Abstract
This study introduces a Bi-Directional Interpolation Model (BDIM) designed to address the challenge of intermediate density elements—often referred to as "gray areas"—in structural topology optimization. By integrating BDIM into a Gradient Projection Method framework, the researchers established a mathematical model for optimizing food vehicle frames. The study demonstrates that applying BDIM specifically to sensitivity analysis (BDIM01) ensures global convergence and produces clear structural boundaries, offering a robust solution for designing efficient, manufacturable transportation frames.
Introduction
In structural engineering, topology optimization seeks the most efficient material distribution within a design space to maximize stiffness while minimizing weight. A common hurdle in the "Variable Density Method" is the emergence of elements with intermediate densities (e.g., 0.5), which are physically impossible to manufacture. Traditional penalization methods like SIMP can sometimes lead to numerical instabilities or "checkerboard" patterns. This paper proposes the Bi-directional Interpolation Model (BDIM) as a more stable alternative to force elements toward fully solid (1) or void (0) states.
Methodology
The core of this research is the Bi-directional Interpolation Model (BDIM), which utilizes a threshold-based cosine function. Unlike unidirectional models, BDIM penalizes elements with densities between a minimum value and 0.5, while encouraging elements between 0.5 and 1 to become solid.
Two primary strategies were evaluated:
- BDIM01: The interpolation model is applied only to the sensitivity analysis.
- BDIM02: The model is applied to both the stiffness matrix and the sensitivity analysis.
The optimization process is driven by the Gradient Projection Method, iteratively refining the structure to minimize flexibility under a specific volume constraint.
Experimental Setup
To validate the model, the researchers used a standard 2D benchmark: a 20×40 rectangular design domain. The material properties were defined with an elastic modulus of 1 and a thickness of 1. The left boundary was fixed, and a concentrated load was applied to the midpoint of the right boundary. The optimization goal was to achieve the stiffest possible structure using only 30% of the original material volume.
Results and Discussion
The experimental results highlighted several key advantages of the BDIM approach:
- Superiority of BDIM01: Applying the penalty only to sensitivity (BDIM01) yielded the most practical results. It produced distinct "black and white" topologies with clear boundaries, eliminating the need for additional complex filtering techniques that often blur the final design.
- Convergence and Efficiency: The introduction of a bi-directional factor ($q$) significantly impacted the speed of the optimization. While the final topology remained consistent, increasing $q$ from 1 to 4 reduced the number of iterations required for convergence from 43 to just 18.
- Stability: Across various penalty factors, the objective function values consistently converged to a stable range (approximately 7.8 to 7.9), proving the model's reliability.
Conclusion
The Bi-directional Interpolation Model, particularly when applied to sensitivity analysis, provides a powerful new method for resolving the problem of intermediate density elements in topology optimization. By producing clear, manufacturable designs with high computational efficiency, this framework serves as a valuable tool for the structural design of food transportation vehicle frames. Future work will focus on extending this 2D proof-of-concept to complex 3D vehicle structures and non-linear material behaviors.
Summary updated