The sound which is produced when a water drop impacts into a water pool is a prominent example for acoustics produced by multiphase flow. In this work the feasibility of numerical methods for simulating this challenging test case is evaluated. First the multiphase flow needs to produce the correct physical mechanisms, e.g. the bubble entrapment. For this an in-house block-structured finite-volume solver with the volume-of-fluid method is used. For the curvature computation a standard finite difference method within the continuum surface force model is employed, including some necessary improvements. A high resolution in space and time is essential and therefore the method is parallelized by domain decomposition. The acoustic part is simulated with the linearized Euler equations which are valid in each phase but need to be adapted in the interface region. The results are compared with numerical and experimental data. It is shown, that the methods are suitable for simple test cases. A coupled drop impact test case corresponds with equivalent experiments until the drop detachment. The acoustic pressure shows a significant rise in the vicinity of the bubble detachment within both phases. However, an oscillation of the cavity bottom can not be observed in the multiphase neither in the acoustic outputs of the airborne signal.

The accurate prediction of the curvature of fluid-fluid interfaces is crucial for appropriately modeling the surface forces when computing two-phase flows with immiscible fluids. The volume of fluid (VOF) method is often used for these computations to specify the different fluids and the interface by the so-called volume fraction field. In this study, a deep artificial neural network is trained to predict the interface curvature from the volume fraction values. This approach is investigated within an algebraic VOF framework. A rudimentary interface resharpening algorithm is introduced for the input stencils to enhance the accuracy and robustness when the interface can not be captured entirely sharp. The performance of different neural network architectures is evaluated by generic test data and the computation of two oscillating droplet flow configurations.