Minimal time required for material characterization.

Minimize time, equipment and material costs while being able to run multiple tests in minutes.

Extract mechanical
properties from intricate
components

Ability to map the local mechanical properties in components & structures with complex geometries at small scales

Tests final components to ensure mechanical properties and process reliability

A distinct contrast from the fully destructive nature of traditional mechanical testing

Cloud based neural network that seamlessly integrates with existing instrumented indenters

Removes the need to buy new expensive hardware to perform accurate mechanical tests

In instrumented indentation testing there is no closed form analytical solution to the inverse problem (deriving the stress-strain curve from the load displacement curve). Today, the industry standard is to use various brute force FEM simulation approaches combined with some form of fitting functions to correlate the FEM results to the stress-strain behavior. However, this type of approach is bound to be hypersensitive to systematic test errors and produces highly inaccurate results especially for the inverse problem.

Due to this error sensitivity of using fitting functions we have introduced a new approach that is a paradigm shift from previous ways of addressing this problem.

We have utilized recent advances in deep learning and trained neural networks to accurately extract a material’s stress-strain behavior from an indentation test.

Our deep learning approach, unlike previous methods that attempt to include all potential behavior inside fitting functions, trains a neural network with a vast number of physics-informed indentation data. Our physics-informed neural network is trained to recognize all the physical behaviors the system (an indentation test) may encounter.

At Anailytica, we have developed a brain that continues to learn and get smarter over time as more and more indentation data is introduced. This is a stark difference from previous static methodologies.