Deep material networks for fiber suspensions with infinite material contrast

Benedikt Sterr,Sebastian Gajek,Andrew Hrymak,Matti Schneider,Thomas Böhlke

2024-06-17

Abstract:We extend the laminate based framework of direct Deep Material Networks (DMNs) to treat suspensions of rigid fibers in a non-Newtonian solvent. To do so, we derive two-phase homogenization blocks that are capable of treating incompressible fluid phases and infinite material contrast. In particular, we leverage existing results for linear elastic laminates to identify closed form expressions for the linear homogenization functions of two-phase layered emulsions. To treat infinite material contrast, we rely on the repeated layering of two-phase layered emulsions in the form of coated layered materials. We derive necessary and sufficient conditions which ensure that the effective properties of coated layered materials with incompressible phases are non-singular, even if one of the phases is rigid. With the derived homogenization blocks and non-singularity conditions at hand, we present a novel DMN architecture, which we name the Flexible DMN (FDMN) architecture. We build and train FDMNs to predict the effective stress response of shear-thinning fiber suspensions with a Cross-type matrix material. For 31 fiber orientation states, six load cases, and over a wide range of shear rates relevant to engineering processes, the FDMNs achieve validation errors below 4.31% when compared to direct numerical simulations with Fast-Fourier-Transform based computational techniques. Compared to a conventional machine learning approach introduced previously by the consortium of authors, FDMNs offer better accuracy at an increased computational cost for the considered material and flow scenarios.

Computational Engineering, Finance, and Science

What problem does this paper attempt to address?

The paper aims to address the problem of predicting the effective stress response of fiber suspensions in non-Newtonian solvents. Specifically:

1. **Research Background**: Fiber-reinforced composites are widely used in transportation and energy sectors due to their high specific stiffness. In the design process, accurately predicting the viscosity of local fiber suspensions is crucial for correctly estimating process parameters and other engineering quantities. However, in component-scale simulations, fully resolving the entire microstructure of interest to calculate suspension viscosity is computationally infeasible. Therefore, analytical and computational homogenization methods become important tools for providing viscosity estimates.

2. **Main Contributions**: This paper extends the Direct Deep Material Networks (DMNs) framework and proposes a Flexible DMN (FDMN) architecture to handle rigid fiber suspensions with infinite material contrast. Specifically:
- It proposes two-phase homogenization blocks to handle incompressible fluid phases and infinite material contrast.
- It establishes the necessary non-singularity conditions to ensure that the effective properties of coated layered materials are non-singular even when one phase is rigid.
- It constructs and trains FDMNs to predict the effective stress response of shear-thinning fiber suspensions, demonstrating validation errors below 4.31% under different fiber orientation states, loading conditions, and shear rates relevant to engineering processes.

3. **Experimental Results**: FDMNs perform excellently in predicting the viscous response of fiber-polymer suspensions, capable of handling incompressible fluid phases and infinite material contrast. Compared to previous methods, FDMNs offer better accuracy and flexibility, albeit at a higher computational cost.

Through this work, the paper addresses the accurate prediction of the effective viscosity of fiber suspensions under complex flow conditions, providing strong support for practical engineering applications.