Multi-objective topology optimisation for acoustic porous materials using gradient-based, gradient-free, and hybrid...

Area of Science:

• Acoustics
• Materials Science
• Computational Engineering

Background:

• Designing passive sound-attenuation structures presents a challenge in optimally distributing acoustic porous materials.
• The objective is to maximize sound absorption while minimizing material usage, a multi-objective problem.

Purpose of the Study:

• To compare gradient, non-gradient, and hybrid topology optimization strategies for acoustic material distribution.
• To identify efficient optimization methods for maximizing sound absorption and minimizing material usage.

Main Methods:

• Gradient approaches: Solid Isotropic Material with Penalisation (SIMP) and a gradient-based constructive heuristic.
• Gradient-free approaches: Hill climbing with weighted-sum scalarisation and Non-Dominated Sorting Genetic Algorithm-II (NSGA-II).
• Development and testing of two hybrid approaches combining gradient initiation with non-gradient local improvements, including Pareto-slope-based weighted-sum hill climbing.

Main Results:

• Gradient methods offer fast convergence and high-quality solutions.
• Gradient-free strategies can identify improvements in specific regions of the Pareto front.
• Hybrid methods consistently outperform parent methods within a given computational budget.

Conclusions:

• Hybrid topology optimization strategies are effective for the multi-objective problem of acoustic material distribution.
• Combining gradient and non-gradient methods provides a robust approach for designing efficient sound-attenuation structures.
• The proposed Pareto-slope-based weighted-sum hill climbing enhances local improvements in hybrid methods.