See

discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/264233919

Simulation of the firebox of thermal cracking

furnaces

ARTICLE · JANUARY 2000

READS

37

2 AUTHORS, INCLUDING:

S.M. Sadrameli
Tarbiat Modares University
70 PUBLICATIONS 544 CITATIONS

SEE PROFILE

Available from: S.M. Sadrameli

Retrieved on: 26 December 2015
 PETROCHEMICALS

Simulation of the firebox of

thermal cracking furnaces
This article deals with the simulation of thermal cracking furnaces of olefin
plants, in particular those of the Tabriz plant in Iran

M Sadrameli Tarbiat Modares University

H A Zahmatkesh Amirkabir Petrochemical Company

P
roduction of ethylene has grown
exponentially over the past three
decades. One of the major sources
of this versatile building block is the
pyrolysis of hydrocarbons, notably
ethane, propane, butane, naphtha and
other heavier hydrocarbons. Thermal
cracking of hydrocarbons for olefin
production is carried out in long tubu-
lar reactors, inserted in a furnace.
With the help of an accurate simula-
tion model for the radiative heat
exchange in the furnace, the calcula-
tion of the temperature and heat flux Ro ( = 0)
distribution in the firebox and the Ri ( = 0)
reactor simulation can be coupled. Ro ( = 90)
The combined furnace and reactor R i ( = 90)
design can then be optimised in order
to obtain a favourable product distri-
bution or to reduce unwanted side
effects, such as fouling of the heat
transfer surface by carbonaceous Figure 1 Tube skin temperature results along the reactor length
deposits.
The mathematical model of the fur- expected performance. Thus, faulty the reactor coil inside the furnace, con-
nace consists of the multi-zone radia- instruments, poorly operating equip- sisting of a number of a straight verti-
tion model for the firebox coupled ment and any other operating prob- cal tube called passes connected by
with the kinetic model for the tubular lems can be easily discovered. bends, and the positioning and type of
reactor inside the furnace. The model Simulators are also useful when the burners require special attention.
equations are solved numerically. changes in operation are contemplat- With the help of an accurate simula-
The results of the furnace model are ed. They can be used to test new oper- tion for the radiative heat exchange in
tested by the coupled simulation of ating conditions or plant strategies to a furnace, the calculation of the tem-
heat transfer and reaction in an indus- change operation with minimal impact perature and heat flux distributions in
trial thermal cracking furnace in the on production. Simulators employing the firebox and the reactor can be
Tabriz olefin plant, Iran, and also with advanced kinetic models can be espe- achieved. The simulation of the firebox
3,4,
the design data using the Spyro simula- cially useful for the prediction of was developed by Hottel Vercam-
7 1 5
tion package. important products related to the mar- men , Froment and Paramenswaran .
ket demands. The multi-zone mathematical model
Uses of process simulators Finally, the simulator can be used to has been used by Sadrameli for the
9

A process simulator is a powerful tool train new operators and/or provide simulation of the cracking furnaces
for plant operators in a variety of ways. refresher training for current operators. assuming one dimensional conduction
The ability to simulate the current Thus, operating experience equivalent heat transfer in the tubes.
operation of a plant is an important to several years on the job can be In the present work this model is
capability. A simulator can provide a acquired in a relatively short time. Ben- extended to three-dimensional model
wealth of information not readily efits can be realised in both smoother and has been applied to simulate the
available from control instrumenta- and safer operations. thermal cracking furnaces of Tabriz
tion. It provides this information in an Furnace model olefin plant. The furnace walls, the
easily understandable form for both Thermal cracking of naphtha is one the reactor coils and the flue gas volume
the operators and plant engineers. most important routes in production are discretised into a number of isother-
The simulation results can be used to of olefins such as ethylene, propylene mal surface and volume zones with
check plant actual performance against and butadiene. The configuration of uniform properties. For the calculation

135
P T Q SPRING 2000
 PETROCHEMICALS

Cracking furnace characteristics

R = 121
Furnace:
R = 124
Length 10480mm
Height 11473mm R = 127
Width 2100mm R = 130
Number of side wall burners 108
Firing conditions:
Quantity Fuel 137.kmole/hr
Air Excess 15%
Total Fired Heat input 23.846E+6 Kcal/hr
Reactor:
Total coil length 45m
Internal diameter of passes 85mm
wall thickness 7mm
Internal diameter of passes 121mm
wall thickness 9mm
Operating condition:
Feedstock
Hydrocarbon feedrate/tube 1126 kg/hr Figure 2 Circumferential tube temperature results on the reactor wall
Steam dilution 778.4 kg/hr
Coil inlet temperature 596.5°C
Sarofim4. It involves subdivision of the power Ei. The emissive of a black body
Coil outlet pressure 1.55 (atm)
radiating enclosure into isothermal with temperature Ti is given by Stafen-
Material properties:
Emissivity of Furnace wall 0.7
volume and surface zones. Mathemati- Boltsmann law. The calculation of the
Emissivity of the tubes 0.94 cal models based on the zone method total exchange area for the flue gas vol-
vary in complexity, depending on the ume, the refractory wall zones, and the
number and arrangement of zones. tube surface zones were discussed in
6
Table 1 Three main types of zone models can detail by Rao et al .
be identified: The non-radiative heat flux emitted
of the direct and total-exchange areas, — Single-gas zone model (assumes uni- by zone Zi is represented by Qi. Solving
a fundamental approach considering form gas temperature and composi- the set of energy balances yields the heat
individual band absorption by carbon tion) flux and temperature distribution in the
dioxide and water, are taken. In addi- — Long-furnace models (one dimen- furnace. When the reactor tubes are dis-
tion, the position of the burners in the sional model consists of a longitudinal cretised in axial direction only, heat flux
furnace walls and the flue gas flow pat- series of well stirred zones where the profiles q(z) and external tube skin tem-
tern in the firebox are explicitly enthalpy of the combustion products perature profiles Text (z) are obtained.
accounted for. leaving one zone is the same as input) When the tubes are also discretised
Simulation results are given for — Multi-dimensional zone models. along the perimeter, heat flux profiles
naphtha cracking furnaces in Tabriz This model is used when both longi- T(r, 0-, z) are obtained. The conduction
petrochemical complex. tudinal and cross sectional or radial equation for each axial tube zone with
Multi-zone model variation in temperature and heat flux the appropriate boundary conditions
The development of the zone method are expected to be important. In this are:
of analysis is attributed to Hottel and model the furnace refractory walls, the
tubes and the flue gas volume are (2)
Firing details divided into a number of isothermal
surface and volume elements with uni- By assuming that the temperature in
Calculation basis = 25°C form properties. each small element is is constant, Equa-
Fuel molecular weight The energy balance, containing con- tion 1 can be can be simplified as:
1.4136E+01 kg/kmole tribution of radiative, convective
Lower heating value and/or conductive heat exchange, is (3)
1.7355E+05 kcal/kmole constructed for each of these elements,
Lower heating value resulting in a set of non-linear algebra-
1.2277E+04 kcal/kg ic equation in the form shown in Equa- Notation
Fuel gas temperature 1.0000E+01°C tion 1 below: Cp Heat capacity of gas species
Air temperature 1.0000E+01°C (J/mol.K)
Excess air 1.5000E+01 E Black body emissive power (W/m2)
Firebox pressure 8.5000E-01 kg/cm
2 Hp Process gas convection
coefficient (W/m2K)
O2 Stoic 1.7936E+00
k Thermal conductivity of tube (W/m2K)
O2 Act 2.0626E+00 q Heat flux (W/m2)
Air (kmole) 9.9728E+00 air/fuel Q Heat flux (W/m2)
Fired heat 2.3846E+07 kcal/hr r Tube radius (m)
Heat flux of one burner Re External tube radius (m)
The matrix element ZiZj represents
2.2080E+05 kcal/hr Ri Internal tube radius (m)
the total exchange area between the T Temperature (°C)
Fuel flow to the side burners
zone Zi and Zj. This is the amount of Z Axial reactor co-ordinate (m)
1.3740E+02 kmole/hr
Air flow to the side burners
radiative energy emitted by the zone Zi ZiZj Total exchange area between
and absorbed, both directly and after zones i and j (m2)
1.3703E+03 kmole/hr
reflection on other zones, by the zone α Thermal diffusivity (m2/s)
0- Tube perimeter angle
Table 2 Zj, divided by the black body emissive

136
P T Q SPRING 2000
 PETROCHEMICALS

Flue gas properties

Molecular weight. Flue

2.7660E+01 kg flue/kmol flue
Flue 1.0902E+01 mole.flue/mole.fuel
TOT (Flue gas flow rate)
1.4980E+03 kmol/hr
TOT (Flue gas flow rate)
4.1435E+04 kg/hr
Flue gas specific heat (CP)
3.3527E-01 kcal/kg.C
Flue gas enthalpy (hfg)
4.3580E+02 kcal/kg
3D MODEL Enthalpy in flue gas leaving
1.8057E+07 kcal/hr
SPYRO Flue gas viscosity 5.4898E-02 CPoise
1D MODEL

Table 4

Furnace dimensions
Figure 3 Comparison of the simulation results with Spyro
Floor to roof 11.4730m
Fuel gas composition leaving the stack End wall to end wall 10.4800m
Side wall to centre line of tubes 1.0500m
Component kmole/kmolefuel kmole/hr mole% weight% Nr of end wall zones 22.0000
Nr of roof zones 4.0000
CO2 8.644E-01 1.188E+02 7.929E-02 1.262E-01 Nr of bottom zones 4.0000
H2O 1.986E+00 2.729E+02 1.822E-01 1.186E-01 Nr of cold plane zones 44.0000
O2 2.690E-01 3.697E+01 2.468E-02 2.855E-02 Nr of gas cube zones 44.0000
Average beam length 1.5400m
N2 7.690E+00 1.057E+03 7.054E-01 7.144E-01
Zone dimensions 1.05m
Convection coefficient 29.2145 kcal/hr.m2.C
Table 3 Gas emissivity at (1220.00)C 0.2240
Gas absorptivity at (1064.92)C : 0.2283
The boundary conditions for Equa- coke thickness and composition profile Weight factor at TG (AG) 0.3348
tion 3 are now defined in the internal along the reactor. Weight factor at TR (AR) 0.3622
and external surfaces of the coils as:
Naphtha cracking furnaces
(4) The main dimensions and operating
conditions of the naphtha cracker fur- Table 5
(5) nace with the staggered row configura-
tion reactors are given in Table 1. The 108 burners positioned in eight rows in
(6) furnace considered in this work contains both front and rear walls. The furnace is
eight split coils in a staggered row con- divided into zones by means of four
(7) figuration, each making eight passes equidistant horizontal planes. For lack
through the furnace. Because of the of symmetry, all eight circumferential
The discretisation of the above equa- symmetry, only two reactor coils had to zones per axial division had to be simu-
tions was discussed in detail by Heynd- be simulated. The furnace is heated by lated.
erickx et al2. The calculation module for
the circumferential temperature profiles Refractory temperature profile (°C)
is coupled to the furnace and reactor
simulation model. Height, m Zone 1 Zone 2 Zone 3 Zone 4
Kinetic model
Thermal cracking of light and heavy 10.95 990.80 1007.12 1027.16 1048.52
hydrocarbons is simulated based on the 9.91 991.02 1007.09 1027.15 1048.54
radical mechanism. Data bank of chem- 8.87 1033.24 1047.08 1067.04 1091.69
ical kinetics contains over 1000 reac- 7.83 1045.28 1057.37 1076.20 1117.54
tions. Solution of unsteady state
6.79 1068.45 1079.22 1097.08 1129.96
differential equations set is developed
5.75 1083.68 1092.75 1109.94 1125.90
based upon the Gear method. Details of
the kinetic model and the solution may 4.71 1083.47 1090.83 1107.10 1111.08
10
be obtained from Towfighi et al . 3.67 1072.98 1078.87 1094.30 1125.82
Coil outlet temperature, feed condi- 2.63 1059.61 1064.33 1078.60 1092.69
tions, run length or maximum tube skin 1.59 1075.75 1078.35 1090.13 1103.42
temperature and recycle components 0.55 1102.08 1102.12 1109.56 1121.20
are the main parameters which can be
altered in the model. The results of the
model include: coil outlet conditions, Table 6

138
P T Q SPRING 2000
 PETROCHEMICALS

Flue gas temperature profile (°C)

“The development of the
Height 157m Spyro 2.62m Spyro 3.67m Spyro 4.72m Spyro zone method of analysis is
10.95 1099.12 1097.5 1116.80 1116.4 1137.22 1135.8 1160.42 1159.4 attributed to Hottel and
9.91 1134.38 1134.2 1150.64 1151.5 1172.51 1172.5 1195.89 1195.5 Sarofim. It involves sub-
8.87 1183.78 1185.9 1197.59 1200.9 1218.54 1220.6 1240.20 1242.1 division of the radiating
7.83 1201.48 1202.2 1213.92 1216.0 1234.13 1234.8 1254.52 1255.3
enclosure into isothermal
6.79 1232.12 1228.9 1242.66 1241.1 1262.10 1259.2 1281.11 1278.6
volume and surface zones.
5.75 1249.53 1252.6 1258.37 1263.4 1277.20 1280.6 1294.54 1298.8
4.71 1245.72 1249.7 1252.88 1259.6 1271.51 1276.1 1287.30 1293.3
Mathematical models
3.67 1227.77 1234.4 1233.79 1242.7 1251.85 1258.7 1266.37 1274.6 based on the zone method
2.63 1208.86 1190.7 1213.62 1210.7 1231.11 1226.2 1244.23 1240.7 vary in complexity,
1.59 1236.66 1212.6 1239.37 1230.4 1254.38 1243.6 1266.44 1257.8 depending on the number
0.55 1267.36 1255.0 1267.20 1268.8 1275.74 1276.6 1286.13 1290.1 and arrangement of zones”
Table 7

Tube skin temperature profiles (°C) Results and discussion

The results of the simulation program
Height Tubes 1-8 Tubes 9-16 Tubes 17-20 Tubes 21-24 are listed in Tables 2 to 9 and are illus-
trated in Figures 1 to 3.
Model Plant Model Plant Model Plant Model Plant Figure 1 represents the calculated
10.92 796.5 795 892.43 890 903.64 898 955.68 952 tube skin temperature profiles along
the reactor length at two angular posi-
9.88 835.7 829 909.68 906 923.49 916 970.81 963
tions and on its external and internal
8.84 869.2 862 926.26 922 944.06 936 984.66 976 surfaces. The discrepancies between
7.80 890.7 882 947.48 942 957.92 948 993.04 983 the temperature values at zero (front
view of the burner) and 90 degrees
6.76 909.5 900 955.84 949 969.32 959 998.68 988
reflect the shadow effects on the tem-
5.72 918.3 912 959.68 953 976.29 966 1000.25 989 peratures.
4.68 932.6 913 955.82 950 976.05 967 994.31 985 Circumferential non-uniformities
are of particular importance for severe
3.64 936.7 925 946.81 941 971.37 962 984.08 975
operating conditions, whereby the
2.60 934.3 925 939.41 932 968.87 958 975.83 965 tube skin temperature is high and close
1.56 943.5 932 942.65 933 973.87 962 974.79 963 to the limits imposed by the tube mate-
rial properties. Therefore, a representa-
0.52 952.2 943 945.26 938 978.91 969 973.32 965
tive tube skin temperature is a
Table 8 necessity and the measurements
should be correctly interpreted.
Figure 2 illustrates the circumferen-
tial temperature variations at different
Tube wall heat flux profile (kCal/m2.hr
radial positions of the tube. They clear-
ly show that the maximum tempera-
Height Tubes 1-8 Tubes 9-16 Tubes 17-20 Tubes 21-24
tures are always at the zones which are
Model Spyro Model Spyro Model Spyro Model Spyro facing the front walls (0 and 180
10.92 56452.97 52630 36525.75 35550 36138.59 35480 28244.71 29297 degrees) which receive the maximum
radiation from the furnace wall and
9.88 58017.18 57620 42865.08 41760 42548.32 41620 33696.48 32980 the burners and the minimum temper-
8.84 63995.94 63590 50938.80 49750 49325.61 48340 40451.52 39630 atures occur in the zones at both sides
7.80 66207.57 65350 52148.24 50910 51945.29 50740 43711.70 42580 of the tubes (90 and 270 degrees), due
to the shadow effect of the neighbour-
6.76 69287.41 67610 57044.04 55270 55442.81 53570 47823..01 46130 ing tubes.
5.72 71593.33 69640 60096.79 58740 57063.77 55640 50249.27 49030 Cracking furnaces are controlled on
4.68 66835.92 68790 59922.63 58910 55633.98 54380 49813.60 48780 the basis of the process gas exit tem-
perature and/or composition. Thermo-
3.64 61648.64 61840 57876.81 57030 52286.60 51270 47672.90 46770 couples inserted in intermediate
2.61 61457.54 57120 56636.58 54720 49341.90 47380 45725.64 43910 locations of the coil do not last very
1.56 61325.74 58700 59753.80 57320 50931.11 48440 47998.43 45740 long since the cracking temperature is
too high.
.52 62644.21 60700 61812.05 60100 51158.30 49750 49384.48 48180 Therefore, the external tube skin
temperature is measured periodically
Table 9 by the radiation pyrometers, through

140
P T Q SPRING 2000
 PETROCHEMICALS

peepholes in the furnace walls. difference between maxima and mini- been involved in the development of most of
The simulated results show that the ma can be of the order of 30°C. Yet the petrochemical plants, especially olefins,
observed temperatures depend upon these differences cannot always be in Iran. He has also held various positions in
the location of the peephole with detected since temperature measure- National Petrochemical Company
respect to the tubes. The results of the ment relies upon infrared pyrometer,
model have been compared with the through peepholes in the walls of the
design data obtained from the plant. furnace, so that certain view angles References
Unfortunately, except for the tube skin only are possible. 1 Froment G F, Vercammen H A .J, “Simula-
tion of thermal cracking furnaces”, Chem
temperatures which are measured by The model and simulation software Reaction Eng, Houston, 271, 1978.
the pyrometer, other thermal charac- presented here is used as a guide for 2 Heynderickx G J, Cornelis G G “Circumfer-
teristics of the furnace such as gas and plant operators in Tabriz petrochemi- ential tube skin temperature profiles in
thermal cracking coils“ AIChE J,38 (12)
refractory temperatures and heat flux cal complex to control the furnace ,1905, 1992.
measurement are impossible or inac- parameters. 3 Hottel H C, “Radiant heat transmission”
curate. Trans Am Soc Mech Engrs, 53, 265 1931.
4 Hottel H C, Sarofim A F, Radiative Heat
Figure 3 shows the comparison Transfer, McGraw Hill, New York, 1967.
between the present three-dimension- 5 Paramenswaran A K, Sharma, V K, “Mod-
al model and the experimental results M Sadrameli is an associate professor of elling of naphtha pyrolysis in the swaged
for the tube skin temperatures. The chemical engineering at Tarbiat Modares coils”, Can J Chem Eng, 66, 957, 1988.
6 Rao M V R, Pleheirs P M, Froment G F, “The
agreement between the present simu- University, Tehran. He graduated from Sharif coupled simulation of heat transfer and
lated results and the experimental data University, Tehran, in 1980. He holds MSc reaction in a pyrolysis furnace”, Chem Eng
are observed in the figure. and PhD degrees in chemical Sci, 43, 1223, 1988.
7 Vercammen H A J and Froment G F, “An
The values of the tube skin temper- engineering from Leeds University, UK. His improved zone method for the simulation
atures and the heat flux on the tubes research interests are heat recovery, heat of radiation in industrial furnaces”, Int J
are also listed in Tables 8 and 9. There- transfer, modelling and simulation of Heat Mass Trans, 23, 329, 1980
8 CRACKSIM, Simulation software for the
fore, the comparison with plant data is petrochemical processes, especially olefin naphtha cracking furnaces; Tarbiat Modares
made for only tube skin temperature technology. He is presently head of the University, Chemical Engineering Depart-
and for the other parameters that are chemical engineering division in the ment, Tehran, Iran, 1999.
9 Sadrameli M, “Heat Transfer Calculation in
based on the design data obtained department and consultant on energy the Firebox of the ethylene plant Furnaces“,
from the standard package of Spyro. conservation projects in the oil and energy Int J of Eng, 10, 4, pp 219-228, Iran, 1997.
Non-uniformity of temperature ministries. 10 Towfighi J, Karimzadeh R, “Modelling and
simulation of thermal cracking of heavy
along the perimeter of the vertical H A Zahmatkesh is managing director of hydrocarbons through radical
coils shown in Figure 2 is considered Amirkabir Petrochemical Complex, Iran. He mechanisms”, 3rd National Congress of
to be significant. Even with a coil the graduated from Shiraz University, and has Chem Eng, Tehran, Iran, 1998.

141
P T Q SPRING 2000