 |
| Intelligent
Systems in Process Safety On the taming of the shrew? |
| |
ABSTRACT |
| |
While
safety in the process industries has received widespread
attention for several decades, recent advances in the
field of Artificial Intelligence (AI) promise to bear
choice tools for plant personnel. Artificial Intelligence
has been described as "the part of computer science concerned
with designing systems that exhibit characteristics we
associate with intelligence in human behaviour". |
 |
| |
From
its early beginnings in the Thinking Machines of the
1960s to the Mars Rover Sojourner of the 1990s, the
field and its applications have come a long way. These
advancements in natural language processing, computer
vision, robotics, machine learning and expert systems
have lead to practical applications in diverse fields
of human endeavour.
|
 |
| In process systems engineering,
a wide variety of areas including process design, modelling
and simulation, process planning and operations, process
control, and product design have gained from these developments
in AI. Process design and operations are inherently expertise-based
fields. This has lead to the development of expert systems
that assist humans tackle different aspects of safety
such as hazards analysis, process monitoring, preventive
control and emergency training. |
| |
Process Hazards Analysis (PHA) is the proactive identification,
evaluation, mitigation and prevention of process hazards.
A team of experts performs PHA and the analysis is repeated
periodically during the entire life of the plant from
design until decommissioning.
|
 |
| Given the enormous amounts
of time, effort and resources involved in performing such
reviews, there exists considerable incentive to develop
intelligent systems for automating the process hazards
analysis of chemical process plants. Recently, expert
systems that automate Hazard and Operability analysis
and other hazard identification and evaluation techniques
for both continuous and batch processes have been reported.
These systems exploit the fact that the analysis is systematic
and logical and a large part of any PHA study involves
routine aspects that are common across different plants.
Therefore, by suitable knowledge capture and using process
knowledge, these systems can automatically identify hazards,
their causes and adverse consequences. The appeal of such
automated systems is that they can reduce the time, effort
and expense involved in a PHA review, make the review
more thorough and detailed, minimize human errors, and
free the team to concentrate on the more complex aspects
of the analysis." |
| |
The other side of the safety
coin is ensuring the safety of a process during its operations.
The onus for this normally rests on plant operations personnel.
Any failing can result in repercussions ranging from poor
product quality and schedule delays to equipment damage,
human injury or worse. A recent study reported that such
mishaps have an economic impact of at least $20-billion
annually in the US petrochemical industry and highlighted
the need to bolster support to the plant operations personnel
especially during abnormal situations. While humans are
better at recognising patterns, solving original problems,
learning from past experience and improvising and adapting,
machines are better at storing and recalling large quantities
of information quickly, performing routine repetitive
and precise operations over long periods of time. Based
on this, recently there is a significant effort from industry
and academia to develop and deploy intelligent systems
that can act as an intelligent associate to operations
personnel. This extra member of the operations team has
skills that compliment the human operators' and is based
on AI-based collaborative decision support technologies
and is expected to significantly improve safety during
process operations. |
| |
There are several such
examples of process design and operations responsibilities
being leveraged through the use of intelligent systems.
Many of these developments are now well beyond proof of
concept and ready for industrial applications. In the
future, such intelligent systems are certain to become
plant personnel's treasured partner in taming the safety
shrew. |
 |
| |
Rajagopalan Srinivasan
Dept of Chemical & Environmental Engineering
National University of Singapore |
| |
|
|
|
|
