Fused deposition modelling (FDM) is a versatile additive manufacturing technique involving extruding thermoplastic polymers layer by layer. FDM enables the fabrication of complex geometries, however prediction and minimization of warpage deformation is critical for high-precision applications such as medical implants. Simulations based on finite element analysis (FEA) have been shown to accurately predict warpage deformation in FDM; however, this approach is computationally intensive and has limitations for process optimisation and monitoring. In this study, a feed-forward neural network (FFNN) model was trained using deformation values from FEA simulations. Input features included nozzle temperature, material deposition rate, bed temperature, layer height, layer number, and spatial coordinates, while warpage deformation served as the output. The trained FFNN was then combined with the particle swarm optimization (PSO) algorithm to find the optimal FDM parameters to minimize the warpage deformation. The combination of a trained FFNN based on FEA with PSO identified optimal FDM parameters, producing a minimum warpage deformation of 0.062 mm. The proposed PSO–FFNN approach demonstrates the feasibility of using machine learning and metaheuristic algorithms to minimize warpage defects in additive manufacturing while reducing computational cost and computation time.

A novel finite element modelling approach is presented which incorporates representative binder yarn compaction, for simulating the low-velocity impact (LVI) and compression after impact (CAI) response of 3D woven layer-to-layer carbon/epoxy composite architectures. Simulations of out-of-plane drop-weight impact tests were performed at energies of 32 J and 42 J. Warp and weft layers were modelled as continuous plies and three different approaches were explored to model the binder reinforcement; (i) with a rectangular cross-section and non-compacted, (ii) with an elliptical cross-section and non-compacted, and (iii) an elliptical cross-section which accounts for compaction. Predictions were compared with experimental results from literature and it is shown that modelling the binder reinforcement as an elliptical cross-section with compaction leads to a predicted damage area, on the impacted side, which is within 2%, and the non-impacted side within 6% of experimental measurements. The predicted CAI strength is within 11% of the experimental values.

Semi-crystalline polymers develop higher amounts of residual stress and part distortion (warpage) compared to amorphous polymers due to their crystalline nature. Additionally, the FDM processing parameters such as ambient temperature play an important role in the resulting residual stresses and part distortion of the printed part. Hence, in this study, the effect of ambient temperature on the in-built residual stresses and warpage of amorphous acrylonitrile-butadiene-styrene (ABS) and semi-crystalline polypropylene (PP) polymers was investigated. From the results, it was observed that increasing the ambient temperature from 50 °C to 75 °C and further to 120 °C resulted in 0.22-KPa and 0.37-KPa decreases in residual stress of ABS, but no significant change in the amount of warpage. For PP, increasing ambient temperature from 50 °C to 75 °C led to a more considerable decrease in residual stress (0.5 MPa) and about 3% increase in warpage. Further increasing to 120 °C resulted in a noticeable 2 MPa decrease in residual stress and a 3.4% increase in warpage. Reduction in residual stress in both ABS and PP as a result of increasing ambient temperature was due to the reduced thermal gradients. The enhanced warpage in PP with increase in ambient temperature, despite the reduction in residual stress, was ascribed to crystallization and shrinkage.

In fused deposition modelling (FDM) based on the selected raster pattern, the developed internal thermal residual stresses can vary considerably affecting the mechanical properties and leading to distinct part distortions. This phenomenon is more pronounced in semi-crystalline than amorphous polymers due to crystallisation. Hence, this study focuses on the simulation of the FDM process of a semi-crystalline polymer (polypropylene) with raster patterns such as line (90°/90°), line (0°/90°), zigzag (45°/45°), zigzag (45°/−45°), and concentric from Cura (slicing software). The simulation provides visualisation and prediction of the internally developed thermal residual stresses and resulting warpage with printing time and temperature. The sample with a line (90°/90°) raster pattern is considered as the reference sample in order to compare the relative levels of residual stress and warpage in the other printed/simulated samples. Among the considered raster patterns, the concentric pattern displays the lowest amount of warpage (5.5% decrease) along with a significant drop in residual stress of 21%. While the sample with a zigzag (45°/−45°) pattern showed the highest increase of 37% in warpage along with a decrease of 9.8% in residual stresses. The sample with a zigzag (45°/45°) pattern, exhibited a considerable increase of 16.2% in warpage with a significant increase of 31% in residual stresses. Finally, the sample with a line (0°/90°) raster pattern displayed an increase of 24% increase in warpage with an increase of 6.6% in residual stresses.

 Abstract — Fused Deposition Modelling (FDM) is one of the additive manufacturing techniques that has become highly attractive in the industrial and academic sectors. However, parts fabricated through FDM are highly susceptible to geometrical defects such as warpage, shrinkage, and delamination that can severely affect their function. Among the thermoplastic polymer feedstock for FDM, semi-crystalline polymers are highly prone to part distortion due to polymer crystallization. In this study, the influence of FDM processing conditions such as chamber temperature and print bed temperature on the induced thermal residual stress and resulting warpage are investigated using 3D transient thermal model for a semi-crystalline polymer. The thermo-mechanical properties and the viscoelasticity of the polymer, as well as the crystallization physics which considers the crystallinity of the polymer, are coupled with the evolving temperature gradient of the print model. From the results it was observed that increasing the chamber temperature from 25 °C to 75 °C leads to a decrease of 3.3% residual stress and increase of 0.4% warpage, while decreasing bed temperature from 100 °C to 60 °C resulted in 27% increase in residual stress and a significant rise of 137% in warpage. The simulated warpage data are validated by comparing it with the measured warpage values of the samples using 3D scanning.

The printing conditions in Fused Deposition Modelling (FDM) affect the amount of induced residual stresses within the printed part and its dimensional accuracy. Among the thermoplastic feedstock for FDM, semicrystalline polymers are more prone to part distortion due to crystallisation. Therefore, this study aims to numerically investigate the behaviour of semi-crystalline polymer under various FDM printing conditions (namely print speed and ambient temperature) and the resultant residual stress and warpage in the printed parts. For this, the coefficient of thermal expansion (CTE) and the thermo-mechanical properties of the polymer under study (polypropylene), and the crystallisation kinetics are coupled with the evolving temperature and time during printing. The values of residual stress and warpage are calculated and compared for the bottom and top layers of the samples. From the results, it was observed that increasing the nozzle speed from 30 mm/s to 60 mm/ s resulted in the bottom and top layers exhibiting a 15% and 13% decrease in residual stress, respectively. Similarly, a drop in warpage (-30%) was observed for both layers. The reduction in residual stress and warpage with increased printing speed is attributed to the improved heat transfer between the deposited roads and the reduced cooling rate. Increasing the ambient temperature from 25 degrees C to 75 degrees C resulted in a 2% and 3% decrease in residual stress in the bottom and top layers, respectively. In terms of warpage, an insignificant increase (-1%) was observed in both top and bottom layers. This is explained by the counter effects of reduced thermal gradients (i.e., lower cooling rate) and increased crystallisation on the overall amount of residual stress and warpage. 3D scanning of experimentally printed samples was used for verification of the simulation results, and good agreement between these is reported.

Residual stresses and warpage adversely affect the dimensional accuracy and performance of 3D-printed semi-crystalline polymers in Fused Deposition Modelling (FDM). One of the main challenges in FDM is to understand and relate the impact of printing conditions on part distortion for optimizing the 3D-printing process to achieve good print quality. Hence, the effect of various printing parameters, particularly print bed temperature, layer bonding, layer thickness and raster pattern, on built-up residual stresses and warpage is simulated in this work, by building a relationship between the crystallisation kinetics, viscoelastic and thermo-mechanical properties of the polymer in relation to changes in temperature during FDM using element activation in COMSOL. To the best of our knowledge, this is a novel approach for quantitative prediction of part distortion in FDM of semi-crystalline polymers under various printing conditions. Based on the simulation results, it is observed that a decrease in layer thickness from 0.5 mm to 0.1 mm results in an 89% drop in warpage and a reduction in residual stress of 24%. Applying a line raster pattern reduces warpage and residual stresses by 16% and 36%, respectively in comparison with a zigzag raster pattern. Very good agreement is observed between simulation and experimental results for warpage under various printing conditions. The results of this study can be used to predict and/or minimise part distortion in a semi-crystalline, 3D-printed polymer by simulating the effect of printing parameters on residual stresses during FDM. Crown Copyright (C) 2021

Polymers have been widely used in the field of fused deposition modelling (FDM). The part integrity of the final printed part is affected by parameters such as processing conditions and the material properties of the polymer. Build-up of residual stresses are the main cause of shrinkage and warpage (i.e., part distortion) in the FDM parts. Among the thermoplastic polymers, semi-crystalline polymers are more prone to part distortion due to crystallisation. Therefore, it is important to understand and predict part distortion in FDM of polymers to achieve good quality prints with desirable mechanical properties. Several studies have investigated the resulting part distortion in FDM parts through empirical, analytical, and numerical approaches. In most cases, the simulation results are not quantitatively validated, mainly because the temperature dependent properties of the polymers and the crystallinity of semi-crystalline polymers are often overlooked. In this study, the thermal-mechanical properties of the polymer of study such as specific heat capacity, thermal conductivity and density and the crystallisation kinetics are invoked as a function of temperature. Furthermore, an amorphous polymer was also simulated with consideration of its respective material properties. Both the semicrystalline and the amorphous polymer models were simulated under various layer thickness (0.1 and 0.5mm), in order to investigate the effect of layer thickness on the induced thermal stress and resulting warpage. Based on the simulation results, for 0.1mm layer thickness, the amorphous polymer model exhibited a warpage drop of 77%. And for 0.5mm, the warpage noted was found to decrease by 63%, on comparison with the warpage noted from semi-crystalline polymer model. These warpage values from the simulated models were then measured against the 3D scan results of the printed samples for quantitative validation. An excellent agreement was observed between the experimental and the simulated samples.