Design of a Computational Model for 3D Concrete Printed Geometry Using Machine Learning and Genetic Optimization

3D Concrete Printing (3DCP) is an emerging technology with well-established benefits and the potential to dramatically change the construction industry. While research in material and process optimization is gaining traction, the architectural application of 3DCP remains relatively underdeveloped. Among many challenges, a lack of suitable computational modelling techniques is often identified as a major obstacle, resulting in simplistic design solutions that do not take full advantage of 3DCP technology. This study proposes a fabrication-aware design model using machine learning (ML), specifically genetic optimization, to address the research gap. 3DCP is used to produce sacrificial formwork for freeform reinforced concrete shell structures. The model conceptualizes a module-based approach to establish interlinked feedback loops across the various stages of a project, enabling fabrication and assembly considerations in the early design phase. Structural behaviour, printability, and segmentation constraints are translated into evaluative criteria within a unified computational workflow implemented in Rhino/Grasshopper, using the Galapagos genetic optimization solver. The framework enables iterative exploration of design options while accounting for both geometric and fabrication-related constraints. Three shell typologies are used to demonstrate the method, including a cantilever, a bridge, and a wall element, supported by a full-scale 3D printed segment for initial validation. This approach enables designers to develop geometries that are specifically tailored to the constraints and opportunities of 3DCP, opening new possibilities for meaningful interaction with the design-to-fabrication pipeline of complex 3DCP geometries.