Beruflich Dokumente
Kultur Dokumente
h i g h l i g h t s
A complete coupled simulation method of convection section is proposed.
Velocity fields are not uniform along width direction due to asymmetrical structure.
Recirculation zones cause a longer residence time of flue gas and local overheating.
Process gas and tube skin temperature and heat flux have axial and radial profiles.
Changes of flow pattern are effected by gravity and centrifugal force.
a r t i c l e
i n f o
Article history:
Received 30 July 2015
Received in revised form 26 October 2015
Accepted 28 October 2015
Available online 2 November 2015
Keywords:
Coupled simulation
Convection section
Dual stage steam feed mixing
Cracking furnace
Computational fluid dynamics
Evaporation
a b s t r a c t
A complete coupled simulation of the convection chamber and tubes with dual stage steam feed mixing
of an industrial ethylene cracking furnace has been carried out with the computational fluid dynamics
(CFD) method for the first time. In the convection chamber, the standard ke model and discrete ordinates (DO) radiation model were respectively used in the descriptions of turbulence characteristics
and radiative heat transfer. In the tubes, renormalization group (RNG) ke model and volume of fluid
(VOF) model were respectively applied to the turbulence flow and the liquidvapor two phases flow.
Simulation results agree well with the industrial data. Based on the coupled result, a dynamic simulation
was calculated in the feedstock preheater (FPH). Simulation results show that the velocity and temperature fields are inhomogeneous distributions along the width direction due to the asymmetrical structure
of convection chamber. Two recirculation zones occur at the corner both near and away from the
entrance to the convection chamber, which will cause a longer residence time of flue gas and local overheating in furnace wall of convection chamber. The process gas temperature, tube skin temperature and
heat flux profiles are respectively different along the axial and radial direction of the high temperature
coil (HTC-I). The changes of flow pattern from bubble flow to spray flow are effected by gravity and centrifugal force during evaporation. The results will be helpful for the design and operation in cracking
furnace.
2015 Elsevier B.V. All rights reserved.
1. Introduction
Ethylene cracking furnace is the key equipment in the production of ethylene, which influences the yield efficiency of feedstock,
selectivity of important products, equipment energy consumption,
etc. In the world 99% of ethylene production has been adopted by
tubular cracking furnace [1].
Tubular cracking furnace mainly includes two parts: radiation
section and convection section. For the study of radiation section,
many achievements have been gained at home and abroad.
Corresponding author at: No. 130 of Meilong Road, Shanghai 200237, China.
E-mail address: fqian@ecust.edu.cn (F. Qian).
http://dx.doi.org/10.1016/j.cej.2015.10.093
1385-8947/ 2015 Elsevier B.V. All rights reserved.
437
Nomenclature
I
radiation intensity (J/m2/s)
_ pq , m
_ qp mass transfer from phase p to phase q and from phase q
m
to phase p respectively (kg/m3/s)
_ l!v , m
_ v !l rates of mass transfer due to evaporation and conm
densation respectively (kg/m3/s)
n
refractive index ()
~
r
position vector ()
~
s
direction vector ()
~
s0
scattering direction vector ()
s
path length (m)
Saq
mass source term of the phase q (kg/m3/s)
Su
source term ()
t
time (s)
T
local temperature (K)
Tl
temperature of the liquid phase (K)
Tv
temperature of the vapor phase (K)
u
fluid velocity (m/s)
The results show that the method can greatly improve the computational efficiency of the process gas side.
Convection section study early focused on the calculation of the
macroscopic phenomenon. For example, He et al. [14] developed a
simulation software of convection section based on Pro-II. Liu et al.
[15] and Zhou and Yang [16] respectively established their convection section programs based on Aspen Plus [17] platform. However,
the previous studies focused on the process modeles of convection
section, and greatly simplified the fluid flow and heat transfer processes inside the tube, thus, the improved understanding of the
occurring processes could not be taken into account. With
the development of CFD technique and computer technology, the
research in microscopic phenomena of the convection section has
also been paid more and more attention. The difficulty of numerical simulation study in the convection section lies in vaporization
of hydrocarbon feedstock in the tube. The existence of two-phase
flow makes the fluid flow, heat transfer and mass transfer become
more complicated. De Schepper et al. [18] added source terms of
the energy and mass source to the control equations by using the
volume fraction method, thereby preliminarily studying the fluid
flow boiling process of convection section. Mahulkar et al.
[19,20] and De Schepper et al. [21] studied numerical simulation
of coking phenomena of heavy feedstock pyrolysis, which provided
suggestions for reducing thickness of coke layer. Mertinger et al.
[22] studied the cause of heat tube corrosion in the convection section with ANSYS FLUENT 14.0 software, providing theoretical reference for the dangerous case judgement.
In the convection section, vaporization process of feedstock
hydrocarbon is an endothermic process, and the vaporization heat
required is provided by the waste heat of the flue gas. Therefore,
flue gas flow field has a close relationship with heat transfer and
vaporization inside tubes. Only efficient coupling heat transfer process between the convection chamber and tubes does accurately
get the vaporization and heat transfer and other such characteristics inside tubes, thus accurately predicting outlet temperature of
flue gas and process gas in convection section. De Schepper et al.
[23] used the CFD method to carry out coupled simulation of flue
gas side and the internals of the heat exchanger tubes in the convection section. However, they simulated convection section with
only one steam flux, and did not consider the heat changes of feedstock in the evaporation section, which may impact the coupled
simulation results.
Greek letters
absorption coefficient (1/m)
volume fraction of the phase q ()
gas density (kg/m3)
density of the liquid phase (kg/m3)
density of the vapor phase (kg/m3)
density of the phase q (kg/m3)
Cu
generalized diffusion coefficient ()
r
StefanBoltzmann constant (r = 5.672 108 W/m2K4)
rS
scattering coefficient (1/m)
u
dependent variable ()
U
phase function ()
X0
solid angle ()
a
aq
q
ql
qv
qq
@qu
divquu divCu gradu Su
@t
438
"
n
X
!
@
_ pq m
_ qp
m
aq qq r aq qq v q Saq
@t
p1
#
2
The volume fraction equation will not be solved for the primary
phase; the primary-phase volume fraction will be computed based
on the following constraint:
aq 1
q1
This paper adopted the evaporationcondensation model proposed by Lee [25]. The model added the mass and energy exchange
source terms of between vapor and liquid to the control equations,
and correlated fluid saturation temperature. Based on the following temperature regimes, the mass transfer can be described as
follows:
If Tl > Tsat
_ l!v bl al ql
m
T l T sat
T sat
If Tv < Tsat
_ l!v bv av qv
m
T sat T v
T sat
r I~
r;~
s~
s a rs I~
r;~
s an2
rT 4 rs
p 4p
Z 4p
I~
r;~
s0 U~
s ~
s0 dX0
6
Domain-based weighted-sum of gray gas model (WSGGM) is
used for the calculation of the emissivity of the flue gas mixture
[13].
2.4. Physical properties
In the industry, there are many kinds of hydrocarbon
feedstock which usually include naphtha, diesel oil of ordinary
pressure, light hydrocarbon, heavy-vacuum gas oil, etc. And
online detection instrument is expensive, which is seldom
adopted in actual production, therefore it is difficult to get the
detailed physical property data of hydrocarbon feedstock in industry, thus leading to difficulty for the numerical simulation of
convection section.
In this work, the combinations of COILSIM1D [28] and Aspen
plus give rise to the detailed information of material using industrial data. The detailed steps are described as follows: first of all,
the collected industrial data such as the density, distillation range,
PIONA value of naphtha are input to COILSIM1D. The industrial
indices of the naphtha feedstock are described in Table 1. Secondly,
COILSIM1D is adopted to build feedstock model and to obtain the
detailed feedstock component data. Thirdly, these components
are input to Aspen Plus, which is used to calculate the physical
properties of mixture based on SoaveRedlichKwong equation
[29]. Finally, based on the data of calculation above, polynomial
functions expressing the variation of the physical properties with
temperature are obtained by least square method, as shown in
the Supporting information. Among these physical properties, the
polynomial functions of the flue gas, the dilution steam and the
mixture of naphtha and dilution steam are calculated by Aspen
Plus, similar to the last two steps of the physical properties calculation of naphtha. These polynomial functions of physical properties are input into Ansys Fluent 14.0 [30] for the numerical
simulation work in the next step. The detailed calculation process
is shown in Fig. 1.
3. Simulation procedure
3.1. Geometry and operating conditions
The structure diagram of dual stage steam feed mixing of convection section of ethylene cracking furnace is shown in Fig. 2. This
section consists of eight tube sections, which are the FPH, economizer (ECO), high temperature coil I (HTC-I), high temperature coil
Table 1
Industrial indices of the naphtha feedstock.
Specific density (20 C) (g/cm3)
Specific density (15 C) (g/cm3)
0.715
0.7196
318.15
342.15
365.15
434.15
442.15
464.15
PIONA (wt%)
n-Paraffins
i-Paraffins
Olefin
Naphthenes
Aromatics
30.16
41.56
1.37
16.13
10.78
439
Computing feedstock
information with
COILSIM1D
Input
information
1. Density
2. Distillation range
3. PIONA value
NAP
FPH
BFW
ECO
Drum
DS
Computing feedstock
properties with Aspen Plus
Output
information
HTC-I
1. Density
2. Thermal conductivity
3. Heat capacity
4. Dynamic viscosity
HTC-II
Drum
HPSSH-I
Desuperheater
HPSSH-II
DS
DSSH
HTC-III
Radiation section
Inlet of convection
section
Fig. 3. Diagram of convection section process with dual stage steam feed mixing.
Fig. 2. Schematic diagram of convection section with dual stage steam feed mixing.
440
Table 2
Convection section dimensions and operating conditions.
Convection chamber specifications
Length (z-direction) (m)
Width (x-direction) (m)
Height (y-direction) (m)
Number of tube sections
Number of total tubes
Straight length of each tube pass
10.3689
2.26
13.385
8
432
10.368
24
6
0.0603
0.00554
8.4861
333.15
0.48
18
6
0.0603
0.00874
10.4475
418.15
11.92
6
6
0.1143
0.00602
8.4861
0.8486
406.15
0.40
3
10
0.0603
0.00874
10.2428
0.5850
597.15
656.15
11.69
11.55
4
6
0.1143
0.00602
4.2431
458.15
0.40
4
6
0.1413
0.00655
8.4861
5.0917
719.15
0.33
where each tube consists of 6 straight tubes and 6 bend tubes. The
grid cell number of each tube is 101,036.
HPSSH-I tube section consists of 10 tubes where each tube consists of 3 straight tubes and 2 bend tubes. The grid cell number of
each tube is 48,650.
HPSSH-II tube section structure is the same as the HPSSH-I. The
grid cell number of each tube is 48,804.
441
Start
Calculate
equations of
DSSH
Calculate
equations of FPH
Convergence?
N
Convergence?
Y
Calculate equations
of HTC-I
Calculate
equations of
HPSSH-I
N
Convergence?
Calculate
equations of
ECO
N
Convergence?
Calculate equations of
HPSSH-II
Calculate equations of
mass, momentum, energy
and species
Convergence?
Convergence?
Calculate equations of
HTC-II
Y
N
Convergence?
Y
Convergence?
Calculate equations
of HTC-III
Convergence?
N
N
Convergence?
End
Fig. 4. Coupled calculation procedure of tubes and furnace with dual stage steam feed mixing.
Table 3
Comparison of simulation results and industry data.
Items
Industry data
Simulation results
385.15
430.15
523.15
609.15
677.15
695.15
793.15
844.15
884.15
388.65
437.41
527.78
615.75
681.61
698.66
792.44
843.18
887.38
consistent along z axial direction. At the corner both near and away
from the entrance to the convection chamber, flue gas velocity is
less than that in the other places. Fig. 6 shows flue gas velocity vector field at cross section z = 5 m. At the corner both near and away
from the entrance to the convection chamber, there are two recirculation zones obviously. The recirculation zone will make residence time of high temperature flue gas lengthen in the
convection chamber. At the same time, it will cause local overheating in furnace wall of convection chamber, thus being not conducive to the stable operation of the cracking furnace. When
high temperature flue gas entering the convection chamber, its
velocity is larger and will scour the furnace wall of opposite side.
In the convection chamber bottom, flue gas velocity distribution
along the x axial direction is not uniform. This is because the structure of the convection chamber is asymmetrical. With the increase
of height the flue gas passes through the staggered tube bundles
and its velocity field along the x axial direction is gradually uniform. While arriving at the outlet of convection chamber, the flue
gas velocity is basically consistent along x axial direction.
Fig. 7 shows flue gas temperature contours at section z = 0.5, 5
and 10 m of convection chamber respectively for dual stage steam
feed mixing. One can see that flue gas temperature distribution
along the x axial direction is not uniform in the bottom of convection chamber. With the increase of height, the flue gas passes
through the staggered tube bundles and its temperature field along
the x axial direction is gradually uniform. Fig. 8 shows flue gas temperature profiles along x and z axial direction. In Fig. 8a, when the
height of 3.10 m, flue gas temperature at x = 0.1 m is smaller than
that of x = 1.0 m. This is because the flue gas velocity of x = 0.1 m
which is near the entrance to convection chamber is smaller than
that of x = 1.0 m (this is can be seen in Fig. 5), leading to a smaller
heat transfer coefficient, and thus a lower flue gas temperature.
The flue gas of upward flow carries out heat exchange with the
tubes, and the flue gas temperature field also gradually becomes
uniform. From Fig. 8a, the temperature basically arrives at
consistence at y = 8.88 m, showing that the flue gas temperature
field gradually becomes uniform through the heat exchange with
staggered tubes. While arriving at the outlet of convection
chamber, the flue gas temperature is basically consistent along x
axial direction, showing staggered tube facilitates the exchange
of heat. Both figures show with the increase of height, flue gas
temperature gradually decreases. As can be seen from Fig. 8b, the
flue gas temperature field is basically consistent along z axial
direction, which is caused by the uniform flow field along z axial
direction in Fig. 5.
442
Fig. 7. Flue gas temperature contours with dual stage steam feed mixing. (a)
z = 0.5 m; (b) z = 5 m; (c) z = 10 m.
Fig. 5. Flue gas velocity contours of furnace with dual stage steam feed mixing. (a)
z = 0.5 m; (b) z = 5 m; (c) z = 10 m.
1400
1200
1000
800
600
y=3.10 m
y=5.76 m
y=7.31 m
y=10.34 m
400
200
(a)
y=4.84 m
y=6.53 m
y=8.88 m
0
0
0.2
0.4
0.6
0.8
1.2
1.4
Width (m)
1400
1200
1000
800
600
200
(b)
0
0
Fig. 6. Flue gas velocity vector filed at cross section z = 5 m. (a) Velocity vectors
enlarged diagram at corner away from the entrance to the convection chamber; (b)
velocity vectors enlarged diagram at corner near the entrance to the convection
chamber.
y=3.10 m
y=5.76 m
y=7.31 m
y=10.34 m
400
y=4.84 m
y=6.53 m
y=8.88 m
10
Length (m)
Fig. 8. (a) Flue gas temperature distribution along x axial direction in convection
chamber; (b) flue gas temperature distribution along z axial direction in convection
chamber.
Fig. 10. Process gas temperature contours of different cross sections along z axial
direction in HTC-I. (a) 3 m; (b) 6 m; (c) 9 m.
443
Fig. 12. External tube skin temperature profiles of different passes in HTC-I.
the two ends of tube lie in near the furnace wall, where the flue gas
temperature is low, leading to a low difference between the flue
gas temperature and tube skin temperature, thus resulting in a
low heat flux. One can also see from this figure that in the Pass
1, Pass 3 and Pass 5, the heat flux increases along tube axial position, while in the Pass 2, Pass 4 and Pass 6, the heat flux decreases
along tube axial position.
Fig. 10 shows the process gas temperature contours of
different cross sections along z direction. Along the hydrocarbon
feedstock flow direction, the temperatures of the hydrocarbon
feedstock near the tube wall firstly rise, then the heat is gradually
transferred to the center of tube. It shows that temperatures near
the tube wall are higher than that of tube center. When reaching
the tube outlet, the temperature profile has been average in z axial
direction.
Fig. 11 shows the process gas temperature profiles of different
passes along tube axial position in HTC-I. The process gas temperature increases from Pass 1 to Pass 6. This is because the process
gas absorbs the heat transferred by flue gas, resulting in the
increase of temperature.
Fig. 12 shows the external tube skin temperature profiles of different passes along tube axial position in HTC-I. The external tube
skin temperature increases from Pass 1 to Pass 6 except near
the two ends of tubes, that is, in the crossover position of two tube
passes, the staggered external tube skin temperature exists. This is
because the two ends of tubes connects the bend tubes that is
outside of the convection section, which are not heated by flue
gas, leading to the lower tube skin temperature than inside the
convection section.
444
Fig. 14. Mass fraction profile of FPH straight tube: (a) bubble flow; (b) elongated
bubble flow; (c) stratified flow; (d) spray flow.
(1) There is recirculation zone at the corner both near and away
from the entrance to the convection chamber. The recirculation zone will make residence time of high temperature flue
445
Acknowledgments
This work is supported by Major State Basic Research Development Program of China (2012CB720500) and National Natural
Science Foundation of China China (Key Program: U1162202).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.cej.2015.10.093.
References
[1] J.L. Qian, Tubular Heating Furnace, China Petrochemical Press, Beijing, 2003.
[2] G.J. Heynderickx, A.J.M. Oprins, G.B. Marin, E. Dick, Three-dimensional flow
patterns in cracking furnaces with long-flame burners, AIChE J. 47 (2001) 388
400.
446
[26] Z. Yang, X.F. Peng, P. Ye, Numerical and experimental investigation of two
phase flow during boiling in a coiled tube, Int. J. Heat Mass Transf. 51 (2008)
10031016.
[27] A. Habibi, B. Merci, G.J. Heynderickx, Multiscale modeling of turbulent
combustion and NOx emission in steam crackers, AIChE J. 53 (2007) 2384
2398.
[28] K.M. Van Geem, R. Zajdlk, M.F. Reyniers, G.B. Marin, Dimensional analysis for
scaling up and down steam cracking coils, Chem. Eng. J. 134 (2007) 310.