Decentralized proportional-integral controller based on dynamic decoupling technique using Beckhoff TwinCAT-3.1

International Journal of Electrical and Computer Engineering (IJECE)
Vol. 13, No. 3, June 2023, pp. 2721~2733
ISSN: 2088-8708, DOI: 10.11591/ijece.v13i3.pp2721-2733  2721
Journal homepage: http://ijece.iaescore.com
Decentralized proportional-integral controller based on
dynamic decoupling technique using Beckhoff TwinCAT-3.1
Nomzamo Tshemese-Mvandaba, Mkhululi Elvis Siyanda Mnguni
Department of Electrical, Electronics and Computer Engineering, Faculty of Engineering and the Built Environment,
Cape Peninsula University of Technology, Cape Town, South Africa
Article Info ABSTRACT
Article history:
Received Jun 22, 2022
Revised Aug 22, 2022
Accepted Aug 31, 2022
An improved technique for the design of decentralized dynamic decoupled
proportional-integral (PI) controllers to control many variables of column
flotation was developed and implemented in this paper. This work was
motivated by challenges when working with multiple inputs and multiple
outputs (MIMO) systems that are not controllable by conventional linear
feedback controllers. Conventional feedback control design consists of
various drawbacks when it comes to complex industrial processes. The
introduction of decentralization, decoupling, and many advanced controls
design methods overcomes these drawbacks. Hence, the design and
implementation of control systems that mitigate stability for MIMO systems
are important. The developed closed-loop model of the flotation process is
implemented in a real-time platform using TwinCAT 3.1 automation software
and CX5020 Beckhoff programmable logic controllers (PLC) through the
model transformation technique. The reasons for using the CX5020 as an
implementation environment were motivated by the reliability, and is built
according to new industry standards, allowing transformation, which makes it
more advantageous to be used more than any other PLCs. This is done to
validate the effectiveness of the recommended technique and prove its
usability for any multivariable system. Comparable numerical results are
presented, and they imply that industrial usage of this method is highly
recommended.
Keywords:
Model transformation
Modeling
Multivariable systems
Pole placement
Real-time implementation
Set-point tracking
This is an open access article under the CC BY-SA license.
Corresponding Author:
Nomzamo Tshemese-Mvandaba
Department of Electrical, Electronic and Computer Engineering, Faculty of Engineering and the Built
Environment, Cape Peninsula University of Technology
Symphony Way, Bellville, Cape Town, South Africa
Email: tshemesen@cput.ac.za, nltshemese@gmail.com
1. INTRODUCTION
The results presented by Tshemese-Mvandaba et al. [1] indicated possibilities of enhancing the
performance of multivariable systems using well-recognized controllers such as proportional-integral (PI) or
proportional-integral-derivative (PID) with relevant design methods and decouplers. The performances of the
closed-loop multivariable processes are affected by the interactions among the controlled and manipulated
variables [2]. If a multivariable closed-loop process presents minor interactions, the decentralized control
system is preferred and if interaction effects are more, then the decentralized control system with decouplers
is required to reduce these interactions. It is still challenging to implement the control applications of
multivariable industrial processes such as flotation columns, petrochemical, water purification, paper mill, and
many more, because of their complexity [2]–[4], and the special need for an intelligent optimized PID controller
to handle multi-purposes [4]–[6].

 ISSN: 2088-8708
Int J Elec & Comp Eng, Vol. 13, No. 3, June 2023: 2721-2733
2722
Flotation is commonly used in many industrial processes such as wastewater treatment to remove
contaminants that are otherwise difficult to separate like floating solids, residual chemicals, droplets of oil, and
fat, and in mineral processing to recover valuable minerals [7], [8]. Control of flotation processes is an
important area of research for the metallurgical industry, and according to [9]–[11]. The introduction of
flotation columns in mineral processing plants caught the attention of many researchers in the last two decades
[12], [13] indicating that froth flotation is an important method for the separation of minerals. These minerals
contained multiple metals such as copper or zinc and can be selectively extracted by using froth flotation. The
main control objective of the column flotation process is to improve the metallurgical performance to ensure
that the column operation comes to the reference values necessary for the specified recovery and position of
the concentrate stream [1], [11].
Control strategies aimed at controlling multiple variables at the same time need to consider the
interactions between controlled and manipulated variables. However, these strategies are not easy when real
plant implementation is required. Hence, most industrial implementations still make use of PID tuning
strategies, although they are time-consuming [12]. Many control strategies have been reviewed [5], [14];
however, this paper presents the implementation of a decentralized PI controller design based on a dynamic
decoupling flotation process using the CX5020 programmable logic controllers (PLC) with TwinCAT 3.1
software environment. The reason for using this PLC as an implementation environment is motivated by the
reliability of this platform and the fact that CX5020 is built according to new industry standards, allowing
transformation which makes it more advantageous to be used more than any other PLC. The application of
personal computers (PC) and PLC technologies produces better runtime results that can be easily implemented
in the industrial environment as presented in this paper. This paper aims to simplify the implementation of a
multivariable process by using a transformation approach that allows MATLAB/Simulink model to be
transformed into a PLC via function block programming. Control strategies applied to the column flotation
process are presented in different journals [15]–[17] presented two simple control strategies for stabilizing the
process. In one of those schemes, the froth layer height is controlled by manipulating the non-floated flow rate,
and the wash water is manually controlled, however, manual control of industrial processes needs to be
avoided.
The introduction of various control design strategies is triggered by various drawbacks of the
conventional feedback control design, especially with the introduction of complex industrial processes. Hence,
the introduction of the improved controller design, decentralization, and decoupling of a system to overcome
these drawbacks. Taking from the analysis presented by other researchers in the field of control, the paper
presented by [1] is based on advancing the most implemented conventional controller called proportional-
integral-derivative (PID), which has been widely used in a variety of processes. This paper is used to execute
the runtime implementation of a decentralized PI controller design based on a dynamic decoupling flotation
process using a PLC environment. PI and PIDs have been regarded as powerful strategies for regulatory control,
but if the process is interrupted and optimal operating conditions change continuously, the performance of
these techniques decreases. This issue arises from the fact that the PID does not use constraints, making it
difficult to adapt to changes. Their high sensitivity to interactions between process variables adds a further
drawback to these types of controllers. To compensate for these drawbacks, Shenoya and Kini [18] has
presented a robust control configuration for multi-inputs multi-outputs (MIMO) systems using decentralized
controllers. In processes with very complex dynamics, such as column flotation, traditional PI or PID control
parameter design are not sufficient to keep the plant in optimal conditions without improvement. Hence, many
researchers addressed these problems by complementing the PIDs with advanced control techniques such as
optimal control, decentralization, and decoupling design methods, PID [18]. The adopted control parameter
design allowed decent closed-loop response, and it requires some experience to decide which pole locations
are the best for any problem, without tuning requirements. This technique based on the pole-placement scheme
is used for the decentralized PI controller design to achieve the desired set-point tracking performance and
disturbance rejection.
The real-time implementation of the designed technique is useful for industrial applications [19].
The most convenient PLC implementation of the closed-loop controlled plant in a real-time environment
requires function block programming. Although this can be complex when dealing with multivariable
systems, however, this can be solved through direct usage of the models developed in Simulink.
MATLAB/Simulink is a well-established powerful tool for modeling and industrial used software [20], [21],
hence the solution towards runtime control implementation of this flotation column system is modeled in this
environment. Section 2 of this paper presents mathematical modeling and an overview of the transformation
approach. The results are presented in section 3, followed by the discussion in section 4 and the conclusion
in section 5.

Int J Elec & Comp Eng ISSN: 2088-8708 
Decentralized proportional-integral controller based on dynamic … (Nomzamo Tshemese-Mvandaba)
2723
2. METHOD
To succeed in model transformation, it is important to understand the software platforms involved in
the process of executing the task at hand [22]. In this case, MATLAB/Simulink for model development and
TwinCAT 3.1 for integration and runtime implementation were used. MATLAB/Simulink software can
generate codes from Simulink models to multiple targets using the Simulink Embedded Coder, formerly known
as Real-Time Workshop. The Simulink Embedded Coder, in combination with Beckhoff automation's
TwinCAT 3.1 target for MATLAB/Simulink (TE1400), enables the development of C++ code, which is
subsequently compressed into a standard TwinCAT 3.1 module format. TwinCAT Standard simply uses the
basic part of Visual Studio, with all its benefits in terms of processing and connecting to source code
management tools. As the name suggests TwinCAT Integrated itself into Visual Studio. With support for the
3rd
edition of IEC 61131-3, the C/C++, VB.NET programming languages, and links to MATLAB/Simulink in
Microsoft Visual Studio, it is possible to program automation objects in parallel. The MATLAB/Simulink
TwinCAT 3.1 Interface facilitates communication between MATLAB/Simulink and the TwinCAT 3.1 runtime
[20], [23]. TwinCAT 3.1 delivers real-time parameter acquisition and visualization. The selected interfacing
type displays the parameters and variables in TwinCAT 3's graphic interface, allowing real-time viewing and
modification in runtime mode. The TwinCAT System Manager has been integrated into the development
environment. Possible configuration, parameterization, program, and troubleshooting within the software
environment are shown in Figure 1 for TwinCAT runtime program modules in different languages.
Figure 1. eXtended Automation Engineering (XAE): language support
The following steps need to be appropriately executed for the transformation of the model from
Simulink to TwinCAT for real-time PLC implementation. Configuration: at this stage model Parameter
Configurations such as simulation time, fixed-step time, and sample time are configured. To some extent, each
computer-based model will have to use some discretization technique to reflect data flow in a physical system.
Hence, it is important not to have variable step time, as this can create instability or error in the real-time mode
of this operation. Code generation: the code generation and the building of the model process are started via
the Simulink menu, tools, code generation and build model. Then the building process will start and can be
monitored on the MATLAB workspace. Building process: the building process is initiated and executed via
code generation, as a result, a TwinCAT module is generated. The generated file (TwinCAT module) is then
used to build the TwinCAT component object model (TcCOM) module manually with Microsoft Visual Studio
in a TwinCAT environment.
The developed experimental setup is presented in Figure 2. Figure 2(a) illustrates the power supply
used to supply 24 V power to the PLC. On top of the power supply is the PLC used to achieve hardware
implementation. The PLC is used as a physical control device to control the flotation system. All parameter
setup and modifications are made on the device via the TwinCAT software as presented in Figure 2(b) and
then sent or downloaded into the PLC using serial communication. The investigations of real-time setpoint
tracking control for both the froth layer height and air holdup are made and various responses are presented
under the results section.
2.1. Model overview and decouple design
A graphical representation of column flotation, with a gas bubble generator system utilized for the
generation of the bubbles, is presented in Figure 3. Added at the top of the operational tank is a froth washing
system, whose purpose is to wash away the infected minerals from the valuable minerals within the froth [16],
[24]. The flotation column system as presented by [25], [26] is adopted, as it shows significant interactions

 ISSN: 2088-8708
2724
between the controlled and manipulated variables. From the matrix transfer function, the bias system is affected
by the transfer functions of both the froth layer height (ℎ) and the air holdup (𝜀𝑔) systems. With the focus on
simplifying the complex flotation system to the matrix transfer function presented by (1). This equation
specifies the dominant features of this flotation process.
(a) (b)
Figure 2. Experimental setup: (a) PLC and power supply and (b) industrial PC for engineering
Figure 3. Schematic diagrams of the pilot flotation column and associated instrumentation [1]
𝐺𝑃(𝑠) = [
h(s)
εg(s)
] = [
−(1.029×10−3s+2.3×10−5)
(s+4.02×10−4)(s+1.92×10−2)
−1.59×10−4s+4.33×10−7
(s+4.02×10−4)(s+7.981×10−3)
7.6×10−5
(𝑠+1.92×10−2)
7.78×10−5
(s+7.81×10−3)
] [
QW(s)
Qg(s)
] (1)
G11 and G12 are the transfer functions related to the froth layer height, and the transfer functions related
to the air holdup process in the recovery zone are G21 and G22. Since the flotation system is an interconnected
plant, a need to design a decoupling method to minimize the interconnections was identified. Figure 4
represents a 2×2 column flotation system with a decoupler model connected at the input of the plant model.
The desirable decoupling strategy to implement for the column flotation process would be dynamic decoupling,
thus, ensuring the process interactions are completely compensated for within the system [27]. The process’s
transfer function matrix (𝐺𝑝(𝑠)) requires the creation of a transfer function matrix 𝐷(𝑠), such that 𝐺𝑝(𝑠) • 𝐷(𝑠)
is a diagonal transfer function matrix 𝑀(𝑠) as shown in (2), and the resulting (7).

2725
𝑀(𝑠) = 𝐺𝑃(𝑠) • 𝐷(𝑠) (2)
[
𝑀11(𝑠) 𝑀12(𝑠)
𝑀21(𝑠) 𝑀22(𝑠)
] = [
𝐺11(𝑠) 𝐺12(𝑠)
𝐺21(𝑠) 𝐺22(𝑠)
] ∙ [
𝐷11(𝑠) 𝐷12(𝑠)
𝐷21(𝑠) 𝐷22(𝑠)
] (3)
Figure 4. Block representation of the decoupled controlled system
The matrix in (2) can be represented by its components, as shown in (3). The idea is to design the
elements of the decoupler in such a way that the interconnections between the process model outputs are
eliminated which will allow the design of the process controllers to be implemented in a decentralized dynamic
decoupled system. M12=M21=0 and D11=D22=1, so, setting the diagonal matrix for the representation of the
decoupled process is applied and summarized in (4).
[
𝑀11(𝑠) 0
0 𝑀22(𝑠)
] = [
𝐺11(𝑠) 𝐺12
𝐺21(𝑠) 𝐺22(𝑠)
] ∙ [
1 𝐷12(𝑠)
𝐷21(𝑠) 1
] (4)
Multiplication of the matrixes in (4) produces (5) and, using (5), solves the matrix M(s) for M(s) to be
diagonal. The obtained dynamic decouplers are presented in (6), with the model of the process. The obtained
decoupled model of the process is represented by (7).
[
𝑀11(𝑠) 0
0 𝑀22(𝑠)
] = [
𝐺11(𝑠) + 𝐺12(𝑠) ∗ 𝐷21(𝑠) 𝐺11(𝑠) ∗ 𝐷12(𝑠) + 𝐺12(𝑠)
𝐺21(𝑠) + 𝐺22(𝑠) ∗ 𝐷21(𝑠) 𝐺21(𝑠) ∗ 𝐷12(𝑠) + 𝐺22(𝑠)
] (5)
[
𝑀11(𝑠) 0
0 𝑀22(𝑠)
] = [
𝐺11(𝑠) 𝐺12(𝑠)
𝐺21(𝑠) 𝐺22(𝑠)
] . [
1 −
𝐺12
𝐺11
−
𝐺21
𝐺22
1
] (6)
[
h(s)
εg(s
] = [
𝐺11(𝑠) −
𝐺21(𝑠)
𝐺22(𝑠)
𝐺12(𝑠) 0
0 𝐺22(𝑠) −
𝐺21(𝑠)𝐺12(𝑠)
𝐺11(𝑠)
] = [
−873.68×10−6
𝑠+0.000402
0
0
6.6057×10−5
𝑠+0.00781
] (7)
The general model representation of the decoupled process with diagonal transfer function matrix as
shown in (2) and the completed decoupled model of the process presented in (7) might be undesirable
processes. However, the process to design the dynamic decoupled model for this system is explained in [1].
The transfer functions of the designed decoupled process are integrated with the controller design technique
presented in the following section. The description of the process followed when designing the decentralized
PI controller for the decoupled model of the flotation process is also explained in [1].
2.2. Transformed models and controller design technique
The process to design the dynamic decoupled model for this system is completed. Therefore, the
transfer functions of the designed decoupled process are integrated with the designed controller. The aim is to
design controllers to keep the system outputs as close to the target values as possible by reducing the errors
between input and output or feedback to zero at a steady state using shorter response times. The control integral
action is very helpful in eliminating the system’s steady-state error. As a result, it is only reasonable to utilize
a PI controller that ultimately has the transfer function (8).
𝐶(𝑠) = 𝐾𝑝(1 + 𝐾𝐼
1
𝑠
) (8)

 ISSN: 2088-8708
2726
where KP and KI are the controller regulation parameters demonstrating the controller gain constant and the
integral gain constant, respectively. The pole placement design technique purely attempts to find controller
settings that give the desired closed-loop poles. The arrangement is built in such a way that the control strategy
originates from the desired system response, making it simple to find the controller gains mathematically [1].
The closed-loop poles can be freely selected by adjusting the controller's gains KP and KI. As a result, the
system's controller transfer matrix C(s) is presented by (9).
𝐶(𝑠) = [
𝐶1(𝑠) 0
0 𝐶2(𝑠)
] = [
𝐾𝑝1(1 + 𝐾𝐼1
1
𝑠
) 0
0 𝐾𝑝2(1 + 𝐾𝐼2
1
𝑠
)
] (9)
The PI controllers' parameters for the process's loops are designed, using (7) and (9) for M(s) and C(s)
respectively. Thus, the decoupled system is built by including the two diagonal PI controllers. As shown in
(10) is the indication of M(s)*C(s), which is the decoupled model to be used further for the controllers’
parameter design.
𝑀(𝑠) ∗ 𝐶(𝑠) = [
−873.68e−6KP1(1+KI1
1
s
)
s+0.000402
0
0
6.6057e−5KP2(1+KI2
1
s
)
s+0.00781
] = [
h(s)
εg(s)] (10)
The technique followed when designing the decentralized PI control parameters for the decoupled
model of the flotation process is developed following the flowchart in Figure 5 [1], [28]. Generally, it is
necessary to specify where closed-loop poles must lie, and to calculate the controller parameters. Investigations
of the performance of the closed-loop system are performed to verify the system's capability for set-point
tracking. The control area is shown by two PIs in Figures 6(a) and 6(b) receive the error signal, which is
produced by the difference between the set point and the system’s output feedback signal. The control action
aims at correcting the offset to make sure each system loop follows any reference point.
Figure 5. Flow-chart of the summary pole selection procedure [1]

2727
(a)
(b)
Figure 6. Function blocks representation of the decentralized dynamic decoupled system model, (a) Simulink
block diagram and (b) TwinCAT engineering
Table 1 introduces various set-point changes used to evaluate the performance of the proposed
multivariable system. The set-points of ℎ = 60 cm and 𝜀𝑔𝑐𝑧 = 18% as tabled below are used as the first case
study of the investigations. The input/output environment within TwinCAT will allow real-time viewing and
modification in runtime mode.
Table 1. Outline Set-points for closed-loop systems’ performance evaluation
Simulated and
Implemented Study Case
Setpoint Transformed processes with PI controllers
h 𝜀𝑔𝑐𝑧 Froth layer height (cm) Air holdup (%)
1 40-60 (cm) 12-20-15 (%) Step set-point of 40 to 60 (cm) Pulse set point of 12-20-15(%)
2 50-70-60 (cm) 10-18 (%) Pulse changes of the setpoint
from 50 to 70 to 60 (cm)
Step changes of the setpoint
from 10 to 18 (%)
3 80-60-80 (cm) 4-5-4 (%) Rectangular pulse change of the
setpoint signal
Rectangular pulse change of the
setpoint signal
2.3. Model evaluations
The closed-loop flotation process as developed in Simulink is transformed into TwinCAT 3.1 software
environment for hardware implementation. Because the data and parameter connections are the same in both

 ISSN: 2088-8708
2728
platforms, there is a one-to-one correspondence of function blocks between Simulink and TwinCAT 3.1,
according to the transformation methodology. The real-time implemented cases are presented below, followed
by the results illustrated in section 3.
In case study 1, the set-point of the froth layer height is established from 40 cm and increased to
60 cm, while the holdup is set to start at 12%, increased to 20% at 10 seconds, and decreased to 15% at
60 seconds. The simulated and implemented results for this case are presented in Figures 7 and 8 under the
results section.
Case study 2 establishes the set-point of the froth layer height to start at 50 cm, the set-point change
is applied at 0.17 minutes to move the signal from 50 cm to 70 cm and decreased to 60 cm at 1.08 cm, while
the holdup is set to start at 10% and increased to 18% at 0.17 minutes, as presented in Figures 9 and 10,
respectively. In real-time or let us say real-life situations, this is applied by collecting a huge number of mineral
particles from the collection zone that would increase the froth layer. Then the drop of froth layer height at
1.08 minutes is made possible by adding wash water on the top of the column, and this is done to wash away
an unwanted mineral that might be present.
In case study 3, both set points of the air holdup and froth layer height altered their states by changing
the set point, to observe the system’s operational behavior. Figure 11 presents a setpoint change in the holdup
from 4% to 5% at 0.17 minutes and a decrease from 5% to 4%. The froth layer height in Figure 12 starts at
80 cm, while the applied holdup is at 4%. Irregular changes are applied at 0.17 minutes, where the set-point of
the layer height is dropped to 60 cm. Adding the wash water on top of the column result in a drop of froth layer
height. This is done to see how well the closed-loop system tracks any set-point or if the behavior of the set-
point change can be successfully tracked under any circumstances.
As specified by Ng et al. [29] and Vieira et al. [30] flotation is the process of splitting valuable
minerals from a mixture by creating a froth on which minerals separate. The following section presents the
simulation results and real-time PLC implemented results of the flotation presented in TwinCAT environment.
These figures are put next to each other for easy comparison of the results. The behavior and the values are the
same for both environments. This indicates a successful transformation between the two environments. The
transformation has made the PLC real-time implementation much easy if the configuration and the building
process are done correctly. The desired variables were achieved and viewed using TwinCAT 3 runtime mode
which integrates with Beckhoff PLC CX5020. The results are presented in the following section.
3. RESULTS AND DISCUSSION
A comparison of the obtained results demonstrated successful real-time PLC implementation results
as displayed in the TwinCAT environment. The designed controller based on dynamic decoupling is an
effective strategy to be used irrespective of the set-point variations and disturbance influence. Decoupling the
model of the process has proven to be an effective strategy to reduce the influence of the interactions in the
closed-loop control and consistently keep the system stable. The presented results have proven possibilities of
model transformation using MATLAB/Simulink and TwinCAT software environments. Figures 7 to 12 present
simulation results on (a) the left-hand side in comparison to Beckhoff PLC real-time implemented results on
(b) the right-hand side.
(a) (b)
Figure 7. Comparing the performance of the air holdup system: (a) Simulink results and
(b) real-time PLC results

2729
(a) (b)
Figure 8. Comparing the performance of the froth layer height: (a) Simulink results and
(a) (b)
(a) (b)

 ISSN: 2088-8708
2730
(a) (b)
(a) (b)
Since the data and parameter connections are the same in both platforms, there is a one-to-one
correspondence of function blocks between Simulink and TwinCAT 3.1, according to the transformation
methodology. Table 2 presents the performance indices simulated using Simulink software, and this is in
comparison with Table 3, which presents the real-time characteristics behavior. The Real-time results show the
effectiveness of set-point tracking control and disturbance rejection as assessed in this section. The desired
variables were achieved and viewed using TwinCAT 3 runtime mode. One of the reasons for using the
Beckhoff PLC CX5020 as an implementation environment was motivated by the reliability of this platform
and the fact that the hardware used for this implementation is built according to new industry standards and
allows transformation which makes it more advantageous for industrial use more than any other PLC.
Table 2. Performance indices of simulated decentralized dynamic decoupled for the flotation process
Study Case MATLAB/Simulink results analysis
Loops Rise time (s) Settling time (s) Peak overshoot MP (%)
1 Height 4.91 18.05 0.02
Air 3.4 66.17 0.05
2 Height 4.908 69.5 0.02
Air 3.4 15.0 0.05
3 Height 4.9 69 0.02
Air 3.4 66.3 0.05

2731
Table 3. Performance indices of the real-time decentralized dynamic decoupled for the flotation process
Case studies TwinCAT PLC real-time results analysis
Loops Rise time (s) Settling time (s) Peak overshoot MP (%)
1 Height 5.1 20 0.01
Air 2.55 69 0
2 Height 5.1 74 0.01
Air 2.55 14 0
3 Height 5 73 0
Air 2.55 68 0
The following are the real-time simulation results from the investigations and experiments conducted
for the decentralized dynamic decouple system. It can be noted from the characteristic behavior both loops are
not poorly influenced by the set-point changes applied in the holdup loop and froth layer height at the same
time. Comparing the characteristics of behavior in Tables 2 and 3, it can be noted that the decentralized
decoupled systems are a good option for the rejection of the random variations that might occur in the control
signal when the column flotation system is implemented in runtime mode. These results also agree with results
obtained from the Simulink environment. This implies that industrial usage of this method can be highly
recommended.
The suitability of closed-loop control systems following function block programming concepts is
demonstrated through simulations. The model transformation between the two environments
MATLAB/Simulink and TwinCAT 3.1 is created in this paper based on modeling, data analysis, and runtime
implementation. The ability to combine the MATLAB/Simulink control function blocks into the TwinCAT 3.1
function blocks for real-life industrial implementation has been demonstrated by the real-time implementation
outcomes of the closed-loop process for the presented situations and more performed investigations. The
paper's contributions are designed to give a foundation for understanding the principles of model
transformation, the TwinCAT software engineering and runtime environments, PLC implementation, and its
application to industrial distributed control systems.
The generated software model from MATLAB/Simulink environment to TwinCAT 3.1 environment
is utilized to perform the real-time execution of the closed-loop flotation process with different control
conditions to demonstrate the effectiveness of the transformation. Software integration of Simulink and
TwinCAT 3 gives possibilities of implementing linear or nonlinear controllers with MIMO processes in
runtime mode. The proposed controller design scheme and its application of PC and PLC technologies have
produced better real-time results in comparison with the classical controller design. These results are used to
motivate industrial use of the developed algorithm, other than classical control methods only. The good thing
about model transformation applied in this paper it automatically translates each state flow block into
customized basic function blocks with the inputs, outputs, and parameters as their Simulink counterparts.
Although the result could not be included in this paper due to the limited number of pages, it was noted that
the characteristic behavior of the system was not poorly influenced by the applied random disturbance and the
set-point changes applied at the same time.
4. CONCLUSION
Column flotation is a process in which different minerals are separated selectively. Investigations and
developments of the decoupled multivariable models of the column flotation process were conducted. A
comparison between the characteristics behavior of the closed loop decentralized coupled flotation system and
dynamic decoupled flotation system was conducted and presented before. Through these results it was noted
that the algorithm based on a decentralized-coupled system has limitations in terms of variations that can be
applied, random variation resulted in several overshoots. Hence, a decentralized dynamic decoupling approach
was introduced and simulated.
This paper’s focus is based on the real-time implementation using PLC, and TwinCAT real-time
environment in comparison to the simulation results. The modeling of the column flotation process and the
design of the control parameters for the decoupled flotation system were described. The model transformation
between MATLAB/Simulink and TwinCAT 3.1 is generated in this paper based on modeling, data analysis,
and runtime implementation that will be useful for the implementation of industrial projects. The ability to
combine the Simulink control function blocks into the PLC via TwinCAT function blocks for real-life industrial
implementation has been demonstrated by the real-time implementation outcomes of the presented closed-loop
flotation process. The generated software algorithms from MATLAB/Simulink to TwinCAT 3 engineering
environment are utilized to perform the real-time execution of the closed-loop flotation process with different
control conditions to demonstrate the effectiveness of the transformation. Different case studies based on the
closed-loop dynamic decoupled system are outlined in this paper. It can be noted from the characteristic

 ISSN: 2088-8708
2732
behavior both loops are not poorly influenced by the set-point changes applied in the holdup loop and froth
layer height at the same time. This indicates that the decentralized decoupled systems are a good option for the
rejection of the random variations that might occur in the control signal when the column flotation system is
implemented in runtime mode. Additionally, these results agree or are the same for both environments. The
combined usage of the Beckhoff PLC and TwinCAT 3.1 environments has opened more possibilities for the
industry to implement beneficial academic findings, through the integration of Simulink and TwinCAT 3.1
software. This technique can also be advantageous to be used for the industrial implementation of different
processes.
ACKNOWLEDGEMENTS
This work is supported by the THRIP/NRF project, “Centre for Substation Automation and Energy
Management System (CSAEMS) development and growth”, Reference No. TP211061100004.
REFERENCES
[1] N. Tshemese-Mvandaba, R. Tzoneva, and M. E. S. Mnguni, “Decentralised PI controller design based on dynamic interaction
decoupling in the closed-loop behaviour of a flotation process,” International Journal of Electrical and Computer Engineering
(IJECE), vol. 11, no. 6, pp. 4865–4880, Dec. 2021, doi: 10.11591/ijece.v11i6.pp4865-4880.
[2] E. C. Goud, S. Rao A., and M. Chidambaram, “Improved decentralized PID controller design for MIMO processes,” IFAC-
PapersOnLine, vol. 53, no. 1, pp. 153–158, 2020, doi: 10.1016/j.ifacol.2020.06.026.
[3] K. Anbumani and R. Rani Hemamalini, “Three-interacting tank controlled with decentralized PI controller tuned using grey wolf
optimization,” in International Conference on Communication, Computing and Electronics Systems, 2020, pp. 485–500. doi:
10.1007/978-981-15-2612-1_47.
[4] P. Quintanilla, S. J. Neethling, and P. R. Brito-Parada, “Modelling for froth flotation control: A review,” Minerals Engineering,
vol. 162, Mar. 2021, doi: 10.1016/j.mineng.2020.106718.
[5] X. Xu, Y. Tian, Y. Yuan, X. Luan, F. Liu, and S. Dubljevic, “Output regulation of linearized column froth flotation process,” IEEE
Transactions on Control Systems Technology, vol. 29, no. 1, pp. 249–262, Jan. 2021, doi: 10.1109/TCST.2020.2974430.
[6] A. Faruq, M. F. Nor Shah, and S. S. Abdullah, “Multi-objective optimization of PID controller using pareto-based surrogate
modeling algorithm for MIMOevaporator system,” International Journal of Electrical and Computer Engineering (IJECE), vol. 8,
no. 1, pp. 556–565, Feb. 2018, doi: 10.11591/ijece.v8i1.pp556-565.
[7] A. Bahadori, G. Zahedi, S. Zendehboudi, and M. Bahadori, “Estimation of air concentration in dissolved air flotation (DAF) systems
using a simple predictive tool,” Chemical Engineering Research and Design, vol. 91, no. 1, pp. 184–190, Jan. 2013, doi:
10.1016/j.cherd.2012.07.004.
[8] R. Bürger, S. Diehl, M. C. Martí, and Y. Vásquez, “Simulation and control of dissolved air flotation and column froth flotation with
simultaneous sedimentation,” Water Science and Technology, vol. 81, no. 8, pp. 1723–1732, Apr. 2020, doi: 10.2166/wst.2020.258.
[9] M. T. Carvalho and F. Durão, “Control of a flotation column using fuzzy logic inference,” Fuzzy Sets and Systems, vol. 125, no. 1,
pp. 121–133, Jan. 2002, doi: 10.1016/S0165-0114(01)00048-3.
[10] J. Bouchard, A. Desbiens, R. del Villar, and E. Nunez, “Column flotation simulation and control: An overview,” Minerals
Engineering, vol. 22, no. 6, pp. 519–529, May 2009, doi: 10.1016/j.mineng.2009.02.004.
[11] A. Riquelme, A. Desbiens, R. del Villar, and M. Maldonado, “Predictive control of the bubble size distribution in a two-phase pilot
flotation column,” Minerals Engineering, vol. 89, pp. 71–76, Apr. 2016, doi: 10.1016/j.mineng.2016.01.014.
[12] M. A. Ahmad, S. Azuma, and T. Sugie, “Performance analysis of model-free PID tuning of MIMO systems based on simultaneous
perturbation stochastic approximation,” Expert Systems with Applications, vol. 41, no. 14, pp. 6361–6370, Oct. 2014, doi:
10.1016/j.eswa.2014.03.055.
[13] M. Bilal et al., “Effects of coarse chalcopyrite on flotation behavior of fine chalcopyrite,” Minerals Engineering, vol. 163, Mar.
2021, doi: 10.1016/j.mineng.2021.106776.
[14] M. Maldonado, A. Desbiens, and R. del Villar, “Potential use of model predictive control for optimizing the column flotation
process,” International Journal of Mineral Processing, vol. 93, no. 1, pp. 26–33, Sep. 2009, doi: 10.1016/j.minpro.2009.05.004.
[15] L. G. Bergh and J. B. Yianatos, “Control alternatives for flotation columns,” Minerals Engineering, vol. 6, no. 6, pp. 631–642, Jun.
1993, doi: 10.1016/0892-6875(93)90117-6.
[16] L. G. Bergh and J. B. Yianatos, “Flotation column automation: state of the art,” Control Engineering Practice, vol. 11, no. 1,
pp. 67–72, Jan. 2003, doi: 10.1016/S0967-0661(02)00093-X.
[17] Y. Shen, W.-J. Cai, and S. Li, “Multivariable process control: Decentralized, decoupling, or sparse?,” Industrial & Engineering
Chemistry Research, vol. 49, no. 2, pp. 761–771, Jan. 2010, doi: 10.1021/ie901453z.
[18] K. P. Shenoya and S. Kini M., “Proportional-integral controller with decouplers for an interacting TITO process,” Turkish Journal
of Computer and Mathematics Education, vol. 12, no. 10, pp. 280–285, 2021, doi: 10.17762/turcomat.v12i10.4144.
[19] W. Dai and V. Vyatkin, “Redesign distributed PLC control systems using IEC 61499 function blocks,” IEEE Transactions on
Automation Science and Engineering, vol. 9, no. 2, pp. 390–401, Apr. 2012, doi: 10.1109/TASE.2012.2188794.
[20] C. (John) Yang and V. Vyatkin, “Automated Model Transformation between MATLAB Simulink/Stateflow and IEC 61499
Function Blocks,” IFAC Proceedings Volumes, vol. 42, no. 4, pp. 205–210, 2009, doi: 10.3182/20090603-3-RU-2001.0302.
[21] B. Adhikari, E. J. Rapos, and M. Stephan, “Simulink model transformation for backwards version compatibility,” in 2021
ACM/IEEE International Conference on Model Driven Engineering Languages and Systems Companion (MODELS-C), Oct. 2021,
pp. 427–436. doi: 10.1109/MODELS-C53483.2021.00066.
[22] L. Prenzel, A. Zoitl, and J. Provost, “IEC 61499 runtime environments: A state of the art comparison,” in International Conference
on Industrial Informatics, 2020, pp. 453–460. doi: 10.1007/978-3-030-45096-0_55.
[23] C. Yang and V. Vyatkin, “Transformation of Simulink models to IEC 61499 function blocks for verification of distributed control
systems,” Control Engineering Practice, vol. 20, no. 12, pp. 1259–1269, Dec. 2012, doi: 10.1016/j.conengprac.2012.06.008.
[24] Z. Li, M. Huang, W. Gui, and Z.-P. Jiang, “Data-driven adaptive optimal control for flotation processes With delayed feedback and
disturbance,” IEEE Access, vol. 7, pp. 163138–163149, 2019, doi: 10.1109/ACCESS.2019.2952396.

2733
[25] M. A. M. Persechini, F. G. Jota, and A. E. C. Peres, “Dynamic model of a flotation column,” Minerals Engineering, vol. 13,
no. 14–15, pp. 1465–1481, Dec. 2000, doi: 10.1016/S0892-6875(00)00131-X.
[26] M. A. M. Persechini, A. E. C. Peres, and F. G. Jota, “Control strategy for a column flotation process,” Control Engineering Practice,
vol. 12, no. 8, pp. 963–976, Aug. 2004, doi: 10.1016/j.conengprac.2003.11.003.
[27] B. A. Ogunnaike and W. H. Ray, Process dynamics, modeling, and control. Oxford University Press, 1994.
[28] S. Nalan-Ahmadabad and S. Ghaemi, “The design of pole placement with integral controllers for gryphon robot using three
evolutionary algorithms,” International Journal of Materials, Mechanics and Manufacturing, vol. 5, no. 2, pp. 127–131, May 2017,
doi: 10.18178/ijmmm.2017.5.2.303.
[29] C. Y. Ng, H. Park, and L. Wang, “Improvement of coal flotation by exposure of the froth to acoustic sound,” Minerals Engineering,
vol. 168, Jul. 2021, doi: 10.1016/j.mineng.2021.106920.
[30] S. M. Vieira, J. M. C. Sousa, and P. O. Durao, “Combination of fuzzy identification algorithms applied to a column flotation
process,” in 2004 IEEE International Conference on Fuzzy Systems (IEEE Cat. No.04CH37542), 2004, vol. 1, pp. 421–426. doi:
10.1109/FUZZY.2004.1375763.
BIOGRAPHIES OF AUTHORS
Nomzamo Tshemese-Mvandaba received a B.Tech. degree in electrical
engineering from the Cape Peninsula University of Technology in 2011 and her Master of
Technology degree from the Cape Peninsula University of Technology in 2015. She is currently
a doctoral student in the Department of Electrical, Electronic, and Computer Engineering, Cape
Peninsula University of Technology (CPUT), Cape Town. Her research interest is in linear and
non-linear multivariable system modeling, controller design, process control instrumentation,
automation, and PLC applications. In addition, she is a chair of IEEE PES WIP South Africa
Chapter. She can be contacted at tshemesen@cput.ac.za.
Mkhululi Elvis Siyanda Mnguni received his M.Tech. (2013) and Doctor of
Engineering (2018) degrees from Cape Peninsula University of Technology (CPUT), Cape
Town, South Africa. He is a lecturer/researcher in the Department of Electrical, Electronic and
Computer Engineering (DEECE) at CPUT, where he is also the leader of the Centre for Real-
Time Distributed Systems (CRTDS). His research interests are in the fields of power systems
stability, substation automation and control, digital substations, protection and control, IEC
61850 Standard Communication, process and instrumentation, adaptive protection schemes, and
smart grids. He can be contacted at mngunim@cput.ac.za.

Decentralized proportional-integral controller based on dynamic decoupling technique using Beckhoff TwinCAT-3.1

More Related Content

Similar to Decentralized proportional-integral controller based on dynamic decoupling technique using Beckhoff TwinCAT-3.1 (20)

More from IJECEIAES (20)

Recently uploaded (20)

Decentralized proportional-integral controller based on dynamic decoupling technique using Beckhoff TwinCAT-3.1