This item is under embargo and not available online per the author's request. For access information, please visit http://libanswers.wustl.edu/faq/5640.

ORCID

http://orcid.org/0000-0002-1560-999X

Date of Award

Spring 5-15-2020

Author's School

McKelvey School of Engineering

Author's Department

Mechanical Engineering & Materials Science

Degree Name

Doctor of Philosophy (PhD)

Degree Type

Dissertation

Abstract

Rapid development of computer power and computational fluid dynamics (CFD) technology has made CFD a critical component of analysis in many engineering disciplines; however its use in standardization of cardiovascular assist devices through analysis and design is still in its infancy. U.S. Food and Drug Administration (FDA) created a “Critical Path Initiative” and reached out to industry professionals and academics with knowledge in CFD to participate in simulation of flow in a blood contacting nozzle seen in many medical devices and in a blood pump. These two “Critical Path Initiative” projects were aimed at determining the best practices for CFD simulations such as the grid quality, numerical algorithms, turbulence models and boundary conditions, etc. A large number of people participated in these projects including the author in the CFD group at WUSTL under the direction of Dr. Agarwal. The results for the nozzle case have been compiled in a report and released by the FDA. However, due to the complexity of the blood pump project, since 2014 only a portion of the results have been released by the FDA with the full data analysis to be released in the future. Some highlights using data from this author include acknowledgement from the FDA for being one of a small group of participants in the pump benchmark CFD simulation and his data was used in a paper in the ASAIO’s annual conference which was the winner of the Willem Kolff Top Abstract prize and was also published in an ASME conference paper. Simulations of the nozzle, blood pump and a left ventricular assist device (LVAD) cannula are conducted in this dissertation and the machine learning technique of a genetic algorithm (GA) is applied to optimize the shape of these devices in order to minimize the blood damage which characterizes the hemolysis or rupture potential of red blood cells. These “shape optimizations” are conducted by automation with no user interaction after the initial setup for definition of boundary conditions, mesh generation and evaluation of intermediate results. “Optimizations” differ from “design improvements” where the physics and the resulting shape of the profile are not easily determined by intuition or “guess and improve.” Here, the results are loft surfaces of various orders that best describe the fluid physics as opposed to conventional parametric studies or design of experiments approaches. Results for the original and shape optimized devices are presented to demonstrate the effectiveness of optimized devices in reducing the blood damage. For the original configurations, CFD computations are validated using the experimental data or other computations depending upon their availability.

Language

English (en)

Chair

Ramesh K. Agarwal

Committee Members

Kenneth L. Jerina, David A. Peters, Palghat A. Ramachandran, Swami . Karunamoorthy,

Comments

Permanent URL: https://doi.org/10.7936/hnsd-5332

Available for download on Wednesday, March 05, 2121

Share

COinS