development and application of an les based cfd code … · 2018. 2. 12. · particle trajectory...

12
9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub ) April 26-29, 2011/Estoril, Portugal Page 1 DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE TO SIMULATE COAL AND BIOMASS COMBUSTION IN GENERAL REACTOR CONFIGURATIONS Ahti Suo-Anttila, Ph.D., [email protected] , Joseph D. Smith 1 , Ph.D., [email protected] , Lawrence D. Berg, P.E., [email protected] Systems Analyses and Solutions, Inc., Owasso, Oklahoma, USA ABSTRACT This paper describes a new Large Eddy Simulation (LES) based computational fluid dynamics (CFD) code that has been developed to simulate multiphase coal and biomass combustion and gasification in an transient eulerian framework. This work is based on previous efforts aimed at simulating radiant heat transfer and combustion characteristics of large buoyancy driven pool fires. The original code has been applied to large industrial flares to predict flame shape, radiant flux from the flare, and to predict the effect of cross winds on combustion efficiency. In the current application, the code is extended to simulate multiphase reacting flow of coal and biomass. Submodels have been added to approximate coal/biomass devolatilization and char oxidation. Simulations have been performed to demonstrate and validate this capability. This code simulates fast and slow fuel devolatilization, fuel vapor and oxygen transport, chemical reaction and heat release, soot formation, pollutant formation and destruction, diffuse radiation within the combustion zone and radiation from the flame to surrounding objects. Reaction rate and soot radiation parameters are based upon a local equivalence ratio combustion concept which has been calibrated against experimental data. Code validation includes comparing predicted results to experimental data reported in the Brigham Young University data book. Model results indicate the importance of small particles in flame stabilization and combustion efficiency. KEYWORDS: Coal, Biomass, Combustion, Gasification, Large Eddy Simulation, Computational Fluid Dynamics I. INTRODUCTION Simulation of pulverized coal combustion processes involves modeling a number of complex simultaneous interdependent physics as shown conceptually in Figure 1. Several driving motivations exist in today’s coal fired power generation industry including: 1) meeting increasingly tighter environmental regulations focused on reducing NOx, CO, and more recently CO2, 2) identifying the “best” burner, furnace design, or other equipment to use for a specific combustion application, 3) optimizing plant operations to increase production efficiency and plant safety while reducing maintenance costs, and 4) improving plant operating factor by reducing time spent during routine plant shutdowns. Developing and applying a comprehensive computational fluid dynamics (CFD) based multi-physics code allows a user to address each of these issues. Accurate CFD modeling of coal combustion requires that the code allow for turbulent particle laden gas flow with homogeneous gas phase combustion chemistry, particle mass change due to moisture evaporation, particle devolatilization and char combustion, trace chemistry associated with NO x /SO x formation, and radiation heat transfer from the flame to the surrounding furnace walls with participating media including non-grey gases and absorbing/emitting particles. A significant amount of work has been done over the years in developing and applying advanced CFD models for coal combustion [1-9]. One of the main challenges for these models has been accurately capturing the strong multi-phase coupling between reacting coal particles and their effect on the gas phase. Most of the previous work has adopted a Lagrangian/Eulerian treatment for the particles/gas coupling [1-7]. The main justification for using a particle trajectory model has been to accurately approximate the combusting coal particle history effects. 1 Corresponding author

Upload: others

Post on 08-Oct-2020

0 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 1

DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE TO SIMULATE COAL AND BIOMASS COMBUSTION IN GENERAL REACTOR CONFIGURATIONS Ahti Suo-Anttila, Ph.D., [email protected], Joseph D. Smith1, Ph.D., [email protected], Lawrence D. Berg, P.E., [email protected] Systems Analyses and Solutions, Inc., Owasso, Oklahoma, USA ABSTRACT

This paper describes a new Large Eddy Simulation (LES) based computational fluid dynamics (CFD) code that has been developed to simulate multiphase coal and biomass combustion and gasification in an transient eulerian framework. This work is based on previous efforts aimed at simulating radiant heat transfer and combustion characteristics of large buoyancy driven pool fires. The original code has been applied to large industrial flares to predict flame shape, radiant flux from the flare, and to predict the effect of cross winds on combustion efficiency. In the current application, the code is extended to simulate multiphase reacting flow of coal and biomass. Submodels have been added to approximate coal/biomass devolatilization and char oxidation. Simulations have been performed to demonstrate and validate this capability. This code simulates fast and slow fuel devolatilization, fuel vapor and oxygen transport, chemical reaction and heat release, soot formation, pollutant formation and destruction, diffuse radiation within the combustion zone and radiation from the flame to surrounding objects. Reaction rate and soot radiation parameters are based upon a local equivalence ratio combustion concept which has been calibrated against experimental data. Code validation includes comparing predicted results to experimental data reported in the Brigham Young University data book. Model results indicate the importance of small particles in flame stabilization and combustion efficiency.

KEYWORDS: Coal, Biomass, Combustion, Gasification, Large Eddy Simulation, Computational Fluid Dynamics

I. INTRODUCTION

Simulation of pulverized coal combustion processes involves modeling a number of complex simultaneous interdependent physics as shown conceptually in Figure 1. Several driving motivations exist in today’s coal fired power generation industry including: 1) meeting increasingly tighter environmental regulations focused on reducing NOx, CO, and more recently CO2, 2) identifying the “best” burner, furnace design, or other equipment to use for a specific combustion application, 3) optimizing plant operations to increase production efficiency and plant safety while reducing maintenance costs, and 4) improving plant operating factor by reducing time spent during routine plant shutdowns. Developing and applying a comprehensive computational fluid dynamics (CFD) based multi-physics code allows a user to address each of these issues.

Accurate CFD modeling of coal combustion requires that the code allow for turbulent particle laden gas flow with homogeneous gas phase combustion chemistry, particle mass change due to moisture evaporation, particle devolatilization and char combustion, trace chemistry associated with NOx/SOx formation, and radiation heat transfer from the flame to the surrounding furnace walls with participating media including non-grey gases and absorbing/emitting particles. A significant amount of work has been done over the years in developing and applying advanced CFD models for coal combustion [1-9]. One of the main challenges for these models has been accurately capturing the strong multi-phase coupling between reacting coal particles and their effect on the gas phase. Most of the previous work has adopted a Lagrangian/Eulerian treatment for the particles/gas coupling [1-7]. The main justification for using a particle trajectory model has been to accurately approximate the combusting coal particle history effects.

1 Corresponding author

Page 2: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 2

However, to obtain a detailed distribution of particle velocity and concentration for comparison with experimental data, a very large number of particle trajectories are required. Alternatively, Fiveland and Wessel [10] developed an Eulerian model for modeling pulverized coal combustion, while neglecting the velocity slip between the gas phase and coal particle phase, and assuming that the temperature of coal particle phase is equal to the temperature of gas phase, the temperature distribution of the gas-particle mixture can be obtained by solving the overall energy equation. More recently, Thornock et.al [11] has extended a previous Large-Eddy Simulation (LES) based CFD model to coal combustion using a Direct Quadrature Method of Moments (DQMOM). This work has shown the ability to accurately capture particle history effects and simulate particle size effects related to size segregation which is thought to cause large particles to have higher carbon in ash than small ones.

In this paper, a pure two-fluid model for reacting gas-particle flows is developed, using a comprehensive Eulerian treatment for both gas phase and particle phase. Velocity and temperature slip between coal particles and gas phase can be calculated by solving the momentum equations and energy equations of gas phase and particle phase respectively. In addition, a general model of pulverized coal devolatilization and char oxidation is included in the comprehensive model. For volatile and CO combustion as well as radiation heat transfer, the conventional EBU-Arrhenius model and the six heat-flux model were used. The model is designed to address issues related to 1) coal conversion or Loss on Ignition (LOI), 2) NOx and CO emissions from coal furnaces, 3) radiant heat transfer from the flame to the pendants and water walls inside the furnace, and 4) coal-biomass combustion and gasification.

A critical part of the model development process includes validation and verification/uncertainty analysis (VV/UA) work necessary to quantify uncertainty in model predictions [12]. For this work, a general V&V hierarchy has been established as shown in Figure 2.

II. PREVIOUS CODE VALIDATION AND VERIFICATION

A previous version of the code has been used to simulate large pool fires and gas flares. Simulation of a well documented experiment involving a 4.66-m diameter culvert pipe located on the leeward edge of an 18.9 m diameter pool fire in a crosswind with average speeds of 4.6 m/s test is included to illustrate code validation for large combustion systems. The code accurately calculated the time-dependent temperature distributions for this combustion system. A second validation example is included that shows predicted flame shape and temperature for a multi-tip ground flare burning ethylene under wind and no-wind conditions compared against test results. Here the code accurately predicted the measured flame height, radiant flux profile to the ground, and smoke generation for this fuel under these wind conditions.

To accomplish rigorous code verification and validation with a detailed uncertainty analysis, several levels of simulations and comparisons are required. Initially, efforts are focused on analyzing the gas phase combustion chemistry and coal devolatilization/char oxidation kinetic mechanisms required to simulate the homogeneous and heterogeneous chemistry inside a coal furnace. Efforts reported in this paper focus on this level of analysis. The second level of analysis considers “Unit Problems” using the reaction model, the mixing model, and the radiation model required to simulate coal combustion and gasification [12]. Next, several “Benchmark” cases are used to verify model predictions for a complete coal combustion system [13]. In the Benchmark simulations, predictions are compared to measured results on several planes through a test reactor. Predicted scalars include gas temperature, H2O, O2, Volatiles, and CO. Comparisons of the particle phase include plots showing particle temperature, char concentration, and particle burnout (LOI) as well as profiles of wall heat flux and wall temperature. Lastly, model VV/UA work focuses on applying the code to both “Sub-system” and “Complete-system” cases [14]. The work reported in this paper focuses on analyzing the kinetic mechanisms and Unit Problems with one application to a bench scale test case. Future work will focus on comparing model predictions for additional bench mark cases as well as a full scale industrial case. Future work will also include comparing code predictions to results from a RANS based code such as Fluent.

Page 3: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 3

Figure 1- Components of a Pulverized Coal CFD model

Figure 2 - Required steps for validation and verification of a pulverized coal combustion CFD

model

CFD BASED ANALYSIS

Particle/Droplet Reactions(Surface Chemistry)

VaporizationDevolatilization

Pyrolysis

Pollutant Formation (Trace Chemistry)

SOx – Non-EquilibriumNOx – Fuel and Thermal

PIC – Incineration

Numerical ApproximationValidation Data

Radiation Properties

Radiation Heat Transfer

Turbulent DispersionWall Deposition

Nucleation/Agglomeration

Particle Fluid Mechanics(Multiphase Flow)

Momentum EquationsEnergy Equation

Turbulence Model

Gaseous Fluid Mechanics(Turbulent Flow)

Gaseous Reactions(Homogeneous Chemistry)

Local EquilibriumTurbulence Coupling

Eddy Dissipation

CFD BASED ANALYSIS

Particle/Droplet Reactions(Surface Chemistry)

VaporizationDevolatilization

Pyrolysis

Particle/Droplet Reactions(Surface Chemistry)

VaporizationDevolatilization

Pyrolysis

Pollutant Formation (Trace Chemistry)

SOx – Non-EquilibriumNOx – Fuel and Thermal

PIC – Incineration

Pollutant Formation (Trace Chemistry)

SOx – Non-EquilibriumNOx – Fuel and Thermal

PIC – Incineration

Numerical ApproximationValidation Data

Radiation Properties

Radiation Heat Transfer

Numerical ApproximationValidation Data

Radiation Properties

Radiation Heat Transfer

Turbulent DispersionWall Deposition

Nucleation/Agglomeration

Particle Fluid Mechanics(Multiphase Flow)

Turbulent DispersionWall Deposition

Nucleation/Agglomeration

Particle Fluid Mechanics(Multiphase Flow)

Momentum EquationsEnergy Equation

Turbulence Model

Gaseous Fluid Mechanics(Turbulent Flow)

Momentum EquationsEnergy Equation

Turbulence Model

Gaseous Fluid Mechanics(Turbulent Flow)

Gaseous Reactions(Homogeneous Chemistry)

Local EquilibriumTurbulence Coupling

Eddy Dissipation

Gaseous Reactions(Homogeneous Chemistry)

Local EquilibriumTurbulence Coupling

Eddy Dissipation

Page 4: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 4

The LES based code described here produces more accurate simulations by involving larger time and length scales which reduces the dependence on modeling assumptions related to time averaging to approximate turbulent mixing that other RANS based CFD codes rely on. This LES based code produces temporally and spatially resolved predictions of a turbulent reacting gas flow with particles to predict particle segregation and clustering which are clearly shown important in the particle devolatilization process. This feature is critical to simulating highly turbulent lean coal flames predicted to blow out by RANS based CFD codes.

The code being developed for this work is based on previous research activities related to external buoyant reacting flow systems including pool fires and gas flares. The following sections summarize previous VV/UA work completed for this code with references given to more extensive documentation of this work.

A. CULVERT PIPE POOL FIRE

The heat transfer from large pool fires to engulfed objects has been analyzed to predict the total heat transfer to an object and the general object temperature distribution for a variety of fire and wind conditions [15]. The simulation accounted for fuel evaporation rate and radiation heat transfer to accurately model large fire heat transfer using a relatively coarse computational mesh. Reaction rates and soot radiation model parameters were based on experimental data reported elsewhere [15]. In this validation, model calculations were performed to simulate the conditions of three experiments that measured the temperature response of a 4.66 m diameter culvert pipe located at the leeward edge of 18.9 m and 9.45 m diameter pool fires in crosswinds with average speeds of 2.0, 4.6, and 9.5 m/s (see Figure 3). The model accurately calculated the time dependent temperatures in all three experiments (see Figure 4).

Figure 3 - Culvert test for model validation

Thermocouple locations in culvert

Experimental layout for culvert

Page 5: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 5

Figure 4 - Comparison of predicted (left) vs measured (right) culvert temperature profile B. ETHYLENE FLARE RADIATION

A detailed analysis of a high pressure low profile gas flare has been performed [16]. A three-step reaction mechanism, similar to that used by Greiner [17] for military grade jet fuel (JP8) fires, was used in the present simulation. The first reaction in this mechanism burns half the hydrogen contained in the hydrocarbon, and any free hydrogen while the second reaction burns the remaining hydrogen in the hydrocarbons and most of the carbon and produces some soot and the final reaction burns the soot produced in the second reaction.

To validate the model, predicted flame height and flame radiation were compared to measured flame height and radiation for a single flare burner (Figure 55). Two simulations were performed using a course mesh and a refined mesh to test for mesh sensitivity. The finer mesh density had 97,000 cells (45 x 44 x 49) for a physical domain of 6 m X 6 m X 25 m which was used for all predictions compared to flare measurements. Two burner sizes operating at two tip pressures were tested with flame height observed and radiation data taken at both 15 meters and 30 meters from the flare burner. The measured flame height was between 13.8 m and 15.3 m and compared well to the predicted height of approximately 15 m.

Radiation predictions included the effect re-radiation from the heated ground plus radiation reflected by the ground. In addition, radiation calculations included atmospheric attenuation using Hamins’ model [18] which depends on ambient temperature, source temperature, and relative humidity and accounts for radiation absorption by water vapor and carbon dioxide in clear air. As shown in Table 11, when ground re-radiation and atmospheric transmissivity effects are included, the predicted radiation fluxes compare very well with the measured values.

C. COAL AND BIOMASS COMBUSTION MODEL.

The coal and biomass combustion model is based upon a set of both parallel and series chemical reactions. Typically coal has a particle size distribution. The fine particles will heat up rapidly and coarse particles will heat much slower resulting in different devolatilization and combustion response. The size distribution could be modeled as a sum over several distinct species representing different diameters, and each species has its own set of chemical reaction rates to include the effect of particle size on reaction rates. Each species has its own set of rates to include the effect of particle size on reaction rates. In this example a 50 micron coal dust will be modeled as a sum over two sizes, fine and coarse. Each species has its own set of chemical reactions to model low and high temperature devolatilization, and char combustion.

300

500

700

900

1100

1300

1500

1700Top

Lee Top

Leeward

Lee Bot

Bottom

Wind Bot

Wind

Wind Top

Left EndLeftCenterRightRight End

300

500

700

900

1100

1300

1500

1700Top

Lee Top

Leeward

Lee Bot

Bottom

Wind Bot

Wind

Wind Top

left endleftmiddlerightright end

Page 6: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 6

Figure 5 – Comparison of measured fluctuating flare flame from a single burner (left hand image

documents observed range of observed flame heights) to predicted results (snap shot shows range of predicted flame heights from transient LES simulation). Note: flame surface estimated

using 0.5 ppm soot iso-surface. The devolatilization process has two ranges, fast and slow, similar to the Kobayashi model [19], and each process occurs sequentially where the fast process produces a new species which has the second, slower rate of devolatilization. The devolatilization process is modeled by Arrhenius kinetics, where the activation energy (or temperature) is modeled by physio-chemical processes characteristics of the coal (or biomass) type. Therefore the activation energy is independent of particle size and assumed identical for both fine and coarse particle sizes. On the other hand heat transfer and surface area effects are dependent on the pre-exponential Arrhenius coefficient as are numerical modeling assumptions such as the number of particle species chosen to model the actual size distribution. In this model the activation energy is chosen based upon the work of others, whereas the pre-exponential coefficient is adjusted to match experimental data.

The experimental apparatus and data chosen for this sample validation is the BYU coal combustor. The combustor consists of a cylindrical chamber 0.2 m in diameter, with a co-annular nozzle at one end (see Figure 6). The inner injector is 2.2 cm in diameter which injects a mixture of coal particles and gas, (mostly air) of a specified composition and temperature. The outer annulus is 8.4 cm diameter and injects air at a specified mass flow rate, temperature and swirl. Several datasets were gathered using this apparatus with varying diameter nozzles, flow rates and coal types and many of those data sets appear in the ACERC combustion data book as reported by Rasband [13].

15 m (49 ft)

15 m (49 ft)

20 m (66 ft)

20 m (66 ft)

Flare tip located 2 m (~7 ft) above

grade

13.8 -15.3 m (48’-53’)

Non-luminous

Flame ~1 m

Page 7: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 7

Table 1 – Radiation Predictions and Test Results Comparison

Tip Size

Position (m)

Burner Pressure

(psi)

Predicted Radiation

(W/m2)

Measured Radiation

(W/m2) Difference

(%) 3 15 2.8 2700 3344 -20.0 % 3 15 11.4 6150 6192 -0.7 % 3 30 2.8 650 671 -3.0 % 3 30 11.4 1650 1532 +8.0 % 4 15 2.8 4325 6371 -32.0 % 4 15 11.4 10000 9536 +5.0 % 4 30 2.8 1150 1513 -23.0 % 4 30 11.4 3250 2747 +18.0 %

For this validation, the Asay coal combustion dataset was selected. As reported, the Asay data

includes oxygen concentration distribution at several axial and radial locations through out the combustor. The vertical cylindrical combustor has refractory lined walls and consists of several flanged sections stacked vertically (approximately 200 cm high when fully assembled) into which coal and air were fed through the top co-annular burner nozzle. Each flanged section included access ports through which suction probes were inserted to collect gas, temperature, and particle samples along the axial length of the reactor.

Conditions for this experiment are a central injector mass flow of 2.835e-3 kg /sec coal and 6.228e-3 kg/sec gas and an inlet temperature of 300 K. The outer annulus has a mass flow rate of air at 0.019 kg/sec and a temperature of 589 K. The combustor outlet was modeled as a constant pressure boundary that allows free exit of particles and gases.

Figure 6 - Experimental apparatus used to collect the Asay coal combustion data reported by

Rasband [13]

Page 8: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 8

Since ash and char concentrations were not available for this data set, direct comparisons to coal particle conversion were not possible. Thus, this validation should only be considered as a partial validation of the coal combustion model.

The kinetics based combustion model developed to simulate coal combustion included 11 distinct species and eight reactions including:

1. Fine Coal Particles (FCP) 2. Coarse Coal Particles (CCP) 3. Volatile Organic Hydrocarbons from the devolatilization process (VOC) 4. Partially Devolatilized Fine Particles (PDFP) 5. Partially Devolatilized Coarse Particles (PDCP) 6. Soot articles (Soot) 7. Fine Char Particles (FPchar) 8. Coarse Char Particles (CPchar) 9. Ash (Ash) 10. Oxygen (O2) 11. Products of Combustion (PC)

The chemical reactions and parameters2 developed to model this experiment included the following steps.

Fine coal particles (FCP) undergo a fast devolatilization reaction with associated Arrhenius kinetics:

1.0 FCP 0.5 VOC + 0.5 PDFP (1)

The Activation temperature for this equation set to 12,581 K with the Pre-exponential coefficient set to 1.0e11 sec-1.

Partially devolatilized fine coal particles (PDFP) undergo slow devolatilization with associated Arrhenius kinetics:

0.5 PDFP 0.2 VOC + 0.3 FPchar (2)

The Activation temperature for this equation is also set to 12,581 K with the Pre-exponential coefficient set to 1.0e7 sec-1.

Fine Char Particle (FPchar) combustion uses the global combustion form:

0.3 FPchar + 0.72 O2 0.7 PC + 0.05 Ash (3)

The Activation temperature for this equation was set to 7,337 K, the Pre-exponential coefficient was set to 1.4e7 sec-1, the oxygen exponent was set to 0.5, the Char particle exponent was set to 0.33 and the Temperature exponent was set to 0.6.

Coarse Coal Particles (CCP) undergo fast devolatilization with associated Arrhenius kinetics:

1.0 CCP 0.2 VOC + 0.8 PDCP (4)

The Activation temperature for this equation set to 12,581 K and the Pre-exponential coefficient set to 1.0e8 sec-1.

Partially Devolatilized coarse particles (PDCP) undergo slow devolatilization with associated Arrhenius kinetics:

0.8 PDCP 0.25 VOC + 0.55 CPchar (5)

The Activation temperature for this equation was set to 581 K and the Pre-exponential coefficient set to 2.0e7 sec-1.

2 Arrhenius parameters for each reaction were established by adjusting them to obtain the best fit to the Asay coal combustion data set.

Page 9: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 9

Coarse Char particle (CPchar) combustion uses the global combustion form as:

0.55 CPchar + 1.32 O2 PC + 0.143 Ash (6)

The Activation temperature for this equation was set to 7,337 K, the Pre-exponential coefficient was set to 0.43e8 sec-1, the oxygen exponent set as 0.5, the Char particles exponent set as 0.33 and the Temperature exponent set as 0.6.

Volatile Organic Hydrocarbon (VOC) combustion and soot formation follow 2nd order Arrhenius kinetics as:

1.0 VOC + 2.5 O2 3.29 PC + 0.21 Soot (7)

The Activation temperature for this equation was set to 15,500 K and the Pre-exponential coefficient set to 1.0e15 sec-1.

Soot combustion uses the global combustion form as:

1.0 Soot + 2.66 O2 3.66 PC (8)

The Activation temperature was set to 25,500 K, the Pre-exponential coefficient set as 2.0e10 sec-1, the oxygen exponent set as 1.0, the Char particles exponent set as 0.33 and the Temperature exponent set as 0.6.

The coefficients in the chemical reaction formulas (Eqs. 1-8) are mass based instead of mole based. The net reaction energy for all the reactions combined is 21 MJ/kg raw coal. The activation temperature is the activation energy divided by the universal gas constant. The pre-exponential coefficients were adjusted to match the results from the current experiment. For global reaction rates, the concentration and temperature exponents were taken either from work by others, or from previous work on sooty pool fires.

The result of these computations revealed an interesting trend. Fine particle devolatilization is required to maintain combustion! Without its contribution to the local energy profile, the flame goes out. The coarse particles release VOC’s too slowly to maintain combustion in the current flow configuration. The fine particles release VOC’s much easier which maintains a flame close to the nozzle exit.

III. COMPARISON TO ACERC DATABOOK TESTS.

Figure 7 and Figure 8 shown below compares the predicted radial oxygen profile at several locations downstream of the nozzle to the measured data from the Asay test case. In each figure shown below several lines appears which represent a 0.25 second time averaged profile of the oxygen concentration. The number of lines is 8 which correspond to a total of 2 seconds comparison time. Since the code used here (called C3d) is an LES based CFD code, the large scale turbulence is resolved which results in time varying fluctuations of all variables. The time varying fluctuations will average out to a fixed profile after 10’s of seconds, however in this work the time averaging was limited to 8 X 0.25 second average profiles to illustrate the magnitude of the fluctuations. Each figure demonstrates reasonable agreement with the measured values for the current combustion model. Data types, other than oxygen concentration, were not available for this experiment so this work represents only a partial validation of the combustion model for this coal combustion application.

Page 10: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 10

a) b)

Figure 7- Comparison of predicted radial oxygen concentration at various times along axial reactor length (lines) compared to experimental data from Asay (solid dots): a) 0.41 m from nozzle

and b) 0.71 m from nozzle

Figure 8 - Comparison of predicted radial oxygen concentration at various times along axial

reactor length (lines) compared to experimental data from Asay: a) 1.01 m from nozzle and b) 1.32 m from nozzle

Figure 9 through Figure 12 show predicted 2-dimensional contour slices of selected species including oxygen, coarse particle Char, Volatile Organic Compounds (VOC), and Ash on a normalized linear scale from 0 (blue) to 1 (magenta).

Figure 9 - Snapshot of transient 2D contour slice of the oxygen concentration

Page 11: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 11

Figure 10 - Snapshot of transient 2D contour slice of coarse char particles

Figure 11 - Snapshot of transient 2D contour slice of Volatile Organic Compounds (VOC)

Figure 12- Snapshot of transient 2D contour slice of ash Concentration

IV. SUMMARY AND CONCLUSIONS

A comprehensive numerical model for pulverized coal combustion (C3d) has been developed based on Large Eddy Simulation. The gas and particle phases have been modeled using an Eulerian framework as opposed to previous work using a Eulerian/Lagrangian framework. This model approximates multiple particle sizes and considers particle devolatilization and char oxidation for each particle size. Gas phase chemistry is approximately using Eddy-Dissipation type equations with Arrhenius kinetics. The predictions have been compared to experimental results from the ACERC databook for the Asay case. Model validation and verification work for this code is ongoing to further investigate the application of C3d to full scale combustion systems for coal and biomass materials. Particle radiation is also included in this analysis to estimate particle heat up due to radiation transfer and wall heat transfer from the flame. The predicted results illustrate the impact that fine particles have on the combustion process in flame stabilization. The CFD code described in this paper is commercially available to public and can be obtained by request from Systems Analyses and Solutions.

V. REFERENCES

[1] - Smoot , L. D. and Smith, P. J., Coal Combustion and Gasification. Plenum Press, New York, 1985.

[2] - FLUENT User’s Guide. Fluent Inc., Centerra Resource Park, 10 Cavendish Court, Lebanon, NH.

[3] - STARCD User’s Guide. CD adapco group, Melville, NY.

[4] - Lockwood, F.C., Papadopoulos, C., and Abbas, A.S.,”Prediction of a Corner-Fired Power Station Combustor,” Comb Sci & Tech, 58, 5 (1988).

Page 12: DEVELOPMENT AND APPLICATION OF AN LES BASED CFD CODE … · 2018. 2. 12. · particle trajectory model has been to accurately approximate the combusting coal particle history effects

9th European Conference on Industrial Furnaces and Boilers (www.cenertec.pt/infub) April 26-29, 2011/Estoril, Portugal

  

Page 12

[5] - Wennerberg, D., “Prediction of Pulverized Coal and Peat Flames,” Comb Sci & Tech., 58, 25 (1988).

[6] - Truelove, J.S., “Prediction of the Near-Burner Flow and Combustion in Swirling Pulverised Coal Flames,” 23rd Symposium (International) on Combustion, The Combustion Institute, 963, (1990).

[7] - Papadakis, G. and Bergeles, G., “Prediction of Staged Coal Combustion in Three–Dimensional Furnaces,” J of the Institute of Energy, 67, 156 (1994).

[8] - Coimbra, C.F.M., Azevedo, J.L.T. and Carvalho, M.G., “3-D Numerical Model for Predicting NOx Emissions from an Industrial Pulverized Coal Combustor” Fuel, 73, 1128 (1994).

[9] - Lockwood, F.C., Romo-Millares, C.A., “Mathematical Modeling of Fuel-NO Emissions from PF Burners,” J. of the Institute of Energy, 65, 144 (1992).

[10] - Fiveland, W.A. and Wessel, R.A. “A Numerical Model for Predicting the Performance of Three-Dimensional Pulverized-Fuel Fired Furnaces,” J Eng for Gas Turb and Power, 110, 117 (1988).

[11] - Thornock, J.N., Reid, C.M., Pedel, J., and Smith, P.J., “Toward Prediction of Reacting Coal Systems using Large Eddy Simulation (LES) with the Direct Quadrature Method of Moments (DQMOM),” American Flame Research Committees - International Pacific Rim Combustion Symposium, Advances in Combustion Technology: Improving the Environment and Energy Efficiency, Sheraton Maui, Hawaii - September 26 –29 (2010)

[12] - Smith, J.D., Smith, P.J., Hill, S.C., "Parametric Sensitivity Study of a CFD-Based Coal Combustion Model," AIChE Journal, Vol. 39, No. 10, October (1993).

[13] - Rasband, M.W., “PGCC-2 and the Data Book: A Concurrent Analysis of Data Reliability and Code Performance,” M.S. Thesis, Brigham Young University, Provo, Utah, USA, (1988).

[14] - Gorner, K. and Zinser, W. “Prediction of Three-Dimensional Flows in Utility Boiler Furnaces and Comparison with Experiments,” Comb Sci & Tech., 58, 43 (1988).

[15] - Greiner, M., and Suo-Anttila, A., "Validation of the ISIS Computer Code for Simulating Large Pool Fires Under a Varity of Wind Conditions," ASME J. Pressure Vessel Technology, Vol. 126, pp. 360-368 (2004).

[16] - Smith, J.D., Suo-Ahttila, A., Smith, S.K., and Modi, J., “Evaluation of the Air-Demand, Flame Height, and Radiation Load on the Wind Fence of a Low-Profile Flare Using ISIS-3D,” AFRC-JFRC 2007 Joint International Combustion Symposium, Marriott Waikoloa Beach Resort, Hawaii, October 21-24, (2007).

[17] - Greiner, M., and Suo-Anttila, A., "Radiation Heat Transfer and Reaction Chemistry Models for Risk Assessment Compatible Fire Simulations," Journal of Fire Protection Engineering, Vol. 16, pp. 79-103 (2006).

[18] - Fuss S.P., A. Hamins. “An Estimate of the Correction Applied to Radiant Flame Measurements Due to Attenuation by Atmospheric CO2 and H2O”, Fire Safety Journal, Vol. 37, pp. 181-190 (2002).

[19] - Kobayshi, H., Howard, J. B., and Sarofim, A. F., “ Coal Devolatilization at High Temperatures”, 18th Symposium (International) on Combustion, The Combustion Institute, Pittsburgh, PA, 411 (1977).