Available online at www.sciencedirect.

com

ScienceDirect
Procedia CIRP 41 (2016) 723 728

48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015

The effect of communications on networked monitoring and control

of manufacturing processes
Alexios Papacharalampopoulosa, Panagiotis Stavropoulosa,b, John Stavridisa, and
George Chryssolourisa*
a

Laboratory for Manufacturing Systems and Automation, Department of Mechanical Engineering and Aeronautics, University of Patras, Patras
26500, Greece
b
Department of Aeronautical Studies, Hellenic Air Force Academy, Dekelia Air-Force Base, TGA 1010 Athens, Greece

* Corresponding author. Tel.: +30 2610 997262; fax: +30 2610 997744. E-mail address: xrisol@lms.mech.upatras.gr.

Abstract
Manufacturing systems utilize data communications in order to monitor and control signals to be communicated either between the
process and the controller or within the Manufacturing Execution System (MES), in general. Such communications, mostly
wireless, which are usually studied in the framework of Networked Control Systems (NCS), have certain effects on the control
performance. The current study deals with the modeling of such a production subsystem as a NCS and with pointing out the main
aspects, namely delay in the signals due to queuing, packets loss due to routing and corresponding protocols, or even measurements
distortion under additive channel noise.
2015
2015 The
TheAuthors.
Authors.Published
Publishedby
byElsevier
ElsevierB.V.
B.V.

This is an open access article under the CC BY-NC-ND license

Peer-review under responsibility of the Scientific Committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
CMS 2015.under responsibility of the scientific committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015
Peer-review
Keywords: Manufacturing Process; Networked Control System; Monitoring System; Measurement Error

1. Introduction
Both control and monitoring are applied to
manufacturing processes achieving a better performance,
especially in terms of cycle time [1] and energy
consumption [2]. However, modern manufacturing
systems require more complicated loop schemes (Figure
1), since remote control techniques are adopted [3, 4].
Networked control may seem to be the solution to the
remote control of manufacturing; however, according to
the specifications dictated by the networked control [58], there is inevitable intervention in the control scheme
and as a result, the performance of the controller is
affected [5]. It is then, a matter of the controllers design
to add robustness with respect to communication and
networks configurations. Regarding the technical details
of data communications, a review of the industrial
networks structure is presented in [9]. This offers a
qualitative overview of the communications complexity.
A fact that has to be pointed out is the communication
through wireless ad-hoc networks [10] and the
controlling of routing or the topology.

Fig.1. Generic schematic of networked control

Lacking in a pre-existing infrastructure, such as routers

or access points, each node has to be used in routing by
forwarding data to other nodes. This, in combination
with the throughput and the queued data, causes a delay
in the signals/data. The exact mechanism, however, of
the way that communications interfere with control, is
explicitly discussed in Section 3.
The networked control thematic area is a well
established topic; however, complicated techniques [5,
6] have not been applied to it widely. In manufacturing
research, various approaches have been made, such as
inclusion of holonic manufacturing execution system in
networked production systems architecture [7] and
modeling of networked systems [8]. Specifically, in [7],
the work presented consists of a virtual enterprise,
utilizing network nodes to achieve heat treatment of

2212-8271 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the scientific committee of 48th CIRP Conference on MANUFACTURING SYSTEMS - CIRP CMS 2015
doi:10.1016/j.procir.2015.12.041

724

Alexios Papacharalampopoulos et al. / Procedia CIRP 41 (2016) 723 728

metallic materials. The concepts of "information

sensitivity" (and therefore non disclosure) and noncompliance of components are also discussed, since
semi-open systems are utilized. Finally in [8], the
appropriate augmentation of a reactor-distillation
systemic model is given for the application of the
networked control.
Regarding other specific applications of industrialized
networked control of manufacturing processes, a work
on networked control of CNC machines is referred to, in
[11]. OPC servers and transmission over TCP were the
main concerning issues after having also taken into
account the data transmission times. Results were
acquired regarding a test bench for the remote control of
a grinding process. Furthermore, robot control has been
taken into consideration in another work [12], where the
manager is able to control the robot position by remote
access to the operational server, through the internet.
The tests run, concerned a table with discrete
coordinates, which helped in adjusting the robots
position. Furthermore, in a more abstract level, Flexible
Machining Cells have also been addressed [13]. Web
pages (in terms of HTML interface) can be designed and
created by the users, and then be uploaded along with
the cell control program. Therefore, new complex taskoriented options are introduced for a cell, consisting of a
lathe and a milling machine. Finally, an internet-based
quality control approach is presented in [14]. During the
production, inspection routines can be remotely adjusted
by operators and new features, part size variations, and
quality criteria can be handled.
As shown, some research has been performed in the
thematic area of Networked Control Systems; however,
it needs further investigation, especially regarding its
industrial applications. Firstly, the environment can be
different from that of laboratory results, whilst the nature
of processes causes different model requirements [7].
Secondly, there is a need for theoretical and
experimental work on the proof of concept that the
efficiency, the repeatability, the precision and the
robustness of the controllers towards communicational
effects can be feasible [3]. Finally, the controllers design
(and implementation) has to be simple, by engaging
controllers that will take into account the framework of
"producing more with less" [15].
In this study, the main effects of networks
communication on manufacturing systems performance
are indicated, mostly in terms of process quality [1]. The
subject of quality is addressed using specific
performance indicators, defined in dynamic systems
theory. First, an illustration is given of the way that a
controller succeeds in enhancing the performance of
manufacturing processes. Simple control schemes,
achieving low level of both implementation and design
complexity, have been taken into consideration.

Fig.2. Open (a) System & Closed (b) Loop

Fig.3. (a) response of open system (b) response of closed system of

Fig. 2

The following sections deal with the main cases of

factors to be blamed for, namely the effect of delays on
control signals, the packets loss (drop-out) when
communicating through networks and the measurement
distortion.
2. Manufacturing Systems & Control Design
The main objective of a controller is to enhance a
system's (process) performance (or occasionally its
stability) [16]. Herein, a simple feedback scheme is
adopted (Figure 2), utilizing through a sum ( 6 ) simple
static gains as a tracking [16] controller (K) and a
steady-state-error [16] fixing controller (K1). The
desired signal (u) is considered being a discrete-time
step function, while the system that has to be controlled
(TF) is governed by the following transfer function

G (z)

0.75
,
0.75  z 1

(1)
1

utilizing the lag operator z .

The responses of both the open and the closed system
are shown in Figure 3. It is apparent that the steadystate-error has been improved and the rise-time has been
reduced; both of which lead to better performance
tracking [16].
From a manufacturing point of view, the
correspondence between the variables used herein and
the physical measures could be as in indicative cases
found below:
x Laser welding [17]
The systems input could be the laser power (in
Watts), while its output and input could be the actual and
the desired depth of the weld (measured by melt-pool
front in mm).

Alexios Papacharalampopoulos et al. / Procedia CIRP 41 (2016) 723 728

Fig.6. Variable delay in the forward direction

Fig.4. Manufacturing example of a closed loop controller: A laser
machining head

Fig.5. Manufacturing example of a closed loop controller: A heavy

load handling industrial robot

Fig.7. Closed system response - variable delay in the forward direction

x Robot movement [18]

The systems input could be the motor voltage (in
Volts), while its output and input could be the actual and
the desired position of a link (in rads).
In any case, the rise time refers to the time that the
phenomenon has been completed, whilst the steadystate-error is related to the process quality since the
desired process parameter value has been achieved.
The extraction of the relationship between the process
parameters is the first step missing. However, it is not
imperative that analytical modeling should take place.
Having a sufficient number of experiments, an
identification technique that would lead to a discrete
time model could be performed. It has to be mentioned
that the first order linear time invariant system chosen as
TF is the simplest case one can have to prove the
concept in this study.
In the present study, the controllers have been
calibrated semi-empirically, as a mixture of pole
placement and steady-state-error compensation, whilst
the values that have come up are K1=0.95 and K=1.35.

Consequently, in the case that the control/monitoring

signals are routed through such a network, the existence
of delays will have certain effects on the control
performance. It is noted here that the delay may take any
value. Therefore, the assumption in this section is that
the delay is an integer multiple of the sampling period.
Practically, this means that time-stamps have to be used
and an artificial delay has to be added to the signals. In
the figure below (Figure 6), the closed loop in Figure 2,
solely for simplicity reasons, is equipped with variable
delay in the control signal. . The delay block's transfer
d
function is z , utilizing the stochastic variable
d ^0,1, 2, 3` of discrete uniform distribution.
This systems response is shown in Figure 7. It can be
observed there that the rise time has not been affected,
however, an oscillation with damping has occurred and
thus, an overshoot of 30% is introduced [16]. The
steady-state-error, on the other hand, is the same
(approximately equal to 0 dB).

3. Delays Effect

The stability definition [16, 20, 21] depends on the

modeling. This is synoptically shown in this section, by
using as empirical stability criterion. As it can be seen
through the simulations run, the use of a constant delay
D
z , results in the unstable behavior for D>2. However,
d
when using a variable delay z with the stochastic
variable d being:

3.1. Effect on Performance

According to the standards of networks [19] and the
controllers design and implementation, various kinds of
delays can take place within a network such as:
x transmission
x propagation
x queuing
x computation

3.2. Effect on Stability

0, with prob. 50%

Dmax , with prob. 50%

725

726

Alexios Papacharalampopoulos et al. / Procedia CIRP 41 (2016) 723 728

Fig.8. Response of closed system with stability effect for Dmax=30

Fig.11. Quantizing the control signal

even for Dmax=30, the system remains stable, despite its

behavior being far from tracking a constant unitary
input, as seen in Fig. 8, below. In addition the time that
the system needs to settle back to "normal", in this case,
is one order of magnitude larger, resulting in a far slower
process.
4. Packets Loss
Packets Loss [22] is another phenomenon, causing
circumstantial communication black-outs and therefore,
irregular control signal absences. In Figure 9, this is
simulated. A switch has been added to the closed loop
system of Figure 2, causing the control signal to be
either equal to zero (with probability 50%), or equal to
the actual control signal, acquired from the loop scheme.
The constant zero value is denoted with a (C) block.
This systems response has quite a deteriorated
tracking performance. As seen in Figure 10, an undamped oscillation takes place, whilst the steady-stateerror has worsened, leading to a response mean value of
approximately 0.8.

Fig.9. Simulating packets drop-out

Fig.12. Response of closed system with quantized control signal

(Figure. 11)

5. Measurement Distortion
Measurement or control signal distortion is another
case of network-related devices introducing performance
deterioration. A specific case is the quantization of the
corresponding signal. This is shown in Figure 11, where
a quantizer with quantization intervals, equal to 0.125,
has been used to change the control signal. The unitary
delay in the feedback branch has been added for
solvability reasons.
The corresponding response is shown in Figure 12,
proving to be better than that in the other two cases.
Only a small-scale oscillation has been added to the
signal. Nevertheless, the systems output does not seem
to converge to the desired value, indicating a lack of
damping in its behavior.
Apart from quantization, signal distortion may also
occur due to communication errors. The link between
quantization and channels noise, however, can be the
signal-to-noise ratio (SNR). According to the data
communications theory [23], the SNR in dB, due to
quantization (also known as SQNR), is estimated by the
following formula:

SQNR | 6.02 N
where N is the number of quantization bits.
Fig.10. Response of closed system with packets drop-out (Figure 8)

(2)

727

Alexios Papacharalampopoulos et al. / Procedia CIRP 41 (2016) 723 728

Table 1. Performance Indicators values for delay effect

1.4

Performance
Indicators

1.2

Delay Dmax (Fig. 7) in

samples
2
4
6
8
8
34
38
46

step responses

Response Time (R)

in [msec]
Overshoot (S%)

0.8
0.6
0.4

target
closed - SNR=120dB
closed - SNR=60dB
closed - SNR=20dB

0.2
0

0.005

0.01

0.015
time [sec]

0.02

0.025

17

49

(-)59

69

Table 2. Performance Indicators values for Packets Drop-out effect

(N/M denotes Non-Measurable, as the output behavior is stochastically
oscillating)
0.03

Fig.13. Responses of closed system with AWGN in the control signal

for various values of SNR.

Nevertheless, the simulation of a noisy channel and the

cyclic redundancy checking exceeds the purposes of this
study, so additional white Gaussian noise (AWGN) is
put into the control signal. In the following figure
(Figure 13), various responses are given to this case,
using SNR as parameter. The delay in the feedback
branch has also been recorded.
As observed, while the SNR increases, the systems
behavior, in terms of tracking performance, clearly
worsens. It should also to be noted that the value for
fixing steady-state-error is K=1.245 (due to the unitary
delay in the feedback branch). Visualization in Figure 13
is not the same as in the other cases, since a different
tool has been used for its extraction. The dots indicate
the value at the corresponding time points and the lines
in between are of first order instead of zero-th.
6. Control Performance Indicators and Sensitivity
Analysis
After the effects of the various delays in closed loop
systems have been examined, an approach for the
establishment of control performance indicators has been
performed. The declaration of these indicators has
enhanced a sensitivity analysis, which explored the most
crucial parameters (delays, packets loss, and
measurement distortion) that affected the performance of
such closed loop control systems, in the manufacturing
sector. The response time and the overshoot of the
system were selected as the most suitable indicators
based also on previous work regarding PID controllers
[24]. Response Time is defined as the time required for
the systems output to be bounded by a variation of 5%
of the final (steady state) value. Overshoot is the
percentage of the final value reached by the maximum
value of the output. The following tables show the
extracted values of the performance indicators for each
kind of delay.

Performance
Indicators
R
S%

Packets Drop-out Probability (Fig. 9)

0,1%
1%
10%
50%
4
5
N/M
N/M
0
(-)25
(-)44
(-)58

Table 3. Performance Indicators values for Measurement Distortion

effect (P/I stands for Potentially Infinite, as the output seems to
perform a damp-less oscillation)

Performance
Indicators
R
S%

Quantization Interval (Fig. 11)

0.001
0.01
0.1
0.5
0.001
0.001
P/I
P/I
0
0.001
3
13.5

In order for each of the obstacles on the wireless

communication of networked manufacturing systems to
be better visualized and for conclusions to be drawn
about the sensitivity of the systems, in terms of
performance indicators, the following diagrams have
been deployed.
7. Conclusions, Challenges & Outlook
Nowadays, manufacturing systems can be quite
complex and as a result, the current trends have asked
for communication among various units. The need for
remote control, to a great extent, has also contributed to
this. An implication of data communication can be the
deterioration of the controlled systems performance.
Consequently, the robust design of such systems and / or
controllers must inevitably take into consideration any
structural
modeling
uncertainties
related
to
communication, namely delay, distortion or packets
drop-out.
Moreover, a sensitivity analysis for the most crucial
parameters should be performed in order for the
engineers to be able to choose the right hardware
protocols thus, avoiding the aforementioned effects in
their systems operation. Additionally, even a greater
challenge is the use of techniques and controllers that
require design and implementation in accordance to lean
manufacturing, while maintaining a non-complex
structure. The use of static gains as controllers seems to
be addressing these kinds of criteria.

728

Alexios Papacharalampopoulos et al. / Procedia CIRP 41 (2016) 723 728

Fig.14. Sensitivity of a first order closed loop system to the

communication uncertainties.

Further research has to be conducted, however, in order

for this to be performed for the cases of higher order, or
even non-linear systems.
Furthermore, different production performance
criteria may have to be used. Apart from quality implied
to the examples of this study and expressed by control
performance indicators (Overshoot and Response Time),
additional criteria such as cost, flexibility or even energy
consumption should be included. In such a case,
different kinds of studies have to be combined. Control
in the networks routing could be an additional measure
to all the above.
In addition, the controllers will have to be adaptive
since several models of manufacturing processes are
case-dependent comprising many uncertainties, which
may even be up to the manufacturing conditions (i.e.
dry/wet in the general case of machining) and process
parameters (cutting depth in the case of turning).

Acknowledgements
The work reported in this paper was partially
supported by EU Project "Multimodal spectrAl control
of laSer processing with cognitivE abilities - MAShES",
under Grant agreement no: 637081.

References
[1] Chryssolouris, G., 2006. Manufacturing Systems: Theory and
Practice. 2nd Edition, Springer-Verlag, New York.
[2] Paralikas, J, Salonitis, K, Chryssolouris, G., 2008. Robust
optimization of the energy efficiency of the cold roll forming process.
IJAMT, Volume 69, Issue 1-4, pp 461-481.
[3] Michalos, G, Kaltsoukalas, K, Aivaliotis, P, Sipsas, P, Sardelis, A,
Chryssolouris, G., 2014. Design and simulation of assembly systems
with mobile robots. CIRP Annals-Manufacturing Technology, Volume
63, Issue 1, Pages 181184.
[4] Larreina, J., Gontarz, A., Giannoulis, C., Nguyen, V.K.,
Stavropoulos, P., Sinceri, B., 2013. Smart Manufacturing Execution
System (SMES): The Possibilities of Evaluating the Sustainability of a
Production Process. (GCSM) 11th Global Conference on Sustainable
Manufacturing, Berlin, Germany, pp. 517-522.

[5] Dritsas, L, Tzes, A., 2009. Robust stability analysis of networked

systems with varying delays. International Journal of Control, Vol. 82,
No. 12, pp. 23472355
[6] Garcia, E, Antsaklis, PJ., 2013. Model-Based Event-Triggered
Control for Systems with Quantization and Time-Varying Network
Delays. IEEE Transactions on Automatic Control, Vol. 58, No. 2.
[7] Saint, G.B., Valckenaers, P., Van Brussel. H., Van Belle, J., 2013.
Networked manufacturing control: An industrial case. CIRP Journal of
Manufacturing Science and Technology, pp.324326
[8] Xu, S., Bao, J., 2011. Plantwide process control with asynchronous
sampling and communications. Journal of Process Control 21, pp. 927
948.
[9] Galloway, B., Hancke, G.P., 2013. Introduction to Industrial
Control Networks. IEEE Communications Surveys & Tutorials, Vol.
15, NO. 2, Second Quarter.
[10] Santi, P., 2005. Topology Control in Wireless Ad Hoc and Sensor
Network. ACM computing surveys (CSUR) 37.2, pp: 164-194.
[11] Torrisi, N.M., Oliveira, J.F.G., 2008. Remote control of CNC
machines using the Cyber OPC communication system over public
networks. IJAMT, 39, 570577.
[12] Jeong, H.Y., Hong, B.H., 2011. Web Based Remote Robot Control
for Adjusting Position on Manufacturing System. Communications in
Computer and Information Science Volume 266, pp 96-103
[13] Rogowski A.M., 2015. Remote programming and control of the
flexible machining cell. International Journal of Computer Integrated
Manufacturing, Volume 28, Issue 6.
[14] Kwon, Y.J., Chiou, R., 2009. Automated vision inspection in
network-based production environment. IJAMT, 45, 8190.
[15] Chryssolouris, G., Papakostas, N., Mavrikios, D., 2008. A
perspective on manufacturing strategy: Produce more with less, CIRP
Journal of Manufacturing Science and Technology, Volume 1, No.1,
pp. 45-52.
[16] Dorf, R.C., Bishop, R.H., 2011. Modern Control Systems.
Prentice Hall.
[17] Salonitis, K, Stavropoulos, P, Fysikopoulos, A, Chryssolouris, G.
2013. CO2 laser butt-welding of steel sandwich sheet composited, pp.
69, 245-256.
[18] Makris, S., Tsarouchi, P., Surdilovic, D., Kruger, J., 2014.
Intuitive dual arm robot programming for assembly operations. CIRP
Annals-Manufacturing Technology, Volume 63, Issue 1, Pages 1316.
[19] Tanenbaum, A.S., 2003, Computer Networks, Fifth Edition,
Prentice Hall.
[20] Lee, K., Halder, A., Bhattacharya, R., 2014. Mean Square Stability
for Stochastic Jump Linear Systems via Optimal Transport.
arXiv:1403.2471
[21] Daafouz, J., Riedinger, P., Iung, C.,2002, Stability Analysis and
Control Synthesis for Switched Systems: A switched Lyapunov
function approach. IEEE Transactions on Automatic Control, Vol. 47,
No. 11.
[22] Zhang, D., Wang, W., 2012. Static output feedback control of
networked control systems with packet dropout. International Journal
of Systems Science, Vol. 43, No. 4, pp. 665672.
[23] Proakis, J., Manolakis, D., 1996/ Digital Signal Processing. Third
Edition, Prentice Hall.
[24] Soloklo, H.N., Ghaibi, A., Tavanaei, Z., Soltani, I., Design and
Optimization of Fuzzy Controller and Comparing with PID Controller
for DC motor, MAGNT Research Report (ISSN. 1444-8939), Vol.2 (5)
pp. 977-983.