An immersed boundary method, similar to the one used by Kim et al. [1] was developed to implement a varying wall topography into an existing DNS code for pipe flow. Validation using a semi-analytical solution and numerical results showed that the method yields accurate results for laminar flow. Four simulations for turbulent flow where performed, each with a different wall geometry. Wall topographies varying in the axial direction and topographies varying in the azimuthal direction have been considered. Time-averaged as well as instantaneous results are presented for the different geometries. The results for turbulent flow are consistent with the expected physical behaviour. They confirm the hypothesis that flow in the outer layer is largely unaffected by the wall topography.

This content is only available via PDF.
You do not currently have access to this content.