You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269099314

Predicting catalyst lifetime

Article  in  Petroleum Technology Quarterly · January 2013

CITATIONS READS

0 2,715

1 author:

Sepehr Sadighi
Research Institute of Petroleum Industry (RIPI)
174 PUBLICATIONS   391 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Computational Fluid Dynamics View project

Adsorption Processes View project

All content following this page was uploaded by Sepehr Sadighi on 15 March 2017.

The user has requested enhancement of the downloaded file.


Predicting catalyst lifetime

A simulation model compares accurately with actual operating data to


provide reliable predictions of catalyst lifetime in naphtha reforming

S REZA SEIF MOHADDECY and SEPEHR SADIGHI


RIPI

T
he catalytic reforming of active phase sintering and present in the oil, both of
heavy naphtha (heavy poisoning during the catalyst’s which can change during a
straight-run gasoline, or life cycle. Consequently, the typical commercial run.
HSRG) is a favoured process in main problem with fixed-bed Moreover, depending on the
petroleum refineries for reactors is the loss of catalyst process used, the catalyst’s life
producing high-octane gaso- activity over time, which cycle may vary from a few
line. From the viewpoint of reduces the length of continu- seconds, as in fluid catalytic
process operation, there are ous operation. cracking (FCC), to several
three kinds of widely used The catalyst’s life cycle is years, as in ammonia synthesis;
catalytic reforming units: further complicated by numer- therefore, a reliable prediction
semi-regenerative, continuous ous technical, environmental of catalyst life has been
catalyst regeneration (CCR) and organisational issues. In extremely difficult.
and cyclic regenerative principle, different companies Before the catalyst is loaded
processes. Semi-regenerative can be involved in each of the in the reactor, many laboratory
catalytic naphtha reforming is life cycle steps. The life cycle of experiments are carried out to
the oldest type and it is gener- a catalyst starts with the initial identify its operability. An
ally carried out in three or four production of fresh catalyst accelerated method for a
fixed-bed reactors in series oxide, which is pre-sulphided one-day laboratory test is
with intermediate preheaters prior to being used in the refin- proposed here for the progno-
operated adiabatically at ery process. During its use sis of activities and an
temperatures between 450°C when the catalyst deactivates estimation of the lifetime of the
and 520°C, total pressure and it does not meet perfor- catalyst. The method is based
between 10 and 35 bar, and mance targets within the limits on knowledge of the deactiva-
molar hydrogen-to-hydrocar- of the reactor’s operating tion behaviour of catalysts
bon ratios between 3 and 8. conditions, the reactor is shut under different reaction condi-
Among all commercial technol- down. Therefore, the catalyst’s tions, including extreme
ogies for the catalytic life cycle typically involves a conditions at high liquid hourly
reforming process, those using long chain of operations, space velocities (LHSV). Hence,
fixed-bed reactors in series are normally performed by differ- the method enables prediction
the most common configura- ent specialised companies.1,2 of lifetime and performance
tions due to the relative Catalyst life is dependent using data obtained from a
simplicity of scaling up an essentially on process condi- one-day test in laboratory
operation. However, the main tions and also, to a large extent, conditions. The curve of LHSV
disadvantages of fixed-bed on the distribution of species versus lifetime is an individual
reactors are coke formation, such as sulphur and nitrogen property of a given catalyst,

www.digitalrefining.com/article/1000840 PTQ Q3 2013 1


Recycle
R

C-1
Vapour
Purge
R-4
R-3 Gas to
R-2
R-1 LPG

Platcharge V-1
H-1 H-2 H-3 H-4 LPG

Stabiliser
E-1 E-2 E-3 column
Air cooler Water cooler

Reformate

Figure 1 Block flow diagram of the catalytic reforming unit of the target oil refinery

which has to be established Process description reformer is regenerated during


experimentally. Any point from A commercial fixed-bed cata- routine shutdowns of the
this curve can be used for an lytic naphtha reforming unit process every 18-24 months.
estimation of lifetime.3,4 Since called Platformer, licensed by Normally, the catalyst can be
deactivation of the catalyst Chevron and with a nominal regenerated three or four times.
takes place over a long time capacity of 16 500 b/d, was Then, it must be returned to
scale, it is difficult to mimic this chosen as a case study. The the manufacturer for reclama-
in a pilot plant or at lab scale. feed to the plant, prior to enter- tion of platinum and/or
Consequently, any study on an ing the catalytic reformer, rhenium.
industrial scale to predict the should undergo hydrodesul- In Figure 1, Platcharge is
life cycle of a catalyst plays an phurisation (HDS) in the preheated by the first furnace
important role regarding hydrotreatment unit. Then, the (H-1). It then enters the first
economic, operational and produced naphtha, called reactor (R-1), where naph-
environmental issues.5,6 Platcharge, is introduced to the thenes are dehydrogenated to
Works on the life cycle of an catalytic reforming process. aromatics. Next, the product
industrial-scale catalytic naph- The most commonly used stream from the first reactor
tha reforming unit are scarce. types of catalytic reforming passes through the second
In our present work, the units have three or four reac- reactor (R-2), and the outlet
semi-regenerative catalytic tors, wherein each has a fixed stream of that enters the third
reforming process of a catalytic bed. For such a unit, reactor (R-3). Similarly, the
commercial-scale oil refinery the activity of the catalyst product stream from the third
was simulated using the Petro- reduces during operation due reactor enters the fourth reac-
Sim simulator (KBC to deposited coke and loss of tor (R-4). The overall reforming
Profimatics, 2009). After vali- chloride. Hopefully, the activ- reactions are endothermic, so a
dating the simulation, the ity of the catalyst can be preheater (H-1, H-2, H-3 and
lifetime of the next cycle was periodically regenerated or H-4) is provided ahead of each
predicted by the previous restored using in situ high- reforming reactor.
simulated cycle lives, and the temperature oxidation of the Next, the product stream
results were compared with coke followed by a chlorination from the fourth reactor enters a
actual test runs of the plant at process. Therefore, the catalyst separator, V-1, in which hydro-
the end of the cycle. of a semi-regenerative catalytic gen produced during the

2 PTQ Q3 2013 www.digitalrefining.com/article/1000840


reforming process (the gas
Catalyst distribution in reforming reactors
stream) is recycled, then mixed
with Platcharge. Finally, the
1st reactor 2nd reactor 3rd reactor 4th reactor
liquid product leaving the Catalyst weight, kg 5077.25 7615.87 12 693.13 25 386.25
separator is introduced to the Catalyst distribution, wt% 10 15 25 50
gasoline stabiliser, in which
LPG and light gases are sepa- Table 1
rated from gasoline. The
vapour pressure of the gasoline gathered. These parameters
can be set according to market Operating conditions in the mainly consist of inlet and
catalytic reforming of the target
requirements. outlet reactor temperatures,
oil refinery
The distribution of catalyst in catalyst weight, specification of
the reactors is shown in Table Process variable Value feed and products, operating
1. The normal operating condi- Inlet temperature, °C 490-515 pressures, flow rates of makeup
Hydrogen/hydrocarbon ratio, mol/mol 3-7
tions of the unit are shown in LHSV, h-1 1-2
hydrogen, and recycle rates and
Table 2. Yield, vol% 70-85 flow rates of all gaseous and
liquid products.14 After prepar-
Process simulation Table 2 ing a test run and running
and validation Ref-Sim, tuning parameters are
Catalytic reforming is often Petro-Sim, external calibration sent to the Petro-Sim environ-
modelled and simulated based for determining the kinetic ment to simulate and optimise
on the number of reactive parameters is needed. For this the operating conditions.
species and the type of kinetic unit, the Ref-Sim module is Moreover, the life of the catalyst
model used.7,8 The presence of provided for the task. For can be predicted in the Petro-
many components, either as generation of the required Sim environment. It is obvious
reactants or as intermediate parameters, such as frequency that after simulating catalytic
products in the reactive factors, activation energies and naphtha reforming (see Figure
mixture, with resulting new decay constants, by Ref-Sim, 2), the simulated variables
reactions, creates challenges for actual test runs from the unit should be compared against
modelling the process. To under study should be actual data to validate the
reduce these complications,
reactants in the mixture are
classified in certain and limited CRU Gas
(R-1 to R-4) H2 product
to LPG
groups called pseudo- compo-
nents. Additionally, Arhenius
and Langmuir-Hinshelwood
kinetics are widely used for Liquid
kinetic-based catalyst model- Platcharge to LPG
ling and simulation of the Stabiliser
catalytic naphtha reforming column

process.9-13 Petro-Sim software


is a simulator capable of simu-
lating commercial-scale
catalytic reforming units,15 and
so enables us to simulate reac- R
Furnace
tors with different catalyst
weights and sizes.16,17 In this
research, Petro-Sim has been
Reformate
used to simulate and analyse a (gasoline)
catalytic reforming unit. Air cooler Water cooler
To simulate a catalytic naph-
tha reforming process using Figure 2 Simulation of catalytic naphtha reforming in the target refinery

www.digitalrefining.com/article/1000840 PTQ Q3 2013 3


octane. To represent this
Comparison of outlet temperature of first reactor for previous cycles (first,
second and third cycles) of catalytic reforming of the target oil refinery behaviour, Petro-Sim tracks
catalyst decay by predicting
Variable Unit Test run 1 Test run 2 Test run 3 AAD% decay factors through each
Simulation Actual Simulation Actual Simulation Actual reactor. This calculated decay
1st cycle °C 441.8 446 441 447 440 449 1.44
2nd cycle °C 466 447 471.5 440 473.2 448 5.68 information forms part of the
3rd cycle °C 452.1 451 446 430 458.1 431 3.42 model and is available on the

* Absolute average deviation per cent calibration factor tab for
predicting the life cycle of the
Table 3 catalyst.

Comparison of outlet temperature of second reactor for previous cycles (first, Results and discussion
second and third cycles) of catalytic reforming of the target oil refinery From the start of run to the
end of run, actual test runs
Variable Unit Test run 1 Test run 2 Test run 3 AAD%
were gathered from the target
Actual Simulation Actual Simulation Actual Simulation reforming unit. Since, for a
1st cycle ° C 486.3 487.2 488 486 489 483.6 0.57 specified catalyst life, the simu-
2nd cycle ° C 488 488.2 482 486.5 493.5 488 0.64
3rd cycle ° C 497 466.8 465 474.1 468 474.5 3.14 lator was used for predicting
the steady-state condition of
Table 4 the Platformer, the data used
for calibration should be
Comparison of outlet temperature of third reactor for previous cycles (first, selected from the normal
second and third cycles) of catalytic reforming of the target oil refinery condition when no
abnormalities, such as tower
Variable Unit Test run 1 Test run 2 Test run 3 AAD% flooding, emergency depressur-
Actual Simulation Actual Simulation Actual Simulation
1st cycle ° C 500 510.8 502 509.8 502.5 507.8 1.59 isation and pump or
2nd cycle ° C 502 503.4 498 502 508 505.5 0.54 compressor shutdown, were
3rd cycle ° C 511 483 480 496.9 481 490.5 3.66
happening during the opera-
tion. Before using these data to
Table 5 estimate the tuning parameters,
it was necessary to validate
Comparison of outlet temperature of fourth reactor for previous cycles (first, them. If a reasonable overall
second and third cycles) of catalytic reforming of the target oil refinery mass balance (±5%) cannot be
calculated, the validity of the
Variable Unit Test run 1 Test run 2 Test run 3 ADD% test run is compromised. After
Actual Simulation Actual Simulation Actual Simulation validating the test runs, gener-
1st cycle ° C 509.9 510 511.8 509 511.6 507.1 0.48
2nd cycle ° C 511.8 509.4 507.5 509.4 517.8 511.5 0.69 ating the calibration factors
3rd cycle ° C 520.1 494 490.2 503.3 493 499.7 3.02 and sending them to the Petro-
Sim environment, the
Table 6 comparison of the simulated
outlet temperatures of reform-
process simulation; otherwise, Petro-Sim environment is ing reactors for the first, second
the accuracy of the predictions predominantly a function of and third cycles are shown in
can be compromised. reformer type and severity of Tables 3 to 6. From these
Since the catalyst in catalytic operation. The latter is deter- tables, it can be concluded that
reforming loses activity over mined by the feed — the Petro-Sim can simulate reactor
time, predicting the life cycle of percentage amounts of naph- temperatures with an absolute
a commercial catalyst must thenic, aromatic and paraffinic average deviation percentage
take into account catalyst hydrocarbons — therefore, (AAD%) of less than 4%.
production, sulphidation, during operation, reactor To gain better insight, the
usage and oxidative regenera- temperatures should be raised most important process varia-
tion. The rate of decay in the to achieve the same reformate bles of Platformer — RON of

4 PTQ Q3 2013 www.digitalrefining.com/article/1000840


product, yield of gasoline and
Comparison of product RON for previous cycles (first, second and third
life of catalyst — were also
cycles) of catalytic reforming of the target oil refinery
simulated. The AAD% of each
of the simulation results (for
Variable Unit Test run 1 Test run 2 Test run 3 AAD%
the first, second and third Actual Simulation Actual Simulation Actual Simulation
cycles) is shown in Tables 7, 8 1st cycle - 97.3 97.47 97.2 97.25 96.6 96.9 0.18
2nd cycle - 94.2 94.55 92.2 92.22 93.9 92.7 0.29
and 9. It is supposed that the 3rd cycle - 92.3 92.91 95.5 94.9 93.6 93.8 0.08
main source of deviation was
the possibility of errors arising Table 7
from measurements in gather-
ing data obtained from faults Comparison of product yield for previous cycles (first, second and third
such as signal transmission, cycles) of catalytic reforming of the target oil refinery
calibration and power fluctua-
tion of instruments, which Variable Unit Test run 1 Test run 2 Test run 3 AAD%
Actual Simulation Actual Simulation Actual Simulation
could not be excluded from the 1st cycle vol% 75.8 74.33 75.5 74.8 74 75.3 1.5
actual data. However, from the 2nd cycle vol% 80.2 76.2 80.4 79.8 81.1 79.1 2.8
presented simulation results, it 3rd cycle vol% 79.5 79.22 80.2 76.2 78.8 77.9 2.25

can be concluded that the


developed simulation program Table 8
was reliable enough to be
applied for predicting the Comparison of total life time prediction for previous cycles (first, second
behaviour of the catalytic and third cycles) of catalytic reforming of the target oil refinery
reforming unit under study.
After evaluating and vali- Variable Unit Test run 1 Test run 2 Test run 3 AAD%
dating the simulation, the Actual Simulation Actual Simulation Actual Simulation
1st cycle days 920 895.6 920 889.6 920 904 2.63
process variables were 2nd cycle days 891 859.6 891 820.2 891 856 5.45
predicted for the fourth cycle. 3rd cycle days 591 568.7 591 569.2 591 600 3.08
After starting up the process,
predicted results from the Table 9
simulator were compared
against actual results versus
days on stream. Figures 3 to 8 460
Actual
show comparisons between
Simulation
the reactor temperatures, RON 455
and yields of products against
Temperature, ºC

the actual values. As these


450
figures show, there are close
mappings between the meas-
ured and the predicted process 445
variables. It should be
mentioned that the AAD% of 440
predictions for the first,
second, third and fourth outlet
reactor temperatures were 435
0 100 200 300 400 500 600
0.75, 0.58, 0.54 and 0.61, Days on stream
respectively. Moreover, the
AAD% of the prediction for Figure 3 Comparison of outlet temperature of first reactor for the current cycle
product RON and product
volume yield were 1.45 and unit — the life of the next the next cycle, the life of the
0.03, respectively. cycle — was also studied by cycle (end of run) was
The most important parame- the validated simulation predicted by Petro-Sim for
ter of a catalytic reforming program. Before starting up different days on stream (see

www.digitalrefining.com/article/1000840 PTQ Q3 2013 5


Conclusion
490 Prediction of the process varia-
Actual bles and life of the catalyst in a
Simulation commercial process is vital. To
485
perform such a task, a catalytic
Temperature, ºC

naphtha reforming unit with a


480 nominal capacity of 16 500 b/d
was studied. Data gathered
475 from the first three cycles were
chosen to calibrate the simula-
tor, then the plant was
470
simulated using the Petro-Sim
process simulator. The signifi-
465 cant process variables — outlet
0 100 200 300 400 500 600 temperatures of the first,
Days on stream second, third and fourth reac-
tors, product RON and volume
Figure 4 Comparison of outlet temperature of second reactor for the current yield — were simulated. After
cycle comparing results against
actual data, the accuracy of the
505
simulator was validated due to
Actual the low deviation of the simu-
Simulation lated process variables.
500 In the next step, the simula-
Temperature, ºC

tor was used to predict the


495 process variables correspond-
ing to the next cycle.
Comparing the results with the
490
real outputs showed that the
per cent of absolute average
485 deviation (AAD%) of the main
parameters were about 0.75%,
480 0.58%, 0.54%, 0.61%, 1.45%
0 100 200 300 400 500 600 and 0.03%, respectively.
Days on stream Additionally, before starting
the cycle, the simulated life of
Figure 5 Comparison of outlet temperature of third reactor for the current cycle the catalyst was estimated at
about 679 days and recorded at
Table 10). From this table, it As was expected, on day 719, about 719 days at the end of
can be seen that the average the temperature of the reactor the cycle under study.
life of the catalyst for the needed to be set at values
fourth cycle is about 679 days. higher than 510°C to preserve References
To validate this result, after the RON; this was catastrophic 1 Eijsbouts S, A A, van Leerdam G C, Life
starting up the fourth cycle, for the catalyst. As a result, cycle of hydroprocessing catalysts and
the operation of Platformer the control room shut down total catalyst management, Catalysis
Today, 2008, 130, 361-373.
was accurately monitored. the unit to decide whether to
2 Girardier F, Meerstadt R P, Oude
After 710 days, it was found regenerate or replace the cata-
Groeniger H, Docter S, Akzo Nobel
that the control room had to lyst. By using the developed Catalyst Symposium, 1998.
raise the inlet temperatures of simulation program, the life 3 Petrov L, Vladov Ch, Neshev N, Bonev
the reactors to around 510°C cycle of the reforming unit can Ch, Prahov L, Kirkov N, Vasileva M, Filkova
to keep the RON of the prod- be estimated with acceptable D, Dancheva S, Method for Forecasting
uct at the desired value (>94). accuracy. Lifetime of Industrial Catalysts for

6 PTQ Q3 2013 www.digitalrefining.com/article/1000840


Hydrogenation, Bulgarian Patent 41960,
1986. 510
Actual
4 Petrov L, Vladov Ch, Bonev Ch, Prahov
Simulation
L, Kirkov N, Eliyas A, Neshev N, Filkova D, 505
Dancheva S, Prognosis of the lifetime of

Temperature, ºC
industrial hydrogenation and oxidation
catalysts, EuropaCat-2, Book of Abstracts, 500
Maastricht, The Netherlands, 1995, 646.
5 Moulijn J A, Van Diepen A E, Kapteijn F, 495
Catalyst deactivation: is it predictable?
What to do?, Applied Catalysis A: General,
490
2001, 212, 3-16.
6 Parrott S L, Adarme O, Lin F, Catalyst
Life Prediction in Hydrodesulphurization, 485
0 100 200 300 400 500 600
US Patent 5341313, 1994.
Days on stream
7 Shakoor Z M, Catalytic reforming of
heavy naphtha, analysis and simulation,
Diyala Journal of Engineering Sciences, Figure 6 Comparison of outlet temperature of fourth reactor for the current cycle
2011, 4, No. 02.
8 Arani H M, Shirvani M, Safdarian K,
Dorostkar E, Lumping procedure for 100
Actual
a kinetic model of catalytic naphtha
Simulation
reforming, Brazilian Journal of Chemical 98
Engineering, 2009, 26(4), 723-732.
9 Mohadesi M, Mousavi H S, Kinetic
96
investigation of catalyst deactivation in
RON

catalytic reforming of naphtha, Chemical


Product and Process Modeling, 2012, 7. 94
10 Saidi M, Mostoufi N,
SotudehGharebagh R, Modeling and 92
simulation of continuous catalytic
regeneration (CCR) process, International 90
Journal of Applied Engineering Research, 0 100 200 300 400 500 600
Dindigul, 2011, 2(1). Days on stream
11 Arani H M, Shokri S, Shirvani M,Dynamic
modeling and simulation of catalytic
Figure 7 Comparison of product RON for the current cycle
naphtha reforming, International Journal
of Chemical Engineering and Applications,
2010, 1(2).
12 Hou Weifeng, Hongye Su, Yongyou
Hu, Jian Chu, Modeling, simulation 84
Actual
and optimization of a whole industrial 82 Simulation
catalytic naphtha reforming process on
Volume yield, %

Aspen Plus Platform, Chinese Journal of 80


Chemical Engineering, 2006, 14(5), 584-
78
591.
13 Rahimpour M R, Esmaili S, Bagheri 76
Yazdi S A, A kinetic and deactivation
model for industrial catalytic naphtha 74
reforming, Iranian Journal of Science and
72
Technology, 2003.
14 Liang Ke-min, Guo Hai-yan, Pan 70
Shie-wei, A study on naphtha catalytic 0 100 200 300 400 500 600
reforming reactor simulation and Days on stream
analysis, Journal of Zhejiang University
Science, 2005, 6B(6), 590-596.
15 Petro-Sim User Guide, KBC Advanced Figure 8 Comparison of product volume yield for the current cycle

www.digitalrefining.com/article/1000840 PTQ Q3 2013 7


Technologist, KBC Profimatic, 2005. Reaction Engineering Department, in chemical engineering from Universiti
16 Mohaddecy S R, Sadighi S, Ghabouli Research Institute of Petroleum Industry Teknologi Malaysia.
O, Lumping kinetic model and simulation (RIPI), Tehran, Iran. He holds a MS Email: Sadighis @ripi.ir
of catalytic naphtha reforming process, in chemical engineering from Sharif
Chemical Technology: An Indian Journal, University of Technology.
6(4), 2011. Email: Seifsr @ripi.ir LINKS
17 Mohaddecy S R, Sadighi S, Optimising
a reactor’s catalyst distribution, PTQ, Q3 Sepehr Sadighi is a Project Manager More articles from the following
2008, 101-107. in the Catalysis and Nanotechnology categories:
S Reza Seif Mohaddecy is a Project Division, Catalytic Reaction Engineering Catalysts & Additives
Manager in the Catalysis and Department, Research Institute of Process Modelling & Simulation
Nanotechnology Division, Catalytic Petroleum Industry (RIPI). He holds a PhD Reforming

8 PTQ Q3 2013 www.digitalrefining.com/article/1000840

View publication stats

You might also like