Condensation-induced water hammer, which appears when horizontal pipe filled with hot vapor is being slowly flooded with cold liquid, is known to be very stochastic phenomena. Time and position of the slug that is followed by the rapid condensation of the bubble behind the slug, is very sensitive to the minor changes in the experimental setup and consequently: results of the simulations are also very sensitive to the minor changes of the physical and numerical parameters in the model. Selected condensation-induced water hammer experiments performed on PMK-2 (AEKI, Hungary) device were numerically modeled with three-dimensional two-fluid model of computer codes NEPTUNE CFD. In most of the experimental cases, slow flooding of the pipe was abruptly interrupted by a strong slugging and water hammer, while in the experimental runs selected in the present work, the transition from the stratified into the slug flow was not accompanied by the water hammer pressure peak. That makes these cases more suitable tests for evaluation of the various condensation models in the horizontally stratified flows and puts them in the range of the available CFD (Computational Fluid Dynamics) codes. The key models for successful simulation appear to be the condensation model of the hot vapor on the cold liquid and the interfacial momentum transfer model. The surface renewal types of condensation correlations, developed for condensation in the stratified flows, were used in the simulations and were applied also in the regions of the slug flow. The CFD simulations quantitatively capture the main phenomena of the experiments, while the stochastic nature of the particular condensation-induced water hammer experiments does not allow detailed prediction of the time and position of the slug formation in the pipe. It is shown that even the selected experiments without water hammer present a tough test for the applied CFD codes.
展开▼
