Design and fabrication of biphasic cellular materials with transport properties – A modified bidirectional evolutionary structural optimization procedure and MATLAB program

Shiwei Zhou, Joseph Cadman, Yuhang Chen, Wei Li, Yi Min Xie, Xiaodong Huang, Richard Appleyard, Guangyong Sun, Qing LiInternational Journal of Heat and Mass Transfer 2012 · DOI

TL;DR

This paper introduces a novel, modified Bidirectional Evolutionary Structural Optimization (BESO) method to design complex cellular materials with tailored heat and mass transport properties. By incorporating a "soft-kill/soft-revive" approach and a gradually decreasing step size, the method achieves more robust and efficient optimization, leading to designs that can be fabricated using 3D printing.

What Problem Does This Paper Address?

Designing materials with specific transport properties (like thermal conductivity or permeability) is crucial for advanced applications in fields ranging from aerospace to tissue engineering. These properties are heavily influenced by the material's microstructure, often composed of repeating unit cells. While homogenization techniques can predict the bulk properties of such materials, designing microstructures to achieve desired properties has been challenging. Traditional topology optimization methods, like the original BESO, often struggle with numerical instability and can be inefficient due to their "hard" element-based updates. This paper aims to overcome these limitations by developing a more sophisticated BESO algorithm that allows for finer control over material distribution and a more systematic approach to inverse homogenization.

Methods

The core of this work is a modified Bidirectional Evolutionary Structural Optimization (BESO) procedure. The key innovations include:

  • Soft-kill/Soft-revive Scheme: Instead of completely adding or removing material from an element, this method allows for a partial alteration of the material fraction within an element. This provides finer control and helps avoid oscillations, especially when sensitivities are similar.
  • Relaxed Volume Constraint: The strict volume constraint is relaxed, allowing for fluctuations around the target value, which diminishes as the optimization converges. This offers more flexibility in the design process.
  • Gradually Decreasing Step Size: The step size for updating material fractions decreases over iterations. This allows for rapid initial progress towards a general design sketch and then finer convergence to the optimal microstructure.
  • Nonlinear Diffusion Technique: A nonlinear diffusion term is incorporated into the objective function to suppress checkerboard patterns, a common issue in topology optimization that can lead to unrealistic designs.
  • Hashin-Strikman (HS) Bounds: These theoretical bounds are used as target properties for the optimization, providing a rigorous framework for designing materials with extremal transport characteristics.
  • MATLAB Program: A comprehensive MATLAB program is provided to facilitate the implementation and understanding of the proposed algorithm.

Datasets

Not discussed in the paper. The paper focuses on computational design and experimental validation of specific designs, rather than training on large datasets.

Key Insights and Ideas

  • "Soft" updates are crucial for complex material design: The transition from hard, element-wise updates to soft, fractional updates significantly improves the robustness and convergence of the optimization process for microstructural design.
  • Gradual convergence is key: A decreasing step size allows the algorithm to explore a wider design space initially and then refine the microstructure with greater precision, leading to better final designs.
  • HS bounds provide a powerful target: Using theoretical HS bounds as optimization targets allows for the design of materials with predictable and extremal transport properties.
  • Topology optimization can yield known high-performance structures: The method successfully reproduces well-known microstructures like Vigdergauz-like structures and Kagomé lattices, which are recognized for their superior transport properties.
  • 3D printing bridges theory and practice: The ability to fabricate the designed microstructures using multi-material 3D printing demonstrates the practical relevance and manufacturability of the computational designs.

Key Results

The paper presents several 2D and 3D examples demonstrating the effectiveness of the modified BESO method:

  • Successful design of biphasic cellular materials: The algorithm successfully designed microstructures targeting both lower and upper Hashin-Strikman bounds for transport properties.
  • Achievement of target properties: The effective transport properties of the designed microstructures were found to be very close to the targeted HS bounds (e.g., achieving K = 1.6667I for the lower bound and K = 1.800I for the upper bound in 2D examples).
  • Fabrication of prototypes: Several designed 2D and 3D microstructures were successfully fabricated using stereolithography (SLA) multi-material 3D printing, validating the manufacturability of the designs.
  • Comparison with traditional BESO: The modified BESO method showed improved convergence and robustness compared to a traditional BESO approach, particularly when dealing with intermediate density elements.

Limitations and Future Work

  • Computational Cost: While improved, topology optimization can still be computationally intensive, especially for 3D problems with fine meshes.
  • Experimental Validation: While prototypes were fabricated, the paper acknowledges that "validating theoretical design and predictions via experiment is left to subsequent research." This suggests a need for further experimental testing to confirm the predicted transport properties.
  • Material Properties: The current work focuses on isotropic materials with two distinct phases. Future work could explore anisotropic materials and more complex multi-phase composites.
  • Other Transport Phenomena: The framework is presented for general harmonic equations, suggesting potential for application to a wider range of transport phenomena beyond heat and mass transfer.

Summary updated

BibTeX
@article{zhou2012design,
  author = {Shiwei Zhou and Joseph Cadman and Yuhang Chen and Wei Li and Yi Min Xie and Xiaodong Huang and Richard Appleyard and Guangyong Sun and Qing Li},
  title = {{Design and fabrication of biphasic cellular materials with transport properties – A modified bidirectional evolutionary structural optimization procedure and MATLAB program}},
  journal = {International Journal of Heat and Mass Transfer},
  year = {2012},
  doi = {10.1016/j.ijheatmasstransfer.2012.08.028},
}
Figure