Impact of gas composition variations on flame blowout and spectroscopic characteristics of lean premixed swirl flames

Autor: Jo-Han Ng, Cheng Tung Chong, Norshakina Shahril, Mohd Shiraz Aris, Meor Faisal Zulkifli, Sing Tung Ting, Guo Ren Mong
Rok vydání: 2019
Předmět:
Zdroj: Process Safety and Environmental Protection. 128:1-13
ISSN: 0957-5820
Popis: The injection of gases from different oil and gas fields and external sources such as liquefied natural gas increases operational risks for the relevant gas turbine power plant operators. In practice, the absence of interchangeability specifications in the gas supply network code has caused combustion blowout, wear and tear to occur due to combustion dynamics and diluent effects. Understanding the effects of diluents on natural gas combustion is essential to ensure the safe operation of existing facilities. The present work investigates the flame stability and spectroscopic characteristics of diluents-containing natural gas by using a swirl flame burner. Stable and continuous swirl flames were successfully established using different types of gas compositions, including those diluted with nitrogen and carbon dioxide. Diluting the modelled natural gas with CO2 and N2 results in higher blowout limit as compared to the baseline pure methane case. Preheating the burner and mixtures can extend the flame blowout limits, although the effect of CO2 on flame blowout is more pronounced than that of N2 due to its higher heat capacity. This work shows the effects of non-reactive diluents on gas turbine flame can be significant, particularly at high-level dilutions. Mitigation measures such as gas composition and flame spectroscopy monitoring can be deployed to ensure safe operation of the system. By using the statistical analysis technique of linear regression, the proportions of all the fuel mixture components of CH4, C2H6, CO2 and N2, alongside temperature were found to be significant factors in determining flame blowout limits. The developed predictor equations for OH intensity and lean blowout equivalence ratios show the predictive capability of >89% at 90–95% confidence level.
Databáze: OpenAIRE