In aerospace, thermal applications demand compact, lightweight, and efficient heat exchangers. Additive manufacturing processes offer the potential to create highly complex structures that are not achievable through traditional manufacturing methods. This work presents the development of an additively manufactured fluid-fluid heat exchanger that shows the potential to enhance the performance, reduce weight, and increase compactness compared to a conventional plate heat exchanger. A numerical model of the conventional plate heat exchanger was created, and fluid dynamics simulations with heat transfer were performed. Validation of the simulations was done by experiments. Then, a novel heat exchanger was designed using a bottom-up approach and investigated at different levels of complexity using computational fluid dynamics. The internal structure of the final heat exchanger consists of a repeating triply periodic Schwarz diamond minimum surface elongated in the direction of flow. The heat exchanger was manufactured with laser powder bed fusion process using AlSi10Mg. It had a 108% higher compactness and 54% lower weight compared to the plate heat exchanger. Numerical analysis yielded the pressure loss in pascal was reduced by 50%–59% while heat transfer in watts was improved by 3%–5%. Future researches should experimentally investigate the thermal and fluid mechanical characteristics of the novel additively manufactured heat exchanger and increase compactness and heat transfer further by analyzing the minimum partition wall thickness and the impact of wall roughness and deposit formation.

Laser powder bed fusion of metals (PBF-LB/M) is a process widely used in additive manufacturing (AM). It is highly sensitive to its process parameters directly determining the quality of the components. Hence, optimal parameters are needed to ensure the highest part quality. However, current approaches such as experimental investigation and the numerical simulation of the process are time-consuming and costly, requiring more efficient ways for parameter optimization. In this work, the use of machine learning (ML) for parameter search is investigated based on the influence of laser power and speed on simulated melt pool dimensions and experimentally determined part density. In total, four machine learning algorithms are considered. The models are trained to predict the melt pool size and part density based on the process parameters. The accuracy is evaluated based on the deviation of the prediction from the actual value. The models are implemented in python using the scikit-learn library. The results show that ML models provide generalized predictions with small errors for both the melt pool dimensions and the part density, demonstrating the potential of ML in AM. The main limitation is data collection, which is still done experimentally or simulatively. However, the results show that ML provides an opportunity for more efficient parameter optimization in PBF-LB/M.

Powder bed fusion of metals using laser beam (PBF-LB/M) is a commonly used additive manufacturing process for the production of high-performance metal parts. AlSi10Mg is a widely used material in PBF-LB/M due to its excellent mechanical and thermal properties. However, the part quality of AlSi10Mg parts produced using PBF-LB/M can vary significantly depending on the process parameters. This study investigates the use of machine learning (ML) algorithms for the prediction of the resulting part density of AlSi10Mg parts produced using PBF-LB/M. An empirical data set of PBF-LB/M process parameters and resulting part densities is used to train ML models. Furthermore, a methodology is developed to allow density predictions based on simulated meltpool dimensions for different process parameters. This approach uses finite element simulations to calculate the meltpool dimensions, which are then used as input parameters for the ML models. The accuracy of this methodology is evaluated by comparing the predicted densities with experimental measurements. The results show that ML models can accurately predict the part density of AlSi10Mg parts produced using PBF-LB/M. Moreover, the methodology based on simulated meltpool dimensions can provide accurate predictions while significantly reducing the experimental effort needed in process development in PBF-LB/M. This study provides insights into the development of data-driven approaches for the optimization of PBF-LB/M process parameters and the prediction of part properties.

Additive manufacturing processes such as the material extrusion of metals (MEX/M) enable the production of complex and functional parts that are not feasible to create through traditional manufacturing methods. However, achieving high-quality MEX/M parts requires significant experimental and financial investments for suitable parameter development. In response, this study explores the application of machine learning (ML) to predict the surface roughness and density in MEX/M components. The various models are trained with experimental data using input parameters such as layer thickness, print velocity, infill, overhang angle, and sinter profile enabling precise predictions of surface roughness and density. The various ML models demonstrate an accuracy of up to 97% after training. In conclusion, this research showcases the potential of ML in enhancing the efficiency in control over component quality during the design phase, addressing challenges in metallic additive manufacturing, and facilitating exact control and optimization of the MEX/M process, especially for complex geometrical structures.

Additive manufacturing processes such as the material extrusion of metals (MEX/M) enable the production of complex and functional parts that are not feasible through traditional manufacturing methods. However, achieving high-quality MEX/M parts require significant experimental and financial investments for suitable parameter development. In response, this study explores the application of machine learning (ML) to predict the surface roughness and density in MEX/M components. The various models are trained with experimental data using input parameters such as layer thickness, print velocity, infill, overhang angle, and sinter profile enabling a precise predictions of surface roughness and density. The various ML models demonstrate an accuracy up to 97 % after training. In conclusion, this research showcases the potential of ML in enhancing the efficiency in control over component quality during the design phase, addressing challenges in metallic additive manufacturing, and facilitating exact control and optimization of the MEX/M process, especially for complex geometrical structures.

The metal additive manufacturing (AM) sector is limited by the scarcity of certified materials. Developing new materials is costly and time-consuming due to the complexity of AM processes where numerous parameters impact the part’s quality. The goal of this study is using simulation and empirical data of AlSi10Mg and an alloy from the Al5000 series (AlMgty80) to train machine learning (ML) models to predict the part density in the laser powder bed fusion of metals. The models are used to predict process windows for a new alloy from the Al7000 series (AlZnty) to identify the potential of recognizing process windows prior to empirical testing. By integrating simulated data, material influences are indirectly accounted for in the dataset. Various ML algorithms were trained and evaluated using root mean squared error (RMSE), R2, and qualitative process map assessments. The results show that ML models based on AlSi10Mg and AlMgty80 data accurately predict the density for the respective materials, achieving an RMSE of around 1% and R2 close to 0.9. Predictions for AlZnty showed larger errors (RMSE 3%–7%, R2 near zero), yet the qualitative evaluation of process maps shows that high-density regions align with experimental data due to the model systematically underestimating the density. Overall, the methodology demonstrates the ability to identify process windows for novel alloys using simulations and ML models based on similar alloys, despite limitations in density prediction accuracy. Future work will involve training models including AlZnty data to iteratively enhance accuracy. Furthermore, limited data availability continues to challenge ML in AM, highlighting the need for extensive databases.