Skip to Main content Skip to Navigation
New interface
Theses

Development of Multigrid Sequential Data Assimilation Strategies for Complex Unsteady Flows

Abstract : The analysis and control of complex high-Reynolds-number flows of industrial and practical interest is one of the most distinctive open challenges that the scientific community must face for fluid mechanics applications in the coming decades. Modelling bias and uncertainty may strongly affect the predictive capabilities of both numerical simulations and experimental measurements. Under this perspective, data-driven tools from Data Assimilation, and, in particular, sequential tools such as the ensemble Kalman filter (EnKF), have been recently used to obtain a precise estimation of the physical flow state accounting for bias or uncertainty coming from real conditions in the performance of the investigative tool. A newly developed sequential Data Assimilation algorithm, combining multi-grid aspects and the ensemble Kalman Filter, is presented in this PhD-study. The so-called MGEnKF algorithm (Multi-Grid Ensemble Kalman Filter) exploits physical states obtained on multiple grids of different resolution to perform the state estimation and parametric optimization using EnKF procedures. More precisely, an ensemble of low-fidelity simulations of the flow is run on a coarse grid level together with a single higher-resolution simulation on the finest mesh level. The state estimation obtained at the coarse level and the associated ensemble statistics are used to filter the finest mesh solution and optimise a set of parameters describing the model (boundary conditions, model parameters…). This procedure allows to i) reduce the computational costs of the EnKF and ii) ensure the conservativity and the smoothness of the final solution.The assessment of the method is performed via the analysis of one-dimensional, two-dimensional and three-dimensional test cases, using different models of increasing complexity. The results show that the MGEnKF can successfully update the state of a system with the available observations to increase the global accuracy of the model. In addition, the parametric description of the numerical problem (in terms of prescribed boundary conditions, turbulence closures...) can be adequately optimized taking into account the different mesh resolutions employed in the algorithm. The MGEnKF opens interesting perspectives for potential application to in-streaming Data Assimilation techniques.
Document type :
Theses
Complete list of metadata

https://theses.hal.science/tel-03813677
Contributor : ABES STAR :  Contact
Submitted on : Thursday, October 13, 2022 - 2:51:42 PM
Last modification on : Friday, October 21, 2022 - 3:46:07 AM

File

2022ESMA0013_moldovan.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-03813677, version 1

Collections

Citation

Gabriel Ionut Moldovan. Development of Multigrid Sequential Data Assimilation Strategies for Complex Unsteady Flows. Other. ISAE-ENSMA Ecole Nationale Supérieure de Mécanique et d'Aérotechique - Poitiers, 2022. English. ⟨NNT : 2022ESMA0013⟩. ⟨tel-03813677⟩

Share

Metrics

Record views

0

Files downloads

0