AVS 55th International Symposium & Exhibition | |
Vacuum Technology | Tuesday Sessions |
Session VT-TuM |
Session: | Vacuum Pumping Technologies, Large Vacuum Systems, Vacuum Modeling |
Presenter: | M.E. Roos, TNO, the Netherlands |
Authors: | M.E. Roos, TNO, the Netherlands R. Versluis, TNO, the Netherlands L. Thielen, TNO, the Netherlands |
Correspondent: | Click to Email |
Three flow regimes can be distinguished (ordered in decreasing pressure): the continuum regime, the transitional or rarefied regime and the molecular regime. Modeling of the continuum flow regime is covered by Computational Fluid Dynamics. The fluidum is regarded as a continuum and the discrete character of molecules is not taken into account. For the molecular regime network modeling can be applied. This is the equivalent of electric network modeling applied to flows. It has always been a problem to model the transitional regime. Transitional flow conditions appear in a wide range of applications where vacuum is applied or small dimensions prevail. Some examples: the design of precision instruments (newest generation lithography machines), physical and chemical vapour deposition processes (fabrication of solar cells or coating of metals) or space technology (plume from a rocket nozzle). TNO has developed in cooperation with the Delft University of Technology a tool to model the transitional regime. This tool is based on the Direct Simulation Monte Carlo (DSMC) method. In this model flow properties are determined by simulating the movement and collisions of molecules. A mixture of different gases can be modeled as well as chemical reactions between gases and on surfaces. The DSMC model is implemented in the parallelized numerical 3D flow solver CVD-X (developed in-house by TNO). The DSMC model is also valid in the molecular and continuum regime. The development of the model is an ongoing process. Recently the model has been extended to be able to simulate moving parts in the transition regime. In our method moving parts are implemented as special boundaries which are moving through the domain. A moving boundary has all properties of a normal stationary boundary (like temperature and accommmodation coefficient). In addition the exact time and position in space of the interaction between molecules and a moving boundary is determined throughout time. Multiple collisions of molecules in one time step with moving and non-moving surfaces are taken into account. This method makes it e.g. possible to model the spinning rotors of a turbomolecular pump and optimize the efficiency. We are performing DSMC simulations on a turbomolecular pump using this method. The results of these simulations and the validation with experimental data are presented.