OCR Text |
Show been designed to agree with the experimental data available for such selective processes (Muzio, et al., 1977). A system of 351 chemical reactions (Boyle, et al., 1991) describing the C/HIN/O system were numerically integrated using a Gear method to show the evolution of species concentrations as a function of time. Modeling of NH2 injection into typical post-combustion gases was performed to validate the feasibility of the radical injection process. Post-combustion gas concentrations were specified to resemble slightly fuel-lean combustion of CH1.8, which is the approximate atomic balance for fuel oil #2. An oxygen concentration of 3.5% was assumed. The species present were modeled for one second prior to the introduction of radicals to allow species such as CO, OR and 0 to achieve equilibrium. Reactions between those species and NO, as well as NH2, may effect the success of the process. Radical injection was modeled at various temperatures from 300 K to 1700 K. Initially 600 parts per million (ppm) NO were present in the stream at all temperatures. Simulations followed the chemical dynamics of the system following injection of NH2 at 600 ppm and 1200 ppm (Figure 1). Significant reduction occurs between 300 K and 1 ()()() K. In the SNCR process, this region is ineffective due to the inefficiency of ammonia breakdown. When the radicals themselves are introduced, amidogen attack on nitric oxide at these low temperatures is favored over oxidation by reaction with Ch, OR or CCh. At low temperature, and for conditions of excess radical injection, NOx concentrations drop to well below 0.1 % of their initial value. Thus externally induced radical generation and injection provides a means by which the low temperature limit of ammonia injection can be removed. This provides additional opportunities for more effective NO reduction. The high temperature production of NO by oxidation of the NH2 radicals is expected. Generation of relatively stable concentrations of NH2 must be investigated to provide the injection stream. Kinetics modeling has been performed using a subset of the original mechanism which consists only of the NIH reactions. Figure 2 shows the calculated normalized concentration of ammonia radicals plotted versus the logarithm of time. Radical concentrations have been normalized relative to the initial ammonia concentration. High temperature results show a large but short-lived radical supply. In order for sufficient mixing of the radical stream with the post-combustion gases to occur, the radicals must survive for at least one millisecond. This is a generally accepted typical time scale for turbulent mixing. Lower temperature modeling shows an increase in time scales and a corresponding decrease in peak concentration. Radicals are produced and consumed more slowly. Currently, the sink for the ammonia radicals is simply a continued dissociation to form nitrogen and hydrogen (R7,R8). The previous NH2 injection R7) NH2 + R8) NH + H H => => NR + N + modeling has indicated, however, that as these radicals are introduced into NO laden gases, the preferred reaction will be with NO causing overall NOx reduction. Although, as stated above, one millisecond is a typical mixing time for turbulent flow, some portion of the stream will come into contact with the stack gases upon introduction to the stream. Additional modeling is aimed at producing a more uniform radical profile across many time scales. A more desirable concentration profile may be obtained through the addition of secondary species in the radical formation chemistry. Results of this modeling, Figure 3, show that the combination of NH3 with nitrogen and hydrogen streams creates such a profile. Nitrogen and hydrogen alone are modeled and show only a small radical 3 |