Large eddy simulations of coal gasification in an entrained flow gasifier

Fuel 104 (2013) 664–680

Contents lists available at SciVerse ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Large eddy simulations of coal gasification in an entrained flow gasifier Neerav Abani a,⇑, Ahmed F. Ghoniem b a b

Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, 3-342A, Cambridge, MA 02139, United States Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, 3-342, Cambridge, MA 02139, United States

h i g h l i g h t s " Multi-phase turbulent flow and coal gasification in an entrained flow gasifier is simulated using LES and RANS. " LES captures unsteady flow structures in both combustion and gasification zones of the gasifier. " Unsteady flow structures affect mixing and char-conversion efficiency. " LES accurately predicts char-conversion efficiency compared to RANS.

a r t i c l e

i n f o

Article history: Received 28 December 2011 Received in revised form 21 May 2012 Accepted 3 June 2012 Available online 23 June 2012 Keywords: Coal gasification Multi-phase flow Entrained flow gasifier CFD Large eddy simulations

a b s t r a c t In this paper, we investigate multi-phase reacting flow in a coal-fed entrained flow gasifier using largeeddy simulations and Reynolds-averaged Navier Stokes models, respectively. An axial-flow type lab-scale gasifier is investigated. The simulations are performed using a Lagrangian–Eulerian method in which the coal particles are modeled using a Lagrangian approach and the gas phase is solved using an Eulerian approach. We compare the performance of LES and RANS results. The coal particle models include devolatilization, char consumption that uses heterogeneous chemistry and two-way coupling of mass, momentum and energy with the surrounding gas phase. The gas phase combustion is modeled using homogenous chemistry and the effect of turbulence on combustion and gasification is modeled using a partially stirred reactor approximation. Results show that LES captures the unsteady flow structures inside the gasifier. We show that modeling the unsteady mixing is critical to the accurate predictions of the gas phase species and carbon conversion in these gasifiers. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Entrained Flow Gasifiers (EFGs) are key component in Integrated Gasification Combined Cycle (IGCC) power plants designed to reduce emissions along with the potential for carbon dioxide capture and the integration with synthetic fuel production. In EFG, gasification converts coal into clean synthetic gas. Entrained flow gasifiers have been designed using zero-dimensional models. There is renewed interest in designing EFGs that can handle different feedstock, and to gain better insight into the gasification process for, e.g., the development of compact and more reliable gasifiers [1]. The flow inside an entrained flow gasifier is inherently unsteady and involves complex turbulent mixing of two phases; coal particles in solid phase and oxygen and steam in the gaseous phase. The gasification process involves phenomena such as devolatilization, heterogeneous surface reactions and complex gas phase ⇑ Corresponding author. Present address: Achates Power Inc., 4060 Sorrento Valley Bl., San Diego, CA 92121, United States. Tel.: +1 6083586266. E-mail addresses: [email protected], [email protected] (N. Abani), [email protected] (A.F. Ghoniem). 0016-2361/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2012.06.006

chemistry. To develop a fundamental understanding of this multiphase reactive process, a comprehensive, high fidelity Computational Fluid Dynamics (CFD) model is needed. These higher fidelity models can be used to improve injection strategies for higher carbon conversion efficiency, detect potential failure modes such as the formation of hot-spots that cause thermal wear, and examine the overall reliability and fuel flexibility of different designs. There are few experimental investigations of entrained flow gasifiers in literature. The available data are mostly limited to lab scale gasifiers and much less for pilot scale gasifiers. Intrusive probe are often used for sampling. Brown et al. [2], performed such experiments in the oxygen fired Brigham-Young University labscale gasifier. They obtained temperature and syngas composition measurements along the gasifier axis. This lab-scale gasifier operated at atmospheric pressure and the results covered four different types of coal. Hill and Smoot [3] performed CFD simulations using RANS model on these gasifiers and compared their results with the measured data. In the BYU axial flow gasifier coal particles were injected along with oxygen from a central nozzle and steam was injected from the surrounding secondary nozzle hole as shown in Fig. 1. The Mitsubishi Heavy Industries (MHI) gasifier, on the other

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Fig. 1. Schematic of BYU gasifier.

hand, is air-blown swirling flow type gasifier. Both research-scale and pilot-scale versions were built, with coal flow rates of 2 tons/ day and 200 tons/day, respectively. These gasifiers have two sections, a lower combustion section and an upper gasification section. Coal particles are injected along with air jets tangentially in the combustor section to create swirling flow inside the gasifier. In the gasification section more coal is injected. Various researchers have performed simulations of the MHI two-stage gasifier for both the research scale (Watanabe and Otaka [4]) and the pilot scale (Chen et al. [5,6]). Two approaches have been used for multiphase reactive flows similar to what is encountered in gasification. These are Lagrangian–Particle Eulerian–Fluid models (LPEF) and Eulerian–Particle Eulerian–Fluid models (EPEF). In the LPEF method, the solid phase particles are tracked using a Lagrangian approach, while the surrounding gas phase is modeled using Eulerian approach. In both cases, the two phases are coupled through source terms in the conservation equations of mass, momentum and energy. In the EPEF method, both the solid and gas phases are solved using an Eulerian approach and an additional equation is solved the ‘‘volume fraction’’, which represents the fractional volume of the solid-phase locally. Typically the EPEF is a good choice for cases in which the solid phase occupies high volume and the velocities of the flow are relatively small [7]. The EPEF method is better for calculating the group effect of the solid phase in regions where the local volume fraction of the solid phase is high. On the other hand, the LPEF method is widely used for flows in which the solid particles are widely dispersed within the flow and the flow velocities are much higher as typically found in EFGs [8]. The LPEF methods have also been used in many other combustion systems, such as diesel engines and gas turbine combustors where liquid droplets are tracked. Recently there have been several investigations of EFGs using one-dimensional and RANS approach. Monaghan and Ghoniem [9] developed a 1D gasification model using a reactor network. Their model showed good agreement with experimental data along the length of the gasifier. Watanabe and Otaka [4] performed multi-dimensional CFD-RANS simulations using the LPEF method for the research scale MHI gasifier. Kumar and Ghoniem [10] performed computations of entrained flow gasifiers that range from commercial, research and pilot scale gasifiers using different turbulence models in RANS. They found that while k–e performs well for gasifiers with straight injection, k–x performs much better for swirling-flow type gasifiers. Gasification processes at the singleparticle level have also been modeled at various complexity levels.

665

Kobayahsi et al. [11] and Ubhayaker et al. [12] investigated different coal types to characterize char consumption kinetics and devolatilisation processes. Goetz et al. [13] performed simulations to characterize the impact of char-consumption kinetics. In recent years, Katijani et al. [14] studied char-kinetics and gasification in carbon dioxide and steam in a pressurized drop tube furnace (PTFD). Singer and Ghoniem [15] focused on developing sub-particle structure models and their impact on char consumption. So far, higher fidelity CFD simulations such as LES of coal gasification have not been attempted. In this paper, we apply LES to simulate coal-gasification processes. LES models can capture some of the unsteady structures that affect mixing, turbulent-chemistry interactions, turbulence dispersion of particles, etc. Data from the Brigham Young University’s lab scale gasifier is used to validate the LES simulation. These data include the exit temperature, and gas composition along the length of the gasifier for four different coal types. The measured data also include gas composition distribution in the radial directions of the gasifier at various axial locations. We use the LPEF approach in the open source code OpenFOAM, along with turbulence modeling using the one-equation eddy viscosity model, and compare the results with the k–e RANS model. Heterogeneous surface reactions, devolatilization, and two-way coupling of particle-gas phase are included. 2. Solid phase numerical model The solid-phase model accounts for fuel conversion via pyrolysis and char consumption, and particle transport. The particle transport model solves the mass, momentum and energy equations of the particle along the jet trajectories [10]. In the context of coal gasification, we discuss key models used in this investigation, viz., pyrolysis and char conversion chemistry. The BYU gasifier operates at high temperatures near the nozzle region and hence pyrolysis occurs at a very fast rate.

Coal ! a1 CHx þ a2 H2 þ a3 CO þ a4 CO þ a5 H2 O þ a6 N2 þ a7 Char ð1aÞ X

ai ¼ 1

ð1bÞ

i

Consistent with Badzioch’s and Hawksley’s [16] approach, the devolatilization rate is given by a single kinetic rate that is similar to the Arrhenius form.

  dmV E0 mV ¼ A exp  dt RT

ð2Þ

where mV is the mass of the volatiles remaining in the particle, A = 2.1106 s1 and E0 = 2.1107 J/kmol, and T is the temperature of the particle. The devolatilization process is assumed to energetically neutral since the heat of devolatilization is negligible as compared to heat of reactions due to char consumption and combustion reactions. 2.1. Char conversion chemistry The coal particle is left with char and ash after all the volatile components are released. Char reacts in the presence of steam, oxygen and carbon dioxide and gets converted into carbon monoxide and hydrogen.

C þ 1=2O2 ! CO

ð3aÞ

C þ CO2 ! 2CO

ð3bÞ

C þ H2 O ! CO þ H2

ð3cÞ

666

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Reaction (3a) is the oxidation reaction of char which is exothermic. Reactions (3b) and (3c) on the other hand are endothermic gasification reactions. The rates for these reactions depend greatly on the coal types and the gasification conditions. Brown et al. [2] reported different rates for bituminous, lignite and sub-bituminous coals at atmospheric pressure. Kajitani et al. [14] performed experiments in a pressurized drop tube furnace and measured reaction rates in the pressure range 0.2–2.0 MPa and temperature range of 1400–1800 K. In this study, we simulate gasification process at atmospheric pressure and the reaction rate constants are taken from Brown et al. [2] analysis of the BYU gasifier. The ash content is assumed to be carried along with the particle, exiting the gasifier without taking part in any reactions. Char consumption rates have been modeled widely using the single film model [16,17] that accounts for the pore diffusion of the gasifying agents within the char particle. A more detailed model is the double film model [3,18] where the gasifying agents CO and O2 react in a diffusion flame that envelops the particle to produce CO2. This coupled model is computationally expensive. Zhang et al. [19] proposed a simplified model of char consumption called the moving flame front model which is suitable for practical CFD simulations. In all of these models char consumption rate is evaluated as explicit functions of kinetic and diffusion rate. We implemented this model in our LES. The overall reaction rate for a ith species with a partial pressure in the gas phase surrounding of particle, pi,g is given as:

r c;i ¼ pi;g

r diff;i rkin;i r diff;i þ r kin;i

r kin;i ¼ Ai T bpi exp

r diff;i ¼ C i



Ei RTp

ð4Þ

Table 1 Coal properties. Contents

Utah bituminous

Moisture Ash Volatiles Fixed carbon High heating value (MJ/kg, dry)

2.4 8.3 45.6 43.7 29.8

Elemental analysis, dry (wt.%) Ash H C N S O

8.5 6 71.0 1.3 0.5 12.7

Table 2 Coal properties. Contents

Utah bituminous

(Eq. (7)) AO2 (ms1 K1) (Eq. (7)) BO2 (ms1 K2) (Eq. (3b)) ACO2 (ms1 K1) (Eq. (3c)) AH2O (ms1 K1) (Eq. (3b)) ECO2 (J kmol1) (Eq. (3c)) EH2O (J kmol1) (Eq. (3b)) bCO2 (Eq. (3c)) bH2O Ci (i = O2, CO2, H2O) (s K0.75)

1.68102 1.32105 8.3 45.6 43.7 43.7 1 1 51012



½0:5ðT g þ T p Þ0:75 d

ð5Þ

ð6Þ

where rc,i is the net reaction rate with the ith species, rdiff,i is the diffusion rate and rkin,i is the kinetic rate. The kinetic and diffusion rates are given by Eqs. (5) and (6), respectively. Tp and Tg are particle and gas phase temperatures, d is the particle diameter and Ci is the mass diffusion constant. Ai, b and Ei are the parameters typical of the Arrhenius forms of kinetic rates. Different coal types have different constants for kinetic and diffusion rates. In the present study we investigate Utah Bituminous coal of which the composition is given in Table 1. There have been other forms of reaction kinetics of char oxidation reaction (3a). Brown et al. [2] used a parabolic form of the reaction rate as reported by Field [22] as following:

r kin;O2 ¼ T p ðAO2 þ BO2 T p Þ where AO2 ¼ 1:68  102 ms1 K1 ; BO2 ¼ 1:32  105 ms1 K2

ð7Þ

These constants are valid for Utah Bitimunous and Illinois Bitimious coal types and have also been used in the study conducted by Brown and Smoot. In our study, we use Eq. (7) to model reaction kinetics of char oxidation (reaction 3a) and Eq. (5) for reaction kinetics for char gasification by carbon dioxide (reaction 3b) and water (reaction 3c). The constants used for reaction kinetics of Utah Bituminous coal are listed below in Table 2. 3. Gas phase model The gas phase is solved using an Eulerian approach formulated as either RANS or LES. While RANS solves for mean values, LES solves for the resolved scale values as determined by the local mesh size. In this work, we use a one equation eddy viscosity

LES model as proposed by Menon et al. [20] which was based on previous contribution by Yoshizawa and Horiuti [21] in statistically deriving sub-grid kinetic energy equation. The sub-grid scale kinetic energy equation is: 3=2

@k @k @ui k @ ¼ sij þ Ce þ þ ui @t @xi @xi @xj D



mtk @k rk @xi

 ð8Þ

where k is the sub-grid scale kinetic energy, mtk is the turbulent kinetic viscosity or sub-grid scale viscosity, sij is the sub-grid scale stress. The constants Ce = 1.0 and rk = 1.0. Turbulent viscosity and sub-grid scale stress are given as follows:

mtk ¼ Ck k1=2 D

ð9aÞ 2 3

sij ¼ 2mtk Sij þ kdij

ð9bÞ

The constant Ck = 0.005. The addition of a transport equation for the SGS kinetic energy improves the accuracy of the LES sub-grid scale stress. The SGS kinetic energy is always positive as shown by Ghosal et al. [22]. The gas phase chemistry is modeled using Jones and Lindstedt [23] kinetics mechanism for hydrocarbon fuels. The hydrocarbon species in our case is CH2 and we adopt the rate constants for lower carbon atom gaseous species, CH4. This mechanism has also been widely used by other researchers, such as Anderson et al. [24]. The reaction mechanism consists of 6 species, CH2, CO, CO2, H2, H2O and O2. N2 and AR are the inert gas species and are also solved in the species transport equations. The mechanism includes 5 reactions with their rate constants are as shown in Table 3 below: Abbreviation FORD represents a forward reaction. Reaction R1 and R2 are the consumption of hydrocarbon through oxidation and steam reaction. Reaction R3 and R4 are hydrogen consumption and water dissociation reactions. Reaction R5 is the reversible water–gas shift reaction. Reaction R4 is the reverse reaction rate

667

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680 Table 3 Gas Phase Chemistry Mechanism (units: cm, s, cal, mol and K). Pre-exponential factor An REACTIONS CH2 + 0.5 O2 ) CO + H2 FORD/CH2 0.5/ FORD/O2/ CH2 + H2O ) CO + 2H2 H2 + 0.5 O2 ) H2O FORD/H2 0.25/ FORD/O2 1.5/ H2O + 0 O2 + 0 H2 ) H2 + 0.5 O2 FORD/H2–0.75/ FORD/O2 1/ FORD/H2O 1/ CO + H2O , CO2 + H2

Reaction order

Activation energy

0

30.0  103

(R1)

30  103 40  103

(R2) (R3)

97.9  103

(R4)

20  103

(R5)

E 7.82  1013

3.0  1011 1.209  1018

7.06  1017

0.275  1013

for H2–O2 reaction and the constants had been derived by Anderson et al. [24] by dividing the forward rates by the equilibrium constant at a series of temperatures and then fitting an Arrhenius expression to the results. Note that the reverse H2–O2 reaction, H2 and O2 have been added as reactants with 0 as stoichiometry coefficients as a convenient way for implementation in the Chemkin format mechanism. The effect of turbulence on homogeneous combustion is modeled using the partially stirred reactor (PaSR) model [25]. Kärrholm [26] used the PaSR model in spray combustion. The turbulent combustion time is the sum of the mixing and chemical reaction times and assumes that the rates of individual reaction steps are equal at any moment. Coal particle combustion shows an initial diffusion flame lifted off the nozzle injector, formed by the burning of the volatiles. Two-way coupling of the gas and the particle phases is an important part of the numerical model for multi-phase reactive flow simulations. The mass transfer of the coal particles to the gas is accounted for by transferring the volatiles and char consumption to the gas phase. Similarly, due to the presence of the particles and the heterogeneous reactions, gas phase species such as O2, CO2 and H2O are consumed. The momentum transfer is accounted for by the drag force on the particle. The particles are also subjected to turbulent dispersion which is modeled using a random walk method proposed by Gossman and Ioanides [27]. This method takes into account the effect of the local eddies affecting the particle trajectories. For both RANS and LES a stochastic turbulent dispersion model has been adopted in this study. The energy transfer terms consists of the heat of reaction from the heterogeneous surface reaction, the latent heat of vaporization of water from the particle to the gas phase, and the convective and radiative transfer terms. Radiation effects are important in the context of coal combustion and gasification. In this work the P-1 model for radiation is adopted. The P1 model is the simplest approximate of more general P–N model which solves partial differential equations for the intensity of thermal radiation. Sazhin et al. [28] found that the P1 model is particularly useful for modeling thermal radiation exchange between the gas and the particles, particularly in coal combustion processes in industrial furnaces. Sazhin et al. [28] showed that an explicit term involving radiation transfer can be substituted into the enthalpy equation. His results for coal combustion agreed well with experiments conducted on industrial furnaces. This explicit approach makes it very suitable for its use in CFD solvers. 4. Test and simulation conditions The geometry of the gasifier is shown in Fig. 2. Coal particles are injected along with oxygen and argon through the primary nozzle.

0 1

0.877

0

The nozzle diameter is 1.3 cm. Steam is injected into the secondary nozzle which is annular and is concentric with the primary nozzle. The outer diameter of the secondary nozzle is 2.86 cm. The gasifier operates at atmospheric pressure and the test conditions for the two coal types are given in Table 4. The total mesh size consists of 333,618 cells and the mesh is refined close to the injector and near the axis of the gasifier. The primary nozzle has about 58 cells, the secondary nozzle has 90 cells and the minimum dimension of the cell near the axis and near the primary injector is 0.5 mm. Both RANS and LES require boundary conditions and initialization. The experiments were conducted by initially igniting a methane/oxygen flame and then establishing a coal/oxygen flame. For

5 cm

180 cm

Primary nozzle hole Secondary annular nozzle hole (1.3 cm diameter) (outer diameter2.85 cm)

Fig. 2. Three dimensional mesh for BYU gasifier.

Table 4 Test conditions. Contents

Utah bituminous

Primary flow rate (kg/s) Primary component mole fraction O2 Ar H2O Primary gas temperature (K) Secondary flow rate (kg/s) Secondary component mole fraction H2O Secondary gas temperature (K) Primary particle loading (kg coal/kg primary gas)

0.00729 0.85 0.126 0.024 367 0.00184 1 450 0.910

668

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Gasification Zone

Combustion Zone

RANS, a high initial temperature field was given with a mixture of CO2 and H2O as by-products of methane/oxygen flame. The walls were assumed to be adiabatic as the heat loss is negligible. For the LES, a converged RANS solution was used for initialization. Using this approach, LES simulations become statistically stationary in about 4–5 flow through times. A sample of coal particles are injected at the rate of 50,000 particles per second. The particles are injected using a stochastic technique and the Rosin Rammler distribution is chosen to represent the variation in particle size with minimum diameter of 10 lm, maximum diameter of 80 lm, and mean diameter of 35 lm, with a Rosin Rammler distribution index of 3.5. 5. Results Results are presented for both RANS and LES of the experiments conducted on BYU gasifier using Utah Bituminous coal. Comparisons are made with measured data available in the literature [2]. 5.1. RANS Fig. 3 shows results from the RANS. Fig. 3a shows the temperature distribution in a plane passing through the axis of the gasifier. The temperature distribution reveals that the gasifier has two zones:

[T·]2 (K2)

T (K)

YC(s)

U (m/s)

(1) A high temperature combustion zone, which is located near the nozzle and is marked by high heat release and rise in temperature. Volatile hydrocarbons get consumed in the presence of oxygen in this zone. (2) A low temperature gasification zone that exists just downstream of the combustion zone in which gasification reactions are dominant at relatively uniform lower temperatures as compared to previous zone.

Fig. 3. Flow features from the RANS simulations in a plane passing through the axis of the cylinder. (a) Temperature distribution, (b) variance of temperature fluctuations, (c) velocity magnitude and (d) coal particle distribution colored with concentration of mass fraction of Char.

In the combustion zone volatiles from the coal particles get released to the surrounding gas phase, mix with oxygen and release heat due to combustion. It can be seen from Fig. 3a that the peak temperature exists away from the axis, in a region between 10 to 15 Dprimary downstream of the primary nozzle. This is typical of jet diffusion flames and in this case occurs because the gaseous volatiles that get released near the nozzle exhibit the characteristics of a gaseous fuel jet. Fig. 3b shows the distribution of temperature fluctuations. The fluctuations are high where the colder oxygen jet issuing from the primary nozzle mixes with the surrounding steam and burnt gases. The peak fluctuations are observed further downstream of the nozzle almost towards the end of the combustion zone. Fig. 3c shows the velocity distribution. The velocity decays due to the jet expansion in the gasifier. Coal particles are subjected to two-way coupling and the stochastic turbulent dispersion that influences both the gas phase and the distribution of the coal particles inside the gasifier. Fig. 3d shows the coal particle distribution according to their size and colored with the mass fraction of char. Near the nozzle, the particle has some volatiles left and hence, the mass fraction of char is low. Very close to the gasifier axis the particles have higher char concentration. This is because these are the large particles with higher momentum that do not get dispersed away from the axis. The smaller particles are easily influenced by the shear layer in the combustion zone because of their smaller momentum; these get dispersed away from the axis. Char consumption is dependent upon the temperature and concentration of carbon dioxide and water vapor. Near the axis, both the temperature and concentration of the gasifying agents are predicted to be lower by the RANS models and hence the particles remain closer to the axis with higher char

concentration. Fig. 3d shows that particles dispersed away from the axis have lower concentration of char because the char gets consumed and the particles are left with only ash. A more detailed analysis of the flow features reveals the presence of a shear layer throughout the gasifier covering both zones as shown in Fig. 4. Fig. 4a shows the axial velocity (gas phase axial velocity, from here on the velocity referred in the paper pertains to the gas-phase velocity) distribution in the range of 0–10 m/s, where the axial velocity with value 0 and lower is colored in blue and with 10 m/s and higher values is colored in red. The shear layer can also be observed in the velocity vector plots of the two zones as shown in Fig. 4b. The vector dimensions are blown up and are not to scale for better visualization of the flow features in the region of shear layer. Because of the high momentum of the jet and the compact dimension of the gasifier, the shear layer exists all the way to the bottom of the gasifer, where the flow is almost fully developed. There is also a corner vortex ring at the upper end of the gasifier close to the injector. Fig. 5 shows the azimuthal component of vorticity in a plane passing through the axis of the cylinder. Fig. 5 also shows the vorticity associated with the shear layer both in the region closer to the injectors and downstream of the injectors. These flow features along with the particle distribution of Fig. 3d reveal that the char consumption is dependent on the mixing. The particles within the shear layer mix with the oxidizing and gasification medium in both the combustion and gasification zones and react. The particles located closer to the axis experience less char consumption due to the lower concentration of steam and carbon dioxide. The gas phase species distribution is shown in Fig. 6a–f. Carbon monoxide and hydrogen concentration is higher near the

(a)

(b)

(c)

(d)

669

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Ux (m/s)

Jet Shear layer

Ux (m/s)

1

(a)

2

(b)

Fig. 4. Shear Layer features from the RANS simulations in a plane passing through the axis of the cylinder. (a) Velocity field scaled from 0 to 10 m/s. Blue colored region exhibit reverse flow, (b) velocity vector in the two zones colored with axial velocity in the range of 0–10 m/s. Zone 1: Combustion zone; zone 2: gasification zone.

ω normal (1/s)

axis and carbon dioxide and water vapor have lower concentration near the axis. 5.2. LES The same operating conditions are used here. Converged results from RANS were used as initial solutions for LES. Because LES models capture the large scale unsteady structures of the flow, it can model mixing more accurately as compared to RANS, especially in turbulent reactive flows such as those found in entrained flow gasifiers. Results in this section are analyzed to examine the unsteady flow features and their contribution on gasification process. Distribution of various quantities, such as the velocity, temperature and species concentration are shown at two instants of time in a plane passing through axis of gasifier. The two times are separated by an interval of 0.25 s, which is close to one-half of an average particle residence time in the gasifier. These two instants are denoted as time t and time t + Dt, where Dt = 0.25 s. Fig. 7a–f shows the distribution of the instantaneous resolved-scale value of temperature, velocity magnitude and coal particle distribution colored with char concentration at time t and t + Dt. The results show that even though the boundary conditions are steady, the large-scale flow features are unsteady. Similar to the RANS results, the peak temperatures are located off-axis and around the shear layer established between the two streams of the particle-laden jet of the coal-oxygen and steam. This shear layer exhibits unsteady flow features that impact mixing and effects the combustion and the temperature distribution. The resolved-scale temperature distribution shows that the length of combustion zone also changes with time as observed from Fig. 7a and d. The location of peak resolved-scale temperature is different at time t and t + Dt, as a result of the unsteady flow features. Fig. 7g and h shows the distribution of the mean temperature and mean temperature variance. The near-nozzle region is characterized by high temperatures and large temperature variation due to combustion.

Fig. 5. Azimuthal component of vorticity distribution in a plane passing through the axis of the cylinder.

The particle distribution shown in Fig. 7c and f exhibits unsteady variations that is effected by the shear layer large-scale structures. The particles are scaled according to their size and colored according to the char concentration remaining in the particle. In the initial high momentum jet in the combustion zone the larger particles are located near the axis of the gasifier. The smaller and lighter particles are entrained by the shear layer, and the char consumption of these particles appears to be faster as compared to the heavier particles located near the axis. The particles with the low char concentration are the blue colored particles away from the axis. The heavier particles continue to be near the axis region at least in the combustion zone. Further downstream, even the heavier particles are entrained by the large-scale eddies and can also be seen away from the axis. In the RANS simulations (Fig. 3d), the heavier particles were not transported away from axis because the unsteady large scale structures were not captured. Fig. 8 show the distribution of the axial velocity and the velocity vectors in a plane passing through the axis at time t and t + Dt. The axial velocity is colored within the range of 0–10 m/s so as to identify the shear layer zone and the regions that exhibit flow reversal. The dark blue colored regions show negative axial velocity in the vicinity of the shear layer between the particle-laden jet and the steam jet. The shear layer breaks up into large-scale structures in the lower half of the gasifier, where the low temperature gasification zone exists. This is evident from the velocity vectors depicted in the Zone 2 in Fig. 8b and d, where the large-scale unsteady structures survive into the gasification zone. RANS simulations did not capture these unsteady flow structures in Zone-2. Comparing Fig. 4a with Fig. 8a and c it can be seen that while in RANS the shear layer appears to be steady along the length of the gasifier,

670

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

YCO2

YCO

(a)

YH2

(b)

YH2O

(c)

YCH2

(d)

YO2

(e)

(f)

T (K)

Gasification Zone

Gasification Zone

Combustion Zone

Combustion Zone

Fig. 6. Predictions of species mass fraction distribution using RANS simulations in a plane passing through the axis of the cylinder. (a) CO2 (b) CO (c) H2 (d) H2O (e) CH2 (f) O2.

YC(s)

U (m/s)

(a)

(b)

T (K)

(c)

YC(s)

U (m/s)

(d)

Time=t

(e)

[T·]2(K2)

T (K)

(f)

(g)

(h)

Time=t+Δt

Fig. 7. Flow features from LES simulations in a plane passing through the axis of the cylinder at two different instant of times with Dt = 0.25 s. (a) Temperature distribution at time t, (b) velocity magnitude at time t, (c) coal particle distribution colored with concentration of mass fraction of char at time t, (d) temperature distribution at time t + Dt, (e) velocity magnitude at time t + Dt, (f) coal particle distribution colored with concentration of mass fraction of Char at time t + Dt, (g) mean temperature distribution and (h) temperature variation distribution.

in LES the shear layer breaks down into large-scale structures. The existence of these large-scale unsteady structures is also evident in the azimuthal component of the vorticity in the central plane as shown in Fig. 9. The vorticity distribution at times t and t + Dt

shows the existence of large-scale structures all along the length of the gasifier. While small scale structures are formed near the injector, the large-scale structures are observed more clearly in Zone-2 downstream of the injector. Comparing the vorticity

671

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Jet Shear layer

Jet Shear layer

Ux (m/s)

Ux (m/s)

2

1

(a)

(b)

1

(c)

2

(d)

Fig. 8. Shear Layer features using LES simulations at times t and t + Dt in a plane passing through the axis of the cylinder. (a) Velocity field scaled from 0 to 10 m/s at time t. Blue colored region exhibit reverse flow, (b) velocity vector in two zones (1 for combustion zone and 2 for gasification) colored with Velocity in the range of 0–10 m/s at time t, (c) velocity field scaled from 0 to 10 m/s at time t. Blue colored region exhibit reverse flow and (d) velocity vector in 2 zones (1 for combustion zone and 2 for gasification) colored with Velocity in the range of 0 to 10 m/s at time t + Dt.

(a)

(b)

Fig. 9. Predictions of average values using LES simulations in a plane passing through the axis of the cylinder (a) Azimuthal component of vorticity at time t and (b) azimuthal component of vorticity at time t + Dt.

distributions of RANS and LES, the vorticity is more uniformly distributed in RANS, while LES predicts the formation of unsteady structures due to the roll up of the shear layer. The impact of the large-scale structures on the species distribution can be seen in Fig. 10a–f and Fig. 11a–f at times t and t + Dt. These figures show the ‘‘resolved scale’’ species mass fraction. CO and H2 concentration are higher in LES as compared to RANS predictions. This is because of the enhanced mixing in LES in zone-2 associated with the large-scale eddies. In RANS simulations, the particles located around the axis had high concentration of char. The char from the heavier particles was consumed at a slower rate due to the low concentration of CO2 and H2O in the near-axis region. In LES, particle mixing is better in zone-2 and CO2 and H2O available in the region away from the axis also gets consumed in the char gasification reactions. This is can be observed from Figs. 10 and 11, where there are pockets of low CO2 and H2O away from the axis of the gasifier in the zone-2. These pockets have low concentration of the gasifying agents because more particles are subjected to char gasification reactions in these locations. Similarly CO2 and H2O that are available in high concentration away from the axis also get transported to the near axis region by the unsteady large flow structures. This also increases char gasification for the particles located in the near-axis region as compared with RANS. Fig. 12a–f shows the mean values of the species concentration in a central plane passing through the axis for LES. The upper half of the gasifier is the high temperature combustion zone that is marked by the presence of hydrocarbons and oxygen. The length of the combustion zone is determined by the extent of the presences of these two species. As compared to RANS, the length of the combustion zone is longer. The temperature distribution in this zone will be discussed in detail later in this section. The distribution of CO and H2 in the combustion zone is concentrated in a cylindrical jet region that is formed by the jet. The combustion of volatiles and the gasification reactions are the main cause of high concentration of CO and H2 near the axis in this zone. The lower portion of the gasifier is the gasification zone and is marked by

672

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Y CO2

Y CO

(a)

Y H2

(b)

Y H2O

(c)

Y CH2

Y O2

(e)

(d)

(f)

Fig. 10. Predictions of species mass fraction distribution using LES simulations in a plane passing through the axis of the cylinder at time t. (a) CO2 (b) CO (c) H2 (d) H2O (e) CH2 (f) O2.

YCO2

YCO

(a)

YH2

(b)

YH2O

(c)

YCH2

(d)

YO2

(e)

(f)

Fig. 11. Predictions of species mass fraction distribution using LES simulations in a plane passing through the axis of the cylinder at time t + Dt. (a) CO2 (b) CO (c) H2 (d) H2O (e) CH2 (f) O2.

the region that does not have oxygen. The lower-half of the gasifier makes up this zone. The species distribution is more uniform in the radial direction in this zone, as a result of the improved two-phase

mixing caused by the jet-shear layer structures. The unsteady flow structures have significant effect in improving the gasification reaction rates in this zone and hence the concentration of species.

673

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

YCO

YCO2

(a)

(b)

YCH2

YH2O

YH2

(c)

(d)

YO2

(e)

(f)

Fig. 12. Predictions of average values of species mass fraction using LES simulations in a plane passing through the axis of the cylinder (a) CO2 (b) CO (c) H2 (d) H2O (e) CH2 (f) O2.

0.35 LES

0.2

RANS

0.15 0.1 0.05

O2 mass fraction

CH2 mass fraction

0.25

0.3 0.25

LES RANS

0.2 0.15 0.1 0.05 0

0

Distance away from nozzle (m)

Distance away from nozzle (m)

Fig. 13. Comparisons of axial distribution of hydrocarbon and oxygen mass fraction using RANS and LES simulations.

A more detailed discussion of mixing and the particle char consumption in this zone will be discussed in later in this section. 5.3. Comparison of RANS and LES 5.3.1. Mixing in the combustion and gasification zones The results from the simulations show that the mechanisms of mixing in the two zones are different, being small-scale dependent in the early part combustion zone, and transitioning soon to being large-scale dependent in the later part of that zone as well as in the gasification zone. Whether the turbulence model can capture these mechanisms accurately determines the quality of the predictions of the mixing-dependent reaction rates. Compared to LES, RANS over-predicts mixing in the combustion zone and hence shows a shorter penetration of the gaseous volatile jet which is formed as a result of volatile matter release from the coal particles. This is consistent with previous work on transient gaseous jets [29] and diesel sprays [30]. Valentino et al. [29] and Jagus et al. [30] both concluded that RANS under-predicts the gaseous jet and spray tip penetration. This results in shorter combustion zone that impacts the temperature distribution in the combustion zone and will

be discussed next in detail. LES captures the unsteady mixing in the combustion zone and predicts longer penetration of the gaseous jet formed by volatile matter, also consistent with previous findings. Mixing in the gasification zone is mainly dependent upon the larger unsteady flow structures. While RANS fails to capture these large structures in the gasification zone, LES shows improved mixing of the two phases as it is capable of predicting the flow unsteadiness there. This impacts the distribution of species in the gasification zone. This will also be discussed in detail ahead in this section. 5.3.2. Temperature distribution The combustion zone of the gasifier is the region marked by the presence of hydrocarbon and oxygen. This zone is characterized by high temperatures due to combustion, and the peak temperatures are within the mixing zone observed away from the axis. This is typical of jet diffusion flames and, in this case, the volatiles released from the coal particles form the gaseous fuel jet. The consumption rate is dependent on the mixing of the jet with surrounding gases. Fig. 13 shows RANS and LES predictions of

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Temperature (K)

674

Axial distance away from nozzle (m) Fig. 14. Comparisons of predicted axial mean temperature profiles at various radial locations of the gasifier using RANS LES simulations.

the axial distribution of hydrocarbon and oxygen mass fractions, respectively, along the axis. These two species determine the length of the combustion zone. LES predicts a longer combustion zone as compared to RANS predictions. The turbulent combustion model adopted in these simulations is the partially stirred reactor [25] that accounts for unburned pockets in the flame zone. The model compares the mixing time scale and the chemical time scale to determine the effect of mixing on the reaction rates. Since the mixing times determined by RANS and LES are different, it is

expected that the predicted reaction rates to be different in the combustion zone. Fig. 14 shows the temperature axial distribution at three different radial locations, viz., r = 0 mm, r = 10 mm and r = 20 mm away from the axis as predicted by RANS and LES. RANS shows higher peak temperatures in the combustion zone and lower temperatures in the gasification zone, indicating faster axial mixing in both zones (recall that combustion is exothermic and gasification is endothermic). This is consistent with the previous conclusion that LES predicts a longer combustion zone as compared to RANS and hence, the peak temperatures are lower in this region for LES. The resolved scale temperatures show significant differences between LES and RANS predictions. Temperature fluctuation in RANS arises from transient coupling of particle to the surrounding gas phase. Temperature fluctuation in LES arises from both transient coupling of particle with the surrounding gas phase and also due to unsteady structures that affect both mixing, combustion and gasification processes. The kinetic energy is present in different scales using both RANS and LES. The kinetic energy using the two models represent the kinetic energy in the eddies that are resolved by the mesh. The finer scales are modeled. In addition, the two-phase coupling also introduces gas-phase velocity fluctuations at the particle level that also affects the kinetic energy at the small scales. The temperature fluctuations for the two models can be compared through temperature

n= -5/3

n= -5/3

(a)

(b)

n= -1

(c) Fig. 15. Predicted temperature fluctuation spectrum at various axial location along the axis of the gasifier using LES and RANS simulations (solid lines – LES simulations; dash lines-RANS simulations) (a) x = 0.28 m, (b) x = 0.8 m and (c) x = 1.73 m.

675

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

n= -5/3

n= -5/3

(a)

(b)

n= -5/3

(c) Fig. 16. Predicted Kinetic Energy spectrum at various axial location along the axis of the gasifier using LES and RANS simulations (solid lines – LES simulations; dash linesRANS simulations) (a) x = 0.28 m, (b) x = 0.8 m and (c) x = 1.73 m.

spectra at particular location within the gasifier. The Fig. 15 shows the temperature spectra on a log-plot at three locations along the axis of the gasifier, viz. x = 0.28 m, x = 0.8 m and x = 1.73 m, in the near-nozzle region, near the end of combustion zone and near the exit of the gasifier, respectively. From Fig. 15a–c it can be seen that LES shows higher temperature fluctuations than RANS, especially at low frequency. The low frequency variations are caused by the large-scale structures in the shear layer. These structures can cause strong variations in the temperature, etc. The slope of the curve at x = 0.28 and x = 0.8 m is 5/3, and at x = 1.73 is 1. At all three locations, the temperature fluctuation is higher for LES, signifying the effect of unsteady flow structures on combustion and gasification reaction. The relatively smaller temperature fluctuation in RANS arises due to transient interaction of particle-gas phase and the stochastic particle model. In LES, this transient interaction between particles and unsteady gas phase structures causes higher variations in combustion and gasification reactions, and hence higher temperature fluctuations. The kinetic energy spectra are shown in Fig. 16 at the same three locations. The kinetic energy spectra at x = 0.28 show that the low frequency content are similar in both the models. At x = 0.8 m, LES predicts higher kinetic energy. In this region large unsteady structures are resolved in the LES simulations. Towards the exit of the gasifier, LES shows that the kinetic energy at the lower frequency are smaller than predicted by RANS, but the values are higher in the medium frequency range.

5.3.3. Particle distribution and char consumption Figs. 17a and b and 18a and b shows the particle distribution predicted by RANS and LES, respectively. Fig. 17 shows the radial location of the particle plotted against the particle diameter and Fig. 18 shows the radial location of the particle plotted against the particle char concentration. Fig. 17a shows that more particles accumulate near the axis in RANS, as marked by the circled region. This is because the unsteady flow structures captured by LES are able to mix particles more effectively in the radial direction. Fig. 18a and b shows the variation of the particle radial location with respect to the particle char concentration. It can be seen from the circled region that RANS predicts more accumulation of particles with high char concentration near the axis as compared to LES predictions. This is because the particles in the gasification zone mix more efficiently and can find the pockets with higher concentration of the gasifying agents (CO2 and H2O) responsible for char consumption. These trends were also observed while discussinf the species distribution in the gasification zone predicted by. A more quantitative comparison of the particle char consumption predicted by RANS and LES can be seen in Fig. 19, showing the percentage of particles that have not been converted fully in the gasifer. For ease of interpretation, the gasifier has been divided into four regions in the cylindrical coordinate system and given designation A, B, C and D. These four regions are the two cylindrical region (region-A and region-B) bound by radius 5 cm and length varying between 0–1 m and 1–1.8 m, and the two cylindrical

676

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Distance from axis (m)

Particle diameter distribution inside gasifier (RANS) Wall

Particle density is high near the axis

Particle diameter (m)

(a) Distance from axis (m)

Particle diameter distribution inside gasifier (LES) 0.12 0.1

Wall

0.08 0.06

Particle density is relatively low near the axis

0.04 0.02

Axis

0

Particle diameter (m)

(b) Fig. 17. Comparisons of predicted particle distribution in terms of radial position and particle diameter using LES and RANS simulations.

Distance from axis (m)

Char distribution inside gasifier 0.12 0.1

Wall

0.08 0.06

RANS predicts particles with high char concentration are located close to the axis of the gasifier

0.04 0.02

Axis

0 0

0.2

0.4

0.6

0.8

1

Particle char mass fraction

(a) Distance from axis (m)

Char distribution inside gasifier Wall LES predicts relatively less particle accumulation close to the axis that have high char concentration. Axis

Particle char mass fraction

(b) Fig. 18. Comparisons of predicted particle distribution in terms of radial position and particle char concentration using LES and RANS simulations.

annular regions (region-C and region-D) surrounding region-A and B, respectively. LES predicts fewer particles with remaining char in Region-A, C and D, respectively. Region-B shows similar number of particle with remaining char in RANS and LES. The improved char

conversion predicted by LES can be attributed to the improved mixing that causes more particles to react with the gasifying agents. The measured char conversion efficiency for Utah bituminous coal under these conditions was reported as 82%. The

677

Particles not fully converted (%)

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Fig. 19. Prediction of particle not converted completely in the four regions of the gasifier. Region-A&B are in the combustion zone and Region-C&D are in the gasification zone.

predicted char conversion from LES simulations is 80.2% and from RANS simulations is 74.3%.

6. Conclusions Multi-phase reactive flows typically encountered in an entrained flow gasifier have been investigated using RANS and LES. Results show significant differences in terms of mixing, combustion, gasification and char conversion. Consistent with previous research on jets and sprays, we observe that in the context of coal-oxygen jet, the near-nozzle mixing is over-predicted by RANS and results in shorter combustion zone. LES on the other hand improves the prediction of the penetration of the volatile gaseous jet and shows a longer combustion zone. RANS predictions of the axial distribution of species along the gasifier axis compare well with the measurement except for carbon monoxide. However RANS compare poorly with the measurements for the radial distribution of species at several axial location of the gasifier. On the other

CO2 (mole fraction)

H 2 O (mole fraction)

5.3.4. Axial and radial species distribution Fig. 20 shows the predicted concentrations of species using RANS and LES compared with measurements. The species concentrations from the simulations are shown at a radial distance 10 mm away from the axis. The measurements are reported at the axis of the gasifier. The predictions at the axis of the gasifier (r = 0 mm) do not compare well and is evident from the flow structures near the nozzle. We note that the measurements were performed using probes located at various axial distances away from the gasifier axis. The presence of probes may have affected the flow. Hence, the measurements are compared to predictions at 10 mm away from the nozzle. The figures show that the LES predictions of CO and CO2 concentration agree better with the measurements in the gasification zone. As mentioned before, this is because of the more accurate predictions of the two-phase mixing in this zone. The concentration of H2O and H2 is similar for both RANS and LES and compares well with the measurement along the axis of the gasifier.

Fig. 21–24 show comparisons of the predicted radial distribution of species and the measurements at four axial location, viz., x = 0.28 m, x = 0.8 m, x = 1.21 m, x = 1.73 m. The first two axial locations are in the combustion zone, the last two locations are in the gasification zone. Fig. 21 shows that the radial prediction of species compares well with the measurements, except for the H2O mass fraction. It should be noted that this location is near the nozzle and could have been subjected to high flow variations affecting the measured values. Figs. 22 and 23 shows that the LES agree better with the measurements than the RANS predictions. RANS on the other hand shows significant gradients in the radial direction that implies that the two-phase flow is not well mixed at these locations. Fig. 24 shows that RANS continues to predict gradient in the species distribution as opposed to the measurements. Compared to RANS, LES compares very well with the measurements of species distribution both trend wise and quantitative comparison. The differences in prediction with measurements at the exit location could be due to interference with the exit boundary conditions as well. Similar to the measurements, LES predicts more uniform distribution of species towards the exit of the gasifier, while RANS shows gradient in the distribution of species.

Distance along the axis (m)

Distance along the axis (m)

(b) H2 (mole fraction)

CO (mole fraction)

(a)

Distance along the axis (m)

(c)

Distance along the axis (m)

(d)

Fig. 20. Species concentration predictions along the axis of the gasifier using RANS and LES simulations for Utah Bituminous coal. (a) CO2 (b) H2O (c) CO (d) H2.

H2O (mole fraction)

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

CO2 (mole fraction)

678

Distance from the axis (m)

(a)

(b) H2 (mole fraction)

CO (mole fraction)

Distance from the axis (m)

Distance from the axis (m)

Distance from the axis (m)

(c)

(d)

CO2 (mole fraction)

H2O (mole fraction)

Fig. 21. Predictions of radial distribution of species concentration at x = 0.28 m inside the gasifier using RANS and LES simulations. (a) CO2 (b) H2O (c) CO (d) H2.

Distance from the axis (m)

Distance from the axis (m)

(b) H2 (mole fraction)

CO (mole fraction)

(a)

Distance from the axis (m)

(c)

Distance from the axis (m)

(d)

Fig. 22. Predictions of radial distribution of species concentration at x = 0.8 m inside the gasifier using RANS and LES simulations. (a) CO2 (b) H2O (c) CO (d) H2.

hand, LES improves the prediction of the axial and radial distribution of species. This is especially significant in the gasification zone, where the measurements show near uniform distribution of species, which RANS fails to capture. LES captures these uniform radial distributions in the gasification zone because of the improved two-phase flow mixing. LES also accurately predicts the exit temperature. LES captures unsteady flow structures of various sizes throughout the gasifier domain. With regards to the two zones, the following additional conclusions can be made.

6.1. The high temperature combustion zone 1. In this zone, volatiles are consumed by the combustion reactions in the presence of oxygen resulting in a high temperature region. Both RANS and LES predict that the near-nozzle region is subjected to very high temperatures. 2. RANS predicts shorter combustion zone with almost half the size predicted by the LES predictions. This is because RANS over-predicts the mixing and releases heat at a faster rate as compared to LES. The turbulent combustion model based on

679

CO2 (mole fraction)

H2 O (mole fraction)

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

Distance from the axis (m)

Distance from the axis (m)

(b) H2 (mole fraction)

CO (mole fraction)

(a)

Distance from the axis (m)

Distance from the axis (m)

(c)

(d)

CO2 (mole fraction)

H2O (mole fraction)

Fig. 23. Predictions of radial distribution of species concentration at x = 1.21 m inside the gasifier using RANS and LES simulations. (a) CO2 (b) H2O (c) CO (d) H2.

Distance from the axis (m)

Distance from the axis (m)

(b) H2 (mole fraction)

CO (mole fraction)

(a)

Distance from the axis (m)

(c)

Distance from the axis (m)

(d)

Fig. 24. Predictions of radial distribution of species concentration at x = 1.73 m inside the gasifier using RANS and LES simulations. (a) CO2 (b) H2O (c) CO (d) H2.

partially stirred reactor approach depends on the mixing time scales predicted by turbulence. The impact of RANS on the turbulent combustion model is to raise the reaction rates because it over-predicted mixing. 3. The low frequency temperature fluctuation is higher in LES as compared to RANS model signifying that the larger flow structures affect the reaction kinetics and thereby temperature.

6.2. The low temperature gasification zone 1. In this zone only char gasification reactions occur and the degree of char consumption depends upon the two-phase mixing of particles with the surrounding gasification agents of steam and carbon-dioxide. This zone also shows lower temperatures because gasification reaction are endothermic.

680

N. Abani, A.F. Ghoniem / Fuel 104 (2013) 664–680

2. The two-phase flow continues to feature a jet shear layer way into the gasification zone. RANS models predict the existence of a steady shear layer all along the length of the gasifier. LES model shows the breakdown of the jet shear layer in the gasification zone. LES captures the unsteady larger scale flow structures that mix the particles with the surrounding gasifying agents. The result is that LES creates a more uniform two-phase flow mixture in this zone as compared to RANS. This also results in uniform radial distribution of species in this zone as predicted by LES and observed in measurements as well. 3. The particle distribution in the gasification zone shows that RANS simulations predict more accumulation of particles near the axis as compared with LES. The particles near the axis in RANS have high char concentration. The particles near the axis in LES have relatively lower char concentration because of the availability of carbon monoxide and steam that result in more char consumption. 4. LES models predict char conversion efficiency of 80.2% that compares well with the measurement of 82%. RANS model predicts char conversion efficiency of 74.3%. The flow inside an entrained flow gasifier is inherently unsteady and LES models are able to capture the unsteady flow structures. The unsteady flow structures affect mixing and particle dispersion and hence the char conversion efficiency of the gasification process. LES model accurately predicts the char conversion efficiency and species distribution throughout the gasifier. Acknowledgments Authors would like to acknowledge BP research to provide financial support for this research. Authors would also like to acknowledge valuable discussion regarding frequency plots with Dr. Santosh Shanbhogue of department of mechanical engineering at Massachusetts Institute of Technology. References [1] Hoffmann J, Tennant J, Stiegel GJ. Comparison of pratt and whitney rocketdyne IGCC and commercial IGCC performance. Technical report, DOE/NETL, technical report no. 071200021; 2010. [2] Brown BW, Smoot LD, Smith PJ, Hedman PO. Measurement and prediction of entrained-flow gasification processes. AIChE J 1988;34:435–46. [3] Hill SC, Smoot LD. A comprehensive three-dimensional model for simulation of combustion systems. PCGC-3. Energy Fuel 1993;7:874–83. [4] Watanabe H, Otaka M. Numerical simulation of coal gasification in entrained flow coal gasifier. Fuel 2006;85:1935–43. [5] Chen C, Horio M, Kojima T. Numerical simulation of entrained flow coal gasifiers. Part I: Modeling of coal gasification in an entrained flow gasifier. Chem Eng Sci 2000;55:3861–74. [6] Chen C, Horio M, Kojima T. Use of numerical modeling in the design and scaleup of entrained flow coal gasifiers. Fuel 2001;80:1513–23. [7] Zimmermann S, Taghipur F. CFD modeling of the hydrodynamics and reaction kinetics of FCC fluidized-bed reactors. Ind Eng Chem Res 2005;44:9818–27.

[8] Wu Y, Zhang J, Smith PJ, Zhang H, Reid C, Lv J, et al. Three-dimensional simulation for an entrained flow coal slurry gasifier. Energy Fuels 2010;24:1156–63. [9] Monaghan RFD, Ghoniem AF. A dynamic reduced order model for simulating entrained flow gasifiers. Part I: Model development and description. Fuel 2012;91(1):61–80. [10] Kumar M, Ghoniem AF. Multiphysics and multiscale simulations of entrained flow gasification. Part II: Constructing and validating the overall model. Energy Fuel 2012;26(1):464–479. [11] Kobayahsi H, Howard JB, Sarofim AF. Coal devolatilization at high temperatures. Proc Int Sympos Combust 1977;16(1):411–25. [12] Ubhayaker SK, Stickler DB, von Rosenberg CW, Gannon RE. Rapid devolatilization of pulverized coal in hot combustion gas. In: Proc 16th int symp on combustion, combustion inst. Pittsburgh; 1977. p. 427–36. [13] Goetz GJ, Nsakala NY, Patel RI, Lao TC. Combustion and gasification characteristics of chars from four commercially significant coals of different rank. Combustion engineering, lnc. Windsor, CT, USA; 1982. [14] Kajitani S, Hara S, Matsuda H. Gasfication rate analysis of coal char with a pressurised drop tube furnace. Fuel 2002;81:539–46. [15] Singer SL, Ghoniem AF. An adaptive random pore model for multimodal pore structure evolution with application to char gasification. Energy Fuel 2011;25(4):1423–37. [16] Badzioch S, Hawksley PGW. Kinetics of thermal decomposition of pulverized coal particles. Ind Eng Chem Process Des Develop 1970;9(4):521–30. [17] Liu G, Niksa S. Coal conversion submodels for design applications at elevated pressures. Part II. Char gasification. Prog Energy Combust Sci 2004;30:679–717. [18] Hong J, Hecker WC, Fletcher TH. Modeling high-pressure char oxidation using langmuir kinetics with an effectiveness factor. Proc Combust Inst 2000;28:2215–23. [19] Zhang M, Yu J, Xu X. A new flame sheet model to reflect the influence of the oxidation of CO on the combustion of a carbon particle. Combust Flame 2005;143:150–8. [20] Menon S, Yeung PK, Kim WW. Effect of subgrid models on the computed interscale energy transfer in isotropic turbulence’’. Comput Fluids 1996;25(2):165–80. [21] Yoshizawa A, Horiuti K. A statistically-derived subgrid-scale kinetic energy model for the large-eddy simulation of turbulent flows. J Phys Soc Jpn 1985;54(8):2834–9. [22] Ghosal S, Lund TS, Moin P, Akselvoll K. A dynamic localization model for largeeddy simulation of turbulent flows. J Fluid Mech 1995;286:229–55. [23] Jones WP, Lindstedt RP. Global reaction schemes for hydrocarbon combustion. Combust Flame 1988;73(3):233–49. [24] Anderson J, Rasmussen CL, Giselsson T, Glarborg P. Global combustion mechanisms for use in CFD modeling under oxy–fuel conditions. Energy Fuel 2009;23:1379–89. [25] Williams FA. Combustion theory. 2nd ed. New York: Benjamin/Cummings Publishing Co., Inc.; 1985. p. 680. [26] Kärrholm FP. Numerical modelling of diesel spray injection, turbulence interaction and combustion, PhD thesis. Chalmers University Technology, Sweden; 2008. [27] Gosman AD, Ioannides E. Aspects of computer simulation of liquid–fueled combustors. J Energy 1983;7(6):482–90. [28] Sazhin SS, Sazhina EM, Faltsi-Saravelou O, Wild P. The P1 model for thermal radiation transfer: advantages and limitations. J. Fuel 1996;75(3):289–94. [29] Valentino MF, Jiang X, Zhao H. A comparative RANS/LES study of transient gas jets and sprays under diesel conditions. Atomization Spray 2007;17(5): 451–72. [30] Jagus K, Jiang X, Dober G, Greeves G, Milanovic N, Zhao H. Assessment of largeeddy simulation feasibility in modelling the unsteady diesel fuel injection and mixing in a highspeed direct-injection engine. In: Proc inst. mechanical engineers., Part-D: Jl. Automobile, Engineering; August, 2009. vol. 223. p 1033–48. doi:http://dx.doi.org/10.1243/09544070JAUTO1052.