Accurate predictions of heat losses in turbulent combustion are critical for accurate predictions of pollutant emissions such as nitrogen oxides and soot due to their extreme sensitivity to the underlying gas-phase temperature and composition. Reduced-order manifold approaches to modeling turbulent combustion require an additional enthalpy or enthalpy-like parameter to account for these heat losses. The particular focus of this work is radiation heat losses in turbulent nonpremixed combustion. A general and mathematically consistent two-dimensional reduced-order manifold model is developed that can account for arbitrary radiation heat losses in nonpremixed combustion. From this model, two further assumptions are made, considering the specific characteristics of radiation heat losses in turbulent nonpremixed combustion, resulting in two models. First, the radiation heat loss is presumed to be sufficiently weak as to prevent flame quenching. In this case, the two-dimensional manifold equation is shown to be reduced to a pseudo-unsteady one-dimensional manifold model. Second, the radiation heat loss is presumed to be much slower than the gas-phase chemistry governing the flame structure. In this case, the pseudo-unsteady one-dimensional manifold model is reduced to a quasi-steady one-dimensional manifold model. These two models are implemented within a Large Eddy Simulation (LES) context and applied to Sandia Flame D to determine whether or not these assumptions are physically correct. The results indicate that major species and temperature are relatively insensitive to these assumptions. However, results for nitrogen oxides indicate that the quasi-steady assumption is invalid and worsens with increasing downstream distance from the nozzle, with differences between the quasi-steady model and pseudo-unsteady model as large as the differences with an adiabatic model. Further analysis reveals that the chemical time scales rapidly slow with increasing downstream distance, and the assumption that the radiation heat losses are much slower than the chemistry breaks down.

由于湍流燃烧对基础气相温度和成分极为敏感，因此准确预测湍流燃烧中的热损失对于氮氧化物和烟灰等污染物排放的准确预测至关重要。用于模拟湍流燃烧的降阶歧管方法需要附加的焓或类似焓的参数来解决这些热量损失。这项工作的重点是湍流非预混燃烧中的辐射热损失。建立了通用且数学上一致的二维降阶歧管模型，该模型可以解决非预混燃烧中的任意辐射热损失。根据该模型，考虑到湍流非预混燃烧中辐射热损失的具体特征，做出了两个进一步的假设，从而产生了两个模型。首先，假定辐射热损失足够弱以防止火焰猝灭。在这种情况下，二维流形方程被简化为拟非定常一维流形模型。其次，假定辐射热损失比控制火焰结构的气相化学反应要慢得多。在这种情况下，拟非稳态一维流形模型被简化为准稳态一维流形模型。这两个模型是在大型涡流仿真（LES）上下文中实现的，并应用于Sandia Flame D，以确定这些假设在物理上是否正确。结果表明，主要种类和温度对这些假设相对不敏感。但是，氮氧化物的结果表明，准稳态假设是无效的，并且随着距喷嘴下游距离的增加而变差，准稳态模型和拟非稳态模型之间的差异与绝热模型之间的差异一样大。进一步的分析表明，化学时间尺度随着下游距离的增加而迅速变慢，并且假设辐射热损失比化学物质分解要慢得多。