CFD modeling of premixed hydrogen-air mixtures: from slow deflagration to DDT

During a severe accident, in a water-cooled nuclear reactor, the release of large quantities of hydrogen occurs due to the exothermal oxidation of fuel claddings. The hydrogen mixed with the air contained in the power plant containment can potentially ignite. The acceleration caused by the interaction of the flame with the equipment in the containment produces a large dynamic pressure load, which is detrimental for the structural integrity of the containment and the equipment. Therefore, the prediction of flame acceleration and dynamic pressure loads is an important safety issue. This thesis work, developed at the Nuclear Research Group(NRG), follows the broad research framework of the SAMHYCO-NET project concerning hydrogen explosions. In the past assignments, NRG developed a CFD-based method to simulate combustion in the density-based solver of ANSYS FLUENT v6 through User Defined Functions(UDF). In the framework of this thesis, the CFD-based method has been updated to work in ANSYS FLUENT v17. A novel experimentally deducted turbulent flame speed closure model has been implemented, which requires the tracking of the flame front, with the previous Turbulent Flame Speed Closure model by Zimont and the Extended Turbulent Flame Speed Closure by Lipatnikov. The main objective of this thesis is to validate further the TFC, ETFC, and the novel model against slow deflagration, upward propagation, and deflagration to detonation (DDT) regimes of hydrogen-air homogeneous mixtures. The particular test selected test cases for each regime are the CNRS-ICARE spherical bomb experiment, the THAI-HD22 experiment, and the ENACCEF-2 benchmark. In the slow deflagration regime, the implemented combustion models predicted a good qualitative sensitivity of flame velocity to hydrogen concentration and the initial turbulence field. However, all the models can not quantitatively reproduce the experimental data because of the incorrect evaluation of the acceleration caused by the thermo-diffusive instabilities. In the THAI-HD22, the upward flame front propagation is overpredicted by all the models. Meanwhile, the peak pressure is overestimated by 3% by the TFC and ETFC models and by less than 1% by the novel model. The maximum mean pressure rise is evaluated in the range of 9% by the ETFC models and by the new model compared with the experiments. The flame acceleration and maximum speed are consistently predicted in the DDT regime by the TFC and ETFC model for the ENACCEF-II benchmark. Moreover, the maximum flame velocity is predicted within 9% by the TFC and ETFC models compared to the experimental values. The pressure history is reproduced consistently by all the combustion models. Moreover, the shock wave interaction with the flame front is correctly accounted for by all the combustion models.