Damien Poitou

Analysis of Radiation Modeling for Turbulent Combustion: Development of a Methodology to Couple Turbulent Combustion and Radiative Heat Transfer in LES

Radiation exchanges must be taken into account to improve large eddy simulation (LES) prediction of turbulent combustion, in particular, for wall heat fluxes. Because of its interaction with turbulence and its impact on the formation of polluting species, unsteady coupled calculations are required. This work constitutes a first step toward coupled LES-radiation simulations, selecting the optimal methodology based on systematic comparisons of accuracy and CPU cost. Radiation is solved with the discrete ordinate method (DOM) and different spectral models. To reach the best compromise between accuracy and CPU time, the performance of various spectral models and discretizations (angular, temporal, and spatial) is studied. It is shown that the use of a global spectral model combined with a mesh coarsening (compared with the LES mesh) and a minimal coupling frequency N it allows to compute one radiative solution faster than N it LES iterations while keeping a good accuracy. It also appears that the impact on accuracy of the angular discretization in the DOM is very small compared with the impact of the spectral model. The determined optimal methodology may be used to perform unsteady coupled calculations of turbulent combustion with radiation.

Analysis of high performance conjugate heat transfer with the OpenPALM coupler

In many communities such as climate science or industrial design, to solve complex coupled problems with high fidelity external coupling of legacy solvers puts a lot of pressure on the tool used for the coupling. The precision of such predictions not only largely depends on simulation resolutions and the use of huge meshes but also on high performance computing to reduce restitution times. In this context, the current work aims at studying the scalability of code coupling on high performance computing architectures for a conjugate heat transfer problem. The flow solver is a Large Eddy Simulation code that has been already ported on massively parallel architectures. The conduction solver is based on the same data structure and thus shares the flow solver scalability properties. Accurately coupling solvers on massively parallel architectures while maintaining their scalability is challenging. It requires exchanging and treating information based on two different computational grids that are partitioned differently on a different number of cores. Such transfers have to be thought to maintain code scalabilities while maintaining numerical accuracy. This raises communication and high performance computing issues: transferring data from a distributed interface to another distributed interface in a parallel way and on a very large number of processors is not straightforward and solutions are not clear. Performance tests have been carried out up to 12 288 cores on the CURIE supercomputer (TGCC/CEA). Results show a good behavior of the coupled model when increasing the number of cores thanks to the fully distributed exchange process implemented in the coupler. Advanced analyses are carried out to draw new paths for future developments for coupled simulations: i.e. optimization of the data transfer protocols through asynchronous communications or coupling-aware preprocessing of the coupled models (mesh partitioning phase).

Parallel Computation for Radiative Heat Transfer Using the DOM in Combustion Applications: Direction, Frequency, Subdomain Decompositions, and Hybrid Methods

In the framework of coupled large-eddy/discrete ordinates method (LES/DOM) computations of turbulent combustion problems, various decompositions for parallel calculations of the radiative heat transfer based on the DOM are investigated. The methods analyzed are: (A) a task decomposition on the discrete directions and frequencies with two numeric strategies: Message Passing Interface (MPI) with distributed memory and OpenMP with shared memory for the direction decomposition; (B) a new algorithm for a DOM subdomain decomposition, which is proposed and tested using MPI; and (C) hybrid methods combining an OpenMP strategy for direction and MPI for tasks and subdomain decomposition. It is shown for the case of coupled simulations that the convergence and the parallel efficiency of the domain decomposition (B) are optimal. This method is limited in this work to 25 sub-domains, at which point the efficiency stagnates. Combining the directions with frequency and/or domain decompositions in a hybrid method (C) results in very good efficiency up to 1,200 processors. This hybrid strategy is also very efficient in terms of memory usage. This work shows that the best way to perform massively parallel computation for radiative heat transfer with the DOM is to combine different decomposition levels. The analysis performed in this work shows the best parallel strategy to be used in coupled simulations between radiation and LES on massively parallel architectures.

COUPLING RADIATION MODELLING WITH TURBULENT COMBUSTION IN LARGE EDDY SIMULATION

Simulation of turbulent combustion has gained high potential with the Large Eddy Simulation (LES) approach, allowing to predict unsteady turbulent reactive flows. In this approach only the largest scales of the turbulence are solved while the smallest scales are modelled. This approach permits to simulate complex industrial geometries on a wide range of Reynolds numbers. Previous works have shown the ability of LES to predict unsteady combustion behaviors such as : instabilites, ignitions and extinctions in industrial systems [1, 2, 3]. It has been demonstrated [4] that it is necessary to take into account thermal radiation losses in combustion caculations to increase their level of accuracy. The radiation is important as well for an accurate prediction of the temperature and the wall heat fluxes. Because the chemistry of polluting species is very sensitive to the temperature, the radiation is a key point for good predictions of the polluting species (CO, NOx, soot, . . . ). Radiation has also an influence on the life time of combustion chambers, so it is necessary to predict accurately the wall fluxes. In this context, taking into account radiation rises new fundamental and practical questions. The physics involved in radiation and combustion are completely different: combustion is controlled by local exchanges and finite times whereas radiation is instantaneous and based on non-local exchanges. In order to couple radiation with turbulent combustion a methodology is needed regarding both physical and numerical aspects. In a first step, the impact of LES modelling on radiation in turbulent combustion is regarded. In LES, the resolved fields are spatially filtered and the unclosed terms are modelled. This question is treated in the more general frame of the turbulence-radiation interaction. From theoretical and numerical studies, it is shown that this interaction is weak in the LES context so that LES solutions can be directly coupled to radiative calculations, without further modelling [5]. This result have been confirmed more recently by other studies [6, 7, 8]. In LES context the nearwall dynamic and thermal boundary layer have to benn modeled. Such models are often derived from Direct Numerical Simulations (DNS). To include the effects of radiation, DNS of an an isothermal reacting turbulent channel flow with and without radiative source terms has been performed to study the influence of the radiative heat transfer on the optically non-homogeneous boundary layer structure [9]. It has been shown that the global structure of the thermal boundary layer is not significantly modified by radiation. However, the radiative transfer mechanism is not negligible and contributes to the heat losses at the walls. The standard wall’s law for temperature can thus be improved for RANS/LES simulations taking into account the radiative contribution by adding the radiative heat flux. The objective of the study is to perform the unsteady coupling of radiation and turbulent combustion that was here quite challenging. First, the reduction of calculation time of radiation, and several strategies are proposed. In particular, a new global spectral model (FS-SNBcK) is introduced [10], ensuring a sufficient level of accuracy. The time calculation of radiation is decreased using of a tabulation technique of the spectral model. Also, larger grids for radiation are employed according to a criterium of temperature homogeneity. The radiative time calculation is finally decreased by two orders of magnitude reaching a ratio of CPU times tradiation/tcombustion ≤ 1, which enables the coupling with a turbulent premixed flame. The studied configuration is a premixed V-shaped turbulent laboratory flame of propane that has been previously studied at EM2C [11] ; [12], see Fig. 2. It has been shown that radiation can decrease the level of temperature by more than 100 K. This effect does not change significantly the mean velocity of the flame or the production of H2O, CO2. However, it has been shown that the total mass fraction of CO decreases by about 20% when the radiation is considered. As the wall temperature is unknown both experimentally and numerically a cold wall temperature (300 K) is assumed. This assumption has an important effect beacause it can lead to an over-prediction of radiative heat losses. The only way to define the accurate temperature is to solve thermal heat transfer inside the solid wall. This demands to couple another solver (AVTP) developped at CERFACS to solve the thermal heat transfer in solids, this work is actually under progress. Figure 1: Calculation of the temperature profiles at the first cell are wrong without law. The law is not modified by radiation Figure 2: Instantaneous field of temperature in the studied configuration. The dimensions of the configuration are 50× 80× 300 mm. References [1] M. Boileau, G. Staffelbach, B. Cuenot, T. Poinsot, and C. Bérat. Combustion and Flame, 154(1-2):2–22, 2008. [2] G. Boudier, L.Y.M. Gicquel, and T. Poinsot. Combustion and Flame, 155:196– 214, 2008. [3] A. Roux, L.Y.M. Gicquel, Y. Sommerer, and T. Poinsot. Combustion and Flame, 152(1-2):154–176, 2008. [4] P. J. Coelho. Progress in Energy and Combustion Science, 33(4):311–383, August 2007. [5] D. Poitou, M. El Hafi, and B. Cuenot. Turkish Journal of Engineering and Environmental Sciences, 31:371–381, 2007. [6] P. J. Coelho. Combustion and Flame, 156(5):1099–1110, May 2009. [7] M. Roger, C. B. Da Silva, and P. J. Coelho. International Journal of Heat and Mass Transfer, 52(9-10):2243 – 2254, 2009. [8] Maxime Roger, Pedro J. Coelho, and Carlos B. da Silva. International Journal of Heat and Mass Transfer, 53(13-14):2897–2907, 2010. [9] J. Amaya, O. Cabrit, D. Poitou, B. Cuenot, and M. El Hafi. Journal of Quantitative Spectroscopy and Radiative Transfer, 111(2):295–301, 2010. [10] D. Poitou, J. Amaya, Bushan Singh C., D. Joseph, M. El Hafi, and B. Cuenot. In Proceedings of Eurotherm83 – Computational Thermal Radiation in Participating Media III, 2009. [11] R. Knikker, D. Veynante, J.C. Rolon, and C. Meneveau. In Proceedings of the 10th international Symposium on Applications of Laser Techniques to Fluid Mechanics, 2000. [12] R. Gonçalves dos Santos, M. Lecanu, S. Ducruix, O. Gicquel, E. Iacona, and D. Veynante. Combustion and Flame, 152(3):387–400, February 2008.