Fiber suspensions, both naturally occuring and synthetic, have many of applications, such as rheological modifiers and composite reinforcement. The rheological properties strongly depend on the microstructure, which varies with fiber properties, concentration, and processing methods. Rigid fiber suspensions form liquid crystalline phases. Flexible fiber suspensions form homogeneous networks and aggregates. In this thesis, fiber-level simulations were applied to systematically investigate the relationship between the microstructure and macroscopic properties, including the viscosity, normal stress differences, and diffusivities. Simulation programs were accelerated and parallelized for GPUs via CUDA. Rigid fibers were modeled as spherocylinders that interacted only through a soft repulsive force. Brownian dynamics simulations were employed to obtain the translational and rotational diffusivities, matching reported values of hard spherocylinder suspensions. Liquid crystalline phases, including nematic, smectic, and solid phases, were obtained. For suspensions that were isotropic at rest, flow curves, which contained two shear thinning regions bracketing a viscosity plateau at intermediate Pe̹clet numbers, qualitatively matched those for suspensions of cellulose nanocrystals. For suspensions that were nematic at rest, system-wide domains that aligned and kayaked about the vorticity direction, domains that rotated in the gradient direction, and layered domains were observed under shear. The transient rheological properties depended on the domain dynamics. Flexible fibers were modeled as spherocylinders that were connected by joints with bending and twisting potentials. The fibers were non-Brownian, and interacted through soft repulsion, short-ranged attraction, and friction. A drying and rehydration process was implemented. The volume fraction of the suspensions were raised 4 fold and subsequently lowered to the original value. The simulation box was shrunk and expanded with a constant number of fibers. During drying, the fibers were moved affinely after the equations of motion were integrated. When sufficient friction and attractive forces were applied, the viscosities for suspensions that were dried and rehydrated were lower than those for suspensions that were not dried and rehydrated. The reduction of viscosity was associated with the formation of dense and persistent flocs, which were observed experimentally for microfibrillated cellulose suspensions.

When elastic fibers are immersed in a Newtonian fluid, the behavior of the system, or the "fiber suspension" becomes non-Newtonian. Understanding the dynamics of such systems is of particular interest in a wide variety of fields, including locomotion of microorganisms, paper and pulp industry, microfluidics etc. When these fibers are immersed in the fluid at low Reynolds number, the elastic equation for the fibers couples to the Stokes equations, which greatly increases the computational complexity of the problem. I have simulated buckling behavior of a single fiber suspended in a shear flow and have applied two numerical method, a slender body approximation known as Local Drag model and a regularized Boundary Integral method known as Regularized Stokeslet method. We have extended the Local Drag model to simulate naturally bent fibers based on the target or resting curvature of the fibers. Microorganisms possess fiber-like organs known as the appendages and swim by flapping these organs. We have also simulated swimming motion of a simple microorganism model by imposing a time dependent resting curvature of these fibers.

Fiber suspensions, both naturally occuring and synthetic, have many of applications, such as rheological modifiers and composite reinforcement. The rheological properties strongly depend on the microstructure, which varies with fiber properties, concentration, and processing methods. Rigid fiber suspensions form liquid crystalline phases. Flexible fiber suspensions form homogeneous networks and aggregates. In this thesis, fiber-level simulations were applied to systematically investigate the relationship between the microstructure and macroscopic properties, including the viscosity, normal stress differences, and diffusivities. Simulation programs were accelerated and parallelized for GPUs via CUDA. Rigid fibers were modeled as spherocylinders that interacted only through a soft repulsive force. Brownian dynamics simulations were employed to obtain the translational and rotational diffusivities, matching reported values of hard spherocylinder suspensions. Liquid crystalline phases, including nematic, smectic, and solid phases, were obtained. For suspensions that were isotropic at rest, flow curves, which contained two shear thinning regions bracketing a viscosity plateau at intermediate Pe̹clet numbers, qualitatively matched those for suspensions of cellulose nanocrystals. For suspensions that were nematic at rest, system-wide domains that aligned and kayaked about the vorticity direction, domains that rotated in the gradient direction, and layered domains were observed under shear. The transient rheological properties depended on the domain dynamics. Flexible fibers were modeled as spherocylinders that were connected by joints with bending and twisting potentials. The fibers were non-Brownian, and interacted through soft repulsion, short-ranged attraction, and friction. A drying and rehydration process was implemented. The volume fraction of the suspensions were raised 4 fold and subsequently lowered to the original value. The simulation box was shrunk and expanded with a constant number of fibers. During drying, the fibers were moved affinely after the equations of motion were integrated. When sufficient friction and attractive forces were applied, the viscosities for suspensions that were dried and rehydrated were lower than those for suspensions that were not dried and rehydrated. The reduction of viscosity was associated with the formation of dense and persistent flocs, which were observed experimentally for microfibrillated cellulose suspensions.

Sheet Molding Compounds (SMC) are discontinuous fiber reinforced composites that are widely applied due to their ability to realize composite parts with long fibers at low cost. A novel Direct Bundle Simulation (DBS) method is proposed in this work to enable a direct simulation at component scale utilizing the observation that fiber bundles often remain in a bundled configuration during SMC compression molding.