Cutting edge materials science without the 20 year lead time to commercialisation. We accelerate the time between first discovery and utilisation in the field.
Our advanced form of armour provides superior protection against armour piercing projectiles and fragmentation through ultra-fast thermodynamic processes. This patent-pending material technology, we call SynTech.
We provide an alternative to ultra-hard ceramics and steels, reducing cost, weight and providing improved performance.
SynTech is a step change in armour not seen since Kevlar revolutionised the use of textiles for armour. Our composite structures will drastically improve the ballistic and armour piercing performance over currently available technology for any application, be it vehicle, personnel or building.
Contact us for a white paper and more information on our thermodynamic armour technology.
AI enhanced FEM modelling
Current methods of material numerical simulation, such as Finite Element Modelling (FEM), use a linear process:
Specific laboratory techniques are used to characterise materials. E.g. Uniaxial tensile tests to get the Young’s Modulus and the Yield Strength.
Analytical models are fitted to this data. E.g. A linear elastic model with a Mohr-Coulomb yield surface is an example of an analytical model for most metals.
The FEM simulation is constructed with this material input data. Mesh refinement and other methods to optimise the numerical simulation are run.
A large scale validation test is designed. This may be a complex structure or a non uniform loading geometry.
Key validation measurements are compared, and the model is either deemed acceptable or not.
If the model is rejected at the validation phase, there is no way to easily update or test the underlying assumptions introduced at each stage, or to know which underlying assumption is incorrect. Is it the analytical model form that has been chosen? Is it a systematic error in the laboratory characterisation?
Our new method uses the advances in optimisation and machine learning to remove the linear process and replace it with a continuously updating system. As more data is introduced, the model improves. There are optimisation algorithms introduced at each stage to quantify uncertainty and direct future experiments. Underpinning all the optimisation, by replacing the underlying analytical material model with a physics based neural network, complex material behaviour including phase changes and time dependence can be introduced without having to solve complex analytical equations.