Prudent utilities consider all aspects of power production – from fuel procurement to firing hardware to waste disposal practices – to achieve their NOX compliance targets at minimum cost. Among the available software tools to manage the fuel quality impacts, EPRI’s NOXLOI Predictor™ is the clear choice. When you need to know how an impending fuel switch will change NOX and loss-on-ignition (LOI), the NOXLOI Predictor™ uses your company’s operating experience at each plant to forecast the new emissions as accurately as possible. Normally, the predicted NOX emissions can be validated against full-scale test data within 0.025 lb NOX/MMBtu. The NOXLOI Predictor™ uses only standard fuel properties and the most basic firing conditions, so cases can be set up in minutes by nontechnical professionals; specialized technical training is unnecessary. It operates on ordinary PCs and completes each calculation set in seconds. So dozens of candidate fuels can be screened at the procurement stage, to identify bad actors before they cause problems at a plant. Potential emissions reductions for coal blending and for co-firing coals and opportunity fuels can be accurately estimated from performance data on individual coals only. Every case study predicts LOI as well as NOX, to flag the fuels that will disrupt flyash management strategies.
Other attempts to relate NOX emissions to fuel quality use conventional fuel properties assigned from the proximate and/or ultimate analyses. Such correlations depict the correct qualitative trends, but they have not been able to predict NOX within useful quantitative tolerances. Whereas the ASTM proximate analysis provides a general indication of differences in fuel properties, it is widely known that it does not accurately represent the fuel’s behavior under the rapid heating conditions in large flames. For the rapid heating rates imposed on pulverized grinds in a utility furnace, the volatile yield and nitrogen release is often substantially different than indicated by the ASTM proximate analysis. Both of these quantities can have a major impact on subsequent NOX formation mechanisms.
The NOXLOI Predictor™ uses two of the most advanced reaction mechanisms available to predict a fuel’s behavior in flames: FLASHCHAIN® for devolatilization and the Carbon Burnout Kinetics (CBK) Model for char burnout. These mechanisms were independently validated against test data on hundreds of coals, and have already been used to predict the behavior of thousands of coals from every geographical region worldwide. In the NOXLOI Predictor™, they are the “Virtual Fuel Laboratories” that describe a fuel’s distinctive volatiles yield, N-release patterns, and char burnout kinetics, based only on the proximate and ultimate analyses and the conditions in p. f. flames.
The NOX predictor portion of the program uses a correlation-based methodology. The user enters the proximate and ultimate analyses of the currently fired coal (or of the components in a baseline coal blend) and of the fuels to be screened, which can be coals and blends of coal with other coals, biomass, or petroleum coke. In addition, measured NOX emissions with the baseline coal or coal blend are entered along with basic information on the firing configuration, whether low-NOX burners are present, an approximate age criterion, and the type of NOX control technology (separated or close-coupled overfire air). After the fuel and boiler information are input, the program processes each fuel through FLASHCHAIN® which calculates the volatiles yield and volatile-N production. The FLASHCHAIN® results are then used as the regression variables in a series of correlations to predict the expected change in NOX emissions for the alternate fuels. The correlations were developed using numerous sets of pilot-scale data available in the literature and from various other sources, and are related to the emissions characteristics of a particular boiler with the data on NOX emissions for the baseline fuel.
In this calculation scheme, the role of the baseline emissions data is key. Baseline data represent all the boiler-related and burner-related factors in the prediction algorithms. If burner settings are varied or if the overfire air level is adjusted, then the baseline data entered into the software must change. The NOXLOI Predictor™ calculates the expected change in emissions for other coals and blends of coals with pet coke and biomass fired under the baseline conditions.
Whereas the NOX prediction is based on correlations, the LOI prediction is phenomenologically based. The primary elements of the LOI prediction include a fuel grinding submodel; the char formation submodel (FLASHCHAIN®); a char density submodel; and the char oxidation submodel (CBK8). The grinding submodel estimates the coal grind size for the alternate coals given values of the Hardgrove Grindability Index (HGI) for the baseline coal and the screening fuels. Heating values determine the pulverizer throughput for the screening fuels. The grinding submodel also represents the way that the coarser fractions of fuel blends are enriched in the components that have lower HGI values. CBK, the char oxidation submodel, predicts the burning rate, the char particle temperature, and the changes in the particle diameter as combustion proceeds, given a fuel’s proximate and ultimate analyses. A key feature of CBK is its ability to describe char oxidation throughout long residence times. This is important for predicting LOI, which is especially sensitive to the last fractions of the carbon left to burn. Under typical p. f. firing conditions, a conventional char oxidation model will basically predict zero LOI; whereas the CBK model, which includes char annealing and ash inhibition, can simulate the latest stages of char oxidation.
In the LOI predictor, the user specifies the excess O2 concentration and LOI measurements for the baseline fuel and firing conditions. These data determine a mixing effectiveness parameter for the subject furnace. The LOI prediction algorithms also account for preferential O2 consumption by the more reactive components in fuel blends. Generally speaking, the LOI predictions are accurate enough to identify fuels that produce excessive LOI, but not to distinguish fuels that will exceed an LOI threshold for commercial flyash utilization, such as additions to concrete or flowable fill.
The NOXLOI Predictor™ has already been distributed to over 80 American utility companies and one Taiwanese company. Predictions from the software have already been independently validated against full-scale emissions data on diverse coals, biomass, and pet cokes for the most common furnace firing configurations. Technical publications illustrating such performance are available on request. Generally speaking, NOX predictions are within about 15 ppm or 0.025 lb NOX/MMBtu of measured values for full-scale utility furnaces. The LOI predictions are accurate enough to identify fuels that produce excessive LOI, but not to distinguish fuels that will exceed an LOI threshold for commercial flyash utilization, such as additions to concrete or flowable fill.
J. Sun, R. H. Hurt, S. Niksa, L. Muzio, A. Mehta, and J. Stallings, “A simple numerical model to estimate the effect of coal selection on pulverized fuel burnout,” Combust. Sci. Technol., 175(6):1085-1108 (2003).
S. Niksa, M. Lanning, D. L. Hill, B. Nguyen, L. Muzio, R. H. Hurt, A. Kornfeld, J. Stallings, “Assess coal quality impacts on NOX and LOI with EPRI’S NOX LOI PREDICTOR™,”EPRI-DOE-EPA-A&WMA Combined Utility Air Pollution Control Symposium: The MEGA Symp. 2003, EPRI, Washington, DC, 2003.
S. Niksa, A. Kornfeld, L. Muzio, T. Fang, R. H. Hurt, J.-Q. Sun, A. Mehta, J. Stallings, W. Gibb, M. Cloke, T. Lester, “Assess coal quality Impacts on NOX and LOI with EPRI’s NOX LOI Predictor™,” Proc. EPRI-DOE-EPA Combined Utility Air Pollutant Control Symp., Washington, D. C., EPRI, Palo Alto, CA, 1997.
S. Niksa, R. H. Hurt, J. Stallings, A. Mehta, L. Muzio, A. Kornfeld, “EPRI’s NOX LOI Predictor™,” Proc. EPRI-DOE-EPA Combined Utility Air Pollutant Control Symp., Washington, D. C., EPRI, Palo Alto, CA, 1999.