{"id":145757,"date":"2026-07-01T08:59:20","date_gmt":"2026-07-01T00:59:20","guid":{"rendered":"https:\/\/www.curtin.edu.au\/research\/?post_type=hdr-r-projects&p=145757"},"modified":"2026-07-01T08:59:20","modified_gmt":"2026-07-01T00:59:20","slug":"physics-informed-machine-learning-for-predicting-damping-behavior-in-3d-printed-metamaterial-vibration-isolators","status":"publish","type":"hdr-r-projects","link":"https:\/\/www.curtin.edu.au\/research\/hdr-r-projects\/physics-informed-machine-learning-for-predicting-damping-behavior-in-3d-printed-metamaterial-vibration-isolators\/","title":{"rendered":"Physics-Informed Machine Learning for Predicting Damping Behavior in 3D-Printed Metamaterial Vibration Isolators"},"content":{"rendered":"\n

Vibration isolation is critical across Western Australia’s strategic industries\u2014mining, renewable energy, and advanced manufacturing. Current vibration isolators rely on conventional elastomeric materials or springs with limited damping control. 3D printing offers unprecedented freedom to design metamaterial structures with tailored vibration properties, but fundamental knowledge gaps exist: we cannot predict how 3D printing parameters affect damping behavior in manufactured metamaterials.<\/p>\n\n\n\n

Damping ratios in 3D-printed materials are strongly influenced by internal structure, yet these effects are not predictable from standard FEA or material science models. This project addresses this knowledge gap by developing a physics-informed machine learning framework that predicts damping behavior directly from printing process parameters (nozzle temperature, print speed, infill pattern, layer height, etc.).<\/p>\n\n\n\n

The research bridges materials science, experimental mechanics, and artificial intelligence to enable rational design of 3D-printed metamaterial isolators optimized for Western Australia’s extreme operating environments (Pilbara mining: -5\u00b0C to 50\u00b0C, vibration frequencies 50-500 Hz, harsh dust conditions).<\/p>\n\n\n\n

Novel contribution: First predictive model linking 3D printing parameters \u2192 material damping in metamaterials, enabling engineers to optimize print settings for target vibration performance without expensive trial-and-error prototyping.<\/p>\n\n\n\n

Aim  <\/p>\n\n\n\n

To establish the first quantitative, physics-informed machine learning framework that predicts damping behavior in 3D-printed metamaterials, enabling rational design of vibration isolators for Western Australia’s mining, energy, and advanced manufacturing industries.<\/p>\n\n\n\n

Objectives <\/p>\n\n\n\n

    \n
  1. Generate experimental dataset: 3D print and test 150-200 metamaterial samples with systematically varied printing parameters, generating 10,000+ data points on damping behavior.<\/li>\n\n\n\n
  2. Develop predictive ML models: Train neural networks, random forests, and Gaussian process models achieving >90% accuracy in predicting damping ratio from printing parameters.<\/li>\n\n\n\n
  3. Quantify parameter sensitivity: Perform global sensitivity analysis (Sobol indices) to identify which printing parameters have greatest influence on damping and their interactions.<\/li>\n\n\n\n
  4. Rank feature importance: Use Shapley value analysis to determine which printing parameters are critical for controlling damping behavior.<\/li>\n\n\n\n
  5. Enable design optimization: Leverage trained ML models to recommend optimal printing parameters for achieving target damping specifications at specific vibration frequencies.<\/li>\n\n\n\n
  6. Validate framework: Test predictions on 5-10 prototype designs, comparing ML predictions vs. experimental measurements with <5% error tolerance.<\/li>\n<\/ol>\n\n\n\n
    \n

    Significance <\/p>\n\n\n\n

    Fundamental research significance:<\/p>\n\n\n\n

    \n