NOx emissions characteristics of the partially premixed combustion of H2/CO/CH4 syngas using artificial neural networks

Abstract

Recently, global interest in energy depletion and the issue of rapid climate change have emerged. To address these issues, research and development related to clean coal technology are currently active. In this study, a combustion experiment is performed to investigate the combustion characteristics of H2/CO/CH4 syngas in a partially premixed model gas turbine combustor. Chemiluminescence measurements are undertaken to study the flame structure and characteristics of syngas combustion at an equivalence ratio range of 0.7–1.3. The Abel inversion method is applied to obtain 2-D chemiluminescence flame images from 3-D accumulated chemiluminescence images. EINOx is measured to investigate the relationship between the flame structure and the exhaust gas. In addition, a process using an artificial neural network (ANN) is employed to establish the EINOx prediction model. As a result, the EINOx characteristics of a partially premixed flame differ from those of a diffusion jet flame. Moreover, they are proportional to the flame temperature and correlated with the flame length at varying equivalence ratios ranging from 0.7 to 1.3 at the same heat input level. Overlapped OH*, CH*, and C2* chemiluminescence images are more accurate for estimating EINOx for various H2/CO/CH4 syngas compositions. This is confirmed from the ANN estimation results. The flame temperature, length, and input air flow rate are used in ANN to predict EINOx. The coefficients of correlation used for the predictions are 0.78 and 0.62 when using overlapped OH*, CH*, and C2* chemiluminescence images and only the OH* chemiluminescence image as input parameters, respectively. The weight partition method of this ANN process also confirms that EINOx formation is affected more by the flame temperature and length than by the air mass flow rate.