REAL TIME OPTIMIZATION AS A TOOL FOR INCREASING PETROLEUM REFINERIES PROFITS

Abstract

The increasing competition in refinery industries, reducing refinery plant’s costs, minimizingmeasurement errors and environmental issues lead to growing interest in modeling, simulation andoptimization of refineries. Real time optimization (RTO) of the process units is one of the most effectiveways for enhancing economic performance and reducing overhead costs of chemical plants. Thismethod has a fully automated system, which intelligently collects and processes main outputs of theplant. Modifications of plant operating conditions have been implemented in order to reduce costs andmeet constraints. Many objectives can be reached by implementation of RTO in refinery industriessuch as: automatically optimizing plant’s performance, automatically performing fault detection,elimination and modification of random errors (Data reconciliation), modification of nonrandom errors(Gross Error Detection), intelligent computation of data which are not measurable, calculating andreporting consumption of raw materials, products production and energy consumption of the entireplant and the equipments at any time. In this article we studied the RTO implementation profits in oilindustries and the effect of different parameters on the refinery processes performance. The mediumterm plan of the Research Institute of Petroleum Industry (RIPI) on developing RTO technology inIRAN oil industry is presented.

Full text

Petroleum & Coal
ISSN 1337-7027
Available online at www.vurup.sk/pc
Petroleum & Coal 51 (2) 110-114, 2009
REAL TIME OPTIMIZATION AS A TOOL FOR INCREASING
PETROLEUM REFINERIES PROFITS
Saeid Shokri *, Reza Hayati, Mahdi Ahmadi Marvast, Mohammad Ayazi,
Hamid Ganji
Research Institute of Petroleum Industry (RIPI), Tehran 14665-1998, IRAN
Email: shokris@ripi.ir
Received February 2, 2009, accepted April 15, 2009
Abstract
The increasing competition in refinery industries, reducing refinery plant’s costs, minimizing
measurement errors and environmental issues lead to growing interest in modeling, simulation and
optimization of refineries. Real time optimization (RTO) of the process units is one of the most effective
ways for enhancing economic performance and reducing overhead costs of chemical plants. This
method has a fully automated system, which intelligently collects and processes main outputs of the
plant. Modifications of plant operating conditions have been implemented in order to reduce costs and
meet constraints. Many objectives can be reached by implementation of RTO in refinery industries
such as: automatically optimizing plant’s performance, automatically performing fault detection,
elimination and modification of random errors (Data reconciliation), modification of nonrandom errors
(Gross Error Detection), intelligent computation of data which are not measurable, calculating and
reporting consumption of raw materials, products production and energy consumption of the entire
plant and the equipments at any time. In this article we studied the RTO implementation profits in oil
industries and the effect of different parameters on the refinery processes performance. The medium
term plan of the Research Institute of Petroleum Industry (RIPI) on developing RTO technology in
IRAN oil industry is presented.
Keywords: Simulation; Real time optimization; Profit, Refinery.
1. Introduction
Chemical and petrochemical process industries are increasingly compelled to operate
profitably in a very dynamic and global market. The increasing competition in the
international arena and stringent product requirements mean decreasing profit margins
unless plant operations are optimized dynamically to adapt to the changing market
conditions and to reduce the operating cost. Hence, the importance of real-time or on-line
optimization of an entire plant is rapidly increasing. Applying this technology to oil refineries
has the potential to provide competitive benefit for oil refiners.
Real Time optimization (RTO) is an effective approach for economic improvement and source
reduction in chemical and petrochemical plants. Real Time optimization uses an automated
system which adjusts the operation of a plant based on product scheduling and production
control to maximize profit and minimize emissions by providing optimal set points to the
distributed control system. This optimization approach is an attractive research field of
computer aided process engineering (Marlin & Hrymak
[9]
, Perkins
[12]
. The decrease in
hardware and software costs has resulted in several implementations of this technology,
showing quite attractive economical results (Basak et al.
[2]
, Lauks et al.
[6
,White
[14]
). The
efforts in this area have been focused on specific components of the system, to mention

data acquisition and validation, gross error detection, data reconciliation and, of course,
modeling and optimization (Bagajewicz
[1]
, Brown & Rhinerhart
[3]
, Crowe
[4]
, 1996). Besides,
particular attention has been paid to the effects of uncertainty and noise over the final
implementation (Forbes & Marlin
[5]
, Loeblein & Perkins
[8]
, Miletic & Marlin
[10]
, Yip &
Marlin
[15]
). Also numerous successful applications of the RTO in oil refinery industry have
been reported (Lid & Strand
[7]
, Zanin et al.
[16]
).
The objective of this paper is to demonstrate that using real time optimization
technology in some industries can improve the their performance and increase profit by
reducing the offset and maintain the process in optimum condition in spite of unknown
disturbances or changing in desired operating point.
2. Elements in the RTO loop
Real-time operations optimization relies on model updating as feedback that corrects for
model errors and disturbances and enables the RTO system to closely track the true plant
optimum. A typical structure of an RTO loop is shown in Fig. 1.
Figure 1. Typical Architecture of Real-Time Optimization
The main elements in the RTO loop consist of the model updater, model-based optimizer,
results analysis and process control. Real-time measurements, z, are collected via the
distributed control system, checked for reliability and low pass filtered. Then, the process
parameters, β, are estimated using the data in the model updater. The estimated
parameters are then sent to the optimizer, in which model-based optimization is performed.
In the results analysis, statistical tests are used to evaluate the optimizer results before the
values are transmitted to the process controllers. Only significant changes in optimization
variables are forwarded to the process controllers for implementation.
Main applications of real time optimization can be stated as follows:
• Elimination and modification of random errors
• Dynamically describing performance deviation of key equipment from their set points.
• Automatically optimizing plant’s performance
• Automatically performing fault detection (instrumentation’s mal function, leak)
• Intelligent computation of data which there measurements are unavailable (temp.
pressure, flows)
• Calculating and reporting consumption of row materials and production of product at any time
• Assessing energy consumption of the entire plant and the equipments at any desired
time instant
• Determining equipment’s performance loss over time (reactors, exchangers, columns)
• Studying and monitoring history of the equipments and monitoring.
• Sending computed data and results via computer networks to the desired locations in the
plant.
Saeid Shokri et al./Petroleum & Coal 51(2) 110-114 (2009)
111

3. Data Reconciliation
Real Time optimization heavily relies on process measurements and accurate process
models. Process measurements are inevitably corrupted by errors during the measurement
itself but also during its processing and transmission stages. The total error in a
measurement, which is the difference between the measured value and the (definitely
unknown) value of a variable, can be conveniently represented as the sum of the
contributions from two types of errors: random and gross errors. Random errors which are
inherent to the measurement process are usually small in magnitude and are most often
described by the use of probability distributions. On the other hand, gross errors are caused
by nonrandom events such as instrument malfunctioning, miscalibration, wear or corrosion
of sensors and so on. The nonrandom nature of these errors implies that at any given time
they have certain magnitude and sign which may be unknown. Thus, if the measurement is
repeated with the same instrument under identical conditions, the contribution of a
systematic gross error to the measurement value will be the same (Narasimhan &
Jordache
[11]
). It is the reason why gross errors are also called systematic errors or biases.
The process in which the accuracy of data could be improved by detecting two mentioned
types of error and eliminating them is currently known as Data Reconciliation (DR) which is
an important foundational activity and performed in Validation block shown in fig. 1. The
estimates of unmeasured variables as well as model parameters are also obtained as a part
of data reconciliation problem.
There are several proposed methods for accomplishing this important stage but we are
not aim to explain them here. This paper just aim to show the importance of doing this data
processing phase.
4. Some Industrial Cases
Some industrial units using RTO as tool for increasing their economic profit are mentioned in
Table1. Some important points could be implied from data collected in this table:
1. Against the most of new technologies tested in developing countries for insuring
profitability, RTO technology was implemented on units in developed countries even for
testing. It shows that applicability of this new technology is trustable and scientists and
engineers who live in developed countries need not to test this method on the industrial
units of developing countries.
2. These results are reported by industrial units on which RTO technology has been
implemented not companies who are the owner of this technology so these results are
very likely to be correct.
3. It could be seen that the payback period of RTO technology is short. It has positive
impact on general improvement in overall industry economics.
Table 1 Some industrial units using RTO
Company Process Location Capacity RTO
Technology
Benefit Year Payback Reference
Borealis Group
Tech nology
Ethylene
Plant
Finland 300000
tpy
Neste Jacobes 12.5 M$/yr 2004-
2005
One -
month
www.Hydrocarbon
processing.com
Not report Low –sulfur
gasoline HDS
plant
France 870000
metric tpy
Axens 1.1 M€/yr 2006 Not report www.Hydrocarbon
processing.com
Eastman
Chemical
Utility plant USA 3.6M lb/hr
of steam,
176MW of
electricity
Emerson
(AMS suite)
1M$/yr Not
report
Not report www.pmo.assetweb.com
ConocoPhillips Alkylation’s
plant
USA Not report Emerson
(AMS suite)
1.2 M$ / yr Not
report
Not report www.pmo.assetweb.com
BASF, Seal
Sands
Chemicals
Optimization
(acrylics &
nylon
polymer)
U.K. Not report Emerson
(AMS suite)
Not report Not
report
Less than
1 year
www.pmo.assetweb.com
Shell Nederland
Chemie,
Moerdijk,
Petrochemica
ls
Optimization
Nederland Not report Emerson (AMS
suite)
Not report Not
report
Less than
6 months
www.pmo.assetweb.com
Sannazzaro
refinery
Refinery (Fcc
unit)
Italy 200000
Barrels per
day
Aspen Tech
Inc (ASPEN
HYSYS)
10 cents
/barrel
Not
report
Not report www. Aspen tech
.com
Saeid Shokri et al./Petroleum & Coal 51(2) 110-114 (2009)
112

Company Process Location Capacity RTO
Technology
Benefit Year Payback Reference
Refineria Isla,
Curacao
Crude Unit Nederland 180000
Barrels per
day
SimSci&
Foxboro
(ROMeo
&MRA)
2 M$ / yr 2001-
2002
Not report www.eptq.com
Yeosu Yeochun Ethylene
Utilities
Korea Not report Emerson
(AMS suite)
1.038M$ / yr Not
report
Not report www.pmo.assetweb.com
Hyundai
Petrochemical
Co.
Olefins Plant Korea 350000
ton/yr
M.W.Kellog
Co.
12%(increased
profit)
Not
report
Not report Oil & gas journal
ConocoPhillips, Refining
Closed-Loop
Optimization
U.S.A Not report Emerson
(AMS suite)
Between
600,000$ -1.2
M$/yr
Not
report
Not report www.pmo.assetweb.com
Not report Boiler
Performance
Monitoring,
Entergy
U.S.A. Not report Emerson
(AMS suite)
240,000$/yr Not
report
Not report www.pmo.assetweb.com
Sriracha
Refinery
Refinery
Utility
Thailand Not report Emerson
(AMS suite)
1M$ /yr
Not
report
Less than
3 months
www.pmo.assetweb.com
5. The potential of applying RTO in Iran refineries
RTO technology could be implemented in some cases which have certain conditions as
explained below:
1. The frequency of occurring disturbances is sufficient so that one could give a good reason
why applying RTO is necessary or sensible.
2. Total profit should considerably change with changing in optimization parameters.
3. Determining the appropriate value for optimization parameter should be more complex
and not determinable with common methods.
4. An operator faces a numerous numbers of data so that he/she cannot trace
The plant which RTO technology will be implemented on, should has some conditions
explained below:
1. Access to plant data should be possible
2. The process has potential to use Distributed Control System (DCS).
With regard to conditions mentioned above, refinery distillation units, hydrotreating units,
olefin units and ethylene units in petrochemical complexes have several decision variables
to apply real time optimization technology.
So some units without DSC technology are not capable of implementing mentioned technology.
Based on statistical data collected from chemical and petrochemical industries, the most
of RTO consumers in the world are gas and petroleum refineries because these companies
have wide range of decision variables, which affect process optimization considerably. In
addition there are a wide range of products in these units which provides suitable conditions
for applying RTO. Table 2 is the list of Iranian companies with the capability of applying RTO
technology.
Table 2: Gas and oil refining company in Iran since year 2007 which have DSC
Bandar Abbas Oil Refining Company
Arak Refinery Company
South Pars Refinery Phase 1
South Pars Refinery Phase 2 & 3
South Pars Refinery Phase 4 & 5
South Pars Refinery Phase 6, 7 & 8
South Pars Refinery Phase 9 & 10
Parsian Refinery 1
Parsian Refinery 1
6. Conclusion
In this work, the concept of Real Time Optimization (RTO) technology is presented and
the potential of applying this method is evaluated. It is concluded that there are many
refining companies in Iran which RTO could be implemented on to maximize their
profitability, reduce off spec products and minimize energy consumption.
Saeid Shokri et al./Petroleum & Coal 51(2) 110-114 (2009)
113

References
[1] Bagajewicz, M.: A brief review of recent developments in data reconciliation and gross
error detection/estimation. Latin American Applied Research, 30, 335, (2002).
[2] Basak, K., Abhilash, K., Ganguly, S., & Saraf, D.: On-line optimization of a crude
distillation unit with constraints on product properties. Industrial Engineering and
Chemical Research, 41, 1557, (2002).
[3] Brown, P., & Rhinerhart, R. :. Automated steady-state identification in multi-variable
systems. Hydrocarbon Processing, 79 (2000).
[4] Crowe, C.: Data reconciliation—Progresses and challenges. Journal of Process Control,
6, 89 (1996).
[5] Forbes, J. F., & Marlin, T. E.: Design cost: a systematic approach to technology
selection for model-based real-time optimization systems. Computers and Chemical
Engineering, 20 (6/7), 717 (1996).
[6] Lauks, V., Vasbinder, R., Vallenburg, P., & van Leuwen, C.: On-line optimization of an
ethylene plant. Computers and Chemical Engineering, 16(Suppl), S213 (1992).
[7] Lid T., Strand S., Real-time optimization of a cat cracker unit, Computers & Chemical
Engineering, Volume 21, Supplement 1, 20 May 1997, Pages S887-S892.
[8] Loeblein, C., & Perkins, J.: Economic analysis of different structures of on-line process
optimization systems. Computers and Chemical Engineering, 22, 1257 (1998).
[9] Marlin, T. E., & Hrymak, A. N.: Real-time optimization of continuous processes. Fifth
International Conference on Chemical Process Control. American Institute of Chemical
Engineering Symposium Series 93, pp. 316, 156 (1997).
[10] Miletic, I., & Marlin, T.: Results analysis for real-time optimization: Deciding when to
change the plant operation. Computers and Chemical Engineering, 20(Suppl), 1071
(1996).
[11] Narasimhan, S., & Jordache, C.: Data reconciliation and gross error detection—an
intelligent use of process data. Houston, TX, USA: Gulf Publishing Company, (2000).
[12] Perkins, J.: AIChE Symp. Ser. Vol. 94 of 320. Plant-wide optimization: Opportunities
and challenges (p. 15) (1998)
[13] White, D.: On line optimization: What, where and estimating ROI. Hydrocarbon
Processing, 43 (1997).
[14] White, D.: On line optimization: What have we learned? Hydrocarbon Processing, 55
(1998).
[15] Yip, W., & Marlin, T.: Multiple data sets for model updating in realtime operations
optimization. Computers and Chemical Engineering, 26, 1345 (2002).
[16] Zanin, A. C., Gouvêa, M. T., Odloak, D. : Integrating Real-Time Optimization into the
Model Predictive Controller of the FCC system. Control Engineering Practice, 10(8),
819–831 (2002).
Saeid Shokri et al./Petroleum & Coal 51(2) 110-114 (2009)
114