Research Assistant

Research title: Physics-Informed Additive Processing Data-Driven Finite-Element Model for Durability Prediction of AlSi10MgAlloy

Research summary:

This study aims to develop a physics-informed, data-driven finite element analysis (FEA) framework combined with machine learning to predict the fatigue life of AlSi10Mg produced by selective laser melting (SLM). Fatigue failure remains a critical limitation for additively manufactured (AM) aluminium alloys due to process-induced porosity, anisotropy, and surface roughness, which significantly reduce fatigue resistance compared with conventional materials. Existing empirical and purely numerical methods struggle to capture the combined influence of process parameters, microstructural features, and multiaxial stress states, leading to uncertainty in designing components for fatigue-critical applications. The study begins with the fabrication of dumbbell specimens under controlled SLM parameters (laser power, scanning speed, and build orientation). Multi-scale characterisation is performed to measure porosity, microstructure, residual stresses, and surface roughness, providing a comprehensive dataset linked to process conditions. High-cycle fatigue tests under fully reversed loading supply SN curves and fracture surface analysis to establish defectfatigue relationships. A physics-informed FEA (PI-FEA) framework is then applied, embedding experimentally measured defects, anisotropy, and surface roughness into simulations. Inverse FEA is used to calibrate cyclic plasticity models, while defect-sensitive fatigue laws are implemented to ensure mechanistic consistency. These physics-informed outputs, combined with process and experimental data, are used to train machine learning models such as artificial neural networks. A physics-informed loss function constrains predictions to remain consistent with equilibrium conditions and fatigue laws. Validation will be carried out on a pipeline geometry subjected to internal pressure and bending, using build conditions outside the training dataset. The expected outcome is a robust and generalisable fatigue life prediction framework with 90% accuracy within a factor-of-two life band, while also identifying dominant fatigue-controlling factors for AM AlSi10Mg. This study will bring significant impact in enabling safer AM component design, reducing experimental qualification costs, and accelerating industrial adoption of SLM-produced AlSi10Mg in fatigue-critical applications.

Responsibilities

1. Candidate will register as Masters student in the Faculty of Engineering and Built Environment, UKM
2. Candidate will perform experimental tests.
3. Candidate will perform Finite Element Analysis.
4. Candidate will perform machine learning modelling.
5. Candidate will perform fatigue life analysis
6. Candidate will prepare final report as journal publication and thesis

Qualifications

Hold a Bachelor's Degree in Mechanical Engineering

Good verbal and written communication skills

Basic skill in programming language, Finite Element software and MATLAB