ONGOING COLLABORATIONS

Physics-Informed Neural Networks for Damage Mechanics

Continuum Damage Mechanics (CDM) numerical algorithms are widely employed to predict material degradation in engineering applications. However, these methods typically require iterative computations at each time step, posing significant challenges, particularly for complex geometries. To overcome these limitations, Physics-Informed Neural Networks (PINNs) offer a promising alternative by embedding governing physical equations directly into the network’s loss function, improving computational efficiency.

Project Leader: Mohammad Mashayekhi (Isfahan University of Technology)

Designing architectures using Finite Element Analysis and Machine Learning

For designing architecture materials, there are theoretically no limit to potential designs since unlimited interface and building block shapes can be used. This is absolutely interesting, since it provides the opportunity to design different architectures for different purposes. It is however required to understand the effect of these arbitrary shaped interfaces and building blocks on the performance of a hierarchically architectured material. In this project, we are using finite element analysis and artificial neural networks to develop computationally efficient numerical algorithms for analyzing the effect of arbitrary interfaces and building block shapes on the performance of architectured materials.

Project Leader: Mohammad Mirkhalaf (Queensland University of Technology)

PREVIOUS COLLABORATIONS

Optimal design and 3D printing of a bicycle helmet

Helmets are crucial safety equipment which prevent a bike rider from getting seriously injured in the event of an accident. Conventional manufacturing processes have been successfully used to fabricate helmets. But recently, 3D printing has gained a considerable interest for that purpose. This is due to (i) Its capability for printing very complex geometries, (ii) The possibility of customized printing in a timely and cost-wise manner. In this project, we are designing a bike helmet, manufacturing the helmet via 3D printing, and testing its response under impact loads.

Project Leader: Mohammad Heidari-Rarani (University of Isfahan)

Deep-Learning-Enhanced multi-physics modelling of atherosclerosis

Atherosclerosis is a medical condition which involves hardening and/or thickening of arteries' walls. To predict and analyze development of atherosclerosis, multi-physics models have been. However, one of the main challenges of these models is the associated computational cost. In this project, we are using artificial neural networks (ANNs), to enhance the computational efficiency of these models.

Collaborators: Meisam Soleimani (Leibniz Universität Hannover), Behdad Dashtbozorg (Netherlands Cancer Institute), and Mohammad Mirkhalaf (Queensland University of Technology)

Pulp fibres-starch composites: Fabrication, testing and micromechanical modelling

In this research, bio-composites were fabricated by combining thermoplastic starch as the matrix and pulp fibers as reinforcements through a process involving reactive extrusion, followed by compression molding. The thermal and tensile properties of these bio-composites were subsequently evaluated. To establish the relationship between processing, structure, and properties, micromechanical simulations were conducted using both full-field Representative Elementary Volume (RVE) finite element simulations and a two-step orientation averaging micromechanical model.

Collaborator: Giada Lo Re (Chalmers)

Polymer concretes reinforced with recycled PET bottles: fabrication, experiments and modelling

We improved the tensile strength and ductility of epoxy-based polymer concrete (PC) by incorporating recycled polyethylene terephthalate (PET) fillers. We examined two PET filler sizes (fine and coarse) using Brazilian disk tests to measure indirect tensile strength. Micromechanical damage analyses via finite element simulations considered void content and PET fillers. We established a correlation between micromechanical and macro mechanical tensile strength for Brazilian disk tests.