Feasibility Study of Co-Firing of Torreﬁed Empty Fruit Bunch and Coal through Boiler Simulation

: Torreﬁed empty fruit bunch (EFB) co-ﬁring is a promising technology to reduce emissions from coal-ﬁred power plants. However, co-ﬁring can inﬂuence the combustion and heat transfer characteristics in a coal boiler. In order to study the feasibility of co-ﬁring application of torreﬁed EFB (T-EFB) in boilers, the combustion characteristics, gas emissions and heat ﬂux distribution were analyzed, respectively. First, the kinetic parameters of T-EFB devolatilization and char oxidation were obtained by experimental analysis. Second, the computational ﬂuid dynamics (CFD) analysis was applied to the actual 500 MWe boiler simulation to further evaluate the di ﬀ erences in the co-ﬁring performance parameters (combustion characteristics and emissions) of the T-EFB and the heat transfer characteristics within the boiler. Numerical results show that T-EFB co-ﬁring can improve the ignition characteristics of pulverized coal, reduce the formation of unburned particles. When the blending ratio was increased from 10% to 50%, signiﬁcantly NO x (oxides of nitrogen) reduction (levels from 170 to 98 ppm at 6% O 2 ) was achieved. At a blending ratio above 40%, boiler combustion e ﬃ ciency decreases as the total heat ﬂux of the boiler decreases due to an increase in the amount of unburned carbon. In addition, T-EFB co-ﬁring can a ﬀ ect the heat transfer characteristics of the boiler.


Introduction
Globally, issues such as climate change and the depletion of fossil fuels have promoted the development and utilization of biomass renewable energy to reduce the dependence of power plants on fossil fuels [1]. At the same time, new technologies to convert raw biomass into more beneficial biomass fuels were significantly promoted [2].
Demand for palm oil has increased as economic growth that led to large-scale palm cultivation in Indonesia and Malaysia. The other reason is that palm act as the dominant agricultural crop in developing countries and the two mentioned countries produce more than 80% of the global palm oil supplement. According to data released by the Indonesian Ministry of Agriculture, the area under palm cultivation has increased 2.3 times from 2010 to 2020 [3]. At the same time, the mass production of palm oil generates a large amount of solid waste, including empty fruit bunches (EFB), palm kernel shells (PKS) and palm pressed pericarp fibers [4,5]. Generally, part of the EFB is directly used as fuel to power palm oil processing plants, part as organic fertilizer, and the remaining part is discarded or incinerated. It has attracted significant attention and interest in utilizing these industrial wastes efficiently.
General as a fuel, biomass combustion has lower input cost and technical research shows more complete than other biomass heat conversion technologies. Currently, the utilization of biomass as an alternative fuel to partial coal and added to combustion systems for co-firing has gradually developed This study aims to explore the feasibility of T-EFB mixed with coal for application in large boilers. In this research, the effect of torrefaction on the properties of EFB fuel was first analyzed. Then, thermogravimetric analysis (TGA) was conducted to determine the kinetic reaction parameters of the T-EFB and the flaring behavior of pyrolytic biomass residues. Finally, the combustion characteristics, contaminant generation and the effect on heat transfer in the boiler at different mixing ratios (10-50%) were analyzed by modeling the actual boiler. The study provides positive reference data for the actual operation of thermal power plants. Figure 1 shows a conceptual graph of T-EFB co-firing system. As shown, the raw EFB (R-EFB) is torrefied an inert atmosphere, in which the reaction is carried out at a temperature of 200-300 • C. This reaction usually requires some minutes to 2 h to improve the calorific value and property of the R-EFB. The obtained T-EFB and coal fuel were ground to a specific particle size required for boiler combustion. Then, the two kinds of fuels were mixed uniformly and were injected into the boiler through a burner for combustion.

Torrefaction-Based Co-Firing System
Energies 2020, 13, x 3 of 27 contaminant generation and the effect on heat transfer in the boiler at different mixing ratios (10-50%) were analyzed by modeling the actual boiler. The study provides positive reference data for the actual operation of thermal power plants. Figure 1 shows a conceptual graph of T-EFB co-firing system. As shown, the raw EFB (R-EFB) is torrefied an inert atmosphere, in which the reaction is carried out at a temperature of 200-300 °C. This reaction usually requires some minutes to 2 h to improve the calorific value and property of the R-EFB. The obtained T-EFB and coal fuel were ground to a specific particle size required for boiler combustion. Then, the two kinds of fuels were mixed uniformly and were injected into the boiler through a burner for combustion.

Material Properties
The EFB biomass material used in the experiment was an Indonesian R-EFB obtained from the Korea Institute of Energy Research (KIER). T-EFB is obtained by torrefying R-EFB using TGA in an inert atmosphere and a temperature of 250-300 °C for 10 min. Moreover, in this study, the coal used from the actual operating power plant. As an energy source, EFB has high power generation potential in waste biomass. The basic properties of each fuel are listed in Table 1. It contains results of a proximate analysis (TGA-701, LECO Co., Anyang, Korea), ultimate analysis (Tru. Spec. elemental analyzer and SC-432DR sulfur analyzer, LECO Co., Anyang, Korea) and calorific value (AC600 semiauto calorimeter, LECO Co., Anyang, Korea) for all samples. Each analysis was based on American Society for Testing and Materials (ASTM) D3172, D3176 and D4326, respectively. In addition to bituminous coal, the possibility of utilization of R-EFB, and T-EFB gradually increased as the fuel heating value increased. At the same time, compared with R-EFB, the volatile matter of T-EFB was further reduced after torrefaction at 250 °C while having a higher carbon content, but the ash content was increased.

Material Properties
The EFB biomass material used in the experiment was an Indonesian R-EFB obtained from the Korea Institute of Energy Research (KIER). T-EFB is obtained by torrefying R-EFB using TGA in an inert atmosphere and a temperature of 250-300 • C for 10 min. Moreover, in this study, the coal used from the actual operating power plant. As an energy source, EFB has high power generation potential in waste biomass. The basic properties of each fuel are listed in Table 1. It contains results of a proximate analysis (TGA-701, LECO Co., Anyang, Korea), ultimate analysis (Tru. Spec. elemental analyzer and SC-432DR sulfur analyzer, LECO Co., Anyang, Korea) and calorific value (AC600 semiauto calorimeter, LECO Co., Anyang, Korea) for all samples. Each analysis was based on American Society for Testing and Materials (ASTM) D3172, D3176 and D4326, respectively. In addition to bituminous coal, the possibility of utilization of R-EFB, and T-EFB gradually increased as the fuel heating value increased. At the same time, compared with R-EFB, the volatile matter of T-EFB was further reduced after torrefaction at 250 • C while having a higher carbon content, but the ash content was increased. As shown in Figure 2, the Van Krevelen diagram [29] shows the correlation between the atomic H/C and O/C ratios of coal, R-EFB and T-EFB obtained by torrefaction at 250 • C. Combined with the data in Table 1, we know that the energy density of R-EFB is much lower than that of coal. However, the T-EFB after torrefaction gradually reached the level of sub-bituminous coal, which means that low-rank biomass fuel can be increased by torrefaction technology to reach the standard of high-rank fuel, especially lower calorific value R-EFB [13].  As shown in Figure 2, the Van Krevelen diagram [29] shows the correlation between the atomic H/C and O/C ratios of coal, R-EFB and T-EFB obtained by torrefaction at 250 °C. Combined with the data in Table 1, we know that the energy density of R-EFB is much lower than that of coal. However, the T-EFB after torrefaction gradually reached the level of sub-bituminous coal, which means that low-rank biomass fuel can be increased by torrefaction technology to reach the standard of high-rank fuel, especially lower calorific value R-EFB [13].

Experimental Studies on TG and DTG and Kinetics Tests of T-EFB Biomass
To determine the changes in the T-EFB reaction characteristics after torrefaction, we performed a thermogravimetric analyzer (TGA, SDT Q600) test under inert nitrogen conditions and atmospheric conditions. In total, 15 mg of R-EFB and T-EFB were taken for the analysis. First, the two forms of matter

Experimental Studies on TG and DTG and Kinetics Tests of T-EFB Biomass
To determine the changes in the T-EFB reaction characteristics after torrefaction, we performed a thermogravimetric analyzer (TGA, SDT Q600) test under inert nitrogen conditions and atmospheric conditions. In total, 15 mg of R-EFB and T-EFB were taken for the analysis. First, the two forms of matter were heated from room temperature at a constant heating rate (20 • C/min) to 107 • C in a nitrogen atmosphere at a flow rate of 100 mL/min. An isothermal time of approximately 10 min was maintained Energies 2020, 13, 3051 5 of 23 to evaporate the moisture completely. Then, during the pyrolysis process, the temperature was heated from 107 • C to 950 • C at a constant heating rate (20 • C/min) under a helium atmosphere. The matter was kept in an isothermal situation for 10 min, and real-time weight loss data could be obtained according to the pyrolysis of the samples. The kinetics of the two biomasses in the devolatilization process were observed. Second, the air was used as the oxidant at a flow rate of 100 mL/min, and the temperature was heated to 950 • C at a constant heating rate (20 • C/min). The same temperature was maintained for 10 min to observe the oxidation kinetics of char.

Boiler Information
The study was conducted in a 500 MWe tangential firing boiler with configuration and dimensional parameters, as shown in Figure 3a. The boiler is a tangential spray one-through boiler that achieves a 500 MWe load, and the heat receiver is horizontally located above the boiler. The efficiency of the boiler is 89.3%. Figure 3a shows the shape of the boiler. The height was 86.6 m, and the horizontal-section width and depth of the furnace were both 16.5 m. As shown in Figure 3b, the pulverized coal and the T-EFB particles were delivered by primary air to five burners (AB, CD, E and F (un-operated burner)) in the burner zone of the furnace. As shown in Figure 3c, the fuel and air ejected through the four corners form clockwise concentric combustion in the center of the furnace. Meanwhile, as the auxiliary combustion air, the secondary air directly entered the furnace through the nozzle near the burner and part of the air passed through the overfire air (OFA) port above the main combustion area, which played an active role in the burning of the fuel. The coal-fired burners used the pollution minimum (PM) burners. Each group of burners had two fuel injection ports, and the flow ratio of the fuel-rich (conc) nozzles to the fuel-lean (weak) was 7:3. Then, there were an economizer (Eco), two reheaters (RH), and three superheaters (SH) in the main flow direction at the top of the furnace outlet zone. From the hopper inlet to the boiler exit, the shape of the water wall was simplified. The combustion performance of a boiler using pulverized coal had a close relation to the operating conditions of the pulverized coal, OFA distribution and the air supply conditions of the burner. When constructing the model, the effective area of the actual burner nozzle was thoroughly considered.

Numerical Simulation
The commercial code ANSYS FLUENT (version 17.1) used in this study performs numerical simulations including turbulence, solid particle motion, particle-gas interactions, combustion and radiative heat exchange, and contaminant NO x formation in boilers [30]. CFD analysis can provide cost-effective solutions for new combustion systems [31]. A k-ε-realizable turbulence model was chosen to calculate the effect of turbulence in the boiler [30,32]. Coal and biomass reactions were modeled using a species transport model. The trajectories of pulverized coal and biomass particles were calculated by the Lagrangian method. Based on the literature [3,25], this study considered the drag force that was crucial for the orbital calculation of discrete particles. The biomass particles were assumed to be nonspherical and larger than the spherical surface area of the coal and the drag coefficients were calculated based on Haider and Levenspiel's equation regarding to [6]. The coal particle is assumed to obey the rosin-rammler distribution with a particle size of 10-150 µm, an average of 75 µm and a distribution index of 1.15, respectively. However, in this study, the particle size of the T-EFB sample taken in the experiment was 400 µm, so the same particle size was also used in the simulation. Discrete ordinate models [33] were used to simulate radiation heat transfer effect, which approximates the solution of radiative transfer equations by discretizing the entire solid angle into a finite number of solid angles. A weighted-sum-of-gray-gases model [6,32] was applied to calculate the absorption coefficient of the gas. Both the particle emissivity and scattering coefficient used a default value of 0.9 [31]. A finite-rate/eddy-dissipation model [25,26,34] are used to calculate the volumetric homogeneous reaction. The fuel particle and gas-phase reactions were simplified into two stages, and the reactions were as follows: In this study, since the volatile combustion of torrefied biomass pellets occurs over a short period of time, the devolatilization model used a single-rate model [35]. The fuel volatile fraction single rate reaction model is shown below.
where m 0 (kg) is the initial volatile fraction mass, m (kg) is the post-reaction volatile fraction mass, T p (K) is the particle temperature, R is the universal gas constant (8.3144 J mol −1 K −1 ), A is the pre-exponential factor (1/s) of the kinetic parameters, and E is the reaction activation energy (kJ/mol). The kinetic/diffusion limitation model is used for surface charcoal combustion. The rate of combustion reaction depends on the diffusion of the particle surface and the kinetics of the inherent chemical reaction [30,34]. The kinetic parameters of coal were referenced from the same type of coal [36] based on the simulation accuracy considerations in this study, while the kinetic parameters of T-EFB biomass were obtained from the experiments in this study. Detailed kinetic data for these two fuels are presented in Table 2. In this study, only fuel NO and thermal NO were considered, as prompt NO was very low. Posttreatment methods have been widely used to simulate NO x generation in recent years, and by comparing field tests and simulation results, posttreatment NO x calculation models can accurately simulate NO x generation [37,38]. Among them, thermal NO refers to the formation of NO in the atmosphere by oxidizing N 2 in the relatively high-temperature region of the combustion reaction, the reaction rate of which was derived from the experimental studies of Hanson and Salimian [39], and the reaction had a certain dependence on high temperature. Heat formation is determined by a series of chemical reactions that are highly temperature-dependent (called the extended Zeldovich mechanism). The main reactions that control the formation of thermodynamic NO from molecular N are as follows.
Fuel NO accounts for the largest share of NO x from coal combustion. Fuel NO consists mainly of the chemical bond cleavage of nitrogen in the solid fuel fraction, which is formed by the combination of oxygen and some of the fuels formed in the devolatilization phase. This is described using the De Soete model [40] and the reaction mechanism is shown in Figure 4. Furthermore, based on the comparison of simulations and experiments in the literature, the conversion rate of HCN and NH 3 emitted by volatile constituents is assumed to be 9:1 for bituminous coal [41]. Whereas in the case of torrefied biomass, the conversion rate of HCN and NH 3 was assumed to be 1:9 based on data from the literature [42], and the char of both was completely oxidized to NO. Other calculation submodel descriptions used in the study can be found in the literature [13,42]. In this study, only fuel NO and thermal NO were considered, as prompt NO was very low. Posttreatment methods have been widely used to simulate NOx generation in recent years, and by comparing field tests and simulation results, posttreatment NOx calculation models can accurately simulate NOx generation [37,38]. Among them, thermal NO refers to the formation of NO in the atmosphere by oxidizing N2 in the relatively high-temperature region of the combustion reaction, the reaction rate of which was derived from the experimental studies of Hanson and Salimian [39], and the reaction had a certain dependence on high temperature. Heat formation is determined by a series of chemical reactions that are highly temperature-dependent (called the extended Zeldovich mechanism). The main reactions that control the formation of thermodynamic NO from molecular N are as follows.
Fuel NO accounts for the largest share of NOx from coal combustion. Fuel NO consists mainly of the chemical bond cleavage of nitrogen in the solid fuel fraction, which is formed by the combination of oxygen and some of the fuels formed in the devolatilization phase. This is described using the De Soete model [40] and the reaction mechanism is shown in Figure 4. Furthermore, based on the comparison of simulations and experiments in the literature, the conversion rate of HCN and NH3 emitted by volatile constituents is assumed to be 9:1 for bituminous coal [41]. Whereas in the case of torrefied biomass, the conversion rate of HCN and NH3 was assumed to be 1:9 based on data from the literature [42], and the char of both was completely oxidized to NO. Other calculation submodel descriptions used in the study can be found in the literature [13,42].

Boundary Conditions and Simulation Scenarios
In this study, all simulated cases were based on field measurements for full load (500 MWe) operation (i.e., case 1 in Table 3). Detailed performance data for the boiler operation can be confirmed in Table 3. The amount of air required for simulated combustion is calculated separately based on the T-EFB mixing ratio of 10-50% (thermal basis), but the air excess ratio is 1.12. During the simulation, fuel is injected uniformly into 24 burners (A-E burner area) and the relative ratios of primary air (PA), secondary air (SA) and OFA are kept the same with pure coal combustion conditions. In addition, as temperature, carbon content concentration in the fly ash, NOx emission and other data were also measured at the monitoring port at the boiler exit zone. The measurement results data can be confirmed in Table 4. Normally, the emissivity of the boiler water wall was set at 0.6 and the temperature was 427 [43]. The calculation case is based on the velocity-pressure coupling using the SIMPLE algorithm and the second-order upwind format to improve the calculation accuracy.

Boundary Conditions and Simulation Scenarios
In this study, all simulated cases were based on field measurements for full load (500 MWe) operation (i.e., case 1 in Table 3). Detailed performance data for the boiler operation can be confirmed in Table 3. The amount of air required for simulated combustion is calculated separately based on the T-EFB mixing ratio of 10-50% (thermal basis), but the air excess ratio is 1.12. During the simulation, fuel is injected uniformly into 24 burners (A-E burner area) and the relative ratios of primary air (PA), secondary air (SA) and OFA are kept the same with pure coal combustion conditions. In addition, as temperature, carbon content concentration in the fly ash, NO x emission and other data were also measured at the monitoring port at the boiler exit zone. The measurement results data can be confirmed in Table 4. Normally, the emissivity of the boiler water wall was set at 0.6 and the temperature was 427 [43]. The calculation case is based on the velocity-pressure coupling using the SIMPLE algorithm and the second-order upwind format to improve the calculation accuracy.

Comparison of R-EFB and T-EFB Combustion Characteristics with TG/DTG
To compare the thermal properties of the R-EFB and the T-EFB, we performed TG (Weight Loss Curves) and DTG (Derivative Thermogravimetric) tests on the two biomasses at a heating rate of 20 • C/min in TGA. Figure 5 shows the TG analysis curve and the DTG analysis curve, respectively. As shown in Figure 5, the two biomasses were dehydrated in the temperature range of 50-150 • C. The comparison showed that the R-EFB mass loss was higher than that of the T-EFB, which directly reflects the torrefaction technology that can remove more moisture from the R-EFB containing high moisture.
The initial degradation temperature of the birth material can be judged from the initial point of the drop in the TG curve. According to Novianti S. et al. studies [44], the initial degradation temperature of hemicellulose and cellulose in the biomass sample occurred between 170 • C and 250 • C. As shown in Figure 5a, the R-EFB after basic drying had the highest mass loss compared to T-EFB, reaching 78%, probably owing to the higher volatile content. Generally, hemicellulose was decomposed at less than 350 • C cellulose was degraded between 250 • C and 500 • C, and residual lignin was degraded at around 400 • C [45]. In the DTG curve, a small subpeak of the R-EFB was observed at about 295 • C, but the T-EFB did not form a subpeak, indicating that the R-EFB changed the hemicellulose structure during torrefaction. At the same time, owing to the thermal decomposition of cellulose, the main peaks of the T-EFB and the R-EFB after torrefaction were observed at around 360 • C. Because the cellulose peak of T-EFB was higher than that of R-EFB, the T-EFB reaction after torrefaction was less activated than the R-EFB because the R-EFB properties after torrefaction were closer to those of sub-bituminous coal.

Comparison of R-EFB and T-EFB Combustion Characteristics with TG/DTG
To compare the thermal properties of the R-EFB and the T-EFB, we performed TG (Weight Loss Curves) and DTG (Derivative Thermogravimetric) tests on the two biomasses at a heating rate of 20 °C/min in TGA. Figure 5 shows the TG analysis curve and the DTG analysis curve, respectively. As shown in Figure 5, the two biomasses were dehydrated in the temperature range of 50-150 °C. The comparison showed that the R-EFB mass loss was higher than that of the T-EFB, which directly reflects the torrefaction technology that can remove more moisture from the R-EFB containing high moisture. The initial degradation temperature of the birth material can be judged from the initial point of the drop in the TG curve. According to Novianti S. et al. studies [44], the initial degradation temperature of hemicellulose and cellulose in the biomass sample occurred between 170 °C and 250 °C. As shown in Figure 5a, the R-EFB after basic drying had the highest mass loss compared to T-EFB, reaching 78%, probably owing to the higher volatile content. Generally, hemicellulose was decomposed at less than 350 °C cellulose was degraded between 250 °C and 500 °C, and residual lignin was degraded at around 400 °C [45]. In the DTG curve, a small subpeak of the R-EFB was observed at about 295 °C, but the T-EFB did not form a subpeak, indicating that the R-EFB changed the hemicellulose structure during torrefaction. At the same time, owing to the thermal decomposition of cellulose, the main peaks of the T-EFB and the R-EFB after torrefaction were observed at around 360 °C. Because the cellulose peak of T-EFB was higher than that of R-EFB, the T-EFB reaction after torrefaction was less activated than the R-EFB because the R-EFB properties after torrefaction were closer to those of sub-bituminous coal.

Grid Independence and Validation of the CFD Simulation
Grid independence studies were performed to the accuracy of the data results. we constructed three high-quality grids, each consisting of approximately 1.39 million, 1.7 million and 2.01 million hexahedral units. By comparing the gas temperature and the horizontal velocity component Vy, the horizontal cross section of the main combustion zone was along the centerline (x = 0-16.5 m, y = 8.25 m and z = 15.7 m). As can be seen from Figure 6, the simulation results with the three grids tended to be similar, but owing to the fineness of the grid, the 1.7 million and 2.01 million grids could accurately express the accuracy of the results. In addition to considering the economics of the time required for the simulation calculations, 1.7 million grids were eventually used for this simulation study.

Grid Independence and Validation of the CFD Simulation
Grid independence studies were performed to the accuracy of the data results. we constructed three high-quality grids, each consisting of approximately 1.39 million, 1.7 million and 2.01 million hexahedral units. By comparing the gas temperature and the horizontal velocity component V y , the horizontal cross section of the main combustion zone was along the centerline (x = 0-16.5 m, y = 8.25 m and z = 15.7 m). As can be seen from Figure 6, the simulation results with the three grids tended to be similar, but owing to the fineness of the grid, the 1.7 million and 2.01 million grids could accurately express the accuracy of the results. In addition to considering the economics of the time required for the simulation calculations, 1.7 million grids were eventually used for this simulation study.
Comparing the simulated data with the experimental data obtained by the power station, the temperatures are 333.5 • C and 353 • C, respectively. The simulated oxygen concentration is almost identical to the experimental data. At the same time, the simulated and experimental NO X yields are 170 ppm and 159 ppm (at 6% O 2 ), respectively, with an error of only 6.5%. The UBC (Unburned Carbon) content simulation results and experimental data were 3.48 and 3.21, respectively, with an error of 8%. Therefore, since the UBC content is slightly less, the NO X generation increases. At the same time, compared with the NO X and UBC obtained by the DTF experiment, the NO X obtained by the DTF experiment was reduced by 28.6% compared with the pure coal combustion at the time of 30% co-firing, which is almost identical to the amount of NO X (28.2%) reduced by the case corresponding to the boiler simulation. At the same time, the number of SO 2 (Sulfur Dioxide) produced was 474.2 and 476.1 ppm (at 6% O 2 ), respectively, and the error was less than 0.4%. In addition, in Table 4, the heat absorption data of the tubes (superheater, reheater and economizer) in the upper part of the furnace obtained by simulation and experiment can be compared, and the maximum error is not more than 8%. Among them, the total endothermic value error is only 1.8%. Therefore, the above simulation results are an excellent example of the boiler combustion and pollutant emission characteristics of this study. At the same time, all the above experimental data provide a forceful verification of numerical simulation, indicating that the grid and calculation model selected in this study can accurately simulate the combustion characteristics of coal and T-EFB biomass in boilers.

Effects of T-EFB Co-Firing Substitutions on the Combustion Characteristics
The effect of T-EFB co-firing on the combustion characteristics of the boiler was analyzed by simulation. Figure 7 shows the temperature distribution of the vertical section of the boiler for T-EFB co-firing with different blending ratios. Apparently, the simulation results show the same trend in the gas temperature distribution inside the boiler under various calculation conditions. The T-EFB has similar characteristics to sub-bituminous coal, so T-EFB pretreated by torrefaction technology revealed replace some of the coal for mixing, without affecting the combustion stability of the boiler.
Based on the temperature distribution profiles along the furnace height (from the bottom of the ash to the furnace outlet) under various T-EFB blending ratio in Figure 8a, it can be seen that the gas temperature distribution is similar for each calculation conditions, but there is some temperature difference between the main combustion zoon and the OFA burnout zoon. In the main combustion zoon, the peak temperature shows a decreasing trend as the amount of biomass added increases. According to the peak temperature data in Table 4, when the T-EFB mixing ratio reaches 50%, the peak temperature of the furnace was 25 • C lower than the highest temperature produced by the pure coal combustion. Finally, there was a clear tendency for the gas temperature decrease with Increased T-EFB substitution in the OFA burnout zoon, due to the different degree of burnout of unburned particles. Compared to the pure coal case, the average temperature under co-firing conditions was lower in the OFA burnout zone which may result from a high percentage of pulverized coal, delayed ignition in the main combustion zoon, low burnout rate and the oxidation reaction of unburned fuel in this zoon result in high temperatures. As shown in Figure 8a, the gas temperature near each burner increases slightly as the T-EFB co-firing ratio increases. The main reason is that biomass has a higher reaction rate than coal dust and contains a large amount of volatile matter, which generates much heat from combustion and conducive to improving the ignition and combustion characteristics of coal. At the same time, the heat generated by the combustion of volatile fractions of matter accelerates the rate of gas rise and shortens the particle residence time. Figure 8b shows the average O 2 concentration curve along the furnace height under various T-EFB blending ratio. In each case, higher O 2 concentrations were displayed near the burners in the main combustion zone but were gradually depleted as the combustion reaction occurred. Compared with pure pulverized coal combustion, T-EFB fuel can devolatilize in a short period of time, and combustion produces high temperature to promote pulverized coal ignition, so with the co-firing ratio, O2 concentration consumption increases, and the O 2 residual concentration was decreased in the main combustion zone. When comparing Case 1 and Case 6, there is a 1.5% difference in oxygen concentration near the burner zones. Therefore, pulverized coal combustion is under high temperature and low oxygen reduction conditions, which will suppress the production of fuel-based NO x in the main combustion zone.   The iso-surfaces of Figure 9 show the local high-temperature zones of 1476 • C and 1526 • C in the furnace of Case 1 (pure coal), Case 6 (50% co-firing). However, it is interesting to note that at an equivalent temperature of 1476 • C, the high-temperature zones of the 50% co-firing case is more widely distributed than the pure coal case. However, when the temperature is at 1526 • C, the T-EFB co-firing cases are again significantly less than the high temperature zones produced by the pure coal cases. The main reason for this phenomenon is that the T-EFB containing higher volatile substances in the mixed burning case is burned near the burner to generate a high temperature zone (1476 • C). On the contrary, the reduction in the amount of pulverized coal in the co-firing case caused the formation of a high temperature area (1526 • C) in the center of the furnace. The distribution of the high-temperature zones is closely related to the heat transfer of the furnace wall and the thermal NO formation. The detailed analysis will be discussed in the next section. The iso-surfaces of Figure 9 show the local high-temperature zones of 1476 °C and 1526 °C in the furnace of Case 1 (pure coal), Case 6 (50% co-firing). However, it is interesting to note that at an equivalent temperature of 1476 °C, the high-temperature zones of the 50% co-firing case is more widely distributed than the pure coal case. However, when the temperature is at 1526 °C, the T-EFB co-firing cases are again significantly less than the high temperature zones produced by the pure coal cases. The main reason for this phenomenon is that the T-EFB containing higher volatile substances in the mixed burning case is burned near the burner to generate a high temperature zone (1476 °C). On the contrary, the reduction in the amount of pulverized coal in the co-firing case caused the formation of a high temperature area (1526 °C) in the center of the furnace. The distribution of the high-temperature zones is closely related to the heat transfer of the furnace wall and the thermal NO formation. The detailed analysis will be discussed in the next section.

Effects of T-EFB Co-Firing Substitutions on the Heat Flux Distribution
As shown in Table 4, the furnace exit gas temperature (FEGT) was compared under the various T-EFB blending ratio. The field test data (1257 °C and simulated data (1258.8 °C) of the case 1. (pure coal) is almost the same. Moreover, the FEGT increases as the amount of T-EFB substitution increases. The maximum temperature difference between cases is about 10 °C. The temperature of the furnace outlet is very important for the operation of the boiler of the power plant, the general temperature is maintained at about 1200 °C [46]. Inside the boiler, there are two main types of heat transfer from the water walls, radiant and convective. Radiation heat transfer accounts for more than 95% of this. The water wall radiation heat transfer was mainly determined by the gas temperature. Therefore, the heat flux distribution of the furnace wall and the gas temperature distribution should have similar properties, the maximum heat flux of the furnace wall occurs at the highest temperature (i.e., the top burner area). This indicates that the heat absorption at the surface of the furnace wall was mainly determined by thermal radiation proportional to the temperature. Figure 10 shows the heat flux distribution of the water wall at different co-firing ratios of T-EFB. It can be observed that with the increase of T-EFB, the area of heat flux distribution of the furnace wall in the main combustion area becomes more extensive, but the maximum degree of heat flux decreases, which shows a similar pattern to the gas temperature distribution. However, in the burnout area above the main combustion zone, the furnace wall heat flux shows a clear downward trend due to the lower gas temperature. Second, based on some heat exchanger heat absorption data Figure 9. Iso-surface temperature of 1476 • C and 1526 • C for cases 1 (pure coal) and 6 (T-EFB 50% co-firing).

Effects of T-EFB Co-Firing Substitutions on the Heat Flux Distribution
As shown in Table 4, the furnace exit gas temperature (FEGT) was compared under the various T-EFB blending ratio. The field test data (1257 • C and simulated data (1258.8 • C) of the case 1. (pure coal) is almost the same. Moreover, the FEGT increases as the amount of T-EFB substitution increases. The maximum temperature difference between cases is about 10 • C. The temperature of the furnace outlet is very important for the operation of the boiler of the power plant, the general temperature is maintained at about 1200 • C [46]. Inside the boiler, there are two main types of heat transfer from the water walls, radiant and convective. Radiation heat transfer accounts for more than 95% of this. The water wall radiation heat transfer was mainly determined by the gas temperature. Therefore, the heat flux distribution of the furnace wall and the gas temperature distribution should have similar properties, the maximum heat flux of the furnace wall occurs at the highest temperature (i.e., the top burner area). This indicates that the heat absorption at the surface of the furnace wall was mainly determined by thermal radiation proportional to the temperature. Figure 10 shows the heat flux distribution of the water wall at different co-firing ratios of T-EFB. It can be observed that with the increase of T-EFB, the area of heat flux distribution of the furnace wall in the main combustion area becomes more extensive, but the maximum degree of heat flux decreases, which shows a similar pattern to the gas temperature distribution. However, in the burnout area above the main combustion zone, the furnace wall heat flux shows a clear downward trend due to the lower gas temperature. Second, based on some heat exchanger heat absorption data in Table 4 and Figure 11, it can be seen that in the pure coal case, the heat absorption simulation predicted for the furnace wall is 488 MWe, which is very close to the actual operating data (482 MWe). However, when the amount of T-EFB was increased, the corresponding heat absorption at the surface of the water wall decreased. In its place, the convection tube bundle above the furnace increases heat absorption accordingly. Furthermore, the UBC reached 4.2% when the T-EFB blending ratio reached 50%, which resulted in a significant reduction in radiative heat transfer from the furnace wall. Compared to the pure coal case, heat transfer reduction is 32 MWe. Despite a small increase in the heat exchanger RH and SH above the burner zones, the total heat absorbed by the boiler is still on a downward trend and the efficiency of the boiler was slightly reduced. Therefore, T-EFB co-firing under different blending ratio conditions change the heat flux distribution during boiler operation.
Energies 2020, 13, x 19 of 27 in Table 4 and Figure 11, it can be seen that in the pure coal case, the heat absorption simulation predicted for the furnace wall is 488 MWe, which is very close to the actual operating data (482 MWe). However, when the amount of T-EFB was increased, the corresponding heat absorption at the surface of the water wall decreased. In its place, the convection tube bundle above the furnace increases heat absorption accordingly. Furthermore, the UBC reached 4.2% when the T-EFB blending ratio reached 50%, which resulted in a significant reduction in radiative heat transfer from the furnace wall. Compared to the pure coal case, heat transfer reduction is 32 MWe. Despite a small increase in the heat exchanger RH and SH above the burner zones, the total heat absorbed by the boiler is still on a downward trend and the efficiency of the boiler was slightly reduced. Therefore, T-EFB co-firing under different blending ratio conditions change the heat flux distribution during boiler operation.

Effects of T-EFB Co-Firing Substitutions on Gas Emission (NOX, SO2)
According to the results of the boiler exit data in Table 4, the NOx reduction was observed with the increase of the co-firing ratio. When the co-firing ratio reaches 50%, the NOX reduction is 42% compared to pure coal case. The positive effect of T-EFB combustion on NOx emission reduction was demonstrated. The amount of NOx produced along the furnace height was shown in Figure 8c. As the combustion reaction proceeds, NOx gradually accumulates along the chamber height. After the OFA region, there was no significant change in NOx production. Comparison of the results obtained under different conditions shows that the total NOx decreases as the proportion of T-EFB addition increases. At the same time, as shown in Figure 12, by comparing the main distribution area of NOx formation, it can be seen that the area of high concentration of NOx from coal combustion was mainly located in the center of the furnace. With the increase of T-EFB co-firing ratio, the area near the burner gradually showed a high concentration of NOx distribution. This is caused by the rapid combustion reaction of T-EFB compared to coal to produce NOx. Therefore, T-EFB has a specific effect on total NOx formation, so fuel NO and thermal NO production should be explicitly analyzed.

Effects of T-EFB Co-Firing Substitutions on Gas Emission (NO X , SO 2 )
According to the results of the boiler exit data in Table 4, the NOx reduction was observed with the increase of the co-firing ratio. When the co-firing ratio reaches 50%, the NO X reduction is 42% compared to pure coal case. The positive effect of T-EFB combustion on NO x emission reduction was demonstrated. The amount of NO x produced along the furnace height was shown in Figure 8c. As the combustion reaction proceeds, NO x gradually accumulates along the chamber height. After the OFA region, there was no significant change in NO x production. Comparison of the results obtained under different conditions shows that the total NO x decreases as the proportion of T-EFB addition increases. At the same time, as shown in Figure 12, by comparing the main distribution area of NO x formation, it can be seen that the area of high concentration of NO x from coal combustion was mainly located in the center of the furnace. With the increase of T-EFB co-firing ratio, the area near the burner gradually showed a high concentration of NO x distribution. This is caused by the rapid combustion reaction of T-EFB compared to coal to produce NO x . Therefore, T-EFB has a specific effect on total NO x formation, so fuel NO and thermal NO production should be explicitly analyzed.
Based on the analysis of the fuel composition data in Table 1, theoretically the fuel NO is reduced because the total supply fuel nitrogen content is inversely proportional to the T-EFB blending ratio. According to Figure 13a, the fuel NO formation and reduction rate along the height of the furnace can be seen, the fuel NO in the burner area is a higher formation phenomenon. Still, with the increase of the ratio of co-firing, the formation rate is reduced. At the same time, the rapid combustion reaction of T-EFB consumes oxygen. It creates a hypoxic-reducing atmosphere in the region near the upper and lower burners, which further reduces NO x production. Second, since thermal NO generation is proportional to temperature. According to the temperature distribution of high temperature (1476 • C and 1526 • C) zones in Figure 9, although the high temperature zone (1476 • C of co-firing cases is slightly more widely distributed, the high temperature zone (1526 • C is less than pure coal cases. As can be seen from the thermal NO generation rate diagram (Figure 13b), the NO generation rate peaked at 28 m high in the furnace under all conditions, but the thermal NO is relatively low under co-firing conditions.   Moreover, then, based on the predicted SO 2 concentrations at the boiler exit listed in Table 4. The SO 2 emissions at the boiler exit were expected to gradually decrease from 476 ppm for the pure coal case to 292 ppm for the 50% T-EFB co-firing case. A reduction of about 38%. The main reason was that the T-EFB contains only 0.15% of S content is lower than pulverized coal (0.4%), so co-firing can significantly reduce SO 2 emissions.

Conclusions
This study analyzed the effects of torrefaction techniques on the properties of R-EFB and also evaluated the feasibility of co-firing T-EFB and coal in large-scale boiler through simulation. The effects of blending on combustion characteristics, heat flux of furnace wall and gas emissions were analyzed for different blending ratios and the and the following conclusions were obtained: T-EFB after torrefaction technology had similar characteristics to sub-bituminous coal, low moisture, high volatility and faster reaction activation. The combustion characteristics of boilers under different blending ratio conditions were analyzed by simulation, and it was found that due to the rapid combustion of T-EFB high volatile fractional matter to generate heat, the high temperature zone inside the boiler diffused towards the burner near, but the furnace peak temperature decreased slightly. Second, T-EFB reacts faster, oxygen consumption was proportional to the co-firing ratio, which could form a low oxygen high temperature reducing atmosphere, while T-EFB co-firing promotes the fire of coal. The temperature distribution in the furnace depends mainly on the degree of fuel burnout. T-EFB co-firing could affect the heat transfer characteristics of the boiler. As the temperature of the OFA burnout area decreased, the heat flux of the water wall decreased, the furnace exit gas temperature increased, and the heat absorption of the heat exchange tubes increased. At a blending ratio above 40%, boiler combustion efficiency decreased as the total heat flux of the boiler decreased due to the rise in the amount of unburned carbon. Compared to pure coal, NO X and SO 2 emissions were significantly reduced with increasing T-EFB substitution. However, the potential adverse effects of T-EFB co-firing in coal-fired boilers, such as slagging, corrosion and economics, require further study.