Education for Chemical Engineers 34 (2021) 33–46

Contents lists available at ScienceDirect

Education for Chemical Engineers

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

Full Length Article

Process safety concerns in process system digitalization

Faisal Khan a,∗ , Paul Amyotte b , Sunday Adedigba a
a
Centre for Risk, Integrity, and Safety Engineering (C-RISE), Faculty of Engineering and Applied Science, Memorial University, St. John’s, NL, A1B 3X5, Canada
b
Department of Process Engineering and Applied Science, Dalhousie University, Halifax, NS, B3H 4R2, Canada

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

Article history: While efforts to use digital solutions in process operations are gaining wider acceptance, there are serious
Received 15 May 2020 safety concerns that need to be addressed when adopting digitalization. Process operations have evolved
Received in revised form 18 October 2020 from batch operation to continuous operation, and from smaller plants to large-scale plants. Automa-
Accepted 10 November 2020
tion and digitalization of processes, especially in process monitoring, instrumentation, and control are
Available online 12 November 2020
becoming the norm. Safety issues have also evolved with these developments, from simple equipment
failure to failure of process systems (equipment with electronic systems), monitoring and control sys-
Keywords:
tems, data encryption systems, and most recently, software systems. How these evolving process safety
Process safety
Safety education
issues should be taught in the classroom to educate and train the next generation of chemical engineers is
Digitalization a challenge with an opportunity. If such issues are not taught in academia, this will create a gap between
Operational safety education and practice, which would have a negative impact on the overall safety of process facilities.
Safety in digital operation Therefore, proactively converting this challenge to an educational opportunity and bringing digital pro-
cess safety issues into the classroom are of paramount importance to help reinforce the concept of making
process safety learning a conscious choice. This will hopefully lessen our reliance on learning from acci-
dents. The current paper presents the need to incorporate digital process safety as part of the chemical
engineering curriculum to adequately address the process industry’s emphasis on digital solutions in
process operations.
© 2020 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction chemical engineering curriculum. It is essential that undergradu-

ate chemical engineering programs capture these emerging process
As a specialized discipline aimed at managing the integrity of safety concerns related to process digitalization.
process systems, process safety is an emerging area in undergradu- The current opinion article reviews process safety education
ate chemical engineering curricula. The primary focus of process in chemical engineering curricula with the goal of identifying
safety is the prevention and mitigation of incidents that arise potential inadequacies in terms of digital process safety education.
from the realization of process hazards. These incidents include Emphasis is placed on the significance of digital process safety edu-
– although are not limited to – fire, explosion, and the release cation in the process digitalization era, and a new set of topics is
of toxic materials, often due to human factors, equipment fail- proposed to remedy the gap identified in the current curricula. The
ure, and external events (Khan et al., 2016). Process safety is proposed revised curriculum explicitly considers evolving process
well-established with its own vocabulary to communicate the tech- safety issues with respect to process digitalization to develop the
niques of hazard identification, risk assessment, risk management, requisite knowledge and skills in the next generation of chemical
and safety-critical decision making. engineers via the university route.
Process digitalization, which is currently gaining wider global This paper is organized in the following manner. Section 2 gives
acceptance, is defined as the integration of digital technologies in a historical perspective of process safety as needed background
process operations for greater efficiency, increased product qual- to the detailed discussion in subsequent sections. Section 3 then
ity, and improved process safety (the latter, for example, by means presents various aspects of process digitalization and its associated
of safety instrumented systems). It has created new challenges for process safety concerns comprised of elements of process mon-
process safety that need to be understood and addressed in the itoring, predictive control, and abnormal situation management.
Section 4 suggests chemical engineering curricula improvements in
process safety education with respect to digitalized process oper-
∗ Corresponding author. ations; existing and proposed course material discussed includes
E-mail address: fi[email protected] (F. Khan). abnormal situation management, process automation and control,

https://doi.org/10.1016/j.ece.2020.11.002
1749-7728/© 2020 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

Fig. 1. Major contributions to the evolution of process safety (Amyotte et al., 2016).

process reliability and asset integrity management. Lastly, Section has revealed that such attempts to control process hazards are
5 gives summary remarks in the form of a conclusion to the paper. not always effective. Intermittent industrial catastrophes have
occurred due to the complex industrial technologies that have
been developed. Examples of major process accidents over the past
2. Process safety: a historical perspective 50 years include: Flixborough (1974), Bhopal (1984), Piper Alpha
(1988), Philips 66 (1989), Sodegaura (1992), Stockline (2004), BP
The term process safety gained international awareness due to Texas City (2005), and Macondo/Deepwater Horizon (2010) (Okoh
significant process accidents that occurred between 1960–1990 and Haugen, 2013). The 1984 Bhopal accident remains the world’s
because of progress in technology development and industrial- worst – and hence most cited – major industrial accident (Qi et al.,
ization during this period (Khan et al., 2015). In the early stages 2012; Amyotte et al., 2016; Kletz and Amyotte, 2019). A compre-
of process safety formalization, contributions by Professor Trevor hensive investigation of these accidents has revealed a variety of
Kletz and others emphasized the importance of techniques such as causes largely related to process safety cultural, organizational,
Hazard and Operability (HAZOP) studies (Gowland, 2012) for iden- and management issues. Process digitalization has had a positive
tifying potential hazards in a process system (Kletz, 2012). HAZOP impact on these causation factors and thus has the potential to help
is defined as the application of a formal, systematic critical examina- minimize the occurrence of major process events.
tion of the process and the engineering intentions of new or existing Fig. 1 presents one view of the various contributions made to
facilities to assess the potential for malfunctioning of individual pieces enhance process safety over the years. Process safety improve-
of equipment, and the consequential effects on the facility as a whole ments were initially focussed on technical aspects – for example,
(Dunjó et al., 2010). The HAZOP procedure begins by breaking down improved and more reliable safety devices. Later, the importance
process flow diagrams and relevant piping and instrumentation of the human dimension in safety performance in the process
diagrams into manageable units called nodes. A process parameter industries was acknowledged, leading to significant developments
(e.g., “flow”) is chosen for each node and a guideword (e.g., “no”) is in the management and behavioural sciences with accompanying
applied to the chosen process parameter to suggest possible devi- advances in system and cultural approaches. The current authors
ation from design intent. If the deviation is applicable, the possible have argued that the previous decade has focussed on managing
causes are determined and the associated consequences evaluated process safety information and communicating these lessons as
(Crowl and Louvar, 2019). knowledge. A similar case can now be made for the need to focus
The concept of inherent safety was also introduced into the on the impact of process system digitalization on process safety –
vocabulary of process safety by Professor Kletz while at Imperial especially as understood and taught in the curricula of university
Chemical Industries (ICI). Kletz proposed modifications in basic engineering programs.
safety technology primarily to eliminate hazards from industrial Process operations themselves have also evolved through dif-
processes instead of adding safety measures and control devices ferent stages. Fig. 2 presents a timeline for the changing stages of
to manage the hazards. The four primary building blocks or prin- industrial process operations from the era of batch operations to
ciples of inherent safety are: (1) minimization, which entails the the digitalization era. Digitalization in the present context means
use of a smaller quantity of a hazardous substance, (2) substitution, enhancement of autonomous process operation by using advanced
which deals with the replacement of a hazardous substance with methods of process monitoring, modelling, optimization, and con-
a less hazardous one, (3) moderation, which involves using haz- trol. While this may positively impact many aspects of process
ardous materials in less hazardous forms or under less hazardous safety (e.g., human factors), it may also introduce new process
conditions, and (4) simplification, which includes the elimination of hazards that act against the inherent safety principles mentioned
unnecessary process complexity through improved process design earlier.
(Khan and Amyotte, 2003; Kletz and Amyotte, 2010; Gowland, The smaller circles within the larger circles in Fig. 2 represent
2012). the relative number of process parameters in each of the succes-
Along with inherently safer design measures, numerous pas- sive stages. As indicated by the relative sizes of the larger circles,
sive or active engineered and procedural controls are typically the scale and complexity of process operations has progressively
put in place to minimize the risk of process accidents. Never- increased, with more process parameters being integrated into pro-
theless, the occurrence of many major accidents over the years

34
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

Fig. 2. Timeline of evolving industrial process operations.

cess systems based on sophisticated technological advances over As industrialization progressed and became increasingly cou-
time. Systemic failure can occur as a result of fragility in a com- pled with technological development, there was a systematic shift
plex, nonlinear system that is coupled with human interaction and from batch operation to continuous processing methods. This was
the surrounding environment. This poses significant challenges driven in part by the significant weaknesses of batch processes as
in ensuring the control and safety of engineered systems. The described above, which are eased by the continuous flow of both
cumulative effect of multiple abnormalities due to dysfunctional reactants and products. Continuous processes offer the following
interconnections among components and subsystems can propa- advantages over their batch counterparts: (i) better process con-
gate in diverse ways to increase the degree of inherent operational trollability, (ii) more opportunities for energy integration, and (iii)
risk (Venkatasubramanian, 2011). lower inventories of hazardous materials as well as reduced waste
Process operation stages and their various safety implications amounts.
are briefly summarized in the following sections. Major indus- Nevertheless, a continuous process can be somewhat deficient
trial accidents such as Bhopal have occurred at different stages in flexibility. If there is a fault, the entire system is affected, and
of the process operation life cycle; these accidents have devel- the energy efficiency of the continuous process is relatively low.
oped much-needed global awareness of process safety. Process These disadvantages have serious safety implications for contin-
system complexity continues to evolve in a manner demanding uous processes. Efforts to improve upon their safety features in
more sophisticated methods to assess and manage process safety. industrial plants led to the revolutionary concept known as process
intensification or PI (Harmsen, 2010).
2.1. Batch processing to continuous processing: safety
implications 2.2. Process intensification (PI): safety implications

Industry has in many respects evolved from batch to continuous Process intensification (PI) is defined as any chemical engi-
processing, although batch processes are still employed, especially neering development that leads to a substantially smaller, cleaner,
by specialty chemical tolling manufacturers (Seggerman, 2017). safer and more energy efficient technology (Harmsen, 2010). PI
Batch or semi-batch processes – involving discontinuous or semi- combines innovative equipment and process configurations to
continuous flow of raw materials and/or products – found broad overcome prevailing limitations and challenges through improv-
application in various process industries in the early stages of ing heat, mass and momentum transfer, minimizing equipment
industrialization. This method is generally used for plants produc- volume, increasing efficiency, and at the same time, maintaining
ing high-quality and low-volume products (Lu et al., 2018). The operability, feasibility, controllability and safety (Demirel et al.,
main benefits of batch processing are large equipment size, poten- 2019). PI facilitates multi-tasking by simultaneously allowing reac-
tially lower financial investment, flexibility, and adaptability of tion, separation, and various other process phenomena within a
flow sheeting configurations. Some of these same features give rise single unit, leading to improved chemical processes with sub-
to several safety implications for batch processes: (i) control diffi- stantial associated economic and environmental benefits (Demirel
culties, which can result in thermal runaway due to wide operating et al., 2017). Process intensification has brought about tremendous
ranges, (ii) energy integration challenges, and (iii) a high inventory change to process design by providing novel processing tech-
of hazardous materials, which can significantly increase the con- niques and equipment that offer substantial advantages in energy
sequences of process failure. Recalling the Bhopal gas tragedy, the and material efficiency, reduced environmental impact and – par-
large amount of MIC released from a ruptured storage tank was an ticularly relevant to the discussion in the current manuscript –
intermediate product of the batch process used for pesticide pro- enhanced process safety (Nikačevíc et al., 2011).
duction. A continuous processing alternative would have consumed Techniques to accomplish process intensification have been
MIC essentially as it was produced, leaving only a few kilograms at categorized as knowledge based/ heuristic, mathematical opti-
a given time for possible release; this would greatly reduce the mization, and hybrid. A heuristic technique is based on rules
consequence severity component of risk (Etchells, 2004). acquired through experience coupled with process insight at the

35
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

operational level of a given unit. The heuristic category is subdi- a dysfunctional interrelationship among system components, are
vided into three main models: data, data mining, and application; not effectively restrained by existing control mechanisms, cas-
the method is simple and fast but lacks generalization and requires cading into violation of underlying safety measures (Leveson and
extensive expert knowledge (Sitter et al., 2019). Optimization- Stephanopoulos, 2014). Accordingly, some process safety practi-
based methods require the formation of a superstructure consisting tioners have substituted “loss of control” for “loss of containment”
of every possible flowsheet alternative. The superstructure gen- as the primary reason for the occurrence of process safety events.
erated is transformed into mixed integer nonlinear programming Drawing on the discussion in Leveson (2004), further insight
(MINLP) optimization problems, with the main objective function can be gained by examining Leveson’s accident causation model,
being related to process economics. While these methods may yield which uses the concept of process control and system theory as an
more comprehensive results than a heuristic approach, MINLP sim- analogy to explain how systems can be viewed. The sociotechni-
plification may preclude new solutions. Hybrid methods integrate cal structure of a process system is considered in term of levels of
the advantages of both mathematical optimization methods and controls. Each of these levels exercises control of process threats
heuristic methods. A hybrid approach couples the simple struc- arising from various factors such as component failure, dysfunc-
ture of heuristic methods with thermodynamic insight (Sitter et al., tional interaction, and environmental disturbances. Process safety
2019). management in Leveson’s model mandates a thorough identifica-
Process intensification has played an important role in the evo- tion of constraints on process behaviors required to ensure safety;
lution of the process industries by increasing the efficiency of these groupings of constraints are enforced through either process
operations. It has, however, introduced numerous complexities operation or design. Leveson’s model thus clearly focuses attention
that have raised significant challenges for process safety. PI efforts on the role of constraints in process safety management (Leveson,
often generate new process hazards that have a higher potential to 2004).
cause operator error and equipment failure (Etchells, 2004). Pro- The advantages of advanced control in an industrial process
cess intensification can occasionally adversely affect the process include handling complex multivariable interaction, control over a
dynamic scheme, which might result in a more substantial devia- wide operating range, and the ability to satisfy a variety of oper-
tion that could cascade into failure (Luyben and Hendershot, 2004). ating constraints related to safety limits, input saturation, and
These safety implications of PI have led to the development of reli- product quality. Extreme process complexity that results in uncer-
able and precise instrumentation systems to control and manage tainty leading to control failure is the primary safety implication of
process operations – and hence a greater dependence on engi- advanced process control (Tade et al., 2008). The Richmond Refin-
neered and procedural safety measures. For example, Baldissone ery fire that occurred on August 6, 2012, is a classic example of
et al. (2014) describe the requirement for additional layers of pro- a process accident resulting from system safety deficiencies (CSB,
tection when process intensification was applied to a catalytic 2015a; Adedigba et al., 2018).
combustion treatment plant designed for the removal of volatile
organic compounds. This may be contrasted with the often-quoted
inherent safety PI benefit of minimization of hazardous inventories. 3. Process digitalization: process safety concerns
Nevertheless, process intensification rightfully commands strong
research interest globally due to its potential for efficient and eco- Process digitalization is defined as the integration of digital tech-
nomic process operations (Tian et al., 2018). nologies in process operations for the purposes of greater efficiency
and increased product quality. It also provides a means of capturing
2.3. Process control and advanced control: safety implications and converting an analog signal to a digital form that can be elec-
tronically processed or stored. Digitalization of industrial processes
To control industrial process operations efficiently and pro- is the ultimate need of today’s industry (Kayikci 2018; Vaidya et al.,
duce products with consistent quality, process control systems 2018). Process digitalization is therefore gaining wider acceptance
are adopted. Process control has been traditionally implemented in industrial operations; it offers promising benefits in term of pro-
through tight control of the main process variables at predeter- ductivity, flexibility, and quality for the chemical process industries.
mined setpoints (Moshgbar and Hammond, 2010). Process control Process digitalization, by virtue of its fundamental and insep-
automatically regulates an output variable by identifying the arable adoption of digital technologies, can cause major changes
amplitude of an output parameter from the process, subsequently at both the process and organization levels. The benefits of digi-
comparing it with the setpoint, and finally propagating or feed- talization are found in the generation of digitized operational data
ing the error signal generated back to an input variable (Dunn, and the replacement of manual process operations with software,
2005). The central principle behind this approach is the established which permit automatic data collection for effective process mon-
or assumed relationship between process inputs and process out- itoring (Parviainen et al., 2017).
puts. However, this control strategy does not adequately account In digitalized processes, operations are thus monitored and
for common-cause variations in the actual process operation. An maintained for greater efficiency and better control. However, this
early perspective paper by Christofides at al. (2007) helped shape may cause overreliance on instrumentation and data utilization,
many of the developments that have subsequently occurred in which can have adverse implications for process safety. It is well-
smart manufacturing to address this and other concerns with tradi- established that such a dependence on engineered devices and
tional process control (as discussed later in the current paper) – e.g., administrative safety measures has been a leading cause of pro-
model predictive control, plant-wide optimization and supervisory cess accidents. An important example in this regard is the disaster
control, process monitoring, and fault-tolerant process control. that occurred on December 11, 2005 at the Buncefield oil storage
Advanced process control uses a mathematical control algo- depot in the UK. The primary cause of overfilling of a petrol (gaso-
rithm that adopts different kinds of techniques such as adaption, line) storage tank was failure of the automatic tank gauging system
prediction, and optimization to control or regulate multi-input pro- (Paltrinieri et al., 2012). A similar accident occurred on October
cess variables and multi-output process operations. Such a control 23, 2009 during offloading of gasoline from a tanker ship at the
strategy is designed to produce a product with consistent attributes Caribbean Petroleum Corporation (CAPECO) facility in Bayamon,
(Moshgbar and Hammond, 2010). Some process control practi- Puerto Rico. (CSB, 2015b). Although process digitalization is a clear
tioners view process safety as a control problem, such that an step in the right direction for enhancing the efficiency of process
accident occurs when external disturbances, hardware failures or systems, its potentially adverse impacts on process safety need

36
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

to be studied and taught as part of the university undergraduate No single process monitoring technique possesses all the desir-
chemical engineering curriculum. Important elements of process able characteristics for an ideal diagnostic system. As reviewed
digitalization and their process safety implications are now briefly in the current section, detailed information about process moni-
discussed. toring techniques can be found in the suite of papers written by
Venkatasubramanian et al. (2003a, 2003b, 2003c).

3.1. Process monitoring 3.2. Predictive control

Modern industrial systems are intricate in both their basic struc- Model predictive control (MPC) is a type of control technique
ture and the level of automation. Reliability and safety concerns that obtains the current action by solving a finite horizon open-loop
regarding such complex systems are increasingly viewed as being optimal control problem at every interval of the sampling period,
among the most critical factors for process system design (Yin et al., using the current information or state of the process plant as the
2014). Process monitoring techniques are therefore adopted in pro- input or initial state (Mayne et al., 2000). This optimization process
cess operations as tools for assurance of both process safety and gives an optimal control sequence, the foremost of which is used
product quality. by the process plant (Mayne et al., 2000). MPC shows an excellent
Process monitoring with the ultimate aim of fault detection capability to handle complex constrained optimization control dif-
and diagnosis are inseparable components of modern industrial ficulties in various aspects of advanced manufacturing, aerospace
process operations (Ge et al., 2013). A comprehensive treatment applications, and the energy industry. This capability comes from
of research on process fault detection and fault-tolerant control the ability of MPC to predict future features or characteristics of
is given in the book by Mhaskar et al. (2013). In a recent study, dynamic systems.
Alauddin et al. (2018) have provided a bibliometric analysis of fault There have been several recent studies relating various aspects
detection methods applicable to process systems. Their analysis of process control with process operational safety. These efforts
demonstrates the strong linkages among process monitoring, fault emphasize taking a system perspective of process safety, which
detection, and subsequent safety interventions. facilitates analysis of process incidents as events that occur because
Fault diagnosis methods are generally classified according to of changes in process state over a period of time. (This is similar to
the type of model on which they are based: quantitative, qualita- the previously mentioned incident causation view of loss of con-
tive, or process history (Venkatasubramanian et al., 2003a, 2003b, trol leading to loss of containment.) Detailed examination of the
2003c). Qualitative model-based methods express a relationship dynamic interaction between feedback control and chemical pro-
concerning the qualitative function of the various units in each pro- cess safety systems in the context of both classical and model-based
cess. Usually, the qualitative model is developed using abstraction control was conducted by Zhang et al. (2018). An approach that
hierarchies (structural and functional) or qualitative causal models coordinates safety and control systems through an indexing (com-
(fault trees, digraphs, and qualitative physics). Qualitative model- mon metric) procedure has also been undertaken (Albalawi et al.,
based methods are typically not very robust (Venkatasubramanian 2017a).
et al., 2003b). Albalawi et al. (2017b) proposed incorporation of a distributed
Process history-based methods make use of the availability of architecture with Lyapunov-based economic model predictive con-
large amounts of historical process data. In these methods, the trol (LEMPC) for process systems (a theme further pursued by Wu
data is transformed to extract salient features that are fed as et al., 2019). Chen et al. (2020) have presented a framework for iden-
prior information or knowledge to a diagnostic system. The fea- tifying various forms of cyber-attack on process systems. In their
ture extraction process can be undertaken either qualitatively or work, a neural network-based detector is combined with a two-tier
quantitatively (Venkatasubramanian et al., 2003c). Widely used control architecture that reconfigures the control system to enable
qualitative extraction methods include expert systems and qual- a return to the original steady state whenever the process system
itative trend analysis (QTA). An expert system has the advantages is threatened by cyber-attack. Additionally, a real-time machine
of clear reasoning, rapid development, and a robust capability to learning approach was used by Wu et al. (2020) to optimize process
reason under uncertainty. However, knowledge-based systems are operational safety for nonlinear systems undergoing time-varying
usually generated from expert rules, are system specific, have lim- disturbances. Other recent developments in the peer-reviewed lit-
ited representation power, and can be challenging to update. QTA is erature concerning process operational safety via model predictive
advantageous because it offers detailed explanations for different control and control design are described by Albalawi et al. (2018).
categories of events taking place in the process due to malfunction- Although MPC has successfully gained wide acceptance in
ing diagnosis; it can also predict the future state. However, filtering industrial process operations as an active control technique for
is needed to obtain a signal trend that will not be susceptible to overcoming constrained optimization challenges, it exhibits sev-
variation because of noise. Widely used quantitative extraction eral limitations that affect the safety of industrial process systems
methods include neural networks and statistical techniques such as (Xi et al., 2013). These limitations are related to the following fea-
PCA (principal component analysis) and statistical classifier. Quan- tures of current industrial MPC algorithms that are (Xi et al., 2013):
titative extraction methods are prone to shortcomings related to
classification error and adaptability (Venkatasubramanian et al.,
• primarily appropriate for slow dynamic processes with extensive
2003c).
sampling opportunities,
Quantitative model-based methods generally rely on input-
• principally limited to linear or quasi-linear processes and there-
output models and state space models. First-principle models and
fore unable to handle nonlinear control problems, and
frequency-response models have also been developed. The princi-
• limited in effectiveness due to their inherent heavy computa-
pal advantage of quantitative-based model techniques is that they
tional burden.
offer control of the characteristic of the residuals. Nevertheless,
many process system attributes, such as process nonlinearity, lack
of data and high dimensionality, limit their application in industrial The effects of these weaknesses of predictive control on the
processes (Venkatasubramanian et al., 2003a). The general limita- reliability and safety of complex industrial processes concern-
tion of quantitative model techniques is multiplying uncertainties ing process digitalization need to be adequately understood and
due to parameter drifts in this model type. emphasized in undergraduate chemical engineering curricula.

37
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

3.3. Abnormal situation management abnormal situations in real time cannot be understated. The timely
detection of abnormal situations is of paramount importance to
Abnormal situations usually occur due to disturbances that prevent devastating accidents that negatively impact humans,
cause a process to deviate from its tolerable operating range. The equipment and other assets, business operations, and the natural
management of such process deviations is termed abnormal situa- environment (Khan et al., 2014).
tion management or ASM (Dai et al., 2016; Eljack and Kazi, 2016). In summary, digitalization has provided numerous advantages
Abnormal situation management provides early warning of atyp- to process operation including improvements in product quality,
ical situations coupled with timely diagnosis of abnormality root system reliability, process monitoring, fault detection and diagno-
causation, and offers decision-making support to process operators sis, and process control. One of the distinct advantages of process
to facilitate reasonable actions to restore the process to normalcy digitalization is the possibility of remote and integrated operations.
(Shu et al., 2016). In other words, ASM identifies deviations from In addition to these advantages, process digitalization offers seri-
normal operation, which may lead to failure conditions. This infor- ous challenges. These include – although are not limited to – the
mation is used in assessing risk, a further step that helps to provide relationship between digitalization and the previously described
information to prevent faults from ultimately becoming failure con- inherent safety principles, the most notable of which is simpli-
ditions. fication. Digitalized processes are, by nature, complex and thus
In digitalized processes, fault detection/diagnosis and ASM can introduce new pathways for hazard realization. Other potential
all occur in a closed loop. Fault detection approaches determine challenges include the replacement of experienced personnel with
the actual fault conditions and attempt to identify the possible digital logic solvers, susceptibility to external events, alarm flood-
hazards. Accident modelling evaluates the probabilities and con- ing, and cyber security. In addition to the examples subsequently
sequences of identified hazards when ASM fails to contain the described, the recent cyber security incident at the Saudi Aramco
relevant faults. ASM is not, however, limited to exposing process refinery in Saudi Arabia has drawn further drawn attention to some
deviations and appropriately diagnosing corresponding causes; it of the potential challenges of process digitalization (Rashad, 2020).
also focuses on decision-making to bring a process back to normal
operation. Abnormal situation management is therefore an integral
part of process safety. ASM plays a key role in dynamic risk manage- 4. Process safety education in digitalized process operation
ment by serving as a link between fault detection/diagnosis and risk
management decision-making. Dai et al. (2016) have commented Chemical engineering (ChE) is a discipline of applied science
that risk assessment is the first essential step of ASM to obtain that focuses primarily on the design and optimization of chemi-
an initial and clear outline of the risk scenario to manage, while cal and biological processes. Chemical engineering education has
fault detection/diagnosis identifies the process deviations and root made steady progress over the centuries, beginning with the first
causes of a given risk scenario. ChE course offered by the Massachusetts Institute of Technology in
As previously discussed, modern process control has introduced 1888 and the creation of the first ChE department at the Univer-
a high degree of automation to chemical processing and has con- sity of Pennsylvania in 1892 (Van Antwerpen, 1980). While various
tributed significantly to production efficiency, product quality, and paradigm shifts have brought the discipline to its current state as
process safety. Automation of abnormal situation management has presented in this paper, chemical engineering has not strayed from
not, however, been fully realized. Economic loss of about 20 billion its founding principles based on methods of processing. What has
USD (2016) has been estimated to result from abnormal situations evolved and expanded is the nature of the entities being processed
in the chemical and petroleum industries in the United States (Shu – be they commodity chemicals, petrochemical feedstock, crude oil,
et al., 2016). biological organisms, nano-size materials, and so on – even now to
In essence, the primary cause of process incidents is ultimately the inclusion of knowledge vectors such as process data.
traceable to the occurrence of an abnormal event or events. Major Emphasizing educational needs, Mkpat et al. (2018) refer to pro-
accidents are typically the result of seemingly insignificant inci- cess safety as the learning of operating disciplines and safety through
dents, near misses, or unsafe behaviours – all of which suggest a systematic approach, with a view to preventing major accidents in
abnormal situations. The primary causes of an abnormal situa- the process industry. They further remark that process safety educa-
tion in process systems are usually human error (brought about tion can be acquired through three distinct avenues, namely: (i) a
by inadequate human factor considerations), process excursions university-based route involving the spectrum of curricula from the
outside the design envelope (recall the previous discussion on Bachelor’s degree to the Doctoral degree, (ii) a professional route
HAZOP), equipment failure, or any combination thereof. Inability involving industry internships, on-the-job training, and continuous
to recognize abnormal situations, poor appreciation of their effects, professional development (CPD), and (iii) training by government
and improper techniques for handling abnormal conditions, are regulatory agencies (Mkpat et al., 2018).
responsible for the often devastating failures of abnormal situation An additional avenue for formally enhancing process safety
management (Eljack and Kazi, 2016). knowledge is the establishment by technical colleges of train-
The inherent characteristics of industrial system data create ing programs for personnel working in the chemical, oil, and gas
numerous challenges for effective process fault diagnosis. Such data industries as operators. All of these approaches are, of course,
are often nonlinear, highly time-variant, and non-Gaussian in dis- not mutually exclusive; they clearly emphasize the desirability
tribution. Descriptions of satisfactory application of fault diagnosis of a continuum view of learning that incorporates contributions
techniques in real industrial processes are rare (Shu et al., 2016). from academia, industry, and government. Process safety educa-
Although process safety in conjunction with abnormal situation tion should not be limited to only chemical engineers, but should
management has undergone significant improvements globally be provided to all who work in the process and related industries.
over the past couple of decades, this has not translated into a sig- The current paper is focussed primarily on suggested revisions
nificant reduction in major process accidents. Further advances in to the ChE undergraduate curriculum in light of the rise of process
real-time process fault detection and diagnosis (abnormal situation system digitalization and the commensurate concerns for process
management) will play a large role in preventing major industrial safety assurance. Here, and in subsequent sections, we also com-
accidents (Shu et al., 2016). ment briefly on the need for process safety education in general
For safe and efficient operation of modern process systems in and in what form this might occur. There appears to be an ongoing
the digitalization era, the urgent need for a protocol that detects discussion – a healthy debate of sorts – among chemical engineer-

38
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

ing educators on whether process safety should be taught at the Inc.), (iii) UCL – IChemE (Institution of Chemical Engineers), (iv)
undergraduate level by means of a dedicated, stand-alone course USM – BEM (Board of Engineers Malaysia) via EAC (Engineering
(e.g., the course described by Amyotte, 2013 and Amyotte et al., Accreditation Council), and (v) Wits – ECSA (Engineering Council
2019), or by integration into core ChE courses (i.e., throughout of South Africa). The USM and Wits web sites also reference the
the entire chemical engineering curriculum; Venkatasubramanian, Washington Accord, which is an international agreement con-
2011). As with the general approaches identified by Mkpat et al. cerning the substantial equivalency of engineering accreditation
(2018), these two possibilities for process safety undergraduate systems. Accreditation is an important factor to which we return
teaching are not mutually exclusive. Both are desirable and fea- in later discussion.
sible. Effectively addressing the common refrain – “Who will teach
this material? and How can we accommodate this material in an We first make a few general observations on the curricula in
already full curriculum?” – requires collegial dialogue and collab- Table 1. One sees clear evidence of progression from an early
orative endeavours (see, for example, Amyotte, 2004 and Amyotte emphasis on the natural sciences (primarily chemistry and physics)
and McCahill, 2004). and mathematics, to drawing in engineering science concepts
from chemical engineering and other engineering disciplines (e.g.,
4.1. Core chemical engineering curriculum and process safety mechanical and electrical) – with the extent of the latter depending
on the particular approach to breadth of scope. As time-in-program
Table 1 lists the chemical engineering undergraduate courses continues, undergraduate students gain exposure to, and experi-
taught at six different universities. Memorial is the home institution ence with, discipline-specific engineering science content such as
for two of the current authors, while U of T, Yale, UCL, USM, and Wits ChE thermodynamics as well as components of engineering design,
represent a cross-section of well-established programs located in culminating in a significant capstone design project. Along the way,
various parts of the world. Table 1 is not intended as a global gen- the various ChE programs all cover what might be called the “heart
eralization of chemical engineering/process safety education. Its of chemical engineering” in terms of engineering science and engi-
purpose is to provide an entry point to subsequent discussion of neering design content: mass and energy balances, heat, mass and
pedagogical strategies and regulatory regimes. momentum transfer, reaction kinetics (to account for molecular
Apart from an attempt to ensure some geographical coverage changes by chemical means), separation processes (to account for
while also limiting the sample size, no other specific criteria were molecular changes by physical means), and process control. The
used in choosing the programs shown in Table 1. None, other than route is not linear; elements of engineering design, for example,
of course Memorial (and to a lesser extent U of T), are particu- are sometimes introduced at early stages in the curriculum.
larly well-known to the current authors. In that regard, we did Four of the programs have a stand-alone pro-
consciously avoid programs with active process safety researchers cess safety course or courses (identified in Table 1 in
and educators with whom we have collaborated (e.g., Texas A&M bold-faced italics with underlining): Memorial – Safety & Risk
University and Delft University of Technology, among others). Engineering; U of T – Risk Based Safety Management (Technical
Before analyzing the curricula given in Table 1, it is important Elective); UCL – Advanced Safety & Loss Prevention; USM – Plant
to note the following points: Safety, and Advanced Process Safety Engineering. Those programs
that do not appear to have a separate process safety course (Yale
• Course titles and groupings by year were taken directly from the and Wits), incorporate the teaching of relevant material in the
university web sites (last accessed on September 28, 2020). capstone design project and related courses (identified in Table 1
• The course order in a given year does not necessarily correspond in bold-faced italics) – as do the other four programs (Memorial, U
to the term or semester of delivery. of T, UCL and USM). The formal teaching of process safety by what-
• As per the previous point, words such as “course”, “term”, ever arrangement generally occurs toward the latter part of each
“semester”, “year”, “credit hour”, etc. are not employed con- program’s curriculum. This is not to say that integration of process
sistently throughout global academia. Therefore, whether a safety principles does not occur in other courses throughout these
program is delivered in 3 or 4 years is a moot point; it is the programs; such a determination is, however, beyond the scope of
curriculum content and delivery that are paramount. Addition- the current analysis.
ally, many chemical engineering programs worldwide extend Descriptions of the process safety courses identified in Table 1
to 5 years with the introduction of co-operative education and reveal the following fundamentals being taught:
industry internships, along with combined Bachelor’s/Master’s
programs. • Hazard identification, including hazard and operability study
• There may be other required courses for a particular program (HAZOP),
beyond those shown in Table 1. For example, Memorial also • Risk assessment, including quantitative risk assessment (QRA),
requires a number of courses termed “complementary studies”; • Risk management,
these typically have content from the humanities and social sci- • Root cause analysis (RCA),
ences. Such courses are shown explicitly in Table 1 for U of T. • Safety management systems,
• The courses in Table 1 are believed, for the most part, to be core • Human factors, and
and therefore required of all graduates from a given program. • Inherently safer design (ISD).
Through our Canadian lens of “counting courses”, it seems there
may be elective courses in the mix for USM Year 3. The university It should also be noted that a somewhat different route for
web site does refer to third- and fourth-year specialization in: (i) teaching process safety has emerged in some countries, in par-
Separation Process Engineering & Catalysis, (ii) Process Control ticular the United States. Here, recent initiatives have been aimed
Engineering, and (iii) Bio-Process Engineering & Environment. at bringing into the classroom process safety professionals with
This point is mentioned for the sake of completeness; it is not years of experience working in industry. With this “practitioner’s
critical to our analysis here. approach”, case study resources such as Kletz and Amyotte (2019)
• The final item of note is the fact that all programs identified in are the primary vehicle used to impart lessons learned from process
Table 1 are accredited by the relevant authority: (i) Memorial incidents.
and U of T – CEAB (Canadian Engineering Accreditation Board), (ii) No matter the approach, gone are the days when teaching safety
Yale – ABET (Accreditation Board for Engineering and Technology, in a ChE program meant reminding students about wearing PPE

39
 F. Khan et al.
Table 1
Undergraduate chemical engineering curricula at six selected universities.

Year Memorial University University of Toronto Yale University University College London University of Science Malaysia University of the
(Memorial) [Canada] (U of T) (Yale) (UCL) (USM) Witwatersrand (Wits)
[Canada] [United States of [United Kingdom] [Malaysia] [Republic of South
America] Africa]

1 • Chemistry I • Orientation to Engineering • General Chemistry & • Computational Modelling & • Engineering Mechanics • Introduction to
• English • Engineering Chemistry & Lab I Analysis • Engineering Practice Process & Materials
• Engineering Statics Materials Science • General Chemistry & • Design & Professional Skills I • Engineering Calculus Engineering
• Introduction to • Engineering Strategies & Lab II • Engineering Challenges • Engineering Materials • Chemistry I
Programming Practice I (includes human • Differential Calculus • Introduction to Chemical • Organic Chemistry • Physics
• Engineering factors) • Integral Calculus Engineering • Chemical Engineering • Mathematics I
Graphics & Design • Engineering Strategies & • Engineering Improv • Mathematical Modeling & Drawing • Critical Thinking*
• Mechanisms & Practice II (Introduction to Analysis I • Electrical Technology • A Social History of
Electric Circuits • Mechanics Engineering • Physical Chemistry • Advanced Engineering Technology
• Calculus I • Calculus I Analysis) • Thermodynamics Calculus
• Calculus II • Calculus II • Introduction to • Transport Phenomena • Physical & Analytical
• Linear Algebra I • Linear Algebra Computing for Chemistry
• Physics I • Fundamentals of Computer Engineers & • Introduction to Project
• Physics II Programming Scientists Management
• Physical Chemistry • Mass Balance
• Concepts in Chemical
Engineering
• Introduction to Chemical
Engineering & Applied
40

Chemistry

2 • Chemistry II • Chemical Engineering & • Organic Chemistry & • Chemical Reaction • Fluid Flow for Chemical • Computing for
• Engineering Applied Chemistry – Lab I Engineering I Engineering Process Engineers
Professionalism I Laboratory I • Organic Chemistry & • Design & Professional Skills II • Energy Balance • Process Engineering
• Engineering • Chemical Engineering & Lab II • Engineering • Mass Transfer Fundamentals
Mathematics Applied Chemistry – • Physics I Experimentation • Computer Programming & • Energy Balances &
• Introduction to Laboratory II • Physics II • Modelling & Analysis II Applications Applications
Process Engineering • Process Engineering • Multivariable • Separation Processes I • Biotechnology for Engineers • Chemistry II
• Thermodynamics I • Fluid Mechanics Calculus • Particulate Systems & • Process Heat Transfer (Metallurgy)
• Chemistry & Physics • Applied Chemistry I – • Ordinary & Partial Separation Processes II • Chemical Engineering • Electrical
of Engineering Inorganic Chemistry Differential • Process Design Principles Thermodynamics Engineering
Materials I • Applied Chemistry II – Equations • Process Heat Transfer • Engineers in Society • Economic Concept

Education for Chemical Engineers 34 (2021) 33–46

• Advanced Calculus Organic Chemistry • Minor I • Mathematical Methods for • Mathematics II
• Process Engineering • Calculus III Chemical Engineering
Thermodynamics • Engineering Economic • Chemical Engineering
• Process Analysis Laboratory I
Mathematical • Communication
Models • Heat & Mass Transfer
• Process Engineering • Process Dynamics:
Calculations Modeling, Analysis &
• Process Fluid Simulation
Dynamics I • Statistics
• Environmental Chemistry
 F. Khan et al.
Table 1 (Continued)

Year Memorial University University of Toronto Yale University University College London University of Science Malaysia University of the
(Memorial) [Canada] (U of T) (Yale) (UCL) (USM) Witwatersrand (Wits)
[Canada] [United States of [United Kingdom] [Malaysia] [Republic of South
America] Africa]

3 • Probability & • Chemical Engineering & • Physical Chemistry I • Advanced Safety & Loss Prevention • Plant Safety • Transport
Statistics Applied Chemistry – • Physical Chemistry II • Process Plant Design Project • Separation Processes Phenomena
• Mass Transfer Laboratory III • Thermodynamics (includes safety issues) • Chemical Reaction • Chemical
• Process Heat • Chemical Engineering & • Fluid Mechanics • Chemical Reaction Engineering Engineering
Transfer Applied Chemistry – • Kinetics & Reactors Engineering II • Process Dynamics & Control Thermodynamics
• Process Equipment Laboratory IV • Transport Processes • Process Dynamics & Control • Environmental Engineering • Chemical
Design I • Engineering • Engineering Elective • Transport Phenomena II & Management Engineering
• Process Equipment Thermodynamics • Minor II • Transport Phenomena Laboratory
Design II • Process Design (includes • Minor III • Reactor Design & Analysis • Mass Transfer
• Chemistry & Physics HAZOP) • Chemical Engineering • Numerical Methods
of Engineering • Reaction Kinetics Laboratory II • Chemical Reaction
Materials II • Professional Engineering • Downstream Processing of Engineering
• Process Modelling & Consultancy Biochemical & • Chemical Design
Analysis • Separation Processes Pharmaceutical Products Principles
• Chemical Reaction • Process Control • Renewable & Alternative • Environmental
Engineering • Chemical Reaction Energies Process Engineering
• Process Fluid Engineering • Biochemical Engineering
Dynamics II • Team Strategies for • Process Intensification in
• Technical Stream Engineering Design Chemical Processes
Course (includes process safety • Nanomaterials in Chemical
management, occupational Engineering
health & safety) • Oleochemical Processing
• Technical Electives
41

• Complementary Studies
Electives

4 • Technical Stream • Chemical Plant Design • Chemical Not Applicable • Advanced Process Safety Engineering• Chemical
Courses: must • Professional Practice Engineering Process • Plant Design & Economics Engineering Design
include • Thesis Design (includes (includes plant safety, (includes HAZOP,
Safety & Risk Engineering • Technical Electives (include safety & hazard reliability and operability, layout, safety
for Process Technical Risk Based Safety Management) analysis) plant layout, plant startup, impacts)
Stream • Complimentary Studies • Separation/Purification commissioning, and • Hydrometallurgy
• Process Engineering Electives Processes shutdown) • Management for
Project I (includes • Free Elective • Process Control • Process Design & Analysis Process Engineers
safety issues) • Chemical • Chemical Engineering • Solid Fluid Systems
• Process Engineering Engineering Lab Laboratory III • Process Control

Education for Chemical Engineers 34 (2021) 33–46

Project II (includes • Engineering • Wastewater Treatment • Advanced Chemical
safety issues) Electives Engineering Engineering B
• Process Dynamics & • Petroleum & Gas Processing • Advanced Chemical
Control Engineering Engineering C
• Process Plant Design • Final Year Project • Advanced Chemical
& Economics • Advanced Control System for Engineering D
• Engineering Industrial Processes • Chemical
Professionalism II • Membrane Science & Engineering
• Elective Course Technology Research Project
• Biorefinery Engineering • Biochemical
Engineering

*
It is interesting to note the inclusion of a course on Critical Thinking in the first year of the chemical engineering program at University of the Witwatersrand. As an essential skill for process safety practitioners, a strong case
can be made for its incorporation (like process safety itself) throughout the undergraduate curriculum of all chemical engineering programs.
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

(personal protective equipment) in the laboratory and posting a A major driver for further process safety and general chemical
copy of the applicable OH&S (Occupational Health & Safety) regu- engineering curriculum reform is the advent of the fourth indus-
lations. These are important matters to be sure; but they are not trial revolution based on smart technology. Industry 4.0 is primarily
the sole safety knowledge chemical engineering undergraduates based on the industrial internet of things (IIoT), Internet of Services
should be taking with them into industry, government, or further (IoS), Internet of manufacturing services (IoMS), and artificial intel-
academic study. The university educational environment cannot ligence and cyber-physical systems (CPS). The basic features of the
neglect its responsibility toward teaching chemical process safety fourth industrial revolution are: (1) digitization and integration of
(Louvar and Hendershot, 2003). vertical value chains, (2) digitization and integration of horizontal
Two factors are central to the evolution of process safety peda- value chains, (3) end-to-end integration across the entire product
gogy. First, we now have excellent resource material on which to life cycle, and (4) acceleration through smart technology (Moktadir
base our teaching efforts – whether these be in a dedicated pro- et al., 2018; Vaidya et al., 2018; Carvalho et al., 2018). Digitization
cess safety course or a program-wide initiative to integrate process and integration of vertical value chains implies full integration of
safety concepts. There are several textbooks now available for this process manufacturing plants with cyber-physical systems (CPS) to
purpose; as illustrative examples, we cite CCPS (2016); Benintendi create a dynamic production entity. The positive impact of smart
(2018), and Crowl and Louvar (2019). Benintendi (2018) and Crowl technology is referred to as acceleration (Moktadir et al., 2018).
and Louvar (2019), in particular, demonstrate to undergraduate Just as overall chemical engineering curriculum reform is needed
students that process safety is a highly quantitative subject as well to ensure graduates are familiar with changes in process opera-
as being also qualitative. What to teach with respect to process tions as a result of process digitalization, process safety curriculum
safety is not a mystery. reform must keep pace. Lee et al. (2019) recently concluded that:
Second, as indicated earlier, engineering accreditation systems Industry 4.0 offers the opportunity to integrate and leverage current
worldwide are driving the imperative for process safety education. technologies and modelling approaches to improve process safety.
Accreditation is one of the educational determinants for a process Table 2 presents some thoughts in this direction by illustrat-
safety course that was identified by Amyotte (2013). In Canada, the ing relevant topics grouped under four subject areas: (i) abnormal
professional accreditation expectation concerning safety is that the situation management, (ii) process automation and control, (iii)
instruction is commensurate with the hazards and risks brought process reliability, and (iv) process safety and asset integrity.
about by the nature of the specific engineering discipline. The need Table 2 is based on the experience of the current authors and
for chemical engineering students to study process safety is thus their colleagues in delivering core and elective undergraduate
evident from their potential exposure to flammable, explosible and courses and graduate courses in these areas. While abnormal situ-
toxic substances over the course of their careers. Simply put, no ChE ation management and aspects of process automation and control
graduate should enter the world of professional engineering with- have appeared in earlier sections of the current paper, the newly
out knowing the meaning of HAZOP and without even rudimentary introduced topics of process reliability and asset integrity are
proficiency in its application. becoming increasingly important as physical assets undergo aging
and degradation. Additionally, the aforementioned increased com-
4.2. Core chemical engineering curriculum and process plexity brought about by digitalization means that the reliability of
digitalization software and associated information technology accessories plays a
critical role in overall safety and security of modern process plants.
The discussion in the previous section is intended to convey a Although “making safety an elective” is inherently subject
sense of guarded optimism about the current state of process safety to misconceptions about a program’s culture and commitment
education. However, it should not be viewed as a broad endorse- toward safety, it can be an effective means to introduce new mate-
ment of the extent of coverage of process digitilization concepts. In rial into the curriculum. This was precisely the approach taken by
absolute terms, there remains much to be accomplished; in relative one of the authors in establishing the now mandatory fourth-year
terms, the role of process safety in chemical engineering education process safety course (Industrial Safety & Loss Management) at
has advanced significantly from even a decade ago. Dalhousie University. The course began about 20 years ago as a
A key part of what remains to be accomplished is full integration graduate level offering that was open to undergraduate ChE stu-
of process safety concepts throughout the undergraduate chemical dents as a technical elective. The subject matter resonated with
engineering curriculum, much like the integral role of mass and undergrad students to the extent that there was a groundswell of
energy balances in a chemical engineer’s skillset. As engineering support for introducing the course into the undergraduate curricu-
educators, we advise our students not to compartmentalize – i.e., lum. Thus, while many of the topics listed in Table 2 might seem
not to tuck away bits of knowledge once a course has been com- more suitable for graduate than undergraduate level courses, this
pleted, never to retrieve those thoughts again for future use. Having alternate approach could provide an effective interim measure for
learned how to do a mole balance in a first- or second-year course, curriculum modification.
the material returns time and again in later parts of the curriculum. Of course, one does not have to view the four columns in Table 2
Ideally, process safety will eventually reach the same state. as representative of four distinct courses. As described previously
In a similar manner, digitalization cannot be comprehensively in this section, they can also be thought of as potential topics for
taught only in a single standalone course. It should also be woven integration into existing ChE courses while the usual content of
throughout the curriculum by embedding relevant digitilization these courses is modified to address process digitalization technol-
concepts in courses such as thermodynamics, reaction engineer- ogy. New digitalization topics such as big data analytics, artificial
ing, process modelling and simulation, and process dynamics and intelligence, and machine learning applications could become part
control. Further, process safety should be considered an essen- of foundational mathematical methods courses that typically focus
tial element of the digitalization topics embodied in the various on traditional numerical modelling and analysis.
courses. For example, a process simulation course typically exposes A decade ago, Venkatasubramanian (2009, 2011) advocated for
students to the use of best-practice, commercial software tools used changes to ChE modelling and control courses to address situa-
for design purposes. Such educational efforts could also demon- tions for which new tools beyond DAEs (differential and algebraic
strate how noise and inaccuracies in process data can threaten equations) were becoming available. Part of his motivation was to
process simulation results and introduce new hazards into the pro- accommodate changing process safety issues with changes in tech-
cessing system. nology. In this vein, the topics given in Table 2 – alarm management,

42
 F. Khan et al.
Table 2
Subject areas and suggested topics to address process safety concerns with respect to process system digitalization.

Abnormal Situation Management Process Automation and Control Process Reliability Process Safety and Asset Integrity

Introduction State Space Theory Failure Distribution Accident Modelling

•Different layers of control and •Modelling in space •Reliability function •Accident process concept
monitoring •Relationship between transfer function and •Mean time to failure •Domino theory
•Abnormal event management state space model •Conditional reliability •STAMP (System-Theoretic Accident Model
(AEM); fault detection diagnosis •Analysis of state space model, controllability, and Processes) model
(FDD) observability •Layers of protection
•Other models in literature
Model-Based Monitoring Estimation of States Time-Dependent Failure Models Consequence Modelling
Techniques •Kalman filter (KF) •Weibull distribution •Fire and explosion modelling
•Modelling of faults •Extended Kalman filter (EKF) •Normal distribution •CFD-based release and dispersion
•Fault detectability, insolubility, •Unscented Kalman filter (UKF) •Lognormal distribution •CFD-based fire modelling (pool fire, fireball,
identifiability •Gamma distribution jet fire, flash fire)
•Parameter estimation, parity •CFD-based explosion and BLEVE modelling
relations, residual generation,
detection properties of residuals,
isolation properties of residuals,
residual evaluation, thresholding
Data-Based Monitoring Introduction to Process Modelling and System Reliability Human Factors in Engineering
Techniques Parameter Estimation •Serial configuration •Definitions: human error, human factors
•Univariate statistical methods: •What is identification or estimation? •Parallel configuration •Human reliability assessment model
43

data sets, sampling interval, •Main issues in model building •Combined series-parallel systems •Various approaches for human error
sample size, data pre-processing, •Model structures •Minimal cuts and minimal path prediction: e.g., SLIM (Success Likelihood Index
lag, order selection •Common mode failures Methodology), HEART (Human Error
•T2 statistics and thresholds Assessment and Reduction Technique)
•Multivariate statistical methods:
principal component analysis
(PCA), reduction order, fault
detection, fault identification, fault
diagnosis, dynamic PCA, other
PCA-based methods; partial least
squares (PLS), PLS algorithms, fault
detection, fault identification, fault
diagnosis; Comparison of PCA and

Education for Chemical Engineers 34 (2021) 33–46

PLS
Steps in Model Building State-Dependent Systems Risk Assessment
Alarm Management
•Design of experiments •Markov analysis •Process hazard analysis (PHA)
•Alarm philosophy codes and standards
•A priori analysis of data •Load-sharing systems •Elements of risk analysis
•Alarm data analysis
•Model structure determination Standby systems Quantitative risk assessment (QRA)
•Alarm rationalization
•Concepts of system identifiability, bias and •Degraded systems •Risk representation and •ALARP (as low as
•Alarm flooding and impacts on process safety
consistency •Three-state devices reasonably practicable)
•Binomial, Poisson and hypergeometric
processes
•Risk simulation techniques
 F. Khan et al.
Table 2 (Continued)

Abnormal Situation Management Process Automation and Control Process Reliability Process Safety and Asset Integrity

Parameter Estimation Methods Design for Reliability Probabilistic Risk Assessment

• Guiding principles behind parameter • Reliability specification and system • Data modelling for risk assessment
estimation methods measurements • Advanced techniques: Bowtie analysis,
• Minimizing prediction errors and the • Reliability allocation Bayesian network analysis
prediction error approach • Design methods • Specialized software packages
• Asymptotic distribution of parameter • Failure analysis • Risk management
estimates • System safety and fault tree analysis

Computing the Estimate Maintainability Uncertainty Analysis

• Linear regression and least squares • Analysis of downtime • Fundamentals of uncertainty
• Subspace methods for estimating state space • Repair time distribution • Uncertainty types
models • Stochastic point process • Analysis techniques
• System repair time • Representation of uncertainty
• Reliability under preventive maintenance
• State-dependent system with repair

Model Validation Methods Availability Asset Integrity

• Strategies for end-use of model (prediction • Inherent availability • Concept of asset
and control) • Operational integrity
• Validation – time domain availability • Reliability centred
44

• Validation – frequency domain • Exponential maintenance (RCM)

availability model • Risk based
Analysis of Multivariable Systems • System availability maintenance (RBM)
• Transfer functions for multiple input, model • Risk based
multiple output (MIMO) systems • Inspection and inspection of static
• Multivariable frequency response repair availability equipment
• Directions in multivariable systems, model
right-half plane (RHP) zeros
• Analysis using eigenvalues
• Analysis using singular value decomposition
(SVD)
• Analysis using relative gain array (RGA)
• Control structure design

Education for Chemical Engineers 34 (2021) 33–46

Controller Design
• Trade-offs in controller design
• Pre- and post-compensator, decoupling, SVD
controller
• State feedback control
• Model predictive control (MPC)
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

model predictive control, Markov analysis, Bayesian networks, and Amyotte, P.R., Khan, F.I., Irvine, Y., 2019. Continuous improvement in process
the like – must be viewed in light of their process safety implica- safety education. Chem. Eng. Trans. 77, 409–414, 16th International Sympo-
sium on Loss Prevention and Safety Promotion in the Process Industries, Delft,
tions in addition to their technical definitions and uses. As a final
Netherlands, June 16-19, 2019).
comment, we note that Table 2 and similar listings are dynamic Baldissone, G., Cavaglia, G., Demichela, M., 2014. Are intensified processes safer and
compilations. With an eye on the technical literature, we should more reliable than traditional processes? An emblematic case study. Chem. Eng.
Trans. 36, 415–420.
also keep in mind changing industry needs related to process secu-
Benintendi, R., 2018. Process Safety Calculations. Elsevier, Oxford, UK.
rity (physical and cyber security (e.g., Chen et al., 2020), both Carvalho, N., Chaim, O., Cazarini, E., Gerolamo, M., 2018. Manufacturing in the fourth
internal and external to a company or facility) and how these con- industrial revolution: a positive prospect in sustainable manufacturing. Procedia
siderations might find their way into our classrooms. (Moreno et al., Manuf. 21, 671–678.
CCPS (Center for Chemical Process Safety), 2016. Introduction to Process Safety for
2018). Undergraduates and Engineers. Wiley, Hoboken, NJ.
Chen, S., Wu, Z., Christofides, P.D., 2020. A cyber-secure control-detector architec-
ture for nonlinear processes. Aiche J. 66, e16907.
5. Concluding remarks Christofides, P.D., Davis, J.F., El-Farra, N.H., Clark, D., Harris, K.R.D., Gipson, J.N.,
2007. Smart plant operations: vision, progress and challenges. Aiche J. 53,
2734–2741.
The chemical process industries (CPI) have gone through dra-
Crowl, D.A., Louvar, J.F., 2019. Chemical Process Safety. Fundamentals With Appli-
matic changes in the past several years due to technological cations, fourth edition. Pearson, Boston, MA.
developments. Process digitalization has rapidly gained ground in CSB (Chemical Safety Board), 2015a. Final Investigation Report – Chevron Richmond
the CPI and has brought with it associated safety concerns. As cor- Refinery Pipe Rupture and Fire. Report No. 2012-03-I-CA. US Chemical Safety and
Hazard Investigation Board, Washington, DC.
responding chemical engineering educational reform takes place, CSB (Chemical Safety Board), 2015b. Final Investigation Report – Caribbean
concurrent changes in process safety pedagogy must also occur. Petroleum Tank Terminal Explosion and Multiple Tank Fires. Report No.
The discipline of chemical engineering has about a hundred year 2010.02.I.PR. US Chemical Safety and Hazard Investigation Board, Washington,
DC.
head-start on process safety in terms of being formally taught in Dai, Y., Wang, H., Khan, F., Zhao, J., 2016. Abnormal situation management for smart
universities. A reasonable goal for process safety educators would chemical process operation. Curr. Opin. Chem. Eng. 14, 49–55.
be to explore avenues within their chemical or process engineer- Demirel, S.E., Li, J., Hasan, M.F., 2017. Systematic process intensification using build-
ing blocks. Comput. Chem. Eng. 105, 2–38.
ing department for both concept integration and development of Demirel, S.E., Li, J., Hasan, M.F., 2019. Systematic process intensification. Curr. Opin.
stand-alone process safety courses. Multiple resources – books, Chem. Eng. 25, 108–113.
journals, technical organizations, and colleagues – are available to Dunjó, J., Fthenakis, V., Vílchez, J.A., Arnaldos, J., 2010. Hazard and operability
(HAZOP) analysis. A literature review. J. Hazard. Mater. 173, 19–32.
assist in these endeavours. Collaborative revision should be the
Dunn, W.C., 2005. Fundamentals of Industrial Instrumentation and Process Control.
aim; adherence to accreditation criteria can provide motivation McGraw-Hill, New York, NY.
when needed. Eljack, F., Kazi, M.-K., 2016. Process safety and abnormal situation management.
Curr. Opin. Chem. Eng. 14, 35–41.
Etchells, J.C., 2004. Process Intensification: Safety Pros and Cons. IChemE Symposium
Declaration of Competing Interest Series No. 150. Institution of Chemical Engineers, Rugby, UK.
Ge, Z., Song, Z., Gao, F., 2013. Review of recent research on data-based process
monitoring. Ind. Eng. Chem. Res. 52, 3543–3562.
The authors declare that they have no known competing finan- Gowland, R., 2012. A journey into process safety with Trevor Kletz. J. Loss Prev.
cial interests or personal relationships that could have appeared to Process Ind. 25, 768–769.
Harmsen, J., 2010. Process intensification in the petrochemicals industry: drivers and
influence the work reported in this paper.
hurdles for commercial implementation. Chem. Eng. Process. Process. Intensif.
49, 70–73.
Kayikci, Y., 2018. Sustainability impact of digitization in logistics. Procedia Manuf.
Acknowledgements 21, 782–789.
Khan, F.I., Amyotte, P.R., 2003. How to make inherent safety practice a reality. Can.
The authors gratefully acknowledge the financial support pro- J. Chem. Eng. 81, 2–16.
Khan, M.A.I., Imtiaz, S.A., Khan, F., 2014. Early warning system for chemical pro-
vided by the Natural Sciences and Engineering Research Council cesses with time delay and limited actuator capacity. Ind. Eng. Chem. Res. 53,
of Canada (NSERC) through the Discovery Grant Program and the 4763–4772.
Canada Research Chair (Tier I) Program in Offshore Safety and Risk Khan, F., Rathnayaka, S., Ahmed, S., 2015. Methods and models in process safety
and risk management: past, present and future. Process. Saf. Environ. Prot. 98,
Engineering. We also wish to thank the anonymous peer reviewers
116–147.
for their suggestions to improve the original manuscript. Khan, F., Hashemi, S.J., Paltrinieri, N., Amyotte, P., Cozzani, V., Reniers, G., 2016.
Dynamic risk management: a contemporary approach to process safety man-
agement. Curr. Opin. Chem. Eng. 14, 9–17.
References Kletz, T., 2012. The history of process safety. J. Loss Prev. Process Ind. 25, 763–765.
Kletz, T., Amyotte, P., 2010. Process Plants. A Handbook for Inherently Safer Design,
Adedigba, S.A., Khan, F., Yang, M., 2018. An integrated approach for dynamic eco- second edition. CRC Press/Taylor & Francis Group, Boca Raton, FL.
nomic risk assessment of process systems. Process. Saf. Environ. Prot. 116, Kletz, T., Amyotte, P., 2019. What Went Wrong? Case Histories of Process Plant
312–323. Disasters and How They Coud Have Been Avoided, sixth edition. Elsevier, Oxford,
Alauddin, M., Khan, F., Imtiaz, S., Ahmed, S., 2018. A bibliometric review and analysis UK.
of data-driven fault detection and diagnosis methods for process systems. Ind. Lee, J., Cameron, I., Hassall, M., 2019. Improving process safety: what roles for digi-
Eng. Chem. Res. 57, 10719–10735. talization and Industry4.0? Process. Saf. Environ. Prot. 132, 325–339.
Albalawi, F., Durand, H., Christofides, P.D., 2017a. Process operational safety using Leveson, N., 2004. A new accident model for engineering safer systems. Saf. Sci. 42,
model predictive control based on a process safeness index’. Comput. Chem. 237–270.
Eng. 104, 76–88. Leveson, N.G., Stephanopoulos, G., 2014. A system-theoretic, control-inspired view
Albalawi, F., Durand, H., Christofides, P.D., 2017b. Distributed economic model and approach to process safety. Aiche J. 60, 2–14.
predictive control for operational safety of nonlinear processes. Aiche J. 63, Louvar, J.F., Hendershot, D.C., 2003. SACHE: 17 years of promoting teaching of safety
3304–3418. to chemical engineering students. J. Chem. Health Saf. 10 (5), 8–10.
Albalawi, F., Durand, H., Christofides, P.D., 2018. Process operational safety via model Lu, P., Chen, J., Xie, L., 2018. ILC based economic batch-to-batch optimization for
predictive control: recent results and future research directions. Comput. Chem. batch processes. IFACPapersOnLine 51, 768–773.
Eng. 114, 171–190. Luyben, W.L., Hendershot, D.C., 2004. Dynamic disadvantages of intensi-
Amyotte, P.R., 2004. A meeting of the minds. Canadian Chemical News 56 (2), 14–15. fication in inherently safer process design. Ind. Eng. Chem. Res. 43,
Amyotte, P.R., 2013. Process safety educational determinants. Process. Saf. Prog. 32, 384–396.
126–130. Mayne, D.Q., Rawlings, J.B., Rao, C.V., Scokaert, P.O.M., 2000. Constrained model
Amyotte, P.R., McCahill, P., 2004. A meeting of the minds – part 2: report on the 2004 predictive control: stability and optimality. Automatica 36, 789–814.
CSChE-PSM summer institute. Canadian Chemical News 56 (9), 27–31. Mhaskar, P., Liu, J., Christofides, P.D., 2013. Fault-Tolerant Process Control: Methods
Amyotte, P.R., Berger, S., Edwards, D.W., Gupta, J.P., Hendershot, D.C., Khan, F.I., Man- and Appplications. Springer-Verlag, London, UK.
nan, M.S., Willey, R.J., 2016. Why major accidents are still occurring. Curr. Opin. Mkpat, E., Reniers, G., Cozzani, V., 2018. Process safety education: a literature review.
Chem. Eng. 14, 1–8. J. Loss Prev. Process Ind. 54, 18–27.

45
F. Khan et al. Education for Chemical Engineers 34 (2021) 33–46

Moktadir, M.A., Ali, S.M., Kusi-Sarpong, S., Shaikh, M.A.A., 2018. Assessing challenges Vaidya, S., Ambad, P., Bhosle, S., 2018. Industry 4.0 - A glimpse. Procedia Manuf. 20,
for implementing Industry 4.0: implications for process safety and environmen- 233–238.
tal protection. Process. Saf. Environ. Prot. 117, 730–741. Van Antwerpen, F.J., 1980. The origins of chemical engineering. Chapter 1. In: Furter,
Moreno, V.C., Reniers, R., Salzano, E., Cozzani, V., 2018. An alysis of physical and W.F. (Ed.), History of Chemical Engineering. American Chemical Society, Wash-
cyber security-related events in the chemical and process industry. Process. Saf. ington, DC.
Environ. Prot. 116, 621–631. Venkatasubramanian, V., 2009. Drowning in Data: Informatics and Modeling Chal-
Moshgbar, M., Hammond, S., 2010. Advanced process control. Qual. Assur. J. 13, lenges in a Data-Rich Networked World. Aiche J. 55, 2–8.
62–66. Venkatasubramanian, V., 2011. Systemic failures: challenges and opportunities in
Nikačevíc, N.M., Huesman, A.E.M., Van den Hof, P.M.J., Stankiewicz, A.I., 2011. Oppor- risk management in complex systems. Aiche J. 57, 2–9.
tunities and challenges for process control in process intensification. Chem. Eng. Venkatasubramanian, V., Rengaswamy, R., Yin, K., Kavuri, S.N., 2003a. A review of
Process. Process. Intensif. 52, 1–15. process fault detection and diagnosis: part I: quantitative model-based methods.
Okoh, P., Haugen, S., 2013. The influence of maintenance on some selected major Comput. Chem. Eng. 27, 293–311.
accidents. Chemical Enginnering Transactions 31, 493–498. Venkatasubramanian, V., Rengaswamy, R., Kavuri, S.N., 2003b. A review of process
Paltrinieri, N., Oien, K., Cozzani, V., 2012. Assessment and comparison of two early fault detection and diagnosis: Part II: Qualitative models and search strategies.
warning indicator methods in the perspective of prevention of atypical accident Comput. Chem. Eng. 27, 313–326.
scenarios. Reliab. Eng. Syst. Saf. 108, 21–31. Venkatasubramanian, V., Rengaswamy, R., Kavuri, S.N., Yin, K., 2003c. A review of
Parviainen, P., Tihinen, M., Kaariannen, J., Teppola, S., 2017. Tackling the digital- process fault detection and diagnosis: part III: process history based methods.
ization challenge: how to benefit from digitalization in practice. International Comput. Chem. Eng. 27, 327–346.
Journal of Information Systems and Project Management 5, 63–77. Wu, Z., Albalawi, F., Zhang, Z., Zhang, J., Durand, H., Christofides, P.D., 2019. Con-
Qi, R., Prem, K.P., Ng, D., Rana, M.A., Yun, G., Mannan, M., 2012. Challenges and needs trol Lyapunov-barrier function-based model predictive control of nonlinear
for process safety in the new millennium. Process. Saf. Environ. Prot. 90, 91–100. systems. Automatica 109, 108508.
Rashad, M., 2020. Saudi Aramco sees increase in attempted cyber-attacks. Reuters Wu, Z., Rincon, D., Christofides, P.D., 2020. Real-time machine learning for opera-
(6 Feb, 2020). tional safety of nonlinear processes via barrier-function based predictive control.
Seggerman, B., 2017. Ensuring process safety in batch tolling. Chem. Eng. Prog. 113 Chem. Eng. Res. Des. 155, 88–97.
(12), 34–40. Xi, Y.G., Li, D.-W., Lin, S., 2013. Model predictive control – status and challenges. Acta
Shu, Y., Ming, L., Cheng, F., Zhang, Z., Zhao, J., 2016. Abnormal situation manage- Autom. Sin. 39, 222–236.
ment: challenges and opportunities in the big data era. Comput. Chem. Eng. 91, Yin, S., Ding, S.X., Xie, X., Luo, H., 2014. A review on basic data-driven approaches
104–113. for industrial process monitoring. Ieee Trans. Ind. Electron. 61, 6418–6428.
Sitter, S., Chen, Q., Grossmann, I.E., 2019. An overview of process intensification Zhang, Z., Wu, Z., Durand, H., Albalawi, F., Christofides, P.D., 2018. On integration of
methods. Curr. Opin. Chem. Eng. 25, 87–94. feedback control and safety systems: analyzing two chemical process applica-
Tade, M.O., To, L.C., Ray, M.S., 2008. Limitations on the applications of advanced pro- tions. Chem. Eng. Res. Des. 132, 616–626.
cess control in mineral processing operations. Dev. Chem. Eng. Miner. Process.
5, 195–207.
Tian, Y., Demirel, S.E., Hasan, M.M.F., Pistikopoulos, E.N., 2018. An overview of pro-
cess systems engineering approaches for process intensification: state of the art.
Chem. Eng. Process. Process. Intensif. 133, 160–210.

46