Comparative Study on Different Treatments of Coal Devolatilization for Pulverized Coal Combustion Simulation

Abstract

Currently, volatile matter is generally treated as a postulate substance or a mixture of light gases and tar with given proportion in pulverized coal combustion (PCC) simulation. Whether those treatments can well characterize the PCC or not remains unknown. Here, current different coal devolatilization treatments are numerically evaluated under the configuration of laminar stagnation PCC (Xia, M.; et al. Proc. Combust. Inst. 2017, 36, 2123–2130). The results show that the one-step model fitted from the Chemical Percolation Devolatilization model prediction together with detailed components and mechanism could give the best prediction of the PCC characteristics. Different detailed volatile components give similar predictions on the gas and particle properties with a slight difference in the CO mass fraction due to their different components fractions. The simplified global mechanism cannot reproduce the PCC characteristics such as those of the detailed mechanisms, with earlier particle ignition and much higher gas temperature and mass fractions of production species. Assuming the volatile matter is pure methane is not reasonable with obvious discrepancies on the particle and gas properties compared with those of the optimal treatments even when a detailed mechanism is used. Devolatilization model also plays a significant role with the two-step model giving a lower flame location, lower NO x , and higher polycyclic aromatic hydrocarbons emissions compared with those of the optimal treatment. The generality of the above conclusions will be further verified on more realistic configurations in the future, such as the benchmark coal flames in the Coal and Biomass Conversion (CBC) workshop.

Large eddy simulation of Cambridge bluff-body coal (CCB2) flames with a flamelet progress variable model

“…The OH contour of the primary gas flame shows similar distribution as the experimental data with the same angle but under-predicted length. Those deviations can be attributed to the boundary treatment of assuming pilot methane/air gas as a mixture of volatiles and carrier gas in the model implementation, which would give quite different predictions of OH mass fraction, especially when heavy hydrocarbon is introduced into the volatile species [31,32]. This can be resolved by extending the present model to a three-mixture-fraction FPV model [33], which, however, is beyond the scope of the present paper.…”

“…While an under-prediction of U z be found at outer radial regions (r>20 mm), U y matches well with the experimental data at x=2 and 10 mm, while is obviously under-predicted at x=20 and 30 mm. This indicates a compact flame located near the centreline, which can be attributed to the boundary treatment of assuming pilot gas as a mixture of volatiles and carrier gas, resulting in an under-prediction of flame angle and length, especially when heavy hydrocarbon is introduced in the volatile components [31,32]. RMS velocities at the outer radial region (10mm<r<20mm) are under-predicted, which may result from the under-prediction of the angle of the pilot flame and viscosity increasing caused by the pilot flame as well as the experimental uncertainty.…”

Hybrid Flamelet/Progress Variable Approach for NO Prediction in Pulverized Coal Flames

“…25 The pulverized coal flame structure obtained with volatiles, including large hydrocarbon species representing tar, could be largely reproduced by the volatiles composed of several light hydrocarbon species. 22,26 Thus, the assumption of light hydrocarbon volatiles omitting tar is applied in the present work. In addition, all fuel-N is assumed to exist in volatiles by neglecting char-N. Generally, both HCN and NH 3 are two commonly used intermediate species for describing volatile-N conversion.…”

Mechanistic Investigations of Particle Ignition of Pulverized Coals: An Enhanced Numerical Model and Experimental Observations

“…Some scholars have used the CPD devolatilization model results to improve the accuracy of the prediction and reduce the difficulty of modeling. Yin and Yan obtained the kinetic parameters of a one-step devolatilization model, while Lou et al and Xing et al developed a two-step CPD-based devolatilization model. They then used a global devolatilization model to predict the pyrolysis behavior of fuels.…”