Topology optimization for 2D swirl flows with application in the design of a ventricular assist device.

• Ano de defesa: 2022
• Autor(a) principal: Alonso, Diego Hayashi
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Bomba tesla; Hemólise; Hemolysis; Otimização topológica; Tesla pump; Thrombosis; Topology optimization; Trombose; Turbulence; Turbulência
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/3/3152/tde-07072022-094033/

Ventricular Assist Devices (VADs) substitute or temporarily assist the heart and blood circulation by means of a small-scale blood pump. Due to the operation with blood, their designs aim to minimize blood damage (hemolysis and thrombosis) and maximize the performance (i.e., more efficient pumping). For that, the topologies of the rotor and the volute play a major role. In this research, the rotor is based on the 2D swirl concept, which considers swirling axisymmetric flow, being represented mainly by the Tesla pump. In a Tesla pump, the boundary layer (viscous) effect pumps the fluid, which may lead to less blood damage than bladed pumps (pumping based on the variation of linear momentum). However, the obtained efficiency is normally low, which may be improved by using the topology optimization method. Therefore, in this research, the topology optimization method is implemented for the 2D swirl flow model, also considering a non-Newtonian model for blood (Carreau-Yasuda) and hemolysis/thrombosis models to quantify the blood damage. Furthermore, the 2D swirl flow topology optimization is extended to the Wray-Agarwal turbulence model (WA2018), which presents advantages from the simulation and topology optimization points-of-view. The solid material is modeled as a porous medium with permeability controlled by the topology optimization. The numerical implementation is performed mainly through the FEniCS platform, by using the dolfin-adjoint library for the automatically derived adjoint model, and also relying on the OpenFOAM® software for considering higher-speed flows, turbulence and 3D simulations. For the optimization, the IPOPT and TOBS solvers are considered. The optimized designs are interpreted and fabricated through additive manufacturing, and then experimentally evaluated.