Papers »

Creation of statistically equivalent periodic unit cell for protein-bound soils

Extraterrestrial construction presents many interesting and new challenges. Unlike Earth, there are very limited naturally occurring resources readily available in extraterrestrial environments, such as the moon or Mars. To take advantage of available resources, this work focuses on a novel class of materials: protein-bound soil composite materials. The composite is produced by desiccating a mixture of soil, water, and a protein binder to create a strong, versatile material. Experimental tests to characterize basic mechanical properties have shown significant variability among tested samples. It is hypothesized that this variability arises from the heterogeneity of the material microstructure observed through microscopy.

To test the hypothesis of the influence of microstructure heterogeneity, we propose the creation of a large population of stochastic Finite Element Models to perform virtual experiments and obtain a probabilistic distribution of the mechanical properties of these composites. The first step to create this population is a framework based on Statistically Equivalent Periodic Unit Cells (SEPUC) to stochastically model images of the protein-bound composite. The model inputs are the soil granulometry and the volume fractions of the phases. Generated ellipsoidal particles are placed in the cell using a Level-Set based Random Sequential Addition algorithm and bridges between particles are created using the same Level Set Function used in placement. 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 was validated by modeling a protein-bound soil composite produced from graded sand and bovine serum albumin (BSA) protein and comparing virtual composite images to experimental specimen images obtained by micro-CT scanning. Results show that the framework is able to rapidly create unique images that capture the particle arrangement and protein bridging of protein-bound sands, marking the first attempt to successfully do so.

Author(s):

Isamar Rosa    
Stanford University
United States

Michael Lepech    
Stanford University
United States

Henning Roedel    
Stanford University
United States

David Loftus    
NASA Ames Research Center
United States