In 2010, NASA was directed to develop technologies to reduce the cost and risk of space exploration and send humans beyond the International Space Station. A central challenge to long-duration space missions is a lack of available construction materials in situ. This work focuses on a novel class of composites that can be produced extraterrestrially in situ by desiccating a mixture of soil, water, and protein binder to create a strong, versatile material. To date, experimental tests of mechanical properties have shown significant variability among samples.
This paper focuses on the creation of Statistically Equivalent Periodic Unit Cells (SEPUC) to stochastically model protein-bound composites for the purpose of creating FE models that provide insights into experimental results. Model inputs include the soil granulometry and volume fractions of the phases. Ellipsoidal particles are placed, and protein coatings and bridges are created, using a Level Set based Random Sequential Addition algorithm. Each image is assigned a statistical descriptor and a simple genetic algorithm is used to optimize for a statistical descriptor close to that of experimental specimens.
The framework is validated by comparing experimental images of protein-bound soils obtained by micro-CT scanning with those obtained through the SEPUC framework.