Browsing by Author "Zirkel, Armin"

Now showing 1 - 1 of 1

Open Access
Numerical investigation of the turbulence mass transport during the mixing of a stable stratification with a free jet
(2011) Zirkel, Armin; Laurien, Eckart (Prof. Dr.-Ing.)
The safety of present and future light-water reactors is a major concern of electrical utilities, politics and research institutes. During a severe accident, hydrogen can be produced by a chemical reaction between the Zircaloy cladding and water and escape into the containment through a leak in the primary circuit. The prediction of the mass transport of hydrogen is vital for an optimised positioning of countermeasures like recombiners. It is possible that a stable stratification of hydrogen and air occurs, due to the different densities of those fluids. This stratification can be mixed with a free jet. This mixing is characterised by the time dependency of the flow, sharp velocity and density gradients as well as the non-isotropy of Reynolds stresses and turbulent mass fluxes. With the use of a Reynolds stress turbulence model, the non-isotropic Reynolds stresses can be simulated. A similar approach is theoretically possible for the turbulent mass fluxes, but only the isotropic eddy diffusivity model is currently available in state-of-the-art cfd-software. The shortcomings of the eddy diffusivity model to simulate the turbulent mass flux are investigated, as well as improvements with the use of a non-isotropic model. Because of the difficulties to get experimental data of flows in real containments, the THAI experimental facility was created to get experimental data for flows in large buildings. The experiments are performed by Becker Technologies. The analysis is using the experimental data of the TH20 experiment as the reference case. For safety reasons the used light gas for the experiments is helium instead of hydrogen. Due to the rotational symmetry of the geometry as well as the boundary conditions, two-dimensional simulations are performed. The grid was built following the best practice guidelines to ensure sufficient grid quality. Several simulations were carried out to investigate the numerical error caused by spatial and time discretisation. An analysis of the currently available turbulence models shows that the eddy diffusivity model yields a poor agreement with the experimental data. This is true regardless of the used model to calculate the Reynolds stresses. Due to the time dependency of the mixing, a comparison between different simulations is not a trivial task with the exception of the time dependent helium concentration on different measuring points. Therefore a theoretical, statistically steady, two-dimensional test case was designed to enable direct comparisons of different models. With steady state results, an investigation of velocities and turbulent values, especially the turbulent mass fluxes, is possible without the need to consider the different mixing progress of a model at a given time. A large eddy simulation is performed as reference for the investigation of the non-isotropic turbulence scalar flux model, TSF-model for short. The new TSF-model is then used to simulate the transient mixing of the TH20 experiments. Results obtained with the new model are showing an improved mixing.