A path and process based simulation approach of 3D concrete printing

Additive manufacturing with concrete has gained significant interest in recent years as a promising technology that could help address some of the major challenges faced by the construction industry. However, 3D printing with concrete poses challenges in terms of buildability, as the deposited strand must support its own weight, as well as the weight of the layers deposited on top of it. Additionally, during the build-up of concrete structures via 3D printing, elastic buckling effects and plastic deformation can lead to structural collapse. Simulation plays a crucial role in predicting structural stability and optimizing process parameters to prevent failure. This dissertation focuses on the implementation of a novel path- and process-based finite element approach to simulate the structural build-up of concrete structures during 3D printing. This approach addresses the limitations in simulation methods using commercial software and enables a close coupling between process parameters and simulation results. The dissertation also details the implementation of time-dependent material behavior of concrete using the Drucker-Prager model. After the implementation of the finite element framework, an analysis and evaluation of the implemented algorithms are presented. A convergence analysis, sensitivity analysis, and the effects of other simulation parameters are examined. Two main numerical studies demonstrating the optimization of process parameters are carried out. The first study focuses on the design process and assesses the achievable wall height based on the strand geometry. The second study investigates the stability of the printed structure and optimizes the interlayer waiting time distribution. A genetic algorithm is implemented to select the optimal solution from a large number of possible permutations. These studies highlight the advantage gained by having the solver and optimization algorithms in the same environment. In the final part of the research, the possibility of using a physics-guided neural network to supplement the solver was demonstrated. An iterative loop within the material model was replaced by a physics-guided neural network to determine the accumulated plastic strain.