Beyond Representative Volume Elements for Random Heterogeneous Materials

Introduction

As a result of the production process, many industrially relevant materials show a heterogeneity of their composition on a lower scale, leading to an uncertainty in their material properties. Traditionally, computational multiscale methods are built around the concept of Representative Volume Element (RVE), which seeks deterministic effective properties on cells that are sufficiently large.

Limitations of RVE

However, the concept of RVE is both too crude and too restrictive for today's engineering requirements.

Crudeness of RVE

• The RVE concept is too crude because of its insensitivity to the material randomness, which is crucial for assessing the reliability of structures.

Restrictiveness of RVE

• The RVE concept is too restrictive because representative volumes need to be truly gigantic for certain multiscale materials, e.g., long-fiber reinforced thermoplastics (LFTs), precluding a computational treatment in reasonable time.

Introduction of BeyondRVE

BeyondRVE introduces and studies microstructure-uncertainty quantifying volume elements (QVEs), which account for the dispersion of the effective properties on cells of finite size.

Methodology

Me and my team will build up a groundbreaking multiscale methodology, taking into account the latest neural-network technology and screening the spurious boundary layers which arise for digital volume images of microstructures.

Goals of BeyondRVE

Methodologically, BeyondRVE intends to provide a novel microscale solver which combines:

1. The efficiency of regular-grid methods
2. The accuracy of boundary-conforming meshes
3. Fast and precise microstructure-generation tools for a variety of heterogeneous and composite materials

Integrated Development

Within BeyondRVE, these complementary pieces of simulation technology will be developed in an integrated and interdisciplinary fashion.

Expected Outcomes

Upon completion, a significant boost for the nonlinear mechanics of heterogeneous materials and lightweight design is expected, providing multiscale methods with more expressive results in shorter time for larger classes of materials with higher complexity.