تجزیه و تحلیل حساسیت عوامل مختلف جنبشی برای مدل سازی عددی صربستان فرآیند تبخیرزدایی ذغال قهوه ای

Numerical modeling is widely used tool for prediction of combustion processes. Computational Fluid Dynamics – CFD models use three kinetic rates for description of the coal combustion processes: coal devolatilization, volatile combustion and char combustion. Reported rates for coal devolatilization vary considerably among the authors depending on the type of experimental systems used in describing the phenomenon. Accurate representation of devolatilization process is necessary in order to perform successful CFD calculations of pulverized coal combustion and gasification. The subject of this work is numerical modeling of Serbian lignite pulverized coal devolatilization in drop tube type laboratory scale reactor. The aim of this study is to evaluate the influence of different devolatilization kinetic factors on total devolatilization time in numerical modeling of pulverized Serbian lignite devolatilization. Nine different devolatilization kinetic rates mostly used in devolatilization numerical modeling are compared in the presented work.

Over the last 20 years CFD became powerful predictive and design tool in combustion research and development. CFD is commonly used for gaseous [1] and solid fuel (coal and biomass) burner’s utilities, [2] and [3], and for test furnaces design and optimization [4] and [5], as well as for investigation of novel pulverized coal combustion technologies [6], [7], [8] and [9]. With increase of computational power CFD is also being successfully applied for combustion modeling of full scale industrial furnaces and whole boilers [10], [11], [12], [13] and [14]. Devolatilization process plays an important role in pulverized coal combustion and gasification processes. Thus, accurate representation of devolatilization process is necessary in order to perform successful CFD calculations of pulverized coal combustion/gasification. Volatiles can account for up to 70% of the coal mass loss during combustion process, significantly increasing surrounding gas temperature in short period of time as a result of released volatiles combustion. Devolatilization has impact to combusting coal particle features from its injection to burnout. It influences particle ignition, trajectories, and eventual fragmentation as well as char intrinsic reactivity. The two main theoretical approaches are used in devolatilization modeling: network devolatilization models and empirical devolatilization models. Models from the first group describe devolatilization behavior of the coal by approximating the breakdown of the macromolecular coal network structure. While network devolatilization models offer detail information about volatile species evolution they are not commonly used in commercial CFD codes that become restrictively slow if large coal network matrix programs were included in main solver body [7] and [15]. Because of this, network devolatilization models are usually employed as pre-processor routines in order to calibrate simpler empirical devolatilization models used for CFD analysis [15]. It is important to underline that in several studies, despite their higher computational demand, network models were fully utilized to predict particle devolatilization behavior during CFD combustion modeling. As example, A. Silaen and T. Wang in their work [16] investigated influence of different turbulence and devolatilization models on coal gasification simulation in entrained-flow gasifier. They compared totally four different deviolatilisation models among which one was Coal Percolation and devolatilization (CPD) network model. Authors concluded that empirical single rate model and CPD model produce consistent devolatilization rates. Recently, Jovanovic et al. compared performance of different devolatilization models in predicting ignition point position during pulverized coal combustion in O2/N2 and O2/CO2 atmospheres with different compositions [17]. Authors used two empirical models (single rate and two competing rates) and two network models (CPD and Functional Group – FG). Although, in general, better agreement with experimental results was achieved using network models their high computational demand was pointed out. Second group models are empirical devolatilization models that utilize global kinetics for modeling of complex devolatilization processes. Arrhenius expressions are used to correlate rates of weight loss caused by devolatilization with temperature. Since empirical devolatilization models require significantly lesser computational resources in compare with network devolatilization models they are widely used in comprehensive CFD codes [18]. However, empirical nature of these models makes them difficult for use for fuels and heating rates beyond those for which Arrhenius parameters were derived. Comprehensive reviews of devolatilization kinetic rates were reported by several authors [19], [20] and [21]. Even though empirical devolatilization models require careful selection of relevant Arrhenius parameters for their successful application, reported results from relevant literature showed that when this condition is met they can predict particle devolatilization with satisfying accuracy. Hart et al. in their work modeled pyrolysis in lab-scale reactor at elevated pressures and high heating rates. Comparing numerical simulations with experimental data they showed that single reaction model could be appropriate for coal combustion CFD modeling as it as long as heating rates for the coal are comparable to the values found in industrial boilers [22]. Wendt and coauthors suggested numerical model for devolatilization and ignition of single coal particles. Particle devolatilization rate was calculated using empirical single rate model in this work. The obtained numerical results showed good agreement with the experiments for all investigated particle shapes [23]. Hashimoto et al. suggested novel devolatilization model – Tabulated-Devolatilization-Process (TDP). This model, similarly to other empirical devolatilization models (single rate model and two competing rates model), uses global kinetics based on Arrhenius expression to correlate particle temperature and mass loss. The main novelty lies in the fact that Arrhenius parameters, which are constant in case of single rate model and two competing rates model, change their values during simulation based on each particle temperature history. Arrhenius parameters database is prepared either based on experimental values or based on information obtained from network devolatilization models. Performed numerical simulations employing the TDP model are in better agreement with the experiments than that predicted by the other empirical models, with a slight increase in computation time [24] and [25]. Although significant efforts to determine the most appropriate Arrhenius kinetic parameters for number of different fuels and combustion conditions [17], [26] and [27] were performed, no information’s were found on influence of different devolatilization kinetic factors on overall combustion model for modeling of the Serbian lignite combustion. The subject of this work is numerical modeling of Serbian lignite pulverized coal devolatilization in drop tube type laboratory scale reactor. The aim of this study is to evaluate the influence of different devolatilization kinetic factors on total devolatilization time in numerical modeling of pulverized Serbian lignite combustion.

Sensitivity analysis of different devolatilization kinetics on pulverized coal devolatilization and ignition time was conducted. Totally nine different devolatilization kinetics and two different devolatilization models were used in this study. Inspection of obtained numerical results pointed out that different kinetic parameters produce very different coal devolatilization behavior. The quickest devolatilization rates were achieved using Badzioch and Hawksley devolatilization rates. Similar, rapid devolatilization rates were obtained using SANDIA, FG-tar, and Hu et al. kinetic parameters. These high devolatilization rates may be explained by the experimental conditions in which coal particles were examined, i.e. vertical reactor with rapid heating of entrained coal particles. Somewhat slower devolatilization rates were produced using Ubhayakar et al. and Stojilkovic’s kinetic. In case of Ubhayakar and coauthors work, slower devolatilization rates are probably produced due to assumption under which kinetic parameters were derived. Namely, it was assumed that devolatilization starts at relatively high temperatures ∼700 K. Slower devolatilization rates obtained using Stojiljkovic Arhennius kinetics can be explained by influence of slow heating rates used in her work. Jankes et al. kinetical parameters produce two different devolatilization rates: fast at low temperatures and slow at high temperatures. This feature is especially appreciable for small particle sizes (90 and 200 μm). The slowest devolatilization rates were obtained using Kobayashi kinetics. In this case it was almost impossible to have stable devolatilization, even for the smallest particles. So slow devolatilization may be caused by authors’ assumption that devolatilization starts at quite high temperatures (more than 850 K). Numerically obtained devolatilization rates were validated against two different sets of experimental data, one derived at rapid heating rates (ignition delay time) and the second set derived at low to moderate heating rates (total volatile yield at different devolatilization temperatures). The best agreement with experimentally obtained ignition delay time is obtained using fast devolatilization models, [26], [36], [37] and [38], developed for high heating rates similar to those used in experiments. It should be mentioned that only single rate model with temperature dependent kinetic parameters by Jankes et al. is able to accurately predict ignition delay time for the biggest, 2000 μm, particles due to experimental conditions under which they were derived. Moreover, it should be noted that experimental values were obtained investigating single particle devolatilization, while in reality, as well as in the proposed numerical model coal particle devolatilise in clouds which could be possible cause in difference between numerical and experimental values. Values for numerically calculated total volatile yield were close to experimentally determined in case of slower devolatilization models, [26] and [39], as it is expected since experimental data used for comparison were obtained using slow to moderate heating rates as it was already explained above. However all applied devolatilization kinetic parameters except those suggested by Jankes et al. over predict total volatile yield at high temperatures. Arrhenius kinetic parameters by derived by Jankes and co-authors are temperature dependant assuming faster devolatilization rates at low temperatures and slower rates in high temperature region thus capturing primary and secondary devolatilization behavior of coal paricles. Based on the previous conclusion it was decide to build entrained flow combustor able to operate at high heating rates and with desired oxidizer composition at desired gas temperatures. Numerical and experimental investigation of Serbian lignite coals devolatilization and combustion will be performed in this reactor, using valuable conclusions from this work.