REDAFLOW · REconstruction-based DAta-assisted frameworks for turbulent reacting FLOWs.
Horizon 2020 — Marie Skłodowska-Curie Actions
- Duration
- 2021-09-01 → 2024-08-31
- EU contribution
- €295,062
- Participants
- 1
- Scheme
- MSCA-IF
Lines connect the coordinator with its partners.
Results in brief
REconstruction-based DAta-assisted frameworks for turbulent reacting FLOWs.
The fellow Dr. Zacharias M. Nikolaou conducted a project to develop a generalised and computationally efficient modelling framework for simulating turbulent and reacting flows by employing two promising methods: signal reconstruction/deconvolution and machine learning. The work was carried out at the CORIA lab, a joint research unit between CNRS, the University of Rouen, and INSA Rouen, under the supervision of Prof. Luc Vervisch. The proposed modelling frameworks had to enable complex reacting flow simulations to be conducted under a unified modelling framework. This had to allow robust and accurate simulations of challenging flows in industrial engineering devices, thereby improving the modelling capabilities for the design of the next generation of greener energy technologies such as gas-turbine combustors, reformers, etc.
Data: CORDIS, © European Union
Project objective
REDAFLOW aims to develop a generalised, computationally efficient and scalable modelling framework for simulating turbulent and reacting flows, aimed at the latest and emerging high-performance computing architectures. Current state of the art classic modelling approaches developed from simplifying assumptions (in-compressible, self-similar, non-reacting) limit the generality and application domain of computational fluid dynamic simulations which is becoming the workhorse in industry for virtual prototyping. At the same time, a large number of flow-dependent and reaction-dependent model parameters limit the predictive ability and robustness of numerical simulations. The novelty of the proposed framework is twofold: reconstruction/deconvolution will be employed for modelling in a generalised and parameter-free framework unresolved terms in the governing equations while machine-learning will be employed to model the chemical kinetics including detailed-chemistry effects. The necessary filtering and interpolation schemes as well as all the deconvolution algorithms and chemistry neural network libraries will be developed in-house in stand-alone libraries, and optimised for use with state of the art parallelisation libraries. The proposed framework is expected to reduce the computational time required for tabulation-based reacting flow simulations, improve the simulation predictions, and allow a wider range of practical flows to be simulated under a generalised framework, irrespective of the flow or reaction regime. The tools and libraries developed are expected to attract the interest of a range of industries (chemical, automotive, aerospace, software, consulting) where simulation is the main tool for developing improved processes and designs for a wide range of engineering devices.
Original text from CORDIS.
Participants
- INSTITUT NATIONAL DES SCIENCES APPLIQUEES DE ROUEN · Saint-Etienne-du-RouvrayCoordinatorFrance
Links
Data: CORDIS, © European Union