Process Parameter Selection for Production of Stainless Steel 316L Using Efficient Multi-Objective Bayesian Optimization Algorithm

Laser powder bed fusion (LPBF) of titanium or titanium alloys allows fabrication of geometrically more complex and, possibly, individualized implants or osteosynthesis products and could thus improve the outcome of medical treatments considerably. However, insufficient LPBF process parameters can result in substantial porosity, decreasing mechanical properties and requiring post-treatment. Furthermore, texturized parts with anisotropic properties are usually obtained after LPBF processing, limiting their usage in medical applications. The present study addresses both: first, a design of experiments is used in order to establish a set of optimized process parameters and a process window for LPBF printing of small commercially pure (CP) titanium parts with minimized volume porosity. Afterward, the first results on the development of a biocompatible titanium alloy designed for LPBF processing of medical implants with improved solidification and more isotropic properties are presented on the basis of conventionally melted alloys. This development was performed on the basis of Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au, a near-α alloy presented by the authors for medical applications and conventional manufacturing, with yttrium and boron additions as additional growth restriction solutes. In terms of LPBF processing of CP titanium grade 1 powder, a high relative density of approximately 99.9% was obtained in the as-printed state of the volume of a small cubical sample by using optimized laser power, scanning speed, and hatch distance in combination with a rotating scanning pattern. Moreover, tensile specimens processed with these volume settings and tested in the as-printed milled state exhibited a high average yield and ultimate tensile strength of approximately 663 and 747N/mm2, respectively, combined with a high average ductility of approximately 24%. X-ray diffraction results suggest anisotropic mechanical properties, which are, however, less pronounced in terms of the tested specimens. Regarding alloy development, the results show that yttrium additions lead to a considerable microstructure refinement but have to be limited due to the occurrence of a large amount of precipitations and a supposed higher propensity for the formation of long columnar prior β-grains. However, phase/texture and microstructure analyses indicate that Ti-0.44O-0.5Fe-0.08C-0.4Si-0.1Au-0.1B-0.1Y is a promising candidate to achieve lower anisotropy during LPBF processing, but further investigations on LPBF printing and Y2O3 formation are necessary.

The optimization of processing parameters is indispensable for the laser powder bed fusion (L-PBF) process. The deposited energy density (DED) is one of the process indexes for the L-PBF process and has a simplified formula of P·v−0.5, where P is the laser power, and v is the scan speed. This parameter describes the change in the relative density and the melt pool morphology with laser power and scan speed well, whereas it does not include the effect of other processing parameters, e.g., hatch spacing (S). In the present study, an attempt was made to incorporate the effect of S into DED. Al–12Si (mass%) alloy cube samples were fabricated by L-PBF under various P, v, and S, for evaluating the relative density and the melt pool morphology. The melt pool depth and width of L-PBF-manufactured Al–12Si alloy increased linearly with P·v−0.5 and did not exhibit a clear correlation with S. Based on the experimental observation, the effect of hatch spacing on DED was estimated to be S−0.5, and a new index of P·v−0.5·S−0.5 was proposed. This index described the change in the relative density of the L-PBF-manufactured Al–12Si alloy with laser conditions (P, v, and S) well when the thermal conduction mode melting was dominant. This study also indicated the limitation of the applicability of P·v−0.5·S−0.5 under the keyhole or transition mode melting.

This study presents a comprehensive numerical investigation of the Laser Powder Bed Fusion (LPBF) process for pure molybdenum, focusing on high-precision modeling and process optimization. The powder spreading behavior is simulated using the Discrete Element Method (DEM), while molten pool dynamics are analyzed through Computational Fluid Dynamics (CFD). Optimization of process parameters is performed using FLOW-3D Release 7 software in conjunction with the HEEDS-SHERPA algorithm. A total of 247 simulations are conducted to assess the effects of four critical parameters: laser power (50–400 W), scanning speed (80–300 mm/s), laser spot diameter (40–100 µm), and powder layer thickness (50–100 µm). The optimal parameter set—350 W laser power, 120 mm/s scanning speed, 50 µm spot diameter, and 50 µm layer thickness—results in an 80% laser absorption rate, a 60% reduction in micro-porosity, and over a 30% enhancement in both molten pool volume and surface area. Utilizing a fine 10 µm mesh resolution enables detailed insights into temperature gradients and phase transition behavior. The findings highlight that optimized parameter selection significantly improves the structural integrity of Mo-based components while minimizing manufacturing defects, thus offering valuable guidance for advancing industrial-scale additive manufacturing of refractory metals.

In the Laser Powder Bed Fusion (L-PBF) process, 3D components with complex geometries are fabricated in a layer-by-layer fashion by using a controlled laser beam to selectively melt particular regions of the metal powder bed. However, due to the stochastic nature of the L-PBF process, the top surface roughness of each solidified layer tends to be different even when the optimal processing conditions for the different positions on the build plate are employed. As a result, the mechanical properties of the built components frequently vary from one component to the next. Accordingly, this study proposes an Intelligent Additive Manufacturing Architecture (IAMA) for controlling the surface roughness of each build layer through an appropriate adjustment of the laser re-melting parameters. The IAMA architecture comprises five modules, namely In-Situ Metrology (ISM), Ex-Situ Metrology (ESM), Automatic Virtual Metrology (AVM), Additive Manufacturing Simulation (AMS) and Intelligent Compensator (IC). The feasibility of the proposed architecture is demonstrated by comparing the top surface roughness of cubic and mechanical strengths of tensile test samples built using the proposed method with those built using a traditional L-PBF approach without surface roughness control. It is found that the samples fabricated using the IAMA approach have an average top surface roughness of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.6~\mu \text{m}$ </tex-math></inline-formula> and a standard deviation is <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$0.7~\mu \text{m}$ </tex-math></inline-formula> . By contrast, the samples produced using the traditional L-PBF approach have an average surface roughness of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$13.45~\mu \text{m}$ </tex-math></inline-formula> and a standard deviation of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2.5~\mu \text{m}$ </tex-math></inline-formula> . In addition, the specimens produced with the assistance of IAMA architecture have an average tensile strength of 1013 MPa with a standard deviation of 69.5 MPa, while those printed without surface roughness control have an average tensile strength of 903 MPa with a standard deviation of 101.4 MPa Note to Practitioners—As L-PBF produce part in a layer-by-layer manner, therefore, the roughness on the top surface of previous layer have a strong influence on the printing quality of current layer. The variations of surface roughness will lead to the fluctuation of mechanical properties of the fabricated components. Additionally, to the best of author’s knowledge, current commercial L-PBF machines can not actively control the surface roughness of samples during the process. The proposed IAMA in this work can predict and control the roughness on the top surface of each layer during L-PBF process. Therefore, the constructed architecture has strong potential for integrating into the commercial L-PBF machine for controlling the surface roughness on the top of the parts during the manufacturing process. As a result, the quality of the fabricated components is expected to be in consistent. Accordingly, L-PBF machine equipped with IAMA will have a strong potential in applying for mass production of the components in aerospace and automobile industries.

PurposeThis study aims to investigate the relationship between melt pool dimensions and acoustic emission (AE) signal magnitudes obtained during laser powder bed fusion (L-PBF) process of austenitic stainless steel. Specifically, by varying laser power and scan speed, the widths, depths and aspect ratios of melt pools were compared with AE signal magnitudes across a frequency range of 100–200 kHz.Design/methodology/approachExperiments were conducted under different laser powers at a fixed scan speed and scan speeds at a fixed laser power. Melt pool dimensions were measured from cross-sectional optical images, and AE signals were obtained using a piezoelectric AE sensor installed beneath the build plate. Short-Time Fourier Transform (STFT) was applied to AE signals, and the magnitudes of frequency components were obtained.FindingsA strong correlation between melt pool dimensions and STFT magnitude was obtained. Pearson correlation coefficients between melt pool dimensions and STFT magnitudes were above 0.9 and the p-values were below 0.05. Increasing the laser energy enlarged the volume of melt pool and intensified the oscillation of melt pool. When scan speed exceeded 1,100 mm/s, STFT magnitude showed a slight increase owing to the increase in the vapor pressure.Originality/valuePrevious studies used AE signals to detect defects, but this study found a correlation between STFT magnitude and melt pool dimensions in L-PBF process. It was also found that STFT magnitude was more affected by vapor pressure at higher scan speeds. Monitoring STFT magnitude can help to understand melt pool dynamics, maintain process consistency and identify irregularities in real time.

Effect of process parameters and build orientations on the mechanical properties of maraging steel (18Ni-300) parts printed by selective laser melting

Fatigue performance of shelled additively manufactured parts subjected to hot isostatic pressing

Additive manufacturing (AM) enables the production of customized and complex metal parts, providing greater flexibility than traditional mass production. However, maintaining consistent product quality for AM processes remains challenging due to process variability and heat emission fluctuations, which can cause defects such as overheating. Overheating occurs when non-uniform temperature distributes across the printed parts, leading to issues such as structural weaknesses. Predicting heat emissions accurately is essential to preventing such issues, but AM facilities, particularly small- and medium-sized enterprises (SMEs), generate limited and heterogeneous datasets due to unique configurations and privacy concerns, making traditional centralized predictive model training paradigms impractical. Federated learning (FL) offers a privacy-preserving solution, it alleviates data privacy and data scarcity concerns by enabling collaborative learning across decentralized datasets while keeping data local. However, its performance suffers under heterogeneous client data and varied computational resources. This creates a need to address data heterogeneity while maintaining both global model performance and individual client personalization. We propose FedScope-KD, a novel heterogeneity-aware FL framework designed to balance the scope of global and personalized learning. Client heterogeneity level is quantified prior to the FL training, to improve the model weight aggregation during training. This is achieved through component decomposition and knowledge distillation. The framework decomposes client data into process-invariant representations, which capture the fundamental nature of the laser powder bed fusion (LPBF) process, and process-variant representations, which reflect local, client-specific configuration settings. This decomposition ensures that global models benefit from collaborative learning on the homogeneous data representations, while personalized predictions remain tailored to client-specific heterogeneous data representations. A knowledge distillation (KD) mechanism further enhances the framework by efficiently transferring process-invariant knowledge from a global teacher model to local student models. We compare the heterogeneity index generation and calculate the Dynamic Time Warping (DTW) distance for both inter-client variant and invariant components. The variant components exhibit greater distances between clients, reflecting their inherent heterogeneity, while the invariant components are more closely aligned across clients. This pattern demonstrates the effectiveness of the decomposition process in separating heterogeneous and homogeneous information. The heterogeneity index is further used for weight aggregation during FL training. The framework’s performance is validated through a comparison with baseline FL methods such as FedAvg, as well as centralized and individual learning paradigms. Performance metrics include personalized performance and global accuracy with MSE. Through experiments on heat emission datasets in the LPBF process, we show that FedScope-KD successfully addresses the challenges of data heterogeneity and resource constraints, providing a robust solution of predictive modeling for SMEs.

Ultra-fine bimodal (α + β) microstructure induced mechanical strength and corrosion resistance of Ti-6Al-4V alloy produced via laser powder bed fusion process

Multiscale modelling of microstructure evolution, and local solidification behaviours of the AlSi10Mg build component in laser powder bed fusion process

Effect of powder bed preheating on the crack formation and microstructure in ceramic matrix composites fabricated by laser powder-bed fusion process

Additive manufacturing has made the fabrication of large-dimensioned bulk metallic glasses (BMGs) achievable; however, questions remain regarding how to control the processing parameters to obtain dense and fully amorphous BMGs with desirable mechanical properties. Here, laser powder bed fusion (LPBF) was used to produce dense and fully amorphous Zr59.3Cu28.8Nb1.5Al10.4 BMG samples from two different starting powders within a large processing window of laser powers and scanning speeds. X-ray diffraction (XRD) revealed that fully amorphous XRD patterns were obtained for samples with high relative densities (>99 %) when the LPBF energy density ranged from ∼20 J/mm3 up to ∼33 J/mm3 for coarse powder with low oxygen and up to ∼30 J/mm3 for fine powder with higher oxygen. For the fully XRD amorphous samples, strength and hardness increased with increasing energy density while the relaxation enthalpy and ductility decreased. Transmission electron microscopy revealed that the softer samples contain larger medium range order clusters within the amorphous matrix. With higher LPBF energy density, high relative density was still achieved along with devitrification and embrittlement. While lower energy densities below ∼20 J/mm3 could retain a fully XRD amorphous structure, such samples had relative densities <99 %. When comparing the two powders, the coarse powder with four times lower oxygen content gave better glass forming ability, compression ductility up to 6 % plastic strain, and fracture toughness up to ∼38 MPa√m. These findings demonstrate that it is possible to tailor the structure and mechanical properties of BMGs by tuning the LPBF process parameters within a wide processing window.

In multi-objective Bayesian optimization, an infill criterion is an important part, as it is the indicator to evaluate how much good a new set of solutions is, compared to a Pareto-front approximation set. This paper presents a deterministic algorithm for computing the Expected R2 Indicator for bi-objective problems and studies its use as an infill criterion in Bayesian Global Optimization. The R2-Indicator was introduced in 1998 by M. Hansen and A. Jaszkiewicz for performance assessment in multi-objective optimization and is more recently also used in indicator-based multi-criterion evolutionary algorithms (IBEAs). In Bayesian Global Optimization, we propose the Expected R2-indicator Improvement (ER2I) as an infill criterion. It is defined as the expected decrease of the R2 indicator by a point that is sampled from a predictive Gaussian distribution. The ER2I can also be used as a pre-selection criterion in surrogate-assisted IBEAs. It provides an alternative to the Expected Hypervolume-Indicator Improvement (EHVI) that requires a reference point, bounding the Pareto front from above. In contrast, the ER2I works with a utopian reference point that bounds the Pareto front from below. In addition, the ER2I supports preference modelling with utility functions and its computation time grows only linearly with the number of considered weight combinations. It is straightforward to approximate the ER2I by Monte Carlo Integration, but so far a deterministic algorithm to solve the non-linear integral remained unknown. We outline a deterministic algorithm for the computation of the bi-objective ER2I with Chebychev utility functions. Moreover, we study monotonicity properties of the ER2I w.r.t. parameters of the predictive distribution and numerical simulations demonstrate fast convergence to Pareto fronts of different shapes and the ability of the ER2I Bayesian optimization to fill gaps in the Pareto front approximation.

In this paper, an adaptive optimization method based on Gaussian process (GP) surrogate model is proposed to minimize the warpage of injection molding parts. GP surrogate model combining design of experiment (DOE) methods is used to build an approximate function relationship between warpage and process parameters, replacing the expensive simulation analysis in the optimization iterations. First, establish an approximation function of the relationship between warpage and process parameters by a small size of design of experiment with GP surrogate model. And then, an enhanced probability improvement criterion is used to determine how additional training samples could be added to optimize the surrogate model. Comparing with expected improvement criterion, proposed enhanced probability improvement criterion can switch to global optima more swiftly. Finally, a front grille molding processing is taken as an example to illustrate the criterion. The results show that the proposed optimization method can effectively decrease the warpage of injection molding parts.

A simplified analytical model of the laser powder bed fusion (LPBF) process was used to develop a novel density prediction approach that can be adapted for any given powder feedstock and LPBF system. First, calibration coupons were built using IN625, Ti64 and Fe powders and a specific LPBF system. These coupons were manufactured using the predetermined ranges of laser power, scanning speed, hatching space, and layer thickness, and their densities were measured using conventional material characterization techniques. Next, a simplified melt pool model was used to calculate the melt pool dimensions for the selected sets of printing parameters. Both sets of data were then combined to predict the density of printed parts. This approach was additionally validated using the literature data on AlSi10Mg and 316L alloys, thus demonstrating that it can reliably be used to optimize the laser powder bed metal fusion process.