HLRB Project pn56ci

Direct numerical simulation of partially-filled pipe flow

Professur für Hydromechanik, TUM

Direct numerical simulation of partially-filled pipe flow

Professur für Hydromechanik, TUM

Professur für Hydromechanik, TUM

Prof. Dr.-Ing. Michael Manhart

Arcisstraße 21

80333 München

We plan to perform numerical simulations of flows in partially-filled pipes, which are the type of flow typically occurring in waste water channels. Partially filled pipe flow can as well be regarded as idealized flow in a river. It has not been paid a lot of attention to this very basic flow case, so we intend to study the establishing flow patterns and their generation mechanisms. In this project we use the method of Direct Numerical Simulation to investigate the flow physics in detail. The results will be used to improve our current understanding of the flows considered and to provide highly resolved data to the scientific community.Partially-filled pipe flows are found in many engineering applications, nevertheless the knowledge of this flow case is very limited and only coarse measurement data is available. We know that the maximum streamwise velocity is located at around 60 to 70 % of the flow depth (dip phenomenon). Furthermore a secondary flow pattern in the cross-section plane has been detected. But how this flow pattern is generated and why it appears in the way it does can only be answered by examining very detailed information of the turbulent structure. In order to do this, we need to do long-run simulations to gather as much statistics of the flow as necessary to get valid and detailed insights.In our simulations, we solve the Navier-Stokes equations which describe the conservation of mass and momentum on an infinitesimally small volume. We use a Finite-Volume method on a Cartesian grid. Complex geometries are described by a so-called Immersed Boundary condition. A local grid refinement can be achieved by hierarchically arranged overlapping zonal grids. By these techniques, an efficient solver has been developed that is highly scalable and flexible in terms of grid resolution and geometrical configuration. The code has been fully parallelized and applied up to more than several thousand cores so far.

Impressum, Conny Wendler