Fernando Marcelo Pereira

Assessment of combustion models for numerical simulations of a turbulent non-premixed natural gas flame inside a cylindrical chamber

ABSTRACT This work presents a Computational Fluid Dynamics (CFD) study of the non-premixed combustion of natural gas with air in an axisymmetric cylindrical chamber, focusing on the contribution of the chemical reaction modeling on the temperature and the chemical species concentration fields. Simulations are based on the solution of mass, momentum, energy and chemical species conservation equations. Thermal radiation heat transfer in the combustion chamber is computed through the Discrete Transfer Radiation Method, and the Weighted-Sum-of-Gray-Gases model solves the dependence of gas absorption coefficient on the wavelength. Turbulence is modeled by the standard k-ε model. Regarding the combustion modeling, it is performed a comparison of solutions obtained with the combined Eddy Break-Up/Arrhenius (EBU/Arrhenius) and the Steady Laminar Diffusion Flamelet (SLDF) models. The finite volume method is employed to treat the differential equations. Among other results, the solution of the governing equations allows for the determination of the region where combustion takes place, the distribution of the chemical species and the velocity fields. The numerical results are compared to experimental measurements, showing varied agreements. Results indicate that, in this case, the EBU/Arrhenius model can predict the flame temperature and the concentration of the most important species with better accuracy than the more sophisticated SLDF model.

Application of Inverse Analysis to Determine the Parameters of the Weighted Multi Point Source Model for the Prediction of Heat Flux From Flames

Thermal radiation is responsible for a substantial portion of the heat transfer from a flame. Radiative characteristics of jet fires are usually expressed through the use of the fraction of heat radiated. Detailed flame simulations provide useful information but may be prohibitive for practical applications due to the significant computational resources that are required. The weighted-multi-point-source model uses point sources with different weights to simulate the contribution of each portion of the flame. This method can give good predictions of the radiant heat flux both in the near and far fields. While previous studies used the trial-and-error approach to determine the weights of the sources, this paper proposes a method to obtain the weight of each source by inverse analysis. In the analysis, the experimental measured radiation heat flux distribution is the input data, while the weights of the sources and the fraction of heat radiated are the sought parameters. The inverse problem is formulated as an optimization problem, which is solved by the generalized extremal optimization. As will be shown in the paper, the inverse method is capable of recovering the weights of the sources that lead to results with more accuracy than the single source model or the multiple sources model with the commonly employed linear variation in the weights.Copyright