Some industrial applications require properties of materials to be determined to evaluate components’ safety in the event of loading and impact. Understanding the behaviour of materials subjected to extreme dynamic loading will aid in enhancing their design. This work is based on developing methods to appraise high loading rate measurements. Different approaches to quantify material properties such as finite element method (FEM), instrumented Charpy testing, and impact testing using servo-hydraulic testing machines are included. Testing is performed at various loading rates, extending existing quasi-static fracture toughness determination to higher loading rates, and accounting for strain-rate dependent properties. The high loading rate servo-hydraulic test machine located at TWI, Cambridge has the capacity to test up to a displacement rate of 20 m/s. The force, displacement, and time parameters are captured by Digital Image Correlation (DIC), which improves the accuracy of the results obtained from experiments. Moreover, the underlying plasticity theory to capture the influence of the strain rate is presented, along with damage constants for FEM calculations adopting the Johnson-Cook model. In addition to the Johnson-Cook approach, analytical solutions using dislocation evolution theory were applied which features the effects of phonon drag and dynamic recovery coefficient in body-centered cubic materials of which X65 grade steel was applied. Also, a deep learning framework was built to predict the tensile curves when given specific test conditions and sample specifications. It was found that high strain rate tests lead to local change at the crack tip which increases plasticity and reduces fracture toughness with single-edged notched three-point bend specimen. The yield strength of the material increased with loading rates during tensile testing leading to a ductile to brittle transition of metals. These strategies were used to establish a revised approach for high strain rate testing and predicting stress-strain curves with a machine learning algorithm.
Read full abstract