Additive manufacturing techniques, such as 3D printing, are nowadays well established in different industrial sectors, including automotive, aerospace, and healthcare, where they enhance production efficiency. However, their adoption in construction remains limited due to many technical challenges and uncertainties. Advances in numerical modelling are expected to partially bridge this gap, offering designers greater control and insight into the processes and materials involved in these new construction approaches. This thesis presents an advanced numerical framework for simulating 3D printing of complex materials. Using the Particle Finite Element Method (PFEM), which combines a Lagrangian FEM formulation with a robust re-meshing algorithm, it efficiently handles large displacements, evolving domains, and automatic free-surface tracking. Within the PFEM context, two FEM solvers were developed, utilizing implicit and explicit time integration schemes respectively. Initially validated for Newtonian fluids, the solvers were extended to handle more complex behaviours, including viscoelastic, viscoplastic, and elasto-viscoplastic materials, potentially undergoing time-dependent evolution or fluid-to-solid phase transitions. The non-Newtonian solvers were benchmarked on well-established case studies involving free-surface flows, such as dam breaks, drop impacts, and jet buckling. At the same time, a set of specialized numerical techniques was developed to accurately simulate the 3D printing process in the PFEM framework. These tools include the use of the Arbitrary Lagrangian-Eulerian (ALE) approach to model nozzle movements, advanced algorithms to accurately capture base and interlayer contact, an adaptive mesh de-refinement technique to decrease the computational cost, and an efficient nodal mapping scheme to transfer historical variables across re-meshing events. The overall framework was validated by reproducing the extrusion and layer-deposition phases of 3D Concrete Printing (3DCP) and comparing results with experimental data. The explicit solver proved efficient in simulating a variety of different printing scenarios, highlighting the influence of material properties (density, yield stress, viscosity) and process parameters (nozzle diameter, nozzle height, printing velocity, extrusion velocity) on the morphology of the printed layers. The outcomes of these simulations enabled the generation of a dimensionless map summarizing all possible cross-sectional shapes and printing regimes in free-flow deposition of yield-stress fluids. Vertical and inclined multi-layer walls were also simulated and compared with experimental data from literature or provided in the framework of the collaboration with TU/e Eindhoven. Numerical results aligned closely with the measured data, also providing insights into the role of material compressibility in 3DCP. The implicit solver was preferred in all those cases involving longer printing times and where the objective was to predict in advance the geometrical conformity and buildability of complex shape objects. In the first example, a rectangular wall characterized by a W-type infill was virtually printed, investigating the role of thixotropy and assessing interlayer bonding quality at intersections. The second example involved the design and virtual printing of three parametric hollow objects, to evaluate their stability and geometrical conformity under varying material properties. This analysis also highlighted how the numerical tool can be used to predict when and where additional support could be placed to enhance the final quality of printed structural components. Overall, the proposed framework proved to be accurate, robust, and potentially applicable to different classes of materials, such as clays and polymers. Future improvements will aim to decrease the computational cost through the integration of AI and machine learning techniques, transforming the framework into an effective predictive tool and enabling real-time optimization and the creation of reliable digital twins for the 3D printing process.
Le tecniche di manifattura additiva, come la stampa 3D, sono ormai consolidate in diversi settori industriali. Tuttavia, il loro uso in edilizia è ancora limitato a causa delle numerose sfide tecniche ed incertezze. I progressi nella modellazione numerica potrebbero colmare parzialmente questo divario, offrendo maggiore controllo sui processi e sui materiali impiegati. Questa tesi propone un framework numerico avanzato per la simulazione della stampa 3D con materiali complessi, sfruttando il Metodo agli Elementi Finiti Particellari (PFEM). Grazie alla sua formulazione lagrangiana e ad un algoritmo di re-meshing, il PFEM può efficacemente riprodurre scenari caratterizzati da domini in evoluzione, soggetti a grandi spostamenti e con superfici libere. Nel contesto PFEM, sono stati sviluppati due robusti solutori FEM, basati sull’integrazione temporale esplicita ed implicita rispettivamente. Inizialmente validati per fluidi newtoniani, i solutori sono stati poi estesi per trattare i casi di materiali viscoelastici, viscoplastici ed elasto-viscoplastici, anche soggetti a transizioni di fase. Le validazioni includono problemi di flussi a superficie libera, come dam-break, impatti di gocce e instabilità di getti. Successivamente, sono state sviluppate o applicate delle tecniche numeriche specifiche per simulare la stampa 3D. Tra queste, l’approccio Arbitrary Lagrangian-Eulerian (ALE) per imporre il movimento dell’ugello, degli algoritmi di contatto per riprodurre correttamente l’aderenza tra gli strati, una tecnica adattiva di de-raffinamento della mesh per ridurre il costo computazionale ed un metodo efficiente per il trasferimento delle variabili durante il re-meshing. Inizialmente, il solutore esplicito è stato utilizzato per validare con dati sperimentali le fasi di estrusione e deposizione nella stampa 3D del calcestruzzo (3DCP). Il modello ha inoltre permesso di simulare diversi scenari di stampa e analizzare l’influenza delle proprietà del materiale e dei parametri di processo sulla morfologia degli strati generati, portando alla creazione di una mappa generale che correla le diverse tipologie di regimi di stampa alle geometrie delle sezioni trasversali dei filamenti stampati. Sono state inoltre simulate pareti multistrato verticali e inclinate, confrontando i risultati numerici con dati sperimentali ottenuti dalla letteratura e dalla collaborazione con TU/e Eindhoven. Il solutore implicito è stato poi impiegato per scenari di stampa più lunghi e complessi, come la simulazione di pareti con riempimento a W e oggetti cavi parametrici, valutando stabilità, conformità geometrica e qualità dell’adesione tra strati. Il framework proposto si è dimostrato accurato e robusto ed è potenzialmente applicabile anche per la stampa con materiali differenti, come le argille e i polimeri. Futuri sviluppi mireranno a ridurre il costo computazionale mediante AI e machine learning, trasformandolo in uno strumento predittivo per l’ottimizzazione in tempo reale e la creazione di digital twin affidabili per la stampa 3D.
A PFEM numerical framework for simulating 3D printing of advanced materials
Rizzieri, Giacomo
2024/2025
Abstract
Additive manufacturing techniques, such as 3D printing, are nowadays well established in different industrial sectors, including automotive, aerospace, and healthcare, where they enhance production efficiency. However, their adoption in construction remains limited due to many technical challenges and uncertainties. Advances in numerical modelling are expected to partially bridge this gap, offering designers greater control and insight into the processes and materials involved in these new construction approaches. This thesis presents an advanced numerical framework for simulating 3D printing of complex materials. Using the Particle Finite Element Method (PFEM), which combines a Lagrangian FEM formulation with a robust re-meshing algorithm, it efficiently handles large displacements, evolving domains, and automatic free-surface tracking. Within the PFEM context, two FEM solvers were developed, utilizing implicit and explicit time integration schemes respectively. Initially validated for Newtonian fluids, the solvers were extended to handle more complex behaviours, including viscoelastic, viscoplastic, and elasto-viscoplastic materials, potentially undergoing time-dependent evolution or fluid-to-solid phase transitions. The non-Newtonian solvers were benchmarked on well-established case studies involving free-surface flows, such as dam breaks, drop impacts, and jet buckling. At the same time, a set of specialized numerical techniques was developed to accurately simulate the 3D printing process in the PFEM framework. These tools include the use of the Arbitrary Lagrangian-Eulerian (ALE) approach to model nozzle movements, advanced algorithms to accurately capture base and interlayer contact, an adaptive mesh de-refinement technique to decrease the computational cost, and an efficient nodal mapping scheme to transfer historical variables across re-meshing events. The overall framework was validated by reproducing the extrusion and layer-deposition phases of 3D Concrete Printing (3DCP) and comparing results with experimental data. The explicit solver proved efficient in simulating a variety of different printing scenarios, highlighting the influence of material properties (density, yield stress, viscosity) and process parameters (nozzle diameter, nozzle height, printing velocity, extrusion velocity) on the morphology of the printed layers. The outcomes of these simulations enabled the generation of a dimensionless map summarizing all possible cross-sectional shapes and printing regimes in free-flow deposition of yield-stress fluids. Vertical and inclined multi-layer walls were also simulated and compared with experimental data from literature or provided in the framework of the collaboration with TU/e Eindhoven. Numerical results aligned closely with the measured data, also providing insights into the role of material compressibility in 3DCP. The implicit solver was preferred in all those cases involving longer printing times and where the objective was to predict in advance the geometrical conformity and buildability of complex shape objects. In the first example, a rectangular wall characterized by a W-type infill was virtually printed, investigating the role of thixotropy and assessing interlayer bonding quality at intersections. The second example involved the design and virtual printing of three parametric hollow objects, to evaluate their stability and geometrical conformity under varying material properties. This analysis also highlighted how the numerical tool can be used to predict when and where additional support could be placed to enhance the final quality of printed structural components. Overall, the proposed framework proved to be accurate, robust, and potentially applicable to different classes of materials, such as clays and polymers. Future improvements will aim to decrease the computational cost through the integration of AI and machine learning techniques, transforming the framework into an effective predictive tool and enabling real-time optimization and the creation of reliable digital twins for the 3D printing process.| File | Dimensione | Formato | |
|---|---|---|---|
|
GR_thesis.pdf
accessibile in internet per tutti a partire dal 18/03/2026
Dimensione
69.07 MB
Formato
Adobe PDF
|
69.07 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/237032