3-day training workshop on Mesh generation and adaptation, Mazurski Raj, Masuria, Poland , Wednesday 24 June - Thursday 26 June 2015.
Day 1: Wednesday 24 June
Introduction to adaptive mesh refinement and mesh movement, Mateusz Gugala(QMUL):
Mesh-related problems in adjoint-based design optimisation, pinciples and approaches for mesh movement (r-refinement), overview of methods for local mesh refinement (h-refinement).
Overview of h-refinement methods and application in finite volume methods, pitfalls of residual-based error estimation in finite-volume methods.
One-shot methods review of existing approaches, Armen Jaworski (WUT) - presentation:
Conceptual and mathematical background of one-shot methods with mesh adaptation; Presentation of different approaches with examples.
Useful tools for mesh adaptation (metric tensor, quality measurement); Local remeshing including how to modify a triangulated surface in keeping a good approximation of the underlying surface
Day 2: Thursday 25 June
Feature-based anisotropic adaptation using global remeshing, Jurek Majewski (WUT)- presentation (140MB):
Introduction to feature-based anisotropic adaptation using global remeshing; Elements of Delaunay unstructured anisotropic grid generation with cell spacing defined by a global metric field; Anisotropic error estimator and its modifications necessary for viscous flow simulations; Results of RANS simulations with adaptation of 2D and 3D flows.
An overview of remeshing method based on metric tensors, Cécile Dobrzynski (INRIA, Bordeaux) :
Application to the generation of a computational mesh from an implicitly defined domain; Anisotropic mesh adaptation for immersed boundary method (involving several adaptation criteria, some of them based on a level-set function, other based on flow features).
Introduction to Uncertainty Quantification and Robust Design, Lukasz Laniewski-Wollk (WUT):
Introduction to Uncertainty Quantification. Common methods: Monte Carlo, Polynomial Chaos (intrusive and non-intrusive). Moments methods. Use of sensitivity information gained from adjoint solver. Relation of design uncertainties to modeling error. Error estimation and quantification. Robust design: Concept, problem statement and basic techniques.
Adjoint-weighted residual mesh refinement for large-scale two and three-dimensional steady-state problems (e.g. drag prediction workshop, high-order workshop cases).
Day 3: Friday 26 June
Adjoint-weighted mesh refinement: steady state, Krzysztof Fidkowski(University of Michigan):
Extension of the adjoint-weighted residual method to unsteady problems, including: unsteady adjoints, incorporating space-time anisotropy, problems on deformable domains, and static/dynamic mesh/order refinement.
Adjoint-weighted mesh refinement: discrete choice sampling, Krzysztof Fidkowski(University of Michigan) :
Anisotropic mesh optimization using discrete-choice sampling, both for structured and unstructured meshes.