Optimizing 3D-Printed Concrete Mixtures for Extraterrestrial Habitats: A Machine Learning Framework

Abstract

The utilization of 3D printing technology for the construction of habitats and infrastructure on celestial bodies such as the Moon and Mars presents an increasingly fascinating prospect in space construction research. The success of 3D printing constructions heavily depends on the rheological and mechanical properties of 3D-printed concrete influenced by several factors such as nozzle speed, interlayer interval time, and environmental conditions. However, existing studies have not addressed the challenge of real-time optimization of concrete mixtures under constantly changing conditions, including temperature, humidity, pressure, and gravity on other planets. Potentially, machine learning (ML) offers advantages in terms of data-driven optimization, flexibility, cost, and time savings and handling complex relationships on optimization tasks. This study aims to identify and propose a framework for applying ML in real-time optimization of concrete mixture and printing parameters. The framework involves identifying influential factors to optimizing a 3D-printed concrete mix and printing parameters, thereby developing an optimization framework specifically tailored to the extraterrestrial environment. The study result explored the feasibility of using machine learning techniques for real-time optimizing concrete mixtures in 3D printing construction on celestial bodies. The expected contribution of the study lies in introducing data-driven optimization techniques that provide adaptability to changing environmental conditions, improving the efficiency and reliability of construction in space. © ASCE.