3D image analysis and microstructure models for simulation of materials properties

Abstract

The rising availability of computational power allows for a steady improvement of simulation techniques. As the awareness for the limits of our natural resources increases, the use of numerical simulation for understanding existing materials and their mechanical behavior is receiving increasing attention. Recently, Diebels and Rjasanow 2019 incorporate knowledge about a material (glass fiber reinforced polymers) w.r.t. the physical scale into a multi-scale approach, leading to a chain of conclusions. They study the microstructure of the composite material on the microscale, develop modelling and simulation techniques for mechanical properties on the micro- and macro scale. Microstructure models help us to gain knowledge at one scale and deducing to the following. In this context, this work contributes to the exploration of mechanical properties of microscopically heterogeneous materials.

I analyze 2D or 3D data of materials’ microstructure by means of stochastic geometry and fit geometric microstructure models. First, I model polycrystalline metals using a Laguerre tessellation. Second, glass fiber reinforced polymers are modeled by a union of cylinders. This modelling step enables the connection between a geometrical microstructure model and a simulation of mechanical properties of microscopically heterogeneous materials.

In ultrasonic non-destructive testing, the scattering theory (Hirsekorn 2014) originally predicts the scattering in a polycrystal w.r.t. grain mean diameters and the frequency of an ultrasonic wave. I implement a new simulation technique for the computation of a scattered ultrasonic time-domain signal in a fitted Laguerre tessellation. Hence, this approach allows to simulate the noise which appears in ultrasound signals caused by microstructers. An investigation of scattering due to microstructural variation in a polycrystal is now possible, which should be employed in context of non-destructive material testing in the future.

Destructive sample extraction from composite components for estimation of fiber orientation distribution belongs to the state-of-the-art. We demonstrate success in two main scenarios: First, virtual experiments for tensile tests now can replace destructive tensile testing. Second, we demonstrate that truly non-destructive scanning of regions of interests is now possible.