Experimental analysis and machine learning based predictions of fatigue fracture in dehydrated polyacrylamide hydrogel

• المؤلف الرئيسي: Tia, Yi Ken
• مؤلفون آخرون: Li Hua
• التنسيق: Final Year Project
• اللغة: English
• منشور في: Nanyang Technological University 2023
• الموضوعات: Engineering::Mechanical engineering; Engineering::Computer science and engineering; Engineering::Materials
• الوصول للمادة أونلاين: https://hdl.handle.net/10356/172821
• المؤسسة: Nanyang Technological University

The polyacrylamide (PAAm) hydrogel, a transparent and nearly elastic material, has applications in diverse fields such as biomedical, agricultural, and water treatment. While extensive research has explored its many mechanical properties, a critical gap exists in understanding the impact of dehydration on PAAm hydrogel fatigue fracture. This is crucial due to PAAm hydrogel’s susceptibility to dehydration given its rich-water structure. Therefore, this project aims to investigate the fatigue fracture of PAAm hydrogel under dehydration, focusing specifically on fatigue life. Machine learning (ML) models will be employed to predict the fatigue life, addressing the limitations of existing mathematical models. The study begins with a pure shear test of PAAm hydrogel under dehydration, gathering 90 raw data samples with variations in loading rates (200/250/300 mm/min) and maximum displacement (20/25/30 mm). A supplementary experiment was conducted to compare fatigue fracture behaviour under both dehydration and constant moisture conditions using 10 samples. The 90 raw data samples were processed using MATLAB to generate graphs for fatigue life determination and hysteresis loss analysis. Feature extraction was also performed using MATLAB, followed by data cleaning and statistical analysis using Jupyter Notebook. Subsequently, augmented data was generated from the original data and integrated with it to create a new dataset through the data augmentation process. Both the original and new datasets were separately fitted with 5 ML models to predict fatigue life under dehydration. The study revealed that dehydration significantly alters fatigue fracture behaviour, leading to an abrupt and extensive fracture at fatigue life and a transition to an opaque state. The fatigue life generally tends to increase with loading frequency and decrease with maximum stretch. Hysteresis loss was also found to be negligible during fatigue. Moreover, the KNN model was identified as the most suitable ML model, achieving a test score of 0.68 for the original dataset and 0.90 for the new dataset. These findings contribute to a better understanding and highlight the adverse effects of dehydration on the fatigue fracture of PAAm hydrogel. They also demonstrate the potential of ML models in predicting the mechanical properties of hydrogels for future studies.