeled using 700 DPD particles. Simulations were performed under a very low Reynolds number.
What've we found?
Results revealed distinct flow patterns with higher velocity magnitudes near the channel's taller edge, creating asymmetric pressure gradients across the channel. The WBC exhibited significantly increased rolling behavior compared to rectangular channels, with anticlockwise rotation in the X-Z plane as it moved through the channel. These findings demonstrate that trapezoidal channel geometry significantly influences cell dynamics.
AMReX-based Hybrid Staggered/Non-staggered Grid Solver for Exascale Simulation of Incompressible Flow
What've we done?
We are developing a hybrid staggered/non-staggered grid solver built upon the AMReX library. The solver is designed to efficiently simulate incompressible flows on exascale supercomputers. Our approach uses the Fractional Step Method which decouples the Navier-Stokes equations. The momentum equation in a semi-implicit form is solved using 4-th order Runge-Kutta method. The pressure corrector is then calculated by solving the Poisson equation.
What've we found?
The solver can efficiently simulate both 2D and 3D flow problems. Test cases include 2D/3D lid-driven cavity flows, and 2D Green-Taylor vortex. The results show that the solver can achieve high accuracy and efficiency, validated by comparing with analytical solutions and benchmark data. Because the solver is still in development, a sample code written in Julia/Matlab is provided for demonstration purposes.
