| dc.description.abstract | State-of-the-art metal additive manufacturing (AM), mainly laser powder-fed AM (LPF-AM) and laser powder-bed AM (LPB-AM), has been used to produce high-quality, complex-shaped, and end-user metallic parts. To achieve desirable dimensional, microstructural and mechanical features of as-built components through fast process optimization or feedback-control-based adaptive processing adjustment, high fidelity and calculation-efficient processing model is urgently needed. The thesis research has been motivated by the need for time-efficient process models of both LPF-AM and LPB-AM. To this end, comprehensive accelerated models for these processes have been built and experimentally verified.
The comprehensive process model of LPF-AM was built by an innovative analytical approach. Firstly, a mathematical module that couples laser heat flux and powder mass flow was developed, while considering the attenuated laser intensity distribution and the heated powder spatial distribution. Correspondingly, a powder catchment module was built in terms of a three-dimensional (3D) melt pool shape and powder stream spatial distribution. Integrating these physical modules into the thermal modeling, a coupled heat and mass comprehensive model of the LPF-AM process was achieved. Experimental depositions of Inconel 625 proves the model’s high accuracy in predicting as-built deposits’ geometry (a maximum error of ~6.2% for clad width, ~7.8% for clad height) and powder catchment efficiency (maximum error of less than ~6.8%). It was found that the predicted real-time melt pool peak temperatures match well with the experimental results in Stainless Steel (SS) 316L deposition. The calculated micro-hardness has a maximum prediction error of ~16.2% compared with the measured results. The predicted microstructural evolutions show reasonable agreement with the experimental observations for both SS 316L and Inconel 625 depositions. Moreover, sensitivity analysis shows that the powder feed rate has the largest positive effect on the clad height.
The time-efficient process model of LPB-AM was achieved by a novel analytical approach that couples the critical physics of the process, while considering the volume shrinkage and the melting regime. The proposed model can perform a time-efficient prediction of the localized-transient thermal field, melt pool temperature distribution, and multi-track overlapping dimension. The powder bed was treated as a homogeneous medium with effective thermophysical properties derived from the randomly packed rain model. In addition, different melting regimes of the LPB-AM process were considered in the built model. A 3D heat source model with variant penetration depths, together with the varying melting regimes, was utilized to solve the transient thermal field. Moreover, the density and top surface roughness of the final parts were empirically modeled using response surface regression under a Box-Behnken design. Subsequently, the mechanical properties of the part and the in-situ build rates were simultaneously optimized by combining the built analytical models and empirical models with employing a multi-objective genetic algorithm. Experimental results with SS 17-4PH show that the predicted melt pool dimensions have a high degree of accuracy under steady melting regimes, with a maximum of ~14% error for the width prediction and ~15% error for the depth calculation. Furthermore, an optimized parameter solution set was provided based on the built 3D Pareto fronts.
The built models’ calculation time for the localized-transient characteristics for LPF-AM and LPB-AM are ~4 ms and ~1.2 ms, respectively. These findings confirm the great potential of the present research to be used for fast process optimization and in-situ process control.
In addition, a new magnetic concentration approach designed with various configurations was explored. This approach is designed to focus the diverging metal particles in the gas-powder stream of LPF-AM, thereby improving powder catchment and deposition accuracy. It was shown that the proposed permanent-magnet-based configurations may not be suitable for concentrating submillimeter-sized particles. However, an additional development, a doublet-electromagnet-quadrupoles-based configuration with high frequency, may be capable of concentrating the non-ferrous metallic particles (e.g., aluminum particle) with a radius of r_p≥150 μm. | en |
