I³MS - Sayadi Seminar

Location: Room 115, AICES, Rogowski

Prof. Dr. Taraneh Sayadi - Adjoint-based Viscous Stress Reduction Using Dynamic Roughness Elements

Department of Aerospace Engineering, University of Illinois

Abstract

Adjoint and direct operators are widely used in theoretical and applied studies related to non-linear optimization and flow control. We have adopted the approach of Fosas de Pando et al. (2012) in forming the discretized adjoint operators to transform an existing compressible flow solver, suitable for large-scale direct and large eddy simulations for the purpose of flow control investigations. Furthermore, the compressible solver is selected in order to showcase the flexibility and performance of the developed framework, which is particularly efficient, since the linearized operators are computed simply by using the local differentiation technique, without explicitly forming the resulting matrices for both forward and adjoint operators. Using the described framework we investigate the effect of the initial condition on the spatial distribution of roughness elements and their variation in time in order to reduce viscous stress on a defined region of the wall. Roughness elements have a great impact on laminar boundary layers, where a small amount of surface deflection causes transition to turbulence. The influence of roughness is also relevant to the study of turbulence, where the introduction of new length scales effects the prediction of turbulent structures. In addition, roughness elements can serve as controllers in both the laminar and turbulent regime to, for example, reduce the skin friction or drag. The roughness elements considered here are of the ``dynamic" type, varying both in space and time, which allows control over the spatial distribution of the roughness but also the inherent timescales of the flow. Dynamic roughness is modeled here using linearized boundary conditions previously introduced by McKeon (2008), where the no-slip and impermeability boundary conditions are replaced by streamwise and wall-normal distributions at the wall.