Predicting creep failure life in adhesive-bonded single-lap joints using machine learning

Area of Science:

• Materials Science
• Mechanical Engineering
• Computational Science

Background:

• Predicting creep failure life in single-lap adhesive joints (SLAJs) is challenging, demanding significant resources and time.
• Ensuring structural integrity and preventing premature failures in adhesive joints necessitates accurate creep failure life predictions.

Purpose of the Study:

• To utilize machine learning (ML) for identifying critical features influencing SLAJ durability under creep conditions.
• To develop and validate an ML model for predicting the creep failure life of SLAJs.

Main Methods:

• Feature identification through correlation analysis and sequential feature selection.
• Application of multiple ML algorithms to model complex relationships between features and creep failure life.
• Development of a predictive model using Python, validated through experimental testing.

Main Results:

• Key features influencing creep failure life include SLAJ creep strain, adhesive tensile strength (UTS), SLAJ creep stress, adhesive surface area (A), and Young's modulus (E).
• The random forest (RF) model demonstrated the highest accuracy in predicting creep failure life.
• The ML model's predictions were experimentally verified, confirming its reliability.

Conclusions:

• Machine learning effectively identifies critical factors for predicting adhesive joint creep failure.
• The developed RF model provides an accurate and experimentally validated method for forecasting creep failure life in SLAJs.
• The study provides valuable insights and tools for enhancing the durability and reliability of adhesive joints in engineering applications.