Many studies have been performed to investigate the combustion behaviour for different types of coal [123], as each of them is unique in terms of their combustion process. Recent studies have demonstrated this in various ways, either through experiment [124] or numerical simulation [125], with the aim of better understanding as well as characterising the processes of coal utilisation. One of the important parameters is ignition delay and this was introduced in section 4.4.2. This characteristic is very
important for designing coal combustion systems as it has significant roles in the prediction of spontaneous ignition and in the production of stable flames [126].
In the experimental study by Levendis et al. [105, 121] reported that the lower rank (lignite) coal has the shorter ignition delay time (tid) compared to the higher rank
(bituminous) coal [105]. This result generally agrees with the studies of several other authors [126-128], and further indicates that the ignition delay time increases from a lower to higher rank coal. In addition, Young et al. [129] reported that lignite coal is more reactive than other types of coal.
The ignition delay was the lapse of time after the coal was injected until the burning occurred. However, the devolatilization reaction of coal initiates the process of combustion [40, 130], therefore potentially linking with the ignition delay. Numerical studies of bituminous coal particle combustion have been performed in this study [131- 133]. The numerical model has been validated by the experimental study by Levendis et al. [105], which was specifically based on the results of the ignition delay time (tid), char
burn out time (tchar), maximum temperature of coal volatility combustion (Tcv), and
maximum char temperature (Tchar) [133, 134]. This section investigates the
devolatilization reaction and how it influences the ignition delay time. A comparison of the ignition delay time was carried out between bituminous and lignite coals, representing, respectively, a high and low ranked coal since they have significantly different chemical compositions. Results could give a better understanding of the devolatilization reaction for further modelling applications.
By using the same procedures, the combustion model of lignite coal (PSOC 1443) in the DTF reactor is developed and the ignition delay time between the results of simulation and experiment is assessed. The devolatilization reaction process is simulated initially with the kinetic parameters of R1 in Table 4-8. This model simulation allows the process of devolatilization to be simulated either by including, or excluding, the process of combustion of coal volatile species. Therefore, the devolatilization process can be simulated independently, or even simultaneously with the other reactions referred to Table 4-2. For identification, the simulation process of PSOC 1443 (lignite coal) combustion with the kinetic parameters in Table 4-8, is named as Simulation A. Other simulations, named respectively accordingly as Simulations B, C and D, are developed
as a part of the investigation. The simulation results of the model devolatilization process can be seen in Figure 4-25, where they do not have any combustion of coal volatile matter. As part of the investigation, the same process of each simulation with the coal volatility matter burning can be seen in Figure 4-26.
Figure 4-25. Devolatilization reaction process without combustion
Figure 4-26. Devolatilization reaction with combustion
Figure 4-25 and 4-26 show the process of devolatilization in terms of the coal volatile fraction profile. Figure 4-25 presents the devolatilization process without volatile
combustion while Figure 4-26 presents it with combustion. The devolatilization process of Simulation A lasts between ~20 and ~40ms with the most rapid coal volatility release occurring at ~30ms, as seen in Figure 4-25. If it is performed with combustion, as in Figure 4-26, the peak of coal volatile profile occurs also at ~30ms, but then it goes down, which indicates its burning out. However, the coal volatile combustion initiates the combustion of coal particles, so at the time when the most rapid combustion occurred, the temperature of the coal particle increased rapidly and initiated its burning. The period between the particle injection and the particle starting to burn is the ignition delay time. Therefore, the ignition delay of Simulation A is determined as ~30ms after the coal injection, but this result does not agree with the experiment [105, 106], and therefore Simulation B, C and D are developed by systematically increasing the pre-exponent factor (A). It should be noted that the reactor condition is the same for each simulation (heat rate and temperature), so the activation Energy (Ea) and temperature exponent (𝛽) are
assumed to be the same. The value of the pre-exponent factor of Simulations B, C, and D is increased 10, 100 and 300 times that of Simulation A, respectively. Finally, the results indicated that the best fit result of the ignition delay time was that obtained by Simulation D. Simulation D took ~10ms, which agrees with the ignition delay time for the lignite coal PSOC 1443 in the experiment [105]. This further indicates that the kinetic parameter value of Simulation D is suitable for the lignite coal combustion. The comparison of PSOC 1443 coal particle in a char fraction for each simulation can be seen in Figure 4- 27.
Figure 4-27 further shows that, from simulation A to Simulation D, the ignition delay time decreases. This is because of the increase in the kinetic parameter of devolatilization reaction. Simulation D agrees best with the experiment and it indicates the best fit value of kinetic parameters for PSOC 1443 coal particle combustion.
A comparison between the results of experiments and simulation for coal PSOC 1443 can be seen in Table 4-9.
Table 4-9. Parameter comparison between the experiments and simulation PSOC 1443 Max Temperature (K) tid (ms) Total Burn
out (ms)
Experimental 2000 10 72
Deviation [105, 106] 93 - 15
Simulation 2042 10 71
Table 4-9 shows all parameters of simulation results are in the limit of tolerance according to the references [105].