2021 60th Annual Conference of the Society of

Instrument and Control Engineers of Japan (SICE)

September 8-10, 2021, Tokyo, Japan

Matlab-Stateflow for automated HAZOP*

Meng Fei Chia, Pavan Kumar Naraharisetti

Abstract— The principle of HAZOP analysis is to identify understand and seek the causes for such deviations. The
hazards arising in a plant due to deviations from the design brainstorms of the causes involve the members to identify
intent or the normal behavior of the plant. A meaningful as many as possible. These causes include both inherent
deviation is generated by coupling a guideword and a design of the system as well as human factors involved
parameter and this is applied to cover all the possible such as negligence and errors. It is also crucial that
deviations that could arise from the design and operation of members, especially the designers, takes a positive and
the plant or the process in the node. HAZOP is normally critical attitude and not a defensive one to establish a
2021 60th Annual Conference of the Society of Instrument and Control Engineers of Japan (SICE) | 978-4-9077-6473-9/21/$31.00 ©202110.1109/SICE54164.2021.00021

done manually or using commercial software. Software is successful HAZOP study.

used both as a means of book keeping and also to generate
plant operation modes. In this work, we present on the use of Evaluation of the consequences derived from each
Matlab-Stateflow to perform automated HAZOP. This cause shall be done to identify all the immediate and
methodology aids safety professionals in automating most of delayed consequences developing in and outside the node
the manual tasks, thus has potential to reduce the time under study. As the processes and nodes are interlinked,
required to perform a HAZOP study. the consequences arise from the particular node can be
I. INTRODUCTION more prominent in another node. A decision to defer the
consequences analysis to another area can be carried out,
but it is important to note and repeat the consequences so
A. Background
as to cover the study comprehensively.
A HAZOP study is a systematic process to conduct a
structured analysis of a system, process or operation Once the consequences are established, the existing
where details of its design are available and carried out by safeguards and protections of the system against the
a team of participants with different subject matter consequences or the scenarios envisaged can be recorded
expertise, from the designer to the operator and vendor. In down as the next step. One approach would be to ignore
the Hazard Study (HS) methodology developed by ICI the existence of any safeguards and derive the worst-case
using six different stages [1] to sequence out safety scenario, which has the advantage of being more prudent
studies, it is often carried out at the third stage, which is and misjudgment of the need for protection can be
the detailed design stage of a plant where the design has minimised. It is argued that such approach can be
no major changes and modification that deviates far from unrealistic to ignore any existing safeguards if the system
the original intended design. The first paper on hazard is well-designed.
and operability study [2] became the procedure and more A risk assessment then assigns the risk level by
widely used after the Flixborough disaster occurred considering the likelihood and the severity of the
during the same year. The term “HAZOP” was not widely consequences. Statistically the level of risk can be
adopted in formal publications until 1983 when Keltz represented as the product of the probability the harm
conducted the workshop and had this term incorporated occurs multiplied by the severity of the harm. In most
into his notes [3-4]. cases neither the likelihood or severity can be indicated
In HAZOP, guide words such as no, more, less, as with accuracy and precision as it is a leading indicator.
well as, other than, part of, reverse, before/after, Risk matrices, when defined properly and used correctly
faster/slower etc. are combined with parameters such as [5] can be applied to quantitatively assess the risk
flow, temperature, pressure, viscosity, speed etc. and presented. With an experienced team and sound
apply to the node to cover all the possible deviations that judgement, the process efficiency of qualifying the risk
could arise from the design and operation of the plant or increases.
the process in the node. Following the risk assessment, recommendations and
Once the list of meaningful deviations has been actions to be taken to remediate the problem are proposed
identified, the next step to the HAZOP study is to and recorded in the HAZOP study. The positive actions
should be unanimously agreed by the team and benefits
the design of the plant to improve its safety.
*M. F. Chia is a graduate student at the Newcastle University in
Singapore, Faculty of Science, Agriculture & Engineering, Newcastle B. Drawbacks of Conventional HAZOP
University, Newcastle Upon Tyne NE1 7RU, UK. (e-mail:
m.f.chia2@[Link]) With increasing complexity of plants and operations,
P. K. Naraharisetti is an Assistant Professor and Degree Programme HAZOP studies become increasingly tedious and time-
Director (Industrial Automation & Machine Learning, PG) at the consuming. A typical chemical process could take
Newcastle University in Singapore, Faculty of Science, Agriculture & anytime from 1 to 8 weeks, involving a team of 4-10
Engineering, Newcastle University, Newcastle Upon Tyne NE1 7RU,
UK. (corresponding author, phone: + 65-6908 6025; e-mail: pavan- members or more [6].
[Link]@[Link]).

978-4-9077-6473-9 PR0001/21 ¥400 © 2021 SICE 287

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on February 23,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 As HAZOP is a very thorough systematic approach to time required to brainstorm for them, hence better efforts
carry out process hazard analysis, much of the time spent can be put into proposing recommendations for mitigation
is used to identify failures and hazards [7]. Of these of these consequences.
identified, most of them are routine causes and
TOPHAZOP is a software [18] using knowledge base
consequences [8] because the process units are common
divided into two segments, process-general knowledge
in many plants, such as pumps, tanks, and valves etc. The
and process-specific knowledge to of 15 process units to
repetitive nature of such analysis results in reduced efforts
propagate the failure cause and consequences.
and concentration over time over mundane tasks like this.
Subsequently, a new software EXPERTOP was developed
Cognition reliability and reasoning capacities may
[19] where it supplements their previous works to add on
decrease drastically when repetitive tasks are carried out,
an inference engine and graphical user interface on top of
or large amounts of data are presented, and the conditions
the existing knowledge base design in TOPHAZOP.
becomes stressful to the HAZOP team [9].
[20] developed a knowledge-based software
As the manual process involves an expert team of
HAZOPExpert that uses the framework of process-
human, it is subjected to cognitive biases [10].
specific and process-general knowledge, [21] and the use
Overconfidence bias, herd mentality, confirmation bias,
of Digraph-based HAZOP (HDG) models [22] for process
framing cognitive bias, representative heuristics are some
units. The models were developed to be context-
of the biases that could implicate the result of HAZOP as
independent manner and can be used to perform analysis
such biases are subconsciously employed during the
over a wider variety of processes.
course of the study and difficult to detect and override.
PHASuite was developed subsequently [23] with an
Although the guidelines for deviations are clearly
extensive knowledge engineering framework, comprising
established in the guide [1], the way the causes and
of four main parts: information sharing, representation,
consequences are written can vary due to the nature of the
knowledge base, and reasoning engine. The approach was
processes. The records of the HAZOP study can be
based on Petri nets and applied to pharmaceutical batch
phrased differently across different studies, and difficult
processes.
to interpret as a result of inconsistency in wordings and
expressions, or can be perceived differently by different ExpHAZOP+ [24] utilises the same key concept as
people. The estimated cost of about 1% sales or 10% optHAZOP [7] and was a further development of
profit in chemical company is spent on carrying out safety EXPERTOP [19], using a unique fault propagation
analysis [11]. algorithm and a knowledge-base to retrieve causes and
consequences downstream from an identified upstream
II. LITERATURE REVIEW event.
One of the earliest attempts of automation was carried The works that have been done in this area aims to
out [12] using qualitative propagation equations to reduce the amount of time spent on conventional HAZOP
propagate faults at each process unit. The causes are study with the aid of computer programs. [25] set up a
generated by searching the initial event and consequences way to estimate the amount of time required to carry out a
by searching for the terminal event. [13] used a quasi- HAZOP based on number of major equipment for
steady state qualitative simulation approach to automate analysis, the complexity of the system as well as the
HAZOP analysis. The approach top-down approach here experience of the team as the considerable factors. [26]
is highly similar to other techniques such as fault-tree proposed a mathematical model that further refined this
analysis with which it was compared. However, the by considering other parameters like preparation time,
approach was highly combinatorial and time consuming meeting time, delay and report writing.
for practical application.
A. Analysis
[14] proposed HAZOPEX as a system for safety
management to support the process designers in the Since the mid-1980s, rule-based approach has been
HAZOP analysis. Its purpose was to identify defects and the kickstart of automation of HAZOP [12]. A little later
error in design in the early stage to have better and more it evolved to using a more advanced rule-based expert
economic possibilities for corrective actions to be taken system using IF-THEN algorithms and an inference
[15]. engine to generate the deviations [15]. This was called the
HAZOPEX, an advanced development environment
[16] developed an inductive and deductive consisting of a Lisp workstation (Symbolics) and a hybrid
methodological approach for chemical reactions hazards expert system shell (KEE).
and the approach was only limited to this aspect. [17]
developed a rule-based approach to map out the networks Knowledge-based prototypes and expert systems were
for consequence reasoning using both forward and developed subsequently. [27] adopted the use of Prolog,
backward chaining. one of the first and the advanced logic programming
language back then to create an expert system consisting
OptHAZOP that was developed [7] proposed a study of a knowledge base, an inference engine and case-
procedure and technique that relied on an expert specific information that forms the backbone of an expert
knowledge base with a large collection of facts, rules and system structure. An inference engine can infer new rules
information regarding various components of process from the case-specific data and new user inputs to
plants to generate cause and consequences to reduce the

288

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on February 23,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 construction a solution. The advantages over shells are efforts being made to automate the process. As a result,
that it allows more flexibility in programming, and has the process is subjected to cognitive biases [10].
advantages of modularising the knowledge base, allowing Groupthink is a phenomenon where a group of people
faster search to be carried out instead of searching from a have common but possibly false beliefs and hence, tend to
single large database [28]. make decisions the same way. Another type of cognitive
bias call group polarization refers to the tendency for
The usage of knowledge-based expert systems become
decisions made in a group to be more extreme than the
widely adopted as the framework to carry out automation,
initial inclination of its members, which is often seen
as seen in the many studies and prototypes being
during conventional HAZOP studies [30]. Unfortunately,
developed [11,17,19,21,24]. This was optimal as causes
biases are subconscious and difficult to detect, and there
and consequences are all interlinked and required to be
are no comprehensive theory or practices to mitigate such
stored in a structured data form, compared to typical
human thinking. However, being aware and understanding
databases.
its role is important so that the expert team are aware of
HAZOPExpert was one such knowledge-based system potentials for poor decision-making during HAZOP
that had much further development from a prototype practice. Even in automation, the biases arise from inputs
compared to the rest of the tools. It was taken further generated from users can form part of the cognitive bias
when the authors further improved the framework [8] with within the errors of the tools unless machine learning and
the use of a digraph-based model [22] that could further deep neural networks can be incorporated into these
enhance the representation of the process system to the programs and tools as a future development.
tool user. The authors developed a more comprehensive
On the technicality limitations, a certain level of
system to perform detailed safety evaluation, and digraph
details and maturity in the design is needed before a
and fault tree models to synthesize and analyse fault trees
HAZOP, whether conventional or automated study is
[6, 21].
carried out. If the design is poor, the large number of
The tool optHAZOP was developed to help improve causes can potentially overload the process [31]. This is
efficiency in carrying out studies compared to also seen in earlier development of automation where the
conventional HAZOP processes. Further works were vast amount of data impedes the performance of the
carried out to speed up optHAZOP in a tool called computers and taking a very long time to derive a HAZOP
TOPHAZOP to identify general and specific cause- study output.
consequences of all probably process-deviations. The
The limitations of the computer system in simulating
entire expert system (termed as EXPERTOP) is also made
the dynamic reasoning mechanism and knowledge
up of knowledge-base, inference engine and user interface
structures of human makes it difficult to obtain a fully
[19]. Finally, the author also proposed a knowledge-based
automated HAZOP by computer [11]. As in the many
framework to conduct HAZOP on offshore process
studies carried out, the primary aim was not to replace
facilities [29].
human involvement in HAZOP study but more of to
A detailed analysis on the time taken for study reduce the amount of time taken for HAZOP studies to be
procedure reveals that around 35-42% of the total time for carried out by automating the routine works such as
a HAZOP study is spent on identifying failures and common causes and consequences associated with the
hazards, and causes of these hazards [7]. The amount of inherent nature of the equipment itself.
time saving with use of optHAZOP takes 45% less time
It is impossible to have a complete knowledge base
than conventional study time with the use of a case study.
that comprises of all the known and unknown causes and
also claimed the tool TOPHAZOP enabled the study to be
consequences with increasing complexity and emergence
completed in 4.5hours, while the same study by
of new technology, of plant process design and operation.
conventional HAZOP study took 2.5 weeks [18]. It was
Data acquisition therefore sets the quality tools apart from
also claimed the HAZOPExpert tool completed the
the others. However, if the efforts and time required to
HAZOP of a sour water stripper plant in 2 hours, and a
input and represent the information in is excessive, the
hydrotreator plant in 4 hours, much lower than the time
value of the tool becomes greatly diminished. It is
spent by the team on the analysis [21].
therefore crucial to figure out an approach to
systematically build up a generic knowledge base that
B. Knowledge Gaps / Limitations could be applied to most, if not all process design plants
in general, and a specific knowledge base for each unique
HAZOP is the most studied PHA method and process design itself.
abundant of research are carried out to improve and re-
adapt HAZOP processes to the modern world. Although The outputs generated by computer tend to be
there have been numerous researches over the past decade excessive which in conventional HAZOP these are usually
on the automation of HAZOP, and many advances have filtered out. Handling a huge number of data output is one
been made in this field of work as summarised above, of the most significant challenges from automation of
there are still gaps that can be addressed in this aspect of HAZOP. While certain tools can provide users with
study. control over the output of the data as such that is applied
to HAZOPExpert [32], the significant causes or
HAZOP studies are still very much conducted by consequences could still be left out unintentionally.
human expert teams in the process industry, despite

289

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on February 23,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 Treatment of the data becomes one of the tedious steps to correspond to the steps taken in conventional HAZOP to
carry out over conventional HAZOP study. define a meaningful deviation relating to each
process/equipment due to the nature of the
There are also issues on the amount of trust that can
process/equipment. Another level of substates is then
be given to automated HAZOP studies. HAZOP is usually
defined for each parameter, which are representation of
carried out by group of experts with considerable
the deviations in HAZOP.
experience, responsibility and authority. In order for the
team to accept a recommendation for hazard mitigation, it The initialisation of a deviation requires an input from
has to be collectively agreed and brainstormed to the user at the start, which triggers the states to be active
understand the basis by which the problem was and results in a propagation along the connected blocks
approached. This cannot be achieved in a computer- given the conditions are fulfilled. Within each block, the
generated HAZOP study. states can then be linked to consequences such as tank
empty, tank overflow as the end state of the block. The
Most of the current research approaches with the aim
interactions between the cause-deviation-consequence are
to demonstrate that HAZOP studies can be automated to a
similar to the bow-tie analysis for process hazard analysis.
certain extent by producing prototypes specific to a
certain area or industry. There is no single software The modelling allows the demonstration of what
capable of emulating a wider approach due to complexity effects user input event simulating a deviation has on the
and the dynamics of process plants and emergent of new simulated plant model and derived at an end-state
technologies. consequence. Such pathways can then be translated into a
HAZOP study worksheet by extraction of the events and
III. HAZOP METHODOLOGY AND AUTOMATION transitions that occurred and processing the data. The
potential for the approach to help in time reduction of
With the emergence of new technologies and software, HAZOP study is considerably high and the purpose of the
the aim is to provide an alternative method of providing a research work aims to demonstrate that, as well as to lay
(semi)automated HAZOP using a commercially available the foundation for further works to be developed.
tool call Stateflow. Stateflow is a component of the
Simulink graphical language used in MATLAB, which
IV. RESULTS AND DISCUSSION
allows representation of state machines and flow charts
within a Simulink model. Stateflow allows modelling on An existing industrial process is used to perform the
sequential decision logic which can be simulated as a HAZOP using the proposed modelling framework. The
block within the Simulink model. Utilizing the system is an LNG Fuel Gas Production Process,
multifunctional tool, along with Simulink and MATLAB, consisting of the following major process units and
a methodology is formulated on how it could be utilised equipment: LNG Storage Tank, LNG Vaporizer (Heat
to emulate a HAZOP study process. Exchanger), Control Valves. A single node and process
was selected to model using Matlab-Stateflow.
The model-based approach uses the idea of state-
transitions to model the sequence of events that emulates LNG from the storage tank is extracted by the pressure
the idea of fault propagation process within a system or of the tank at 10 barg to the shell and tube heat exchanger
process. The tool is capable of extending the qualitative to be heated up to 15°C, before being stored and used as
modelling to complement with quantitative-based to allow fuel gas for the gas consumer. The flow to the heat
event triggers defined by process variable inputs. Such exchanger passes through a control valve that controls the
approaches have been utilised to demonstrate the amount of LNG being sent to heat exchanger.
possibility of using statechart modelling and simulation
[33].
A P&ID for a typical chemical process plant is used to
model the automation of the HAZOP study. The units
involved in the system includes pump, buffer tank, heat
exchanger and the end consumer which is the gas engine.
The units are interconnected by pipelines. This system
contains most process units that are common across all
industries, such as pumps, tanks, and heat exchanger.
The approach here is to model the process units from
the P&IDs into the Stateflow as blocks similar to
construction of a block flow diagram (BFD). These
blocks become the knowledge-base containing all the
possible information for the process unit with input and Figure 1. Modelling methodology showing the initial deviation from
output that could link to other blocks as what a flow the starting point.
diagram could do, and forms the superstate of the
equipment. Within these superstates, the relevant process The methodology involves defining HAZOP inputs
parameters for each of these unit operations are identified. into a format the software recognises words(strings) to
For example, a storage tank would have parameters such numbers, and setting up a Knowledge Base which
as level, temperature, and pressure. These parameters involves a 2-Dimensional Cause and Effect Matrix,

290

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on February 23,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 Inherent Causes Table. The modelling of the process Some of the advantages of Stateflow for HAZOP are:
operations in Stateflow is done by designing process unit
blocks, defining states and developing an algorithm and  Stateflow blocks for process units can be easily
sequencing. The outputs from the simulation are reproduced with drag-and-drop.
translated back into HAZOP format by converting  Applicable for wide range of industries and
numbers to words (strings). The input/output for processes.
HAZOP are defined in a form that Stateflow recognises
by using a hierarchy numbering system, through a Matrix  Using the hierarchy developed, it is easy to
Table mapping all the causes and effects, aided by a extend the database to other process units, process
Lookup Table for causes associated with each individual parameters and guidewords.
equipment. Using this, Causes can be derived from both  Input conditions can be catered to use actual field
inherent and effected causes. and design data. Program can be further
developed for operator training purposes.
In addition, process units are also modelled using  Multiple process parameters deviations can be
Stateflow Blocks. A storage unit is one such example concurrently run and obtained from a single pass.
where, parameters are represented as superstate. This
 Time and effort saving compared to carrying out
allows parallelism for multiple parameters to be simulated
the systematic process conventionally.
concurrently. Each substate represents a process
parameter condition from the upstream process unit and  Database can be extended, modified, to cater to
gives a process parameter condition to pass on to process/plant-specific design and hazards.
downstream unit. Flow, Pressure and Temperature were
selected to represent the parameters for deviations.  HAZOP outputs are standardized
 Records can be saved or re-run easily with the
program and the block flow diagram built.
 High traceability and repeatability.
Some of the limitations of Stateflow for HAZOP are:
 Does not consider anything further upstream or
downstream of the process.
 Multidimensional array requires intense
knowledge and inputs into the database.
Figure 2. Modelling valves using Stateflow Blocks.
 In conventional HAZOP, start up and shut down
Using Stateflow (and Simulink/MATLAB), it is on plants and operations can be studied. This
possible to achieve the following: cannot yet be simulated in this computerised
study.
 Build Stateflow blocks that can represent 3
process units and equipment (pressure vessel, heat  The simulation only can provide deviations from
exchanger, and on/off valves) steady-state/normal operations.
 Reconstruct a simplified P&ID node with drag-  Extensive programming of the databases and state
and-drop function into a Simulink Flow chart flow required.
using pre-defined Stateflow blocks
 Large number of process units/equipment,
 Use states to represent the 4 process parameters parameters and deviations will make it highly
of the process unit/equipment (Flow, Level, combinatorial.
Temperature, Pressure) and state of equipment
 As extensive as the database inputs are, the
such as valves (i.e. valve open or close)
outputs can be excessive and/or not applicable to
 Apply multiple process parameters and state of a specific process.
equipment (e.g. valve open or close) on the same
 Filtering out the large data outputs can be time-
process unit/equipment
consuming, contrary to what we want to achieve.
 Run a limited semi-automated HAZOP process to
 Increasing complexity, emergence of new
identify 3 deviations and causes similar to that of
technology, of plant process design and
the conventional HAZOP based on pre-defined
operations make it tedious for data acquisition.
conditions.
 Program is not able to simulate the dynamic
 Develop a collective database for HAZOP
reasoning mechanism and knowledge structure of
deviation and causes.
human.

291

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on February 23,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.
 V. CONCLUSION [13] A. Waters, J. Ponton, “Qualitative simulation and fault
propagation in process plants.” Chemical engineering research &
Given the enormous amount of effort, manhour, and design, 67, 407-422, 1989.
cost to perform HAZOP of considerable reliability, there [14] J. Suokas, P. Heino, I. Karvonen, “Expert systems in safety
management.” Journal of occupational accidents, 12, 63-78, 1990.
are incentives to develop automation of HAZOP. The
[15] [Link], J. Suokas, I. Karvonen, “An expert system in process
automation reduces the time and effort involved on design-analysis of process safety and reliability.” Proceedings of
routine analytical steps and allows more focus to be the International Workshop on Artificial Intelligence for Industrial
placed on non- routine aspects such as process-specific Applications, 1988. IEEE, 225-231.
[16] C. J. Nagel, “Identification of hazards in chemical process
hazards causes and consequences. systems.” Massachusetts Institute of Technology, 1991.
Automation of HAZOP also allows completeness and [17] H. Chae, Y. H. Yoon, E. S. Yoon, “Safety analysis using an expert
consistency across multiple HAZOP carried out in system in chemical processes.” Korean Journal of Chemical
Engineering, 11, 153-161, 1994.
general. Terminologies can be standardised and easily [18] F. I. Khan, S. Abbasi, “TOPHAZOP: a knowledge-based software
recognised by different teams, and better consequences tool for conducting HAZOP in a rapid, efficient yet inexpensive
recording drawing from the contributions to a large manner.” Journal of Loss Prevention in the Process Industries, 10,
333-343, 1997.
comprehensive knowledge-base to ensue completeness. [19] F. I. Khan, S. Abbasi, “Towards automation of HAZOP with a
The aim of this research is to adopt the use of the tool new tool EXPERTOP.” Environmental Modelling & Software, 15,
called Stateflow in MATLAB and Simulink to emulate 67-77, 2000.
the process of HAZOP study by state-transitions. The [20] R. Vaidhyanathan, V. Venkatasubramanian, F. T. Dyke,
“Hazopexpert: An expert system for automating HAZOP
development of the tool to not only showcase a single analysis.” Process Safety Progress, 15, 80-88, 1996.
process, but to be able to replicate across multiple [21] R. Vaidhyanathan, V. Venkatasubramanian, “Experience with an
processes is desirable to achieve. With the foundations set expert system for automated HAZOP analysis.” Computers &
Chemical Engineering, 20, S1589-S1594, 1996.
in this research, it lays the path for future works to further [22] R. Vaidhyanathan, V. Venkatasubramanian, “Digraph-based
develop the tool into a more robust one for wider models for automated HAZOP analysis.” Reliability Engineering
applications. and System Safety, 50, 33-49, 1995.
[23] C. Zhao, M. Bhushan, V. Venkatasubramanian, “PHASuite: An
The main reason for not able to make HAZOP fully Automated HAZOP Analysis Tool for Chemical Processes: Part I:
automated is that it is much more than just a process for Knowledge Engineering Framework.” Process Safety and
producing the HAZOP table. It is a socio-process Environmental Protection, 83, 509-532, 2005.
[24] S. Rahman, F. Khan, B. Veitch, P. Amyotte, “ExpHAZOP+:
requiring the team members to interact, discuss, and
Knowledge-based expert system to conduct automated HAZOP
jointly conclude on consensus made to safety and be analysis.” Journal of Loss Prevention in the Process Industries, 22,
responsible for the decisions and implementations made. 373-380, 2009.
[25] R. Freeman, R. Lee, T. Mcnamara, “Plan HAZOP studies with an
expert system.” Chemical Engineering Progress, 88, 28-32, 1992.
REFERENCES [26] F. I. Khan, S. Abbasi, “Mathematical model for HAZOP study
[1] F. Crawley, B. Tyler, “HAZOP: Guide to Best Practice”, Elsevier time estimation.” Journal of loss prevention in the process
Science, 2015. industries, 10, 249-257, 1997.
[2] H. Lawley, “Operability studies and hazard analysis.” Chem. Eng. [27] T. Weatherill, I. T. Cameron, “A prototype expert system for
Prog., 70, 45-56, 1974. hazard and operability studies.” Computers & Chemical
[3] T. A. Kletz, “Hazop & Hazan: Hazard Workshop Modules: Notes Engineering, 13, 1229-1234, 1989.
on the Identification and Assessment of Hazard”, Institution of [28] W. Earl, C. Williamson, “Control System Synthesis: A
Chemical Engineers, 1983 Comparison between PROLOG and an Expert System Shell for
[4] T. A. Kletz, “Hazop—past and future. Reliability Engineering & Distillation Column Control.” PSE'88: Third International
System Safety”, 55, 263-266, 1997. Symposium on Process Systems Engineering: In Affiliation with
[5] K. Busby, M. Kazarians, “Pitfalls of Using the Wrong Risk Matrix CHEMECA 88, a Bicentennial Event; Sydney, Australia, 28
In PHA and LOPA”, AIChE Spring Meeting and Global Congress August-2 September, 1998; Preprints of Papers, 1988. Institution
on Process Safety, 2018. of Engineers, Australia, 175.
[6] V. Venkatasubramanian, J. Zhao, S. Viswanathan, “Intelligent [29] F. I. Khan, “Knowledge-based expert system framework to
systems for HAZOP analysis of complex process plants.” conduct offshore process HAZOP study.” 2005 IEEE International
Computers & Chemical Engineering, 24, 2291-2302, 2000. Conference on Systems, Man and Cybernetics, 2005. IEEE, 2274-
[7] F. I. Khan, S. Abbasi, “OptHAZOP—an effective and optimum 2280.
approach for HAZOP study.” Journal of Loss Prevention in the [30] P. Baybutt, “The psychology of decision-making in process hazard
Process Industries, 10, 191-204, 1997. analysis”, Spring Meeting & 11th Global Congress on Process
[8] V. Venkatasubramanian, R. Vaidhyanathan, “A knowledge-based Safety, 2015.
framework for automating HAZOP analysis.” AIChE Journal, 40, [31] S. A. McCoy, S. J. Wakeman, F. D. Karkin, M. L. Jefferson, P. W.
496-505, 1994. H. Chung, A. G. Rushton, Lees, F. P, P. Heino, “HAZID, A
[9] O. E. Iyun, “Plant-wide diagnosis: Cause-and-effect analysis using Computer Aid for Hazard Identification: 1. The Stophaz Package
process connectivity and directionality Information.”, Ph.D. and the Hazid Code: An Overview, the Issues and the Structure.”
Thesis, Imperial College, UK, 2012. Process Safety and Environmental Protection, 77, 317-327, 1999
[10] P. Baybutt, “Cognitive biases in process hazard analysis. Journal [32] R. Vaidhyanathan, V. Venkatasubramanian, “A semi-quantitative
of Loss Prevention in the Process Industries”, 43, 372-377, 2016. reasoning methodology for filtering and ranking HAZOP results in
[11] J. Zhao, L. Cui, L. Zhao, T. Qiu, B. Chen, “Learning HAZOP HAZOPExpert.” Reliability Engineering & System Safety, 53,
expert system by case-based reasoning and ontology.” Computers 185-203, 1996.
& Chemical Engineering, 33, 371- 378, 2009. [33] H. Graf, H. Schmidt-Traub, “An integrated approach to early
[12] J. Parmar, F. Lees, “The propagation of faults in process plants: process hazard identification of continuous and batch plants with
hazard identification. Reliability Engineering, 17, 277-302, 1987. statechart modelling and simulation.” Computers & Chemical
Engineering, 25, 61-72, 2001.

292

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY ROORKEE. Downloaded on February 23,2025 at [Link] UTC from IEEE Xplore. Restrictions apply.