SUMMARY

The goal of this research is to accurately describe the flow of blood in arteriole sized vessels. In small vessels, the cellular nature of blood is of utmost importance. The numerical method employed for the proposed research is based on a hybrid lattice-Boltzmann / finite element (LB-FE) implementation. A spectrin-link (SL) red blood cell (RBC) membrane model has been added to the LB-FE framework to capture deformations associated with high shear rates that exceed the capabilities of a linear elastic FE modeled membrane. For this presentation, the three dimensional LB method will be briefly discussed, but the SL method along with a coarse-graining procedure will be presented in detail. The existing sub-grid scale (SGS) models for contact and lubrication used with the LB-FE are coupled with the SL model for these suspension flows. For validation, the deformation of the axial and transverse diameters of an RBC mimicking the optical tweezer experiments have be performed. The validation of an isolated RBC in shear is also examined to capture tumbling and tank-treading motions for viscosity ratios and shear rates of interest. The deformations of an isolated RBC in the "wheel" configuration and parachuting in a microvessel sized tube have also be compared to previous experimental and numerical investigations. Investigations of rheological properties of blood will be examined leveraging the previously developed methodology used in conjunction with the LB-FE method. These studies will include simulating a semi-infinite suspension of RBCs using the Lees–Edwards boundary condition while altering shear rates, and hematocrit levels to monitor the effects on viscosity, particle pressure, and normal stress differences. The ability to track shear stress accumulation on platelets in a suspension of RBCs has also been added to the code capabilities. The flow of RBCs in arteriole sized rigid tubes has also been performed capturing the hematocrit distribution and cell-free layer near the tube wall. Further, computational comparisons of the LB-SL method to the LB-FE method are given for isolated RBCs. The parallel performance of the LB-SL method on up to 512 computational cores is also discussed.