AM process development for NiTi Alloys with AI & ML

I was glad to participate in this month's ASM advanced materials manufacturing committee meeting for a talk on Nickel Titanium alloys and its additive manufacturing process development via artificial intelligence and machine learning. 

Dermot Brabazon Prof. at the Dublin City University and deputy director at I-Form ,the publicly funded institution leading Ireland's manufacturing innovation, presented the findings of his research group. NiTi alloys ,most famously Nitinol,are known for exhibiting super elasticity and shape memory effect ,which causes material to return to its original shape after being heated. 

This has found several applications in the fields of medicine and actuation. The general processes of manufacturing NiTi alloys are now well developed however additive manufacturing introduces new challenges that are not present in traditional melt cast and heat treated products, that is process uniformity. 

So while AM is successful in producing complex shapes with NiTi alloy powders, the exact phase transformations required for it to exhibit its super elasticity and shape memory are not always reproducible. Additionally because AM components could have residual porosity the absolute mechanical strength is somewhat lower than that of casted materials. 

Predicting the properties of AM produced components could accelerate research. Prof Dermot and his team fed data collected from various sensors (infrared and acoustics) to machine learning programs to ascertain the properties of the final product based on laser power, powder composition,spot size and scan speed. The program came up with density values for the component which was used to infer the porosity of the part and finally determine its properties. 

The building of the model required generation of comprehensive datasets for the program but once that was done the model could accurately predict the properties after only a few layers of deposition. 

His work demonstrated that using ML could be a viable strategy to simulate materials properties programmatically rather than experimentation which can be much slower. 

I thank ASM for continuing to organise these events that gives materials enthusiasts like myself a chance to learn from experts in the field. I-Form is doing incredible work in pushing materials innovation in Ireland and I'm excited to learn what new developments emerge from the group in the future.