ISSN 0005-1179, Automation and Remote Control, 2020, Vol. 81, No. 10, pp. 1929–1943.

 c Pleiades Publishing, Ltd., 2020.

c The Author(s), 2015, published in Avtomatizatsiya v Promyshlennosti, 2015, No. 4, pp. 4–14.
Russian Text 

AUTOMATION IN INDUSTRY

Advanced Process Control:

From a PID Loop up to Refinery-Wide Optimization
P. L. Logunov∗ , M. V. Shamanin∗ , D. V. Kneller∗∗,a ,
S. P. Setin∗∗ , and M. M. Shunderyuk∗∗
∗
LLC “LUKOIL-Nizhegorodnefteorgsintez,” Kstovo, Russia
∗∗
JCS Honeywell, Moscow, Russia
e-mail: a [email protected]
Received February 5, 2015
Revised February 5, 2015
Accepted May 25, 2020
Abstract—Advanced process control (APC) provides distributed control systems with wider
functionality thus improving operations efficiency. At the same time, APC systems create a
framework for higher-level optimization solutions. The paper outlines the key features and
advantages of Honeywell Profit R
Suite product family enabling effective and user-friendly
design of layered optimization solutions. With a case study of a major Russian refinery it
overviews the development and operation history of advanced control (AC) and optimization
solutions from a PID loop up to a group of interrelated process units.
Keywords: advanced process control (APC), model-based predictive control (MPC), dynamic
optimization of several process units, PID control performance monitoring, APC performance
monitoring
DOI: 10.1134/S0005117920100100

1. INTRODUCTION
Advanced process control (APC) is a key segment of industrial automation. Its main goal is to
increase the profitability of existing distributed control systems (DCS) at process units Most of
today’s DCS have some room for improvement, and APC offers a powerful tool to capture potential
benefits.
The term “APC” covers a wide variety of control technologies from relatively simple DCS-
embedded control schemes, such as PID cascades, feedforward, ratio, or lead-lag controls, etc.,
through science-intensive process optimization techniques. The latter may include adaptive or
fuzzy control algorithms, quality estimators (QE, aka soft sensors, or inferential calculations), as
well as model-based multivariable controllers. Over the past 30 years, multivariable control systems
with embedded predictive models have been widely accepted in processing industries under the
abbreviation MPC that stands for model predictive control. The MPC technology has been steadily
proving its high potential for multivariable processes with strong interrelations and time lags. APC
is frequently referred to in the narrower sense as MPC, and we will follow this convention further
on. Some APC fundamentals may be found in [1, 2] with more details, case studies, and discussions
in [3].
An APC system will be further referred to as special software for multivariable control of a
process unit based on its predictive model. APC applications usually operate on dedicated servers
integrated into DCS LAN; some vendors create them however, within native DCS in the form of
one or more complex cascade controllers.

1929
1930 LOGUNOV et al.

APC systems of large process units are typically implemented as separate blocks, called multi-
variable controllers (or simply “controllers” software applications that should not be confused with
conventional hardware devices of the same name)
Among the advantages of AC, which determine its attractiveness for users, the following should
be noted.
• Higher automation level expressed as reduced operator involvement. APC applications under-

take a large part of the routine control functions and, thanks to the predictive model, performs
them better than operators. If a process is running under the APC system, operators have to look
after a much smaller number of process variables than without APC: all they have to do is just keep
an eye on APC applications and type in new operation targets as appropriate.
• Product quality stabilization, that means:

—less off-spec product,

—more consistent product quality (variability decrease typically by 20–30 %, sometimes up to
two or even threefold.
• Improved operations reliability APC applications run the process within its normal operating

envelope This results in reduced number of process upsets and lesser wear and tear of process and
control equipment.
• Process optimization implying higher throughput, improved product yields, or energy conser-

vation. The optimization is activated in APC if the operator sets process variable limits narrower
than the safety ones but still wide enough to ensure sufficient control flexibility. As a result, the
process gets the necessary degrees of freedom allowing APC to select and maintain the optimal
operation mode according to a user-defined objective function.
Honeywell is recognized worldwide as a tier-one APC vendor: according to independent industry
consultants, such as ARC Advisory Group or Frost & Sullivan its market share in oil refining varies
around 15 %—the highest in this segment. The features of Honeywell’s APC technology, which ensure
its leadership, are discussed below.
The first APC applications came to Russia by 2005. Further proliferation has not been easy; the
results obtained by Honeywell Russia APC team are all the more impressive: 25 projects within less
than 10 years with almost complete market domination. Key features and milestones will be further
discussed with the case study of a major Russian oil refinery that has been successfully realizing
the potential of this promising technology over the past 8 years.

2. KEY ADVANTAGES OF HONEYWELL APC

Advanced control algorithms from major APC vendors are based on the fundamental MPC
principles. In process of bringing theoretical results to commercial products and their further
evolution each vendor added his know-how that gave the finished products individual traits.
This Section outlines the key advantages of Profit R
Controller—the main product of ProfitR

Suite process optimization family which is a part of Honeywell Connected Plant portfolio. These
featured properties underlie successful implementation of Honeywell’s APC in Russia and elsewhere
across the globe.

2.1. Convenient Design, Engineering, and Simulation Toolkit

The Profit R
Design Studio (PDS) toolkit allows one to carry out the entire development pro-
cess from raw data preprocessing through applications design and simulation, and the configuring
of executable applications. Owing to easy-to-use model identification and validation functionality,
PDS is easily mastered and effectively used by plant engineers who have passed the minimum

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1931

necessary training. The PDS can be compared to a modern multifunction camera that’s equally
comfortable for both beginners (thanks to its automatic focus, sets of standard shooting modes,
etc.) and experienced professionals who prefer manual settings PDS is fully integrated with other
Honeywell products, in particular, with Profit R
Stepper software, which enables process step-
testing with simultaneous data collection, model identification and validation. The models devel-
oped in Profit R
Stepper can be imported into the PDS, reviewed by the engineer and improved,
if necessary.
For predictive models, Profit
R
Controller basically employs linear transfer functions for their
parsimony, transparency and easy implementation in the multivariable closedloop control PDS
automatically offers the developer the simplest lowest-order models to fit the step-test data.

2.2. Robustness of Multivariable Control Algorithms

The term “robustness” is typically understood by the APC community as the ability of ap-
plications to control the process to targets even based on an inaccurate predictive model. The
robustness of a multivariable controller is a critical requirement for any APC technology.
Predictive models for APC applications are developed on the basis of process history and step-
test data, and the process is step-tested in certain operation conditions. There is no guarantee that
the model will describe process dynamics with admissible accuracy in a wider range of operating
modes. An inaccurate model coupled with an insufficiently robust control algorithm may result in
stability loss with possible devastating effects for the process and equipment. Profit R
Controller is
globally recognized for its highest robustness due to several engineering options aimed at stability
improvement and, first of all, the Honeywell’s patented Range Control Algorithm (RAC). This
unique algorithm enables large control moves for fast compensation of external disturbances such
as feed properties fluctuations, changing weather conditions or new operation targets, while the
control plant remains stable during the transient.

2.3. Minimal Impact on Process

The above-mentioned RAC affects the plant to the extent necessary to compensate for distur-
bances. Most industrial multivariable controllers operate at all times, adjusting relentlessly the
operation mode, pulling the control valves and, eventually, spending excessive energy for nothing
In contrast, ProfitR
Controller would slumber until the model forecast points out possible overrun
by one or more process variables. Once the potential peril is detected the controller wakes up and
starts pushing the process smoothly to the desirable mode.

2.4. Easy Tuning

The simplicity of the predictive models used and the effectiveness of the control algorithm is
combined with the convenience of controller tuning. A predictive model in multivariable control
is actually a set of SISO models altogether named “model matrix.” When a new target (setpoint
or limit) is specified for some controlled variable the multivariable control algorithm calculates
the necessary control actions (“moves”) for one or more manipulated variables associated with the
controlled one through predictive models. The procedure of getting current instrument readings
through the DCS, predicting process dynamics for several time steps ahead, calculating the moves
and transmitting them back to the DCS is a recurring one and runs in real time, typically once in
30–60 seconds. Move sizes can be adjusted during APC application setup thus ensuring either faster
control to the target with possible overshoots or more sluggish one when smooth plant operation
is more relevant than higher speed. In Profit R
Controller, a coefficient named performance ratio
is used to specify control aggressiveness for each controlled variable; it has the optimum default

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

1932 LOGUNOV et al.

value. Note that Profit R

Controller is equipped with many other handles, each one offering the
most preferable default value but still leaving enough room for fine manual tuning (again, as in the
camera).

2.5. Built-in Optimizer

The embedded optimization algorithm selects and maintains in real time optimum operating
mode under sufficient number of degrees of freedom for the controller. Optimization criteria (ob-
jective functions) can have a linear and/or quadratic structure. The optimization speed can be set
up with a special handle as a ratio to control speed; a default value is also available as well as more
handles for fine tuning.

2.6. Quality Estimators and Their Update per Lab Data

To control product qualities to targets, the APC system must “see” their values in real time.
(If no APC is available, the product streams are sampled at the plant and then measured in the
laboratory, sometimes by stream analyzers.) Quality estimators (QE) aka soft sensors are the “eyes”
of an APC system, and it is critically important to ensure its good eyesight. QE performance should
be monitored on a regular basis and their mathematical models adjusted as appropriate based on
lab or analyzer data. Honeywell had developed and has been successfully using a semiautomated
lab update application based on statistical process control technique. When a new value comes from
the LIMS through the refinery network, the application evaluates its statistical significance and the
expediency of QE model update. If these conditions match it calculates the amendment and asks
the user to confirm the automatic bias update or offers a value for manual entry (dependent on
the application settings). During the update procedure, multivariable controllers continue working.
The application is integrated with Honeywell Profit R
Sensor Pro toolkit for QE development and
maintenance.

2.7. Ergonomic User Interface

The operator interface of Profit R
Controller integrated with Windows, is localized (Fig. 1) and
contains only the functions necessary for operators to monitor APC variables and specify new
targets. The engineer interface is an extension the operator one. APC system access levels (operator,
supervisor control engineer) are configured at the commissioning phase.

2.8. Integration with other Honeywell Solutions

Profit R
Controller-based APC systems integrate perfectly with Honeywell Experion R
PKS
(EPKS) DCS. Ease of integration is expressed, in particular, in the ability to use Honeywell Dis-
tributed Server Architecture (DSA)—a network technology which provides easier interaction be-
tween APC and EPKS servers as against the conventional OPC. Another advantage is the use of
Experion R
HMI Web graphic tools which makes it easy to create special APC schematics in the
same style as the rest of the operator graphics.
The convenience of interfacing APC applications with first-principles process models should be
emphasized. Such models can be developed in the Honeywell UniSim R
modeling and simulation
environment and employed in APC for gain scheduling as appropriate. The gains are calculated by
the first-principles model in real time (typically, every 5 or 10 minutes) and transferred to Profit
R
R
Controller via a special Profit Bridge interface. Gain scheduling technique thus enables effective
application of linear transfer function models in a wider class of nonlinear processes such as, say,
polymerization.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1933

Fig. 1. Controlled variables windows of the operator interface Profit

R
Suite Operator Station (PSOS).

The ease of integration with EPKS does not confine the interaction of Honeywell APC applica-
tions with third party DCS. More than a half of Honeywell APC systems in Russia were successfully
integrated with third-party DCS from all key vendors on the basis of OPC technology. Honeywell
has also a dedicated MatrikonOPC Funnel DA software enabling significant reduction of integration
efforts and cost. With regard to third-party first-principles models, the Profit R
Bridge interface
can be also used to connect with them. Honeywell has such experience with SPYRO software for
pyrolysis as well as with various polymerization models developed by technology licensors.
For more than 25 years, Honeywell has been investing heavily in Profit R
Suite development on
a regular basis thus extending its functionality, applicability, and convenience. For example, the
most recent release features advanced developer’s toolkit (Profit R
Suite Engineering Studio) and

R
the possibility to install Profit Controller release upgrade at the APC server as a software patch
without any need for a complete reinstallation.

3. APC AT THE MAJOR RUSSIAN REFINERY: A CASE STUDY

The advantages of APC software can be fully realized only if they are implemented in a real-
life APC system that allows for all process specifics and is easy to operate and maintain. The
development of an APC system typically takes up to 1 year and requires cooperation between the
customer and the contractor. The stages of an APC project at a Russian refinery are described
in [4] with the case study of a crude and vacuum distillation unit.
In this section, we will briefly review the development application and evolution of APC in
“LUKOIL-Nizhegorodnefteorgsintez” LLC (hereinafter referred to as the “Refinery” or “NNOS”).
In the 8 years since the first APC applications were deployed, the Refinery has successfully installed
10 systems with more than 30 multivariable controllers and 150 QE altogether. All applications have
been operated and maintained with high uptime and good profitability—a result still outstanding
for the Russian industry and notable even globally. It is just not surprising that European indus-
trial automation community has shown interest in the Refinery’s APC achievements presented at

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

1934 LOGUNOV et al.

Fig. 2. Block diagram of NNOS process units and main streams: Dark grey blocks–units with APC applications,
white blocks–APC is planned; light grey blocks—APC not planned, dash line—Profit R
Optimizer solution at
R
the VGO complex #1, dotted lines—planned Profit Optimizer solutions at the VGO complex #2 and two
Diesel hydrotreaters.

annual Honeywell User Groups workshops. The Refinery’s experience in APC and other advanced
automation solutions are regularly reported at specialized conferences and in industry press. The
aim of this article is to present a brief overview of the Refinery’s production optimization solutions
with a roadmap for further development.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1935

For a pilot APC project, NNOS has chosen CDU-5 and CDU-6 crude/vacuum distillation units.
This choice was predictable: crude units are among the most likely candidates for APC due to
their high capacity and optimization potential. These units process up to 75 % of Refinery’s crude
altogether. Thus, even minor benefits attained here might have resulted in much higher returns
downstream, that has eventually happened.
Both units appeared to be ready for APC implementation from the technical viewpoint. (The
readiness appeared however to be somewhat overestimated.) Honeywell and NNOS engineering
teams faced numerous technical and organizational challenges which were eventually surmounted
owing to well-organized and creative teamwork and effective support rendered by Refinery’s man-
agement [5]. The high level of confidence among all stakeholders coupled with the overall readiness
to novel solutions was a major key to project success. All works were completed in 1.5 years; as
a result, APC applications were deployed at all key sections of both units: atmospheric, vacuum,
and naphtha rerun. By the time of its completion (March 2009) it was the largest such project in
Russia.
Further APC proliferation over NNOS (Fig. 2) was underpinned by the following considerations:
—priority implementation of APC applications on the units with the best possible project pay-
back;
—further implementation on the units which are less profitable but make altogether a framework
for higher-level optimization solutions;
—inability to implement APC systems at more than two units per year due to financial and
organizational constraints
Due to these considerations, the development of APC applications in the next 6 years followed
the following roadmap:
—LF-35-21/1000 UOP CCR PlatformingTM unit was attractive in terms of payback plus a key
link in the naphtha preparation chain according to “Euro” specifications;
—AGFU gas plant plus UOP ParIsomTM isomerization unit was a step towards further automa-
tion and optimization of the gasoline production chain;
—Fluid catalytic cracking unit (UOP FCCU) with C3 splitter is the core of vacuum gas oil
conversion complex commissioned in 2011–2012; typically, FCCU is among the most attractive
units in terms of APC payback;
—UOP HF Alkylation and vacuum gas oil hydrotreater (VGO HTU) are the basis for higher-level
optimization solutions at the VGO complex; in addition alkylation processes look also lucrative in
APC terms (the highest benefits per unit of alkylate product);
—CDU-1 and CDU-2 crude units—appreciable (albeit rather modest due to the relatively low
throughput) benefits plus the addition of necessary elements to higher-level optimization framework.

4. THE BENEFITS
APC projects are driven by expected benefits. Below is the summary of main APC benefit sources
for the units listed in the previous Section:
• CDU-5 and CDU-6—overall yields improvement towards light ends, in particular, CCR feed-
stock;
• CCR—stabilization and tighter control of reformate octane for improved gasoline blending;
• FCCU—yields improvement towards cracked naphtha and propylene;
• Alkylation—higher alkylate octane.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

1936 LOGUNOV et al.

These expectations were converted into hard monetary benefits that showed the payback of
all APC projects within 5–8 months. It should be noted here, that the evaluation of the benefits
attained looks easy for the units with single product of a tangible market value. However, all the
abovementioned process units produce intermediate products—feedstock for downstream units and
blending facilities. LP-based refinery planning systems, such as PIMS or RPMS are the best toolkit
for estimating APC benefits at such units in monetary terms NNOS, which operates RPMS since
1995 and has a proven refinery LP model, was the Russia first refinery to employ it for APC benefits
estimation. The methodology was developed in parallel with the first APC project implementation;
the benefits were calculated based on the trial operation (sustained performance test) data. The
methodology has been revised and improved since, and the benefits are being recalculated on a
regular basis to ensure their sustainability.
APC benefits for a process unit are not limited to capacity increase, yields improvement, or energy
conservation that can be measured as tangible savings It often happens that more sustained product
quality at one unit delivers tangible benefits at downstream units. An example from NNOS is its
blend facilities with online optimization based on Honeywell OpenBPC software. APC stabilizes the
quality of blend components coming directly from process units, and this is a source of additional
benefits. APC benefits maybe also generated at higher levels of optimization (group or plantwide)
where the underlying APC applications contribute to the overall synergetic effect captured by
upper-level solution(s).
APC provides a variety of intangible savings as well. These include:
• Reduction in plant trips;

• Increased equipment life;

• Reduced downtime for maintenance;

• Improved management information;

• Reduced operators’ workload (smaller number of process variables to control).

Finally it is hardly possible to quantify the leap in production culture and the shift in thinking:
with such power tool as APC, the engineers can now focus on systematic optimization of operations
rather than plugging process and equipment “holes,” as before.

5. APC BENEFITS GUARDIANSHIP VS. DETERIORATION

The deterioration of APC applications, its causes and the ways to prevent or mitigate
are extensively debated in the global APC community [3, 6].
As early as during CDU-5 and CDU-6 APC project implementation, NNOS has realized the
importance of establishing its own team for further maintenance, benefits guardianship and further
development of APC applications. Engineers recruiting, their training, and involvement in ongoing
projects is a multifold task, but the right placement of the APC team within the organization is
all the more important. The issue was resolved by the Refinery’s management in an effective and
timely manner. By the commissioning time of the first two APC systems, the Refinery had a trained
APC engineer on staff with the experience of working with Honeywell and the ambitious intentions
to design and implement APC applications in future. Over the following years, the Refinery’s APC
team has grown up to three employees—not too many against the increasing number of applications
to maintain as well as the support of new APC projects and the Team’s own work on upgrading the
existing APC applications and developing new ones. The latter deserves a special comment: three
of the 10 APC systems operating by 2015, namely, the Gas Plant CDU-1, and VGO Hydrotreater,
were designed and implemented by the NNOS APC team based on Profit R
Controller software with
Honeywell’s minor support, supervision and consultancy. The team has also undertaken a significant
modernization of APC applications at CDU-5 and CCRU.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1937

The placement of an APC team in the company’s organizational structure is not as simple a
task as it may seem. APC helps to attain operation targets faster and more efficiently. There-
fore, helping and advising operators and process supervisors about the correct and effective use
of this powerful tool is a paramount task for APC engineers. Their assistance to the operations
includes:
—selecting a relevant objective function for the optimization if more than one were specified at
the design phase;
—monitoring APC health in terms of the QE models validity compliance of predictive models
and tuning parameters with the current state of operations;
—advising operators about correct limits, setpoints, QE updates, etc., helping to resolve problems
they may face;
—planning development and/or modernization of APC application, software release updates, etc.,
in contact with applications developer.
The Refinery has positioned its APC team in the production planning group, i.e., at the “cross-
road” of planning and production functions. Further evolution of APC activities has proved the
expediency of the decision. It was later successfully replicated across other LUKOIL refineries and
eventually resulted in the establishment of more formal corporate APC community headed by the
APC specialist from NNOS with the objective to accumulate and share the best APC operation
and maintenance practices within LUKOIL.
Maintenance and benefit guardianship of several APC systems is a labor-intensive, largely rou-
tine activity requiring, among other things, frequent visits to control rooms. To facilitate the APC
maintenance exercises Honeywell and NNOS have cooperated in the development and implemen-
tation of a simple but effective APC monitoring tool. The software collects data from Honeywell
Uniformance R
PHD real-time database via the Refinery’s corporate network and provides timely
information on all APC applications and their key functional elements in a convenient user-friendly
form. As a result, APC engineers can see on their computers:
—lists of all Refinery’s APC applications (multivariable controllers and QE) categorized by
process units;
—status of each multivariable controller and of its manipulated, controlled and disturbance
variables;
—status of each QE its statistical characteristics and output trends vs. lab data;
—copies of all messages sent by APC applications to operators.
The developed solution provides the concise status information about APC applications with no
or minimum analytics. For more in-depth monitoring and performance analysis Honeywell offers
Control Performance Monitor for APC (CPM APC). This software suite enables APC monitoring
and analysis based on key performance indicators (KPIs) both standard and user-specified. The
design concepts and user interface and of CPM APC are similar to Honeywell Control Performance
Monitor for base-level control (CPM PID) discussed in the next section.
For over 8 years of APC support and development NNOS has accumulated extensive observations
and practices which are unique for Russia. Here is a short collection of such observations in the form
of typical APC misconceptions and delusions.
• “APC reduces the number of operators.” This is disproved by the worldwide practice: APC

applications facilitate operators’ work but do not replace them. Reducing operations staff may be
a policy where APC deployment is among the least critical components.
• “Just make a turnkey contract for APC; the contractor will do the rest.” This statement is refuted

by this article, as well as by the experience described in [4, 5]; customer’s deeper involvement ensures
smoother applications buy-in and their longer lifecycle with sustained benefits.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

1938 LOGUNOV et al.
• “Turn on and forget.” A “forgotten” application will be sooner or later turned off by operators
for good—as unnecessary. Only qualified and systematic monitoring, maintenance and benefit
guardianship will prevent the deterioration of APC applications and the ensuing lost profit. The
variety of maintenance and benefit guardianship tools, from fairly simple to sophisticated, whether
local or cloud-based cannot fully replace a human individual. The tools are intended just to help
the authorized person by saving him/her valuable time for more creative and intellectual work.
• “A good QE is a substitute for lab analysis.” A QE can sometimes reduce the sampling frequency

but, generally, lab analyses are still necessary to check and maintain APC vigilance. Otherwise, the
blinding APC system will start knocking against the limits set by operators, and sooner or later
will be retired.
• “One or more QE demonstrating the high correlation with lab data can replace a multivariable

controller.” With few exceptions, that is not true. Indeed, even a single well-performing QE may
help operator to run the process without APC so it is valuable on its own. However, the world
practice confirms that an APC system runs a process unit better than operators even those ones
who check their actions against QE readings—no operator can handle the operation target changes as
smoothly as a properly tuned APC application does. Therefore, it may be reasonable to consider QE,
if available, as just the first step towards further APC deployment except for rare cases where
the implementation of APC is hampered by serious technical issues (process bottlenecks) or low
anticipated benefits.

6. BASE-LEVEL CONTROL PERFORMANCE: A FRAMEWORK FOR APC BENEFITS

APC rests upon base-level controls. An APC application may be considered as a master controller
for several PID loops (slaves). Therefore, the efficiency of the total control solution depends on the
BLC performance as well.
Honeywell offers Control Performance Monitor (CPM): an effective user-friendly tool for extended
PID monitoring from a single loop up to the whole enterprise. Along with Profit R
Suite, the solution
is a part of Honeywell process optimization portfolio.
Based on KPIs and the best practices, CPM allows to monitor, diagnose, identify, and remedy
possible BLC problems. Monitoring can be based on the data stored either in DCS or in real-time
databases, such as Uniformance R
, PI System, SQL, etc. The number of PID loops and their location
do not matter the availability of an appropriate network infrastructure is only required.
The system has a user-friendly interface (Fig. 3), including both overview frames (tree maps)
and windows with detailed information. Various types of KPI reports are generated, which can
be automatically placed in dedicated folders and/or sent to predefined addresses, including mobile
devices, with user-defined frequency and/or per requests.
CPM has a number of featured properties, such as:
—vendor-independence;
—a unified software platform for BLC (CPM PID module) and APC (CPM APC module) mon-
itoring;
—the availability of TaiJi PID tool for loop tuning (up to 5 interacting loops simultaneously);
—the convenience of creating plant-wide solutions, flexibility and easy configuring;
—workflow function, which allows to focus on critical loops;
—Honeywell’s extensive control performance knowledge base.
Due to its many advantages CPM is used by many global companies in processing industry such
as ExxonMobil, British Petroleum, ConocoPhillips, Quatar Petroleum BASF, Shell Chemicals, etc.;
some have chosen it as a corporate standard; CPM consistently holds the largest market share
among similar products.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1939

Fig. 3. An example of CPM user interface window.

Fig. 4. PID loop performance before and after tuning with the help of Honeywell TaiJi PID software.

NNOS was Russia first enterprise to implement CPM. The scope included CPM PID and TaiJI
PID modules. The solution was deployed on a dedicated server at the VGO conversion complex
(5 process units). The data for analysis come from the Uniformance R
PHD database through
the Refinery network. The controller settings calculated in TaiJi PID are transmitted directly to

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

1940 LOGUNOV et al.

DCS. The solution is operated by APC engineers through several workstations; the APC team is
responsible for the overall BLC performance at the VGO complex. Once a problem is detected
and identified APC engineers would either retune the loop themselves or inform the Refinery’s
instrumentation engineers. During CPM trial operation, several “bad actors” were identified, most
of them quickly and efficiently retuned with the help of TaiJi PID (Fig. 4). In the future it is planned
to extend the solution to other process units.

7. REFINERY-WIDE OPTIMIZATION

Plantwide optimization solutions comprising several APC systems have been around in oil refin-
ing industry for at least twenty years. These solutions typically employ first-principles models for
optimal steady states calculation and mechanisms to transfer controlled plants from one steady-state
condition to another along the simplest linear trajectory. Calculation results are being further sent
as setpoints/limits to related APC applications which then execute the new tasks with respect to
existing process constraints. Practice has shown that this approach is limited, basically due to high
first-principles model maintenance cost. With this in mind, Honeywell had developed an alternative
approach to multi-layered optimization on the basis of Profit R
Optimizer software included in the
R
Profit Suite family.
ProfitR
Optimizer allows to create a layered optimization structure over multivariable con-
trollers that are part of one or more APC systems. Its linear-quadratic objective function looks
the same as in Profit R
Controller. When APC applications connect to the higher-level optimizer,
their local optimizers are blocked and Profit R
Optimizer performs coordinated optimization of
all related controllers with regards to a combined set of constraints. As a result, the combined
optimization solution works more efficiently than the related APC applications altogether. A sim-
ple explanation looks the same as for the higher efficiency of APC application as against BLC.
An APC system acts as a “super-operator” with respect to the set of associated PID controllers,
each one addressing only its own variable. Similarly, a higher-level optimizer acts as a “super-
dispatcher” in relation to several multivariable controllers, each handling its own process unit or
section.
Dynamic optimization should be noted as Profit R
Optimizer’s key feature. Unlike traditional
solutions based on first-principles models; it goes on optimizing without waiting for the process
units to line out. This results in additional benefits generated during transients.
Not surprisingly NNOS pioneered Russian oil refining in the development of this high-tech
optimization solution. The VGO conversion complex was selected again as the test bench for a
pilot project. The obvious arguments in favor of this choice were:
(a) the unified control room with centralized control of the entire complex, and
(b) APC applications commissioned recently at its key process units.
The Profit R
Optimizer-based optimization solution comprises FCC and Alkylation units with
8 multivariable controllers altogether.
The solution addresses:
• common optimization tasks (as soon as the optimizer is on):

—stabilization of operation modes of both units due to the coordinated levels and product
qualities control
—minimization of C3+ in tail gas against sulfur content in product propylene,
—maximization of conversion on the C4 isomerization section (“Butamer”) of the Alkylation unit;
• mass balance improvement (any task from the below list per production planning department’s

choice dependent on the market demand), i.e., the maximization of:

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1941

Fig. 5. Optimization mode selection window in the higher-level optimization solution for NNOS based on
Honeywell Profit R
Optimizer software.

—cracked naphtha yield against the constraints such as propylene and C4 yields naphtha octane,
etc.,
—propylene yield against the constraints such as naphtha and C4 yields C3 splitter throughput,
etc.,
—alkylate yield against the Alkylation unit throughput and cracked naphtha yield,
—total processing margin of the VGO conversion complex.
All optimization handles of the complex are located at the FCCU, while the multivariable
controllers of the Alkylation unit add additional constraints to the whole solution.
In order to optimize product draws they are continuously calculated in accordance with the
daily targets received from the production planning department. This function is implemented in a
special computing module (a “virtual” controller), which receives all flowmeter readings corrected
for product density and measurement conditions.
For processing margin maximization, a linear objective function is calculated as:

a1 x1 + a2 x2 + a3 x3 + a4 x4 ,

where the coefficients ai are the current prices of cracked naphtha, alkylate, propylene and C4 and
the variables xi are the respective product draws. The prices come from the Refinery LP model twice
or thrice per month after the recalculation of the Refinery’s total product plan; APC engineers then
enter them into the optimizer.
The optimization software is deployed on the APC server of the Alkylation unit and OPC-
connected with APC applications of the FCCU. The main user of the system is the VGO complex
shift supervisor. A simple user-friendly interface enables easy switching between optimization tasks
(Fig. 5).

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

1942 LOGUNOV et al.

Fig. 6. Cracked naphtha yield has increased after the commissioning of the multi-unit optimization solution.

A year-long sustained performance test has shown a ca. 6-month payback period, predominantly
due to the increased naphtha yield (Fig. 6).
Profit R
Optimizer enables the development of plant-wide optimization solutions over the unlim-
ited number of multivariable controllers. Below are three examples, each comprising more process
units than in the NNOS case described above:
—Sinopec Shanghai Gaoqiao (PRC)—the optimization solution for atmospheric reduced crude
processing facilities including a crude and vacuum distillation unit, two delayed coking units, and
three FCCU;
—Chevron (USA)—the optimization solution for the Diesel fuel production complex covers crude
and vacuum distillation unit, visbreaking, and two Diesel hydrotreaters;
—Helennic Petroleum (Greece)—aromatics optimization comprises 2 naphtha rerun sections, a
catalytic reforming, an isomerization unit, and a benzene column.

8. CONCLUSIONS AND FUTURE WORK

The article has briefly reviewed Honeywell process optimization products with a case study of
the major Russian refinery. The case demonstrates impressive results obtained owing to the fruitful
combination of best-in-class software, contractor’s and customer’s profound engineering expertise
coupled with creative task-oriented cooperation, timely and tactful support by the customer’s
management deeply interested in further adoption of state-of-the-art advanced solutions and best
engineering practices in process automation.
NNOS plans for the near future include further deployment of APC and optimization in collabo-
ration with Honeywell at LC-35/11-600 fixed-bed catalytic reformer, two Diesel hydrotreaters, the
visbreaking unit, and, last but not least, the VGO conversion complex #2 comprising of FCCU,
Alkilation unit, cracked naphtha hydrotreater and the new VDU-2 vacuum distillation unit. The
further step will be the deployment of multi-unit optimization solutions at two Diesel hydrotreaters
and at the VGO conversion complex #2. In the longer run, deeper integration between APC and
Refinery’s production planning systems is possible based on Honeywell’s software.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020

 ADVANCED PROCESS CONTROL 1943

REFERENCES
1. Comacho, E.F. and Bordons Alba, C., Model Predictive Control , London: Springer Verlag, 2007.
2. Tatjewski, P., Advanced Control of Industrial Processes, London: Springer Verlag, 2010.
3. www.APC-network.com
4. Fairuzov, D.Kh., Bel’kov, Yu.N., Kneller, D.V., and Torgashov, A.Yu., Advanced Process Control System
for a Crude Distillation Unit. A Case Study, Autom. Remote Control , 2016, vol. 72, no. 13, pp. 345–354.
5. Makarova, T., Progress Makes a Profit, iTime, 2009, no. 1(11).
6. Kern, A.G., Summitting with Multivariable Predictive Control, Hydrocarbon Process., 2007, vol. 6,
pp. 63–64.

AUTOMATION AND REMOTE CONTROL Vol. 81 No. 10 2020