비정상 화염편 연소모델을 사용한 비예혼합 및 부분예혼합 난류 화염장 수치해석

Abstract

In the practical combustion systems, the partially premixed flame characteristic considerably influences the flame stabilization and pollutant formation and largely depends on flow configurations, fuel type, heat losses, and mixing conditions. The partially premixed turbulent flames involve many fundamental mechanisms, which contain flame ignition, local extinction, re-ignition, and flame propagation. Thus, it is quite important to numerically predict the partially premixed combustion which involves the strongly coupled and highly nonlinear physical processes. In the present study, the transient flamelet model has been developed to realistically simulate the partially premixed turbulent flames due to the computational efficiency, the effective incorporation of complex combustion phenomena such as detailed chemical kinetics, differential diffusion, and soot formation. It is well known that the steady flamelet model has the serious shortcoming to deal with the slow processes including pollutant (thermal NOx, soot) formation and radiative heat transfer. To overcome these limitations, the present approach has adopted two representative interactive flamelet (RIF) approaches including the Lagrangian flamelet model and the Eulerian particle flamelet model. Numerical results indicate that these transient flamelet models correctly predict the transient development of the local flamelet structure. The sequence of the benchmark cases were used to validate the present transient flamelet models. First, to verify the prediction capability of the Lagrangian flamelet model, the CO/H2/N2 turbulent jet flames [Barlow et al., 1998] are chosen as the validation cases. In terms of unconditional and conditional means, numerical results are in good agreement with experimental data. It is also identified that the realistic modeling of the radiative cooling is essential to correctly predict the precise flame structure. In order to validate the Eulerian particle flamelet model, the measurement (Dally et al., 1998) of the bluff-body flames is selected as the example case. The predicted profiles of unconditional and conditional means are reasonably well agreed with the measured distributions. Moreover, it is found that the present transient flamelet approaches realistically predict the dilution effects on NOx formation of syngas non-premixed turbulent flame as well as the flame structure and pollutant formation characteristics of the oxygen-enriched syngas flames. However, these two transient flamelet models are unable to correctly predict the local extinction or re-ignition phenomena encountered in the actual combustion devices because they do not consider the unstable branch region in the S-shaped curve. To resolve these shortcomings of the transient Lagrangian and Eulerian flamelet models, the present study has devised the flamelet/progress approach which is capable of considering the unstable branch region in the S-shaped curve. the unstable branch region, is then developed to predict the partially premixed lifted jet flame. To demonstrate the prediction capability of the steady and unsteady flamelet/progress approach, the experiment (Cabra et al., 2002, 2005) of the lifted jet flames with the vitiated coflow was chosen as the validation cases. In terms of the liftoff height, flame structure and autoignition characteristics, both the steady and unsteady flamelet/progress variable models yield the reasonably good conformity with experimental results. However, numerical results clearly indicate that the unsteady flamelet/progress variable approach has the much better capability to predict the precise flame structure, autoignition and liftoff characteristics in the turbulent partially premixed flames. Based on numerical results obtained in this study, the detailed discussions are made for the essential combustion characteristics and turbulence-chemistry in turbulent nonpremixed and partially premixed flames as well as the capability and limitations of the flamelet-based turbulent combustion models.