chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49

Contents lists available at ScienceDirect

Chemical Engineering Research and Design

journal homepage: www.elsevier.com/locate/cherd

Creativity and heuristics in process control

engineering

Leonid Yakovis a , Leonid Chechurin b,∗

a Department of Mechanics and Control Processes, St. Petersburg State Polytechnical University, Saint Petersburg,
Russia
b School of Business and Management, Lappeenranta University of Technology, Skinnarilankatu 34, PO Box 20,

FI-53851 Lappeenranta, Finland

a r t i c l e i n f o a b s t r a c t

Article history: We study the relationship between heuristics and standard tool application in the design of
Received 7 March 2015 process control systems. This relationship is illustrated by classical control history exam-
Received in revised form 12 June ples. Features of modern engineering objects are highlighted that challenge classical control
2015 approach and leave substantial space for creative solutions. We focus on inevitable heuristics
Accepted 9 July 2015 in control design for modern complex process systems. The study suggests new systematic
Available online 17 July 2015 approach to heuristic control design. The approach is based on simplification technique
and follow three steps. First, certain assumptions (e.g. negligible nonlinearity or dynamic
Keywords: behavior) help to develop the simplified model and problem setup. Second, standard control
Process control design is performed on this reduced model. Third, the perturbation theory methods help to
Creativity find corrections to the obtained control to cope with the real plant. Special attention is given
Heuristics to the paradigm of concurrent (or integrated) plant and control design. The study suggests
Concurrent design finding a proper compromise between investments in the process and its control. Several
cement manufacturing process examples demonstrate that this method can generate better
effect comparing to autonomous approach.
© 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction measured signals and inform operators on the current process

Creativity is known as the individual ability to accept and gen-
erate conceptually new ideas, ideas that differ to traditional or
conventional schemes. Whether this competence is required
for process control design engineers nowadays? Walking
through a modern enterprise we can hardly encounter a
human being. Should we see one, she or he is sitting in
front of a monitor with blinking interface of graphs, diagrams
and schemes. Everything seem to have been done and solved
already. We take a risk to advocate the opposite, assuming that
the most important problems of automations are to be solved
yet. What has been done is just a preparatory work for the most
difficult and desired goal: design of integrated intellectual
manufacturing control. Indeed, if we analyze the monitor’s
interfaces we conclude that most of them just visualize
status. Field engineers of different levels are mostly responsi-
ble for hard control decisions that influence the economic life
of the enterprise. And the choice of the proper control decision
is not an easy procedure since the problems are substantially
difficult. The result is the excessive use of energy or materi-
als, high reject rates, manufacturing plan failures. Why not to
use powerful methods of modern control theory? The theory
has already been successfully applied for control of such com-
plex objects like spacecrafts, airplanes and robots! This is the
main discussion point for the paper. We describe the difficul-
ties, which automation engineers encounter when applying
standard control theory in process automation practice. And
we present an approach how to overcome the difficulties
by combing the modern theory with heuristic (intuitive)
ideas.

∗
Corresponding author. Tel.: +358 504419193.
E-mail address: leonid.chechurin@lut.fi (L. Chechurin).
http://dx.doi.org/10.1016/j.cherd.2015.07.010
0263-8762/© 2015 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49 41

Fig. 2 – Modern reconstruction drawing of first invented

thermostat. The temperature in the incubator is
proportional to the vent size, that is changed by lever arm,
connected to the mercury plunge (drawing from http://nautil.
us/issue/12/feedback/the-vulgar-mechanic-and-his-magical
Fig. 1 – Idea of self-controlling roasting jack drive, from A -oven).
Treatise of Mechanics (1826). Rotation rate of the roasting
jack, connected to the belt F, is proportional to the fire (Fig. 3), although it could be just an adaptation of windmill
strength due to the turbine places in the chimney. speed controller known by that time (Fig. 4). The same idea
but based on deviations from gyroscopic axis was widely used
2. Automation evolution trajectory: From for mechanical course control devices for ships and airplanes
inventions to abstract science until the mid of XX. And we can find essentially the same idea
in the invention for the system of different physical nature:
In this paragraph we speculate on the subject of retrospective steam boiler feed system level control by the float linked to the
analysis of control system evolution as a stage of inventive valve. Various mechanical hydrostats can be found nowadays
and systematic design interplay. as modern analog of this level control system.
All known examples of early automation widely presented The general principle of all these inventions is called “nega-
in books on the history of technics are examples of inven- tive feedback control” or “deviation control” in modern control
tions, original engineering designs. These are artifacts of theory. Another definition that can be better understood by
technical creativity rather than science since they descrip- non-professional in the field is “self-stabilization”.
tion lack generalization. The principles of system design is not The beginning of the control theory is marked by the
there yet. Although they obviously demonstrate the knowl- study of new phenomenon emerged in first inventions on
edge and understanding of some basic physics: mechanics, automation. These are two well known studies of instability
hydraulics or thermodynamics (see for example, temple door of steam turbine controlled by the abovementioned governor
drive automation and other inventions of Heron of Alexandria,
10–70 AD). We can also cautiously state the almost all those
early invention are served as drives, power amplifiers or pro-
grammed control devices.
Ensuring constant level of a liquid turned to be the first
stabilization challenge for first engineers. A watercourse
dam could be claimed as the simplest example of a solu-
tion. Ancient water clocks and oil lamps with float level
regulator mechanism, known already by Greeks and Arabs,
demonstrate more sophisticated design for this action. One
of Leonardo da Vinci’s invention seems to have been the first
advanced automation design that provides motion stabiliza-
tion control (Fig. 1). Still, with contrast to other mechanical
inventions, Leonardo does not provide a general principle of
stabilization.
Another famous invention in different physical field is Cor-
nelis Drebbel’s temperature stabilization system or thermostat Fig. 3 – Watt’s governor principal scheme. Turbine steam
(1624), see Fig. 2. And possibly the most famous in control his- supply valve opening is proportional to the rotation rate
tory system for rotation rate stabilization is Watt’s governor due to centrifugal forces effect.
42 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
Fig. 4 – Windmill running stone gap adjustment
mechanism. The gap is proportional to the rotation rate due
to centrifugal forces effect (drawing form http://www.texva.
com/holland/The%20Dutch%20Windmill.htm).
done by Maxwell (1868) and Vyshnegradsky (1877). There two
successful analysis used differential equations to model the
mechanical system and a number of mathematical manipu-
lations to derive the stability region for system’s parameters.
The same roadmap became standard for the next era of
automation: invention of governing mechanism or device –
(when necessary) stability or performance analysis by math-
ematical modeling – device improvement. Here we have to
refer to another famous example: Harrold Black’s invention
of feedback amplifier (Black, 1932). The invention suggests a
breakthrough structure for electric signal transmission, where
the feed backing is used to attenuated exogenous noises. But
the patent reveals just a industry-specific solution only again
(check out terms in the patent “electric wave”, “current”, “volt-
age”, “amplifier” and not “controller”, “signal” etc.). Similar to
steam turbine governor improvement based on mathemat-
ical analysis, it was Black’s colleague, Henry Nyquist who
analytically derived the stability conditions for the feedback
amplifier and generalized the conclusions in the form of the-
orem (Nyquist, 1932).
A thoughtful description of this stage of control system/and
theory evolution is given by H. Black himself: “Pertinent to
invention in engineering is the familiar experience that inven-
tion fosters research and vice versa.” (Black, 1958).
A new trend in control theory evolution appeared in the
mid of the last century. At one hand, control theory became
the field of application of mathematics. At the other hand
the focus of the research in control theory moved to pure
mathematical problems migrating further and further from
the actual problems in industrial practice. The object of con-
trol became the set of differential equations, the research
question was how to influence their trajectories by controlled
inputs. As the result we obtained the academic theory of
control, some chapters of which are the monument for math-
ematical instrumentation. And at the same time, most of
solutions were true but useless. For example, formal meth-
ods of optimal control with the quadratic performance index
do provide the solution. But the controller is given in the form
of differential equations of high order. The solution is correct
but unrealistic for practical implementation: high derivatives
would amplify noises, that always exist. The studies in control
theory in second part of the last century could be described
by the paradigm inversion. Namely, from mathematical mod-
eling and analysis of equations for a given invention to the
adapting of formal mathematical solution (standard con-
troller) to the object that is believed to be unchangeable.
Inventive component in control moved to heuristic compo-
nent in finding a solution to a mathematical problem or a proof
for a statement. For example, the history of studies of meth-
ods to reduce the dimension of controller or object description
(reduction problem) experienced distinct paradigm changes.
In general, the approximation problem for a transfer matrix
G(s) looks for a transfer matrix Gred (s) of lower order that is
somehow close to the origin (let us assume the G(s) is scalar
for the simplicity of further discussion). The closeness can be
described by a norm
 
G (s) − Gred (s) .
However, the modal reduction method, known from 60s,
removed from G(s) components with higher oscillation modes.
So, the underlying idea had physical meaning. Then the
paradigm for model reduction became rather mathematical.
First,
 the optimal approximation
 in respect to quadratic norm
or G (s) − Gred (s) . was introduced in 70s, where
2

 ∞
  1  2
G (s) = 
 2 G (jω) dω.
2
−∞
Then balanced reduction approach came in 80s, with the
estimation of approximation in respect on H∞ norm (since the
H∞ norm is not smooth, the optimal approximation has not
been found). The H∞ norm of transfer function G(s) is defined
by
   
G (s) = sup G (jω) .
∞
ω
This norm, or Hardy space norm, became popular due to its
use for robust system analysis and design. Finally, optimiza-
tion with Hankel norm became the focus of approximations in
90s since it provides better H∞ approximation. We emphasis
that the heuristic design of solutions migrated from physical
to mathematical meaning.
Another example of paradigm shifts is robust control
synthesis. Several concepts were suggested (invented): poly-
nomial synthesis (frequency domain), state space methods
based on Lurie–Riccati equations solving, linear matrix
inequalities methods etc.
At the present time the amount of mathematically involved
papers on control theory is greater in powers than the amount
of patents. The efforts of practicing automation engineers
address the adaptation of rigor book solutions to real problems
like PID/PI controller tuning, approximate solutions to control
problems or stability evaluation, complexity reduction etc.
The interest to systematic creativity or conceptual design
in engineering is gaining momentum as well. Regarding chem-
ical engineering design, for example, we can refer to (Ottino,
2000; Eekels, 2002; Kossack et al., 2008). We strongly rec-
ommend an exhaustive review on approaches for chemical
process control problems presented by (Sharifzadeh, 2013). It
contains nearly all possible ideas for process control although
the heuristics seems to be mostly used to resolve mathe-
matical difficulties. In respect to systematic creativity, one
of the very popular approaches to generate successful inven-
tive ideas is TRIZ, the theory for inventive problem solving
 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
limestone raw mix
raw
clay
grinding
Fig. 5 – General scheme of cement processing.
(Altshuller and Shapiro, 1956; Salamatov, 2005). There are
studies on direct adaptation of TRIZ for chemical process
design (Srinivasan and Kraslawski, 2006). At the same time we
focus on creativity in process design and automation issues in
the presented paper.
3. Process engineering control problem
setup
The manufacturing automation problem can be given a short
form. Given the limited amount of resources, we are to maxi-
mize the profit by the material flow intensity and specification
parameter control subject to the restrictions on the output
quality and quantity are held.
Without the loss of generality, we consider continuous type
manufacturing process. Any manufacturing is the set of tech-
nological process over the material flows, divided by buffer
tanks. Cement processing scheme is given as the example in
the Fig. 5.
The general scheme for manufacturing control is given in
Fig. 6. Here the control inputs are the intensities of material
flows and technological parameters (e.g. fuel consumption,
raw mixture consumption, furnace rotation rate and air thrust
for roasting process). Process monitoring data and output
quality specifications are used to develop the control.
The difficulties of straightforward control problem solving
are originated by the complexity of the problem. As com-
plexity is known for bearing many meanings, we need to
provide the details. First, by complexity we mean the practi-
cally uncountable amount of possible variants of control that
makes the trial and error approach too time demanding even
for advanced computational resources. Second, complexity
stands for the interconnection of subsystems and system vari-
ables. It becomes hard (inaccurate) to single our certain units
or a subset of variables to reduce the dimension. Third, the
complexity reflects the nature of real control processes, that
flow
intensity
Manufacturing output
technology
parameters
Control Measurement
system system
Fig. 6 – General scheme for manufacturing control.
43
slag
cement
roasting
milling
clinker
cement
gypsum
cannot be described by the extensive language of linear analy-
sis. Thus, some internal nonlinearities may produce complex
even unpredictable behavior of system variables, that is hard
to describe in, for example, “classical” frequency domain.
We have to look for heuristic approaches to reduce the
complexity of the problem, to decompose or to reduce the
dimension, while keeping the same general setup.
4. Heuristic methods of complex control
problem reduction
There are many approaches to reduce the complexity of the
problem, where complexity means the amount of variables.
We refer again to the review (Sharifzadeh, 2013) that con-
tains the description of popular methods as well. We cluster
them into two basic strategies: decomposition and aggregation
(Pervozvanskii and Gaitsgori, 1988). The first means break-
ing down the problem into a number of simpler independent
ones, although linked by a common goal. Heuristic reason for
decomposition is the weak connection or crosstalk between
some subsystems of lower dimension. Aggregation is the
inverse of decomposition: a number of similar variables or
subsystems are replaced by one macro variable or system,
an aggregate. The dimension is reduced since the amount of
variables is reduced.
When designing the control system for a complex object,
the ideas of decomposition and aggregation produce multi-
level hierarchical system with block structure. The integrated
automation control scheme depicted in Fig. 7 provides an
example.
The hierarchy means that every control level (except the
highest, which is driven by demand) obtains orders from
the level above. Each level (except the lowest one) forms
the order for the lower level in the process of the real-
ization of the obtained order. Thus, in similar manner to
cascade control, systems develop orders to other systems.
Direct changes of material flows and regime parameters hap-
pen at the lowest hierarchical level. But if the whole system
is well designed, control goals at each level are harmonized
with the main strategic goal: economic optimization of the
enterprise (Yakovis, 2010).
5. Heuristic principles of inter-level
interactions in hierarchical control system
The diversity of raw materials and processing technologies
is very large. But the automation systems for planning and
control have many things in common regarding their struc-
ture, goals and functions. In addition, the frequency of control
interventions decreases as we go from lower levels to upper
(see Fig. 7). This is the base for heuristic reduction of the huge
44 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
Measurement Control
output planning
scheduling
operative control
distributed control of TP
unit 1 … unit N
…
Manufacturing
unit 1 … unit N
…
Fig. 7 – The scheme of multi-level hierarchical system for manufacturing control (MOM—Manufacturing Operations
Management, MES—Manufacturing Execution Systems, LIMS—Laboratory Information Management Systems,
EAM—Enterprise Asset Management, DCS–Distributed Control System, SCADA—Supervisory Control and Data Acquisition,
APC—Advanced Process Control).
dimension of initial problem. For example, the whole manu-
facturing is considered for material volume planning at the
highest level of control. This does not deal with “frequent”
control designs (material dosing in real time). It analysis
annual amounts of materials used in manufacturing only.
That makes the problem still difficult but solvable.
Lower level control directly manages material flows and
processing regime parameters. As the properties of processed
materials vary stochastically, the control is to be fast to react
them. That means the high frequency control signals. But
the technological processes are slow and inertial, therefore
the future responses of any control input is to be taken into
account. In contrast to planning that is based on static mate-
rial balance, the control design is to be based on dynamic
physical–chemical processes and dosing system models. It
complicates the control design a lot if we try to approach it
directly. Instead, placing intermediate buffer tanks between
process stages makes them independent. In general, manu-
facturing site with shop structure reflects the heuristic idea of
decomposition, where each shop can be considered indepen-
dently.
Even after decomposition or problems of lower control level
studied above, the control problem is still difficult since the
large amount of variables and dynamic properties of the sub-
systems. We are going to suggest a heuristic idea of further
manufacturing control decomposition by two-level control
design for certain processes (Yakovis, 2008).
Let us assume that the optimization problem is solved at
the upper control level while the lower level stabilizes the pro-
cess parameters within the given specification obtained from
above.
It is worth mentioning that in cost-sensitive optimization
problems the solution generally belongs to the boundaries
of the available set. For example, economy of cement batch
manufacturing tells to increase the share of slag usage (that
is cheaper than clinker) and to decrease the roasting tem-
perature (to save fuel). The same situation happens in beer
brewing, where water share in the product is to be maximized
for maximum profit. But theses trivial means are restricted
by quality specification. The latter is given the form of tech-
nological tolerances on various physical–chemical properties
year, quarter, ERP
month (Enterprise Resource
Planning)
week, APS
(Advanced Planning &
day Scheduling)
MOM,
day, shift,
MES,
hour LIMS, EAM
DSC,
shift, hour, SCADA,
minute APC
of intermediate and final products. These results in “bound-
ary” control regimes. But if the upper control level should not
make effective the “boundary” technological process orders
to the lower control level since there are disturbances always.
Since these disturbances are stochastic and therefore unpre-
dictable, the lower level is not able to observe the upper level
control specification. Therefore, the process parameters that
were supposed to be within specification boundaries, would
exceed them in 50% cases approximately. It will lead to quality
degradation, high reject rate, claims etc., in other words to seri-
ous economic losses. Thus the upper level control should take
into account certain parameter instability. It is done by setting
certain “emergency” margins, in other words by stepping back
from the dangerous boundaries inside the admissible set. The
described situation is shown in Fig. 8. The setpoint Ỹ of index
Y changes due to its instability.
The figure explains the economic meaning of stabilization:
the more index stability is attained at the lower control level
the closer the upper level controlled technological parame-
ter to the most profitable value Ȳ. It is worth mentioning
that incorrect “emergency” margin evaluation leads to eco-
nomic losses regardless its sign. Indeed, the underestimation
of the instability leads to quality drops while overestimation
to unnecessary deviation from the most profitable regimes.
If there many indexes (that is general case) to be optimized
and kept at the required level than two level control becomes
more complex but the effects of narrowing the admissible
regime parameter sets are the same for parameter instability
analysis. To illustrate this statement we consider the problem
with two indexes Y, that has to belong to a polygonal depicted
in Fig. 9. Some index values and the index gradient are also
shown. Obviously, the optimum of Y is attained at the least left
point of the admissible set Ỹ 1 . But it assumes no disturbances
that bring about index Y instability, that produce an elliptic set
around the setpoint Ỹ. Therefore, we have to choose much less
economic regime Ỹ 3 . Instead, having optimized the regime in
respect to stabilizing (stabilization reduced the size of the devi-
ation ellipse), we obtain the setpoint Ỹ 2 , corresponding to the
correct interaction between upper and lower control levels.
In addition to “emergency” margins evaluation we also
have to define proper priorities for stabilization at the lower
 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49 45

Fig. 8 – Economical meaning of output stabilization.

control into two levels where the upper one ensures the opti-
mal regime while lower one ensures stabilization. The scheme
of system interactions for this two level structure can be sug-
gested as in Fig. 11.
There are two information flows in the scheme. One down-
stream of program controls as well as index function forming
request from upper level to lower level. As the result, the lower
level system operates with the common to the whole sys-
tem optimization criteria, as usual. But the difference in the
scheme in Fig. 11 is that flows from lower level to upper one
are shown. In particular, these flows are used to evaluate those
emergency margins that define the restrictions of upper level
problem. This is typical situation for any neighboring control
levels in the multilevel hierarchical control system.

6. Heuristic approaches to real time control

problems

Real time stabilization control takes place at the lower level

Fig. 9 – Two-level process control. of the two-level control scheme. To simplify the problem we
split it into quasi-static regime optimization and dynamic
optimization. Upper level control can use static models of
level. The latter is illustrated in Fig. 10. Let the optimum is
process (they are corrected from time to time) and a num-
found at the least left point of two-dimensional admissible
ber of methods for numerical static problem optimization.
set as earlier. Let the deviation ellipse take the from of circu-
At the lower level simplification could mean usage of lin-
lar shape in the case that means that instability characteristics
earized dynamic models. Linearization is performed in the
are the same in both indexes.
vicinity of the regime, that the lower level ensures. Moreover, if
The optimal process regime without stabilization corre-
emergency margins are properly evaluated by the upper level,
sponds to the center of circle in the left triangle in Fig. 10.
then the lower level does not have take into account multiple
As this point is far from the least left corner, the economic
restrictions on the variables that define the process status. As
losses are high due to index instability. The middle part of
the result we can successfully use linear feedback control of PI
figure shows what the uniform stabilization can provide. Sta-
or PID type (Johnson and Moradi, 2005; Yakovis and Sporyagin,
bilization recourse is shared between two indexed equally.
2011, 2013) or more advanced control algorithms (Camacho
There is some effect (the circle center moves to the most prof-
and Bordons, 1999).
itable point) but it is small. Right part shows the achievable
However, there are still difficulties in control algorithm
effect if we spend all available stabilization resources on one
design and tuning. Although modern multivariable control
index only. For this, the upper level control has to provide not
theory is very powerful, it is almost unable to cope with the
only the optimal regime as well as priorities to the lower level.
set of typical features of manufacturing plants.
In other words, the upper level is to design the criterion for
The most challenging features that make the control anal-
optimal stabilization.
ysis and synthesis difficult are:
There are so called extremal objects and extremal control
systems (Ariyur and Krstić, 2003) where the index optimum
argument is inner point of admissible set. We do not con- • delays (output variables are not immediately affected by
sider this case here in details but economy losses due to index control variables). Time delays can be originated by contin-
instability take place here as well. It is also good idea to split uous nature of controlled objects and processes requiring

· · ·
Fig. 10 – The effect of non-uniform resource assignment of stabilization system.
46 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
Emergency
margin
evaluation
Statistical
processing
Fig. 11 – The scheme of two-level control system architecture.
partial differential equations for modeling or transport
delays,
• multivariablity of the dynamic objects, that can reflect the
continuous nature of the controlled plant again. Partial
differential equations may require many variables for finite-
element approximation for reasonable accuracy,
• subsystem’s interconnection, any of control input influence
every output,
• complex and time-variant output dependence from control
and disturbance inputs,
• stochastic and time-variant uncontrollable external distur-
bances,
• tough boundary restrictions on output and control vari-
ables,
• significant measurement noise, and
• “nonstandard” control quality criteria.
The most powerful MIMO control theory synthesis meth-
ods use state space approach (Åström, 2012). But this
representation is not comfortable to cope with delays and
boundary restrictions. Moreover, it has been already men-
tioned that they lead to controllers of high orders in general.
Heuristics and simulations can help an engineer in this sit-
uation. Control design (that ideally is be done simultaneously
with the plan design, we return to this point later) should be
assisted by modelling as well as control itself can use sim-
ulations in its algorithms (Perelman, 1978). The modelling
and simulation nowadays are well automated and does not
require intensive mathematical knowledge (see, for example
SIMULINK and MATLAB). Visualization of these modelling and
simulation tools is similar to typical SCADA interfaces, that are
known by automation engineers. A study of the role of simu-
lation and visualization in modern digital manufacturing can
be found in (Mourtzis et al., 2015).
Once the idea of control algorithm is found, the simulation
tools can numerically help to tune its parameters, to anal-
yses stability and robustness, to evaluate the performance
indexes etc. The focus of the problem so far is how to find the
conceptual idea for the control. And modern control theory
can help provide those principles for control concept design.
Namely, the basic approach can be derived by disturbed
system analysis methods. First, we consider idealized (sim-
plified) problem set up. Since many complicating details are
released, this generating problem is solved by classical control
Static process model
Regime Index function
optimization
Restrictions
system
Regime Index function
stabilizing
system Process dynamic model
Measurement
system
Controlled
process
theory methods. Then the obtained solution is heuristically
(inventively) modified and the closed loop performance is eval-
uated by simulations.
The question of the choice of the successful (neither over-
simplified nor lost in details) is also of heuristic nature.
However we formulate here typical patterns:
• Systems with weak nonlinearity. The “weakness of nonlinear-
ity” does not seem to have been given a rigor definition but
the main idea is the following. Let an operator A over x
be nonlinear A(x) = y. Let the operator A has a linear com-
ponent such A0 such as y0 = A0 x. If ||y − y0 || = ||A(x) − A0 x||
is small then “the main part of A(x)” is linear, so we
can call this nonlinearity weak. Thus, the basic model is
chosen to be linear (A0 ) and the control is designed by
standard linear control design procedures. Then simula-
tions are used to check the performance of the control on
actual nonlinear plant with A(x). If necessary, the linear
control solution can be corrected: the first approximation
is evaluated with neglected nonlinearities and the per-
turbed problem is solved again by linear theory methods
(Pervozvanskii and Gaitsgori, 1988).
• Systems under weak disturbances. The basic model assumes
zero disturbances. Then program control methods can be
used and, in particular, static optimization for stationary
systems. The upper level control provides the optimal pro-
gram or regime setpoints at the second stage. At the same
time the setpoint deviations are stabilized by additional
feedback controller, that attenuates relatively weak distur-
bances. Linear dynamic control synthesis methods can be
used for that based on linearized object description. This
heuristic idea was in fact implemented for two level process
control scheme described above.
• Systems with weak dynamic properties. The impulse func-
tion of stable dynamic objects is known to vanish as time
increases ad infimum. It means that the dependence of future
outputs from past inputs decreases with time difference.
Thus, the basic model can be one with the “truncated”
dynamic memory. Then the obtained solution is upgraded
with respect to truncated parts. For example, if the con-
trol inputs intervals in the basic model are longer than the
“truncated” memory depth, we arrive at simper problem of
static object control. Now, the control is to be upgraded by
lower level control design that moves the operation regime
 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
to the upper (static) level control setpoints as fast as possi-
ble.
• Systems with weak control. Let us consider manufacturing
systems that consist of a sequence of technological pro-
cesses with material flow between them. Since the volume
of buffer tanks is limited we face the complex problem of
simultaneous control design with one or several criteria.
These criteria describe the manufacturing and economic
indexes of the whole plant. Some of these controls (let us
call them “of first type”) are material consumption per time
unit. Others (let us call them “of the second type”) are regime
parameters for every technological process. It is reasonable
to assume that if all the controls belong to the admissible
set than the first type control inputs influence the material
output flow for some processes. Therefore we can simplify
the problem to so called operating control problem, in other
words, the problem of routing of material flows through the
technological processes. To solve it the regime parameters
are fixed first at certain nominal value, for example in the
median of admissible interval. Second, the initial manu-
facturing and economic criteria and the obtained routing
problem solution are taken into account: every local tech-
nological process is given a performance index. The best
regime parameters are defined as local optimization prob-
lem solution. The operating control problem solution can be
corrected at this stage since local regime parameters vary
the process productivity (Gaytsgori et al., 1986).
7. Systematic design of automated
technological complexes
The abovementioned features of technological processes seri-
ously complicate the control and force the control design
engineers to engage sophisticated and not always reliable con-
trol algorithms. A natural question appears: is it possible to
address control difficulties at the stage technology equipment
design? In other words, what if it is allowed to change the
plant to be controlled? These inventive plant (re)design, or
concurrent (simultaneous) plant and control system design, or
“design-for-control”, or integrated process and control design
would provide more radical means to increase control qual-
ity. Regarding the process control it would mean systematic
design of automated equipment or, in more general sense,
automated processing complexes (APC).
The systematic design of automated processing complexes
is becoming more and more popular topic among tech-
nology and automation designers both (Ambartsumian and
Kazansky, 2008). The general goal is to maximize the economic
output by concurrent choice of the structure and parameters
of technological complex (TC) as well as control system (CS).
There are studies on concurrent automated process design,
the problems of buffer tank volume choice is attacked in
Gel’fand et al. (1980) under condition of operating control and
maintenance of equipment. The problem of static multicri-
teria optimization is studied in Volin and Ostrovskii (2007),
where the choice of regime parameters during normal oper-
ation is assumed at the stage of chemical processes design.
(Sharifzadeh, 2013) has already been mentioned for its review
on integrated design practice. However, no general concurrent
“TC + CS” design method is known so far. Therefore, engineers
have to rely on heuristics, where it means the blend of science,
creativity and experience.
47
We consider an example of cement manufacturing plant
(Fig. 5) as an illustration. More precisely, we focus on the raw
material mixing followed by the roasting. The first heuristic
idea is that the roasting in the rotating oven is the process of
very complex physical and chemical nature. The development
of reliable dynamic model of such a process is very difficult
time demanding effort. Therefore, the problem of quality con-
trol of roasting process is difficult too. The less stable the
chemical stability of the raw material inflow the more difficult
the roasting quality control. But the mixing process is easy to
control as it is described by universal material balance equa-
tions. Thus, maximum efforts should be allocated to chemical
contents stabilization system design. Having reduced the vari-
ations of main external disturbances for the roasting oven,
we essentially simplify the control design for this sensitive
process.
Let us turn to the chemical contents stabilization problem
for raw materials to be mixed. We face chemical nonuniform-
ity of the materials that come from a quarry. Two approaches
can help to achieve the contents stability: “by control” and “by
technology”. “By control” means careful share control for the
materials to be dispatched to the mill. It will obviously main-
tain the required chemical contents. “By technology” means
just dosing the raw materials in constant specified proportions
and averaging the milled mixture in huge homogenizing tanks
with mixing facilities.
To control the mixed materials share we need frequent
contents measurements and precise dosing equipment. Sys-
tems for frequent control of bulk materials are expensive and
not precise. Therefore chemical contents sensors for milled
mixture are engaged. But it generates substantial transport
delay. The control system attenuates low frequency distur-
bances only due to this delay. On the other hand, we need
giant averaging tanks to reduce low frequency disturbances,
that are originated by raw material chemical contents vari-
ation. Capital costs as well as operation costs are too high
for these big volume averaging tanks. But they filter high
frequency contents fluctuations efficiently. These heuristic
analysis result in the conclusion that an optimal design would
combine the control system as well as averaging tanks of rela-
tively small volume. With this conceptual idea we can turn
to quantitative design by approximate analytical stochastic
control methods as well as numerical simulation.
The general scheme of this quantitative evaluation is the
following. First, the achievable (under control) output prod-
uct stability is estimated for each set of selected technological
parameters. Then comparing the derived estimation with
the specification we define admissible solutions. Finally, the
cheapest design is selected over all admissible designs. This
man-machine approach was implemented in practical design
of averaging-mixing complexes or cement manufacturing
plant (Gorenko et al., 1987; Doroganitch et al., 1989). MATLAB-
Simulink software toolbox was developed for analysis and
optimization of averaging-mixing automated complexes. The
toolbox perform multi-scenario simulations for design and
upgrading of cement manufacturing plants. These simula-
tions assist analytically based recommendations on the type,
amount and volume of averaging tanks, type and character-
istics of measurement tools, structure and parameters for
control system.
We have to emphasize that we need reliable mathematical
model of the processes at the stage of process design for the
abovementioned simulations. It seriously complicates the
design since the model identification requires some active
48 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
experiments performed on real plants. Designers can use
the mathematical models of similar plants in the case. It
should be also pointed out that the comparative evaluation of
various technological processes belongs to the earlier stages
of design and is aimed at finding the substantial effects only.
Thus, simplified and inaccurate models would be sufficient.
8. Conclusions
The role of creativity and one of its instruments, heuris-
tics, has been changing in automation and control system
design. The trend is illustrated by a sequence of historical
examples. This changes follow a spiral pattern. First, at its
“pre-scientific” stage, the automation systems were inven-
tions. The heuristic share in the design was maximal. The
progress of mathematics based control theory brought an illu-
sion that all automation and control problems would be solved
by its standard procedures. The complexity of modern intel-
lectual manufacturing process control forces design engineers
to consider search for creative and heuristic ideas along with
standard automation solutions.
At the same time, the study warns the design engi-
neers from another extreme. No serious industrial automation
problem can be successfully solved without control the-
ory assistance. Heuristics and numerical simulation are not
enough. Although without formulae, we report that heuris-
tic principles of decomposition and aggregating, hierarchical
control, disturbance theory based motion separation require
modern control theory instruments. But this instruments are
creatively (heuristically, inventively) applied. An interested
reader can find references with high mathematical rigor on
these approaches in the bibliography.
Finally, the study presents the concept of concurrent design
of the plant and control system. The traditional “sequential”
approach means that technology engineers, who are proba-
bly not the experts in control, develop the process and choose
the equipment. Afterwards, the automation team is able only
to beautify the picture by nice process information inter-
faces on monitors. New systematic design approach accounts
requirements and possibilities of automation at the early
stage of design or modernization of automated technological
complexes. It can generate substantial synergy and therefore
profits. Thus, the study adds to the discussion on creativity
role and place in process automation, provides general pattern
of heuristic moves as well as a roadmap for bringing together
creativity, control theory and computational resources.
Further work can be of two types. First, it can address the
development of more systematic approach to the heuristic
part of process and control design. The case study, best prac-
tices and literate analysis can produce better classification of
automation problems as well as heuristic solutions and, more
important, to link them to each other. It would make the cre-
ative part of design more controllable and predictable. Second,
further study and development of concurrent design or inte-
grated of process control algorithms are needed to be able to
convert the ideas into industrial practice.
Acknowledgments
One of the authors would like to acknowledge TEKES, The
Finnish Funding Agency for Innovation and its Finnish Dis-
tinguished Professor (FiDiPro) program that supported the
research.
References
Altshuller, G.S., Shapiro, R.B., 1956. About the psychology of
technical creativity (O пcихологии изобрetɑteльcкого
tворчectвɑ). Questions Psychol. 6, 37–49, Retrieved from
http://www.altshuller.ru/triz/triz0.asp.
Ambartsumian, A.A., Kazansky, D.L., 2008. Complex automation
of technological processes with involving event model in
feedback control scheme. In: Proceedings of the 17th World
Congress, July 6–11, The International Federation of
Automatic Control (IFAC), Seoul, Korea,
pp. 28–33.
Ariyur, K.B., Krstić, M., 2003. Real-Time Optimization by
Extremum Seeking Control. John Wiley & Sons, Inc., Hoboken,
NJ, USA, pp. 236p.
Åström, K.J., 2012. Introduction to Stochastic Control Theory.
Dover Publications, Inc., Mineola, NY, pp. 320p.
Black H.S. (1932). Wave Translation System. U.S. Patent 2,102,671.
Application filed in 1932. Patent Issued to Bell Labs in
1937.
Black, H., 1958. Invention in engineering. Elec. Eng. 77,
722–723.
Camacho, E.F., Bordons, C., 1999. Model Predictive Control.
Springer-Verlag, London, pp. 327p.
Doroganitch, S.K., Edvabnik, J.A., Shtengel, E.G., Jakovis, L.M.,
1989. Optimisation of parameters of automated process
complex for raw mix preparation. In: Proceedings of the
Second NCB International Seminar on Cement and Building
Materials, 30 January–3 February, New Delhi,
pp. 56–63, 4.
Eekels, J., 2002. On the logic and methodology of engineering
design. Chem. Eng. Res. Des. 80 (6), 615–624.
Gaytsgori, V.G., Itskovich, E.L., Pervozvanskiy, A.A., Sorkin, L.R.,
1986. Interrelation of on-line control and local optimization of
installations in continuous processes. Autom. Remote Control
47 (6), 851–860.
Gel’fand, Y.Y., Lur’ye, Y.M., Savitskiy, Y.P., 1980. Choosing the size
of buffer capacities in chemical processes with allowance for
maintenance management strategies. Autom. Remote Control
41 (1), 110–116.
Gorenko, I.G., Doroganitch, S.K., Yakovis, L.M., 1987. Multilevel
process control in multicomponent mixtures blending. In:
Proceedings of the 10th World Congress on automatic Control,
July 27–31, Munich, Federal Republic of Germany, pp. 218–222,
vol. 2.
Johnson, M.A., Moradi, M.H., 2005. PID Control: New Identification
and Design Methods. Springer-Verlag, London, pp. 537p.
Kossack, S., Kraemer, K., Gani, R., Marquardt, W., 2008. A
systematic synthesis framework for extractive distillation
processes. Chem. Eng. Res. Des. 86 (7), 781–792.
Maxwell, J.C., 1868. On governors. Proc. R. Soc. London 16,
270–283.
Mourtzis, D., Papakostas, N., Mavrikios, D., Makris, S.,
Alexopoulos, K., 2015. The role of simulation in digital
manufacturing: applications and outlook. Int. J. Computer
Integr. Manuf. 28 (1), 3–24 (2015).
Nyquist, H., 1932. Regeneration theory. Bell Syst.Tech. J. 11,
126–147.
Ottino, J.M., 2000. The art of mixing with an admixture of art:
viewing creativity through PV Danckwerts’s early work. Chem.
Eng. Sci. 55 (15), 2749–2765.
Perelman, I.I., 1978. Dynamical optimization in process control
based on causal forecast algorithms. Avtomatika i
Telemekhanika 9, 146–160.
Pervozvanskii, A.A., Gaitsgori, V.G., 1988. Theory of Suboptimal
Decisions. Decomposition and Aggregation. Kluwer Acad.
Publ., Dordrecht, pp. 384.
Salamatov, Y.P., 2005. TRIZ: The Right Solution at the Right Time:
A Guide to Innovative Problem Solving, second ed. Institute of
Innovative Design, Krasnoyarsk, pp. 254 (O. Kraev, Trans.).
Sharifzadeh, M., 2013. Integration of process design and control:
a review. Chem. Eng. Res. Des. 91 (12), 2515–2549.
 chemical engineering research and design 1 0 3 ( 2 0 1 5 ) 40–49
Srinivasan, R., Kraslawski, A., 2006. Application of the TRIZ
creativity enhancement approach to design of inherently
safer chemical processes. Chem. Eng. Process. 45 (6), 507–514.
Volin, Yu.M., Ostrovskii, G.M., 2007. Multicriteria optimization of
technological processes under uncertainty conditions. Autom.
Remote Control 68 (3), 523–538.
Vyshnegradsky, I.A., 1877. On Controllers of Direct Action. Izv.
SPB Tekhnolog. Inst.
Yakovis, L.M., 2008. Approximate solution for problem of
dynamical optimization of process control. In: Proceedings of
Sixth EUROMECH Nonlinear Dynamics Conference, June
30–July 4, 2008, Saint Petersburg, Russia (electronic
publishing).
49
Yakovis, L.M., 2010. Multi-level production control: modern state,
problems, prospects. Autom. Remote Control 71 (10),
2247–2258.
Yakovis, L.M., Sporyagin, K.V., 2011. Controller tuning for inertia
objects with time delay. Autom. Remote Control 72 (1),
208–217.
Yakovis, L.M., Sporyagin, K.V., 2013. Typical controllers
adjustment for multivariable objects of industrial automation.
In: Proceedings of Seventh IFAC Conference on Manufacturing
Modelling, Management, and Control, 2013, June 19–21, 2013,
Saint Petersburg, Russia, pp. 1792–1797, vol. 7 part
no. 1.