Titulo Estágio
Quality-mesh generation for simulation optimisation
Áreas de especialidade
Sistemas Inteligentes
Local do Estágio
DEI
Enquadramento
Modelling, simulation and optimisation are currently essential design tools in virtually all engineering domains. In chemical, civil, electrical, and mechanical engineering, among other disciplines, many important phenomena are described through partial differential equations (PDEs), the solution of which often relies on numerical methods such as finite-difference, finite-element and finite-volume methods (FDM, FEM and FVM, respectively). Furthermore, in settings such as aerospace, automotive, biomedical and marine engineering, joint modelling and simulation of multiple physical phenomena is required in order to allow system optimisation to be performed globally across the various disciplines, but this may be computationally very expensive.
Objetivo
This work is concerned with the study of constrained Delaunay approaches to quality unstructured mesh generation in order to support a tighter integration between FEM/FVM solvers and gradient-free optimisation algorithms such as direct-search methods and evolutionary algorithms. More specifically, incremental approaches to quality-mesh generation are to be implemented and integrated into a FVM solver. Application examples will be derived from the aeronautics domain, in partnership with Embraer, Brazil.
Plano de Trabalhos - Semestre 1
1. Literature review on curve/surface parametrisation, data structures and algorithms for quality-mesh generation, and available mesh-generation software.
2. Prototype implementation of guaranteed-quality triangular mesh generators for two-dimensional domains with curved boundaries.
3. Prototype implementation of incremental re-meshing under domain-boundary perturbations.
4. Preliminary evaluation of the performance of the algorithms developed.
Plano de Trabalhos - Semestre 2
5. Production implementation of the algorithms developed in the first-semester.
6. Implementation of the remaining stages of a two-dimensional flow-solver and application to an airfoil optimisation problem.
7. Validation and benchmarking. Evaluation of the performance benefits afforded by incremental re-meshing.
8. Dissertation writing.
Condições
A five-month research grant is available to support this work.
Orientador
Carlos M. Fonseca (DEI) and Luís Carlos Castro Santos (USP, São Paulo)
cmfonsec@dei.uc.pt 📩