Development of Generative AI to Predict Hydrogen Embrittlement Failure on Welded and 3D Printed Parts

The project is on development of Artificial Intelligence (AI) Framework on Hydrogen Embrittlement (H2E) failure prediction mainly focusing welded and 3D printed parts.

In a nutshell the work analyses Hydrogen Embrittlement failures through development of methods / algorithms on AI infused Machine Learning (ML) models by primarily using scanning electron microscopic (SEM) images of hydrogen exposed welded and 3D printed parts

This is by generating synthetic images by AI and predicting failures for varied pressures and conditions. The hydrogen exposure test will be conducted on welded areas on metals (306 L/ 314L stainless steel) and 3d printed (CFRP) parts at certain pressure levels (10/20/30 bars) and time period (24 hrs).

Artificial intelligence (AI) is one of the priorities of the present industry 4.0 era / 4th industrial revolution which helps us to advance in many domains to improve sophistication and comfortableness. Some examples of AI that are well known in our day-to-day life include use of voice over are ‘hey google’ in android phones, ‘hey siri’ in apple based mobile phones, ‘hey alexa’ in our homes. This AI advancement shows promising direction when it comes on solving the effects on the ‘El Niño’ effect on weather conditions all over the world in the form of adverse cold, rain and temperatures. This would be through achieving net zero carbon emissions by adopting Hydrogen as fuel and as agreed by all countries. But using hydrogen is challenging as it requires highly pressured atmosphere for transportation and usage. Imagine a truck carrying hydrogen in the motor way (or) a pipeline that supplies Hydrogen to lab facilities and / or our homes, may leak / burst? and before it happens, we need to predict the failure. Do we have confidence in using them like we use in our natural gas? If it fails can we weld it or use 3D printing to close the gap. What would be effect of these replaced areas? what effect it will have of Hydrogen? This is a challenging problem and requires more than human mind for prediction. Using AI, this research focusses on hydrogen usage to predict failures also known as Hydrogen embrittlement (H2E) on welded areas and 3D printed parts. It will involve usage of research facilities and expertise at Coventry University (CU) and Vellore Institute of Technology (VIT). Because of sustained nationwide emphasis on both India and UK this collaborative research effort will support AI infused research to support achieving the target of net zero emissions.