The reactor network is a computational environment that accommodates realistic chemical reaction mechanisms; indeed, mechanisms with several hundred elementary chemical reactions can now be simulated in minutes on ordinary PCs, provided that the flow structures are restricted to the limiting cases of plug flow or series of continuously stirred tank reactors (CSTRs). The network is “equivalent” to the CFD flowfield in so far as it represents the bulk flow patterns in the flow. Such equivalence is actually implemented in terms of the following set of operating conditions: The RTDs in the major flow structures are the same in the CFD flowfield and in the section of the reactor network that represents the flow region under consideration. Mean gas temperature histories and the effective ambient temperature for radiant heat transfer are also the same. The entrainment rates of surrounding fluid into a particular flow region are evaluated directly from the CFD simulation. To the extent that the RTD, thermal history, and entrainment rates are similar in the CFD flowfield and reactor network, the chemical kinetics evaluated in the network represents the chemistry in the CFD flowfield.
To illustrate the main differences, we consider the case of NOX emissions from a coal-fired furnace. In conventional CFD post-processing, a first-pass calculation imposes a radically reduced set of chemical species with a rudimentary reaction scheme to predict the heat release and its impact on the flowfield, but not the emissions. Then the converged solutions for the flowfield, temperature field, and major species concentration fields are re-analyzed with additional species and more global reaction processes to predict emissions. In contrast, ChemNet™ utilizes the flow and temperature fields but not the species concentration fields from the first pass CFD simulation, because the concentrations were determined with the rudimentary reaction submodels. In addition, ChemNet™ uses fields of the turbulent diffusivity and selected conserved scalar variables, which are always computed in CFD but not normally reported to the user. ChemNet™ then specifies an equivalent reactor network directly from the CFD flow and temperature fields. Finally, realistic elementary reaction mechanisms are used to determine the concentrations of all major and various minor species across the reactor network, including any emissions of particular interest.
From a practical perspective, it is only possible to implement ChemNet™ after the CFD flowfield has first been subdivided into regions. The regions are the rudimental elements of the chemical structure of the flowfield. As such, each region sustains a collection of chemical reaction mechanisms that are distinctive. Regions are usually much more extensive than any distinct flow structures. For example, the core formed by the primary jet within a dual register burner is a region, because the very high loadings of particles and soot in this region will significantly perturb the chemical reaction rates in the gas phase, especially the N-conversion mechanisms. Mixing layers formed by simultaneous entrainment of fuel-rich fluid into secondary or tertiary air streams are also regions, because the temperature profiles along the direction of mixing exhibit similar maximum values across the entire layer. The portion of an overfire jet remaining to be mixed with a process stream is another region, because the absence of fuel essentially eliminates all chemistry.
NEA developed ChemNet™ as a means to couple our fully validated reaction mechanisms for the solid fuel phase with comprehensive elementary reaction mechanisms for chemistry in the gas phase and on soot. This method is especially suitable whenever the levels of emissions and other minor species must be predicted for a diverse assortment of solid fuels across a broad domain of operating conditions; in other words, in the most demanding simulation applications. It has already been demonstrated in pulverized fuel flames at lab-, pilot-, and full-scale; in pilot- and full-scale CFBCs; in pilot- and full-scale gasifiers; and along utility flue gas cleaning systems.
Pulverized coal flames, pressurized coal flames, and NOX reduction via direct radical injection at lab-scale have already been analyzed with ChemNet™ for NOX emissions and minor species like CO from numerous coals. NEA used ChemNet™ to simulate a 1.0 MWt test flame facility to determine the optimal injection configuration for diverse fuels in biomass cofiring on coal. Predicted NOX emissions were within 30 ppm of the measured values for two biomass forms co-fired at three loadings on four diverse coals with no parameter adjustments whatsoever. A boiler OEM in Japan had problems with excessive NOX emissions whenever slag formed near the burner zone of a 550 MW wall-fired furnace. With ChemNet™, NEA connected the substantial increase in NO production due to a relatively small increase in flame temperature, and accurately predicted the magnitude of this effect. NEA used ChemNet™ to accurately predict LOI, NOX and CO emissions from a 550 MW T-fired boiler, based on the client’s in-house CFD simulations and a 444-step mechanism for chemistry on soot and in the gas phase, including fuel-nitrogen conversion. NEA used ChemNet™ to develop a CFBC simulator based on full chemistry that clearly reveals the underlying chemical structure, from hardly any conversion at all in the dense bottom bed to vigorous conversion of both volatiles and char in the splash zone to staged burnout of char along the riser and into the cyclone separator. This simulator has been used to interpret the most important emissions (NO, N2O, NO2, SO2, unburned carbon) in the literature reports for about two dozen CFBCs at pilot- and full-scale.
Coal gasification in two pilot units has been simulated with ChemNet™. In one instance, the goal was to predict the levels of residual CH4 in the syngas from their moderate temperature gasifier for a variety of solid fuels. Whereas virtually all the leading simulation teams in the USA were given opportunities to accurately interpret the reported syngas compositions, NEA’s ChemNet™ simulations were the only ones that accurately predicted the residual CH4 levels over a broad domain of gasifier conditions, scales, and fuel quality. ChemNet™ has also been implemented for moderate-temperature, transport, and entrained-flow gasifiers at commercial scale.
J.-P. Lim, D. Steele, D. del Rio Diaz-Jara, D. J. Eckstrom, R. B. Wilson, S. Niksa, and R. Malhotra, ”A zero CO2-emitting process for transportation fuels from coal and natural gas resources,” J. Sustainable Energy Eng., 1(3):202-219 (2013).
G.-S. Liu and S. Niksa, “Pulverized coal flame structures at elevated pressures. Part 1. Detailed operating conditions,” Fuel, 84(12/13), 1563-74 (2005).
S. Niksa and G.-S. Liu, “Pulverized coal flame structures at elevated pressures. Part 2. Interpreting NOX production with detailed reaction mechanisms,” Fuel, 84(12/13): 1575-85 (2005).
S. Niksa, G.-S. Liu, L. G. Felix, P. V. Bush, and D. M. Boylan, “Predicting NOX Emissions from Biomass Cofiring,” EPRI-DOE-EPA-A&WMA Combined Utility Air Pollution Control Symposium: The MEGA Symp. 2003, EPRI.
S. Niksa, G.-S. Liu, L. G. Felix, P. V. Bush, and D. M. Boylan, “Predicting NOX Emissions from Biomass Cofiring,” 28th Int. Technical Conf. on Coal Utilization and Fuel Systems, Coal Technology Assoc., Clearwater, Fl, March, 2003.
S. Niksa and G.-S. Liu, “Advanced CFD Post-Processing for P. F. Flame Structure and Emissions,” 28th Int. Technical Conf. on Coal Utilization and Fuel Systems, Coal Technology Assoc., Clearwater, Fl, March, 2003.
S. Niksa, G. Liu, L. G. Felix, and P. V. Bush, “Advanced CFD Post-Processing for Pulverized Fuel Flame Structure and Emissions,” Paper No. IJPGC2002-26136, Int. Joint Power Gen. Conf., ASME, Phoenix, AZ, June 25, 2002.
S. Niksa and G. Liu, “Detailed reaction mechanisms for coal-nitrogen conversion in p. f. flames,” Proc. Comb. Inst. 29:2259-2265 (2002).
S. Niksa and G. Liu, “Incorporating detailed reaction mechanisms into simulations of coal-nitrogen conversion in p. f. flames,” Fuel, 81(18):2371-85 (2002).