Large Area Additive Manufacturing (LAAM) of Short Fiber Reinforced Polymers (SFRP) represents an emerging technology for the production of large-scale structural components. Despite its growing industrial relevance, widespread adoption remains hindered by the process-induced anisotropy intrinsic to the deposition process and the absence of standard- ized methodologies capable of linking manufacturing parameters to the final structural performance of printed components. Current research remains largely fragmented, treating fluid dynamics and mechanical characterization as disconnected problems, often with oversimplified microstructural assumptions. This thesis addresses these limitations by developing an integrated multi-scale numerical framework—combining process simulation, fiber orientation modeling, and micromechanical homogenization—with the overarching objective of establishing a physically rigorous and extensible tool for the virtual character- ization of LAAM composites. The methodology is structured as a four-step sequential pipeline. The process simulation stage employs two complementary CFD solvers: ANSYS Fluent, implementing a generalized Newtonian (Cross-William Lander Ferry) rheological model, and ANSYS Polyflow, which introduces a viscoelastic formulation via the Phan- Thien-Tanner (PTT) model to capture die swell and post-extrusion bead spreading. The velocity gradient fields extracted from the CFD are fed into a custom Lagrangian tracking algorithm implementing the principal Anisotropic Rotary Diffusion with Reduced Strain Closure (pARD-RSC) orientation model; the resulting stiff system of Ordinary Differential Equations is integrated via a Backward Differentiation Formula (BDF) scheme to ensure numerical stability in high-shear regimes. A central methodological contribution is the iterative calibration procedure developed for the pARD-RSC model, which adjusts the model parameters (CI , k, Ω) against reference orientation data, enabling physics-informed parameter identification without reliance on purely empirical assumptions. The calibrated orientation tensor field then drives the stochastic generation of a Representative Volume Element (RVE) of the ABS/20% Carbon Fiber composite, upon which Finite Element micromechanical homogenization is performed in Abaqus, incorporating a cohesive fiber-matrix interaction model to account for interfacial behavior and extract the effective orthotropic elastic properties. The framework is applied to a parametric study spanning Pull and Push deposition regimes across a systematic Design of Experiments. Geometric validation of the CFD predictions is carried out against a dedicated experimental printing campaign through cross-sectional bead analysis. The fiber orientation analysis quantifies the sensitivity of the alignment state to deposition kinematics, showing that the Pull regime induces a stronger flow-aligned orientation (A11 = 0.68) compared to the Push regime (A11 = 0.64). A sensitivity analysis of the homogenized elastic modulus with respect to the orientation tensor components demonstrates a pronounced anisotropic response, confirming the dominant role of fiber alignment on the predicted structural stiffness. The complete experimental validation of the micromechanical stage—through direct comparison with mechanical testing of printed specimens—is foreseen within the proposed methodology and represents the natural continuation of the present work. The integrated pipeline constitutes an original and extensible digital framework for the process-informed design of LAAM components, providing a solid foundation for future validation campaigns and macro-scale structural optimization.
La manifattura additiva di grandi dimensioni (Large Area Additive Manufacturing, LAAM) di polimeri rinforzati con fibre corte (Short Fiber Reinforced Polymers, SFRP) rappresenta una tecnologia emergente per la produzione di componenti strutturali su larga scala. Nonostante la sua crescente rilevanza industriale, la diffusione su ampia scala è ancora ostacolata dall'anisotropia indotta dal processo di deposizione e dall'assenza di metodologie standardizzate in grado di collegare i parametri di fabbricazione alle prestazioni meccaniche finali del componente stampato. La letteratura esistente rimane in larga misura frammentata, trattando la fluidodinamica e la caratterizzazione meccanica come problemi separati, spesso con assunzioni microstrutturali semplificate. Questa tesi affronta tali limitazioni sviluppando un framework numerico multi-scala integrato—che combina simulazione di processo, modellazione dell'orientazione delle fibre e omogeneizzazione micro-meccanica—con l'obiettivo generale di costruire uno strumento fisicamente rigoroso ed estensibile per la caratterizzazione virtuale di compositi prodotti mediante LAAM. La metodologia è strutturata come una pipeline sequenziale a quattro fasi. La simulazione di processo impiega due solver CFD complementari: ANSYS Fluent, con un modello reologico newtoniano generalizzato (Cross-William Lander Ferry), e ANSYS Polyflow, che introduce una formulazione viscoelastica basata sul modello di Phan-Thien-Tanner (PTT) per catturare il rigonfiamento all'uscita dall'ugello (die swell) e la morfologia del cordone depositato. I campi del gradiente di velocità estratti dalla CFD alimentano un algoritmo di tracciamento lagrangiano personalizzato che implementa il modello di orientazione pARD-RSC (principal Anisotropic Rotary Diffusion with Reduced Strain Closure); il sistema di equazioni differenziali ordinarie risultante, di natura stiff, viene integrato tramite uno schema alle Differenze All'Indietro (Backward Differentiation Formula, BDF) per garantire la stabilità numerica nei regimi ad alto shear. Un contributo metodologico centrale è la procedura di calibrazione iterativa sviluppata per il modello pARD-RSC, che calibra i parametri del modello (CI , k, Ω) invece di impiegare dati di orientazione di riferimento da letteratura, consentendo un'identificazione dei parametri fisicamente fondata senza ricorrere ad assunzioni puramente empiriche. Il tensore di orientazione calibrato guida poi la generazione stocastica di un Elemento di Volume Rappresentativo (Representative Volume Element, RVE) del composito ABS/20% fibra di carbonio, sul quale viene eseguita un'omogeneizzazione micro-meccanica agli Elementi Finiti in Abaqus, con un modello di interazione coesiva fibra-matrice per descrivere il comportamento dell'interfaccia ed estrarre le proprietà elastiche ortotrope efficaci. Il framework viene applicato a uno studio parametrico che abbraccia i regimi di deposizione Pull e Push nell'ambito di un piano sperimentale sistematico (Design of Experiments). La validazione geometrica delle previsioni CFD è condotta mediante un'apposita campagna sperimentale di stampa, attraverso l'analisi della sezione trasversale dei cordoni depositati. L'analisi dell'orientazione delle fibre quantifica la sensibilità dello stato di allineamento alla cinematica di deposizione, mostrando come il regime Pull induca un'orientazione preferenziale più marcata nella direzione di flusso (A11 = 0.68) rispetto al regime Push (A_11 = 0.64). Un'analisi di sensitività del modulo elastico omogeneizzato rispetto alle componenti del tensore di orientazione evidenzia una risposta fortemente anisotropa, confermando il ruolo determinante dell'allineamento delle fibre sulle proprietà meccaniche previste. La validazione sperimentale completa della fase micro-meccanica—tramite confronto diretto con prove meccaniche su provini stampati—è prevista nella metodologia proposta e rappresenta la naturale prosecuzione del presente lavoro. La pipeline integrata costituisce un framework digitale originale ed estensibile per la progettazione process-informed di componenti LAAM, fornendo una base metodologica solida per future campagne di validazione e per l'ottimizzazione strutturale a scala macro.
Micro-scale modeling of LAAM processes: an integrated and validable framework from flow dynamics to mechanical characterization
FAVERZANI, ELIA
2025/2026
Abstract
Large Area Additive Manufacturing (LAAM) of Short Fiber Reinforced Polymers (SFRP) represents an emerging technology for the production of large-scale structural components. Despite its growing industrial relevance, widespread adoption remains hindered by the process-induced anisotropy intrinsic to the deposition process and the absence of standard- ized methodologies capable of linking manufacturing parameters to the final structural performance of printed components. Current research remains largely fragmented, treating fluid dynamics and mechanical characterization as disconnected problems, often with oversimplified microstructural assumptions. This thesis addresses these limitations by developing an integrated multi-scale numerical framework—combining process simulation, fiber orientation modeling, and micromechanical homogenization—with the overarching objective of establishing a physically rigorous and extensible tool for the virtual character- ization of LAAM composites. The methodology is structured as a four-step sequential pipeline. The process simulation stage employs two complementary CFD solvers: ANSYS Fluent, implementing a generalized Newtonian (Cross-William Lander Ferry) rheological model, and ANSYS Polyflow, which introduces a viscoelastic formulation via the Phan- Thien-Tanner (PTT) model to capture die swell and post-extrusion bead spreading. The velocity gradient fields extracted from the CFD are fed into a custom Lagrangian tracking algorithm implementing the principal Anisotropic Rotary Diffusion with Reduced Strain Closure (pARD-RSC) orientation model; the resulting stiff system of Ordinary Differential Equations is integrated via a Backward Differentiation Formula (BDF) scheme to ensure numerical stability in high-shear regimes. A central methodological contribution is the iterative calibration procedure developed for the pARD-RSC model, which adjusts the model parameters (CI , k, Ω) against reference orientation data, enabling physics-informed parameter identification without reliance on purely empirical assumptions. The calibrated orientation tensor field then drives the stochastic generation of a Representative Volume Element (RVE) of the ABS/20% Carbon Fiber composite, upon which Finite Element micromechanical homogenization is performed in Abaqus, incorporating a cohesive fiber-matrix interaction model to account for interfacial behavior and extract the effective orthotropic elastic properties. The framework is applied to a parametric study spanning Pull and Push deposition regimes across a systematic Design of Experiments. Geometric validation of the CFD predictions is carried out against a dedicated experimental printing campaign through cross-sectional bead analysis. The fiber orientation analysis quantifies the sensitivity of the alignment state to deposition kinematics, showing that the Pull regime induces a stronger flow-aligned orientation (A11 = 0.68) compared to the Push regime (A11 = 0.64). A sensitivity analysis of the homogenized elastic modulus with respect to the orientation tensor components demonstrates a pronounced anisotropic response, confirming the dominant role of fiber alignment on the predicted structural stiffness. The complete experimental validation of the micromechanical stage—through direct comparison with mechanical testing of printed specimens—is foreseen within the proposed methodology and represents the natural continuation of the present work. The integrated pipeline constitutes an original and extensible digital framework for the process-informed design of LAAM components, providing a solid foundation for future validation campaigns and macro-scale structural optimization.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_07_Faverzani_Frignati.pdf
non accessibile
Dimensione
61.66 MB
Formato
Adobe PDF
|
61.66 MB | Adobe PDF | Visualizza/Apri |
|
2026_07_Faverzani_Executive Summary.pdf
non accessibile
Dimensione
5.8 MB
Formato
Adobe PDF
|
5.8 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/10589/259597