OCR Text |
Show INTRODUCTION The use of natural gas continues to grow as a key fuel for utility, industrial, and residential services. This increased use is expected to continue in the United States because of availability and price. In addition, market demands and both federal and local regulations are placing more stringent requirements on efficiency and emissions. These motivations are driving the natural gas industry to design new burners which will operate beyond of existing design envelopes. Previous design strategies have relied heavily on empirical data, but this approach becomes risky for advanced design concepts. Methods are needed to scale burner designs and to effectively study a variety of parameters such as burner configuration, burner location and orientation, fuel composition and mUltiple fuel requirements, fuel and air staging, operational upsets, and specific applications such as flame shaping. While pilotscale tests will continue to be used to design and test burner concepts, the cost of these tests is significant. Additional tools are needed to optimize and focus the test matrix, and to aid in scale-up of the pilot-scale designs, thus reducing developmental costs and risk of nonperformance. Numerical modeling is a relatively new engineering tool that is ideal for evaluating combustion systems. The evolution of these methods since the mid-1970's has produced computational tools that are now applicable to the prediction of the complex physical processes involved in fossil fuel combustion: turbulent fluid dynamics, gas-phase species transport, chemistry, and heat transfer (Fiveland, et aI., 1984; Fiveland and Wessel, 1991; Fiveland and Jessee, 1994). These models are now being used more frequently in both the utility and industrial burner markets to provide additional insight and guidance to the design and application of efficient, low emission natural gas burners (Peters and Weber, 1994). The use of these models introduces several new issues, since they are neither foolproof, nor simple to use. These issues include suitability of the modeled physics to the design, validation of the software, correct use, and interpretation of the results. Most issues rely on an experienced engineer to determine the suitability of the model, and to correctly apply it and interpret the predictions. The coupling of the complex, non-linear mechanisms simulated by these models, however, requires that the models must be thoroughly validated before they can be routinely applied to design problems. This validation process must also be repeated as the models are refined and extended. Validation of the software, however, requires a set of comparison data for a representative condition with characteristics similar to the case of interest. For satisfactory validation of the numerical model, this data must meet several criteria. The data must be of good quality: repeatable, measured with appropriate instrumentation to sufficient resolution, and with error bounds that are understood. Just as important as the quality of the measurements is the need to provide information on all aspects of the physics concerned, and the need to characterize critical spatial areas where the physical processes are active. For an industrial natural gas flame, measurements are needed for velocities, temperature, and species both within the flame and external to the flame, as well as characterization of the heat transfer around the flame. In addition to the measurement data themselves, sufficient information must also be provided on the geometry, operating conditions, and surrounding environment to make the validation meaningful. The application of reasonable boundary conditions is essential to the validation process, since unrealistic or uncharacteristic boundary information can result in misleading predictions. For a natural gas burner, information must be obtained on the burner and furnace geometry, oxidant and natural gas flow rates, velocities and turbulence levels, and heat transfer and material information on the boundary. For a validation benchmark that is to be used consistently at different times and for different analysis software, this information must be documented clearly so that compatible predictions can be obtained and compared. Although a significant amount of experimental data is available for natural gas fired burners, the data is typically collected as part of specific burner test programs, and does not provide the detailed in-flame data and boundary condition information required to verify combustion models. As a result, no comprehensive data sets are 2 |