A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement

Energy xxx (2015) 1e10

Contents lists available at ScienceDirect

Energy journal homepage: www.elsevier.com/locate/energy

A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement Norbert Modlinski a, *, Pawel Madejski b, Tomasz Janda b, Krzysztof Szczepanek b, Wlodzimierz Kordylewski a a b

Division of Boilers, Combustion and Energy Processes, Faculty of Mechanical and Power Engineering, Wroclaw University of Technology, Poland Research and Development, EDF Polska S.A., Poland

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 December 2014 Received in revised form 25 April 2015 Accepted 21 May 2015 Available online xxx

The main objective of this work was to examine the capability of CFD (Computational Fluid Dynamics) on properly predicting temperature distribution in the combustion chamber. Numerous approaches were employed to verify CFD models of large-scale utility boilers. Furnace Exit Gas Temperature is one of the key values used for verification studies. Harsh environment and large dimensions inside the furnace make temperature measurement a complex task. Traditionally used suction pyrometry provides only local information. With this technique, while extremely accurate, it is practically impossible to obtain a representative temperature distribution at the furnace exit as measurements in different locations are not taken at the same time. Acoustic Pyrometry technique is the most appropriate for comprehensive CFD flame shape prediction verification. Not only average temperature value in a certain boiler crosssection can be continuously measured but also its complete two-dimensional distribution. CFD code was used to simulate the OP-650 front-fired boiler operation. The boiler is equipped with Acoustic Gas Temperature Measuring system located in a horizontal plane approximately 4 m under the furnace exit. Comparison of simulation results with measurements proves good accuracy of CFD results. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Computational Fluid Dynamics Pulverized coal Front-fired boiler Acoustic Pyrometry

1. Introduction According to the International Energy Agency, coal will become the world's top source of energy, before oil, in the coming years [1]. Global coal consumption will grow by 1.1% per annum by 2035, driven mainly by non-OECD countries [2]. Although deployment of renewables, need to improve coal plant efficiency and increase in natural gas utilization tends to decrease coal consumption in OECD, coal will long remain a key energy fuel for electricity generation in a number of developed countries. Albeit pulverized-fuel firing technology was first established almost a century ago, researchers and boiler operators still look for a reliable tool able to describe complex phenomena inside the

* Corresponding author. Division of Boilers, Combustion and Energy Processes, Faculty of Mechanical and Power Engineering, Wroclaw University of Technology, 27 Wybrzeze Wyspianskiego St, 50-370 Wroclaw, Poland. Tel.: þ48 606 219 270; fax: þ48 71 328 38 18. E-mail address: [email protected] (N. Modlinski).

furnace, including gasesolid flow, combustion and heat transfer. Performance and environmental concerns as well as utility maintenance issues have increased the use of CFD (Computational Fluid Dynamics) codes to investigate and understand processes inside large scale boilers. CFD application to pulverized coal combustion has been extensively applied. Boyd [3] presented a fully three-dimensional model of a tangentially-fired furnace almost 30 years ago. However, detailed validation studies of pulverized coal combustion simulations have been mostly concerned with pilot scale combustors. Andre et al. [4] carried out a mathematical modeling of a 2.4 MW swirling pulverized coal flame. Experimental measurements provided comprehensive data on velocity components in the near-burner zone, temperature, radiative heat flux and species distribution along the furnace. Hashimoto et al. [5] proposed a novel approach to devolatilization modeling. Suggested tabulateddevolatilization-process model was validated by performing simulation of a pulverized coal combustion field behind a low-NOx burner in a 100 kg-coal/h test furnace. The results show that drastic differences in the gas flow patterns and coal particle behavior

http://dx.doi.org/10.1016/j.energy.2015.05.124 0360-5442/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

2

N. Modlinski et al. / Energy xxx (2015) 1e10

appear between novel and standard approach. Numerous measurement ports provided detailed distribution of temperature and O2 at the furnace axial cross section. A detailed validation of large-scale utility boiler modeling studies is difficult. Majority of the CFD validation studies utilize measurements provided by the Distributed Control System. However, the useful information is limited to average species concentrations at the outlet, local temperatures, unburned carbon in ash. Very often these information are supplemented by design information on FEGT (Furnace Exit Gas Temperature) and heat absorbed by the furnace. FEGT can also be evaluated from zero-dimensional boiler energy balance model [6]. Additional local data can be provided using suction pyrometers or aspiration probes. Xu et al. [7] predicted flow with combustion in a front fired boiler. Measured (estimated) data of furnace outlet temperature, unburned carbon, O2, CO2, CO and NOx concentrations at the furnace outlet as well as heat absorbed by walls and platen super heaters have been used for model validation. Yin et al. [8] investigated a furnace and part of the rear pass in the tangentially fired boiler. The simulation have been validated with global design parameters including O2 at the furnace outlet, heat transfer in the furnace and furnace exit temperature. Site operation data was used to verify NOx predictions. Pallares et al. [9] simulated a front fired boiler. The work concentrated on char burnout predictions. A limited furnace modeling validation included only O2 plant measurements. Choi et al. [10] have used various measured and design values to validate tangentially-fired furnace. Local temperatures at different furnace locations, total heat flux to the furnace walls, O2, CO2, and NOx concentrations at the boiler exit have been compared with computed values. Karampinis et al. [11] have evaluated the effect of co-combustion of cardoon with lignite in a 300 MWe boiler. Validation of the simulations was performed using plant data for the reference case of pure lignite combustion (furnace outlet temperature, O2 and NOx concentrations). Asotani et al. [12] predicted pulverized coal ignition behavior in a 40 MW tangentially fired boiler. Ignition image was obtained from high temperature resistant camera and compared to simulation results. Accuracy of general simulation approach was confirmed by available operating and design data. Gubba et al. [13] have applied CFD simulations to the tangentiallyfired boiler co-firing coal with biomass. Predicted temperatures have been compared to local measurements at three boiler heights. Furnace Exit Gas Temperature is one of the key values used for verification studies. A tool being able to measure not only the average value but also the temperature distribution is needed for a comprehensive CFD model validation. The aggressive environment of high temperatures and ash particles in addition to large dimensions of the furnace make temperature measurement a complex task [9]. Traditionally used suction pyrometry is extremely accurate. However, single probe provides only local information [14]. As the combustion chamber is a dynamic and turbulent environment, representative temperature distribution in the furnace cross-section can be obtained only by performing simultaneous measurements. Number of available test ports pose a technical limitation. Using six pyrometers simultaneously has been reported [14], which would probably not be enough for the crosssection of a large scale furnace. A turning point in CFD furnace models general accuracy assessment came with Acoustic Pyrometry [15]. This technology is more appropriate for two-dimensional temperature mapping than suction pyrometry. It can provide average value in the selected cross-section and the information is available on-line. Flame shape and its location is a critical parameter influencing combustion process performance. Combustion process optimization can improve thermal efficiency of existing boilers up to 0.84% [16]. Homogenous temperature field promotes lower emissions of

NOx, CO and minimizes loss on ignition. The plant operators often rely on 3-D simulation tools to acquire information necessary for combustion process improvement. A comprehensive large scale furnace CFD model should be capable of properly predicting the temperature field and peaks associated with high combustion rates. In the current work a commercial CFD code Ansys Fluent was used to simulate the OP-650 front-fired boiler operation. The boiler is equipped with AGAM (Acoustic Gas Temperature Measuring) system located in a horizontal plane approximately 4 m under the furnace exit. The simulation results were compared with measurements in terms of average temperature as well as temperature profile. 2. Case study boiler and operating conditions The evaluations were performed for a drum type radiant unit installed at the EDF power plant in Rybnik (Poland) utilizing bituminous coal. The case-study boiler (OP-650) is a front-wall pulverized coal boiler with maximum capacity of 220 MWe. The boiler produces 650 tons of steam per hour. Main/reheat steam temperature is 540  C. The boiler is a dual-pass type with unique air/fuel supply system. General scheme of OP-650 boiler low-emission installation with the firing system is depicted in Fig. 1. 12 swirl burners are located on the front wall in two rows (6 in Burner Level I and 6 in Level II). Combustion air is separated into core, primary, secondary and tertiary air as demonstrated in Fig. 2. Swirl vanes fixed at 45 angle are mounted in primary air-coal mixture tube to ensure burner stability. Additional diluted coal-air mixture is provided through 12 drop tubes located in a single row above vortex burners with additional 6 nozzles for optional biomass injection (Burner Level III) as shown in Fig. 1. Approximately 40% of fuel and primary air is transported to the top level. Burners design details are shown in Fig. 2. Coal is supplied through five pulverizers (notation A to E in Fig. 1). During normal operation one pulverizer is always out of service to reassure undisturbed fuel supply in case of unexpected break-down. OFA (Over-fire air) ports are located on three levels (Fig. 1) to complete the oxidation of any unburned combustibles and assure minimum NOx formation. In current boiler configuration the OFA I is shut down. 6 OFA II ports and 10 OFA III ports are installed on front and rear wall respectively. Additionally, air-slots installed on front, rear and side walls are used to provide protective air to the membrane walls against over temperature operations and the influence of the combustion products. To minimize unburned carbon in bottom ash and rear wall corrosion 20 air nozzles were installed on the front wall of the ash pit (Fig. 3). Simulations were conducted for two boiler loads of 135 and 200 MWe. The excess air levels were 15 and 20%, respectively. For lower load two pulverizers were out of service. Operating conditions have been retrieved from plant on-line monitoring system by averaging two-hours measurements during steady-state boiler operation. Air/fuel distribution, temperatures and pulverizers activity are given in Table 1. In certain conditions, due to dynamic boiler environment, deposits are unevenly formed on the platen superheaters surface. As a result, uneven cooling water injection level between left and right live steam line was noticed. The boiler operator tends to adjust flame shape to compensate water injection imbalance. To cool the specific side of the furnace more over-fire air is injected through nozzles on that side. That was the situation for 135 MWe case. Operator directed more air to the left side of combustion chamber to lower the temperature in that region. The OFA II and OFA III distribution through specific ports is given in Table 2. In 200 MWe case the over-fire air was distributed evenly on each level.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

N. Modlinski et al. / Energy xxx (2015) 1e10

Fig. 1. General OP-650 boiler low-emission installation (a) and the firing system (b).

Fig. 2. Swirl burner and drop-tube burner design.

3. Acoustic Gas Temperature Measuring system in OP-650 boiler OP-650 boiler is equipped with AGAM (Acoustic Gas Temperature Measuring) system, utilizing the principle of Acoustic Pyrometry. Acoustic pyrometry is a well established technology in power engineering sector and often used in boilers [17]. This method has been applied to optimize combustion processes depending on actual needs [18]. Temperature monitoring system is a key element of Selective Non-Catalytic Reduction process control system. This technology allows for continuous and contactless real gas temperature measurement inside the combustion chamber and determines temperature profiles [19]. Several individual measurements in a specific boiler cross-section are taken at the same moment to provide temperature distribution map and its average value. That is a very valuable information for furnace CFD models verification. The system is mainly based on the fact that speed of sound depends on temperature. The principle of operation is the measurement of the acoustic wave time delays, which depend on the temperature in the environment of propagation. The relation is defined by the following formula:

v¼

rffiffiffiffiffiffiffiffiffiffiffiffi k·R ·T M

where:

v ¼ speed of sound, (m/s) R ¼ universal gas constant, (kJ/kmol k) M ¼ gas molar mass, (kg/kmol) T ¼ temperature, (K) k ¼ adiabatic index (cp/cv)

(1) Fig. 3. Bottom air supply system. 1 e air injection nozzles, 2 e cold air supply.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

3

4

N. Modlinski et al. / Energy xxx (2015) 1e10

Table 1 Boiler operating conditions at 135 and 200 MWe. Capacity: 200 MWe Inactive burners/pulverisers

Burner Level I

Burner Level II

Burner Level III

Sum:

All active

E2, E3

E1, E4

Coal (kg/s) Core air (kg/s) Primary air (kg/s) Secondary air (kg/s) Tertiary air (kg/s) OFA II (kg/s) OFA III (kg/s) Protective air (kg/s) Bottom air (kg/s) Total air (kg/s) Total air excess Burner belt air excess Primary/secondary air temperature (K)

7.76 0.73 25.4 5.14 7.53 33.79 24.84 36.35 4.97 205.47 1.15 0.57 386/573

5.54 0.73 16.26 5.14 7.53

10.88 0 34.38 2.93 0

24.18 1.46 76.04 13.21 15.06 33.79 24.84 36.35 4.97 205.47 1.15 0.57

Burner Level I

Burner Level II

Burner Level III

Sum:

D1, D4, F1, F4,

All Active

D2, D3, F2, F3

1.98 0.2 7.54 1.42 2.08 18.22 13.39 32.96 2.68 128.11 1.2 0.57 386/543

5.7 0.2 21.4 1.42 2.08

6.3 0 23.7 0.81 0

Capacity: 135 MWe Inactive burners/pulverisers

Coal (kg/s) Core air (kg/s) Primary air (kg/s) Secondary air (kg/s) Tertiary air (kg/s) OFA II (kg/s) OFA III (kg/s) Protective air (kg/s) Bottom air (kg/s) Total air (kg/s) Total air excess Burner belt air excess Primary/secondary air temperature (K)

Table 2 Uneven over-fire air distribution in 135 MWe case. OFA II Nozzle no.

1

2

3

4

5

6

Mass flow rate (kg/s)

3.4

3.42

3.02

3

2.69

2.69

OFA III Nozzle no.

1

2

3

4

5

6

7

8

9

10

Mass flow rate (kg/s)

1.54

1.54

1.54

1.54

1.54

1.14

1.14

1.14

1.14

1.14

The adiabatic index and molar mass of gas are calculated based on the typical flue gas composition from the combustion of hard coal. For fuel fired in OP-650 boiler the volume fractions of main combustion products are N2 ¼ 76.5%, O2 ¼ 3%, CO2 ¼ 13%, H2O ¼ 7.5%. Velocity of a wave is determined by the propagation time of an acoustic impulse between transmitters and receivers. The distance between them is fixed and known. The temperature is expressed with the relation presented below:

T¼

l2 ·106  273:15 B·t2

(2)

where: l e distance between transmitter and receiver (propagation path), (m) t e propagation time (time of delay), (s)   B e acoustic constant k·R M

13.98 0.4 52.64 3.65 4.16 18.22 13.39 32.96 2.68 128.11 1.2 0.57

Presented system is dedicated to measure temperature distribution in a single boiler level about 4 m under the furnace exit. AGAM consists of 8 transceivers (operating as transmitter and receiver) (Fig. 4). Each transmitter emits sound, recorded in turn by other receivers giving 21 measurement paths. AGAM system uses audio signals in a frequency range between 200 and 3000 Hz. The generated sound pulse is a pneumatically created high intensity white noise signal. Compressed air is used as a sound source medium. It is initialized by the opening of electromagnetic valve at a transmit side of the transceiver unit. The signal created at the transmitter side is present time-delayed with respect to all other transceiver units (receive sides). All signals at the transceiver units are recorded at the same time by the control computer. The registered signals undergo an analysis of propagation correlation within the specific environment and the flight time for all directions is obtained. The transceiver units creates and records the compressed airgenerated sound signals. The units have identical design and differ only in their function for the respective path. A transceiver unit consists of:  An enclosure (junction box) with preamplifier, terminal block and service switch  An acoustic horn for connecting to boiler/furnace openings  A piezo microphone at the flange of the acoustic horn  A solenoid valve with sound nozzle at the beginning of the acoustic horn. The main properties of measurement system are presented in Table 3.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

N. Modlinski et al. / Energy xxx (2015) 1e10

2880

5

Zone 1

Zone 2

Zone 3

Zone 4

Zone 5

Zone 6

Zone 7

Zone 8

Zone 9

Zone 10

Zone 11

Zone 12

2880

Rear wall 6000

6000

Fig. 4. AGAM transceivers set-up on boiler OP-650k in Rybnik Power Plant. Measurement paths inside the boiler with 12 measuring zones.

The temperature field can be reconstructed by the twodimensional tomography algorithm [20] on a user-selected number of zones (12 zones were selected for this work as shown in Fig. 4). Temperature distribution, the zone minimal and maximal temperature values, standard deviation, imbalance information are calculated by the software after the path temperatures had been sent to the diagnostic computer. Temperature profile can be presented either as zone average temperature diagram or as an isothermal contour plot. Described temperature monitoring system is a powerful tool for boiler combustion process performance assessment. The identification of possible uneven temperature distribution causes can be explained by boiler CFD simulation. Modeling results accuracy should than be verified with measured temperature values from AGAM. 4. Mathematical model The three-dimensional geometry was created using ANSYS Design Modeler e FLUENT pre-processor [21]. Mesh is one of the most important factors in the simulation of the large-scale boilers [22]. A body fitted mesh was created containing mostly hexahedral elements. Mesh quality was evaluated with two parameters:  Orthogonal Quality e defining to what extent the mesh is not orthogonal. It is computed for cells using the vector from the cell centroid to each of its faces, the corresponding face area vector, and the vector from the cell centroid to the centroids of each of the adjacent cells. The best cells are close to 1. In the generated mesh Minimum Orthogonal Quality was 0.2, with an average value of 0.95  Aspect Ratio e defining the ratio of cell sizes in different dimensions. For best results this value should be close to one. Presented mesh is characterized with Maximum Aspect Ratio of 7 in the cell of poorest quality, while the average value was equal to 1.4. Because of the discrepancy of scale between the burner and of the much larger volume of the furnace the solution domain was subdivided into fine grid regions around the burners and coarser regions elsewhere. The circular drop-tube burners nozzles have been replaced by rectangular ones to ease the meshing process. The platen super heaters have been modeled as zero-thickness

horizontal planes, since it is practically impossible to model actual tubes. Radiant superheaters geometry were included for the better development of gas outflow. A mesh independence analysis was performed. The solution evaluation criterion was related to predicted two-dimensional temperature distribution in the AGAM cross-section. Initially mesh consisting of 4.0 million control volumes was generated and the reference solution was obtained. Reduction of cell number to about 3.3 million did not substantially changed the predicted temperature field, while introducing better compromise between solution accuracy and computational time. Computational domain and grid system in the swirl burners region is shown in Fig. 5. The mathematical model used here is based on commercial CFD code FLUENT [21]. Simulation of the following processes takes place in the furnace: turbulent flow, coal combustion, gas phase combustion, particle transport and radiative transport. The gas phase is modeled assuming an Eulerian approach, whilst for the solid phase, the Lagrangian approach is more suitable. Realizable k-ε model [23] was used as a closure of turbulent Reynolds equations. The realizable k-ε model is relatively widely used for engineering applications and provides better performance in many industrial turbulent flows than the standard k-ε model. The pulverized coal combustion process can be divided into two parts, devolatilization and char combustion. A single-rate kinetic devolatilization model [24] is used to predict the volatiles yield rate, which assumes that the rate of devolatilization is dependent on the amount of volatiles remaining in the particle via a first-order reaction. This model was extended by using FG-DVC network devolatilization model [25,26] as a pre-processor. It predicts the rate of the production and high temperature yields for the char, tar, volatiles and the composition of key species during the devolatilization of any coal. The results as well as the proximate and ultimate analysis for the used coal are given in Table 4. FG-DVC output was used in the devolatilization sub-model. In the CFD modeling of turbulent flow with combustion it was assumed that volatiles are produced as a single compound that undergoes instantaneous break up reaction into tar, light hydrocarbons, CO, CO2 and H2O. FG-DVC calculates devolatilization rate of tars and other species. The most significant mass drop of fuel particle occurs when tar is evolved. Tar is produced as a first specie. For this reason tare release rate is used in devolatilization model.

Table 3 Operation parameters of AGAM system. Measurement range

Accuracy

Resolution

Update time

0e2000  C

1% (path value)

10  C (for 10 m path length)

2e4 s for 1 to 6 path

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

6

N. Modlinski et al. / Energy xxx (2015) 1e10

Fig. 5. CFD model geometry of a front-fired boiler with the grid system around the burners.

Knowing the volatile fraction of dry ash free coal (fvolatile) and assuming that residual char is pure carbon we can calculate lower heating value of volatiles.

LHVvolatile ¼

LHVcoaldaf  fchar ·LHVchar

(3)

fvolatile

Assuming that lower heating value of light hydrocarbons is approximately equal to that of methane (LHVgas ¼ 50 MJ/kg) we can easily calculate lower heating value of tars from the instantaneous break-up reaction of volatiles:

LHVtar ¼

  LHVvolatile  ygas ·LHVgas þ yCO ·LHVCO

(4)

ytar

where ygas, ytar, yCO stand for mass fraction in volatiles. EDC (Eddy dissipation concept) [27] was used as a general concept for treating interaction between turbulence and chemistry in flames. In this model the total space is subdivided into fine structures and the surrounding fluid. All reactions of the volatile Table 4 Coal analysis and FG-DVC output. Proximate analysis (wt%, as received) Ash

Volatile matter

Moisture

Fixed carbon

22.34

25.77

12.75

39.14

Ultimate analysis (wt%, daf) C

H

N

S

O

84.7

5.39

1.55

1.23

7.13

FG-DVC high temperature yield (wt%, daf) Volatiles

Char

47.8

52.2

components are assumed to react only in these spaces which are locally treated as a perfectly stirred reactors. Four-step global mechanism mainly based on the one demonstrated in Ref. [28] was employed: 1. 2. 3. 4.

CmHn þ (m/2 þ n/4)O2 / mCO þ (n/2)H2O CxHyOz þ (x/2 e z/2)O2 / xCO þ (y/2)H2 CO þ 0.5O2 / CO2 H2 þ 0.5O2 / H2O

Following the above mechanism with known heat of reactions we can further calculate enthalpies of formation of volatiles, light hydrocarbons and tars. Enthalpy of formation of tars was calculated by assuming zero heat of volatiles instantaneous break-up reaction. Heat of pyrolysis was not included in the analysis. Char undergoes heterogeneous oxidation to CO. The reaction rate is calculated on the assumption that the process is limited by the diffusion of oxygen to the external surface of the char particle and char reactivity [29]. It was assumed that the particle absorbs all the heat of the char burnout according to [3]. Discrete phase trajectories are calculated using Lagrangian formulation and the coupling between the phases is introduced through particle sources of Eulerian gas-phase equations. Within the particle transport model, the mass flow of coal particles is represented by a number of trajectories that always represent much larger number of actual particles. The dispersion of particles due to turbulence in the fluid phase was considered. In case of combustion in furnace problem radiation is not only the dominant energy transport mechanism but also one of the most complex problems. The RTE (radiative transfer equation) [30] governs the radiation heat transfer in participating media. It describes the variation of radiation intensity (I) as it travels along a certain path (s) in the medium, in the direction (s, u). Considering absorption, emission and scattering apart from and into the direction s, the RTE can be described as follows:

Volatile composition from FG-DVC (wt%) H2O

CO

CO2

CmHn

CxHyOz

4.8

3.3

2.35

7.12

31.5

Empirical formula for light hydrocarbons (CmHn) and tar (CxHyOz) CmHn (gas)

CxHyOz (tar)

m ¼ 1, n ¼ 7.22

x ¼ 7, y ¼ 4.48, z ¼ 0.72

dIðs; uÞ s ¼ ðk þ sÞ$Iðs; uÞ þ k$Ib þ $ ds 4$p

Z Iðs; uÞ$Fdu

(5)

4p

where k and s denote absorption and scattering coefficients, F is the phase function. The spatial integration of RTE was carried out with the DO (Discrete Ordinates) method [31]. The number of RTE

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

N. Modlinski et al. / Energy xxx (2015) 1e10

depends on the total number of gray gases and takes into account scattering of the particles. The DO method solves the RTE for a set of directions based on the concept of angular discretization scheme. Each octant of the angular space 4p is discretized into polar and azimuthal solid angle. The continuous integral over the solid angle is approximated by a numerical quadrature scheme, where the equations are solved for a series of directions. In a typical combustion chamber H2O and CO2 are the main gaseous absorbers and emitters of radiant energy. The total emissivity of gas is calculated by a number of gray gases using polynomial correlations for weighing factors and absorption coefficient according to Weighted Sum of Grey Gases method [32]. Widely employed coefficients for emissivity [33], fitted from the benchmark exponential wide-band model, have been used in this work. The WSGGM represents the entire spectrum with three gray gases having uniform absorption coefficients. The total gas phase absorption coefficient is calculated from the total emissivity with the mean path length calculated from the characteristic cell size. The gas phase absorption coefficient was corrected according to TayloreFoster model [34], assuming uniform and constant soot concentration (103 kg/m3) in the furnace. As noticed in Ref. [35] the main source of radiative transfer in two-phase mixture is the particle cloud. Therefore coal particles emissivity (εp) treatment is crucial in coal combustion simulation. In this work the particle emissivity was assumed to be a function of unburned carbon in particle (Uc) following the relation [35]:

εp ¼ 0:4·Uc þ 0:6

(6)

The particle reflectivity, and scattering effects are also included in the calculation of heat transfer. Thermal boundary conditions at walls have been expressed in terms of surface temperature and emissivity. It was assumed that the evaporator surface temperature is about 60 higher than the saturation temperature corresponding to the pressure of 16 MPa in the boiler drum. Calculations have been carried out for three

7

emissivity values equal to 0.5, 0.7, 0.9 at fixed wall temperature. All the presented figures correspond to emissivity equal 0.7, which is a typical value found in the literature [11]. We have to emphasize that during real boiler operation temperature and emissivity vary with time and spatially due to water-wall slagging and soot blowers operation. This phenomena can be included in the simulations by implementing deposition model with thermal properties submodel [36].

5. CFD results Fig. 6 shows temperature contours in two vertical planes crossing axis of F4 and C4 burners. It is clearly visible that for 135 MWe F4 burner is out of service. One mill is switched off and only cooling air flows through the burners. This burner out-ofservice technique helps to decrease the local air-to-fuel ratio at the exit locations of in-service burners, leading to reduced furnace temperature and lower NOx emissions. A swirling flow is developed behind each burner, thus stabilizing the flame. Due to large dimensions of the combustion chamber the temperatures inside the OP-650 are considerably lower in comparison to boilers of similar thermal power at full load. The temperature of the flue gas is relatively high under the last over-fire air injection inside the furnace where coal combustion actively takes place. As the flue gas flows from the furnace exit, the temperature gradually decreases due to the heat transfer from the flue gas to the furnace walls, re-heaters, super-heaters. A notable temperature deviation is seen in the burner region. The maximum temperature values reach 1830  C close to the burners nozzles, where combustion has the highest intensity, associated with strong mixing. No flame impingement on the rear wall was observed (Fig. 6) which corresponds to lower corrosion risk. From the temperature contours we can also conclude that OFA II air deeply penetrates combustion chamber breaking up the gas stream.

Fig. 6. Temperature contours ( C) in vertical cross-sections.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

8

N. Modlinski et al. / Energy xxx (2015) 1e10

As shown in Fig. 7 for 200 MWe case the high flue gas temperature zone is distributed approximately along the furnace centerline and close to the rear wall. In some situations deposits are unevenly formed on the platen superheaters surface. As a consequence boiler operator observed large deviations of cooling water injection level between left and right steam line in a 135 MWe case. To compensate this it was decided to manually redistribute the amount of heat released in the combustion chamber. By supplying more over-fire to the left side the maximum temperature zone was transferred to the right side of the furnace (Fig. 7). General CFD simulation performance was verified by comparing AGAM temperature measurement against computations results (Figs. 8 and 9). Widely used furnace exit temperature measurement method using thermocouples and further averaging does not provide precise information about the temperature profile in specific cross-section plane and thus are useful to limited extent.

These studies were focused on comparison of temperature deviations in the boiler cross-section corresponding to AGAM system location level. General temperature profile observation shows good accuracy of CFD results. Maximum temperature location was correctly predicted for both cases (Figs. 8 and 9). For 200 MWe boiler capacity the high temperature zone is situated close to the rear wall along the furnace centerline. Simulations showed flame deflection towards the left wall at 135 MWe load. These results give satisfying agreement between measured and calculated temperature profile. It can be easily noticed that one of the reasons for the difference between predicted and measured temperature distribution is that AGAM does not account for temperature gradients in the near-wall zones, while the CFD model does. Maximum predicted and measured temperature difference in AGAM cross-section plane was 200  C for both loads. To demonstrate how does the temperature field vary over the surface, an Area Based and Mass Based Uniformity Indexes (UIarea, UImass) of

Fig. 7. Iso-surface of temperature equal to 1300/1100  C corresponding to 200/135 MWe boiler capacity.

Fig. 8. Comparison of temperature contours ( C) measured by Acoustic Gas Temperature Measuring system with CFD simulation results for 200 MWe.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

N. Modlinski et al. / Energy xxx (2015) 1e10

9

Fig. 9. Comparison of temperature contours ( C) measured by Acoustic Gas Temperature Measuring system with CFD simulation results for 135 MWe.

temperature distribution were used. The parameters are defined with the following equations:

UIarea ¼ 1 

UImass

 PN  Ti  Taverage $Ai i¼1

2·Taverage ·A



 PN  Ti  Taveragemass $ðri $wi $Ai Þ i¼1 ¼1 PN 2·Taveragemass $ i¼1 ðri ·wi ·Ai Þ

(7)

(8)

where: i ¼ subscript denoting facet defining the surface N ¼ total number of facets (12 zones for AGAM, 2260 facets for CFD) Taverage ¼ is the average temperature in the cross-section plane (K) A ¼ the cross-section area (m) r ¼ density (kg/m3) w ¼ velocity (m/s) and the mass based average temperature (K) in the cross-section plane is defined as:

PN Taveragemass ¼

i¼1 ½Ti ·ðri ·wi ·Ai Þ PN i¼1 ðri ·wi ·Ai Þ

(9)

For the calculation of Mass-Based Uniformity Index the values of average densities and velocities in 12 AGAM zones were taken from CFD results as no such measurements were available. UI values of one indicate the highest uniformity. The Area Based UI calculated from measurements exceeds 90% for both cases, which means that no significant gradients exists. Measured Standard Deviation was 162  C and 128  C for 200 MWe and 135 MWe respectively. A conclusion can be made that it is easier to achieve a homogenous temperature field at lower load. The temperature homogeneity is important not only for NOx reduction but also for platen super-heaters operation, which are located above the combustion chamber. High value of Area Based UI indicates relatively uniform distribution of radiative heat flux emitted from combustion chamber. The temperature distribution Mass Based UI can be associated with convective heat transfer. Calculated values are 79% for 200 MWe and 73% for 135 MWe suggest less evenly distributed convective heat flux in comparison to radiative one. CFD prediction of temperature distribution parameters is satisfying (Figs. 8 and 9). In the worst case the Standard Deviation calculated from CFD results was 12  C higher than the value obtained from measurements at 135 MWe load. Both Uniformity Indexes show good agreement with less than 5% maximum difference for both boiler loads.

For 200 MWe case the measured average temperature in AGAM cross-section was 38  C higher than calculated, what corresponds to 3% relative error. At 135 MWe load the difference was 27  C (relative error equal to 2.5%). Since the heat transfer in the furnace is highly dependent on the emissivity of the furnace wall, the choice of this value has significant influence on the calculated average temperature. In this work calculations have been carried out for three emissivity values equal to 0.5, 0.7, 0.9 at fixed wall temperature. The predicted average temperatures were 1189, 1222, 1271  C at 200 MWe and 1043, 1078, 1112  C at 135 MWe. All the results demonstrated in this work correspond to wall emissivity equal to 0.7. More detailed description of thermal wall boundary conditions can be introduced by implementation of deposition model with thermal properties submodel [36]. 6. Conclusions Homogeneous temperature field and elimination of high peaks is a crucial factor in lowering NOx emission and high boiler performance. The main objective of this work was to show to what extent CFD boiler simulation are capable of properly predicting temperature distribution in the combustion chamber. The results obtained from the numerical calculations were compared with the values obtained from Acoustic Gas Measurement System. Acoustic Pyrometry brings new quality for CFD models verification. One of the key parameters used for this purpose is the Furnace Exit Gas Temperature. This technology allows for continuous and contactless real gas temperature measurement inside the combustion chamber and determines temperature profiles in a selected boiler cross-section. Operating conditions and AGAM temperature measurements have been retrieved from plant on-line monitoring system by averaging two-hours measurements during steady-state boiler operation. CFD simulations demonstrate non-uniformity of temperature distribution that is similar to the one obtained from measurements. For 200 MWe case the highest temperature peak is located in the middle of the plane closer to the rear wall. Simulations are able to properly capture flame deflection towards the left wall at 135 MWe load. Comparison of Standard Deviation and Uniformity Indexes proves good accuracy of CFD results. In the worst case the Standard Deviation calculated from CFD results was 12  C higher than the value obtained from measurements. Both Uniformity Indexes show good agreement with less than 5% differences. Maximum predicted and measured temperature difference in AGAM cross-section plane was 200  C for both loads. When considering average temperature a 3% relative error was obtained for 200 MWe case, which corresponds to 38  C difference between CFD and AGAM. At 135 MWe load the difference was 27  C (relative error equal to 2.5%). This value can be adjusted by

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124

10

N. Modlinski et al. / Energy xxx (2015) 1e10

modifying wall emissivity for fixed wall temperature. During real boiler operation non-stationary process of deposition formation causes spatial changes in the temperature and emissivity of the water-wall surface between soot blower activities. More detailed description of thermal wall boundary conditions can be introduced by implementation of deposition model with thermal properties sub-model [36]. In addition to other intrinsic CFD model inaccuracies, one of the most visible difference in computed and measured temperature distribution is caused by lack of near-wall temperature gradient consideration by AGAM map reconstruction algorithm. Acoustic pyrometry depends on the molecular weight of the gases and generally complete combustion is assumed. The speed of sound also depends on the flue gas flow and relatively low velocities are desirable. All deviations from these conditions increase measurement inaccuracy. The visual and mathematical analysis of temperature distribution shows good accuracy of CFD results. Simulations are able to properly capture the non-uniformity of temperature distribution for both considered cases. Acknowledgments This research was funded from GEKON program by the National Center of Research and Development (GEKON1/O2/213655/9/2014) and National Fund for Environmental Protection and Water Management under research and development project No. GEKON1/O2/ 213655/9/2014. References [1] World Energy Outlook 2014. International energy agency report IEA, Paris. 2014. [2] BP Energy Outlook 2035, January 2014, bp.com/energyoutlook. [3] Boyd R, Kent J. Three-dimensional furnace computer modeling. In: 21st Symposium (international) on combustion, Munich, West Germany; 1986. p. 265e74. [4] Andre A, Peters F, Weber R. Mathematical modeling of a 2.4 MW swirling pulverized coal flame. Combust Sci Technol 1997;122:131e82. [5] Hashimoto N, Kurose R, Hwang S, Tsuji H, Shirai H. A numerical simulation of pulverized coal combustion employing a tabulated-devolatilization-process model (TDP model). Combust Flame 2012;159:353e66. [6] Constenla I, Ferrín J, Saavedra L. Numerical study of a 350 MWe tangentially fired pulverized coal furnace of the As Pontes Power Plant. Fuel Process Technol 2013;116:189e200. [7] Xu M, Azevedo J, Carvalho M. Modeling of a front wall fired utility boiler for different operating conditions. Comput Methods Appl Mech Eng 2001;190: 3581e90. [8] Yin C, Caillat S, Harion JL, Baudoin B, Perez E. Investigation of the flow, combustion, heat-transfer and emissions of a 609 MW utility tangentially fired pulverized-coal boiler. Fuel 2002;81:997e1006. [9] Pallares J, Arauzo I, Diez LI. Numerical prediction of unburned carbon levels in large pulverized coal utility boilers. Fuel 2005;84:2364e71. [10] Choi C, Kim C. Numerical investigation on the flow, combustion and NOx emission characteristics in a 500 MWe tangentially fired pulverized-coal boiler. Fuel 2009;88:1720e31.

[11] Karampinis E, Nikolopoulos N, Nikolopoulos A, Grammelis P, Kakaras E. Numerical investigation Greek lignite/cardoon co-firing in a tangentially fired furnace. Appl Energy 2012;97:514e24. [12] Asotani T, Yamashita T, Tominaga H, Uesugi Y, Itaya Y, Mori S. Prediction of ignition behavior in a tangentially fired pulverized coal boiler using CFD. Fuel 2008;87:482e90. [13] Gubba R, Ingham DB, Larsen KJ, Ma L, Pourkashanian M, Tan HZ, et al. Numerical modelling of the co-firing of pulverised coal and straw in a 300 MWe tangentially fired boiler. Fuel Process Technol 2012;104:181e8. [14] Rinaldi F, Najafi B. Temperature measurements in WTE boilers using suction pyrometers. Sensors 2013;13:15633e55. [15] Bramanti M, Salerno E, Tonazzini A, Pasini S, Gray A. An acoustic pyrometer system for tomographic thermal imaging in power plant boilers. IEEE Trans Instrum Meas 1996;45:159e67. [16] Campbell R. Increasing the efficiency of existing coal-fired power plant, congressional research service report. December 2013. R43343, www.crs. gov. [17] Deuster M. Acoustic gas temperature measurement. In: Proceedings of Wissenforum: temperature measurement technique, Aldenhoven; 2009. [18] Nabaglo D, Madejski P. Combustion process analysis in boiler OP-650k based on acoustic gas temperature measuring system. In: Proceedings of 3rd International Conference on Contemporary Problems of Thermal Engineering CPOTE 2012; September 2012. p. 18e20. Gliwice, Poland. [19] Domels H, Deuster M, Reinitzhuber F. Acoustic gas temperature measurements at the blast furnace top. Stahl Eisen 2009;129(2):27e34. [20] Zhang S, Shen G, An L, Niu Y. Online monitoring of the two-dimensional temperature field in a boiler furnace based on acoustic computed tomography. Appl Therm Eng 2015;75:958e66. [21] ANSYS FLUENT, www.ansys.com. [22] Wang Z, Zhuo Y, Chen C, Xu X. Mesh investigation about crossflow diffusion of computational flow dynamics in tangential combustion flow field. Proc Chin Soc Electr Eng 2007;27:22e8. [23] Shih T, Liou W, Shabbir A, Yang Z, Zhu J. A new k-ε Eddy-viscosity model for high Reynolds number turbulent flows e model development and validation. Comput Fluids 1995;24(3):227e38. [24] Williams A, Backreedy R, Habib R, Jones JM, Pourkashanian M. Modeling coal combustion: the current position. Fuel 2002;81:605e18. [25] Advanced Fuel Research Inc. Usurers guide for FG-DVC, East Hartford, CT. 2004. [26] Williams A, Backreedy R, Habib R, Jones JM, Pourkashanian M. An extended coal combustion model. Fuel 1999;78:1745e54. [27] Magnussen BF, Hjertager BH. On mathematical models of turbulent combustion with special emphasis on soot formation and combustion. In: 16th Symp. (Int'l.) on Combustion. The Combustion Institute; 1976. p. 719e29. [28] Hautman D, Dryer F, Schug K, Glassman I. A multiple-step overall kinetic mechanism for the oxidation of hydrocarbons. Combust Sci Technol 1981;25: 219e35. [29] Baum M, Street P. Predicting the combustion behaviour of coal particles. Combust Sci Technol 1971;3:231e43. [30] Modest M. Radiative heat transfer. 2nd ed. Amsterdam: Academic Press; 2003. [31] Jamaluddin A, Smith P. Predicting radiative transfer in axisymmetric cylindrical enclosures using the discrete transfer method. Combust Sci Technol 1988;62:173e86. [32] Hottel HC, Sarofim AF. Radiative transfer. New York: McGraw-Hill; 1967. [33] Smith T, Shen Z, Friedman J. Evaluation of coefficients for the weighted sum of gray gases model. Trans ASME 1982;104:602e8. [34] Taylor P, Foster P. Some gray gas weighting coefficients for CO2-H2O-soot mixtures. Int J Heat Mass Transf 1974;18:1331e2. [35] Lockwood FC, Rizvi SM, Shah NG. Comparative predictive experience of coal firing. Proc Inst Mech Eng 1986;200:79e87. [36] Modlinski N. Computational modeling of a tangentially fired boiler with deposit formation phenomena. Chem Process Eng 2014;35(3):361e8.

Please cite this article in press as: Modlinski N, et al., A validation of computational fluid dynamics temperature distribution prediction in a pulverized coal boiler with acoustic temperature measurement, Energy (2015), http://dx.doi.org/10.1016/j.energy.2015.05.124