Invited plenary speaker: Sergey V. Utyuzhnikov, the University of Manchester, UK

Non-Overlapping Domain Decomposition for Near-Wall Turbulence
Modelling: State-of-the-Art and Perspectives

Near-wall turbulence modeling remains to be a hot topic in computational fluid mechanics. As well-known, the resolution of a thin near-wall region can take most computational time even with the use of RANS models. Moreover, the problem becomes more complicated in the case of Large-Eddy-Simulation (LES). Most engineering approaches are based on wall-functions. In these approaches the effect of the wall is only realized via off-wall Dirichlet semi-empirical boundary conditions. Inevitably, these techniques are approximate although computationally efficient. An alternative approach based on non-overlapping domain decomposition (NDD) has proved to be efficient even for flows with separation. Since the technique exploits a thin-layer model in the inner region, it represents a trade-off between the accuracy and computational time subject to location of the interface boundary. As was shown, the approach reduces the computational time by one order of magnitude while retaining a relatively good accuracy.

In the talk new advances in the development of NDD are discussed as well as its potential to be extended for LES. The idea of an exact NDD looks attractive for a number of applications. It has been recently realized for a 1D problem yet. In this approach the original (approximate) NDD is used to effectively invert a preconditioning operator. As has been shown, the computational time is reduced by one order of magnitude while retaining the solution practically accurate. The convergence of the algorithm is strictly proven for the advection-diffusion equation. Another direction of the NDD development is related to the interface boundary conditions (IBC) which play a crucial role in the efficiency of the algorithm. As was earlier shown for a model equation, in the case of an essentially unsteady solution, the IBC must be nonlocal in time. In the talk, such IBCs are derived for the Navier-Stokes equations and their properties are analysed. Potentially, they can be applied to hybrid RANS-LES models.

About the author:

Professor Sergey Utyuzhnikov (FIMA) is a Reader in Computational Mechanics at the University of Manchester and life-time Professor of Computational Mathematics (Russian Ministry of Education). He is also an Adjunct Professor at the Moscow Institute of Physics and Technology where he gained his PhD in 1986 and DSc in 1997. He has a broad experience in CFD, active sound control, and engineering optimisation acquired over the course of about 30 years. He has given invited talks at Los Alamos National Laboratory, the Royal Society (Kavli Centre), Schlumberger (Cambridge), CARDC (China), CAAA (China), EDF/CNRS (Paris), Institut Henri Poincare (Paris), ABB, ALSTOM Ltd (Switzerland), VKI (NATO, Brussels), and numerous international conferences. He is the author or co-author of over 100 peer-reviewed journal publications and many conference papers. In recent years, he has obtained fundamental results related to the generalized potentials theory, non-overlapping domain decomposition, active sound control and nonlinear multi-objective optimisation. He is a co-author (with G.A.Tirskiy) of book “Hypersonic Aerodynamics and Heat Transfer” (Begell House: ISBN 978-1-56700-309-3) published in 2014.

Professor Utyuzhnikov website can be found under this link