Laser-assisted additive manufacturing of high-performance thermoplastic composites: process simulation and optimisation

Abstract

High-performance thermoplastic composites (HPTCs), such as carbon fibre reinforced polyether ether ketone (CF/PEEK), play a pivotal role in lightweight structural applications across aerospace, automotive, and biomedical industries owing to their favourable strength to weight ratios and thermal stability. Additive manufacturing of these materials offers advantages in design flexibility and material efficiency compared with conventional routes such as compression moulding or injection moulding. Among material extrusion techniques, fused filament fabrication (FFF) has strong potential for producing complex geometries with minimal waste; however, it faces notable engineering challenges, including warpage induced dimensional inaccuracies, reduced interlayer adhesion, and spatial inconsistencies in internal properties. These issues—closely linked to local thermal gradients, repeated reheating events, and composite anisotropy—have motivated the development of laser assisted fused filament fabrication (LAFFF), which introduces targeted heating to improve consolidation and interlayer performance. Whilst LAFFF can mitigate some limitations of conventional FFF, its effects on warpage and mechanical performance remain variable, and the multi parameter interactions and governing mechanisms require systematic investigation to realise industrial potential.
These challenges arise from (i) incomplete understanding and quantification of coupled thermal and crystallisation mechanisms during LAFFF under realistic deposited states (including void affected heat transfer), (ii) practical inefficiencies in balancing simulation fidelity with computational cost for part scale prediction and optimisation, especially under small sample data regimes, and (iii) the absence of robust multi objective process selection frameworks that explicitly incorporate temperature history as the mediating variable linking process parameters to structure and performance.
This dissertation presents an integrated methodology that combines physics based modelling, experimental characterisation, parameterised finite element (FE) simulations, physics informed machine learning, and systematic optimisation for CF/PEEK LAFFF. A coupled thermo crystallisation FE model was developed, incorporating anisotropic heat transfer and dual non isothermal Avrami kinetics, and validated against synchronised infrared thermography, thermocouple measurements, and differential scanning calorimetry. The model reproduces location specific temperature histories and crystallinity distributions within defined experimental error ranges and provides mechanistic insight into parameter effects on thermal gradients, heat accumulation (thermal redundancy), and microstructural uniformity. To enable efficient temperature history prediction, a kinematics informed surrogate framework—the Kinematics Anchored Temperature Peaks Surrogate Model (KATP SM)—was introduced. Toolpath kinematics analytically prescribe the phases of dominant heating events, while Support Vector Machine / Support Vector Regression (SVM/SVR) learning is restricted to physically interpretable low dimensional thermal features (peak amplitudes and a slowly varying redundancy baseline), enabling millisecond scale prediction of ROI temperature–time histories across a multi parameter domain.
Building on these tools, thermo mechanical–crystallisation simulations were established to quantify residual stress and warpage, with predicted deformation agreeing within ±17% of digital image correlation measurements. Response surface methodology coupled with the NSGA III algorithm was then used to explore the process space and generate Pareto optimal parameter sets, resolving trade offs between interlayer shear strength, flexural strength, and warpage deformation. Three practically validated operating regimes were identified: strength priority, dimensional-accuracy-priority (warpage below 0.5 mm with acceptable strengths), and balanced.
Overall, this research establishes a temperature mediated process–structure–property–performance framework for CF/PEEK LAFFF, integrating analytical, numerical, and data driven elements to deliver mechanistic understanding and actionable optimisation strategies. The methodology is transferable to a wider range of fibre reinforced thermoplastic systems in additive manufacturing.

Date of Award	15 Apr 2026
Original language	English
Awarding Institution	University of Nottingham
Supervisor	Ping Cui (Supervisor), Jian Yang (Supervisor), Kok Wong (Supervisor) & Yingdan Zhu (Supervisor)