Modeling of flame-generated turbulence based on direct numerical simulation databases
Modeling of flame-generated turbulence based on direct numerical simulation databases
复制标题
基于直接数值模拟数据库的火焰产生湍流建模
DOI：
10.1016/s1540-7489(02)80246-2
复制
发表时间：
2002
影响因子：
1.9
通讯作者：
R. Himeno
中科院分区：
文献类型：
--
作者：
S. Nishiki;T. Hasegawa;R. Borghi;R. Himeno
采用单步不可逆反应直接模拟了紊流预混火焰在均匀各向同性湍流中的传播。计算了两种情况，H情况下火焰高密度比pu/pb=7.53， L情况下火焰低密度比pu/pb=2.50，而u′/ ul几乎等于1。我们获得了完全发展的静止湍流火焰的数据库。通过分析湍流动能输运方程对这些数据库进行研究，研究火焰产生的湍流及其模型。我们发现，在火焰刷内，所有组分的湍流波动，尤其是流向组分的湍流波动都被放大了，火焰密度比越大，火焰产生的湍流就越大。基于favre平均输运方程的湍流动能分析表明，压力相关项在火焰刷内产生动能，H情况下平均压力梯度项最重要，l情况下压力功项最重要，而扩散耗散项和速度梯度项则降低了动能。其次，讨论了平衡方程中重要项的建模问题。模拟了平均压力梯度项、压力膨胀项和附加耗散分量，并与直接数值模拟（DNS）结果进行了比较。采用密度假设对平均压力梯度项进行建模，模型与DNS吻合较好。另外两个术语也通过缩放建模，这些模型很好地模拟了DNS。
Turbulent premixed flames propagating in homogeneous isotropic turbulent flows were simulated directly with a single-step irreversible reaction. Two cases were calculated, case H, with a high-density ratio of flame pu/pb=7.53, and case L, low-density ratio of flame pu/pb=2.50, while u′/uLwas nearly equal to unity. We obtained databases of fully developed stationary turbulent flames. These databases were investigated by analyzing the transport equation for turbulent kinetic energy to study flame-generated turbulence and its models. We found that turbulent fluctuations of all components, especially the streamwise component, were amplified in the flame brush and that flame-generated turbulence increased for a larger density ratio of the flame. Analysis based on the Favre-averaged transport equation for turbulent kinetic energy showed that pressure-related terms produced kinetic energy in the flame brush, the mean pressure gradient term was most important in case H and the pressure work term was most important in case L. On the other hand, the diffusion and dissipation term and velocity gradient term decreased kinetic energy. Next, modeling of the important terms in the balance equations were discussed. The mean pressure gradient term, pressure dilatation term, and additional dissipation components were modeled and compared with the direct numerical simulation (DNS) results. The mean pressure gradient term was modeled with assumption on the density, and the model was in good agreement with DNS. The other two terms were also modeled by scaling and these models mimicked DNS well.