Design optimization of regular hexagonal thin-walled columns with crashworthiness criteria

Shujuan Hou, Qing Li, Long Shu-yao, Xujing Yang, Wei LiFinite 2007 · DOI

TL;DR

This paper presents a method for optimizing the crashworthiness of hexagonal thin-walled columns, crucial for energy absorption in vehicles and aircraft. By employing response surface methodology, the researchers efficiently explore various hexagonal column designs to maximize energy absorption while adhering to safety limits on peak impact forces.

What Problem Does This Paper Address?

Modern engineering designs increasingly rely on lightweight yet strong structural components, particularly for energy absorption during impacts. Aluminum alloys are attractive due to their high stiffness-to-weight ratio and ease of manufacturing. Hexagonal thin-walled columns, inspired by honeycomb structures, offer excellent crashworthiness. However, optimizing these complex structures for maximum energy absorption (Specific Energy Absorption - SEA) while ensuring they don't exceed critical peak impact loads (Max PL) presents a significant challenge. Traditional finite element (FE) analysis for optimization can be computationally expensive due to the highly nonlinear nature of crash events. This paper aims to bridge this gap by developing an efficient optimization framework for hexagonal columns.

Methods

The core of this paper's methodology lies in the application of Response Surface Methodology (RSM). Instead of relying solely on computationally intensive FE simulations for every design iteration, RSM builds mathematical models (response surfaces) that approximate the complex relationships between design parameters (like wall thickness 't' and side length 'a') and the desired outcomes (SEA and Max PL).

Key techniques employed include:

  • Design of Experiments (DoE): To strategically select the points in the design space for FE simulations. The paper uses both factorial designs and D-optimal criteria for sampling.
  • Polynomial Basis Functions: These are used to construct the response surfaces. The paper investigates linear, quadratic, cubic, quartic, and quintic polynomials, ultimately finding that quartic polynomials provide the best fit for singly-celled configurations.
  • Least Squares Method: Used to determine the coefficients of the polynomial response surfaces based on the FE simulation results.
  • Successive Linear RS Model with D-optimal Criterion: For triply-celled configurations, a more advanced approach is used where the design space is iteratively refined, and linear RS models are applied within each subregion. This method is particularly effective for complex nonlinear problems.
  • Finite Element (FE) Modeling: Standard FE analysis using four-node shell elements is employed to generate the data for building the response surfaces and to validate the optimized designs. LS-DYNA is used for the simulations.

Datasets

Not discussed in the paper. The "datasets" in this context are generated through FE simulations of various hexagonal column designs under impact conditions.

Key Insights and Ideas

  • Quartic Polynomials are Key for Singly-Celled Hexagons: The study found that quartic polynomial response surfaces provided the most accurate approximations for both SEA and Max PL in singly-celled hexagonal columns, balancing accuracy and complexity.
  • Multiply-Connected (MC) Cells Offer Superior Energy Absorption: For singly-celled hexagonal columns, the multiply-connected (MC) design consistently showed higher SEA per unit mass compared to the single-cell (SC) design, indicating better material utilization.
  • Design Space Exploration is Crucial: The paper highlights how different constraints (like Max PL) significantly influence the optimal design parameters, demonstrating the importance of a comprehensive optimization approach.
  • Triply-Celled Structures Show Promise: The investigation into triply-celled configurations, particularly the side-connected type, revealed potential for even higher SEA, though with increased mass.
  • Successive RSM is Efficient for Complex Designs: For triply-celled structures, the successive linear RS modeling with D-optimal criterion proved efficient, converging to optimal solutions within a few iterations and requiring a limited number of FE analyses.

Key Results

  • Singly-Celled Optimization:
    • With a Max PL constraint of 70 kN, the optimal hexagonal SC profile had SEA of 12.0472 kJ/kg, while the MC profile achieved 12.2384 kJ/kg.
  • Without the Max PL constraint, the SC profile could achieve a higher SEA (13.6140 kJ/kg) but with a significantly higher peak load (79.05 kN).
  • Triply-Celled Optimization:
    • For the side-connected configuration (Max PL constraint of 115 kN), the optimal design yielded an SEA of 18.8331 kJ/kg.
  • For the vertex-connected configuration (Max PL constraint of 145 kN), the optimal design resulted in an SEA of 16.85 kJ/kg.
  • The side-connected triply-celled configuration generally outperformed the vertex-connected one in terms of SEA, Max PL, and structural mass.
  • General Findings: The study demonstrated that the ratio of sectional width to wall thickness (a/t) plays a significant role in energy absorption, with lower ratios generally leading to higher SEA.

Limitations and Future Work

  • Linear Model Limitations: The authors acknowledge that linear RS models, while efficient for successive approximations, may not fully capture the complex nonlinear behavior across the entire design domain.
  • Material Insensitivity: The FE models neglect strain rate effects, assuming aluminum is insensitive to strain rate. This might be a simplification for certain high-speed impact scenarios.
  • Future Work: The paper suggests that the insights gained from this study are significant for developing new classes of structural members. Further research could explore more complex honeycomb configurations, different impact scenarios, and incorporate other performance criteria. The authors also imply that the developed methodology can be extended to other structural optimization problems.

Summary updated

BibTeX
@inproceedings{hou2007design,
  author = {Shujuan Hou and Qing Li and Long Shu-yao and Xujing Yang and Wei Li},
  title = {{Design optimization of regular hexagonal thin-walled columns with crashworthiness criteria}},
  booktitle = {Finite},
  year = {2007},
  doi = {10.1016/j.finel.2006.12.008},
}
Figure