CFD-study of a 230 MWe coal fired boiler to predict the influence of secondary fuels on slagging, fouling, CO corrosion and NOx formation

Transcription

1 Page 1 CFD-study of a 23 MWe coal fired boiler to predict the influence of secondary fuels on slagging, fouling, CO corrosion and NOx formation Johan Vanormelingen*, Alexander Berreth**, Benedetto Risio** * LABORELEC/ELECTRABEL, Rodestraat 12, 163 Linkebeek, Belgium Johan.Vanormelingen@laborelec.be ** RECOM Services, Nobelstr. 1, 769 Stuttgart, Germany Benedetto.Risio@recom-services.com Abstract A reacting-cfd study has been performed to assess the influence of co-firing secondary fuels (cheaper coals, petroleum cokes, sludge, etc.) on the overall combustion performance of a 23MWe coal fired boiler. In order to characterize the boiler a mathematical 3D-furnace model with 1,, calculated volumes has been set up and a baseline situation of a two-fuel scenario has been modeled to compare the future scenarios with. The assessment of the reliability of the numerical predictions was done by a comparison of computational results with measured operational data of the boiler. In a second phase of the project other fuel mixes will be investigated. One option is to study the effect of a coal type with a high pyrite content, e.g. USA CONSOL, on slagging, fouling and NOx formation. Furthermore, co-combustion with a small percentage of sewage sludge will be considered. 1 Introduction The purpose of the present reacting-cfd study is to assess the influence of co-firing secondary fuels (cheaper coals, petroleum cokes, sludge, etc.) on the overall combustion performance of a 23 MWe coal fired boiler. The co-firing of different fuels can influence the regular operation of the 23 MWe boiler in many ways: - The higher nitrogen content of a secondary fuel (e.g. sewage sludge) can have strong influence onto the NOx emission. Due to the dependence of NOx formation on local stoichiometry and temperature the NOx emissions can be controlled by selecting a suitable air distribution and firing position for the secondary fuel. Depending on the characteristics of the secondary fuel it may be suitable for reburning which could even result in a NOx reduction. For example sewage sludge releases most of the nitrogen as NH3 and firing sewage sludge into areas with a convenient composition and within a suitable temperature window can be used to reduce NOx emissions. - The higher ash content and the different ash properties of a secondary fuel can have influence onto the slagging and fouling behavior of the furnace leading to a degradation of the heat transfer to the water walls and the steam side.

2 Page 2 - Differences in the composition and the heating value of the primary and secondary fuel may change the heat release patterns within the furnace having also a negative influence onto the heat transfer to the water walls and the steam side. In order to achieve the project targets the execution of the project is organized into two different phases: - Phase I: Characterization of the boiler - Phase III: Simulation of different co-firing scenarios 2 Characterization of the 23 MWe Coal Fired Boiler The frontwall fired boiler consists of 4 burner rows with each 4 low-nox coal burners. Overfire air is injected through a level of BOOS (burners out of service) and 4 secondary overfire air ports. In addition, there are two lateral overfire air ports. In order to characterize the 23 MWe coal fired boiler a mathematical 3D-furnace model has been set up and a baseline situation firing a South African and a Columbian coal at the same time has been modeled to compare the future scenarios with. The set up of the mathematical furnace model requires a transformation of the geometry data of the real furnace into the so called discretization (numerical grid) of the furnace. The discretization of the furnace has been done using a multi-domain technique suitable for the coupling of cylindrical and cartesian grids. The furnace model geometry is shown in Figure 1. Fig. 1: Multi-Domain grid of the furnace and the burner array with 1,, cells

3 Page 3 Using the multi-domain technique has several advantages over a single domain technique: - A very fine resolution necessary for the burner discretization can be combined with a sufficient coarse resolution in the rest of the furnace. - The time for the model setup is much shorter due to the modular structure of the grid system. - Even complex geometrical changes in one burner can be handled easily by replacing the grid of the specific burner by a modified version. Beside the furnace geometry the operational parameters of the boiler had to be transformed into the input data set of the 3D-furnace model. The relevant operational parameters considered in the model are: - Coal and Air Distribution - Coal fired through the individual burners - Elementary analysis of the coals - Particle fineness delivered by the milling devices - Swirl parameters of the burners - Thermodynamic parameters of the different flows (e.g. temperature, etc.) Fig. 2: Firing arrangement of the baseline situation (Coals, Swirl Direction, Core Air)

4 Page 4 The baseline situation modeled in Phase I is a two-fuel scenario firing Columbian Coal through mill 4 and South African coal through mill 1 to 3. Figure 2 illustrates connection of fuel to mills and mills to burners. The swirl direction of the different burner flows together with the core air mass flow is also shown in Figure 2. In the following paragraphs this situation will be referred to as the fuel mix 1. The fuel parameters that have been utilized for the two coals are summarized in Table 1. South African Coal Columbian Coal Heating Value MJ/kg MJ/kg Ash 16.2 % % Moisture.82 % 4.19 % C % 67.7 % H 3.62 % 4.76 % S.8 %.81 % N 1.49 % 1.4 % O 6.8 % 8.31 % Table 1: Elementary analysis of raw coal for the two utilized coals 3 Computational Results of the Baseline Situation with Fuel Mix 1 The assessment of the reliability of the numerical predictions was done by a comparison of computational results with measured operational data of the boiler. Oxygen concentration [Vol%] Position 1 - Simulation Position 2 - Simulation Position 1 - Measurement Position 2 - Measurement Radial distance from Centerline [m] Fig. 3: Comparison between computed and measured oxygen profiles in front of burner 1 (right) at a distance of 1.4 and 3.9 m from the burner quarl (left)

5 Page Temperature [ C] A comparison of computed and measured oxygen profiles in front of burner 1 (row 1) is shown in Figure 3. The comparison reveals that the calculated profiles are in reasonably good agreement with the measurements. Peak values as well as general trends are reproduced in the calculation Averaged Temperature O 2 [Vol.-%,dry] Averaged Oxygen concentration Oxygen conc. behind Economizer theoretical Oxygen conc. in case of complete burnout Fig. 4: Calculated average temperature and oxygen concentration over furnace height and comparison with measured oxygen concentration at furnace exit (right) Unburned Carbon / C in Ash [%] Unburned Carbon South African Coal Unburned Carbon Columbian Coal Unburned Carbon Total C in Ash Measurement Z1 and Z NO [mg/m n ³] Fig. : Average values of unburned carbon for South African coal and Columbian coal (left) and NOx concentration (right) over furnace height compared to measurements from the boiler Figure 4 and show average values of temperature, oxygen, unburned carbon, and NOx over furnace height and a comparison with measured values in the field (when available). The oxygen and the unburned carbon values at the furnace exit are in good agreement with the 3 values observed in the field. The calculated NOx concentration (826 mg/m n at 6% O2) at the 3 furnace exit is also predicted within the uncertainty of the measured value (863 mg/m n at 6% O2) Measurement Averaged NO at 6% O2 Averaged NO

6 Page 6 Burner Number: NO 4, 8, 12, 16 3, 7, 11, 1 2, 6, 1, 14 1,, 9, 13 [ppm] Fig. 6: 3D-NOx formation illustrated by vertical slices through the axis of the individual burner columns (burners visible in the slice are identified by the numbers depicted above the picture) Examining the NOx formation in the furnace reveals that burner 1 shows a jet burner pattern that is mainly attributed to the high core air momentum of this burner. This destroys the internal recirculation zone and leads to high NOx production rates (see NOx-distribution in Figure 6). Other burners with higher core air momentum also exhibit higher NOx production rates (e.g. burner 6 and 14). An optimization of the core air momentum could lead to a significant reduction of the NOx concentration at the furnace exit. This would reduce operating costs by minimizing the NH3 consumption of the DENOX and enhancing the lifetime of the catalyst. Unburned Carbon [%] mm.21 mm.41 mm.7 mm.112 mm.1 mm Unburned Carbon [%] Fig. 7: Contribution of different particle fractions to the average unburned carbon loss over furnace height for South African coal (left) and Columbian coal (right) mm.21 mm.41 mm.63 mm.93 mm.128 mm

7 Page 7 The 3D-furnace model allows to identify the potential contribution of individual particle sizes of the different fuels to the total unburned carbon loss at the furnace exit. These results are illustrated in Figure 7. A corrosive atmosphere in the vicinity of the walls can lead to severe damages of the water walls. Whether the walls are really damaged due to corrosion depends on a number of parameters, e.g. O2-, CO-, and H2S-concentration as well as temperature near the walls. However, it was observed that the presence of a sufficiently high oxygen level of a few vol.-% establishes a protecting oxygen curtain that can efficiently prevent corrosion at the water walls. Therefore the wall oxygen level at the furnace walls is checked for in Figure 8. Fig. 8: Wall oxygen level on the furnace walls The computational results of the wall oxygen level indicate a potential risk for wall corrosion on wall No. 3 displaying a large area of the furnace wall with wall oxygen values below. Vol.-%. To overcome this problem the boiler can either be tuned by operational modifications (e.g. air distribution, swirl settings) or retrofitted with a sufficiently high number of so-called wall air nozzles to provide air for the build-up of a protecting oxygen curtain. 4 Modeling the Deposit Formation for the Baseline Situation with Fuel Mix 1 The computer-aided simulation of the deposit formation is done by combining the computational results of the conventional combustion prediction (see previous paragraph) with a subsequent ash deposition prediction in a post-processing step. The post-processor predicts the build up of ash and slag-layers onto the water walls and the convective heat exchangers.

8 Page 8 The deposit prediction has to determine the particle arrival rates and the sticking propensity of the particles arriving. The particle transport model that determines the particle arrival rates does account for inertial impaction, turbulent deposition, and thermophoresis. The sticking propensity is calculated by the concept of the critical viscosity. The nature of the critical viscosity scheme is reported in detail by Richter et. al. (see Richter, 4th Intl. Symposium on Coal Combustion, Beijing, China, 1999). The representative chemical composition of the fly ash required for the viscosity model was provided by LABORELEC. The composition was generated by taking fly ash samples at the furnace exit. Although a CCSEM-analysis of the fly ash samples would have been more reliable it was decided to take the element oxide composition of the bulk ash to speed up the project execution. The element oxide composition shown in table 2 was used in the viscosity model. Wt-% SiO2.7 Al2O Fe2O CaO 3.9 MgO.3 K2O.82 TiO Table 2: Element oxide composition of the fly ash for the viscosity model lg µ (µ in Pas) 16, 1, bulk ash 14, 13, 12, 11, 1, 9, 8, 7, 6,, 4, 3, 2, 1,, -1, -2, Temperature [ C] Sticking Probability < 1 Critical Viscosity Sticking Probability = 1 Fig. 9: Temperature dependence of the ash sample viscosity according to the viscosity model

9 Page 9 Paper presented at the 13th IFRF Members Conference in Noordwijkerhout Figure 9 illustrates the behaviour of the ash sample in the viscosity model with the given element oxide composition. The critical viscosity that leads to a sticking probability of 1 % is reached at a temperature of approximately 11 C. The sticking probability is less than one above the critical viscosity line (thick red line in Figure 9). Computational Results of the Deposit Formation Simulation with Fuel Mix 1 The computational results of the deposit formation simulation for the baseline situation with fuel mix 1 are shown in Figure 1. An amount of 18 % of the total incoming fly ash would be retained in the furnace if no measures for cleaning would be taken. 1E- 2.E- E Wall No. 1 Wall No. 2 Wall No. 3 Wall No. 4 Fig. 1: 3D- (left) and 2D- (right) total deposition rates on the furnace walls in kg/m2s The multi-fuel combustion and deposition scheme allows to identify the contribution of individual mills/fuels to the overall deposition pattern. The contribution of different mills to the deposit formation on the furnace walls is illustrated in Figure 11 and 12. The results show that the Columbian coal does not contribute significantly higher to the deposit formation than the South African coal, and that the mills firing South African coal contribute almost equally to the deposit formation.

10 Page1 Paper presented at the 13th IFRF Members Conference in Noordwijkerhout 1E- 2.E- Wall No. 1 E-.1.2 Wall No Wall No Wall No. 4 1E- 2.E- E-.1.2 Wall No. 1 Wall No Wall No Wall No. 4 Fig. 11: Contribution of mill 1 (left) and mill 2 (right) to the deposit formation in kg/m2s 1E- 2.E- Wall No. 1 E-.1.2 Wall No Wall No Wall No. 4 1E- 2.E- E-.1.2 Wall No. 1 Wall No Wall No Wall No. 4 Fig. 12: Contribution of mill 3 (left) and mill 4 (right) to the deposit formation in kg/m2s Analyzing the contribution of different particle sizes to the deposit formation (see Figure 13) shows that the gross of particles with a size range between and 1 µm are almost equally retained in the furnace (around 17 to 2 %). Although the larger sizes (bigger than 2 µm) display a higher contribution per class the mass fractions of these sizes are negligible.

11 Page Ash retained in furnace [%] Ash retained in furnace Particle size [µm] Fig. 13: Ash retained in the furnace depending on the particle size Conclusions Based on the results obtained by the computer-aided analysis of the combustion performance and the deposition behaviour for the baseline situation (fuel mix 1) in the 23 MWe boiler the following conclusions are derived. The model predictions are in good agreement with the measured values in the field. The difference between measured and predicted values are within the uncertainty of the measured data. High NOx formation was observed in front of burner 1 showing a jet burner pattern. This was mainly attributed to the high core air momentum of this burner since other burners with higher core air momentum also exhibited higher NOx production rates (e.g. burner 6 and 14). An optimization of the core air momentum of the burners could lead to a significant reduction of the NOx concentration at the furnace exit. This would reduce operating costs by minimizing the NH3 consumption of the DENOX and enhancing the lifetime of the catalyst. A potential risk for wall corrosion on wall No. 3 was identified displaying a large area of the furnace wall with wall oxygen values below. Vol.-%. A further question is whether the fine grind is really necessary to avoid either slagging or burnout problems. An optimization of the grind back to a slightly coarser level would on the other hand reduce the energy required for the milling devices and thus enhance the efficiency of the power station. In a second phase of the project other fuel mixes will be investigated. One option is to study the effect of a coal type with a high pyrite content, e.g. USA CONSOL, on slagging, fouling and NOx formation. Furthermore, co-combustion with a small percentage of sewage sludge will be considered.