A Study of Pollutant Formation from the Lean Premixed Combustion of Gaseous Fuel Alternatives to Natural Gas

Thesis (Ph.D.)--University of Washington, 2011The goal of this research is to identify how nitrogen oxide (NOx) emissions and flame stability (blowout) are impacted by the use of fuels that are alternatives to typical pipeline natural gas. The research focuses on lean, premixed combustors that are typically used in state-of-the-art natural gas fueled systems. An idealized laboratory lean premixed combustor, specifically the jet-stirred reactor, is used for experimental data. A series of models, including those featuring detailed fluid dynamics and those focusing on detailed chemistry, are used to interpret the data and understand the underlying chemical kinetic reasons for differences in emissions between the various fuel blends. An ultimate goal is to use these data and interpretive tools to develop a way to predict the emission and stability impacts of changing fuels within practical combustors. All experimental results are obtained from a high intensity, single-jet stirred reactor (JSR). Five fuel categories are studied: (1) pure H2, (2) process and refinery gas, including combinations of H2, CH4, C2H6, and C3H8, (3) oxygen blown gasified coal/petcoke composed of H2, CO, and CO2, (4) landfill and digester gas composed of CH4, CO2, and N2, and (5) liquified natural gas (LNG)/shale/associated gases composed of CH4, C2H6, and C3H8. NOx measurements are taken at a nominal combustion temperature of 1800 K, atmospheric pressure, and a reactor residence time of 3 ms. This is done to focus the results on differences caused by fuel chemistry by comparing all fuels at a common temperature, pressure, and residence time. This is one of the few studies in the literature that attempts to remove these effects when studying fuels varying in composition. Additionally, the effects of changing temperature and residence time are investigated for selected fuels. At the nominal temperature and residence time, the experimental and modeling results show the following trends for NOx emissions as a function of fuel type: 1.) NOx emissions decrease with increasing H2 fuel fraction for combustion of CH4/H2 blends. This appears to be caused by a reduction in the amount of NO made by the prompt pathway involving the reaction of N2 with hydrocarbon radicals as the CH4 is replaced by H2. 2.) For category 2 (the process and refinery blend) and category 5 (the LNG, shale, and associated gases), NOx emissions increase with the addition of C2 and C3 hydrocarbons. This could be due to an increased production of free radicals resulting from increasing CO production when higher molecular weight hydrocarbons are broken down. 3.) For category 3 (the O2 blown gasified coal/petcoke), NOx emissions increase with increasing CO fuel fraction. The reason for this is attributed to CO producing more radicals per unit heat release than H2. When CO replaces H2, an increase in NOx emissions is seen due to an increase in the productivity of the N2O, NNH, and Zeldovich pathways. 4.) For category 4 (the landfill gas) the addition of diluents such as CO2 and N2 at constant air flow produces more NOx per kg of CH4 consumed, and N2 is more effective than CO2 in increasing the NOx emission index. The increase in emission index appears to be due to an enhancement of the prompt NOx pathway as the diluents are added and the mixture moves towards stoichiometric. In addition, the presence of CO2 as a diluent catalyzes the loss of flame radicals, leading to less NOx formation than when an equivalent amount of N2 is used as a diluent. For a selected set of fuels, detailed spacial reactor probing is carried out. At the nominal temperature and residence time, the experimental results show the following trends for flame structure as a function of fuel type: 1.) Pure H2 is far more reactive in comparison to CH4 and all other pure alkane fuels. This results in relatively flat NOx and temperature profiles; whereas, the alkane fuels drop in both temperature and NOx production in the jet, where more fresh reactor feed gases are present. 2.) For category 2 (the Process and Refinery blends), H2 addition increases reactivity in the jet while decreasing overall NOx emissions. The increased reactivity is especially evident in the CO profiles where the fuels blended with C2H6 and H2 have CO peaks on jet centerline and CO emissions for pure CH4 peaks slightly off centerline. 3.) For category 3 (the O2 blown gasified coal/petcoke), the temperature profiles for the gasification blend and pure H2 are nearly identical, which is likely due to the high reactivity of H2 dominating the relatively low reactivity of CO. Despite a small temperature difference, the addition of CO causes an increase in NOx production. 4.) For category 4 (the landfill gas), the temperature profiles are virtually indistinguishable. However, the addition of diluent decreases reactivity and spreads out the reaction zone with the CO concentration peaking at 2 mm off of centerline instead of 1 mm. Diluent addition increases NOx production in comparison to pure CH4 for reasons explained above. 5.) For category 5 (the LNG, shale, and associated gases), the temperature profiles are all very similar. The increased reactivity of C2H6 is evident from looking at the CO profiles. Increased C2H6 promotes CO production on jet centerline which is indicative of the hydrocarbon material breaking down earlier in the jet. At temperatures and residence times other than the nominal conditions, the experimental results show the following trends: 1,) The NOx emissions from LPM combustion of pure CH4, H2, C2H6, and C3H8 are shown to vary linearly with residence time and in an Arrhenius fashion with temperature. This occurs because (1) more reaction time leads to more NOx formation, and (2) NOx formation is a strong, non-linear function of temperature. 2.) The addition of both H2 and C2H6 to a LPM CH4 flame is effective at extending its lean blowout limit. The results of both two and three dimensional CFD simulations are presented to illustrate the general flow, temperature, and species structure within the reactor. Since the two dimensional model is far more computationally efficient, it is employed to study various fuel mixtures with more sophisticated chemical mechanisms. The CFD results from the LPM combustion of H2, H2/CO, and CH4 with NOx formation are presented. A three dimensional CFD simulation is run for LPM CH4 combustion that uses a global CH4 oxidation mechanism. While this model does not predict intermediate radicals and NOx, the CO contours and flow field can be used as guidelines to develop a chemical reactor network (CRN), which can incorporate detailed chemistry. In addition, this model runs quickly enough that it is a good way to initialize the temperature and flow field for simulations that do incorporate more complex chemistry. The two dimensional model is used to illustrate the difference in combustion behavior between the various fuels tested. In particular, it illustrates the geometric locations of the super-equilibrium radical fields and shows where and through which pathways NOx is formed. The pathway breakdowns show good agreement with the CRN modeling results. The main goal of the CFD modeling is to use the results of each model to develop Chemical Reactor Networks, CRNs, that are customized for a particular burner. The CRN can then be used to estimate the impacts due to fuel variation