Modal analysis and dimensionality reduction

Description: Flows often present themselves as complex multi-scale phenomena. Nonetheless, often it is possible to identify certain patterns, called coherent structures, that preserve spatial and temporal correlation. Such structures allow us to segregate the flow into distinct dynamic regions, thus forming the keystone to understand how disturbances evolve in the flow and also the response of the flow to external forcing (relevant to control). The aim of this research effort is to combine data processing techniques (POD, DMD, SINDy, etc) and theoretical tools (such as resolvent analysis) to identify relevant flow structures and build simplified models that represent the most relevant aspect of the flow dynamics.

Potential supervisors: Dr. André Fernando de Castro da Silva; Dr. André V. G. Cavalieri

For details, contact: andref@ita.br; andre@ita.br

Flow estimation and control

Description: The end goal of flow estimation is to be able to estimate/reconstruct unobservable flow quantities from available measurements. Such task can be regarded as a way to enhance experimental data or as an intermediary step to closed-loop flow control, which can be used to engineer the stability and response of the fluid system to disturbances. In this research topic, we aim to develop estimators (via Kalman filters or empirical transfer functions) and robust strategies to flow control (control law, sensor/forcing placement, etc).

Potential supervisors: Dr. André Fernando de Castro da Silva; Dr. André V. G. Cavalieri

For details, contact: andref@ita.br; andre@ita.br

Numerical Methods

High-fidelity simulation and analysis of turbulent flows

Description: Transitional and turbulent flows are amongst the most complex topics of interest in engineering. Today certain advanced simulation techniques exist that are excellent tools in the study of such flows. The so-called under-resolved DNS approach via spectral/hp methods offers high-fidelity results at reasonable a computational cost. This approach is however very modern and thus its suitability is being investigated actively. The present research topic aims at studying the quality of under-resolved DNS results obtained with spectral/hp methods.

Potential supervisors: Dr. Rodrigo Costa Moura; Dr. André Fernando de Castro da Silva

For details, contact: moura@ita.br; andref@ita.br

Automated mesh generation and adaptation

Description: The generation of suitable computational grids for practical CFD applications is in general a tedious and time-consuming task. Employing mesh refinement only where necessary is also required for efficient use of computational resources. This research aims at developing algorithms for facilitated mesh generation with integrated grid refinement capabilities for flows featuring rich physics (shocks, turbulence, etc) around geometries of varied complexity.

Potential supervisors: Dr. Rodrigo Costa Moura; Dr. Vinicius Malatesta

For details, contact: moura@ita.br; malatesta@ita.br