CHEMICAL ENGINEERING TRANSACTIONS
VOL. 77, 2019
A publication of
The Italian Association
of Chemical Engineering
Online at www.cetjournal.it
Guest Editors: Genserik Reniers, Bruno Fabiano
Copyright © 2019, AIDIC Servizi S.r.l.
I SBN 978-88-95608-74-7; I SSN 2283-9216
How Big Data & Analytics Can Improve Process and Plant
Safety and Become an Indispensable Tool for Risk
Management
Pankaj Goel
a,b
, Hans Pasman
a
*, Aniruddha Datta
b
a
Mary Kay O’Connor Process Safety Center, Artie McFerrin Department of Chemical Engineering,
b
Department of Electrical and Computer Engineering Texas A&M University, College Station, Texas 77843-3122
hjpasman@gmail.com
With the advances in digitization, Information Technology (IT), and connected devices, data are becoming
plentiful. And with the past 30 years of developments of Artificial Intelligence tools leading to great
enhancements in dealing with various levels and types of uncertainty, much has become tangible, where in
the past it used to remain vague and fuzzy. Tools like neural networks can distil information from datasets,
while probabilistic methods can characterize randomness. Bayesian causation networks enable finding critical
pathways and help to design and monitor effective safeguards, while Petri nets enable analysis of time-critical
events. Interval analysis, Dempster-Shafer theory, and fuzzy logic can assist in delimiting uncertainty in
measurement results and expert judgment. System dynamics modeling and Functional resonance analysis
may unravel interactively degrading processes. All this can improve understanding about communication lines
and mechanisms of interactions within a plant socio-technical system, and the influences on achievement and
performance. This will result in reformed work processes, manufacturing conditions and help in identifying
abnormal trends. Therefore, while planning and prediction are based on observational evidence and trends,
the new technologies will be a strong support for management, in recognizing and evaluating risks, including
safety risks. Although applications of big data and analytics are still young, nevertheless in process control
and reliability prediction of equipment a few achievements have already been demonstrated. However, much
more is possible. For example, in the case of process safety performance indicators, lagging indicators are
usually available but the techniques may stimulate the recording of the more important leading indicators for
the prediction of safety and culture trend in a company in relation to its economic health. The paper will
present more details on the methods and an example of dynamic risk mapping.
1. Introduction
With the advances in technology over past four decades, process plants use different control systems such as
Programmable Logic Controllers (PLC), Distributed Control System (DCS), and Supervisory Control and Data
Acquisition systems (SCADA) for monitoring and controlling the plant operations (Goel et al., 2017b). At the
same time with development in IT, communication methods and connected devices, process plants are
producing incredible amounts of data in different forms stored in ‘data lakes’ (data warehouses). This requires
new and innovative approaches and methods to create Business Intelligence and actionable insights. The
industry can get significant benefits with the use of intelligent systems and big data analytics methods. Several
attributes such as volume, variety, velocity, value, veracity, variability, and valence characterize Big Data
(Goel et al, 2017a). Figure 1a shows different data collected during process plant operations. Static data
means data or reports generated over a period and remains fixed for a considerable amount of time while
dynamic data means data, which changes with time and are continuous. Structured data refers to data in a
table or specific report formats, while unstructured refers to data primarily expressed as text. The collected
data is usually the raw data and requires pre-processing, cleaning and analysis to derive the expected
information for decision making. Figure 1b highlights the various data analysis types such as descriptive,
DOI: 10.3303/CET1977127
Paper Received: 7 December 2018; Revised: 18 May 2019; Accepted: 23 June 2019
Please cite this article as: Goel P., Pasman H., Datta A., 2019, How Big Data & Analytics can improve process and plant safety and become an
indispensable tool for risk management, Chemical Engineering Transactions, 77, 757-762 DOI:10.3303/CET1977127
757