OCR Text |
Show PREDICTING THE COMPLETE DISTRIBUTIONS OF VOLATILE PRODUCTS FROM DIVERSE FUEL TYPES WITH FLASHCHAIN™ Stephen Niksa Niksa Energy Associates, 1745 Terrace Drive, ABSTRACT ci_ASHCHAiN is a reaction mechanism for the rapid devolatilization of solid fuels This paper illustrates how the theory was recently expanded to predict the complete distribution of all major devolatilization products for a vanety of 'uel *ypes. including any kind of coal, biomass, and petroleum coke Noncondensible gases are now resolved n;o the pnmary hydrocarbon species (CH4. C2H,, C2H4, Cihs. C3H,). HCN. H|, H2S, H20. C02, and CO. Tars are characterized by their complete elemental compositions C H/O'N/S) and their molecular weight distributions Chars are characterized by their complete elemental compositions (C'H<0/N/S) and their sizes and densities. The theory also predicts the partitioning of chlorine and alkali species (Na and K) This paper also introduces a mechanism to descnbe the secondary pyrolysis of the pnmary devolatilization products, as occurs naturally at elevated temperatures in all combustion and gasification systems Dunng secondary volatiles. the volatiles are radically transformed, with comoiete conversion of tar into soot and conversion of all gases into H2, CH4. C2H:. HCN. H2S, CO, C02, and H 20. NTRODUCTION Engineers m the utility industry are being called upon to manage power production with a wide spectrum of solid fossil fuels, including coal, cokes, biomass, and various waste materials. They usually use either test data and/or CFD simulations to characterize fuel quality impacts dunng combustion However, traditional fuel quality parameters sucn as the fuel ratio, proximate volatile matter content, and 'uei-nitrogen content are often unable to correlate the most nportant macroscopic combustion characteristics, such as 'uei-nitrogen conversion to NOx, char burnout times, and carbon carryover Measured pyrolysis and combustion behavior performs better, but testing entails a substantial nvestment of professional staff and is usually not used routinely. This paper demonstrates a computational approach for fuel quality impacts based on FLASHCHAIN™ (Niksa, 1995a _ 1995b) that achieves the accuracy of laboratory testing for a fraction of the expense. Evaluations cover a vanety of solid fossil fuels, including coals, petroleum cokes, and /anous forms of biomass. In all cases, the only sample-soecific information in the simulations is the ultimate and proximate analyses of the coals. No model parameters were rme-tuneo to match the predictions to measured values. COAL DEVOLATILIZATION BEHAVIOR Pnmary Devolatilization FLASHCHAIN™ predicts the devolatilization behavior of any coal at any operating conditions. Well over 1000 coals from an geographical regions worldwide have been analyzed to date Recent publications descnbe the theory's mechanisms to predict the yields of the maior volatile products, tar and gas. plus the yields of the gases containing oxygen and nitrogen (Niksa, 1995a). The theory was recently expanded to predict the release of all major hydrocarbons, H2, H 2S and the elemental compositions of char and tar. Consequently, Belmont, CA 94002, USA FLASHCHAIN™ now predicts the yields of char, tar CH4. C2H9, C2H4. C3H,, C3H9, HCN. H2S. H2, H20. CO. C02. plus the C/H/N/O/S compositions of tar and char and the molecular weight distribution of tar Table 1 Evaluation of predicted hydrocarbon and H2 /ieids for test conditions of Xu and Tomita (1989) Subbit Meas 2.00 0 78 0.49 0 40 (SY) Pred 300 1 36 090 0 31 Hv Bit Meas 3 20 1 43 0 85 0.40 (HV) Pred 350 1.58 1.05 0.39 Lv Bit Meas 3 90 0 89 0.37 0 50 KS) Pred 3 50 1 00 0 35 0 42 An evaluation of the predicted yields of hydrocarbons and H2 for subbituminous. hv bituminous and Iv bituminous coals appears in Table 1 These data were reported by Xu and Tomita (1989) for heating at 3000 K/s to 1037 K at atmosphenc pressure with a 4 s reaction period at the ultimate temperature. The predicted hydrocarbon yields are within 15 % of the measured values for the bituminous coals. but about 50 percent too high for the subbituminous coal Hydrogen yields are generally within experimental uncertainty for all coal types Predicted char compositions for hv bituminous coal chars prepared by rapid heating are evaluated in Fig. 1 In these tests, three similar hv bituminous coals were heated under atmosphenc pressure at 1000 K/s to progressively increasing temperatures with a 2 s reaction period at each temperature (Cai. 1995). The total weight loss is evident m Fig. 1 from the reported char yields. The char yields are generally predicted within experimental uncertainty throughout pyrolysis, except for a 5 % discrepancy with the Pit. 8 sample. The predicted char carbon contents increase throughout devolatilization, in accord with the tendency n ail three datasets. Predicted C-contents exceed the measured values m all cases, although the overestimation is significant only for the Gedling coal. The predicted H-contents are within experimental uncertainty throughout all stages of devolatilization with ail three coals. Predicted char-N levels are within exper'mertai uncertainty for all but the highest temperature with the L nby and Gedling coals. But predicted values exceed the data a' all temperatures for the Pit. 8. Whereas the predicted :~3r- N level exceeds the measured values at 1500 =C n this dataset, other datasets do not indicate a similarly ar~e extent of N-release under comparable conditions as explained previously (Niksa. 1995b). Although the char-0 levels were not measured, the predicted values appear n Fig. 1 to illustrate that oxygen (and sulfur) are -Ke pr • elements that are completely expelled dunng devoiatiiiza: zr Such behavior makes them suitable, in pnncipie 'zr " a c " the extent of devolatilization, although laboratory ce'er~-a-tions of char-0 are currently problematic. |