Future laboratory for advanced mechanical testing

Kurzfassung

Climate change and geopolitical conflicts require innovative material solutions to reduce CO2 emissions and manage the transition to a circular economy. Unfortunately, the development of lightweight alloys using classical approaches can take > 10 years . One time-consuming factor that affects the deployment of new materials is their mechanical characterisation. In this context, we have developed an autonomous test rig that enables rapid material testing and high-precision data generation: Robots handle the samples, mount them in the testing machine and carry high-resolution optical sensors for in situ digital image correlation (DIC) analysis. During fatigue crack growth experiments, deep learning-based algorithms detect the crack tip and evaluate the crack tip loadings. The evolution of the plastic zone at the surface of the samples is monitored in situ to provide information about the crack growth mechanisms acting in the bulk of the material, e.g. to detect the transition from flat to slant crack growth. The combination of several data acquisition sources with different evaluation algorithms lead to redundant information that helps enhancing confidence in the experimental results. Computer vision tools are implemented to assist experts and non-experts in the interpretation of the data obtained during testing. In addition, all algorithms developed for the new test rig are available as open-source Python codes to foster FAIR data principles in the materials mechanics community. The full testing and evaluation pipeline is demonstrated by means of fatigue crack growth experiments of an AA2024-T351 alloy showing that the new procedure is faster and allows further insights about the fundamental mechanisms of crack growth than using classical methods . Autonomous systems that combine materials science know-how, robotics, artificial intelligence and computer vision tools will play an essential role to accelerate materials development and testing.