The so-called “sudden death reaction” theory, for a diffusion flame, assumes that the fuel and oxidizer diffuse toward a stoichiometric concentration surface, and then suddenly disappear, due to their combustion which produces water and carbon dioxide. The presence of NOx and CO in the combustion products cannot be explained by the “sudden death” theory. NOx, due to its high activation energy may not be formed prior to the formation of H2O and CO2. NOx is created when both oxygen and nitrogen are present in a high temperature volume; after all the combustible species are consumed. Appearance of CO indicates a lack of oxygen or a low gaseous temperature. Traditionally, when steam is injected into the combustion air, its high heat capacity reduces the flame temperature, which then reduces NOx formation, and this is usually accompanied by high CO formation. This phenomenon is caused by the dilution of oxygen as a quenching effect. This paper describes a novel approach that reverses the traditional wisdom of using steam to control NOx and CO formation, by accelerating the combustion process. This new approach begins with (1) shrinking the flame envelope, (2) enhancing the oxygen diffusion rate, and (3) suppressing the nitrogen concentration diffusion rate. Test results showed that (1) a high temperature volume could form NOx after the combustion of fuel is reduced to a minimum, and (2) that a very high fuel jet momentum increases the oxygen diffusion rate, thus reducing the flame envelope. Also due to the inward movement of the flame envelope, the residential time for NOx formation is also reduced and with the presence of a diluent, the nitrogen penetration rate into the flame is controlled. When all three phenomena are working together, total NOx was reduced downward to below 2 ppm without losing flame stability. Since this process generates enhanced oxygen diffusion, CO has always been seen to be below 2ppm, which indicates extremely high combustion efficiency. The above theory was first simulated by numerical methods using a 3-step reaction for nitrogen and oxygen, and was further expanded to a 28-step chemical kinetic model. The simulation used gas turbine compressor discharge temperatures to produce real adiabatic flame temperatures. Atmospheric tests of real full-scale gas turbine combustors were used with appropriate air temperatures, to simulate adiabatic flame temperatures. Below 2ppm NOx and CO were consistently obtained, independent of turbine types. Actual turbine tests on GE 6B and W501D5A turbines consistently indicated pressure dependent exponents of 0.1.
