J. Iqbal, S. Gao

(1)

( )()

uuu p

t

ρρτ

∂+∇⋅= −∇+∇⋅

∂

(2)

where

( )

( )

2

3

T

u uIu

τη η

= −∇+∇+∇⋅

The burner geometry generation, meshing and calculation of the conservation of mass and momentum equa-

tions are carried out by using the commercial software package ICEMCFD and ANSYS Fluent.

Large eddy and RANS k-model simulations have been performed in this project to investigate the turbulence

effects on the flow [2] [4]. It is well known that LES is a better tool for flows such as jet shear and wall jet im-

pingement [7]. The LES scheme solves the filtered Navier-Stokes equation and can produce more useful infor-

mation, e.g. instantaneous flow fluctuations, than the RANS viscous models. The leading flow parameters for

the three-way assembly are given in Table 1. For the two-way assembly geometry, there is no tertiary entry, and

the secondary is swirling flow.

The 2nd order upwind scheme for flow and QUICK scheme for turbulence are selected with SIMPLE algo-

rithm for the simulations. The turbulence simulation results are compared with each other and also with the

simulation results from Apte and Mahesh [8].

3. Results and Discussion

The geometry from the reference [8] consists of two jets from primary and secondary annular pipe discharging

into a huge cylinder. The effect of the jet velocity ratio shows the complexity of the isothermal models [4], while

the new geometry has an additional annular pipe. It can be seen in Figure 2 that the flow in the new three-way

geometry has better mixing than the two-way assembly from the reference of Apte and Mahesh [8]. The con-

tours in Figure 2 show that the assembly with additional entry is more helpful for better pre-combustion mixing.

The additional air produces additional turbulence and wake effects, and this additional air is also helpful to re-

burn the solid particles like coal/biomass in the burner [1].

Figure 3 and Figure 4 show the mean axial velocity and RMS of mean axial velocity variation with respect to

the characteristic length for different assemblies; the swirling motion governs the flow to increase the rapid

spreading and the reattachment of shear layers [8]. It can be seen that the axial velocity decreases with the char-

acteristic length. The three-way assembly and two-way assembly have been compared with the results from the

simulations of Apte and Mahesh [8] and the experimental work of Somerfield and Qiu [9]. In Figure 3 the mean

axial velocity for the three-way assembly agrees well with the reference data. Comparison of radial and swirl

velocity has also been made with the reference data [8] [9]. The swirl velocity decreases with the characteristic

length as shown in the graphs. The radial velocity graphs show the different flow property than the swirling and

axial velocity; here the radial velocity has the increased value at X/R = 0.8 than the initial point X/R = 0.1,

slightly increase at X/R = 1.6 and starts to decrease and becomes zero at the end point. The increasing and de-

creeing of the radial velocity produce the desired mixing in the combustor. The pressure is also a property of the

moving fluid depending on the axial velocity, which decreases and increases with the axial velocity. The wall

shear problems can also better be solved by using the three-way.

It is the pressure variation created in the fluid due to the kinetic energy fluctuations that produces the wake

and mixing effects, which contribute positively to the mixing phenomena inside the furnace. Clearly, the im-

proved three-way burner model has better mixing efficiency than the simple co-fired burner, mainly due to the

addition of the new annular entry.

Table 1. Flow parameters for the three-way assembly.

Primary (m/sec) Secondary (m/sec) Tertiary (m/sec) Temperat ure Pressure (Pa)

15 10 18 Not considered Atmospheric