Abstract: To capture specific characteristics of non-Newtonian fluids, during the past years fractional constitutive models have become increasingly popular. These models are able to capture in a simple and compact way the complex behaviour of viscoelastic materials, such as the change in power-law relaxation pattern during the relaxation process of some materials. Using the Lagrangian Smoothed-Particle Hydrodynamics (SPH) method we can easily track particle history; this allows us to solve integral constitutive models in a novel way, without relying on complex tasks. Hence, we develop here a SPH integral viscoelastic method which is first validated for simple Maxwell or Oldroyd-B models under Small Amplitude Oscillatory Shear flows (SAOS). By exploiting the structure of the integral method, a multi-mode Maxwell model is then implemented. Finally, the method is extended to include fractional constitutive models, validating the approach by comparing results with theory under SAOS.
Introduction: The Challenge of Viscoelastic History
Modeling complex fluids—such as polymer melts, biological tissues, or food products—presents a significant computational challenge. Unlike Newtonian fluids, these materials exhibit viscoelasticity, meaning their current stress state depends on their entire history of deformation.
Traditionally, researchers have favored differential constitutive models (like the standard Maxwell model) because they are numerically easier to solve. However, integral formulations are often more physically accurate for materials that exhibit complex power-law relaxation rather than simple exponential decay. The primary hurdle for integral models in standard mesh-based simulations (like Finite Element Analysis) is the "bookkeeping" required to track the deformation history of specific fluid elements as they move through a grid. This paper introduces a novel approach using Smoothed Particle Hydrodynamics (SPH) to overcome these limitations.
Methodology: SPH as a Natural Framework for Integral Models
SPH is a mesh-free, Lagrangian numerical method where the fluid is represented by discrete particles. Because these particles move with the flow, they act as self-contained units of fluid. This makes SPH the "natural home" for integral models: tracking history is trivial because the history is simply stored as an attribute of the particle itself.
The Lagrangian Advantage
The researchers developed a framework that directly computes Cauchy and Finger strain tensors from stored particle displacement history. To maintain computational efficiency, they implemented a memory management system that stores configurations at specific intervals and retains history only as far back as necessary (typically ten times the relaxation time). Furthermore, to handle the "weakly divergent" kernels often found in fractional models—where the math can become unstable as time approaches the present—the authors utilized an analytical approximation for the most recent time interval combined with Simpson’s rule for older history.
The Fractional Maxwell Model Framework
A key contribution of this work is the implementation of the Fractional Maxwell Model (FMM). While a standard Maxwell model uses a combination of a spring and a dashpot to represent elasticity and viscosity, fractional models introduce the "spring-pot." This element uses fractional calculus to describe behaviors that fall between a pure solid and a pure liquid.
As shown in the figure above, the FMM serves as a generalized framework. By adjusting the fractional parameters $\alpha$ and $\beta$, the model can be simplified into several special cases:
- Fractional Maxwell Gel (FMG): Occurs when $\beta = 0$.
- Fractional Maxwell Liquid (FML): Occurs when $\alpha = 1$.
- Standard Maxwell Model: The simplest case, recovered when $\beta = 0$ and $\alpha = 1$.
This modularity allows a single SPH "engine" to simulate a vast array of material behaviors simply by swapping the memory function $M(t-t')$.
Results and Validation
The proposed method was validated using Small Amplitude Oscillatory Shear (SAOS) flows, comparing the SPH results against analytical solutions for storage ($G'$) and loss ($G''$) moduli.
- Accuracy: The simulations showed excellent agreement with theoretical predictions for both standard and fractional models across multiple orders of frequency.
- Convergence: The model demonstrated stable results with a spatial resolution of approximately $50 \times 50$ particles.
- Stability: The integral approach was shown to converge toward traditional differential code results as history-saving parameters were increased, confirming the numerical stability of the history-tracking logic.
Conclusion and Future Outlook
This SPH-integral hybrid provides a robust pathway for simulating multi-scale materials that were previously too complex for moving geometries. While the current study focused on the linear regime (SAOS), the framework is inherently capable of handling non-linear deformations, such as those found in industrial plastic extrusion. Future work will focus on stress-testing these non-linear regimes and optimizing the memory requirements for large-scale 3D simulations through GPU acceleration.
Summary updated