Our validated approach with UQ
Our research laboratory conducts development and evaluation of medical devices and patient-specific hemodynamic analysis using Computational Fluid Dynamics (CFD). The application of CFD analysis in the field of cardiovascular medicine requires validation of its reliability, which is a significant issue. The reliability of values computed through CFD can be assessed through standardized validation tests.
The FDA benchmark problem is one such validation test provided by the U.S. Food and Drug Administration (FDA) to validate CFD results for medical devices. In our laboratory, we conducted validation by simulating the Pump and Nozzle models provided by the FDA, comparing them with experimental data and previous research results. Through this validation test, our goal was to ensure the reliability of CFD and establish an appropriate methodology for CFD simulation.
** FDA validation test and UQ results by our researcher Minjun Kang (KHU) directed by Prof. J. Seo
FDA Pump Benchmark Test
FDA Pump Benchmark is a benchmark provided by the US FDA to validate CFD tools. Through modeling the fluid dynamics inside a rotating pump, it allows for understanding velocity and pressure variations within the pump as well as useful for gaining insights into critical areas that are not easily accessible.
We have examined the magnitude of velocity extracted along the radial lines. Our simulation results matched the marked lines from the PIV results. Both graphs exhibit nearly identical peak velocities and peak points, confirming a general consistency in the trend of the graphs.
FDA Nozzle Benchmark Test
The FDA nozzle is a benchmark problem provided by the U.S. FDA for validating Computational Fluid Dynamics (CFD) tools. The nozzle represents various shapes included in medical devices for blood transportation such as blood vessels, blood dialysis sets, catheters, cannulas, syringes, and more. The FDA has designed standardized nozzle shapes for cardiovascular devices for CFD validation and published data from research groups that experimentally measured fluid phenomena occurring here. The designed nozzle model is intended to encompass various behaviors of blood flow, including acceleration, deceleration, shear stress, and recirculation. The flow occurring in the nozzle encompasses laminar conditions to transitional and turbulent conditions. We have successfully validated CFD results from this model against experimental results.
FDA Nozzle - Uncertainty Quanfication
Our study provides an in-depth exploration of how uncertainty in turbulence models affects the reliability of computational fluid dynamics (CFD) simulations, particularly in the context of cardiovascular medical device design. Using the FDA nozzle as a standardized benchmark, we focused on the turbulence model, which is commonly employed in simulating fluid flow. We specifically examined how variations in the empirical constants in the turbulence model influence the accuracy of simulation results.
By treating these constants as stochastic variables and applying Monte Carlo sampling, we quantified the uncertainty and analyzed its propagation through the simulations. The study revealed that empirical coefficients interact nonlinearly, causing significant variations in jet flow characteristics and the recirculation area, which are critical for simulating realistic blood flow conditions.
The findings underline the critical role of precise parameter selection in turbulence models to improve the reliability and predictive power of CFD simulations. We contribute to understanding of how turbulence modeling impacts CFD outcomes but also provide valuable insights for optimizing medical device designs, ensuring both safety and efficacy in clinical applications