For now, this page is assumed to be a static description of your courses. You can convert it to a collection similar to _projects/ so that you can have a dedicated page for each course.
Organize your courses by years, topics, or universities, however you like!
2026
Numerical metallurgy: a modeling framework for gas atomization of molten metal
{Dennis Pieter Leonard} Thuy
Mar 2026
G. Finotella, N. Deens (Promotors) Joris J.C. Remmers (Co-promotor)
In a world with an increasingly higher demand for circular production processes, additive manufacturing (AM) of metal objects is an attractive alternative to traditional manufacturing techniques. Bed-based AM techniques allow the flexible design and manufacturing of objects from metal powders with significantly reduced amounts of waste material. Moreover, a fully circular production process can be achieved if scrap metal can be recycled into metal powder suitable for AM. The gas atomization of molten metals is a production process for fine metal powders which are suitable for AM processing. However, this complex process needs to be significantly better understood for it to consistently meet the powder requirements posed by AM. In gas atomization, a continuous flow of molten metal is ruptured by high-velocity expanding gas jets. This results in a dense spray, in which the melt breaks up into fine droplets and finally solidifies into powder particles. Due to the challenging operating conditions of the gas atomization process, the process is well-suited to be studied numerically using computational fluid dynamics (CFD). While numerical models of different stages of the atomization process exist, the effect of modeling assumptions on the prediction of the final particle size distribution (PSD) of the powder is unclear. Simulation of the gas atomization process consists of two parts. The primary breakup involves initial fragmentation of the melt and requires computationally intensive modeling techniques that resolve the interface in detail. The secondary breakup stage comprises further breakup and solidification of the melt, and uses more computationally efficient Euler-Lagrangian methods. A numerical framework for both breakup stages, as well as their coupling, is presented, validated and applied in this work. The high surface tension of liquid metals and the high interfacial curvature of micrometer-sized droplets in the primary breakup impose significant challenges for the modeling of the interface with the volume-of-fluid (VOF) method. A tensile-force (TF) method was implemented to accurately model surface tension forces. The TF method is combined with a pressure-jump correction (PJC) to reduce spurious velocities. The method is combined with large eddy simulations (LES) and adaptive mesh refinement (AMR) to accurately model the formation of droplets in primary breakup in the gas atomization process. It is found that the use of symmetry boundary conditions to reduce the required domain size for primary breakup simulations significantly affects the spray behavior compared to simulations of the full spray domain. Introduction of the symmetry boundary conditions enhances the growth of a liquid core in the spray and reduces rotational symmetry. The growth of a liquid core does not occur in simulations of the full domain. Droplet sizes after primary breakup are found to decrease significantly with increasing grid resolution. With the increase of resolution, the formation of waves on the melt surface near the nozzle tip is observed. These waves are stretched into ligaments before they break up into small droplets. The bag breakup mode is observed for a number of droplets, at Weber numbers corresponding to the range in which this breakup mode is expected to occur. These breakup mechanisms are not observed at lower resolutions. The morphology of the spray cone is not significantly affected by the grid resolution. In the Euler-Lagrangian modeling approach for the secondary breakup stage, additional numerical models are required to represent the further fragmentation of metal droplets as well as their solidification. Once a droplet is solidified, it can no longer break up. The use of standard breakup models in combination with a parcel approach to model breakup of liquid metals is validated. The rapid solidification and recalescence of metal droplets is included in the model. The competition between breakup and solidification and its influence on the final PSD of the powder is assessed based on their respective time scales. Simulations of the secondary breakup stage show that the time scale of liquid breakup is significantly shorter than that of solidification. Solidification is not likely to impact the final PSD, since typically the breakup process is completed before solidification of the droplets takes place. The droplet properties obtained in the primary breakup simulations are used as the initial condition for the secondary breakup stage. Therefore, the transfer of data from the VOF to the Euler-Lagrangian framework is essential. A statistical coupling approach (SCA) is established for the systematic transfer of the relevant droplet properties. The effect of droplet data representation in this coupling procedure on the final PSD is investigated. Droplet properties are represented by either a mean value, or a continuous or discrete distribution, at the start of the secondary breakup simulations. The appropriate location of the coupling plane is selected based on droplet size and shape criteria. Input data for secondary breakup simulations is then generated from the statistical representation of primary breakup data. The secondary breakup behavior and resulting PSD is sensitive to the mono- or polydispersity of the input data. Simulations with mean value input data result in a narrower PSD compared to simulations with input generated from continuous or discrete distributions. No significant difference is found between the simulations with input from continuous or discrete distributions. The developed framework is comprehensive and gives an overview of the sensitivity of the different breakup stages to the modeling choices. In future research, experimental validation of the framework at industrial operating conditions is paramount. The numerical methods for primary breakup can be used to research the interaction of the melt with turbulent structures and the resulting breakup mechanisms. This insight allows for the development of computationally cheaper models of the gas atomization process, which are applicable on an industrial scale.
2025
Sintered conductive materials under mechanical loading: Numerical insights into electro-mechanical performance and failure
