This action might not be possible to undo. Are you sure you want to continue?

**M**

**ESO-SCALE STRUCTURE ⎯ A CHALLENGE OF COMPUTATIONAL**

**FLUID DYNAMICS FOR CIRCULATING FLUIDIZED BED RISERS**

Wei Ge*, Wei Wang*, Weigang Dong, Junwu Wang, Bona Lu, Qingang Xiong and Jinghai Li*

**State Key Laboratory of Multi-Phase Complex Systems,**

**Institute of Process Engineering, Chinese Academy of Sciences**

**P. O. Box 353, Beijing 100080, China**

*** Email: jhli@home.ipe.ac.cn**

**Abstract - Multi-scale heterogeneous structures are characteristic of the concurrent-up gas-solid**

**flow in circulating fluidized bed (CFB) risers, which stand for a grand challenge to accurate**

**simulation**

**of**

**industrial-scale**

**CFB**

**risers**

**with**

**reasonable**

**computational**

**cost.**

**Coarse-grid**

**simulations**

**using**

**traditional**

**two-fluid**

**models**

**(TFM),**

**although**

**efficient,**

**are**

**not**

**accurate**

**enough**

**for**

**the**

**purpose**

**of**

**design**

**and**

**control**

**of**

**industrial**

**units,**

**while**

**direct**

**numerical**

**simulations (DNS) starting from below the Kolmogorov scale and particle scale are currently**

**unrealistic for this purpose. Coarse-grid simulations with proper sub-grid scale (SGS) models**

**are expected to provide both reasonable accuracy and efficiency, but the closure of SGS models**

**has**

**been**

**long-lasting**

**controversial**

**issue.**

**In**

**this**

**presentation,**

**we**

**review**

**our**

**attempts**

**in**

**providing such a closure with specific stability condition for flows in CFB risers on the grid**

**scale,**

**that**

**is,**

**the**

**so-called**

**analytical**

**multi-scale**

**approach**

**and**

**its**

**later**

**improvements**

**and**

**extensions.**

**The**

**resultant**

**SGS**

**model**

**has**

**been**

**incorporated**

**into**

**commercial**

**software**

**and**

**applied in the simulation of industrial CFB risers with various sizes and bed materials. Further**

**extension of the model to mass transfer processes shows that superficial Reynolds number is**

**insufficient for correlating mass transfer rate as multi-scale heterogeneity is not reflected in this**

**criterion. The presentation will be concluded by prospects in multi-scale computational fluid**

**dynamics of multiphase systems.**

**1.**

**MULTI-SCALE**

**CHARACTERISTICS**

**OF**

**THE**

**HETEROGENEOUS**

**FLOW**

**STRUCTURES IN CFB RISERS**

**Gas-solid fluidization is typically aggregative (Wilhelm & Kwauk, 1948), and thereby displaying remarkable**

**multi-scale heterogeneity. In general, the heterogeneity can be identified on three scales with spatio-temporal**

**characteristics:**

**On the micro-scale, the difference between gas and solid phases presents the intrinsic heterogeneity in the**

**system, although their interfaces are normally clear-cut, they are by no means simple or regular. The solid**

**particles in natural or industrial processes are usually irregular in shape, for example, most catalyst particles**

**are porous, and complex contacting and non-contacting forces may present between these particles.**

**On the meso-scale, heterogeneity of solid particle distribution in CFB risers can be observed in the form of**

**dynamic**

**dense**

**clusters**

**immersed**

**in**

**dilute**

**“broth”,**

**which**

**deform,**

**interact,**

**split**

**and**

**merge**

**constantly.**

Conventionally, they are also called dense and dilute “phases”, but their interface is not so distinct. In fact,

**big controversy still remains on how to define a cluster, as they are in continuous deformation. Although the**

**maximum**

**gradient**

**of**

**solids**

**concentration**

**has**

**provided**

**reasonable**

**criterion**

**for**

**the**

**phase-interface**

**(Werther et al, 1996), the difficulty remains on how to measure local and transient solids concentration in**

**physical experiments and computer simulations. Anyway, such heterogeneity has definite and also strong**

**impact on the hydrodynamics and transport behaviors, as will be detailed in the next section.**

**On**

**the**

**macro-scale,**

**because**

**of**

**the**

**presence**

**of**

**boundaries**

**and**

**the**

**effect**

**of**

**inlets**

**and**

**outlets,**

**the**

**characteristics of meso-scale structures also vary considerably in space and in time. For CFB risers, we can**

**identify a top “dilute” zone and a bottom “dense” zone axially and a so-called core-annulus structure laterally.**

The local mean solids concentrations in the bottom “dense” zone and in the annulus area are relatively higher,

but dense clusters also present in the top “dilute” zone and in the core area, though with relatively lower

**probability.**

**Besides solids concentration distribution, which is most evident in observation and in measurements, the**

flow velocities in the two phases are also highly heterogeneous. The slip velocity in dense clusters is much

lower than that in the dilute “broth”, and the difference between their gas velocities is even higher as the gas

**flow always bypasses the cluster for less resistance.**

**It is rather difficult, in itself, to describe the heterogeneous flow structure embedded on each scale, and the**

correlation among different scales is the focus of this complexity (Li et al., 2004). For example, net backflow

**of solid particles can be observed in the annulus area due to meso-scale particle clustering and macro-scale**

wall effects, while gas phase turbulence dominates and is also affected by the particle distribution in the core

area, with its energy dissipated by eddies and inter-particle collisions over a wide range of scales. All such

**factors**

**add**

**complexity**

**to**

**the**

**flow**

**structure**

**of**

**CFB**

**risers,**

**and**

**bring**

**about**

**significant**

**errors**

**in**

**both**

**experiments and simulations if they are not properly accounted for, as illustrated in Fig. 1.**

**a) Experiment**

**b) Simulation**

**Fig.**

**1:**

**The**

**effect**

**of**

**heterogeneous**

**structure**

**on**

**the**

**results**

**of**

**experiment**

**and**

**simulation**

**in**

**gas-solid**

**flow:**

**(a)**

**Measuring**

**the**

**same**

**structure**

**with**

**different**

**probes**

**gives**

**different**

**results**

**(Reh**

**&**

**Li,**

**1991);**

**(b)**

**Simulating**

**the**

**same**

**system**

**by**

**different**

**models**

**gives**

**different**

**results**

**(air-FCC**

**system**

**in**

**the**

**bottom**

**section**

**of**

**semi-industrial**

**turbulent**

**fluidized bed, I.D=0.71m, height=16m, U**

g

=0.945m/s, d

p

=7.5×10

-5

m, ρ

p

=1503kg/m

3

).

**2.**

**THE**

**CHALLENGE**

**THEREBY**

**PROPOSED**

**FOR**

**SIMULATIONS**

**OF**

**CFB**

**RISER**

**FLOWS**

**Computer capacity has been increasing dramatically, while the development of computation capability is**

comparatively slow. Then, it is natural to ask: what is the reason for such a difference? Our belief is that the

**meso-scale heterogeneity and the interdependence between scales described above present a grand challenge**

**to the development of computer simulation, that is, how to balance between the two ends of computational**

**accuracy and efficiency? Two approaches standing for high accuracy or high efficiency are usually referred**

to in literature, i.e., the direct numerical simulations (DNS) for accuracy and the coarse grid simulations for

**efficiency. In what follows we will discuss these two approaches with regard to CFB risers.**

**2.1 How expensive is direct numerical simulation?**

**At**

**present,**

**theoretically**

**rigorous**

**simulation**

**of**

**gas-solid**

**flow**

**should**

**go**

**down**

**below**

**the**

**scale**

**of**

**solid**

**particles and that of the dissipative eddies in the turbulent gas phase, which is typically on the orders of**

**microns to millimeters. On such scales, the Navier-Stokes equation for fluid flow can be applied directly, and**

**the**

**particles**

**can**

**be**

**treated**

**individually**

**with**

**their interactions**

**described**

**in**

**classical**

**Newton**

**mechanics.**

**Crossed**

**Probe**

**Parallel**

**Probe**

**Although such a description requires details of particle-particle interactions as inputs, which are not easily**

**obtained in experiments, DNS is generally believed to be able to predict the hydrodynamics of well-defined**

**systems reasonably.**

**However, the computational cost involved in DNS is tremendous. In fact, to our knowledge, the largest DNS**

**of**

**gas-solid**

**suspension**

**to**

**date**

**involves**

**only**

**thousands**

**of**

**particles**

**(Ma**

**et**

**al.,**

**2006),**

**and**

**for**

**computationally**

**less**

**demanding**

**liquid-solid**

**systems,**

**the**

**number**

**is**

**still**

**limited**

**to**

**tens**

**of**

**thousands**

**(Nguyen**

**and**

**Ladd,**

**2005),**

**while**

**for**

**industrial**

**units**

**with**

**dimensions**

**in**

**meters**

**or**

**even**

**ten**

**meters,**

**the**

**number of particles are in the range of 10**

10

~10

15

**and the number of numerical elements needed would reach**

**10**

12

~10

20

.

**In**

**terms**

**of**

**temporal**

**evolution,**

**the**

**time**

**step**

**of**

**the**

**DNS**

**simulations**

**are**

**typically**

**finer**

**than**

**microseconds**

**and**

**the**

**computational**

**time**

**for**

**one**

**step**

**can**

**hardly**

**reach**

**milliseconds**

**even**

**for**

**well**

**parallelized code on state-of-the-art high performance computing (HPC) systems, therefore, an reasonable**

**estimate is “days (computational) for seconds (physical)”. Further development of HPC technology is likely**

**to expand the size of the simulated systems dramatically, but reducing the ratio of computational to physical**

**time will be much more difficult because of the strong barriers in the frequency of chips. In**

**short,**

**we**

**may**

**say**

**that**

**conventional**

**DNS**

**of**

**industrial**

**CFB**

**units**

**is**

**yet**

**impossible**

**or**

**at**

**least**

**not**

**economical**

**in**

**the**

**foreseeable**

**future,**

**unless**

**certain**

**breakthrough**

**specially**

**designed**

**for**

**parallelization**

**can**

**be**

**made**

**in**

**both**

**software and hardware.**

**2.2 How accurate is coarse grid simulations?**

**For reasonable computational cost and timely response, most practical simulations on industrial units have**

**used two-fluid model (TFM) with coarse grids on the scale of several millimeters to centimeters. However,**

**as reported by many researchers (Andrews IV et al., 2005; J iradilok et al., 2006; Lu et al., 2005; Qi et al.,**

**2007; Qi et al., 2000; Sundaresan, 2000; Yang et al., 2003a, 2003b), the accuracy of such simulations is not**

**satisfactory.**

**For**

**example,**

**the**

**solids**

**circulating**

**rates**

**are**

**significantly**

**over-predicted**

**and**

**the**

**meso-scale**

**heterogeneity is considerably smoothed, as discussed in the literature (Yang et al., 2003a).**

**The**

**accuracy**

**of**

**numerical**

**simulation**

**depends**

**on**

**the**

**rationality**

**of**

**the**

**physical**

**model,**

**the**

**numerical**

scheme and the grid size used. Assuming that a TFM is reasonable on the scale its averaging procedures are

taken, its best solutions can be approached when the grid size is much smaller than this scale, since a smooth

and linear change can usually be secured then. When coarser grids are used, the numerical solution will

**gradually**

**deviate**

**from**

**the**

**analytical**

**solution**

**of**

**the**

**model**

**and**

**the**

**physical**

**picture,**

**at**

**rates**

**largely**

**dependent**

**on**

**the**

**nature**

**of**

**the**

**model**

**and**

**the**

**system**

**simulated.**

**When**

**significant**

**sub-grid**

**scale**

**(SGS)**

**heterogeneity displays in the system, the linear and smooth variation of the flow variable assumed in the**

**numerical solution loses completely, so that the discrepancy between coarse-grid simulations and the picture**

**really described by TFM could be very large, as reported by Sundaresan (2000) and shown in Fig. 2.**

**The calculation of drag coefficient in TFM can serve as a good example where correlations from seemingly**

**uniform**

**suspensions**

**or**

**fixed**

**beds**

**(Wen**

**&**

**Yu,**

**1966;**

**Ergun,**

**1952)**

**are**

**routinely**

**used.**

**As**

**previously**

estimated (Ge, 1998; Li, 2000; Yang et al., 2003b), these correlations may give predictions even with wrong

**order of magnitude if typical cluster-broth two-phase structures present under the grid scale. This effect can**

**be**

**demonstrated**

**more**

**precisely**

**by**

**Fig.**

**for**

**suspension**

**of**

**mono-sized**

**2D**

**particles**

**under**

**different**

**configurations, where the slip velocity induced by a given pressure drop with constant solids concentration is**

**obtained from DNS using macro-scale particle methods (Ma et al., 2006; 2007) with Lagrangian descriptions**

**for both particles and fluid. The naturally developed heterogeneous suspension, though not as significant as**

**in real gas-solid systems owing to the limitation of the number of particles, has quite different slip velocities**

**from that for fixed hexagonal (most uniform)**

**suspension and the ratio of the highest to the lowest drag**

**coefficient**

**is**

**about**

**6.**

**The**

**simulation**

**also**

**suggests**

**that**

**the**

**form**

**of**

**the**

**heterogeneity,**

**for**

**example,**

**the**

**orientation**

**of**

**the**

**voidage**

**gradient**

**relative**

**to**

**the**

**fluid**

**flow,**

**have**

**great**

**influence**

**on**

**the**

**resistance**

**characteristics of the suspension. Therefore, mean voidage and slip velocity alone are insufficient to define**

the mean drag coefficient within a grid cell (Xu et al., 2007), and the structure inside the grid cells has to be

**considered.**

**(a) 16×64**

**(b) 32×128**

**(c) 64×256**

**(d) 200×800**

**Fig. 2: Snapshots of solids distribution in a periodic domain simulated by using TFM with different numbers of grids.**

**Domain**

**size=15mm×60mm.**

**Time**

**steps=2.0×10**

-4

s.

**Wen**

**&**

**Yu’s**

**relation**

**and**

**algebraic**

**model**

**are**

**adopted**

**for**

**the**

**calculation**

**of**

**drag**

**coefficient**

**and**

**granular**

**temperature,**

**respectively.**

**Restitution**

**coefficient**

**e**

p

=0.9.

**Fluent®6.2**

**is**

**used**

**for**

**simulation.**

**The**

**Second**

**Order**

**Upwind**

**and**

**Quick**

**schemes**

**are**

**used**

**for**

**discretization**

**of**

**the**

**momentum**

**equations**

**and**

**volume**

**fraction,**

**respectively.**

**The**

**simulation**

**initiates**

**from**

**uniform**

**state**

**with**

**zero**

**velocities**

**for**

**gas**

**and**

**solids**

**except**

**in**

**certain**

**grids**

**with**

**small**

**perturbation.**

**Physical**

**parameters:**

**ρ**

p

=1500

**kg/m**

3

,

**d**

p

=75

**μm,**

**ρ**

g

=1.3

**kg/m**

3

, μ

g

=1.8×10

-5

kg/(m·s). Average solids volume fraction remains 0.05. With increasing number of grids from 16×64

**in Fig. 2a to 200×800 in Fig.2d, the structural details revealed by the numerical solution improve.**

**Fig. 3: Temporal variation of drag coefficient for a suspension of 3264 solid particles in a domain of 7.26mm×29.1mm**

**simulated**

**with**

**macro-scale**

**particle**

**modeling**

**(Ma**

**et**

**al.,**

**2006).**

**Initial**

**velocities**

**of**

**the**

**particles**

**are**

**assigned**

**with**

**random**

**directions**

**but**

**equal**

**magnitude**

**(2.68×10**

-2

m/s).

**Physical**

**parameters:**

**ρ**

p

=1225

**kg/m**

3

,

**d**

p

=140

**μm,**

**ρ**

g

=1.225

**kg/m**

3

, μ

g

=1.8 kg/(m⋅s). The smoothing length is 2.8×10

-5

m, the gas particle diameter is 8.4×10

-6

m. As the voidages and

**average**

**drag**

**forces**

**are**

**constant,**

**but**

**the**

**average**

**slip**

**velocity**

**changes**

**significantly,**

**the**

**drag**

**coefficients**

**are**

**very**

different when different flow structures present. a) Steady flow around an array of fixed particles with equal neighbor

**distance, the averaged axial slip velocity is 0.18 m/s, the corresponding drag coefficient is 23.56; b) Enlargement of a**

**local**

**part**

**of**

**the**

**flow**

**field**

**in**

**“a”**

**showing**

**the**

**simulated**

**details;**

**c)**

**snapshot**

**of**

**the**

**flow**

**structure**

**under**

**free**

**suspension,**

**the**

**averaged**

**axial**

**slip**

**velocity**

**is**

**0.336**

**m/s,**

**the**

**corresponding**

**drag**

**coefficient**

**is**

**5.31.**

**d)**

**Another**

snapshot of the flow structure, the averaged axial slip velocity is 0.13m/s, the corresponding drag coefficient is 31.87.

**The**

**difference**

**in**

**slip**

**velocity**

**and**

**drag**

**coefficient**

**under**

**the**

**same**

**voidage**

**and**

**drag**

**force**

**are**

**caused**

**by**

**the**

**heterogeneous structures in the system.**

**The SGS structure may further perplex the heat/mass transfer and reaction behavior of a CFB reactor. As a**

**reason, the transport and reaction processes involve more information of concentrations and velocities in**

local space, which cover a wider, but not fully resolved, range of scales, so that the same coarse-grid flow

**behavior may lead to different performances of overall reactions due to the variation of structures. Therefore,**

even though the coarse-grid flow field can be well predicted, significant errors will still set in if transfer rates

are calculated on the basis of uniform solid distribution within the grid cells. We will return to this topic in

**Section 4.**

In summary, DNS or coarse grid simulation in its own right can only meet one end of our requirement for

**simulation with accuracy or efficiency. To cope with the ever-increasing demand from industries for better**

**design**

**and**

**control**

**of**

**CFB**

**units,**

**more**

**suitable**

**simulation**

**approaches**

**have**

**to**

**balance**

**between**

**the**

**computational cost of DNS and the inaccuracy of coarse-grid simulations. Fine-grid**

**TFM and coarse-grid**

**simulations with SGS models present two mainstream approaches in this respect. The former focuses more on**

**the reduction of grid sizes while the latter focuses more on the dependence of particle-fluid interaction on**

**structures within grids.**

**3. THE APPROACHES TO MEET THE CHALLENGE**

**3.1 Fine-grid TFM**

**The**

**seemingly**

**most**

**straightforward**

**approach**

**is**

**to**

**use**

**set**

**of**

**high-resolution**

**fine**

**grids**

**for**

**TFM**

**simulations, as is the case of Zhang and VanderHeyden (2001). Likewise, as Agrawal et al. (2001) have**

shown in simulations of periodic flow domains, with decreasing grid size, more and more details of the flow

field appear and the overall slip velocity also changes significantly. For Geldart A particles, a plateau can be

**reached finally, which gives the grid-size independent numerical solution to the two-fluid model. In some**

**cases, for instance as shown in Fig. 4, this approach may give results quantitatively comparable with DNS**

results, though their agreement cannot be counted on too much as Wen & Yu’s drag correction function has

**not been validated for cylindrical particles. Owing to the sensitivity of dynamical flow behavior to initial**

conditions and numerical errors, the detailed flow pattern and transient field variables are, as expected, not

**comparable quantitatively.**

**The**

**upper**

**limit**

**of**

**the**

**fine-grid**

**size**

**is**

**roughly**

**of**

**the**

**same**

**order**

**of**

**mean**

**particle**

**spacing**

**and**

**the**

**corresponding**

**time**

**step**

**reported**

**in**

**the**

**literature**

**is**

**in**

**the**

**range**

**of**

**10~100**

**microseconds,**

**so**

**the**

**computational cost of fine grid simulations can be much lower than that of DNS. In this regard, industrial**

**applications can possibly be achieved under some conditions, provided that algorithms with good parallelism**

**can**

**be**

**implemented**

**on**

**modern**

**HPC**

**systems**

**featuring**

**massive**

**parallel**

**processing.**

**However,**

**the**

**best**

**results**

**of**

**fine-grid**

**simulations**

**can**

**only**

**approach**

**the**

**theoretical**

**solution**

**of**

**TFM,**

**which,**

**in**

**itself,**

**has**

**essential limitations and waits for substantial improvements and extensions.**

**In more details, the momentum equation for the gas phase is an averaging of the Navier-Stokes equation over**

**a spatial and/or temporal domain caved by many solid particles, which inevitably contains the second order**

**velocity correlation term (u´u´), similar to the Reynolds stress term in single-phase turbulence (u is the gas**

velocity and the prime denotes its fluctuations). But the composition of this term is more complicated since

**the**

**presence**

**of**

**solids**

**can**

**either**

**enhance**

**or**

**suppress**

**gas**

**turbulence**

**and**

**even**

**induce**

**covariance**

**of**

**gas**

**velocity in the absence of turbulence, though both in form of (u´u´) (Liu, 1993). In fact, the closure of these**

**terms could be more difficult and controversial than that for single-phase turbulence.**

**On**

**the**

**other**

**hand,**

**the**

**solid**

**phase**

**is**

**treated**

**as**

**continuum**

**of**

**granular**

**gas**

**interpenetrating**

**with**

**the**

**continuum of real gas, that is, the so-called “pseudo-fluid” description following the N-S equation. However,**

**the kinetic theory of this granular “gas”, which provides the constitutive laws to the continuum stress, is**

**more**

**complicated**

**than**

**that**

**of**

**real**

**gas.**

**For**

**instance,**

**the**

**solids**

**dissipate**

**mechanical**

**energy**

**through**

collisions and their slippage with the surrounding gas, while real gas molecules conserve mechanical energy

**during**

**collisions**

**and**

**free**

**flight.**

**Further,**

**gas-solid**

**interactions**

**may**

**form**

**long-range**

**hydrodynamic**

**correlation between solid particles that decays very slowly, so that the molecular chaos assumption employed**

**in**

**the**

**derivation**

**of**

**the**

**classical**

**Boltzmann**

**equation**

**may**

**become**

**invalid.**

**Moreover,**

**recent**

**DNS**

**have**

shown (Ma et al., 2006) that the flow velocities of the solids are of the same magnitude or even lower than

**their**

**fluctuation**

**velocities,**

**implying**

**highly**

**non-equilibrium**

**flow;**

**and**

**their**

**thermal**

**velocities**

**are**

**anisotropic so that a second order tensor is not sufficient to describe its stress properties. These features can**

**result**

**in**

**effective**

**transport**

**properties**

**quite**

**different**

**from**

**existent**

**theoretical**

**predictions**

**with**

**quasi-equilibrium**

**assumptions.**

**The**

**solids**

**can**

**introduce**

**further**

**complexities**

**if**

**wide**

**size**

**distribution,**

**irregular shape (for example, deformable fibers) and cohesion between particles have to be considered.**

**Fig. 4: Time variation of axial slip velocities obtained by different simulation approaches (macro-scale particle methods**

**(MaPM;**

**Ma**

**et**

**al.,**

**2006)**

**and**

**two-fluid**

**model**

**(TFM))**

**with**

**characteristic**

**snapshots**

**of**

**solids**

**distribution.**

**Physical**

**parameters:**

**ρ**

p

=160.49

**kg/m**

3

,

**d**

p

=109

**μm**

**for**

**hollow**

**glass**

**beads,**

**and**

**ρ**

g

=1.225

**kg/m**

3

,

**μ**

g

=1.8×10

-5

**kg/(m⋅s)**

**for**

**air**

**under**

**ambient**

**conditions.**

**(a)**

**domain**

**size**

**is**

**(7.72mm×7.68mm),**

**average**

**solids**

**volume**

**fraction**

**is**

**0.24.**

**MaPM:**

**initially**

**389**

**solid**

**particles**

**are**

**uniformly**

**distributed**

**in**

**the**

**domain**

**with**

**velocities**

**of**

**random**

**directions**

**but**

**equal**

**magnitude**

**(0.0268**

**m/s);**

**the**

**smoothing**

**length**

**is**

**2.181×10**

-5

m;

**the**

**gas**

**particle**

**diameter**

**is**

**6.54×10**

-6

m,

**and**

**the**

solid/gas particle diameter ratio is 33; TFM: Fluent® 6.2 is used for the simulation with double-periodic boundaries,

and the number of grids is 40×40; time step is set to be 0.1ms; the algebraic form of the kinetic theory of granular flow

**is**

**adopted**

**for**

**closure**

**of**

**the**

**granular**

**temperature;**

**the**

**restitution**

**coefficient**

**(e**

p

)

**is**

**set**

**to**

**be**

**0.8;**

**as**

**the**

**MaPM**

simulation is undertaken for 2D disks, the drag coefficient for single cylinders is used in the TFM simulation which is

**a) 7.72×7.68mm**

MaPM

**TFM**

**b) 3.9×61.4mm**

**correlated**

**from**

**Tritton**

**(1959,**

**1977),**

**i.e.,**

**C**

d0

=0.67245+280.8898/(1.0+26.63697Re

0.6565

)

**(Re≤100)**

**and**

**C**

d0

≈1.0

(Re>100); owing to the lack of experimental data for cylinders, the form of Wen & Yu correction for particle clusters is

**employed,**

**i.e.,**

**f(ε**

g

)=ε

g

-2.65

,

**to**

**model**

**the**

**effective**

**drag**

**correction**

**in**

**swarm**

**of**

**cylinders.**

**Second**

**order**

**upwind**

**and**

QUICK schemes are used for discretization over momentum equations and volume fraction, respectively. The simulation

**initiates**

**from**

**uniform**

**suspension**

**with**

**zero**

**velocities**

**for**

**gas**

**and**

**solids,**

**small**

**domain**

**in**

**center**

**having**

**non-zero**

**initial velocities is used for perturbation. (b) domain size is (3.9mm×61.4mm), average solids volume fraction is 0.16.**

**MaPM:**

**1024**

**solid**

**particles;**

**TFM:**

**number**

**of**

**grids=30×120,**

**other**

**simulation**

**settings**

**are**

**the**

**same**

**with**

**those**

**for**

**case**

**(a).**

**In**

**general,**

**the**

**axial**

**slip**

**velocities**

**predicted**

**by**

**these**

**two**

**approaches**

**differ**

**from**

**each**

**other**

**with**

**relative**

**errors**

**of**

**34%**

**and**

**5.6%**

**for**

**cases**

**(a)**

**and**

**(b),**

**respectively,**

**though**

**the**

**details**

**of**

**their**

**dynamical**

**behavior**

**are**

**not**

consistent. This is an indication that fine grid simulations may give acceptable results for engineering calculations but

**further validation is still needed.**

**It should be noted further, as a continuum description, that TFM can be numerically solved with arbitrary**

fine grids. The numerical error will certainly decrease before round-off errors dominate. But the physical

**error of the model remains unchanged and the resolution of the model is still on a spatio-temporal scale large**

**enough**

**to**

**host**

**many**

**particles**

**and**

**slow**

**enough**

**to**

**smooth**

**statistical**

**fluctuation,**

**where**

**the**

**continuum**

**description is valid. In addition, because of the multi-scale nature of the heterogeneity in gas-solid systems**

and the fact that solid particles are much larger than molecules, in many cases, the continuum scale overlaps

**with the scale of meso-scale heterogeneity, which presents a more complicated situation for the formulation**

**of constitutive laws. With all these difficulties, it seems necessary to explore other possibilities of CFB riser**

**simulation when further efforts are continued on fine-grid simulations.**

**3.2 Coarse-grid simulation with sub-element-scale models**

**If a continuum model can be established on a scale with characteristic meso-scale heterogeneity directly,**

**smooth**

**variation**

**of**

**the model**

**variables**

**can**

**be**

**reassumed**

**on a**

**relatively**

**larger**

**scale**

**around**

**which**

**the**

**deterioration**

**of**

**accuracy**

**with**

**increasing**

**grid**

**size**

**is**

**not**

**significant**

**and**

**coarse-grid**

**simulations**

**can**

**be**

justified. In fact, the name “SGS models”, though popularly used, is not very reasonable for such continuum

models, since the grid size for these models should, in principle, smaller than the scale the model based on,

and there is no limit to use finer grids. The scale of the elements described by the model is actually more

**characteristic and intrinsic. Therefore, we have coined the name “sub-element-scale (SES) models” for such**

**models. Anyway, this is a practical approach to improve the accuracy of coarse-grid simulations for practical**

**purpose.**

**Most SES models recognize clusters as a key factor to the closure of TFM. For example, Arastoopour and**

**Gidaspow (1979) found it necessary to adjust the particle diameter for obtaining correct axial distribution of**

**solids. O’brien and Syamlal (1993), Wang and Li (2004), Heynderickx et al. (2004) and Cruz et al (2006)**

**suggested modified drag correlations based on the effects of clusters, while Gu and Chen (1998) and Lu et al**

**(2005) proposed their cluster-based approaches, respectively, to simulate CFB risers.**

**In addition to using the concept of cluster directly, some studies tend to extract more accurate inter-phase**

**drag**

**force**

**and**

**particulate**

**phase**

**stresses**

**for**

**TFM**

**from**

**microscopic**

**simulation**

**methods**

**such**

**as**

**lattice**

Boltzmann method (van der Hoef et al., 2006), particle method (Ma et al., 2006) and DNS (Pan et al., 2002),

**or from the fine-grid TFM methods (Agrawal et al., 2001; Andrews IV et al., 2005; Igci et al., 2006; De**

**Wilde, 2005; Zhang and Vanderheyden, 2002).**

**No matter how the SES model is established, a critical issue is to close the equation set with an expanded**

**variable set. Regressing correlations between these variables, either from experimental work or simulation**

**results, may serve as a practical approach but a physical explanation to these correlations and hence their**

**applicability or generality are not easily found. The so-called analytical multi-scale approach (Li & Kwauk,**

**2003; Li et al., 2004; Ge et al., 2007) reviewed in the following presents an attempt to deal with the closure**

**problem another way around.**

**4. THE**

**ANALYTICAL**

**MULTI-SCALE APPROACH**

**PROVIDING**

**CLOSURE**

**TO**

**THE**

**SES MODELS**

**In analytical multi-scale approach, instead of introducing explicit expressions, variational criteria are applied**

**on the variable set which provides a closure to the model. Different from most variational approaches (ref.**

Finlayson, 1972), where the equations are all known but some of them are then presented equivalently by

**variational**

**criteria,**

**the**

**variational**

**criterion**

**here**

**is**

**an**

**natural**

**expression**

**of**

**the**

**stability**

**condition**

**of**

**dynamical**

**SES**

**structures,**

**or**

**the**

**“purpose”**

**of**

**their**

**collective**

**self-organization,**

**whereas**

**the**

**physical**

**meaning of its equivalent equation(s), if exist(s), is/are not so apparent. According to Li & Kwauk (2003),**

**seeking**

**the**

**dominant**

**mechanisms**

**shaping**

**SGS**

**structures**

**and**

**trying**

**to**

**express**

**them**

**as**

**extremum**

**tendencies**

**provides**

**general**

**clue**

**to**

**establishing**

**the**

**stability**

**condition,**

**and**

**the**

**key**

**notion**

**is**

**the**

**compromise**

**between**

**these**

**dominant**

**mechanisms,**

**which**

**is**

**mathematically**

**expressed**

**as**

**mutually**

**constrained extremum.**

This idea is well presented by, and actually stemmed from the energy minimization multi- scale (EMMS)

model (Li, 1987; Li & Kwauk, 1994) for CFB risers. The original model has based on a rather simplified

**physical picture. The flow was considered one-dimensional, fully developed and free of wall effects. The**

**suspension**

**was**

**assumed**

**to**

**separate**

**distinctly**

**into**

**particle-laden**

**mixture**

**called**

**the**

**dense**

**“phase”**

**(denoted**

**by**

**the**

**subscript**

**“c”)**

**and**

**fluid-dominated**

**mixture**

**called**

**the**

**dilute**

**“phase”**

**(denoted**

**by**

**the**

subscript “l”). The flow was thought to be uniform and steady within each “phase”, which was then uniquely

**specified**

**by**

**particle**

**and**

**fluid**

**velocities**

**(U**

p

**and**

**U**

f

),

**and**

**voidages**

**(ε)**

**in**

**each**

**phase,**

**together**

**with**

**the**

volume fraction of the dense “phase” (f) and its “diameter” (l) since it was assumed to occur as spherical

**clusters.**

Six dynamical equations are introduced in the model. Firstly, the effective weight of all particles in the dilute

**“phase” is balanced by the drag of fluid. Secondly, the effective weight of particles in the dense “phase” is**

partially supported by the fluid flow inside, and the rest is supported by the bypassing dilute “phase” fluid

flow. Thirdly, the pressure drop in the dense “phase” is balanced by that in the dilute “phase” plus that over

**the dilute-to-dense “inter-phase”. Finally, the continuity of the fluid and the solid phases and a semi-empiric**

**correlation for the cluster diameter (l) provide three other equations.**

**To**

**close**

**the**

**model,**

**stability**

**condition,**

**minimization**

**of**

**the**

**mass-specific**

**energy**

**consumption**

**for**

**suspending**

**and**

**transporting**

**particles**

**(N**

st

→min),

**is**

**proposed**

**to**

**express**

**the**

**compromise**

**between**

**the**

**tendency**

**of**

**the**

**fluid**

**to**

**pass**

**through**

**the**

**particle**

**layer**

**with**

**least**

**resistance,**

**that**

**is**

**W**

st

→min,

**and**

**the**

**tendency of the particle to maintain least gravitational potential, that is ε**

**→min (Li, 2000; Ge & Li, 2002),**

**which physically results in the aggregation of both phases to display a multi-scale heterogeneous structure.**

**Such a compromise has been verified in micro-scale simulations using pseudo-particle modeling (Li et al.,**

**2004; Zhang et al., 2005).**

**The EMMS model can be directly applied on the equipment scale if boundary effects are negligible. Maybe,**

**its theoretical rationality and significance has been best demonstrated by its ability to provide a physical**

**mapping of fluidization regimes which are partitioned by different saltations and bifurcations (Li et al., 1998;**

**Ge & Li, 2002). Especially, it has given new insight to the long-lasting controversy about “choking” in CFB**

**risers**

**(Ge**

**&**

**Li,**

**2002),**

**that**

**is,**

**the**

**highly**

**heterogeneous**

**regime**

**of**

**“fast**

**fluidization”**

**and**

**the**

**nearly**

**homogeneous regime of “dilute transport” may have the same Nst**

**under certain pairs of U**

g

**and G**

s

, as shown

**in Fig. 5 by the state with (U**

g

=1.5m/s, G

s

=10.5kg/m

2

s) for an air-FCC system. Thus, the “choking” transition

**actually**

**represents**

**jump**

**at**

**these**

**bi-stable**

**states,**

**and**

**both**

**stable**

**branches**

**satisfy**

**the**

**minimum**

**requirement of N**

st

that is denoted by the green and red points near the two ends of the dash-dot line. The

**relevant calculation is now available on the Internet (http://pevrc.ipe.ac.cn/emms/emmsmodel.php3). Note**

**that**

**this**

**important**

**phenomenon**

**of**

**the**

**nonlinearity**

**and**

**multiplicity**

**in**

**particle-fluid**

**flows**

**was**

**hardly**

**captured by using TFM until recently by using an EMMS-based multiscale**

**CFD approach (Wang et al.,**

**2007c), which will be referred to in section 5.**

**To**

**cope**

**with**

**TFM**

**simulation**

**and**

**to**

**describe**

**the**

**lateral**

**and**

**axial**

**flow**

**distributions**

**in**

**CFB**

**risers,**

**the**

**acceleration-related**

**effects neglected in the original**

**EMMS**

**model**

**have**

**to**

**be**

**considered.**

**The**

**acceleration**

**effects**

**were**

**first**

**considered**

**in**

**the context of the EMMS model**

**by**

**Xu**

**(1996)**

**in**

**his**

**attempt**

**to**

**describe the transitional section**

**in CFB risers. Later on identical**

**accelerations**

**were**

**assumed**

**for**

**both dense and dilute phases in**

**the revised EMMS model by Ge**

**and**

**Li**

**(2002).**

**Based**

**on**

**these**

**studies,**

**Cheng**

**et**

**al.**

**(2005)**

**introduced**

**different**

**accelerations**

**for**

**dense**

**and**

**dilute**

**phase**

**particles**

**respectively,**

**which**

**link**

**the**

**particle**

**velocities**

**on**

**different**

riser heights. All these attempts pave the way to integrate the EMMS model into TFM, and to help realize

**three-dimensional prediction of flow distributions.**

**5. SUB-GRID SCALE MODEL BASED ON THE EMMS MODEL**

**For the simulation of dynamic processes or equipments with complex geometries, the EMMS model has to**

**be incorporated into the framework of TFM. In the first attempt made by Yang et al. (2003a), TFM is used to**

provide local velocities as well as local mean voidage in each grid to the EMMS model, which then returns

**its correction of the inter-phase drag coefficient to TFM. A transient local acceleration was allowed in the**

**model,**

**but**

**no**

**difference**

**between**

**the**

**dense**

**and**

**the**

**dilute**

**“phases”**

**was**

**considered**

**then.**

**To**

**facilitate**

**application,**

**user-defined-function**

**(UDF)**

**interface**

**has**

**been**

**developed**

**to**

**dock**

**with**

**commercial**

**CFD**

**software.**

**Such**

**integration**

**has**

**been**

**demonstrated**

**to**

**improve**

**the**

**accuracy**

**of**

**TFM,**

**in**

**terms**

**of**

**solids**

**concentration, solids circulation rate and the resolution for multi-scale heterogeneities (Yang et al, 2003a,**

**2003b, 2004). Later development along this direction is summarized in the following subsections.**

**5.1 Incorporation of the EMMS model and TFM on sub-grid level**

**Recent exploring the physics of the EMMS model indicated that minimization of Nst**

**is only realized on**

**meso- or macro- scale, while on micro- or local scale, Nst**

**manifests no extremum tendency but fluctuates**

greatly with respect to time and space (Zhang et al., 2005). Such a concept helps raise an extended EMMS

model and its coupling scheme named EMMS/Matrix (Wang & Li, 2007). As shown in Fig. 6, for operating

conditions given, TFM assumes uniform distribution at the grid level and then provides mean voidage and

**velocities**

**in**

**each**

**grid**

**to**

**feed**

**the**

**extended**

**EMMS**

**model.**

**The**

**extended**

**EMMS**

**model**

**defines**

**three**

sub-grid phases, i.e. the dense phase, the dilute phase and the meso-scale interphase, and their generalized

**acceleration terms in relation to gas-solid inertial difference. The time-mean meso-scale structure therein is**

**characterized through minimization of N**

st

**under macro-scale operating conditions while its variation in terms**

**Fig.**

**5:**

**Variation**

**of**

**energy**

**consumption**

**for**

**suspending**

**and**

**transporting**

**particles**

**N**

st

**with**

**voidage**

**in**

**clusters**

**ε**

c

**for**

**different**

**solids**

**flux**

**G**

s

**(U**

g

=1.5m/s;

**Air-FCC:**

**d**

p

=54

**μm,**

**ρ**

p

=930

**kg/m**

3

;

**ε**

max

=0.9997,

**ε**

mf

=0.5).

**Solutions**

**are**

**indicated**

**by**

**the**

**lowest**

**points**

**on**

**the**

**curves.**

**At**

**choking**

**point**

**(G**

s

≈10.5kg/m

2

s), two solutions could coexist (Figure generally adapted from

**Wang**

**et**

**al.,**

**(2007d)**

**and**

**the**

**schematic**

**insets**

**taken**

**from**

**Wang**

**et**

**al.,**

**(2007c)).**

of formation, fluctuation and dissolution depends on dynamical equations. In this manner the EMMS/Matrix

**scheme allows bottom-up integration between the EMMS model and TFM. The EMMS-featured structure**

**parameters such as U**

p

, U

f

, U

pc

, U

fc

, l, ε

c

**etc. as well as inertial terms and drag coefficient can hence be**

determined in each CFD grid, which paves the way for developing a sub-grid multiscale mass transfer model

**as will be discussed in the following section (Dong et al., 2007).**

**By using EMMS/Matrix, for a specific system with given physical properties, the drag-coefficient correction**

for the presence of neighboring particles depends not only on local velocities and local voidage but also on

**global**

**operating**

**conditions**

**of**

**gas**

**and**

**solid**

**flow**

**rates.**

**Therefore,**

**it**

**is**

**hard**

**to**

**correlate**

**this**

**structural**

**dependency exclusively as a function of voidage as ever before. In practice, such a structure-dependent drag**

**coefficient is usually lower than the value of widely used relations, such as of Ergun (1952) or Wen and Yu**

**(1966), and may cover a range of variation by several orders of magnitude.**

**As**

**mentioned**

**in**

**section**

**4,**

**this**

**EMMS-based**

**CFD**

**approach**

**was**

**used**

**to**

**simulate**

**the**

**“choking”**

**phenomenon of circulating fluidized beds (Wang et al., 2007c). As shown in Fig. 7, the bell-shaped area**

marking the choking transition was reproduced, over which the solids flux equals the saturation carrying

capacity and the dense fluidization coexists with the dilute transport. Near the critical point, which is the

**summit of the bell-shaped area, simulation shows that the fluctuation of solids volume fraction reaches a**

**maximum, in a manner analogous to what is found in the critical phenomenon of vapor-liquid equilibrium**

**system. The choking phenomenon demarcates the transition between the pneumatic transport and the dense**

**fluidization, and hence is of great importance to the design and operation of CFB reactors. Accordingly,**

**whether or not a CFD simulation can correctly reproduce this phenomenon could be viewed as a touchstone**

for the model used. However, such a fact had been long ignored by simulation practitioners. The long-lasting

**disputes**

**about**

**the**

**definition**

**of**

**choking**

**could**

**be**

**responsible**

**for**

**this**

**ignorance**

**in**

**part,**

**while**

**the**

**most**

**important reason, in our opinion, should be attributed to the neglect of flow structure in most of CFD models.**

**This extended version of EMMS-based CFD approach has been used to simulate the flow behavior of an**

industrial MIP reactor (Lu et al., 2006; Lu et al., 2007). The MIP process developed by SINOPEC (Xu et al.,

**2001) is distinguished from conventional FCC process by an enlarged reaction zone in the middle of the riser**

**tube, to favor the reaction for producing iso-paraffins. As shown in Fig. 8, under its operating conditions,**

**choking occurs in the second reaction zone, in which the solid concentration increases almost linearly with**

the increase of imposed pressure drop but the solids flux remains virtually unchanged. At the same time, the

other two slim sections operate under the condition of dilute transport and undergo little change of solids

holdup. Once the dense fluidized state occupies the entire space of the enlarged zone, the increase of its solid

concentration slows down although the relevant solid flux curve rises. At the same time, the outlet tube is

**gradually**

**filled**

**with**

**catalyst**

**particles,**

**entering**

**the**

**dense**

**fluidized**

**state,**

**and**

**the**

**primary**

**section**

**still**

operates under the dilute transport. From the analysis above, it is evident that a MIP reactor may operate

**synchronously**

**under**

**multiple**

**flow**

**regimes.**

**Quantifying**

**this**

**complex**

**flow**

**behavior**

**in**

**fact**

**helps**

**the**

**relevant design, and then troubleshoots its operation and finally be expected to optimize the MIP refinery**

**process.**

**Fig. 6: Incorporating the EMMS model into two-fluid models (Adapted from Wang et al., 2007c)**

**5.2 Implicit cluster size correlation for the EMMS model**

**Despite all these extensions and improvements, the cluster diameter correlation remained a major source of**

uncertainty in the EMMS model, it also restricts the incorporation of the model into the TFM framework for

**cases**

**with**

**solids**

**flow**

**in**

**gravitational**

**direction,**

**such**

**as**

**downer.**

**To**

**remove**

**these**

**deficiencies, Wang et al. (2007a)**

**has recently made it implicit by**

**analyzing**

**the**

**fluctuation**

**characteristics**

**of**

**solid**

**concentration in CFB risers with**

**stochastic**

**geometry**

**approach**

**of**

**doubly**

**stochastic**

**Poisson**

**processes.**

**The**

**added**

**mass**

**effect**

**is**

**related**

**to**

**the**

**accelerations**

**in**

**the**

**dilute**

**and**

**dense “phases”, which provides**

**an**

**alternative**

**closure**

**to**

**the**

**extended**

**model.**

**cluster**

**diameter**

**can**

**still**

**be**

**derived**

**from the model variable but it is**

**no longer an independent one. It**

**is**

**interesting**

**to**

**note**

**that,**

**although the calculated value of**

cluster diameter is still sensitive to different criteria partitioning the dense and dilute phases, the actual flow

**structure**

**changes**

**very**

**little,**

**suggesting**

**that**

**the**

**model**

**has**

**reflected**

**the**

**essential**

**characteristics**

**of**

**the**

**meso-scale heterogeneity in the system.**

**Fig.**

**7:**

**“Choking”**

**phenomenon**

**in**

**an**

**air-FCC**

**particle**

**CFB**

**system**

**reproduced**

**by**

**EMMS-based**

**CFD.**

**The**

**abscissa**

**variable**

**ΔP**

imp

**refers**

**to**

**the**

**total**

**imposed**

**pressure drop across the riser. The snapshots of voidage**

**profile**

**refer**

**to**

**the**

**transition**

**from**

**the**

**dilute**

**transport**

**(a), the choking transition with different solids inventory**

**in**

**the**

**riser**

**(b)**

**and**

**(c),**

**to**

**the**

**dense**

**fluidization**

**(d)**

**(Wang et al., 2007c).**

**Fig.**

**8:**

**“Choking”**

**in**

**an**

**industrial**

**MIP**

**reactor**

**is**

**reproduced by the EMMS-based CFD. The snapshots of**

**voidage**

**profile**

**refer**

**to**

**the**

**transition**

**from**

**dilute**

**transport,**

**choking**

**transition**

**with**

**different**

**solids**

**inventory**

**in**

**the**

**riser,**

**to**

**dense**

**fluidization**

**(Lu**

**et**

**al.,**

**2006).**

**The**

**solid**

**concentration**

**is**

**calculated**

**from**

**the**

**simulated pressure drop.**

**Fig. 9: Simulation of S-shaped axial voidage profiles (G**

s

=150kg⋅m

-2

s

-1

) and

**the**

**choking**

**phenomenon**

**in**

**0.09m**

**inner**

**diameter**

**and**

**7.2m**

**height**

**riser:**

**U**

g

=4m/s, d

p

=1.0×10

-4

m, ρ

p

=2650kg/m

3

.

**The approach is now capable of simulating both systems with Geldart A or B particles (Wang et al., 2007a)**

**and the flow structures in dense fluidized beds (Wang et al., 2007b). It is shown that the experimentally**

**found**

**S-shaped**

**axial**

**voidage**

**profiles**

**and**

**the**

**choking**

**phenomenon**

**can**

**be**

**well**

**predicted**

**(Fig.**

**9).**

**The**

**computed one-dimensional slip velocities decrease towards the top of the risers and increase with decreasing**

**cross-sectional**

**averaged**

**voidages.**

**The**

**experimentally**

**found dependence**

**of**

**the**

**root**

**mean**

**square**

**of**

**the**

**solid concentration on its mean value at a given position is also well predicted, as shown in Fig. 10. The**

**dynamical flow pattern featured by significant multi-scale heterogeneity is reasonably reproduced, as show**

**in Fig. 11.**

**6. FURTHER EXTENSIONS OF THE EMMS MODEL TO RELATED PROBLEMS**

**The EMMS-based sub-grid scale model has recently been extended to predict mass transfer and reaction**

**behavior in CFB reactors (Dong et al., 2007a, 2007b). With the two-phase structures calculated from the**

**EMMS/Matrix model (Wang and Li, 2007), we further distinguish the mixture concentration in the dense**

phase from that in the dilute phase at the sub-grid level. In this manner, the conventional governing equations

used for mixture concentrations are replaced by the sub-grid mass conservation equations for the dense and

**the dilute phases respectively. The relevant model named EMMS/Mass was integrated into CFD computation.**

**A testing case accounting for sublimation of naphthalene is presented to demonstrate the characters of the**

**model, showing that the conventionally used Reynolds number is not adequate to correlate the structural**

**effects for mass transfer.**

**It was well known, as shown in Fig. 12(a), that the overall Sherwood number reported in the literature is**

scattered in a wide range when correlated as a function of Re, not only for different researchers but also for

the same source of data. For the case of Subbarao and Gambhir (2002), for example, if their data are drawn

**Fig. 10: The computed standard deviation of solid concentration**

**as a function of local mean solid concentration at various heights**

**in**

**large-scale**

**turbulent**

**fluidized**

**bed**

**(inner**

**diameter**

**0.71m,**

**height**

**16m,**

**U**

g

=0.945m/s,

**d**

p

=7.5×10

-5

m,

**ρ**

p

=1503kg/m

3

)

**and**

**comparison**

**with**

**the**

**experimental**

**data**

**of**

**Issangya**

**(1998)**

**(U**

g

=0.69m/s,

**d**

p

=7.0×10

-5

m,

**ρ**

p

=1600kg/m

3

)

**and**

**Bi**

**and**

**Su**

**(2001)**

**(The**

**operating**

**condition**

**and**

**particle**

**properties**

**not**

**given): The comparison is based on the conclusion that standard**

**deviation of solid concentration is a function of local mean solid**

**concentration only.**

**t=10s**

**t=15s**

**t=20s**

**Fig.**

**11:**

**Snapshots**

**of**

**solid**

**concentration**

**distribution**

**in**

**pilot**

**scale**

**CFB**

**riser**

**as**

**simulated by**

**the**

**EMMS/CFD approach**

**(Wang**

**et**

**al.,**

**2007b),**

**particle**

**density=1420kg/m3,**

**particle**

**diameter=0.18mm, riser diameter=0.3048m, riser**

**height=16.5m,**

**superficial**

**gas**

**velocity=4.28m/s,**

**in**

**conventional**

**way**

**as**

**function**

**of**

**Reynolds**

**number,**

**the**

**corresponding**

**Sherwood**

**number**

**will**

**be**

**scattered in a wide range of sub-area 1 in Fig. 12(a). However, when the Sherwood number is redrawn as a**

**function of both U**

g

**and G**

s

, as shown in Fig. 12(b), the EMMS/Mass predictions (the colorful surface) are in

**good agreement with Subbarao and Gambhir’s experimental data (red dots).**

**Fig.**

**12:**

**Comparison**

**of**

**overall**

**Sherwood**

**number**

**between**

**this**

**work**

**and**

**the**

**literature**

**data.**

**a):**

**conventional**

**Sh-Re**

curve of (1) Subbarao and Gambhir (2002); (2) Kettenring and Manderfield (1950); (3) Resnick and White (1949); (4)

**Venderbosch**

**et**

**al.**

**(1999);**

**(5)**

**Gunn**

**(1978);**

**(6)**

**Van**

**der**

**Ham**

**et**

**al**

**(1991);**

**(7)**

**Dry**

**et**

**al**

**(1987);**

**(8)**

**Dry**

**and**

**White**

**(1992); b): EMMS/Mass predicted surface of Sh as a function of Ug**

**and G**

s

**and its comparison with the experimental**

data for the case of Subbarao and Gambhir (2002). Red solid dots are taken from one group of the data (ρ

p

=2600 kg/m

3

,

**d**

p

=196 μm).

**Fig.**

**13:**

**Snapshots**

**of**

**dimensionless**

**ozone**

**concentration**

**at**

**time**

**of**

**30s**

**and**

**related**

**time-averaged**

**radial**

**profiles**

**at**

**different**

**heights**

**(experiment:**

**Ouyang**

**et**

**al.,**

**(1995);**

**U**

g

=3.8

**m/s,**

**G**

s

=106

**kg/m**

2

s,

**k**

r

=57.21

**m**

3

(O

3

)/m

3

**(catalyst)s).**

**a)**

**EMMS/Matrix**

**for**

**flow**

**and**

**EMMS/Mass**

**for**

**mass**

**transfer;**

**b)**

**only**

**flow**

**structure**

**is**

**considered**

**through**

**a)**

**Conventional**

**correlation**

**by**

**Sh**

**vs.**

**Re**

**showing disparity between different sources**

**of data and poor dependence of Sh on Re**

**b)**

**EMMS/Mass**

**description**

**of**

**Sh**

**showing**

**fair**

**agreement**

**between**

**model**

**and**

**experimental data**

**a)**

**Multi-scale**

**models**

**for**

**both hydrodynamics and**

**mass transfer**

**b)**

**Multi-scale**

**model**

**only**

**for hydrodynamics**

**c)**

**Conventional**

**average**

**models**

**for**

**both**

**hydrodynamics**

**and**

**mass transfer**

EMMS/Matrix drag coefficient, mass transfer model is the conventional; c) conventional CFD model for both flow and

**mass transfer without structural consideration. (Color spectrum is in log scale)**

**These results show that Re**

**is**

**insufficient**

**for**

**correlating**

**Sh**

**in**

**heterogeneous**

**structures**

**of**

**fluidized**

**beds,**

**which**

**is**

**the**

**reason**

**why**

**the**

**conventional**

**mass-transfer**

**experimental**

**results**

**could**

**not**

**be unified**

**in**

**Sh∼Re**

**diagram. Further work is needed in this aspect.**

**Fig. 13 gives further comparison between three simulating results of different approaches and experimental**

**data for an ozone decomposition experiment (Ouyang et al., 1995). While using the same experimental data,**

**three**

**subfigures**

**used**

**different**

**simulating**

**models:**

**(a)**

**featuring**

**EMMS/Mass**

**for**

**mass**

**transfer**

**and**

EMMS/Matrix for flow, (b) featuring EMMS/Matrix for flow but conventional model for mass transfer, and

(c) with conventional models for both flow and mass transfer. It is indicated that the conventional model

**without**

**considering**

**the**

**structural**

**effects**

**gives**

**the**

**worst**

**results,**

**considering**

**the**

**effect**

**of**

**structures**

**on**

hydrodynamics gives better results, and the best fit to experimental results can be obtained only when the

**structural effects on both flow and mass transfer are taken into account.**

**7. CONCLUSIONS**

**Multiscale**

**flow**

**structure**

**inherent**

**in**

**gas-solid**

**risers**

**presents**

**grand**

**challenge**

**to**

**CFD**

**simulation**

**of**

**industrial**

**scale**

**CFB**

**reactors.**

**Coarse-grid**

**simulations**

**using**

**traditional**

**two-fluid**

**models**

**(TFM)**

**are**

**not**

**accurate**

**enough**

**owing**

**to**

**the**

**ignorance**

**of**

**structures**

**at**

**sub-element**

**scale**

**(SES,**

**conventionally**

**called**

**sub-grid-scale or SGS) structures, while direct numerical simulations (DNS) starting from below the lowest**

**dissipative scales are currently impractical at least for industrial applications. Coarse-grid simulations with**

**appropriate SGS models are expected to reconcile the contradiction between accuracy and efficiency, but**

**closure of the model with an expanded variable set presents a critical problem. By analyzing the compromise**

**between**

**the**

**dominant**

**mechanisms**

**shaping**

**the**

**SES**

**structures,**

**the**

**analytical**

**multiscale**

**approach**

**can**

**provide such a physically reasonable closure in the form of variational criterion (or stability condition). This**

**approach is exemplified by the so-called energy-minimization multi-scale (EMMS) model for gas-solid riser**

**flow and has been extended to couple with two-fluid models. Coarse-grid simulations with EMMS-based**

**SES**

**models**

**have**

**been**

**incorporated**

**into**

**commercial**

**software**

**(Fluent**

**and**

**CFX)**

**and**

**applied**

**in**

**the**

**simulation of choking transition which is of great importance to industrial operation. Recent efforts have**

**further extended its application to simulate mass transfer in CFB, demonstrating this approach a promising**

**tool to meet the challenges in CFB simulation.**

**The EMMS approach for quantifying meso-scale structures in CFB risers is expected to develop further in**

**three directions. Firstly, the coupling of an axial model considering the acceleration effect and a radial model**

**considering wall and internal frictions with the global minimization of Nst**

**being the stability condition may**

**lead to a complete description of the flow field and the characteristics of its fluctuation behavior in steady**

**operations of the risers. Such a model will be most useful to industrial applications in that the information**

**provided is most relevant while the computational cost is very reasonable. Secondly, with both accelerations**

and frictions considered on the element scale for the dense and dilute “phases”, and the reaction, mass and

**heat transfer processes coupled with the hydrodynamic process, incorporation of EMMS and traditional CFD**

**methods can give comprehensive information of complex industrial units with adequate accuracy. Note that**

**the**

**Reynolds**

**number**

**is**

**insufficient**

**for**

**correlating**

**mass**

**transfer**

**data**

**since**

**more**

**structural**

**factors**

**are**

**involved**

**when**

**multi-scale**

**heterogeneity**

**presents.**

**And**

**finally,**

**by**

**implementing**

**CFD**

**models**

**(both**

**traditional and EMMS based) in particle methods, which can make full use of massive parallel computation,**

**fine-grid simulation of industrial risers can be achieved eventually. With these developments, the role of**

**computer**

**simulation**

**in**

**scaling-up,**

**optimization**

**and**

**on-line**

**control**

**of**

**CFB**

**units**

**can**

**be**

**significantly**

**promoted.**

**In**

**addition**

**to**

**the**

**efforts**

**in**

**upgrading**

**the**

**physical**

**models**

**for**

**CFD,**

**we**

**should**

**be**

**aware**

**of**

**the**

**incompatibility of physical model, mathematical algorithm and computer architecture is, in general, a major**

**reason**

**for**

**the**

**slow**

**development**

**of**

**computing**

**capability**

**along**

**with**

**the**

**dramatic**

**progress**

**in**

**computer**

**capacity.**

**When**

**different**

**algorithms**

**each**

**describing**

**different**

**phenomena**

**are**

**implemented**

**on**

**the**

**same**

**general-purpose**

**computer**

**architecture,**

**low**

**efficiency**

**is**

**almost**

**inevitable.**

**As**

**multi-scale**

**structure**

**is**

**common**

**nature**

**behind**

**various**

**systems**

**and**

**phenomena,**

**designing**

**computer**

**with**

**multi-scale**

**architecture and running multi-scale models may provide an efficient solution to many different problems,**

**and hence bridges high efficiency with reasonable generality.**

**Finally, we would conclude that a breakthrough in computational fluid dynamics is subject to the progress in**

**understanding spatio-temporal multi-scale structures and the efficient use of computer capacity. Therefore,**

**multi-scale nature of the simulated phenomena is a common challenge!**

**REFERENCES**

**Agrawal, K. Loezos, P. N., Syamlal, M., Sundaresan, S. 2001. The role of meso-scale structures in rapid**

**gas-solid flows. J ournal of Fluid Mechanics 445, 151-185.**

**Andrews**

**A.**

**T., Ⅳ**

**Loezos,**

**P.**

**N.,**

**Sundaresan,**

**S.,**

**2005.**

**Coarse-grid**

**simulation**

**of**

**gas-particle**

**flows**

**in**

**vertical risers. Industrial and Engineering Chemistry Research 44(16), 6022-6037.**

**Arastoopour, H., Gidaspow, D. 1979. Analysis of IGT pneumatic conveying data and fast fluidization using a**

**thermo-hydrodynamic model. Powder Technology 22, 77-87.**

**Bi, H. T., Su, P. C. 2001. Local phase holdups in gas-solids fluidization and transport. AIChE J ournal 47,**

**2025-2031.**

**Cheng, C. L., Gao, S., Zhang, Z. et al., 2002. Energy-minimization multi-scale core-annulus (EMMS/CA)**

**model for gas–solid circulating fluidized beds. J ournal of Chemical Industry and Engineering (China)**

**53, 804–809 (in Chinese).**

**Cheng, C. J ., Ge, W., Li, J . 2005. An extended EMMS model with considerations to dense and dilute phase**

**accelerations. Internal Report. Institute of Process Engineering, Chinese Academy of Science, Beijing..**

**Cruz, E., Steward, F. R., Pugsley, T. 2006. New closure models for CFD modeling of high-density circulating**

**fluidized beds. Powder Technology 169, 115-122.**

**De Wilde, J . 2005. Reformulating and quantifying the generalized added mass in filtered gas-solid flows**

**models. Physics of Fluids 17, 113304.**

Dong, W., Wang, W., Li, J ., 2007a. A multiscale mass transfer model for gas-solid riser flows: Part Ⅰ—

**Sub-grid model and simple tests. Chemical Engineering Science, submitted**

Dong, W., Wang, W., Li, J ., 2007b. A multiscale mass transfer model for gas-solid riser flows: Part Ⅱ—

**Sub-grid simulation of ozone decomposition. Chemical Engineering Science, submitted**

**Dry, R. J ., Christensen, I. N., White, C. C., 1987. Gas--solids contact efficiency in a high-velocity fluidised**

**bed. Powder Technology 52, 243.**

**Dry, R. J ., White, C.C., 1992. Gas--solid contact in a circulating fluidized bed: The effect of particle size.**

**Powder Technology 70, 277.**

**Ergun, S. 1952. Fluid Flow through Packed Columns. Chemical Engineering Progress 48, 89-94.**

**Finlayson, B. A. 1972. The Method of Weighted Residuals and Variational Principles, with Application in**

**Fluid Mechanics, Heat and Mass Transfer, Academic Press, New York.**

**Ge, W., 1998. Multi-Scale Simulation of Fluidization. Ph.D. Thesis, Harbin Institute of Technology, Harbin,**

**China.**

**Ge, W., Li, J . 2002. Physical mapping of fluidization regimes－the EMMS approach, Chemical Engineering**

**Science 57, 3993-4004.**

**Ge, W., Chen, F., Gao, J . et al., 2007. Analytical multi-scale method for multi-phase complex systems in**

**process**

**engineering—Bridging**

**reductionism**

**and**

**holism.**

**Chemical**

**Engineering**

**Science**

**62,**

**3346-3377.**

**Gidaspow,**

**D.**

**1994.**

**Multiphase**

**Flow**

**and**

**Fluidization:**

**Continuum**

**and**

**Kinetic**

**Theory**

**Descriptions,**

**Academic Press, San Diego.**

Gu, W. K., Chen, J . C., 1998. A model for solid concentration in circulating fluidized beds. In: L. S. Fan & T.

**M. Knowlton (Eds.), Fluidization**

**. New York: Engineering Foundation. pp. 501 Ⅸ -508.**

**Gunn, D. J ., 1978. Transfer of heat or mass to particles in fixed and fluidised beds. International J ournal of**

**Heat and Mass Transfer 21, 467-476.**

**Halder,**

**P.**

**K.,**

**Basu,**

**P.,**

**1988.**

**Mass**

**transfer**

**from**

**coarse**

**particle**

**to**

**fast**

**bed**

**of**

**fine**

**solids.**

**AIChE**

**Symposium Series 84(262), 58-67.**

**Heynderickx, G. J ., Das, A. K., De Wilde, J ., Marin, G. B., 2004. Effect of clustering on gas-solid drag in**

**dilute two-phase flow. Industrial and Engineering Chemistry Research 43, 4635-4646.**

**Igci,**

**Y.,**

**Sundaresan,**

**S.,**

**Pannala,**

**S.,**

**et**

**al.**

**2006.**

**Coarse-graining**

**of**

**two-fluid**

**models**

**for**

**fluidized**

**gas-particle suspensions. Presented at Fifth International Conference on CFD in the Process Industries,**

**CSIRO, Melbourne, Australia.**

**Issangya, A. S. 1998. Flow dynamics in high density circulating fluidized beds. Ph.D thesis. The University**

**of British Columbia, Canada.**

**J iradilok V, Gidaspow D, Damronglerd S, et al. 2006. Kinetic theory based CFD simulation of turbulent**

**fluidization of FCC particles in a riser. Chemical Engineering Science 61, 5544-5559.**

**Kettenring,**

**K.**

**N.,**

**Manderfield,**

**E.**

**L.,**

**1950.**

**Heat**

**and**

**mass**

**transfer**

**in**

**fluidized**

**systems. Chemical**

**Engineering Progress 46(3), 139-145.**

**Li, J . 1987. Multi-Scale Modeling and Method of Energy Minimization for Particle-Fluid Two-Phase Flow,**

**Doctor thesis, Institute of Chemical Metallurgy, Chinese Academy of Sciences, Beijing.**

**Li,**

**J .,**

**Kwauk,**

**M.**

**1994.**

**Particle-fluid**

**two-phase**

**flow**

**energy-minimization**

**multi-scale**

**method,**

**Metallurgical Industry Press, Beijing.**

**Li, J ., 2000. Compromise and resolution: exploring the multiscale nature of gas–solid fluidization. Powder**

**Technology 111, 50–59.**

**Li, J ., Kwauk, M. 2003. Exploring Complex systems in chemical engineering - the multi-scale methodology.**

**Chemical Engineering Science 58, 521-535.**

**Li, J ., Zhang, J ., Ge, W., Liu, X. 2004. Multi-scale methodology for complex systems, Chemical Engineering**

**Science 59, 1687-1700.**

**Liu,**

**D.,**

**1993.**

**Fluid**

**dynamics**

**of**

**two-phase**

**system.**

**Higher**

**Education**

**Publishing**

**House,**

**Beijing**

**(in**

**Chinese).**

Lu, B., Wang, W., Wang, J ., Li, J ., 2006. CFD simulation of MIP reactors (Report for SINOPEC research

**project X505028).**

Lu, B., Wang, W., Li, J ., Wang, X., Gao, S., Lu, W., Xu, Y., Long, J ., 2007. Multi-scale CFD simulation of

**gas–solid flow in MIP reactors with a structure-dependent drag model. Chemical Engineering Science**

**62, 5487-5494.**

Lu, H., Sun, Q., He, Y., Sun, Y., Ding, J ., Li, X., 2005. Numerical study of particle cluster flow in risers with

**cluster-based approach. Chemical Engineering Science 60, 6757-6767.**

**Ma, J ., Ge, W.,Wang, X.,Wang, J ., Li, J . 2006. High-resolution simulation of gas-solid suspension using**

**macro-scale particle methods. Chemical Engineering Science 61, 7096-7106.**

**Ma, J ., Ge, W., Wang, J ., et al. 2007. Exploring the clustering mechanism of a gas-solid flow from a scale**

**smaller than a single particle. AIChE J ournal, Submitted**

**Nguyen, N. Q., Ladd, A. J . C. 2005. Sedimentation of hard-sphere suspensions at low Reynolds number.**

**J ournal of Fluid Mechanics 525, 73-104.**

**O’brien,**

**T.**

**J .,**

**Syamlal,**

**M.,**

**1993.**

**Particle**

**cluster**

**effects**

**in**

**the**

**numerical**

**simulation**

**of**

**circulating**

**fluidized bed. In: Avidan, A. A. (Ed.), Preprint Volume for CFB-IV. New York. pp 430-435.**

**Ouyang, S., Li, X.-G., Potter, O. E., 1995. Circulating fluidized bed as a catalytic reactor: Experimental study.**

**AIChE J ournal 41(6), 1534-1542.**

**Pan, T., J oseph, D. D., Bai, R. et al. 2002. Fluidization of 1204 spheres: simulation and experiment. J ournal**

**of Fluid Mechanics 451, 169-191.**

Qi, H., You, C., Boemer, A., Renz, U., 2000. Eulerian simulation of gas–solid two-phase flow in a CFB-riser

**under**

**consideration**

**of**

**cluster**

**effects,**

**in:**

**D.**

**Xu,**

**S.**

**Mori**

**(Eds.),**

**Fluidization**

**2000:**

**Science**

**and**

**Technology, Xi’an Publishing House, Xi’an, pp. 231–237.**

Qi, H., Li, F., Xi, B. and You, C., 2007, Modeling of drag with the Eulerian approach and EMMS theory for

**heterogeneous dense gas-solid two-phase flow. Chemical Engineering Science 62: 1670–1681.**

**Reh, L., Li, J . 1991. Measurement of voidage in fluidized beds by optical probe. In Circulating Fluidized**

**Beds Technology III, P. Basu, M. Horio and M. Hasatani(Eds), Pergamon Press, P105.**

Resnick, W., White, R. R., 1949. Mass transfer in systems of gas and fluidized solids. AIChE J ournal 45(6),

**377-390.**

**Subbarao, D., Gambhir, S., 2002. Gas to particle mass transfer in risers. In: Proceedings of 7th International**

**Circulating Fluidized Beds Conference, Canadian Society for Chemical Engineering, Niagara Falls. pp.**

**97-104.**

**Sundaresan,**

**S.,**

**2000.**

**Modeling**

**the**

**hydrodynamics**

**of**

**multiphase**

**flow**

**reactors:**

**current**

**status**

**and**

**challenges. AIChE J ournal 46(6), 1102-1105.**

**Tritton, D. J ., 1959. Experimental on the flow past a circular cylinder at low Reynolds numbers., J ournal of**

**Fluid Mechanics 6, 547-567.**

**Tritton, D. J ., 1977. Physical Fluid Dynamics. Van Nostrand Reinhold company Ltd., New York.**

**Van**

**der**

**Ham,**

**A.**

**G..**

**J .,**

**Prins,**

**W.,**

**Van**

**Swaaij,**

**W.**

**P.**

**M.,**

**1991.**

**Hydrodynamics**

**and**

**mass**

**transfer**

**in**

**regularly**

**packed**

**circulating**

**fluidized**

**bed.**

**In:**

**Basu,**

**P.,**

**Horio,**

**M.,**

**Hasatani,**

**M.**

**(eds.),**

**Circulating**

**Fluidized Bed Technology**

**Ⅲ, Oxford, UK.Pergamon Press. pp. 605.**

van der Hoef, M. A.,Ye, M., van Sint Annaland, M., Andrews, A. T., Sundaresan, S., Kuipers, J . A. M., 2006.

**Multiscale Modeling of Gas-Fluidized Beds. In: G. B. Marin (Ed.) Advances in Chemical Engineering.**

**New York: Academic Press. Volume 31, pp 65-149.**

Venderbosch, R. H., Prins, W., van Swaaij, W. P. M., 1999. Mass transfer and influence of the local catalyst

**activity on the conversion in a riser reactor. Canadian J ournal of Chemical Engineering 77(2), 262-274.**

**Wang,**

**J .,**

**Ge,**

**W.,**

**Li,**

**J .**

**2007a.**

**Eulerian**

**simulation**

**of**

**heterogeneous**

**gas-solid**

**flows**

**in**

**CFB**

**risers:**

**EMMS-based sub-grid scale model with a revised cluster description. Chemical Engineering Science,**

**in press.**

**Wang, J ., Ge, W., Li, J . 2007b. Flow structures inside a turbulent fluidized bed of FCC particles: Eulerian**

**simulation**

**with**

**an**

**EMMS-based**

**sub-grid**

**scale**

**model.**

**to**

**be**

**submitted**

**to**

**Chemical**

**Engineering**

**Science**

**Wang, L., 2002. Multi-scale mass transfer model and experimental validation for heterogeneous gas-solid**

**two-phase flow. Doctor dissertation. Chinese Academy of Science, Beijing.**

**Wang, W., Li, Y. 2004. Simulation of the Clustering Phenomenon in a Fast Fluidized Bed: The Importance of**

**Drag Correlation. Chinese J ournal of Chemical Engineering 12, 335-341.**

**Wang, W., Li, J ., 2007. Simulation of gas–solid two-phase flow by a multiscale CFD approach—extension of**

**the EMMS model to the sub-grid level. Chemical Engineering Science 62, 208–231.**

**Wang,**

**W.,**

**Lu,**

**B.,**

**Li,**

**J .,**

**2007c.**

**Choking**

**and**

**flow**

**regime**

**transitions:**

**simulation**

**by**

**multi-scale**

**CFD**

**approach. Chemical Engineering Science 62, 814–819.**

Wang, W., Ge, W., Yang, N., Xu, G. 2007d. Response to “Evaluating EMMS model for simulating high solid

**flux risers” by P. R. Naren, A. M. Lali and V. V. Ranade, Chemical Engineering Research and Design,**

**85(A8): 1188—1202. Chemical Engineering Research and Design, 85(A9): 1338—1339.**

**Wen, C. Y., Yu, Y. H., 1966. Mechanics of fluidization. Chemical Engineering Symposium Series 62(62),**

**100-111.**

**Werther, J ., Kruse, M., Hage, B., Rudnick, C., 1996. Modeling the fluid mechanics of a circulating fluidized**

**bed based on a local flow strucure analysis. In In-situ measuring techniques and dynamic modeling for**

**multiphase**

**flow**

**systems,**

**Werther,**

**J .,**

**Markl,**

**H(Eds),**

**Technical**

**University**

**Hamburg-Harburg,**

**pp475-516.**

**Wilhelm,**

**R.**

**H.,**

**Kwauk, M.**

**1948.**

**Fluidization**

**of Solid Particles.**

**Chemical**

**Engineering Progress**

**44(3):**

**201-218.**

**Xu M, Ge W, Li J . 2007. A discrete particle model for particle-fluid flows with considerations of sub-grid**

**structures. Chemical Engineering Science 62, 2302-2308.**

**Xu,**

**G.**

**1996.**

**Hydrodynamic**

**Modeling**

**for**

**Heterogeneous**

**Flow**

**Structure**

**in**

**Circulating**

**Fluidized**

**Beds.**

**Doctor Dissertation. Institute of Process Engineering, Chinese Academy of Science, Beijing.**

**Xu,**

**G.,**

**Li,**

**J .**

**1998.**

**Multi-scale**

**interfacial**

**stresses**

**in**

**heterogeneous**

**particle-fluid**

**systems.**

**Chemical**

**Engineering Science 53(18), 3335-3339.**

**Xu, Y., Zhang, J ., Rong, J ., 2001. A modified FCC process MIP for maximizing iso-paraffins in cracked**

**naphtha. Petroleum Processing and Petrochemicals 32 (8), 1–5 (in Chinese).**

**Yang,**

**N.,**

**Wang,**

**W.,**

**Ge,**

**W.,**

**Li,**

**J .,**

**2003a.**

**Choosing**

**structure-dependent**

**drag**

**coefficient**

**in**

**modeling**

**gas-solid two-phase flow. China Particuology 1(1): 38-41.**

**Yang, N., Wang, W., Ge, W., Li, J . 2003b. CFD simulation of concurrent-up gas-solid flow in circulating**

**fluidized beds with structure-dependent drag coefficient, Chemical Engineering Journal 96, 71-80.**

**Yang, N., Wang, W., Ge, W., Wang, L. Li, J ., 2004. Simulation of heterogeneous structure in a circulating**

**fluidized-bed**

**riser**

**by**

**combining**

**the**

**two-fluid**

**model**

**with**

**the**

**EMMS**

**approach.**

**Industrial**

**&**

**Engineering and Chemistry Research 43: 5548-5561.**

**Zhang,**

**D.**

**Z.,**

**VanderHeyden,**

**W.**

**B.,**

**2001.**

**High-resolution**

**three-dimensional**

**numerical**

**simulation**

**of**

**circulating fluidized bed. Powder Technology 116, 133-141.**

**Zhang,**

**D.**

**Z.,**

**VanderHeyden,**

**W.**

**B.,**

**2002.**

**The**

**effects**

**of**

**mesoscale**

**structures**

**on**

**the**

**macroscopic**

**momentum equations for two-phase flows. International J ournal of Multiphase Flow 28(4), 805-822.**

**Zhang J , Wei G, Li J . 2005. Simulation of heterogeneous structures and analysis of energy consumption in**

**particle-fluid systems with pseudo-particle modeling. Chemical Engineering Science 60, 3091-3099.**