Professional Documents
Culture Documents
School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, People's Republic of China
Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin University, Tianjin 300072, People's Republic of China
c
National Engineering Research Centre of Distillation Technology, Tianjin University, Tianjin 300072, People's Republic of China
b
H I G H L I G H T S
art ic l e i nf o
a b s t r a c t
Article history:
Received 30 April 2014
Received in revised form
15 October 2014
Accepted 1 November 2014
Available online 10 November 2014
In this work, we aimed to predict the ue gas side performance of a rening vacuum furnace with oor
gas burners. The computational uid dynamics (CFD) approach was employed to simulate the ow,
combustion, heat transfer and NO emission. Detailed insights into the ue gas velocity, temperature eld
and NO mass concentration distribution were obtained with the aid of velocity vectors and contour
snapshots. The standard k- model was applied to turbulence simulation. The non-premixed turbulent
ames and NO emission were predicted using the Laminar Flamelet model. The discrete transfer model
(DTM) was applied to the radiative heat transfer simulation. Comparative simulation cases were carried
out to investigate the effect of excess air amount on the ue gas temperature distribution and NO
emission. Calculations were performed using the commercial packages ANSYS CFX 14.0.
& 2014 Elsevier Ltd. All rights reserved.
Keywords:
CFD
Rening vacuum furnace
Radiation simulation
Combustion simulation
1. Introduction
Rening vacuum furnace is one of the key facilities in atmospheric
and vacuum distillation units. It is used to raise the atmospheric
residuum temperature high enough to meet the vacuum gas oil
(VGO) yield target, and supply sufcient wash oil ow to prevent the
wash section packing from coking. The combined effect of high
operating temperature and heavy feedstock tends to increase difculties in design, especially in deep-cut unit in which the atmospheric
residue is heated to above 420 1C. Furnaces with design defects could
carry the risk of coking and result in relatively shorter operating life
cycles. It has been long established that the prediction of ue gas side
behavior is necessary for predicting the thermal efciency, coking
n
Corresponding author at: School of Chemical Engineering and Technology,
Tianjin University, Tianjin 300072, People's Republic of China.
Tel./fax: +86 22 27400199.
E-mail address: zhanglvh@tju.edu.cn (L. Zhang).
http://dx.doi.org/10.1016/j.ces.2014.11.001
0009-2509/& 2014 Elsevier Ltd. All rights reserved.
tendency and overall performance of the furnace. Besides, the understanding of the ue gas ow pattern and temperature eld distribution in the furnace is helpful for the ne tuning and optimization of
modern furnace design.
Due to progresses in computer hardware and software and
consequent increase of the calculation speed, the computational
uid dynamics (CFD) modeling technique becomes a powerful and
effective tool for understanding the complex chemical engineering
processes. It provides a theoretical basis and a computational
technique for predicting the ue gas velocity, temperature eld,
species concentrations as well as the process side conditions in a
tube furnace. The simulation of a tube furnace can be divided into
two parts. One is the ue gas side simulation including ue gas
ow, fuel combustion and heat transfer. The other is the process
side simulation including heater transfer, two-phase ow, phase
change and thermal cracking reaction. There have been many
attempts to model petrochemical furnaces by the use of CFD.
Heynderickx et al. (2001) used a 3-D CFD model, which was
71
Fig. 1. Conguration of the radiant section of the furnace. (a) 3D view, (b) Top view.
Table 1
Meanings of , , and S in governing equations.
Equation
Continuity equation
Momentum equation
ui
Enthalpy equation
Species equation
k-equation
h
Zi
k
ht
D
kt
-equation
Here: t
C k2
,
i
P k t u
xj
P
x
x i i xi j g i
i
Sh
0
Pk -
(/k)(C1Pk-C2)
uj
ui
xj xi
Table 2
Arrhenius coefcients for kt and kp.
Reaction rate constant
A (1/s)
Ta (K)
kt
kp
1.8 1011
6.4 106
38370
36510
72
1
S
t
xj
xj
xj
The means of , , and S for each equation are summarized
in Table 1.
2.2. Combustion model
In this work, the furnace was red with non-premixed burners
at oor and methane was used as fuel. Many researchers (Guihua
et al., 2011; Habibi et al., 2007b; Hu et al., 2011; Magnussen and
Hjertager, 1977; Yang et al., 2012) performed similar combustion
calculations using the combined EDM/FRC model (Eddy Dissipation model/Finite Rate Chemistry). A limitation of this model is
that the radical or intermediate species, such as CO and NO, cannot
be calculated with adequate accuracy. Besides, the O concentration is estimated using the O2 concentration and temperature
instead of calculating, and this may lead to inaccuracy for the
prediction of NO emission.
Due to the fact that the fuel and oxidizer (air) undergo fast
combustion which means the combustion rate is dominated by the
rate of mixing of the materials, the Laminar Flamelet model
(Peters, 2000; Sanders et al., 1997) is adopted for its good
performance in modeling turbulent ames at a high Damkhler
number. Flamelet equations are derived from energy and species
transport equations by applying a coordinate transformation and
assuming one-dimensional behavior of the combustion phenomena in the direction normal to the ame front. In these equations,
temperature and species mass fractions are functions of a conserved scalar known as mixture fraction (Z). Non equilibrium
effect, which means the inuence of the outer ow eld on the
inner reaction zone, is described by the scalar dissipation rate ()
at stoichiometric mixture:
2DZ2
be written as:
T 4 s
!!!
!!
I r ; s s s I r ; s n2
4
0
0
! !
! !0
I r ; s s U s d
0
3
In the context of typical combustion systems, the dominant
emitters of radiation are carbon dioxide and water vapor (although
hydrocarbons, CO and SO2 also make a minor contribution) in ue gas.
The weighted sum of gray gases model (WSGGM) (Heynderickx and
Nozawa, 2005; Liua et al., 1998; Stefanidis et al., 2007) was used for
the calculation of the absorption coefcient of ue gas. In the WSGGM,
the emissivity of the real gas is expressed as the weighted sum of the
emissivities of a number of gray gases and a transparent gas, and it is
presented as:
I
;i T1 e ki sP
i0
;i 1
i0
(6)
Table 3
The effect of the grid size on the outlet average temperature of ue gas.
No. of cells
Grid 1
Grid 2
Grid 3
5.74 10
7.06 106
8.19 106
N O2-NOO
(7)
(8)
(9)
HCN O2-NO
(10)
11
It is worth noting that Eq. (11) is derived from the reaction rate
of Reaction (6) multiplying by 2. This is based on the fact that
Reaction (6) tends to be the rate limiting step, producing both an
NO and N species, while Reaction (7) and (8) are assumed to be
fast, so one step of Reaction (6) will produce two NO molecules.
The prompt NO formation source was described by a single
reaction rate proposed by De Soete (1975) as:
M ave 1:5
SNO;p M NO kp O2 0:5 N2 CH 4
12
kt and kp in Eqs. (11) and (12) were the reaction rate constants and
they could be related to temperature by Arrhenius equation as:
k Ae T a =T
13
73
Q air;pri
Aair;pri
14
U air;sec
Q air;sec
Aair;sec
15
Industrial data
Simulation results
973.15
1.20
977.57
1.22a
Table 4
Furnace dimension and operating conditions.
Items
Furnace segment
Height (y-direction) (m)
Length (x-direction) (m)
Width (z-direction) (m)
Number of burners
coils
Number of passes
Number of tubes per pass
Tube length (m)
Tube diameter (mm) number per pass
Tube pitch (mm)
Firing conditions
Fuel composition
Fuel gas ow rate (kg/s)
Feed temperature of fuel (K)
Feed temperature of air (K)
Air composition (wt%)
O2
N2
Primary air ow rate (kg/s)
Secondary air ow rate (kg/s)
Excess air coefcient, q
Total heat input (kW)
Material properties
Emissivity of chamber wall
Emissivity of tube skin
Data
13.14
7.888
4.224
12 (6 of half)
6
24
1.2
152 17/168 2/219 2/273 2/325 1
304/336/438/546/650
Methane
0.354
300 K
370 K
23.2
76.8
4.38
2.92
1.20
19768
0.8
0.9
74
16
where Qfuel was the total fuel gas ow rate and Afuel was the total
cross area of all inlets.
2.5.3. Flue gas outlet
The ue gas outlet boundary was specied as average static
pressure of 30 Pa (gauge pressure) with pressure prole blend of
0.5 over whole outlet. The average constraint was applied by
comparing the area weighted pressure average over the entire
outlet to the specied value. The pressure prole at the outlet was
shifted by this difference such that the new area weighted
pressure average would be equal to the specied value. The ow
direction was an implicit result of the computation.
2.5.4. Wall and symmetry boundary conditions
In this work, no coupled furnace-reactor simulations (De
Schepper et al., 2009b; Guihua et al., 2011; Heynderickx et al.,
2001; Hu et al., 2011; Oprins and Heynderickx, 2003; Yang et al.,
2012) were performed, since only the ue gas side of the furnace
was simulated. Therefore, a xed tube skin temperature prole
(see Table S1 in Supporting Information), taken from industrial
operating condition, was applied and considered to be part of the
boundary conditions of the simulations. The inner wall of the
chamber was assumed to be adiabatic. A no-slip and smooth wall
boundary condition was imposed at all the walls and the
emissivity was specied the same as 0.8. The symmetry plane
of the geometry was specied as symmetry boundary as shown
in Fig. 2.
carried out to obtain converged results with the target value of the
normalized residual for each variable to be 10-5, as generally
recommended in the ANSYS CFX 14.0 User Manual (2011). The
average temperature of the ue gas at the outlet was taken as a
monitor parameter in the output control. During the simulation this
parameter was monitored and quasi-steady state was assumed to
prevail if the value remained constant for a period long enough to
determine the time-averaged values of the various parameters.
75
Fig. 5. Flue gas velocity vectors in XY-plane. (a) Z=0m, (b) Z=1.346m, (c) Z=2.1515m, (d) Z=2.792m.
76
Fig. 6. Flue gas velocity vectors in YZ-plane. (a) X=3.944m, (b) X=5.444m, (c) X=7.244m, (d) X=7.744m.
77
Fig. 9. Flue gas temperature proles along centerlines of the burners in the
symmetry plane. Burners are numbered in the minus X-direction.
Fig. 8. Flue gas temperature snapshots of vertical cross sections. (a) (d) XY-plane; (e) (h) YZ-plane. (a) Z=m, (b) Z=1.346m, (c) Z=2.1515m, (d) Z=2.792, (e) X=3.944m, (f)
X=5.444m, (g) X=7.244m, (h) X=7.744m.
78
5. Conclusions
In this article, we have attempted to predict the ue gas side
behavior of the radiant section of a rening vacuum furnace using
CFD. A steady-state 3D model was developed for modeling of
combustion and radiative heat transfer. Detailed insights into the
Fig. 10. NO mass concentration contours in vertical cross sections. (a) X=3.944, (b) X=4.444.
P
Pk
79
Nomenclature
A
pre-exponential factor
total area of primary air inlet (m2)
Aair,pri
Aair,sec
total area of secondary air inlet (m2)
Afuel
total area of fuel gas inlet (m2)
C1, C2, C constants of k- turbulent model
D
diffusion coefcient of mixture fraction equation
h
enthalpy (J kg-1)
I
radiation intensity (J m-2 s-1)
k
turbulent kinetic energy (m2 s-2)
ki
absorption coefcient of the ith gray gas (m-1)
kp
reaction rate constant for prompt NO formation (s-1)
kt
absorption coefcient of the ith gray gas (m3 kmol-1 s-1)
MNO
molar mass of NO (kg kmol-1)
Mave
mean molar mass of the mixture (kg kmol-1)
n
refractive index
Fig. 12. NO mass concentration distribution along the centerline of each burner in
the symmetry plane.
Fig. 14. Average ue gas temperature distribution along the furnace height.
Fig. 15. Average NO mass concentration distribution along the furnace height.
Fig. 13. Comparison of ue gas temperature snapshots of different excess air coefcients. (a) q=1.2, (b) q=1.5.
80
q
Qair,pri
Qair,sec
Qfuel
!
r
!
s
s
SNO,t
SNO,p
S
t
T
Ta
Uair,pri
Uair,sec
Ufuel
uj
xj
Z
Greek Letters
'
Subscripts
air,pri
air,sec
fuel
primary air
secondary air
fuel
Acknowledgment
We are grateful for the nancial support from the National
Science Foundation of China (No. 21336007).
Appendix A. Supporting information
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.ces.2014.11.001.
References
Chou, C.-P., Chen, J.-Y., Yam, C.G., Marx, K.D., 1998. Numerical modeling of NO formation
in laminar bunsen ames-a amelet approach. Combust. Flame 114, 420435.
De Schepper, S.C., Heynderickx, G.J., Marin, G.B., 2009a. Modeling the evaporation of a
hydrocarbon feedstock in the convection section of a steam cracker. Comput. Chem.
Eng. 33, 122132.
De Schepper, S.C., Heynderickx, G.J., Marin, G.B., 2009b. Coupled simulation of the
ue gas and process gas side of a steam cracker convection section. Aiche J. 55,
27732787.
De Schepper, S.C., Heynderickx, G.J., Marin, G.B., 2010. Modeling the coke formation in
the convection section tubes of a steam cracker. Ind. Eng. Chem. Res. 49, 57525764.
Detemmerman, T., Froment, G., 1998. Three dimensional coupled simulation of
furnaces and reactor tubes for the thermal cracking of hydrocarbons. Oil Gas
Sci. Technol. 53, 181194.
Driscoll, J.F., Chen, R.-H., Yoon, Y., 1992. Nitric oxide levels of turbulent jet diffusion
ames: effects of residence time and Damkohler number. Combust. Flame 88,
3749.
Guihua, H., Honggang, W., Feng, Q., 2011. Numerical simulation on ow, combustion
and heat transfer of ethylene cracking furnaces. Chem. Eng. Sci. 66, 16001611.
Habibi, A., Merci, B., Heynderickx, G., 2007a. Multiscale modeling of turbulent
combustion and NOx emission in steam crackers. Aiche J. 53, 23842398.
Habibi, A., Merci, B., Heynderickx, G.J., 2007b. Impact of radiation models in CFD
simulations of steam cracking furnaces. Comput. Chem. Eng. 31, 13891406.
Heynderickx, G.J., Froment, G.F., 1998. Simulation and comparison of the run length
of an ethane cracking furnace with reactor tubes of circular and elliptical cross
sections. Ind. Eng. Chem. Res. 37, 914922.
Heynderickx, G.J., Nozawa, M., 2005. Banded gas and nongray surface radiation
models for high-emissivity coatings. Aiche J. 51, 27212736.
Heynderickx, G.J., Oprins, A.J., Marin, G.B., Dick, E., 2001. Three-dimensional ow
patterns in cracking furnaces with longame burners. Aiche J. 47, 388400.
Hu, G., Wang, H., Qian, F., Zhang, Y., Li, J., Van Geem, K.M., Marin, G.B., 2011.
Comprehensive CFD simulation of product yields and coking rates for a oorand wall-red naphtha cracking furnace. Ind. Eng. Chem. Res. 50, 1367213685.
Lan, X., Gao, J., Xu, C., Zhang, H., 2007. Numerical simulation of transfer and reaction
processes in ethylene furnaces. Chem. Eng. Res. Des. 85, 15651579.
Launder, B.E., Spalding, D., 1974. The numerical computation of turbulent ows.
Comput. Method. Appl. M. 3, 269289.
Liua, F., Becker, H., Bindar, Y., 1998. A comparative study of radiative heat transfer
modelling in gas-red furnaces using the simple grey gas and the weightedsum-of-grey-gases models. Int. J. Heat Mass. Tran. 41, 33573371.
Magnussen, B.F., Hjertager, B.H., 1977. On mathematical modeling of turbulent
combustion with special emphasis on soot formation and combustion. Symposium (International) on Combustion, Elsevier, 719729.
Masoumi, M., Sadrameli, S., Towghi, J., Niaei, A., 2006. Simulation, optimization
and control of a thermal cracking furnace. Energy 31, 516527.
Oprins, A., Heynderickx, G., 2003. Calculation of three-dimensional ow and
pressure elds in cracking furnaces. Chem. Eng. Sci. 58, 48834893.
Peters, N., 1984. Laminar diffusion amelet models in non-premixed turbulent
combustion. Prog. Energ. Combust 10, 319339.
Peters, N., 1988. Laminar amelet concepts in turbulent combustion. Symposium
(International) on Combustion, Elsevier, 12311250.
Peters, N., 2000. Turbulent Combustion. Cambridge University Press.
Pitsch, H., Cha, C.M., Fedotov, S., 2003. Flamelet modelling of non-premixed
turbulent combustion with local extinction and re-ignition. Combust. Theor.
Model 7, 317332.
Pitsch, H., Peters, N., 1998. A consistent amelet formulation for non-premixed
combustion considering differential diffusion effects. Combust. Flame 114, 2640.
Sanders, J., Chen, J.-Y., Gkalp, I., 1997. Flamelet-based modeling of NO formation in
turbulent hydrogen jet diffusion ames. Combust. Flame 111, 115.
Souza, B., Matos, E., Guirardello, R., Nunhez, J., 2006. Predicting coke formation due
to thermal cracking inside tubes of petrochemical red heaters using a fast CFD
formulation. J. Petrol. Sci. Eng. 51, 138148.
Stefanidis, G., Merci, B., Heynderickx, G., Marin, G., 2007. Gray/nongray gas
radiation modeling in steam cracker CFD calculations. Aiche J. 53, 16581669.
Stefanidis, G., Van Geem, K., Heynderickx, G., Marin, G., 2008. Evaluation of highemissivity coatings in steam cracking furnaces using a non-grey gas radiation
model. Chem. Eng. J. 137, 411421.
Stefanidis, G.D., Merci, B., Heynderickx, G.J., Marin, G.B., 2006. CFD simulations of
steam cracking furnaces using detailed combustion mechanisms. Comput.
Chem. Eng. 30, 635649.
Warnatz, J., Maas, U., Dibble, R.W., 2006. Combustion: Physical and Chemical Fundamentals, Modeling and Simulation, Experiments, Pollutant Formation. Springer.
Yang, J., Tai, N., Wang, L., Xiao, J., Yang, C., 2012. Numerical simulation of the ue gas
and process side of coking furnaces. Ind. Eng. Chem. Res. 51, 1544015447.
ANSYS CFX 14.0 User Manual, 2011. Canonsburg, PA, USA, ANSYS Inc.,.