p–ISSN: 2723 - 6609 e-ISSN: 2745-5254
Vol. 5, No. 9 September 2024 http://jist.publikasiindonesia.id/
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3338
Factors Affecting Control Room Operator Situation
Awareness and ITS Fuzzy Logic Model
Asep Karmana
1
*, Fergyanto E. Gunawan
2
, Muhammad Asrol
3
Universitas Bina Nusantara, Indonesia
Email:
1*
2
,
3
*Correspondence
ABSTRACT
Keywords: alarm
management, cognitive
behavior, control room
operator (CRO), fatigue,
heart rate.
The rapid development of computer-human interface in
process industries control systems challenged control room
operator (CRO) to maintain their situation awareness (SA)
at a high alert level. The paper aims to present physiological
factors affecting the work performance of CRO wellness
status during their duty to measure their SA level. PERCLOS
number of eye-opening recognition and heart rate profile
used as physiological variable to measure fatigue effect.
Alarm management records and response time spent in
controlling process changes behavior are used as cognitive
behavior variables representing competency performance.
Fuzzy logic methods were selected to correlate the input
variables to the SA level as output variables that will be
useful to communicate the CRO situation awareness status
into the computer-human interface as an alert system for the
CRO themselves or further to their supervisor for any
intervention required. Results of the observation and
analysis show a significant correlation between the observed
physiological of the CRO and the cognitive behavior of the
CRO at different stress level working conditions. The
findings support the development of a wellness alert system
in process industries' control systems by enhancing the SA
level of CRO effectively.
Introduction
The role of the control room operator panel operator or board operator in the process
industries is becoming very critical with the continuous and emerging development of
industry automation in particular to control the system. (Rozo et al., 2016). New
inventions and innovations in human-machine interface (HMI) and network system
design capability are intensively growing along with the amazing delivery of
sophisticated technologies for information and telecommunication (IT), computer
science, and engineering, which is collaboratively deployed in the process control system
and its application in the industrial automation. Those inventions and innovations are not
Factors Affecting Control Room Operator Situation Awareness and ITS Fuzzy Logic Model
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3339
limited to hardware and software part but also touch on the human factors aspect and
usability (Wibowo, 2023).
Although the development of machines, hardware, and software is considered a
very advanced stage, incidents and accidents in the process industries such as oil and gas,
petrochemicals, chemicals, and nuclear facilities are still observed as catastrophic with
large consequences in cost and fatality. (Skarphéðinsson & Mohan, 2017). Studies
conducted by Willis Engineering on the Willis Energy Loss Database (WELD) in the last
30 years, in addition to the loss of life, the financial losses for the event with a total loss
value of over USD 50 million cumulatively reached more than USD 25 billion from only
property damage and business interruption. (Rozo et al., 2016).
Human error was the main cause of the most tragic recent incident in the large
processes industrial complex of the oil & Gas, nuclear, and petrochemicals industries.
(Mikelsten, 2019). Marsh JLT Specialty reported the 100 largest losses in the
Hydrocarbon Industry incident from 1974 -2019 where the incident caused a loss value
higher than US$ 500 Million indicated in the below table. (Panel et al., 2014).
Table 1
Hydrocarbon Industry Value Loss from Incidents from 1974 – 2019
The specific case as an example of human errors made by the control room operator
is the incident of the fire and explosion that happened at BP Texas City Refinery
(Pasadena) on March 23, 2005, at 01:20 p.m. The incident occurred during the plant’s
commissioning after plant turn-around maintenance, where the alarm system failed to
notify the control room operator that an unsafe and abnormal situation existed within the
tower, and the blowdown drum – was overfilled with the flammable liquid. As reported
on the final investigation report issued by the US Chemical Safety and Hazard
Investigation Board (CSB), the incident caused 15 fatalities, 180 injuries, and financial
losses exceeding USD 1.5 billion. The release of flammables led to an explosion and fire.
All the fatalities occurred in or near office trailers located close to the blowdown drum.
A shelter-in-place order was issued that required 43,000 people to remain indoors. Houses
were damaged as far away as three-quarters of a mile from the refinery (US CSB, 2005)
(Campbell, 2021).
Despite the other root causes within the above BP Texas City Refinery case, the
control room operator and his supervisor have been pointed out as the major contributing
factor to the incident. The fact, over three hours of flammable liquid filling a distillation
tower without any further doubt or action on the process going on, indicates a degradation
in his situation awareness as a result of mental workload, fatigue, boredom, distraction,
and/or other human factors after long heavy work during the turn-around maintenance
execution. (Kurnia, 2016).
Asep Karmana, Fergyanto E. Gunawan, Muhammad Asrol
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3340
Human error is considered a normal function of humans; it is inevitable but may be
preventable. Although the recent development in industrial Advance Process Control
(APC) systems has been successfully demonstrated and applied in some parts of process
plants, eliminating the human part from the process control of integrated plants is
practically almost impossible. From the thirty years of author experience in the process
industries, in reality, even APC operated unit will need an intervention from a control
room operator in many situations, at least for monitoring and adjusting set point.
However, the control room operator is a normal human being like all others makes
mistakes and has lapses in attention. Therefore, all efforts involving all the stakeholders
should be made, and management systems and tools to help reduce the error rate to the
minimum possible level. (Tanjung & Nasution, 2005).
Safety in process plants is at the top of the list of issues that are yet to be resolved
fully. Recent accidents and their impact on the economy, environment, and human lives
have raised this issue once again. There are many causes for such accidents and many
reports have been published to explain why such accidents happen. (Sakti & Setiyawan,
2024). All of these have a common point of view, which is related to the environment of
the process control room. Human operators are at the heart of the control room as
responsible parties for proper monitoring and controlling of plant operations through
observing information from resources present inside the control room. Massive
information is faced all the time such as sudden bombardment with a lot of alarms under
abnormal situations can make operators paralyzed and lead to overwork, stress, and
fatigue. (Winarsunu, 2008).
A variety of methods are available in literature such as root cause analysis, removal
of alarm chattering, procedure and job aids, etc. to help control room operators and make
the control room much more friendly to control room operators. Nevertheless, their
implementation and usefulness in process plants have yet to see a significant level of
success. The usefulness of these methods depends on the extent to which these methods
can help operators. As an example, from the fellows working in the control room, the
information overflow and alarm flooding often confuse the operator, and it impacts
missing attention to critical alarms that may cause accidents to happen. (Aeberhard-
Hodges, 2019).
Instrumentation and control system designs have been developed based on the latest
advanced technology to reduce delay time and make the instrumentation reliable with
faster response time. In addition, the fascinating computation technique complemented
by software and computer technology made automation systems possible to resolve any
complex and integrated processing requirements. However, this supercomputer and
control generation does not mean reducing the workload of the control room operator but
adversely demands the control room operator to enhance their capability to operate and
be familiar with many elements to be memorizing. This mentally loaded the operator with
additional cognitive tasks and responsibilities.
Hence, a control room operator is a normal human who lives with a certain limited
capacity mentally and physically. They are usually working between 8-12 hours in a shift
Factors Affecting Control Room Operator Situation Awareness and ITS Fuzzy Logic Model
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3341
work schedule. As mentioned by the International Labor Organization (Hadwiger, 2018),
the normal maximum working hours should not exceed 8 hours but it can be extended to
a certain agreed consensus between the parties involved. Thus, the situation awareness
level is unlikely to be the same for the entire duration exceeding 8 hours per day. A
preliminary survey indicated that more than 80% of respondents signaled the need for
physical and mental fitness observation throughout their duty time, and to activate the
supervisor roles more vigilant by intervening in the mental/physical degraded on-duty
control room operator and improve their level of perception, comprehension, and
projection to the standard criteria.
Most of the efforts in supporting high-level situation awareness concentrating on
physical attributes, wellness, and mindfulness of the control room operator were almost
dismissed. Wellness and mindfulness are the keys to human factors to perform and be
able to maintain their healthiness of cognitive action for maintaining situation awareness
effectively. Therefore, the study on the wellness of control room operators will be able
to support the other human factors and concerns related to system awareness.
Method
The objectives of the design research are to assess the variables and parameters
affecting control room operator physiological and cognitive behavior that will provide
input to the advanced human behavior monitoring of control room operators possibly
embedded in human-machine interface (HMI). Physiological and cognitive performance
of control room operation while carrying out his duties will be measured as an inherent
parameter and utilized as input into a situation awareness model that will rate the control
room operator's wellness and mindfulness.
Tremendous workload and expertise requirements to interface with control room
operation elements can be structurally and strategically managed to deliver the proper
steps of actions and maintain their fitness to operate under controllable wellness.
Looking into the facts that none of the provisions to the control room operator
wellness established in the process industry practice, therefore, it is very challenging to
study and deliver the prototype design for the Visual Observer and Wellness Alert as an
integrated part of human factors focus of the control room development. The study is aims
and willing to explore the following objectives:
1) Utilizing simple commonly use devices to capture the physiological and cognitive
behavior and performance of control room operators, that work to evaluate the
wellness and behavior changes while control room operators carry out their duty.
2) To assess and analyze that physiological parameters/variable and their cognitive
performance have having good correlation and are suitable to be used as the basis for
developing and designing wellness alert/control devices that can improve control room
operator situation awareness level.
3) To propose a fuzzy logic-based model to interface the captured information from
physiological parameters/variable and their cognitive performance into the design of
the Wellness Alert system, in terms of situation awareness status, as output signal
Asep Karmana, Fergyanto E. Gunawan, Muhammad Asrol
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3342
received by the supervisory workstation and provide the supervisor on the possible
cognitive-behavioral and physical coaching intervention.
Overall, the target study is aimed to cover all the below workflow described in
Fig.2, however, the discussion in this paper will only focus on the third column of the
chart looking into the physiological and performance Factor Assessment.
Figure 3. The Overall Research Workflow
The ultimate intention of the study is to design a visual Observer and Wellness Alert
that will be embedded into human human-machine interface connected to a supervisory
control desktop for peer or high-level intervention as described in Fig. 4 and Fig. 5 below.
Figure 4. Conceptual Design of Visual Observation and Wellness Alert
Figure 5. Research/Study Framework
Data Correlation Validation
The collected physiological observation data of PERCLOSE and Heart Rate (HR),
cognitive behavior observation data of the Response Time (RT) taken by the operator to
resolve any process condition abnormality, and cognitive performance achieved through
Factors Affecting Control Room Operator Situation Awareness and ITS Fuzzy Logic Model
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3343
counting the number of Alarms appear during the observation period, modeled using
multi-variable linear regression method with #Alarm [Y] as dependent variable, and
PERCLOSE [P], Heart Rate (HR), and the Response Time (RT) as independent variables
following the below scheme.
Figure 8. Inter-variable Correlation Test and Validation
The regression analysis should be sufficient to confirm if between variables have a
significant correlation and be able to see each behavior, character, and wellness trend
while carrying out their tasks and duties.
Results and Discussion
The research study involves 3 professional control room operators (CRO) currently
working in a petrochemical plant with a different range of service years (all above 10
years) and different field experience in oil & gas and petrochemical company, specifically
expert in plant operation control mainly called as a board operator. Observation was taken
during 2 working days of 12-hour day shift and a 12-hour night shift. It was fortunate that
during the observation period, there was a chance of plant start-up, emergency shutdown,
and process anomaly, where a set of normal operation conditions and upset operation
conditions can be collected.
PERCLOS data was collected using a camera recording installed in front of the
CRO DCS console furnished by face recognition of eye blinking and %--closure, while
Heart Rate was recorded using a smart watch worn by CRO for the full observation
period. From the live records, data was extracted manually through a sanitizing process
to eliminate unqualified data such as the data capture while the CRO was not performing
the tasks. PERCLOS data extraction uses a fraction of the pictures generated from a film.
Heart Rate data extracted from the trend on the smartwatch HR record. For Response
Time, data is extracted from the trend of some of the bad actor controllers in the DCS
memory trend, and #Alarm was picked up from the Alarm Management system installed
in DCS.
Table 7. Data Regression Summary
Object
A
Intercept
Multiple
Regression
Coefficient
P-Value
Asep Karmana, Fergyanto E. Gunawan, Muhammad Asrol
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3344
a
[P
]
b
[H
R]
c
[RT
]
[P]
[H
R]
[RT
]
CRO –
1
(Normal
)
+173.7
+0
.9
6
-
2.0
1
+3.
72
1.64
E-
05
0.0
11
7
0.24
88
CRO –
1
(Upset)
-2.06
+0
.3
4
-
0.0
3
+11
.41
0.01
41
0.9
09
5
2.63
E-
06
CRO –
2
(Normal
)
+118.5
+0
.7
7
-
1.5
3
+7.
47
0.00
82
0.1
31
1
0.05
CRO –
2
(Upset)
+23.68
+0
.4
5
-
0.3
3
+5.
14
1.40
E-
03
0.0
25
9
0.00
26
CRO –
3
(Normal
)
+111.15
+1
.0
7
-
1.5
8
+7.
01
4.33
E-
05
0.0
25
9
0.00
26
CRO –
3
(Upset)
+56.28
+0
.2
5
-
0.6
7
+4.
91
1.36
E-
02
0.0
04
6
0.00
86
Aggrega
te
(Normal
)
-21.41
+1
.0
9
-
0.0
2
+9.
88
2.42
E-
13
0.8
96
7
1.27
E-
09
Aggrega
te
(Upset)
-9.07
+0
.4
8
+0
.02
+8.
03
2.70
E-
10
0.7
72
1
5.59
E-
12
Aggrega
te
(Overall
)
-23.35
+0
.7
6
+0
.09
7
+11
.87
1.07
E-
26
0.1
95
3
6.62
E-
32
After segregating the extracted data, 44 points of 10-minute average data were
collected from each mode of operation per each CRO. The total aggregate data is 264.
The quality of data is considered very good with clear trending of the raw data itself.
Physiological and Cognitive Factors Correlation
Referring to Table 7 above, the result of multi-regression indicated a very well-
consistent and significant correlation between the independent variables (PERCLOS,
Heat Rate, and Response Time) with the dependent variable (#Alarm). This is very clear
from the statistical p-value of the 9 sets of linear regression to each independent variable
with p-value <0.05. CRO-1 Normal operation data has having slight anomaly on
Response Time data due to an unplanned operation testing requirement on the analyzer
with the vendor The DCS trend record has been suspended for testing purposes. However,
Factors Affecting Control Room Operator Situation Awareness and ITS Fuzzy Logic Model
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3345
in general, the set of data is satisfactorily represented and aligned with the hypothesis of
the research.
As general findings, PERCLOS and Response Time (RT) hurt the CRO
performance as the more %-Closure increase and longer Response Time to bring
operating control parameter to the control point giving more #Alarm that may reduce the
situation awareness of the CRO. Meanwhile, Heart Rate variable changes seem to be
insignificant to impact the CRO performance and very situational. This is understandable
due to the comfortable place and professionality of the sample CRO for the study that
having long experience and maturity in knowledge and handling the process changes.
Therefore, the variation in heart rate is much less. The findings are also supported by the
P-value from ANOVA indicates less than 0.05 for the three independent variables
(PERCLOS, Heart Rate, and Response Time), which means the correlation between the
variables is significant.
In specific, it is also found that the model can predict the situation awareness
behavior level of individual CROs in response to the operating condition and status of the
plant. The three object CRO in the study demonstrated different characteristics of
performance concerning the tree variable measures in different plant statuses (Normal
and Upset condition). Object CRO-1 shows an opposite behavior compared to Object
CRO-2 and CRO-3 which is indicated by the coefficient value of each variable for normal
and upset situations. Based on their working experiences CRO-1 is the least among them.
Looking into the aggregate model, it was validated that PERCLOS and Response
Time give a very strong correlation to the performance measures, and the Heart Rate is
very situational and very individual physiological wellness status. For future studies, it
may be better to consider using a common comparative HR measurement to give the same
set of wellness levels that negate individual health characteristics.
Fuzzy Logic Deployment in Alert System Design
Based on variable input in Table 5 and variable output in Table 6, a fuzzy set was
developed for each variable using MATLAB Fuzzy Logic Experts version 2017a. The
membership Function of each variable input and output defined as shown on Fig. 9 to
Fig.13 below.
Figure 9. PERCLOS Membership Function
Asep Karmana, Fergyanto E. Gunawan, Muhammad Asrol
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3346
Figure 10. Heart Rate Membership Function
Figure 11 Alarm Membership Function
Figure 12. Response Time Membership Function
Figure 13. Situation Awareness Membership Function
Upon confirming the variable membership functions, the Fuzzy Rule set is defined
as a collective function combination of each category set for variable membership
function according to the following “IF X AND Y THEN Z” algorithm, for example:
IF (PERCLOS is Fresh/Vigilance)
AND (Heart Rate Ratio is Resting Relax)
AND (AM Performance is Acceptable)
AND (TS Performance is Strategic)
THEN (SALevel is Relax/Tune out)
Factors Affecting Control Room Operator Situation Awareness and ITS Fuzzy Logic Model
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3347
For the research study, 156 fuzzy set rules generated by the MATLAB Simulation
Software were used to develop an alert that connected to the Situation Awareness Level
of the control room operator for different sets of variable input from the measurement of
PERCLOS, Heart Rate, #Alarm, and Response Time of the object control room operator.
Figure 14. Rule Viewer Fuzzy Simulation Result
Figure 14 illustrates how the fuzzy logic will work and easily interfaced with the
Human Machine Interface system and further provides an alert to the control room
operator himself or their supervisor for further intervention or self-correction accordingly.
Illustration explanation:
Input Variable
PERCLOS = 23.5% Closure – Fatigue /Drowsi-1
Heart Rate = 1.04 of HRR – Normal Alert
#Alarm = 3.79 per 10 minutes – Manageable/Excessive Demanding
Response Time = 0.385 minutes – Tactical
Output Variable
Situation Awareness Level = 0.851 – Relaxed Awareness (Level-3 SA)
Based on the simulation result, the fuzzy logic is perfectly fit to be used as an
interpretation model to measure the Situation Awareness.
The main focus of the study is to find the relationship and correlation between the
physiological variable and cognitive behavior of CROs while executing their daily work,
as an input to the system alert model, which will contribute to human factor excellence in
the oil & gas and petrochemical industries in specific and commonly in the manufacturing
industries. PERCLOS (%-closure of eye) measured using a face recognition algorithm
using a camera. At the same time, the heart rate of the CRO was recorded to see the impact
of psychological and mental activities while executing their tasks on their physical heart-
pumping characteristic. Combined the two-measurement of the CRO represent
physiological wellness. In addition to physiological wellness, CRO response time in
anticipating or intervening to the process control changes from abnormal condition started
to the point of control recorded as cognitive behavior. Meanwhile, the record of process
alarms per 10 minutes during the observation period was noted as a performance indicator
of the control and monitoring activities results. The collected record of PERCLOS, Heart
Rate, Response Time, and #Alarm was tabulated for data analysis purposes. After several
Asep Karmana, Fergyanto E. Gunawan, Muhammad Asrol
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3348
attempts to trend the data, it decided to choose a multiple regression method to develop a
multivariable linear model.
The result of the statistical multiple regression model for the individual set of
observations as well as the aggregate set of observations provides a significant correlation
between independent variables mainly for PERCLOS and Response Time with statistical
p-value <0.05. Heart Rate correlation was found to vary between the set of observations,
where 5 out of 9 sets of observation provide a p-value slightly higher than 0.05. This
anomaly was expected as the level of heart rate of an individual changes depending on
the psychological environment and healthiness of the CRO by the external factors outside
the workplace including moody personality. From the physical situational observation
note, the set of data is consistent in that personal state at work provides abnormal data
trends which then impacts the regression result. To avoid this personal situation, there is
a normalization process to be treated in the raw data. In general, the result of observation
supports the use of the selected variables as an input to measure CRO wellness and
situation awareness level.
The findings are sufficient to step further for developing a design of a Situation
Awareness Alert System embedded into Human Machine Interface in the oil & gas and
petrochemicals industry. This will become a breakthrough concept of a human factor
improvement area in conjunction with the worker’s Health and Safety Prevention model
to provide early human error prevention and to protect the industries from vulnerable
catastrophic incidents.
Having demonstrated the significant variable correlation, a model of the functional
system is required to process and interpret the variable input into proper messages where
an individual or/and another stakeholder in the operating group can initiate a proper action
or intervention to improve the situation awareness level of the CRO. After reviewing
several functional system applications and the platform available in the market, Fuzzy
Logic Experts has been selected for this study to simulate and interpret the selected input
variable for the CRO Situation Awareness level. Based on the generated 156 fuzzy rules,
the simulation is perfectly run with consistent output correlated to the set of situation
awareness level definitions.
As an internal critique, ideally, the work is to be completed till a prototype design,
however, due to a shortage of resources the study has been completed up to functional
system application modeling which has been sufficient to demonstrate the model design
is workable. This is a notice for the next research.
Conclusion
The research study was inspired by the fact that oil & gas and petrochemical
industries are still suffering and facing huge losses from plant incidents. The concern
about a high percentage of main root causes of incidents triggered by human error brought
an idea to look in specific into the control room operator roles and responsibilities in
carrying out their tasks and duty as central players in the team to monitor, maintain,
control and manage operation activities in every 12 hours shift work schedule.
Factors Affecting Control Room Operator Situation Awareness and ITS Fuzzy Logic Model
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3349
From the research study result, it can be concluded the following 3 points:
(1) The role of the control room operator and their contribution to the safe plant operation
is very critical and crucial. Maintaining high situational awareness is mandatory.
(2) PERCLOS, Heart Rate, Response Time, and #Alarm have significant correlation and
deserve to be selected as variables to monitor the wellness of CRO as well as to use
as input to situation awareness and wellness alert system devices.
(3) Fuzzy Logic Expert works very well to interpret the impact of observation variables
on the situation awareness level and can be deployed as a functional tool for situation
awareness and wellness alert system devices.
Asep Karmana, Fergyanto E. Gunawan, Muhammad Asrol
Jurnal Indonesia Sosial Teknologi, Vol. 5, No. 9, September 2024 3350
Bibliography
Aeberhard-Hodges, J. (2019). Practitioner perspective Women and international treaty
making–the example of standard setting in the International Labour Organization.
In Research Handbook on Feminist Engagement with International Law (pp. 286–
304). Edward Elgar Publishing.
Campbell, F. J. (2021). Human factors: The impact on industry and the environment.
Natural Resources Management and Biological Sciences, 1–14.
Hadwiger, F. (2018). Contracting international employee participation. Global
Framework Agreements, Cham: Springer.
Kurnia, L. (2016). Model Log-Linear Pada Faktor Yang Mempengaruhi Berhenti Studi
Mahasiswa. Sambutan Ketua Panitia, 155.
Mikelsten, D. (2019). Otomasi dan Teknologi Berkembang (Vol. 3). Cambridge Stanford
Books.
Panel, 2014 Organic Contamination, Summons, R. E., Sessions, A. L., (co-chairs),
Allwood, A. C., Barton, H. A., Beaty, D. W., Blakkolb, B., Canham, J., & Clark,
B. C. (2014). Planning considerations related to the organic contamination of
martian samples and implications for the Mars 2020 rover. Mary Ann Liebert, Inc.
140 Huguenot Street, 3rd Floor New Rochelle, NY 10801 USA.
Rozo, E., Rykoff, E. S., Abate, A., Bonnett, C., Crocce, M., Davis, C., Hoyle, B., Leistedt,
B., Peiris, H. V, & Wechsler, R. H. (2016). redMaGiC: selecting luminous red
galaxies from the DES Science Verification data. Monthly Notices of the Royal
Astronomical Society, 461(2), 1431–1450.
Sakti, A. R. P., & Setiyawan, P. (2024). Sistem Manajemen Keselamatan Dan Kesehatan
Kerja K (3) Pada Proyek Pembangunan Gedung Hotel Santika Nagrak Sukabumi.
Universitas Islam Sultan Agung Semarang.
Skarphéðinsson, J. I., & Mohan, V. (2017). Expanding the customer base for DCS in the
Oil, Gas & Chemicals market in Sweden; A Case Study of ABB.
Tanjung, K., & Nasution, M. K. M. (2005). Rancangan dan penerapan kontrol logika
kabur untuk industri. Jurnal Sistem Teknik Industri, 6(2), 75–78.
Wibowo, A. (2023). Internet of Things (IoT) dalam Ekonomi dan Bisnis Digital. Penerbit
Yayasan Prima Agus Teknik, 1–94.
Winarsunu, T. (2008). Psikologi keselamatan kerja. UMMPress.
