TL;DR
This paper presents a novel approach using Smoothed Particle Hydrodynamics (SPH) to directly simulate flexible fiber suspensions, enabling accurate parameter identification for macroscopic fiber orientation models. By introducing a viscous surface traction term and fiber contact forces, the method achieves good agreement with analytical solutions and experimental data, paving the way for more reliable composite material simulations.
What Problem Does This Paper Address?
Simulating the flow behavior of discontinuous fiber-reinforced plastics is crucial for manufacturing high-quality composite parts. A key challenge is accurately predicting how fibers reorient during flow, which significantly impacts the final material properties. While macroscopic models exist, they often rely on parameters that are difficult and time-consuming to determine experimentally. Direct simulation of fiber suspensions, especially for non-dilute cases with fiber-fiber interactions, is computationally expensive and complex due to the fluid-structure interaction. This paper aims to bridge this gap by using a direct simulation method (SPH) to accurately capture fiber behavior and subsequently identify parameters for macroscopic orientation models.
Methods
The core of this work is the application of Smoothed Particle Hydrodynamics (SPH) to simulate fiber suspensions. The fibers are modeled as flexible bead chains, where individual beads are SPH particles.
- Fluid Model: A transport velocity formulation of SPH is used for the fluid phase, which is known for its robustness against tensile instability.
- Fiber Model: Fibers are represented as chains of SPH particles connected by elastic forces (tension and bending).
- Novel Viscous Surface Traction Term: A key innovation is the introduction of a viscous surface traction term. This term accounts for the hydrodynamic forces acting on the fiber's surface, compensating for the fact that the fiber is modeled as a line rather than a 3D object. This is crucial for achieving accurate rotation behavior consistent with Jeffery's equation.
- Fiber Contact Forces: To handle non-dilute suspensions where fibers interact, a model for fiber-fiber contact forces is implemented. This includes surface-surface and end-surface contact scenarios, employing a penalty approach (Hertzian contact force) to prevent penetration.
- Parameter Identification: The identified fiber orientation model is the Reduced-Strain Closure (RSC) model. Parameters for this model (interaction coefficient $C_1$ and phenomenological factor $\kappa$) are determined by fitting the ensemble average of the fiber orientation tensor obtained from SPH simulations to the RSC model's predictions. This fitting is done using a least-squares approach.
- Implementation: The simulations are performed using the PySPH framework, which allows for flexible implementation of custom physics.
Datasets
Not discussed in the paper. The study relies on simulations and comparisons with existing analytical solutions and literature data.
Key Insights and Ideas
- SPH as a Powerful Tool for Fiber Suspensions: The paper demonstrates that SPH, with appropriate extensions, is a capable method for simulating complex fiber suspension dynamics, including fluid-structure interaction and fiber-fiber contacts.
- Importance of Viscous Surface Traction: The introduction of the viscous surface traction term is critical for accurately capturing the rotational dynamics of single fibers, bringing the simulation results in line with Jeffery's equation. Without it, fibers stop rotating prematurely.
- Contact Forces are Essential for Non-Dilute Regimes: For suspensions with higher fiber concentrations, fiber-fiber contact forces are indispensable for realistic behavior and preventing particle penetration.
- Direct Simulation for Parameter Calibration: This work validates the use of direct fiber simulations as a robust method for identifying parameters in macroscopic fiber orientation models, reducing reliance on costly experiments.
- Scale Separation is Key for Macroscopic Models: The study highlights that macroscopic models require sufficient scale separation and a large number of fibers to provide reliable predictions, as evidenced by the decreasing deviation between individual realizations with increasing fiber volume fraction.
Key Results
- Single Fiber Rotation: The SPH simulation of a single stiff fiber in shear flow shows rotation periods that are slightly faster than the analytical solution of Jeffery's equation, attributed to finite Reynolds number and simulation domain effects. However, the agreement is good, especially with the viscous surface traction term.
- Single Fiber Bending: The simulated bending modes of single fibers qualitatively agree with literature results, though differences arise due to the inextensible nature of fibers in some prior models versus the stretching capability in this work.
- Parameter Identification for RSC Model: The paper successfully identifies optimal parameters ($C_1$ and $\kappa$) for the RSC orientation model for different fiber volume fractions. These parameters show good agreement with literature values.
- Ensemble Averaging: The use of ensemble averaging over multiple simulation realizations is shown to be effective in obtaining statistically representative results and reducing the impact of initial configuration randomness.
- Stationary Orientation State: Simulations show that suspensions tend towards a stationary orientation state, with the time to reach this state decreasing for higher fiber volume fractions.
Limitations and Future Work
- Fiber Stiffness: While the method is not inherently limited, the parameter identification examples primarily use stiff fibers ($S=100$) to ensure unambiguous interpretation of the orientation tensor. The authors acknowledge that the method is not limited to rigid fibers.
- Fiber Fracture: The current model includes a criterion for fiber fracture based on bending angle, but further investigation into this aspect could be beneficial.
- Other Flow Types: The current study focuses on shear flow. Future work will extend the investigation to arbitrary initial configurations, flexible and breaking fibers, and other flow types beyond shear flows.
- Standard Deviations: The authors plan to assess standard deviations to enable modeling of uncertainties related to fiber orientation models.
- Resolution: The paper notes that the low resolution and smoothing in SPH might make it less accurate than methods like Stokesian Dynamics for resolving the exact flow field near particles, though it offers computational efficiency.
Summary updated