How to Troubleshoot Apps for the Modern Connected Worker
Gas Detectors & Detectability
1. Copyright 2010 ISA. All Rights Reserved.
Gas Detectors and
Detectability
Simon J. O’Connor
Detector Electronics (UK) Ltd
Standards
Certification
Education & Training
Publishing
Conferences & Exhibits
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
2. Copyright 2010 ISA. All Rights Reserved.
Presenter: Simon J. O’Connor
• ~30 years as a scientist working for Shell Research. Projects
included research into:
– Spark and compression ignition of gases; Open path FTIR; Flare
Spectroscopy; Open path Air Quality monitoring; Differential Absorption
LIDAR
• Last eight years as a Senior Consultant in the Fire and Gas
Detection group of Shell Global Solutions
– Over 50 projects on terminals; offshore facilities (manned and un-
manned); refineries; gas plants; chemical sites; FPSO and FLNG. Shell
and non-Shell customers.
• Joined Det-Tronics April 2010.
2
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
3. Copyright 2010 ISA. All Rights Reserved.
Introduction
• In this presentation we discuss the “detectabilty” of gas
releases.
• A brief introduction to the concept of “volumetric
coverage” and “Gas Detection Mapping” will be
presented.
• Finally, results from simple CFD analysis of releases are
presented and the correlation between these and
Mapping results are shown.
3
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
4. Copyright 2010 ISA. All Rights Reserved.
How good are existing systems?
UK Offshore hydrocarbon release database 2001-2008*
Method of detection
Human Smoke2 Heat2 Flame2 Gas Det. Process3 Hand- Total
observation held4
Major1 9 0 0 0 23 4 0 36
Significant1 34 1 1 1 18 2 1 58
1. See reference (below) for definitions
2. Original database includes smoke, heat and flame detectors for ignited events
3. “Process” – release noticed from changes in process monitors.
4. “Hand-held” – releases noticed during hand-held sniffing of components
* “Offshore hydrocarbon releases 2001 to 2008”; Alison McGillivray, John Hare, UK Health and Safety Laboratory
Research Report RR672; UK Health and Safety Executive, 2008.
4
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
5. Copyright 2010 ISA. All Rights Reserved.
How good are existing systems?
• Approximately 2/3 of “Major” and 1/3 of “Significant”
releases were detected by the fixed gas detection
system:
– Is this acceptable?
– Given the significance of human observation (25% of Major and
60% of Significant), will general reductions in staffing levels lead
to reduced detection?
– Can we identify best practice for fixed gas detection systems?
5
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
6. Copyright 2010 ISA. All Rights Reserved.
The nature of hazardous gas releases
• Liquid spills (including flashing liquids or liquefied gases):
Generally low level, heavy
Gas clouds and wind
dispersed plumes
• Pressurised gas releases:
Release direction dominates wind. Rapid air
entrainment improves dispersion
• Combinations of the above in two phase releases.
6
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
7. Copyright 2010 ISA. All Rights Reserved.
Tools of the trade: Gas detectors
• Point Gas Detectors:
– Measure the concentration of gas at the location of the detector.
May be flammable or toxic gases (Catalytic / Pellister detectors,
Electrochemical cells, Infra Red (IR) point detectors).
• Line of Sight (LOS) Detectors:
– Measure the integrated concentration of gas along a (generally)
IR beam.
• Acoustic Detectors:
– Listen for the characteristic noise associated with high pressure
gas releases.
7
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
8. Copyright 2010 ISA. All Rights Reserved.
Tools of the trade: QRA? Exceedence?
• Release frequency (failure rates) combined with 3D
dispersion and ignition models provides important
information about the risk associated with hazardous
gases.
• However:
– Resolution limited
– Link to “tolerance” unclear
– Can assess incorrect
parameters (average
detectability x average
mitigation efficiency)
• Accept tolerable cloud?
8
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
9. Copyright 2010 ISA. All Rights Reserved.
Consider a release from a site component
Site fence line
Hazardous zone
Point gas detector
9
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
10. Copyright 2010 ISA. All Rights Reserved.
Consider a release from a site component
Release in the “westerly”
direction interacts with site
structures. The jet slows,
this aided by additional
turbulence, forms an
accumulation cloud that is
detected.
10
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
11. Copyright 2010 ISA. All Rights Reserved.
Consider a release from a site component
Release in the “easterly”
exits the site undetected.
11
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
12. Copyright 2010 ISA. All Rights Reserved.
Consider a release from a site component
Larger releases do not
necessarily become easier
to detect; they form
accumulations further
from the source.
LOS ?
Acoustic ?
12
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
13. Copyright 2010 ISA. All Rights Reserved.
Pragmatic view of Gas Detection and
Mitigation
Below acceptable detection frequency
Liquid releases
Gas releases
Ineffective mitigation
Potential consequence
Effective range of application
of Fixed detectors
Event size
13
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
14. Copyright 2010 ISA. All Rights Reserved.
Volumetric coverage protection
• Following on from the Lord Cullen report after Piper
Alpha. Several studies were initiated to assess the
tolerable volumes of gas that could exist on offshore
facilities.
• The concept of a 5m diameter LFL equivalent sphere
was born.
• Detection target was to detect gas clouds before they
could achieve potentially damaging volumes and
escalate the situation.
14
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
15. Copyright 2010 ISA. All Rights Reserved.
Gas Detection Mapping
Uses a 3D model of a facility.
Releases “idealised” clouds from
all points.
Assess detector density.
Idealised sphere constructed to
allow for point or line detector
Produces a coverage “map” and
% coverage of zone to the idealised
target.
15
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
16. Copyright 2010 ISA. All Rights Reserved.
Assessing detectability: NIST 5 FDS*
Module 25x11x10m
Releases of 1.3kg/s
methane
*Fire Dynamics Simulator, Version 5.0, Building and Research Laboratory, National Institute of Standards
and Technology http://www.fire.nist.gov/fds
16
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
17. Copyright 2010 ISA. All Rights Reserved.
Flammable gas release in an enclosed,
partially enclosed and open module
T = 1s
T = 7s T = 18s
17
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
18. Copyright 2010 ISA. All Rights Reserved.
Flammable gas release in an enclosed,
partially enclosed and open module
• 6 point gas and 2 LOS (4 LOS in open module)
simulated.
• 6 release directions for enclosed and partially enclosed
modules with one wind direction. 6 release directions
and 4 wind directions for open module. 36 releases in all.
Percentage of releases
detected by:
LOS Points
Enclosed 100 100
Partially Enclosed 83 50
Open 83 38
18
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
19. Copyright 2010 ISA. All Rights Reserved.
Flammable gas release in an enclosed,
partially enclosed and open module
• Introducing one detector at a time, assessing coverage in
both CFD and volumetric Mapping software. Mapped
coverage shows a correlation to the detectability:
Enclosed Open Area
100% 100%
80% 80%
% D e te c te d
60%
% Detected
60%
40% 40%
20%
20%
0%
0%
0 25 50 75 100
0 25 50 75 100
Mapped Coverage Mapped Coverage
19
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
20. Copyright 2010 ISA. All Rights Reserved.
Toxic gas release on a process unit.
Release point Zone 60x40x10m
Releases of 10% H2S in
methane
X axis
Y axis
20
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
21. Copyright 2010 ISA. All Rights Reserved.
Toxic gas release on a process unit.
Five release rates, four wind directions; four release directions
(80 releases in total)
21
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
22. Copyright 2010 ISA. All Rights Reserved.
Toxic gas release on a process unit.
5m Grid
10m Grid
15
20
15
10
10
5
5
0 0
-20 -15 -10 -5 0 5 10 15 20 -30 -20 -10 0 10 20 30
-5
-5
-10
-10 -15
-20
-15
5m grid 10m grid
• Three grids of detectors were simulated; these were a
5m grid, a 10m grid and a 20m grid (not shown)
• The grids are unrealistic; they are simulated to test the
relationship between detector density and “detectability”
22
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
23. Copyright 2010 ISA. All Rights Reserved.
Toxic gas release on a process unit.
100
90
80 5m grid
Detectability (%) 70 10m grid
60 20m grid
50
40
30
20
10
0
0.002 0.004 0.006 0.7 1.3
Release rate (kg/s)
• Detectors initially set to alarm at 10ppm H2S.
Relationship between detectability, release rate and
detector density clearly evident.
23
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
24. Copyright 2010 ISA. All Rights Reserved.
Toxic gas release on a process unit.
39.0
38.0
Time to the first alarm (s)
37.0
36.0 5m grid
Response time
10m grid
35.0 20m grid
34.0
33.0
32.0
0 20 40 60 80 100 120
Detector alarm setting (ppm)
• Varying the detector alarm set point; the 5m grid at
100ppm alarm level, on average, responds faster than a
10m grid or 20m grid with detectors set at 10ppm.
24
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
25. Copyright 2010 ISA. All Rights Reserved.
Summary
• Guaranteeing high detection rates for hazardous gas
releases on industrial sites is not trivial due to the large
range of possible scenarios and release behaviour.
• A combination of detection technology generally provides
the best solution.
• Simple volumetric Mapping can be used to assess
conformity to target clouds in complex geometries.
• CFD modelling indicates that detectability is a strong
function of detector density (number of detectors) and
this correlates to volumetric Mapping.
25
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org
26. Copyright 2010 ISA. All Rights Reserved.
Summary
• Detector sensitivity (alarm point) is secondary; if
detectors do not intercept the plume, the situation cannot
be corrected by reducing the detector setting.
• It is critically important to assess “upper limit” of
mitigation actions and not to use these events to justify
detection systems. Detection targets should align to
“lower level” tolerable clouds.
Thank you for your attention!
26
Distributed with permission of author(s) by ISA 2010 Presented at ISA Automation Week 2010; http://www.isa.org