Cerfacs Enter the world of high performance ...

@ARTICLE{AR-CFD-20-18, author = {Roy, P. and Jofre, L and Jouhaud, J.-C. and Cuenot, B. }, title = {Versatile Sequential Sampling Algorithm using Kernel Density Estimation}, year = {2020}, doi = {10.1016/j.ejor.2019.11.070}, journal = {European Journal of Operational Research}, abstract = {Understanding the physical mechanisms governing scientific and engineering systems requires performing experiments. Therefore, the construction of the Design of Experiments (DoE) is paramount for the successful inference of the intrinsic behavior of such systems. There is a vast literature on one-shot designs such as low discrepancy sequences and Latin Hypercube Sampling (LHS). However, in a sensitivity analysis context, an important property is the stochasticity of the DoE which is partially addressed by these methods. This work proposes a new stochastic, iterative DoE – named KDOE – based on a modified Kernel Density Estimation (KDE). It is a two-step process: (i) candidate samples are generated using Markov Chain Monte Carlo (MCMC) based on KDE, and (ii) one of them is selected based on some metric. The performance of the method is assessed by means of the C2-discrepancy space-filling criterion. KDOE appears to be as performant as classical one-shot methods in low dimensions, while it presents increased performance for high-dimensional parameter spaces. It is a versatile method which offers an alternative to classical methods and, at the same time, is easy to implement and offers customization based on the objective of the DoE.}, keywords = {Stochastic processes, Design of experiments, Discrepancy, Optimal design, Uncertainty quantification}}

@ARTICLE{AR-CFD-20-19, author = {Campet, R. and Roy, P. and Cuenot, B. and Riber, E. and Jouhaud, J.-C. }, title = {Design Optimization of an Heat Exchanger using Gaussian Process}, year = {2020}, number = {April }, volume = {150}, pages = {119264}, doi = {10.1016/j.ijheatmasstransfer.2019.119264}, journal = {International Journal of Heat and Mass Transfer}, abstract = {The objective of this work is to optimize the internal shape of a single-started helically ribbed heat exchanger. Large Eddy Simulation (LES) is used to simulate the turbulent flow in a wall-resolved periodic channel configuration, heated via a uniform heat flux at the wall. In order to enhance the heat exchange with the flow, the inner surface of the channel features rounded rib. This however increases the pressure loss, and an optimum shape of the rib is to be found. The rib pitch and height as well as rib discontinuities are the geometrical parameters to optimize, allowing a wide variety of inner wall roughness. To limit the number of LES, the optimization procedure is based on a surrogate model constructed from Gaussian Process Regression and adaptive resampling with the Efficient Global Optimization (EGO) method [1]. The optimization consists in the maximization of the cost function proposed by Webb and Eckert [2], which aims at maximizing the heat transfer efficiency for similar pumping power. Results show that a rib induced swirling motion in the near wall region significantly decreases the heat transfer efficiency, leading to an optimum roughness shape featuring large and multiple discontinuities. Moreover, the efficiency of helically dimpled tubes is also found sensitive to the shape of the transitions between the discontinuous parts of the rib. Smoother transitions lead to lower pressure loss but also to lower heat transfer due to smaller recirculation zones.}, keywords = {Heat exchanger, Geometrical design, Optimization, LES, Ribbed tube, Turbulent channel flow}, url = {http://www.sciencedirect.com/science/article/pii/S0017931019348495}}

@ARTICLE{AR-CFD-20-25, author = {Paulhiac, D. and Cuenot, B. and Riber, E. and Esclapez, L. and Richard, S. }, title = {Analysis of the spray flame structure in a lab-scale burner using Large Eddy Simulation and Discrete Particle Simulation}, year = {2020}, number = {february}, volume = {212}, pages = {25-38}, doi = {10.1016/j.combustflame.2019.10.013}, journal = {Combustion and Flame}, abstract = {The numerical study of an academic lab-scale spray burner using Large Eddy Simulation coupled with a Discrete Particle Simulation is presented. The objectives are first, to validate current turbulent combustion modeling approach for two-phase flames, and second, to bring new insight on two-phase flame structure in a complex flow, representative of real configurations. The comparison with the experiment shows a good quantitative prediction of the velocity field of the gas and the liquid phases, in both non-reacting and reacting cases. Experimental and numerical results of the spray flame are also in good agreement. The detailed study of the interaction between the flame front and the droplets shows that both single droplet and group combustion regimes occur in the present configuration. These regimes are investigated from the numerical and physical points of view, highlighting the necessity to further investigate their possible importance for the modeling of two-phase combustion.}, url = {https://doi.org/10.1016/j.combustflame.2019.10.013}}

@ARTICLE{AR-CFD-20-23, author = {Harnieh, M. and Thomas, M. and Bizzari, R. and Dombard, J. and Duchaine, F. and Gicquel, L.Y.M. }, title = {Assessment of a coolant injection model on cooled high-pressure vanes with Large-Eddy Simulation}, year = {2020}, number = {1}, volume = {104}, pages = {643–672}, doi = {10.1007/s10494-019-00091-3}, journal = {Flow Turbulence and Combustion}, abstract = {The high-pressure turbine blades are the components of the aero-engines which are the most exposed to extreme thermal conditions. To alleviate this issue, the blades are equipped with cooling systems to ensure long-term operation. However, the accurate prediction of the blade temperature and the design of the cooling system in an industrial context still remains a major challenge. Potential improvement is foreseen with Large-Eddy Simulation (LES) which is well suited to predict turbulent flows in such complex systems. Nonetheless, performing LES of a real cooled high-pressure turbine still remains expensive. To alleviate the issues of CPU cost, a cooling model recently developed in the context of combustion chamber liners is assessed in the context of blade cooling. This model was initially designed to mimic coolant jets injected at the wall surface and does not require to mesh the cooling pipes leading to a significant reduction in the CPU cost. The applicability of the mod el is here evaluated on the cooled Nozzle Guide Vanes (NGV) of the Full Aerothermal Combustor Turbine interactiOns Research (FACTOR) test rig. To do so, a hole modeled LES using the cooling model is compared to a hole meshed LES. Results show that both simulations yield very similar results confirming the capability of the approach to predict the adiabatic film effectiveness. Advanced post-processing and analyses of the coolant mass fraction profiles show that the turbulent mixing between the coolant and hot flows is however reduced with the model. This finding is confirmed by the turbulent map levels which are lower in the modeled approach. Potential improvements are hence proposed to increase the accuracy of such methods.}, keywords = {Large-eddy simulation, Turbomachinery, Film cooling, Modelling}}

@ARTICLE{AR-CFD-20-24, author = {de Laborderie, J. and Duchaine, F. and Gicquel, L.Y.M. and Moreau, S. }, title = {Wall-modeled Large-Eddy Simulations of a Multistage High-Pressure Compressor}, year = {2020}, doi = {10.1007/s10494-019-00094-0}, journal = {Flow Turbulence and Combustion}, abstract = {This study aims at evaluating the feasibility and the accuracy of a Wall-Modeled Large Eddy Simulation of an actual aeronautical multistage axial high-pressure compressor. The computational domain is composed of 37 blades and a geometrically complex recirculating cavity. The numerical method TurboAVBP is able to handle such a technical challenge thanks to its unstructured and massively parallel approach as well as dedicated rotor-stator interface treatments. The influence of grid resolution, from less than 100 millions to more than 1 billion of cells, is particularly evaluated. The intermediate grid correctly predicts the global aerodynamic performances up to the lowest mass flow rate regime. In comparing with time-resolved measurements, the finest grid is shown to accurately predict flow within rotors, especially in their tip regions that are critical for performances and stability of the whole compressor.}, supplementaryMaterial = {https://link.springer.com/article/10.1007/s10494-019-00094-0}}