You are on page 1of 8

ISA Transactions 51 (2012) 317324

Contents lists available at SciVerse ScienceDirect

ISA Transactions
journal homepage: www.elsevier.com/locate/isatrans

Natural gas operations: Considerations on process transients, design, and control


Flavio Manenti
CMIC Dept. Giulio Natta, Politecnico di Milano, Piazza Leonardo da Vinci, 32 20133, Milano, Italy

article

info

Article history:
Received 6 January 2010
Received in revised form
24 September 2011
Accepted 8 October 2011
Available online 3 November 2011
Keywords:
Process transients
Plant controllability
Plant stability
Dynamic simulation
Plantwide control
Gas processing

abstract
This manuscript highlights tangible benefits deriving from the dynamic simulation and control of
operational transients of natural gas processing plants. Relevant improvements in safety, controllability,
operability, and flexibility are obtained not only within the traditional applications, i.e. plant start-up and
shutdown, but also in certain fields apparently time-independent such as the feasibility studies of gas
processing plant layout and the process design of processes.
Specifically, this paper enhances the myopic steady-state approach and its main shortcomings with
respect to the more detailed studies that take into consideration the non-steady state behaviors.
A portion of a gas processing facility is considered as case study. Process transients, design, and control
solutions apparently more appealing from a steady-state approach are compared to the corresponding
dynamic simulation solutions.
2011 ISA. Published by Elsevier Ltd. All rights reserved.

1. Introduction
The steady-state simulation is the well-known and wellestablished state-of-the-art for preliminary studies and plant
design within the process industry. Nevertheless, in spite of its usefulness, it is unable to account for the process dynamic conditions
such as upset or emergency scenarios or, more frequently, hot and
cold start-ups and plant shutdowns.
Especially in the last years, characterized by strong dynamics in
the oil price and by significant fluctuations in the market demand,
the detailed dynamic process models are becoming the keytool to improve plant stability, flexibility, and controllability [1].
In addition, according to the analyzed scenarios, the dynamic
simulation can push towards certain modifications in the unit
operations and process design.
As an example, the dynamic simulation allows defining
automatic procedures to manage the plant start-up and shutdown
transients, production quality and load changes, moving from a
steady-state operating condition to a more appealing one, as well
as all the operations that affect both the safety and economics
aspects of an industrial process. Also, the dynamic simulation is
the fundamental basis for the operator training simulation that
allows increasing field and control-room operators experience,
preparation, and coordination to face emergency situations and to
be effective and efficient in accomplishing plant operations.
The main goal of the present research activity is to show,
through an appropriate application, how the dynamic simulation

Tel.: +39 02 2399 3273; fax: +39 02 7063 7183.


E-mail address: flavio.manenti@polimi.it.

can significantly assist the process design activity as well as


the plant-wide control development while improving the overall
plant controllability and stability. It would also demonstrate
the importance of an aprioristic dynamic simulation even in
some time-independent fields, such as the process design, where
dynamic simulation studies are traditionally thought as useless or
marginal.
A description of some of the most common process transients
is given in Section 2. The model and the plantwide control of a gas
processing plant, consisting of refrigeration and separation zones,
are explained in Section 3. The discussion of this application is
reported in Section 4.
2. Process transients
Industrial processes are constantly subject to several kinds
of planned or unplanned operating transients. For example, an
industrial plant is scheduled for programmed maintenances, which
require shutdown and start-up procedures.
The plant maintenance involves, sometimes, either whole
processes or plant subsections only, whereas the remaining
portion of the industrial site keeps on operating. This is usually
denoted as slowdown and it brings to a temporary reduced
production capacity. Analogously, the procedure that brings the
plant from the slowdown condition to the nominal one is generally
called speed-up.
Similar dynamics are represented by grade changes, where the
quality of final products is changed to fulfill the market demand.
On this subject, there is an extensive literature on polymer
plants and distillation columns [27]. For instance, polyethylene

0019-0578/$ see front matter 2011 ISA. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.isatra.2011.10.008

318

F. Manenti / ISA Transactions 51 (2012) 317324

Acronyms
CAPE
C3
C4
iC4
nC4
DAE
FC
HEN
LC
LNG
ODE
PC
PDAE
PDE
PID
SW
TC

Computer-Aided Process Engineering


Propane
Butane
iso-Butane
normal-Butane
Differential Algebraic Equation
Flow Controller
Heat Exchange Network
Level Controller
Liquefied Natural Gas
Ordinary Differential Equation
Pressure Controller
Partial Differential Algebraic
Partial Differential Equation
Proportional, Integral, Derivative
Switch
Temperature Controller

terephthalate is mainly produced in three different grades [8,9]


for textile fibers (low intrinsic viscosity), for bottle market
(medium intrinsic viscosity), and for tire-cord resins (high intrinsic
viscosity). It is necessary to perform a grade change transition to
switch from one quality to another.
Unfortunately, also other unplanned process transients are
rather frequent. Specifically, process uncertainties, external disturbances, and sudden malfunctions can induce strong variations in
plant operating conditions. Also, unexpected peaks in the market
demand of final products and the volatility of raw material prices
may significantly affect process conditions, especially in facilities
characterized by just-in-time policies and/or market-driven productions [10,11].
At last, emergency conditions caused by alarms, failures, or
accidents inevitably lead to rapid and unexpected shutdowns.
Besides the aforementioned transients, other unexpected conditions exist (e.g. snowball effect, reactor runaway. . . ). However,
this paper discusses only a few of the aforementioned transients
and proposes different start-up procedures to increase the process
safety, to ensure the operators safety, and to improve the process
design of chemical plants and unit operations.
3. The gas processing plant
The dynamic simulators are useful tools for describing either
complex chemical processes or simple unit operations in terms
of differential or differential-algebraic systems while integrating
them through dedicated numerical techniques. Nowadays, several
packages for the dynamic simulation of processes are commercially available such as DYNSIM (SimSciEsscor, Invensys), AspenHysys (Hyprotech, Aspen Technology), and UNISIM (Honeywell) to
quote some of the most widespread.
These toolkits are mainly equation-oriented where differential equations define energy, momentum, and material balances,
whereas algebraic equations describe thermodynamic, physical,
chemical, hydraulic properties, etc. Notwithstanding these packages are characterized by large databases for process unit models
and component chemicalphysical properties, there is sometimes
the need to custom them by introducing specific compounds, oil
cuts, and more complex models [12,13].
The dynamic simulation always requires more parameters
than the steady-state simulation, since ancillary units, geometric
dimensions, port heights and diameters, etc., must be specified to
account for hold-ups and residence times [14,15]. Moreover, if the

Table 1
Common control scheme for the refrigeration section.
Name

Type

Controlled variable

Manipulated variable

PID1
PID2
PID3
PID4
PID5
PID11

Level
Temperature
Level
Pressure
Ratio
Flow

INLETSEP hold-up
Inlet LTS temperature
LTS hold-up
INLETSEP pressure
Feed flow S3
Feed flow S1

Exiting liquid flow


Chiller coolant
Exiting liquid flow
Flare stream
XV2 valve
XV1 valve

Table 2
Common control scheme for the distillation column.
Name

Type

Controlled variable

Manipulated variable

PID6
PID7
PID7
PID8
PID9
PID10
PID12

Temperature
Level
Level
Pressure
Level
Temperature
Cascade

Tray-10 temperature
Condenser hold-up
Condenser hold-up
Condenser pressure
Reboiler hold-up
Tray-1 temperature
Inlet flow

Coolant flow
Reflux
Reflux
Exiting gas flow
Bottom flow
Heating flow
S16 and PID11

user wants to simulate a closed-loop problem, it is necessary to


address the tuning of control loops [1618].
Here below, certain significant improvements achieved through
the dynamic simulation are discussed for a gas processing plant,
consisting of a refrigeration section and a 12-stage depropanizer.
Fig. 1 shows the refrigeration section. It is aimed at separating
the portion of incondensable compounds from the heaviest
components by means of a series of flash separators and heat
exchangers. The depropanizer of Fig. 2 is fed by the heaviest
components (propane (C3), iso-butane (iC4), normal-butane (nC4))
and a small amount of incondensable components. The feed stream
coming from the cryogenic separation units is preheated up to
10 C. The overhead condenser pressure is about 13 bar by keeping
the condensing temperature rather low; this avoids the use of
water as coolant to refrigerate the top distillate.
Conversely, the heating fluid supplied to the reboiler is a low
pressure steam. The column has two specifications: the butane
fraction (both iC4 and nC4) in the bottom product line must be
greater than 98% molar, whereas the iC4 and nC4 losses in the
distillate must be minimized as possible.
The classical control configuration for a distillation column
[1,14,18,19] consists of 6 controlled variables: the inlet flowrate,
the pressure at the top of the column; the top reflux accumulator
and the bottom reboiler levels, and the temperatures of trays 1 and
10. Control loops are reported in Tables 1 and 2.
The only differences from the common control structure are
the adoption of a ratio controller for the feed flow and a cascade
controller, to improve the regulation of the inlet flowrate to the
column. Different analyses were performed: the first concerns the
plant design configuration (Section 3.1), important to analyze the
plant stability by means of the dynamic simulation analysis; the
second is based on the process control assessment (Section 3.2),
to improve the plant controllability during relevant unsteady
conditions.
3.1. Plant design configurations
The operating column specifications deal with the top and
bottom compositions. With a column pressure of 13 bar, it is
necessary to operate at 10 C in the condenser to achieve the
desired specifications on the distillate outflow: this requires the
usage of an auxiliary fluid as coolant, with a significant increment
in the operating costs.
Among many possibilities [2023], a simple process intensification is to adopt a further heat exchanger to recover some amount
of energy. In Fig. 3(B), the stream from the head of the column is

F. Manenti / ISA Transactions 51 (2012) 317324

319

Fig. 1. Refrigeration section.

Fig. 2. Separation section.

first fed to a cooler (water cooler). Second, it is further refrigerated in the following exchanger, by heating the inlet flowrate to the
column.

feed-effluent heat exchanger that is a candidate solution for process intensification and energy saving. On the other hand, Fig. 3(A)
shows the unintensified column layout.

Actually, by preheating the inlet flowrate, the steam consumption in the reboiler is significantly reduced. Fig. 3(B) outlines the

This manuscript will investigate the pros and cons in stability


and controllability of both configurations A and B, through the

320

F. Manenti / ISA Transactions 51 (2012) 317324

Fig. 3. Alternative plant design solutions.

Fig. 4. Standard (A) and advanced (B) control schemes at the bottom of the depropanizer. The SW unit automatically switches from solution B to A when steady-state
conditions are almost reached.

analysis of plant transients. It is worth underlining that the steadystate simulations clearly indicate that the Design B is the best
economic alternative.
3.2. Improvements in plant-wide control systems
The second study deals with potentialities of the dynamic
simulation in the development of plant-wide control systems (see
also Fig. 4). The classical control scheme discussed in the previous
section ensures the process stability, when the depropanizer
operates close to the design conditions.
Nevertheless, the control loop for the liquid level in the reboiler
is disabled during the transient, at least until the required butane
purity is achieved: during the transient the C4 molar fraction is
off-spec and the product cannot be unloaded. This is an evident
weakness of the standard control system, (see also Fig. 4(A))
[24,25].
Therefore, a controller that directly acts on the reboiler duty
(Fig. 4(B)) is more efficient and allows preventing the phenomenon

of partial wetting of the tube bundle by the liquid hold-up. Only


when the quality specification is achieved, it is possible to switch
to the standard control configuration, which allows to start the
sampling of the bottom product.
The controller LC13 of the configuration B is the only
active automatic controller during the start-up transient, since it
manipulates the steam flowrate to regulate the liquid hold-up in
the reboiler. When the system is approaching the steady-state
conditions, the automatic control is switched from LC13 to the
standard configuration of controllers TC10 and LC09 to definitely
favor the column stability.
4. Simulation results
Dynamic models and control configurations reported in this
manuscript were performed by the dynamic code DYNSIMTM on a
Intel Core2 Duo, 2 GB of RAM, 2 GHz, and Microsoft Windows XP
as operating system. PengRobinson is adopted to characterize the

F. Manenti / ISA Transactions 51 (2012) 317324

321

Fig. 5. Reboiler temperature. Design solution A and control scheme A.

thermodynamics of the system. The simulation times are much


faster than the real time, about 20 times faster for the hardest
process dynamics with the overall process running. This allows
the commercial packages to be effective for both off-line and online needs such as performance monitoring, operator training,
predictive control and dynamic optimization [2628].
The start-up and shutdown of alternative plant design solutions
(see Fig. 3) are simulated and discussed. Next, the alternative
control schemes are compared.
4.1. Analysis of operating transients
The dynamic analysis of process start-ups and shutdowns
allows defining two ad hoc automatic procedures.
Gas processing start-up
In this section, start-up and shutdown procedures are referred
to the standard (control and design) configuration A. The initial
reboiler temperature is 25 C (Fig. 5). When the cold stream is fed
to the reboiler, the temperature slowly decreases while the overall
metal mass of the unit approaches the fluid temperature, which is
close to 0 C. When the reboiler is sufficiently filled up of liquid,
the controller PID10, which controls the bottom temperature of
the column, is activated and the reboiler duty starts increasing the
bottom temperature.
In Fig. 6, the reboiler pressure increases with the rise of the
fluid hold-up (actually, there is a positive inlet flow and no outlet
flows), until the temperature controller PID10 is activated. At that
point, the reboiler starts heating the accumulated fluid and a
liquid fraction is instantaneously vaporized, largely increasing the
pressure.
Only when the controller PID8 (pressure loop at the top of
the column) is switched to the automatic mode, the pressure is
stabilized around 1350 kPa in a few minutes. Propane is a bottom
product impurity and process specifications are reached after 45
min (as reported in Fig. 7).
Gas processing shutdown
The shutdown of the gas processing plant must accomplish
certain important actions: extinction of any inlet flowrate (note
that the extinction of the flowrate S1 inevitably leads to the
extinction of the flowrate S3, since a ratio loop controls the S3

by measuring the molar flow of the S1), switch off of electrical


motors and duties, unloading of any liquid hold-up and unit
depressurization. The dynamics of the column inlet flowrate is
illustrated in Fig. 8. It is achieved by imposing a linear ramp (from
the actual value to zero) to the setpoint of the flow controller
PID11, so as to avoid any sudden responses that would induce
process instabilities during the shutdown procedure.
Moreover, Fig. 8 shows that the plant inlet flowrate is rapidly
decreased since it is controlled directly by the PID11.
On the other hand, the extinction of the column inlet flowrate
is slow since it receives the outlet streams from the INLETSEP and
LTS separators and it has to process all the flows accumulated
across the plant, before closing the valve XV6, which is positioned
between the refrigeration section and the distillation column
(C3/C4 splitter). The chronological actions of the shutdown
procedure can be recognized in the trend of the column inlet
flowrate:

First two minutes, the extinction (by linear ramp) of any


inlet flowrate reduces the mass flow from 7000 kg/h to about
6600 kg/h.
After 2 min: the level controller PID1 is put in manual mode
and the XV3 valve that regulates the INLESEP liquid hold-up is
opened, for unloading the drum liquid accumulation.
Once empted the INLETSEP, the level controller PID3 is put in
manual mode and the valve XV5, which controls the LTS liquid
hold-up, is opened.
At the same time, the linear ramp on the inlet flowrate has been
completely accomplished.
Once the LTS drum is empted, the valve XV6, which is
positioned on the stream S16 that feeds the distillation column,
is definitely closed.
Every unit is depressurized.
It is however worth saying that, in this paper, we consider the
shutdown procedure accomplished when all the aforementioned
operations within the shutdown scenario are carried out, without
accounting for the metal mass, thermal, liquid, and gas holdups still
present and relevant for the process units.
4.2. Selection of the preferable configuration
The presence of long transients, internal and external persistent
disturbances, more and more frequent grade changes for the

322

F. Manenti / ISA Transactions 51 (2012) 317324

Fig. 6. Reboiler pressure. Design solution A and control scheme A.

Fig. 7. Propane molar fraction in the reboiler. Design solution A and control scheme A.

Fig. 8. Inlet flowrates. Design solution A and control scheme A.

F. Manenti / ISA Transactions 51 (2012) 317324

323

Fig. 9. Temperature dynamics across the heat exchanger network (HEN) in the overhead condenser (Fig. 3), after a 5-minute-pulse disturbance occurring in the column
inlet flowrate.

Fig. 10. Overhead butane composition (C4 losses) after the pulse disturbance.

market-driven production, and thermal or material recycle loops


are pushing the design procedure of chemical processes towards
the intensive use of process dynamic simulations and the analysis
of their unsteady conditions.
Actually, the steady-state analysis could be sometimes insufficient to identify the appropriate plant layout [1,29]. As a matter of
fact, Figs. 9 and 10 compare trends obtained with the same process
control system, but with different design configurations. Specifically, Fig. 9 shows that the Design B is characterized by larger temperature oscillations when it undergoes a pulse disturbance in the
temperature coolant. Moreover, the disturbance extends its effect
for a longer time by seriously affecting the process stability of the
Design B.
Conversely, the Design A is able to dampen efficiently and
promptly the pulse effects. Compositions have a longer off-spec
period using the Design B and the product quality is hardly
controlled throughout the transients (Fig. 10). Thus, the overall
process operability is worsened. In presence of more complex
disturbances, only the Design A allows reaching reasonably good
process controllability.
Even though the process steady-state simulation shows without any doubt that the Design B is preferable with respect to
the Design A for its energy-saving layout, the dynamic simulation
allows highlighting how the Design A is more stable and how it increases the overall process operability and controllability. Moreover, due to the existence of long transients for the Design B
layout, the economic losses (off-spec products) can significantly
reduce the benefits coming from the energy recovery related to the
intensified Design B by the introduction of the EXCHANGER_C (see
also Fig. 3).
4.3. Control system improvements
The control schemes of Fig. 4 were tested on the start-up
procedure. The controllers TC10 and LC13 are switched on in the
Case A and B respectively, when the reboiler level reaches the
setpoint (1.2 m). Then, in the Case A, LC09 is activated when the
setpoint of TC10 is achieved.
Conversely, in the Case B, when the temperature is close to the
setpoint of TC10, the controller LC13 is deactivated and the control
configuration is immediately switched to the traditional system of
the Case A.

Fig. 11 illustrates the liquid level within the reboiler. When


TC10 is activated in the Case A, around the 14th min, the level
significantly overshoots the setpoint with oscillation amplitudes of
0.4 m. When the overhead reflux valve is opened after 30 min, the
level slowly converges to the setpoint. By adopting the alternative
control scheme B, the liquid level is practically flat (oscillations are
in the range of 0.1 m).
Actually, the control scheme B immediately dampens the oscillations produced by the start-up procedure, while guaranteeing
the absence of significant variations in the operating conditions. It
performs the plant start-up with minimum holdups while strongly
reducing both transient times and risks of unexpected phenomena,
such as flooding and dumping.
This is particularly important to ensure that the tube bundle
within the reboiler is always wet when the duty is supplied, to
prevent any mechanical failure.
5. Conclusions
This manuscript showed certain potentialities of the dynamic
simulation and the analysis of operating transients as well. Benefits obtained in different fields such as process control, plant safety,
and operator training are discussed in order to show that the dynamic simulation cannot be any longer disregarded for managing
and controlling unsteady operations and upset conditions.
In addition, the study presented appealing solutions in disciplines apparently independent from time-evolving phenomena,
such as the plant design, which is currently carried out a priori,
without considering any interaction with the dynamic simulation.
We have seen as the capability to simulate possible transients during the preliminary plant design can be fundamental in the process layout definition. The dynamic simulation allows pushing the
design activity towards solutions that are often discarded or even
unexpected by looking at the sole steady-state simulation.
When assessing the effectiveness of alternative control
schemes, the dynamic simulation is a crucial tool to discriminate
the best configuration for the plant safety, operability, and controllability, besides respecting the environmental regulations and
product qualities.
To conclude, it is worth remarking that a real plant is no place
for training and an aprioristic experience based on the dynamic
simulation may be fundamental either to shrink human factors

324

F. Manenti / ISA Transactions 51 (2012) 317324

Fig. 11. Liquid level in the reboiler during the plant start-up.

and industrial accidents or to reduce their impact/damages or


even preventing them. In this perspective and looking forward at
the decreasing computational efforts to perform it, the detailed
dynamic simulation could support HAZOP studies and whatif
analyses, which will become a key factor for future CAPE
applications such as virtual reality, operator training simulation,
fault detection, and automaton approaches.
References
[1] Luyben WL, Luyben ML. Essentials of process control. Chemical Engineering
Series, New York: McGraw-Hill; 1997.
[2] Cervantes AM, Tonelli S, Brandolin A, Bandoni JA, Biegler LT. Large-scale
dynamic optimization for grade transitions in a low density polyethylene
plant. Computers & Chemical Engineering 2002;26(2):22737.
[3] Chatzidoukas C, Perkins JD, Pistikopoulos EN, Kiparissides C. Optimal grade
transition and selection of closed-loop controllers in a gas-phase olefin
polymerization fluidized bed reactor. Chemical Engineering Science 2003;58:
364358.
[4] Benamor S, Doyle FJ, McFarlane R. Polymer grade transition control using
advanced real-time optimization software. Journal of Process Control 2004;
14(4):34964.
[5] Flores-Tlacuahuac A, Biegler LT, Saldivar-Guerra E. Optimal grade transitions
in the high-impact polystyrene polymerization process. Industrial & Engineering Chemistry Research 2006;45(18):617589.
[6] Lima NMN, Manenti F, Maciel Filho R, Embiruu M, Wolf Maciel MR. Fuzzy
model-based predictive hybrid control of polymerization processes. Industrial
& Engineering Chemistry Research 2009;48(18):854250.
[7] Dones I, Manenti F, Preisig HA, Buzzi-Ferraris G. Nonlinear model predictive
control: a self-adaptive approach. Industrial & Engineering Chemistry
Research 2010;49(10):478291.
[8] Manenti F, Rovaglio M. Integrated multilevel optimization in large-scale
polyethylene terephthalate plants. Industrial & Engineering Chemistry
Research 2008;47(1):92104.
[9] Manenti F. Considerations on nonlinear model predictive control techniques.
Computers & Chemical Engineering 2011;35:2491509.
[10] Busch J, Oldenburg J, Santos M, Cruse A, Marquardt W. Dynamic predictive
scheduling of operational strategies for continuous processes using mixedlogic dynamic optimization. Computers & Chemical Engineering 2007;31:
57487.
[11] Varma VA, Reklaitis GV, Blau GE, Pekny JF. Enterprise-wide modeling
& optimization an overview of emerging research challenges and
opportunities. Computers & Chemical Engineering 2007;31(56):692711.

[12] Lima NMN, Zuiga Lian L, Manenti F, Maciel Filho R, Wolf Maciel MR,
Embiruu M, et al. Fuzzy cognitive approach of a molecular distillation process.
Chemical Engineering Research and Design 2011;89(4):4719.
[13] Manenti F, Cieri S, Restelli M. Considerations on the steady-state modeling
of methanol synthesis fixed-bed reactor. Chemical Engineering Science 2011;
66(2):15262.
[14] Stephanopoulos G. Chemical Process Control: An Introduction to Theory and
Practice. Englewood Cliffs, NJ: Prentice-Hall; 1984.
[15] Morari M, Zafiriou E. Robust process control. Englewood Cliffs, NJ: PrenticeHall; 1989.
[16] Skogestad S. Simple analytic rules for model reduction and PID controller
tuning. Modeling Identification and Control 2004;25(2):85120.
[17] Skogestad S. Tuning for smooth PID control with acceptable disturbance rejection. Industrial & Engineering Chemistry Research 2006;45(23):
78177822.
[18] Skogestad S. The dos and donts of distillation column control. Chemical
Engineering Research & Design 2007;85(A1):1323.
[19] Barolo M, Papini CA. Improving dual composition control in continuous distillation by a novel column design. Aiche Journal 2000;46(1):
146159.
[20] Jacobsen EW, Skogestad S. Multiple steady-states and instability in distillation
implications for operation and control. Industrial & Engineering Chemistry
Research 1995;34(12):4395405.
[21] Mohideen MJ, Perkins JD, Pistikopoulos EN. Robust stability considerations
in optimal design of dynamic systems under uncertainty. Journal of Process
Control 1997;7(5):37185.
[22] Gani R, Bek-Pedersen E. Simple new algorithm for distillation column design.
Aiche Journal 2000;46(6):12714.
[23] Grossmann IE, Aguirre PA, Barttfeld M. Optimal synthesis of complex distillation columns using rigorous models. Computers & Chemical Engineering 2005;
29(6):120315.
[24] Skogestad S. Dynamics and control of distillation columns: a tutorial
introduction. Chemical Engineering Research & Design 1997;75(A6):
539562.
[25] Skogestad S. Dynamics and control of distillation columns a critical survey.
Modeling Identification and Control 1997;18(3):177217.
[26] Manenti F, Signor S, Grottoli MG, Fabbri P. Adaptive data reconciliation
coupling C++ and PRO/II and on-line application by the field. Computer Aided
Chemical Engineering 2010;28:3738.
[27] Signor S, Manenti F, Grottoli MG, Fabbri P, Pierucci S. Sulfur recovery units:
adaptive simulation and model validation on an industrial plant. Industrial &
Engineering Chemistry Research 2010;49(12):571424.
[28] Manenti F, Buzzi-Ferraris G, Pierucci S, Rovaglio M, Gulati H. Process dynamic
optimization using ROMeo. Computer Aided Chemical Engineering 2011;29:
4526.
[29] Luyben WL. Chemical reactor design and control. Wiley; 2007.

You might also like