close

Вход

Забыли?

вход по аккаунту

?

Modelling and Simulation of a Top-Fired Primary Steam Reformer using GPROMS.

код для вставкиСкачать
Dev Chem. Eng. Mineral Process, lO(1R).p p 77-87. 2002
Modelling and Simulation of a Top-Fired
Primary Steam Reformer using gPROMS
A.J. Dunn, J. Yustos’ and I.M. Mujtaba”
Computational Process Engineering Group, Department of Chemical
Engineering, University of Bradford, West Yorkshire, BD7 1DP, UK
I
Department of Chemical Engineering, University of Valladolid,
Valladolid, Spain
A top-fired primaty steam reformer is modelled and simulated using gPROMS, a
general purpose modelling package Validation of the model is performed against
industrial data A sensitivity analysis of the key operating parameters of pressure,
temperature, steam to carbon ratio, and feed rate are comparable to previous
observations in the published literature Optimisation of the burner profle and the
reactant gas distribution enhanced the balance of the furnace, which will signficantly
improve the operating lifetime of the steam reformer for a given performance
Introduction
Steam reforming is an industrially important chemical reaction process, applied in
several production schemes including ammonia, hydrogen, methanol and oxoalcohols
Hydrocarbons, predominately methane in the current economic climate, and steam
react to form a mixture of hydrogen and carbon oxides, which are major chemical
building blocks. Some production schemes stage the steam reforming across two
reactors, however only the first reactor stage, the primary steam reformer, is
considered in this paper.
Several reactor configurations exist for the primary steam reformer, traditionally a
furnace with suspended catalyst tilled tubes and burners in a terrace, or bottom-, sideor top-fired arrangement. Alternatives to the fired-furnace reforming systems have
been developed, however, ignoring arguments over performance and suitability of the
different reactor designs, Adris et a1 [ 11 report that the top-fired furnace is the “most
popular” arrangement of industrial steam reformers, and hence this arrangement is
studied in this paper.
In this work, using the model reported by Alatiqi et al. [2] as a basis, various
improvements have been implemented to the model. The main improvement is the
extension of the heat flux profile application, from the assumption of one tube profile
representing all the tubes in the reformer to a zone approach. Due to the nature of the
arrangement of top-fired furnaces, (see Figure I), tubes are situated in one of two
locations, either in-between rows of burners or a row of burners and a refractory wall,
* Authorfor correspondence (e-mail
I M Mujtaba@bradford ac uk)
77
A J Durn, J Yustos and LM. Mujraba
hence two zones. Polynomials are developed for the differing zone heat flux profiles
using the same method as Alatiqi et al. [2]. Model validation is performed against
industrial data from a steam reformer within an ammonia process
The general purpose modelling package of gPROMS is used in this work, which
includes several features for simulation such as object-orientation, modularisation and
open-interfaces. A sensitivity analysis is performed by varying the key operating
parameters of temperature, pressure, steam to carbon ratio, and tube wall temperature.
Finally, optimisation of the burner profile and the reactant gas distribution for the two
zones is carried out
Figure 1. Layout of a top-fired primary reformer
Reformer Model
Reformer tube
Within the reformer tube, the only chemical reactions considered are,
CH4 + H 2 0 CO + 3 H2
(methane steam reforming)
CO + H2O CO2 + H2
(water-gas shift)
(methane steam reforming)
CHI + 2H20@ COz+ 4 HZ
For higher hydrocarbons, a ‘methane equivalent’ approach is adopted, as proposed by
Hyman [7] Carbon formation and removal reactions are not considered in the work.
The kinetic expressions developed by Xu and Froment [ 141 are adopted and radial
variation is neglected for all variables. The internal heat transfer coefficient and the
pressure drop expressions of Beek [4] and Robbins [ 1 I], respectively, are modified to
take into account the shape of modern catalysts The orifice plates, on the reactant gas
feed, are established from a general energy balance, assuming isothermal flow of an
ideal gas
78
Modelling and Simulation of a Top-Fired Primary Sieam Reformer
Physical Properties
The polynomial expressions for the physical properties of the gaseous mixture in the
reformer, including heat capacity, viscosity and conductivity, are taken from Yustos
[15] The heat capacity of the mixture is calculated as an average considering the
molar fraction of each component, as recommended by Alatiqi et al. [2] and Beaton
and Hewitt [3] However, for the viscosity and the conductivity, averaged expressions
are not considered suitable. For the gaseous viscosity, Reid et al. [ 101 recommends an
expression by Wilke.
where
,#,rk =
[I 2J5
(%,””]’/[
+(
I)$(+
8( 1
05
and for the gas conductivity, the Wassiljewa equation is recommended:
with the Mason and Saxena approximation. Aik = +ik. In the above equations, Aik is
the mixture conductivity parameter; M is the molar mass [kg/mol]; x is the molar
fraction; (bik is the Wilke’s parameter for mixture viscosity; h is the conductivity
[W/m.K]; h,is the conductivity of gas mixture [W/m.K]; p is the viscosity [Pas]; and
p,,, is the viscosity of gas mixture [Pa.s]. The subscript, i, refers to a component, and
NC is the number of components.
Furnace model
Current reactor simulation capabilities have now extended way beyond the basic onedimensional models entrapped by assumptions, to fully integrated three-dimensional
models such as Detemmerman and Froment [ 5 ] . This enhancement encouraged not
only an increased usage of reactor simulation but also more widespread applications.
However, with the simulation improvements, the need to match the level of modelling
complexity to the application becomes more important.
For previous steam reformer simulations the furnace modelling varies in
complexity, from an assumed tube wall temperature profile, implemented by Ravi et
al. [9], to a coupled ‘zoning’ model, proposed by Plehiers and Froment [8] Although
the solution time and complexity of the ‘zoning’ approach are not justifiable in studies
less concerned about accurate furnace modelling, the assumption that the external
tube wall temperature profile is independent of process changes in the reformer tubes
is flawed. As a compromise to these approaches, Alatiqi et al. [2] proposed a
polynomial representation of the heat flux profile:
q=F,SG(a+bz+cz* + d i 3 +ez4)1
(3)
79
A J Dunn, J Yustos and 1.M Mujtaba
where, a,b,c,d,e are polynomial coefficients; FG is the he1 gas flow rate [m3/s]; q is
the heat flux [W/m*]; SG is the specific gravity [kg/m3], z is the axial position in the
tube [m];and q is the unit efficiency.
However, in a top-fired steam reformer, the heat flux profile is not equivalent for
all the reformer tubes, as the tubes are either situated in-between rows of burners or a
row of burners and a refractory wall, hence two zones. These zones are distinct, as the
refractory wall affects the heat transfer performance, although the actual variation in
external tube wall temperature is limited by furnace strategy. Roesler [ 121 argues that
the fbrnace should be sectioned on a radiation field basis, instead of the geometrical
basis of Hottel and Sarofim [6] which was implemented by Plehiers and Froment [8].
The "two-zones'' theory complements both reasoning, to some extent, and offers an
enhancement to the proposal of Alatiqi et al. [2].
Numerical Solution
The full model of the primary reformer results in a system of differential and
algebraic equations, consisting of sixty-six equations for each zone. A second order
orthogonal collocation finite element method with 10 discretisation points was
applied for the solution technique. The gPROMS package, used in this work offers
several features for simulation including a range of numerical solution methods for
discretisation, parameter estimation, optimisation, and an open architecture for real
time applications. For the reported simulation, the most important capability of
gPROMS is the ability to manage distributed systems. Previously, gPROMS has been
applied to several industrial examples [ 131. These capabilities are not only appropriate
for the reported primary steam reformer model, but also for further developments.
Simulation and Model Validation
The model was validated against industrial data for a top-fired primary reformer
within the ammonia process. The reformer comprises 8 rows of 44 tubes of 12 m in
length. The reformer operates at an inlet pressure and temperature of 35.94 bar and
455"C, respectively. The simulation results are presented in Table 1, which clearly
shows that the simulated data compared well to the industrial data.
Table 1. Comparison between plant and simulated data
Outlet temperature (K)
Outlet pressure (Bar)
CH4
Dry
mol%
co
co2
H2
Inerts
Approach to equilibrium (K)
80
Plant data
1084
33
8.10
10.39
10.96
70.28
0 270
Calculated values
1091
33.54
8.15
10.36
11.17
70.05
0.270
Reaction I
Reaction II
Reaction 111
Deviation (%)
0.66
1.63
0.62
-0.29
1.92
-0.33
0.04
4.9
0
5.5
Modelling and Simulation of a Top-Fired Primary Steam Reformer
The difference between the process gas temperature and the temperature if
equilibrium would have been achieved, or more commonly known as the approach to
equilibrium, demonstrates the proximity of the equilibrium limit, see Table 1. Hence,
further validation is required, so the heat flux profile was compared to that of an
industrial simulation package (see Figure 2) Overall, the profiles compare well
1
X
a
li
08
u
r3 0 6
04
02
0
5
0
10
15
Tube Length [m]
-Simulation
result
-
-Industrial
package
Figure 2. Comparison of heat flux profiles with an industrial simulation package
Sensitivity Analysis
For most applications of steam reforming, the aim of a steam reformer is primarily to
reduce the methane concentration, and hence yield the required composition for
downstream processing. However, the outlet of a primary steam reformer is close to
chemical equilibrium, so the sensitivity of the composition is affected The
consequences of which are observable for the inlet pressure (see Figure 3) and steam
to carbon ratio (see Figure 4).
The choice of material of construction and design parameters for the reformer tube
are not only constrained by the maximum heat flux but also the maximum tube
temperature, hence a suitable range of operating conditions are required. Although the
effect on the reformer exit gas temperature is small with increased inlet temperature
and pressure, along the tube length more significant changes occur and this is
reflected in the tube wall temperature profile (see Figure 5).
The results of the sensitivity analysis presented in this work suffer fiom no strange
quirks or anomalies, and are very similar to those reported by Alatiqi et a1 [2].
A J Dunn, I Yustos and 1.M Mujtaba
-*-*-++
*--x
0 07
0 065
-
0 06
C
0
-
-
-
-
-
I
0055
I;"
005
al
5 0 045
z
004
0 035
...-..CH4
I
20
-x-co
-
35
40
Inlet Pressure [Bar]
25
30
c02
I
45
50
Figure 3. Influence of inlet pressure on outlet mole fractions of CH,CO and C02
009
,
008 -
Y
.g 007 0
I;" 006-
\
c
2 005Q,
004
25
I-..-.
3
-- x
c02
CH4 -X35
4
45
5
Steam to Carbon Ratio
Figure 4. Influence of inlet steam to carbon ratio on outlet molefractions
of C H , CO and C02
82
Modelling and Simulation of a Top-Fired Primary Steam Reformer
1200
1100
E
P
3
c
Fi
E
1000
I-
-
2
8
n
z
900
800
0
1
2
3
4
5
6
7
8
9
1
0
1
1
1
2
1
3
Tubelength [rn]
Inlet Pressure = 25 bar
-Inlet
Pressure = 35 94 bar
-I-
- - -x - - Inlet Pressure = 45 bar
+Inlet Pressure = 30 bar
- - + - - Inlet Pressure = 40 bar
Figure 5. influence of inlet pressure on the average tube wall temperature profile
Optimisation of Furnace Operation
The operational lifetime of a steam reformer tube is fundamentally dependent on the
maximum temperature experienced Ideally, both for ease of maintenance and to
avoid operating the furnace with reduced capacity, the fbmace should have a
consistent tube lifetime or equivalently a uniform temperature profile (see Figure 6).
For top-fired steam reformers, the spread of temperature profiles can be limited by
adjustment of the burner profile or the reactant gas distribution; this technique is
commonly called furnace balancing. For this purpose two optimisation problem
formulations are presented
(a) Furnace balancing via adjustment of the burner profile. The optimisation problem
can be expressed as:
83
A J Dunn, J Yustos and ISM Mujtaba
Minimize Drflerence(u) = (‘Outer Zone’ Tube Wall Temperature
- ‘Inner Zone’ Tube Wall Temperature)’
where u = {Ratio of Fuel Gas Flowrate between Zones} ; subject to 0.5 < u < 1.5
‘Outer Zone’ Outlet Pressure = ‘Inner Zone’ Outlet Pressure
(b) Furnace balancing via adjustment of the reactant gas distribution The
optimisation problem can be expressed as,
Minimize Dzflerence(u) = (‘Outer Zone’ Tube Wall Temperature
- ‘Inner Zone’ Tube Wall Temperature)’
-=
where u = (Orifice Plate Diameter), subject to 0 5 < u 2.3
‘Outer Zone’ Outlet Pressure = ‘Inner Zone’ Outlet Pressure
Orifice Plate Diameter < Internal Pipe Diameter
For optimisation, gPROMS employs the method of successive reduced quadratic
programming. Initialisation from several locations yielded same result.
Implementing optimisation for the burner profile and the reactant gas distribution
between the two zones, reduced the difference in the temperature profiles (see Figures
7 and 8). Both methods enhanced the ‘balance’ of the furnace, however adjustment of
the reactant gas distribution offers a smoother influence on the overall tube profile in
comparison to the burner profile
t
Tube Leneth
Figure 6. ideal tube wall temperature profile
84
Modelling and Simulation of a Top-Fired Primary Steam Reformer
1200
1100
-
E.
2
3
4-
g
-
1000
I-
2
01
0
3
900
800
0
1
2
3
4
5
6
7
8
9
1 0 1 1 1 2 1 3
Tubelength [m]
-..--.
--
-
Optimised Inner Tube Rows
Optimised Outer Tube Rows
Original Inner Tube Rows
-Original
Outer Tube Rows
Figure 7. Comparison of inner and outer zones’ tube wall temperature profiles
Furnace balancing via adjustment of the burner profile Optimal ratio of fuel gas
jlowrate between zones = I 05 CPU time = 130 seconds
Although the optimisation results from the zone approach are logical, it should be
noted that the implications of the operational changes, particularly the effect on the
flow field in the furnace, are not fully portrayed. To accurately display these effects a
full flow simulation of the h a c e would be required. However, from first principles,
adjustment of the burner profile would seem have a more significant effect on the
furnace flow-field.
85
A J Dunn, J Yustos and I.M Mujtaba
1200
1100
E:
e!
3
u
Eal
!I-"t 1000
2
al
n
I-'
900
800
0
1
2
3
4
5
6
7
8
9 1 0 1 1 1 2 1 3
Tubelength [m]
--..---
-
Optimised Inner Tube Rows
Optirnised Outer Tube Rows
Original Inner Tube Rows
-Original
Outer Tube Rows
Figure 8. Comparison of inner and outer zones' tube wall temperature profiles
Furnace balancing via adjustment of the reactant gas distribution Optimal or$ce
plate diameter ratio = I 3 CPU time = 390 seconds
During the design phase, it is industrial practice for this style of top-fued steam
reformer to perform balancing via reactant gas distribution, to limit the effect of flow
field issues, and to simplify the furnace balancing. Over the operational life of the
catalyst charge disturbances will occur, and unequivocally the performance of the
reactor will change. However, adjustment of the orifice plates is an unrealistic
operational practice so furnace balancing via adjustment of the burner profile is
performed. Ultimately, using both balancing techniques reduces the extent of
balancing via the burner profile
86
Modelling and Simulation of a Top-Fired Primary Steam Reformer
Conclusions
Modelling and simulation of a top-fired primary reformer, allowing for the difference
in location of reformer tubes, has been carried out using gPROMS. The model
demonstrated close agreement to industrial plant data A sensitivity analysis of the
key operating parameters of outlet composition, heat flux, and pressure drop was
completed The observations are comparable to those in the published literature
Optimisation of the burner profile and reactant gas distribution, between the zones,
reduced the difference in the tube wall temperature profiles. This is a significant
operational condition overlooked when defining an average reformer tube
Acknowledgements
Financial support fkom EPSRC and technical support from Process Systems
Enterprise, London, and Synetix, Billingham, are gratefully acknowledged.
References
10
11
12
13
14
IS
Adris, A M , Pruden, B B , Lim, C J , and Grace, J R 1996 On the Reported Attempts to Radically
Improve the Performance of the Steam Methane Reforming Reactor, Can J Chem, Eng ,74, 177-186
Alatiqi, M A , Meziou, A M , and Gasmelseed, G A 1989 Modelling, Simulation and Sensitivity
Analysis of Steam Methane Reformers, Int J Hydro Energy, 14,24 1-256
Beaton, C F and Hewitt, G F 1989 Physical Property Data for the Design Engineer, Hemisphere
Publishing Corporation, New York
Beek, J 1962 Advances in Chemical Engineering 3, Academic Press, New York
Detemmennan, T and Froment, F 1998 Three Dimensional Coupled Simulation of Furnaces and
Reactor Tubes for the Thermal Cracking of Hydrocarbons, Rev de L’Instuit Francais ou Petrole, 53,
181- 194
Hottel, H C and Sarofim, A F 1967 Radiative Transfer, Mc Graw-Hill, New York
Hyman, M H 1968 Simulate Methane Reformer Reactions, Hydrocarbon Process, 47, 131-137
Plehiers, P M and Froment, G F 1989 Coupled Simulation of Heat Transfer and Reaction in a Steam
Reforming Furnace, Chern Eng Tech, I2,20-26
Ravi, K , Joshi, Y K , Dhingra, S C and Guha, B K 1989 Simulation of Primary and Secondary
Reformers for Improved Energy Performance of an Ammonia Plant, Chem Eng Tech , 12,358-364
Reid, R C , Prausnitz, J M and Shenvood, T K 1977 The Properties of Gases and Liquids, 3 ed
McGraw-Hill, New York
Robbins, L A 1991 Improve Pressure-Drop Prediction with a New Correlation Chem Eng Prog ,87,
87-91
Roesler, F C 1967 Theory of Radiative Heat Transfer in Co-current Tube Furnaces, Chem Eng Sci ,
22, 1325-1336
Winkel, M L , Zullo, L C , Verheijen, P J T , and Pantelides, C C 1995 Modelling and Simulation of
the Operation of an Industrial Batch Plant Using gPROMS, Comput Chem Eng , 19, SS714576
Xu, J and Froment, G F 1989 Methane Steam Reforming, Methanation and Water-Gas Shift 1
Intrinsic Kinetics, AlChE J ,35(1), 88-96
Yustos, J 1999 Modelling and Simulation of a Primary Steam Reformer Using gPROMS, Internal
Report, University of Bradford, Bradford, UK
Received 6 December 2000; Accepted aper revision 2 April 2001.
87
Документ
Категория
Без категории
Просмотров
0
Размер файла
525 Кб
Теги
steam, gproms, using, simulation, top, reformen, modellina, primary, fire
1/--страниц
Пожаловаться на содержимое документа