Offshore Hydrocarbon Releases 2001-2008

Health and Safety
Executive
Offshore hydrocarbon releases
2001-2008
Prepared by the Health and Safety Laboratory
for the Health and Safety Executive 2008
RR672
Research Report
Health and Safety
Executive

2001-2008
Alison McGillivray & John Hare
Health and Safety Laboratory
Harpur Hill
Buxton
Derbyshire
SK17 9JN
The offshore industry employs about 28 000 personnel involved in a wide range of activities. Since the Piper Alpha
disaster in 1988, health and safety issues concerning offshore platforms have vastly reduced, however, the work
practices involved are not risk free and still have the potential to cause considerable loss of life when things go wrong.
Increases in oil prices, declining reserves and an ageing infrastructure have resulted in increased drilling activity around
marginal fields. Operators have looked for new ways in which to cut costs, which could affect the health and safety of
the workforce.
HSE’s Major Hazards Strategic Programme Plan outlined targets that hope to reduce the number of major and significant
releases from the 2001/02 baseline of 113 to 67 by the end of 2006/07 (10% year-on-year reduction) and to 60 by the
end of 2007/08 (10% year-on-year reduction). However, in recent years there has been an increase in the number of
major and significant hydrocarbon releases on offshore platforms that require investigation. This work hopes to identify
the immediate cause of hydrocarbon leaks, and determine if there are discernable reasons for the increasing trends.
Two databases currently used by HSE when dealing with offshore releases were utilised, namely the Hydrocarbon Release
(HCR) and RIDDOR databases. Cross-referencing between the two catalogues was expected to yield complete information
including platform location, release size and type, as well as possible failure causes. When brought together over a range of
different releases, this information can generate an overall picture of issues related to increasing release frequency.
Once an analysis had been completed, areas that require improvement, such as structural limitations, system and equipment
faults as well as failings in procedural and operational methods, were indicated where possible.
This report and the work it describes were funded by the Health and Safety Executive (HSE). Its contents, including any
opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy.
HSE Books
© Crown copyright 2008
First published 2008
All rights reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted
in any form or by any means (electronic, mechanical,
photocopying, recording or otherwise) without the prior
written permission of the copyright owner.
Applications for reproduction should be made in writing to:
Licensing Division, Her Majesty’s Stationery Office,
St Clements House, 2-16 Colegate, Norwich NR3 1BQ
or by e-mail to hmsolicensing@cabinet-office.x.gsi.gov.uk
ii
ACKNOWLEDGEMENTS
The authors wish to thank Kevin Quinn and Chris Mawdsley for their help in accessing the
appropriate databases.
iii
iv
CONTENTS
1 INTRODUCTION..........................................................................................1
2 DATABASES...............................................................................................2
2.1 HCR Database .........................................................................................2
2.2 RIDDOR Database...................................................................................2
3 HYDROCARBON RELEASES OVERVIEW................................................3
3.1 Yearly and monthly release trends ...........................................................3
3.2 Release trends in relation to process type................................................5
3.3 Release trends in relation to platform.......................................................6
3.4 Release trends in relation to platform age ................................................7
3.5 Release trends in relation to system pressure..........................................8
3.6 Other information....................................................................................10
4 HYDROCARBON RELEASE ANALYSIS .................................................15
4.1 Partial data .............................................................................................15
4.2 Full data..................................................................................................24
4.3 Overall analysis ......................................................................................26
5 REVIEW OF OTHER RELEVANT STUDIES / GUIDANCE.......................28
5.1 OSD hydrocarbon release reduction campaign......................................28
5.2 Loss of containment manual...................................................................31
5.3 Assessment of the major accident hazard aspects of safety cases........34
6 DISCUSSION.............................................................................................39
6.1 Comparison with Hydrocarbon release reduction campaign ..................39
6.2 Comparison with loss of containment manual ........................................39
6.3 Comparison with GASCET .....................................................................40
7 CONCLUSION AND RECOMMENDATIONS............................................42
7.1 Conclusions............................................................................................42
7.2 Recommendations..................................................................................43
8 APPENDIX A – SEVERITY CLASSIFICATION ........................................44
9 APPENDIX B – RELEASE CAUSES PER YEAR .....................................47
10 APPENDIX C – RELEASES IN TERMS OF AGE OF PLATFORM.......50
10.1 Further investigation – Flanges ..............................................................51
10.2 Further investigation – Piping .................................................................52
10.3 Further investigation – Internal corrosion ...............................................53
10.4 Further Investigation – Normal production..............................................53
10.5 Further investigation – Mechanical fatigue .............................................54
11 APPENDIX D – RELEASE CAUSES PER QUARTER..........................55
11.1 Further investigation – Start up ..............................................................58
v
12 APPENDIX E – RELEASE CAUSES FOR EACH PROCESS TYPE.....61
13 REFERENCES .......................................................................................65
14 BIBLIOGRAPHY ....................................................................................66
15 NOMENCLATURE .................................................................................67
vi
EXECUTIVE SUMMARY
Objectives
To utilise the Hydrocarbon Release (HCR) and RIDDOR databases in order to identify the main
causes of major and significant releases between 2001 and 2007. The analysis should also
highlight possible reasons to explain why the number of these releases has been increasing over
recent years, and follows on from previous work [1].
Main Findings
The main findings, based on the sum of significant and major releases, are:
• The total annual number of releases has decreased over the study period. However, this
decrease is due mainly to the reduction of significant releases, as major releases have
stayed fairly constant in number;
• Typically, the number of releases peaks during November, April and August, and
experiences corresponding troughs during May, June, January and September. The
releases per month may be influenced by plant intervention i.e. high interventions in
summer and low in winter;
• Gas releases are the most common followed by oil, non-process and 2-phase (which are
tied) and lastly condensate;
• Older platforms aged 20 years and over experience the most releases;
• Releases from installations in most age ranges seem to be decreasing over time except
the 10-15 age range which seems to be increasing;
Comments on the full set of major and significant releases were:
• Gas compression is the system that results in most releases, then export, then utilities;
• Piping is the most common equipment type that experiences releases, then instruments,
then flanges;
• The most frequent equipment failure cause is mechanical failure, then none, then
mechanical fatigue. Note that the subgroup ‘failure’ is a default under mechanical
causes of releases and may include some cases of ‘fatigue’ or ‘wear-out’ not adequately
described in the Form OIR12;
• Incorrectly fitted equipment is the most widespread operational cause that is explicitly
stated, then improper operation – human factors issue. ‘None’ appears most frequently;
• Non-compliance with a procedure is the most common procedural cause that is

explicitly stated, then deficient procedure – human factors issue. ‘None’ appears most
frequently;
• Most releases occur in the normal production operational mode, followed by start up
and reinstatement.
vii
Main causes per year:
The number of releases for each category is either roughly consistent, or reducing in number
with time. With all causes, except releases due to system failure, the number of blank inputs
has increased in later years, particularly between 2005 and 2007. This is mainly due to the
manner in which the release reports are recorded, so it may be worthwhile checking why these
areas have been left blank. There seemed to be no major issues relating to the year.
Age of platform:
The oldest platforms at 20+ years have the most instances of releases for export and utilities
systems, perhaps confirming the importance of management of ageing installations. Relatively
new platforms, between 5-10 years of age at the time of release, account for the most gas
compression system releases. In fact, these platforms have high numbers of releases in general
when compared to the surrounding platform ages, so it may be useful to investigate the issues
surrounding these installations.
Considering the equipment, piping is an issue with installations aged 20+, this suggests an issue
with the management of ageing installations. Releases concerning flanges are fairly common
with installations aged 5-10 years. Ring Type Joint (RTJ) flange joints are the most
troublesome, again with 5-10 year old platforms and usually with smaller diameter flanges.
This may be due to the population of smaller sized flanges on installations.
Mechanical failure is the most common equipment cause for releases. There are a significant
number of mechanical fatigue releases, again with platforms aged 5-10 years at the time of
release. Internal corrosion also causes a significant number of releases for platforms aged 20+
years compared with smaller numbers for other ages.
Normal production is the most common operational mode in which releases could occur. The
values over all platform ages are fairly consistent except with the oldest platforms experiencing
the most releases. However platforms aged 5-10 years also show a significant number of
releases during normal production.
Main causes per quarter:
There is no obvious pattern that indicates seasonal variance for the number of releases. The
only area with possible questions is the start up operational mode that has the vast majority of
releases in the third quarter (July to September), compared to lower numbers the rest of the year.
Most of the third quarter start up releases involved gas. It may be worthwhile for operators to
investigate if their procedures change during this time or if start up does not usually occur
during the colder months. The number of releases will be influenced by plant interventions, i.e.
high interventions in the summer and low interventions in the winter.
viii
Recommendations
On the whole, older platforms and processes involving gas result in the most releases. Piping
failures in terms of internal corrosion are also an issue, as is mechanical failure of all equipment
in general. As a result, operators should focus on resolving any outstanding concerns dealing
with these areas; taking account of guidance on ageing platforms (Dalzell, 2007).
It is possible to continue this work in the future in order to monitor offshore safety issues, which
could highlight areas that need further improvement. Further investigation could yield more
patterns. The HSE OSD Incident reports could also be sampled. However, relevant incidents
would first have to be identified by platform name and OIR9B number. The recommendation
on the HSE Intranet version of the HCR database also applies here (see below).
The HSE Intranet version of the HCR database should be modified so that all data filters can be
used to fully identify, by platform name and OIR9B number, relevant releases. At present only
a limited number of data filters can be applied, mostly relating to the type of release, size of
release and date. The current arrangement makes data sampling or full identification of releases
(particular types of system, equipment, operational mode and cause type etc) very difficult.
ix
x
1 INTRODUCTION
The offshore industry employs about 28 000 personnel involved in a wide range of activities.
Since the Piper Alpha disaster in 1988, health and safety issues concerning offshore platforms
have vastly reduced, however, the work practices involved are not risk free and still have the
potential to cause considerable loss of life when things go wrong. Increases in oil prices,
declining reserves and an ageing infrastructure have resulted in increased drilling activity
around marginal fields. Operators have looked for new ways in which to cut costs, which could
affect the health and safety of the workforce [3, 4].
HSE’s Major Hazards Strategic Programme Plan [2] outlined targets that hope to reduce the
number of major and significant releases from the 2001/02 baseline of 113 to 67 by the end of
2006/07 (10% year-on-year reduction) and to 60 by the end of 2007/08 (10% year-on-year
reduction). However, in recent years there has been an increase in the number of major and
significant hydrocarbon releases on offshore platforms that require investigation. This work
hopes to identify the immediate cause of hydrocarbon leaks, and determine if there are
discernable reasons for the increasing trends.
Two databases currently used by HSE when dealing with offshore releases were utilised,
namely the Hydrocarbon Release (HCR) and RIDDOR databases. Cross-referencing between
the two catalogues was expected to yield complete information including platform location,
release size and type, as well as possible failure causes. When brought together over a range of
different releases, this information can generate an overall picture of issues related to increasing
release frequency.
Once an analysis had been completed, areas that require improvement, such as structural
limitations, system and equipment faults as well as failings in procedural and operational
methods, were indicated where possible.
1
2 DATABASES
As this project follows on from previous work [1] and because it is recent releases that require
investigating, only data between 01/01/2001 and 31/12/2007 are used in the analysis. In
addition, the scope of this work deals solely with significant and major releases and definitions
and criteria for each can be found in Appendix A.
The HCR and RIDDOR databases were used to obtain relevant information that fully describes
the steps occurring prior, during and after a release. Each has its merits, as described below.
2.1 HCR DATABASE
The HCR database allows the user to download data relating to all hydrocarbon releases for a
particular year. In this case, data between 01/01/2001 and 31/12/2007 was downloaded into
eight Excel spreadsheets. Note that the 2007/08 data was downloaded on 16/01/2008. As the
data is updated periodically, the data for the fourth quarter of 2007 (October to
December) in this report is incomplete. It was manipulated to remove scenarios
corresponding to minor releases and then extracted to form one complete data set. In total, 35
major and 560 significant release records were found, giving a total of 595. However, this
method has its limitations because the data in this format is completely anonymous, so a
particular release cannot be cross-referenced with the RIDDOR database to obtain further
information.
When the search facility was used, however, a table was produced that listed the installation
name, incident date, location, process type, leak size and severity, which was then used to
identify the corresponding releases in the downloaded data. This was feasible for the 35 major
releases, but it was not possible to match all 560 significant releases. In order to encapsulate the
most severe releases, a cut-off point was taken to extract only significant release records where
over 100 kg of gas was released. This produced 55 records, which was more practical to use in
the initial analysis where installation name matching was required.
It is also useful to note that the HCR database supplies a completed copy of the OIR/12 form,
which is submitted by operators following a release. In essence, the downloaded data
mentioned previously is what is contained in this form but in an Excel format.
2.2 RIDDOR DATABASE
This database was used mainly to corroborate information found in the HCR records through
examination of the OIR/9B form, which is available on the RIDDOR database. This form
details information about the incident and injured parties as well as a description of what
happened, which is most useful when cross-referencing.
The particular releases were identified using information in the HCR database. These incident
reports tend to be historical, describing a sequence of events. They show the release detection
method as being either gas detection or by some manual form of detection (see section 3.6.1 for
the HCR information on this). They describe the various shutdown and mustering events that
occur. Sometimes the source of the release is identified but release causes are not generally
described.
2
3 HYDROCARBON RELEASES OVERVIEW
3.1 YEARLY AND MONTHLY RELEASE TRENDS
Before any analysis can be performed on recent data, it is useful to briefly examine release
trends prior to the date range of interest (2001 to 2007). Figure 1 below illustrates the total
number of minor, significant and major releases that have occurred each year between
1992/1993 and 2006/2007; it also indicates how many releases suffered ignition.
350
300
250
Number of releases
200
No. Releases
No. Ignited
150
100
50
0
92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 06/07
Year
Figure 1 Total number of minor, significant and major releases

between1992/1993 and 2006/2007
Between 1992/1993 and 1994/1995 there is a noticeable peak in hydrocarbon releases, before
suddenly dropping during 1995/1996. However, there is a slight trend increase after this date.
After 2004/2005, there is again a noticeable drop in the number of releases. Ignitions, however,
seem to be fairly consistent from year to year despite the variable number of releases.
As the scope of work does not involve minor releases, Figure 1 is not truly representative of the
variation in significant and major releases. Figure 2 can be obtained from a breakdown of the
data used to obtain Figure 1, and illustrates the number of major and significant releases
between 2001 and 2007, which is within the year range of interest. Figure 2 reports the actual
number of releases for the calendar year 2007, not the incomplete data for this year used
elsewhere in this report. Despite the fact that it appears the overall number of releases is
decreasing, it could be seen that there is not a corresponding decrease in the number of major
releases; it does in fact appear consistent between 2001 and 2007. In this report, calendar years
have mostly been used in tables, graphs and in analysis. The internet and intranet versions of
the HCR database tend to use financial years.
Figure 3 illustrates how the number of significant and major releases varies on a monthly basis
between 2003/2004 and 2007/2008. The earlier years of interest between 2001 and 2003 are not
3
depicted here, as there would be too much information on the graph to assess it accurately. The
graph clearly shows that there are distinct peaks in the number of releases during the months of
November (45), April (43) and particularly August (54). There are troughs visible during May
(31), June (33), January (28) and September (39). It is possibly worth investigating why there
are almost twice as many releases in August than January. The releases per month, shown in
Figure 3, may be influenced by plant intervention i.e. high interventions in the summer and low
interventions in the winter.
Figure 1 is based on all recorded releases, including minor, significant and major, while Figures
2 and 3 are based on the full set of data available for major and significant releases.
120
100
80
Number of Releases
Major and Significant Releases

60 Major Releases
Significant Releases
40
20
2000 2001 2002 2003 2004 2005 2006 2007 2008
Year of Release
Figure 2 Number of major and significant releases per year between

2001 and 2007
4
60
50
40
Number of Releases
03/04
04/05
05/06
30
06/07
07/08
Overall
20
10
0
APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR
Month
Figure 3 Number of major and significant releases per month between

2003/2004 and 2007/2008
3.2 RELEASE TRENDS IN RELATION TO PROCESS TYPE
The releases can also be broken down into process type, as illustrated by Figure 4. Gas releases
(273) are the most common kind of release, followed by oil (43), non-process (22), 2-phase (22)
and then condensate (11). This is useful for operators, as more attention should be paid to the
processes involving gas. It appears that safety precautions must be working, as the number of
gas releases has fallen from 51 to 41 between 2003 and 2007, however, the same cannot be said
for releases involving 2-phase and oil. There is a sharp peak in 2-phase releases during 2005
and 2007 for oil, so it would be worth investigating whether these are anomaly values or if there
has been a breakdown in procedures. On the whole, all processes with the exception of gas
appear to have consistent numbers of releases over the seven years in question. Of the non-
process releases, diesel is the most common form substance to be released.
Figure 4 is based on the full set of data for major and significant releases.
5
70
60
50
Number of Major and Significant Releases
40 Gas
2-Phase
Condensate
Oil
30 Non-Process
20
10
2003 2004 2005 2006 2007
Release Year
Figure 4 The number of major and significant releases corresponding to

Hydrocarbon type between 2003 and 2007
3.3 RELEASE TRENDS IN RELATION TO PLATFORM
Factors including operational procedures vary between different operators to some extent, so
some platforms may suffer more hydrocarbon releases than others. Figure 5 illustrates the
number of releases that occur for each platform between the beginning of 2001 and the end of
2007, using anonymous platform codes, in alphabetical order of anonymous code. It indicates
that, on the whole, most platforms experience one release at most, but others suffer multiple
significant and major releases. Four platforms have more than two major releases, and seven
platforms have more than three major or significant releases. Note that two platforms would
appear in both counts. The KP3 inspection reports (HSE, 2007) were examined for most of
these platforms, which seem to have problems with hydrocarbon releases. Topics with red
traffic lights (denoting that issues were found during inspection) were maintenance of safety
critical elements, backlogs, deferrals, measurement of compliance with performance standards,
review of Independent Competent Person (ICP) recommendations/verification and physical
state of plant.
It is important to note here that the platforms described in Figure 5 and section 3.3 are based on
the partial dataset of 35 major releases (all process types) and 55 significant releases (gas) as
detailed in section 2.
6
6
Significant Release
5
Major Release
4
Number of releases
ZMJH
XDB
Y T
YWTX
ZX
JSC
DN
JY Y
PAM
KH X
SEM
SWW
VJN
W XG
YTP
H
H P
H Z
P Y
R XN
TJN
XG F
YND
AYW
BJ Z
EC L
FP Y
H V
XB
KEC
KEE
LBU
LRA
LX T
M E
M X
N U
N Z
VG J
W U
W C
BDR
C K
ELU
EW H
FKM
R R
BSG
EBL
G A
PHH
SK R
S NW
U S
VGD
V N
W RY
ET X
BNH
BP K
G
E
C
Z
X
Y
U
W
U
X
W
H
A
Platform
Figure 5 The number of major and significant releases corresponding to platform
3.4 RELEASE TRENDS IN RELATION TO PLATFORM AGE
It is also interesting to note how the number of major and significant releases is affected by the
age of the installation as shown in Figure 6. The HCR database allows the platforms to be
categorised as < 5 years, 5-10 years, 10-15 years, 15-20 years and 20+ years, noting that the
platform age corresponds to its age at the release date and is not calculated retrospectively. It is
perhaps reasonable to assume that older platforms contribute the most releases as they suffer the
most wear and tear, however it could also be said that younger platforms should be the most
reliable, resulting in fewer releases. Evidence from 2001 and 2002 shows this to be inaccurate
as younger platforms contribute to a significant portion of releases, possibly due to faults with
new equipment. The number then falls, possibly due to platforms ageing and moving into the 5-
10 year bracket. This could also explain the increasing number of releases from platforms aged
10-15 years. The 15-20 year bracket seems to contain the most consistent platform performance,
with the fewest number of releases overall. It is therefore possibly worth investigating if
platforms built in the 1980s and 1990s (corresponding to an age of 15-20 years) were
constructed using different equipment or procedures to current practices. However caution must
be taken in any analysis, as the number of platforms in each age bracket will vary from year to
year.
Figure 6 is based on the full set of data for major and significant releases.
7
40
35
Number of Major and Significant Releases
30
25
20+ years
15-20 years
20 10-15 years
5-10 years
<5 years
15
10
2001 2002 2003 2004 2005 2006 2007
Year of release
Figure 6 The number of major and significant releases corresponding to

platform age between 2001 and 2007
3.5 RELEASE TRENDS IN RELATION TO SYSTEM PRESSURE
The systems present on board offshore platforms are designed to work within a safe range of
pressures. Risk of releases is increased whenever the actual working pressure at the time of
incident is found to be greater than the maximum allowable working pressure (MAWP) of the
system. Having higher pressures than the equipment is able to deal with could cause ruptures at
any weak points in the system, thereby causing a release.
Figure 7 depicts how the maximum and actual system pressure varies for each platform
experiencing major releases, using anonymous platform codes in alphabetical order. In this case
there are only two instances where the working system pressure was greater than the maximum
allowable working pressure (MAWP), so it appears that this is not a factor leading to an
increase in releases. Figure 8 illustrates the same as Figure 7 but deals with platforms
experiencing significant releases, again using anonymous platform codes. As before, system
pressure does not seem to be an important factor in the increasing trend of hydrocarbon releases,
because there are only three cases recorded. In general as wells become depleted, pressures fall
below the original design pressure. However, higher pressures may be needed for gas injection
and/or gas lift.
The figures below are based on the partial dataset of 35 major releases (all process types) and 55
significant releases (gas).
8
Pressure (Barg) Pressure (Barg)
0
100
200
300
400
500
600
W 700
0
50
100
150
200
250
300
350
400
450
500
BD
AYW Z
BJ
H
releases
BDR
Figure 8
Figure 7
BN Z BP
BSK G
C K D
Y BL
EBN D
EC L BL
ELU EC
ELX U
ET
ET X H
ETH FK
ETH Y
E H FP
EWTH A
G
EWM C
P
HM H
P R
H Z V
R
H V JS
R Y
H V
X KH
H B U
X
JYC LX
KEC X
M
KEE LU
KEX N
G
KH X Z
N
KHU G
Z
9
LBU N
U
LBA M
LRA PA
H
Platform
M T
Platform
E PA
PHE H
PX Y PX
R N N
X R
SEM W
R
SER
SE
SK R R
SWW SE
N R
TJ SN
TJS W
MAW P
TJS U
U S TD
T
VGD VG
J
W
Actual Pressure
VJN
U
MAW P
VX N C
W G W
N
W U ZB
R W
W Y
Actual Pressure
Z ZB
XDB XD
XDF F
XG F YN
YTD P
YT
YT T T
YWX
ZJ
YWX H
System pressure (barg) at each platform experiencing major releases
X ZM
H
System pressure (barg) at each platform experiencing significant

3.6 OTHER INFORMATION
There is other information available from the databases which is not crucially important in
explaining why hydrocarbon releases are on the rise. However, information such as how the
release is detected, if the platform is manned and what zone the release occurs in can be useful
to gain a background understanding of lesser issues.
All the tables and figures in section 3.6 use the partial set of data that includes all major releases
and a selection of significant releases.
3.6.1 Detection type
Poor detection systems may not sense smaller releases, which, when given the chance, could
escalate into much more serious releases resulting in potential loss of life. Table 1 indicates
how often each of the four main automatic detector types, namely heat, smoke, flame and gas,
observed a release. Gas detectors detected 23 out of 36 major releases and 18 out of 58
significant releases, and are therefore the most prolific at observing releases, so it is particularly
important for operators to regularly maintain onboard gas detectors. The figures for successful
release detection relate to larger releases (>100 kg). There may be significant issues for
successful detection of smaller significant gas releases (<100 kg). From Figure 4, gas is the
most common form of process release, so it would be reasonable to assume that gas detectors
should be the main detector type. However, it appears that gas detectors fail to spot a large
proportion of releases, which in turn are detected by other means as shown in Table 2.
Table 1 Number of releases detected by different detection equipment
Detection Type Release Type Number of releases detected

Major 0
Heat
Significant 1
Major 0
Smoke
Significant 1
Major 0
Flame
Significant 1
Major 23
Gas
Significant 18
Major 13
Other
Significant 37
Major 36
Total
Significant 58
It appears that manual forms of detection by visual means are the common form of detection
when the four main automatic types fail, followed closely by sound and smell. Operators
should note that there are reliability issues related to automatic detectors and also issues related
to their design and deployment. It appears a significant proportion of releases may be
discovered by chance, especially if there happens to be personnel in the vicinity of a release. As
a result, it may be prudent to investigate why these detectors fail to respond, and if there are any
ways to improve the systems or even to introduce more effective systematic methods.
10
Table 2 Break down of ‘Other’ detection type from Table 1
Detection Type - Other Release Type Number of releases detected

Major 4
Sound
Significant 8
Major 2
Smell
Significant 8
Major 1
Pressure change
Significant 0
Major 3
Visual
Significant 18
Major 0
ROV
Significant 2
Major 3
Level change
Significant 0
Major 0
Hand held detector
Significant 1
Major 13
Total
Significant 37
These tables are based on the 35 major and 55 significant releases mentioned in previous
sections. However, the totals from the automatic detector table (Table 1) give values of 36
major releases and 58 significant releases. This is because, on some occasions, more than one
detector is activated.
3.6.2 Manned / Unmanned platforms
As illustrated in the previous section, automatic detection sometimes fails and it is by manual
means that the alarm is raised, usually when personnel see a release. This could be problematic,
particularly on platforms that are not manned, as releases could continue for some time before
one of the detectors is eventually activated. By this time, the release could escalate into a more
serious situation which could be prevented, provided personnel are present and act swiftly once
a release has been detected. This also raises issues with automatic system shutdown
mechanisms which may suffer failures, so having personnel present on site allows the system to
be shutdown manually, with reduced risk of shutdown failure. Conversely, unmanned platforms
do not have the associated consequences to personnel whenever a serious accident occurs, so in
this respect there are benefits to having no personnel present or at least reduced numbers. The
main issue is maintenance crews approaching a unmanned platform, where there is already a
leak, without knowing it is there.
Table 3 Number of manned and unmanned platforms for major

and significant releases
Platform type Release type Number of releases

Major 33
Manned
Significant 48
Major 2
Unmanned
Significant 7
11
As shown in Table 3, most platforms are manned at the time of release: this is true for 33 out of
35 major releases and 48 out of 55 significant releases.
3.6.3 Zone classification
Zones are classified in the following categories:
• Zone 1: Area in which an explosive atmosphere is likely to occur in normal operation;
• Zone 2: Area in which an explosive atmosphere is not likely to occur in normal

operation, and if it does should be infrequent;
• Unclassified zone: Area not known to contain any concentrations of flammable

materials in the atmosphere e.g. accommodation.
These zones are used to determine the types of electrical equipment and other potential ignition
sources which are allowed in each zone. Releases in the unclassified zone are also a cause for
concern as accommodation blocks can be located within these zones, so there could be a large
number of personnel affected. [Thyer, 2004] states that 3% of releases in zone 1 ignite,
followed by 6% and 16% for zones 2 and unclassified respectively.
Table 4 is based on the partial data set, and illustrates the number of releases that occur in each
zone for both major and significant releases. Zone 2 appears to experience the most releases as
23 out of 35 major and 42 out of 55 significant releases occur here.
Table 4 Number of major and significant releases per zone
Release type Zone Number of releases

1 9
Major 2 23
Unclassified 3
1 6
Significant 2 42
Unclassified 7
3.6.4 Wind
Weather conditions could possibly affect offshore operations as well as influencing the
behaviour of any releases, for example, high wind speeds have the potential to affect the
structural integrity on the side of the platform that faces the oncoming wind. Tables 9 and 10
illustrate the wind speed that was recorded at the time of release for major and significant
releases respectively. Each zero wind speed corresponds to a value of “not known” in the
records.
12
25
Wind speed (m/s)

Linear (Wind speed (m/s))
20
15
Wind speed (m/s)
10
0
Feb-01 Aug-01 Feb-02 Aug-02 Feb-03 Aug-03 Feb-04 Aug-04 Feb-05 Aug-05 Feb-06 Aug-06 Feb-07 Aug-07
Incident date
Figure 9 Wind speed (m/s) observed at time of release for major releases
30
Wind speed (m/s)

Linear (Wind speed (m/s))
25
20
Wind speed (m/s)
15
10
0
Feb-01 Jan-02 Aug-03 Feb-04 Jan-05 Jul-05 Apr-06 Sep-06 Apr-07 Oct-07
Incident date
Figure 10 Wind speed (m/s) observed at time of release for significant releases
13
It is interesting to note that between 2001 and 2007, the wind speeds recorded appear to be
gradually increasing; this may or may not be due to global warming. This means that it may be
important for operators to assess the structural reliability of platforms, particularly areas feeling
the full force of the wind. This could also raise issues for personnel working near the edges of
platforms, as they could experience strong gusts of winds that could potentially cause fatal falls
from height.
The wind can also disperse releases of gas, possibly reducing the puff to a concentration that
may not be sensed by the detectors, or even into zones other than the initial leak zones. As a
result, it may be important to monitor wind conditions or at least to position the most dangerous
process equipment away from the side of the platform that experiences direct wind.
14
4 HYDROCARBON RELEASE ANALYSIS
Section 3 deals with possible minor causes of releases and issues that could cause a release to
escalate. This section investigates further the main causes of a release including:
• The system involved;
• What equipment is used;
• What caused the equipment to fail;
• The operational cause;
• The procedural cause; and
• The operational mode at the time of release.
This investigation should highlight if there are any mechanical issues involved, perhaps due to
wear and tear or the age of the system, as well as identifying if any procedures are deficient.
The partial data section begins a brief analysis on a smaller data set to identify the most
common issues surrounding a failure. These findings are then applied to the full set of data (all
major and significant releases) to see if they are still appropriate.
4.1 PARTIAL DATA
As mentioned in section 2, the initial approach analyses all major releases and only significant
gas releases where quantities greater than or equal to 100 kg are released. This is due mainly to
the large number of significant release records that were found, which means a detailed analysis
could not be carried out.
As a result, the 35 major and 55 significant releases were used to obtain the following sets of
graphs and tables within section 4.1.
4.1.1 System
Figures 11 and 12 demonstrate what system was in use at the time of the release for major and
significant releases respectively. As illustrated, it appears that most major releases occur from
the export system (eight releases), followed by processing (six releases) and gas compression
(four releases). Gas compression (18 releases) is the most common system operation for
significant releases, followed by export (eight releases) and flowlines (six releases).
It is not surprising that gas compression has been identified as the most common system for
significant releases but not with major releases, mainly because the significant data is based
purely on releases of gas. This means, for example, that export concerning oil would not be
highlighted but it would be for gas, so the numbers for export could be lower than expected
compared to using complete data. Therefore, the values for significant releases may not be
completely representative.
4.1.2 Equipment
The equipment in use at the time of release is illustrated in Figures 13 and 14 for major and
significant releases respectively. As with the system analysis, different equipment issues arise
depending on the severity of the release. For example, there are 12 major releases concerning
piping, but with significant releases the most common is valve-actuated, again with 12 releases.
15
The second most common equipment type for major releases is instruments (four releases)
followed by storage tanks, valve manual and flanges, which are tied with three releases each.
For significant releases, piping and instruments are the cause of nine releases each, followed by
valve manual with six releases. Based on this, operators should take particular care in ensuring
the reliability of equipment types mentioned here.
It is worth investigating further why valve-actuated releases appear so prominently for

significant releases but do not feature at all in the top three major equipment types. As before,
valve-actuated could be more commonplace based on the cut-off point of the significant
releases. Analysis of the full set of data in later sections could highlight the reason for this.
4.1.3 Equipment cause
The equipment failure causes are fairly similar for both major and significant releases except
with erosion, which appears to be more common with major releases than significant, as shown
in Figures 15 and 16 respectively. Mechanical failure is the most common type of malfunction
resulting in nine major and 17 significant releases. Note that ‘failure’ is the default option
between the choices of failure / wear-out / fatigue and could include some failures arising from
some other options not fully established at the time of reporting, such as corrosion. However,
the keyword “none” appears for both types of release severity (nine major and 13 significant)
and means that either equipment failure did not cause the release or, depending on who fills in
the details, that investigators were unsure what equipment failed. Mechanical fatigue is the
third most common equipment failure for both major (three releases) and significant (six
releases) releases. Again, reference should be made to the full data analysis in later sections to
see if the same pattern arises.
It is important for operators to ensure that equipment in the previous section is fully maintained
to prevent the most common types of equipment failure as discussed here.
4.1.4 Operational cause
As with equipment cause, “none” appears as the most common operational cause given for both
major (18) and significant releases (32) and can be seen in Figures 17 and 18 respectively.
Improper operation (five releases) and improper maintenance (four releases) are the next two
main issues with major releases; while left open (10 releases) and incorrectly fitted (six releases)
are associated with significant releases.
From this analysis, it seems that there is no pattern to operational causes as different issues
occur with each of the severity types, therefore, reference should be made to the full data
analysis in section 4.2 to see if this is representative using a wider set of data.
4.1.5 Procedural cause
Procedural causes do not seem to be an overwhelming issue with either major or significant
releases, as “none” appears 23 and 40 times respectively. Deficient procedures result in five
major releases, and four releases occur due to non-compliance with a Permit to Work. Non-
compliance with a procedure accounts for six significant releases and deficient procedures for
five, as illustrated in Figures 19 and 20. From this it appears that in general, procedures are
effective in most cases and that there are other reasons causing releases. The full data analysis
should corroborate or contradict this finding.
16
4.1.6 Operational mode
Figures 21 and 22 confirm that normal production is the most common operation in which
releases occur, and is true for both major (19) and significant (27) releases respectively.
Reinstatement causes four major releases and four significant releases while start up failures
occur with three major and 11 significant releases. It is therefore expected that normal
production will be the most common cause when the full data set is analysed.
The partial set of data containing 35 major and 55 significant releases was used to generate the
following figures: The first graph is for the major releases and the second for the significant
releases. Figures 11 and 12 show what offshore system had the release. Figures 13 and 14
show what type of equipment had the release. Figures 15 and 16 highlight the equipment cause,
Figures 17 and 18 the operational cause and Figure 19 and 20 the procedural cause. Figures 20
and 21 show the operational mode at the time of the release.
Refer to Table 5 for a breakdown of the most common failures for both the partial and full sets
of data.
17
9
DRAINS
EXPORT
FLARE
8
FLOWLINES
GAS COMPRESSION
IMPORT
7 MANIFOLD
PROCESSING
SEPARATION
UTILITIES
6
VENT
WELL
Number of releases
Figure 11 System – major releases
20
DRAINS
EXPORT
18
FLARE
FLOWLINES
GAS COMPRESSION
16 IMPORT
MANIFOLD
PROCESSING
14 SEPARATION
UTILITIES
VENT
WELL
12
Number of Releases
WELL OPS
METERING
10
Figure 12 System – significant releases
18
14
STORAGE TANKS
PIPING
XMAS TREES
12
HEAT EXCHANGERS
VALVE MANUAL
DRAIN OPENING
FLANGES
10 INSTRUMENTS
PIG LAUNCHERS
WELLHEADS
VALVE ACTUATED
Number of releases
8 FILTERS
RISERS
Figure 13 Equipment – major releases
14
STORAGE TANKS
PIPING
XMAS TREES
HEAT EXCHANGERS
12
VALVE MANUAL
DRAIN OPENING
FLANGES
INSTRUMENTS
PIG LAUNCHERS
10 WELLHEADS
VALVE ACTUATED
COMPRESSORS
PIPELINES
PRESSURE VESSEL
Number of releases
8 PUMPS
BLANK
Figure 14 Equipment – significant releases
19
10
NONE
9 AWAITING INVESTIGATION
MECHFAIL
MECHFAT
8 MECHWEAR
EROSION
CORREXT
CORRINT
7
BLANK
SPECIFICATION
Number of Major Releases
MANUFACTURING
6
Equipment Cause
Figure 15 Equipment cause – major release
18
NONE
AWAITING INVESTIGATION
16
MECHFAIL
MECHFAT
MECHWEAR
14 EROSION
CORREXT
CORRINT
BLANK
12
Number of Significant Releases
MATLDEF
10
Equipment Cause
Figure 16 Equipment cause – significant release
20
20
NONE
18 LEFTOPEN
INCORRFIT
IMPROPOP
16 IMPROPMAINT
IMPROPINSP
OPENED
14 BLANK
12
10
Figure 17 Operational cause – major release

35
NONE
LEFTOPEN
INCORRFIT
30
IMPROPOP
IMPROPMAINT
DROPOBJ
SPECIFICATION
25
20
15
10
Figure 18 Operational cause – significant release
21
25
NONE
DEFPROC
NONCOMPROC
BLANK
NONCOMPTW
20 QUALITY CONTROL
15
10
Figure 19 Procedural cause – major release
45
NONE
DEFPROC
40 NONCOMPROC
BLANK
35
30
25
20
15
10
Figure 20 Procedural cause – significant release
22
20
PIGGING
18 REINSTATEMENT
NORMPROD
INSPECTION
16 STARTUP
TESTING
ROUTINEMAINT
14 BLANK
FLUSHING
MAINTCOLDWORK
12
10
Figure 21 Operational mode – major release
30
PIGGING
REINSTATEMENT
NORMPROD
INSPECTION
25
STARTUP
TESTING
ROUTINEMAINT
BLANK
WELLOP
20 BLOWDOWN
WELL SERVICES
REPLACEMENT
SHUTTINGDN
15
10
Figure 22 Operational mode – significant release
23
4.2 FULL DATA
The findings obtained in section 4.1 will now be corroborated by analysis of the full dataset of
35 major and 560 significant releases as shown in Appendices B, D and E. They detail the main
causes per year, the main causes per quarter, and the main causes per process type. Appendix C
looks at releases in terms of the age of platform.
4.2.1 Main causes per year
Appendix B gives information on what releases occur on a yearly basis. The number of releases
for each category is either roughly consistent or reducing in number with time. There seem to
be no anomalous values, except with improper maintenance as an operational cause; there are
10 releases in 2001, which drops to one in 2002 and rises again to 10 in 2003. It may be useful
to determine why these values are so varied. It is also noticeable that with all causes, except
releases due to system failure, the number of blank inputs has increased in later years,
particularly between 2005 and 2007. This is mainly due to the manner in which the release
reports are recorded, so it may be worthwhile checking why these areas have been left blank.
On the whole, because there are no major issues relating to the year, further analysis has not
been carried out.
4.2.2 Main causes per age of platform
The tables in Appendix C detail the typical failures that can occur based on the age of the
platform. For example, Table 7a illustrates the top three failures occurring that concern the
system that is in use; it lists gas compression, export and utilities as the most common. The
oldest platform at 20+ years has the most instances of releases during export and utilities. It is
interesting to note that relatively new platforms, between 5-10 years of age at the time of
release, account for the most gas compression releases. In fact, these platforms have high
numbers of releases in general when compared to the surrounding platform ages, so it may be
useful to investigate the issues surrounding these installations.
Table 7b describes releases associated with a particular set of equipment, and lists piping,
instruments and flanges as the most common equipment failures. Furthermore, it is associated
with Table 7c which further describes issues surrounding flange releases. Table 7b highlights
that releases concerning flanges are fairly common with installations aged 5-10 years at the
point of release. It also illustrates that piping is an issue with installations aged 20+. Table 7c
shows that Ring Type Joint (RTJ) flange joints are the most troublesome, again with 5-10 year
old platforms and usually with smaller diameter flanges. This may be due to the population of
smaller sized flanges on installations.
Table 7d illustrates that mechanical failure is the most common cause for equipment failure,
however, there is an anomalous value of 27 mechanical fatigue releases, again with platforms
aged 5-10 years at the time of release. Internal corrosion causes 33 releases for platforms aged
20+ years compared with smaller numbers for other years. It may be that, if inadequately
managed, corrosion allowances run out on these older platforms.
Tables 7e and 7f describe issues for procedural and operational causes respectively, but because
this is unlikely to be affected by the age of the platform, it is not necessary to analyse them
further.
Normal production is the most common operational mode in which releases could occur, as
shown in Table 7g. The values over all platform ages are fairly consistent except with the
24
oldest platforms experiencing the most releases. Platforms aged 5-10 years also show a
significant number of releases (87) during normal production.
4.2.2.1 Further investigation into platform age
On the whole, the number of releases is consistent for each area of interest, for example, the
incorrectly fitted column in Table 7f has a similar number of releases for each of the platform
ages. However, with Table 7b, it is apparent that the flanges of platforms aged 5-10 years are
particularly problematic, resulting in 23 releases compared to releases of six and eight for other
platform ages. This is also true of piping on platforms aged 20+, as there are 54 releases
recorded compared to numbers such as 13 and 23 obtained for other platforms.
Table 7c illustrates that most flange releases for platforms aged 5-10 years is due to Ring Type
Joint (RTJ) flange joints of lower to middle sized diameters. As a result, flanges have been
investigated further as illustrated in Tables 8a to 8c, which were obtained by filtering the entire
dataset to reveal only flange releases occurring on platforms aged 5-10 years. From this, it is
easy to see that mechanical fatigue is the main cause of equipment failure, normal production is
the most common mode of operation, and incorrectly fitted equipment is the main operational
cause.
Piping is also highlighted as a problem for platforms aged 20 years and over at the time of the
release so further examination can be found in Table 9 in section 10.2. From this table it
appears that just under half of the 54 releases that meet this criteria are due to internal corrosion,
particularly for steel pipes with smaller diameters. Flexible pipes seem to experience less
releases, perhaps because they are easier to replace. Fewer releases may however be due to a
small sample population.
Table 10a illustrates further investigation into platforms aged 20 years and over that suffer from
internal corrosion. Oil appears to be the main substance that causes internal corrosion
(especially in pipes) and accounts for half of the 33 records found. As illustrated in the previous
paragraph, piping is the main equipment type that suffers, resulting in 22 releases out of the 33
records.
Normal production is the most common operational mode at time of failure for platforms aged
5-10 years so this is filtered to see if any patterns arise. No particular equipment item seems to
dominate from Table 11a. However, from Table 11b in section 10.4, it appears that mechanical
issues such as failure and fatigue result in most releases.
Section 10.5 does not highlight any patterns that could account for high numbers of mechanical
fatigue in platforms aged 5-10 years old, so further investigation may be required. However it
may be useful to note that 11 out of 27 results indicate a design failure of some kind, which is
the highest of the whole analysis discussed in this section.
4.2.3 Main causes per quarter
Tables 13a to 13f in Appendix D list the number of the releases that occur for each area of
interest, for example, system or equipment during each quarter of the year. This is mainly to
investigate whether or not any seasonal variations could affect the occurrence of releases,
allowing operators to apply caution during particular months as shown below.
1Q – 1 January to 31 March
2Q – 1 April to 30 June
3Q – 1 July to 30 September
4Q – 1 October to 31 December
25
From this Appendix, based on the top three occurrences for each area of interest, there is no
obvious pattern that indicates seasonal variance for the number of releases. The only area with
possible questions is the start up operational mode that has 27 releases in the third quarter,
compared to lower numbers the rest of the year. The first quarter in particular experiences only
three releases, so it may be worthwhile for operators to investigate if their procedures change
during this time or if start up does not usually occur during the colder months. The number of
releases will be influenced by plant interventions, i.e. high interventions in the summer and low
interventions in the winter. Lower values compared to the rest of the year also occur with
instruments (2Q), internal corrosion (3Q) and incorrectly fitted (1Q). Tables 14a to 14f
investigate start up during the third quarter in more depth to see if any patterns arise. On the
whole, these tables do not identify any obvious causes. However most of the releases seem to
involve gas.
4.2.4 Main causes per process type
Tables 15a to 15f details the number of releases that occur for each process type in terms of
particular areas of interest such as system and equipment. Some areas such as internal
corrosion, piping, mechanical failure and improper operation affect all types of process so it is
appropriate to ensure these particular are either maintained or reviewed regularly. Other areas
such as flushing, compressors and top up are applicable to only one or two process types.
4.3 OVERALL ANALYSIS
Table 5 illustrates the common failures for both partial and full datasets. For example, the
major and significant (partial data) columns represent the partial dataset that consists of 35
major and 55 significant releases. The last column, major + significant, describes the full
dataset that is based on 35 major releases and 560 significant releases. This column is based on
Appendices B to D, which indicate the most common causes.
26
Table 5 Most common failures based partial and full data sets
Release Type
Failure
Major Significant
Major + Significant
(partial data) (partial data)
Piping, instruments, Valve-actuated,

Piping, instruments,
Equipment storage tanks, valve piping, instruments,
flanges
manual, flanges valve manual
Export, processing, Gas compression, Gas compression,

System
gas compression export, flowlines export, utilities
Mechanical failure, Mechanical failure, Mechanical failure,

Equipment cause none, erosion, none, mechanical none, mechanical
mechanical fatigue fatigue fatigue
None, improper None, incorrectly

None, left to open,
Operational cause operation, improper fitted, improper
incorrectly fitted
maintenance operation
None, deficient None, non- None, non-

procedure, non- compliance with a compliance with a
Procedural cause
compliance with a procedure, deficient procedure, deficient
Permit to Work procedure procedure
Normal production, Normal production, Normal production,

Operational mode
reinstatement, start up start up, reinstatement start up, reinstatement
27
5 REVIEW OF OTHER RELEVANT STUDIES / GUIDANCE
5.1 OSD HYDROCARBON RELEASE REDUCTION CAMPAIGN
5.1.1 SPC/TECH/OSD/27
Information on this campaign was obtained from SPC/TECH/OSD/27 [6]. The campaign lasted
for four years. The report notes that doubts have been expressed about the reliability and
completeness in the reporting of minor releases via OIR 12. Information is provided on release
rates per installation per year. The report also lists installations with three or more major or
significant major releases in 2000/1 and 2003/4.
The report includes a section on analytical taxonomy. The taxonomy, which was first used in
OTH 2001 055, is not the same as the OIR12 reporting system but there is often sufficient
harmony for comparisons to be made. Data was available for two years, April 2000 to March
2001, and April 2003 to March 2004. During these periods, HSE OSD inspectors investigated
all offshore hydrocarbon releases. The effect of operating mode was examined with the
majority (around 60%) of all releases occurring in normal production with start-up /
reinstatement the next (13 and 21%) highest. The distribution by hydrocarbon type is also
discussed. The majority of all releases are of gas (55% and 51%), with oil the next most
common type (23% and 29.5%). The remaining releases are diesel (both 6%), condensate (8%
and 9%) and two-phase (3% and 2.6%). The majority of major and significant releases were
found to involve gas or two-phase fluids.
Information is provided on release sites for all releases. Pipework was the largest contributor
(at 56%, of which small bore 20%), then valves (17%) and vessels (18%). Small-bore tubing
was the subject of a UKOOA/IP good practice guide designed to improve standards and
competency. As the percentage of small-bore tube failures rose from 18% in 2000/1 to 21% in
2003/4, more work needed to be done in this area. 70% of the associated releases were from
connections and the rest from the pipe body. 75% of the releases were classified as minor and
25% as significant. Flanges accounted for 12.4% of releases (2003/4); this was a decrease from
15% (in 2000/1). This may indicate that the UKOOA/IP guide on bolted pipe joints was having
some effect. Flexible hoses were subject of another UKOOA/IP guide issued in 2003. Releases
from these were at 4% in 2000/1 and 3% in 2003/4. About 20% of releases were due to body
failures of pipe, vessel or valves. The body failure is mostly caused by mechanical degradation
including corrosion and erosion. 16.7% of 2003/4 releases were from seals and packing (14%
in 2000/1). The percentage of releases associated with the temporary repairs was 3%, which
included one major release. This is a problem area that appeared to be growing in importance,
as there were no corresponding releases from this source reported in 2000/1.
Analysis of all immediate causes of releases was provided for 2000/1, and 2003/4. For 2003/4,
the biggest single immediate cause was: (1) corrosion/erosion at 23%, (2) degradation of
material properties at 16% and incorrect installation at 16%, (3) fatigue/vibration at 12% and
operator error at 12%. In 2000/1, operator error and failure to follow procedures were
significant immediate causes, but were not so in 2003/4. The largest immediate cause was
discussed for various principal release sites:
• Flanges – Incorrect installation (40% and 28%) and degradation of material properties
(20 and 20%);
• Small bore tubing – Fatigue and vibration (35% and 29%) and Incorrect installation
(18% and 15%);
• Open ends – Inadequate procedures (47%) and operator error (25%); and
28
• Valve stem releases – Hardening of the stem packing (70%) i.e. a degradation of
material properties.
The report classifies immediate causes as either hardware-related (corrosion, erosion, fatigue,
vibration, degradation of material properties, inadequate equipment and line blockage) or
software-related (inadequate isolation, inadequate installation, inadequate procedures, operator
error and procedural violation). Nearly 60% of all releases had hardware-related immediate
causes. The underlying causes of these releases were:
• Inadequate inspection / condition monitoring (48%);
• Inadequate design (26%);
• Inadequate maintenance (11%); and
• Inadequate installation (4%).
The remaining 40% of releases would have had a software-related cause. The underlying
causes were:
• Inadequate procedures (26%);
• Inadequate compliance monitoring (10%);
• Procedural violation;
• Risk assessment;
• Inadequate installation; and
• Inadequate task specification (4%).
Underlying causes of all releases is also discussed. The most important were inadequate design
(29% in 2000/1 and 21% in 2003/4) and inadequate inspection/condition monitoring (28% in
2000/1 and both 30% and 35% are quoted for 2003/4). Other important underlying causes were
inadequate procedures, inadequate competency and inadequate maintenance. Inadequate design
is often related to inadequate piping support, which left the pipe vulnerable to vibration and
fatigue problems.
Failed safeguards are also mentioned. If these had operated correctly the release might have
been prevented. Failure to operate an effective corrosion/erosion monitoring system was
implicated in 28% of the 2003/4 releases (10% in 2000/1). This is to be expected with ageing
plants. Also implicated were inadequate inspection / condition monitoring (26% of 2003/4 and
32% of 2000/1 releases), procedural review (12% of 2003/4 and 18% in 2000/1), competency
assurance (11% in 2003/4 and 10% in 2000/1) and design review (8% in 2003/4 and 15% in
2000/1).
5.1.2 HSR 2002 002
This section uses information in HSR 2002 002 [7] to complete the analytical taxonomy for the
year 2001/2 in comparison with 2000/1. The taxonomy was first used in OTO/2001/055 in
which all releases reported in 2000/01 were investigated as part of the Process Integrity
Initiative. There were 241 reportable releases in 2001/02, of which 47 investigation reports
were received for analysis, representing 21% of the total. These have been analysed according
to the same taxonomy. Any conclusions are therefore based on that particular sample.
In terms of release sites, again pipework accounted for the majority of releases at, 62%, which
was comparable with the 2000/1 value of 61%. However, there were some differences within
this grouping between the two years. A greater proportion of the 2001/2 release sites were
assigned to small bore pipework and associated connections, including instruments at 25% as
opposed to 18% in the previous year. Perhaps of more significance was the marked reduction in
releases from pipe open ends, down from 16% to 4%. Valves were involved in 21% of releases,
vessels 6% and pumps 6%. Other findings were:
29
• 70% of releases were cracks, splits or holes in the containment envelope. This was an
increase compared to the 47% of the previous year.
• 23% of releases were from the body of the pipe, vessel or valve. This was in line with
21% for 2000/1. Mostly these resulted from degradation of the containment envelope
caused by degradation of material properties.
• 25% of releases were associated with small-bore piping including instruments, the bulk
of these were associated with connections.
• Flanges accounted for 15% of releases, which was exactly the same proportion as the
previous year.
• 19% of releases were from seals or valve stems. This was an increase over the previous
year’s proportion of 14%.
• There were no hose releases in the sample investigated.
In terms of immediate causes, the largest single cause was again degradation of material
properties, which accounted for 28% of the incidents compared with 26% in 2000/1. Incorrect
installation and fatigue/vibration both at 21%, were the second largest contributors. Both of
these were almost double the 2000/1 figures of 12% and 11% respectively, and more significant
than corrosion/erosion which accounted for 13% of incidents, compared to 19% in the previous
year. Procedural type causes were all less than 10%.
When it came to analysis of the most important release sites, the most prominent cause for
flange leaks (15% of all releases) was incorrect installation in 60% of cases, which was a
significant increase compared to the previous year. Degradation of material properties was
much less significant at 15%. For small bore tubing and associated connections (responsible for
25% of releases), the main causes were fatigue (33%), incorrect installation (25%) and
degradation of material properties (17%). This was in line with the previous year’s findings,
although incorrect installation played a more prominent role. Open ends only accounted for
10% of releases in this analysis. The causes were evenly distributed, although most were
procedural rather than hardware-related.
The same hardware versus software classification was used. The immediate causes could be
divided into hardware or software related as follows:
• Hardware: Degradation of material properties, fatigue/vibration, internal corrosion,
erosion. 62% of releases had hardware-related immediate causes. Of these the
underlying causes were mainly inadequate inspection/condition monitoring in 48% of
these incidents. The next most significant underlying cause was inadequate design
(34%).
• Software: Incorrect installation, operator error, procedural violation, inadequate
isolation, inadequate procedures. The remaining 38% of releases had software-related
immediate causes of these the causes were mainly inadequate procedures (40% of these
incidents). The next most frequent cause was inadequate compliance (28%).
The underlying causes of many incidents are complex and more than one cause can be identified
in many cases. When these were analysed independently of immediate cause, the largest
contribution was from inadequate inspection/condition monitoring in 32% of investigated
incidents, closely followed by inadequate design in 30%. These were very similar to the
previous year’s figures of 28% and 29%. The next four most significant underlying causes were
inadequate procedures in 23% of incidents, incorrect installation in 15%, incorrect material
specification/usage in 13% and inadequate risk assessment in 13%. Other underlying causes
were identified in less than 10% of investigated incidents.
As in the previous year, inspection/condition monitoring was the most prevalent failed
safeguarding system that might have prevented the release of 30% of all incidents investigated.
30
Of the remaining safeguarding systems, competency assurance was the next highest with
corrosion/erosion monitoring and change control also featuring. Design review was less
significant than the previous year’s analysis. There was no identifiable pattern for major
releases.
5.2 LOSS OF CONTAINMENT MANUAL
One of the purposes of this document [8] is to reduce offshore hydrocarbon releases. It was
produced during the Key Programme, discussed in section 5.1, on the same issue. It discusses
10 important elements on the management of the containment of the process containment
envelope and provides advice for inspection within each of the elements. The maintenance of
high standards of operation in these areas should help to reduce the frequency of offshore
hydrocarbon leaks. Relevant information under each element will now be outlined below.
5.2.1 Management of Process Integrity
One important aspect of successful process integrity management is the identification of the
basic and underlying causes behind process incidents.
5.2.2 Small bore tubing and piping systems
Small bore tubes, piping systems and flexible hoses have to be successfully managed if they are
used for hydrocarbon duties. The IP/UKOOA guidance on this subject should be referred to.
Vibration induced fatigue was not generally considered in the design of process piping systems.
Vibration induced fatigue can be made worse by both increased flowrates / relaxation of erosion
velocity limits, resulting in more turbulent energy. It also results from the greater use of thin-
walled pipework, made possible by the use of higher strength materials, allowing use of more
flexible pipework and higher stress concentration at small bore connections. Use of small bore
tubing in practice and the control of small bore flexible hoses are also discussed.
5.2.3 Information, instructions and training
The safe operating limits of process plant equipment should be set and clearly documented.
Limits should be set for pressure, temperature, level and flow. Process composition and
material additions may also have to be limited. Related trip and alarm settings should also be
recorded together with a clear record of plant equipment.
5.2.4 Isolations and permits to work
Included here are isolation standards, locked valve controls, long-term isolations, isolations for
relief and vent systems, permit systems and monitoring. The OIAC / HSE guidance documents
‘The safe isolation of plant and equipment’ (HSG 253) and ‘Guidance on permit to work
systems’ (HSG 250) should be referred to.
5.2.5 Process plant protection systems
The systems prevent process excursions from violating the safe operating envelope of the
process equipment. Both instrumented protection and relief & blow down systems are involved.
Issues include instrumented protection and ESD systems, control of inhibits/overrides, control
of programmable systems, alarm systems integrity, and relief/blow down system integrity.
31
5.2.6 Change control
Many of the catastrophic events that have occurred on process facilities are attributable to
changes. There have also been numerous deficiencies in offshore process systems arising from
the failure to control change.
A typical change control problem concerns a failure to re-evaluate relief requirements

adequately when process fluids or operating conditions are changed, or when mechanical
changes are made. Another example involves change from dry gas to wet gas operations. Key
safety issues include different corrosion/erosion rates, liquid slugging effects, increased pigging
frequency, hydrate formation/inhibition, and effects on blow down, flare and vent systems.
Duty holders need to have systems to ensure that changes to the process and its equipment, or to
the management system, are properly evaluated before their introduction. This part of the
manual reviews the way in which change is initiated, communicated, analysed, implemented,
and reviewed. It involves a mix of onshore and offshore inspection; inspectors will need to
decide on the best place to obtain the relevant information.
It is intended that the guidance should be used flexibly. If it is sufficient to open a meaningful
dialogue then the inspection topics/action will have served its purpose.
5.2.7 Maintenance and verification of process safety-critical elements
This part of the manual examines the way in which process safety-critical elements are
managed. OSD’s analysis of incidents indicates that particular attention needs to be given to
systems for managing corrosion and erosion, and leaks at flanges. It involves a mix of onshore
and offshore inspection. UKOOA Guidance sets out various recommendations for industry to
follow e.g.. CP029 Management of Safety-Critical Elements (1996), and CP001 Fire and
Explosion Hazard Management (1995).
5.2.8 Control of miscellaneous process hazards
This part reviews the way in which some miscellaneous process hazards are managed. Not all
the hazards are relevant to every installation, and the notes concentrate on preventive controls
rather than loss of containment etc.
Control of H2S and CO2
Gas streams associated with some reservoirs have to be treated to reduce the Carbon Dioxide
(CO2) or Hydrogen Sulphide (H2S) to levels to meet export pipeline gas specifications. Two
main methods are used. These are:
a) counter current contact in which the gas stream rises through a column against a downward
flowing stream of amine or proprietary solvent, and
b) b) absorption, where the gas stream may be passed through a column packed with zinc
oxide in powder or granular form.
Sand management
Production of sand with well fluids presents several potential hazards for topsides pipework and
equipment, as detailed in SPC/Tech/OSD/19 - Offshore Produced Sand Management. For
example:
• Sand plugging causes valves to seize, potentially compromising ESD action;

32
• Accumulated sand requires operator intervention to remove it, by sand-washing,
digging out, or dismantling of plate exchangers;
• Sand accumulation can prevent corrosion inhibitor reaching the material surface
leading to increased corrosion; and
• Sand accumulation in level instruments and bridles can lead to false readings and poor
control, and may compromise shutdown initiation.
Control of hydrates
Hydrates are ice-like solids that can form when wet gas and light condensate at high pressures
cools to lower temperatures. Hydrates are formed by gas cooling to below its water dew point,
or when free water is present. Cooling may be due to operational pressure drop, or during start-
up when hydrocarbon is introduced into cold pipework or equipment. Once formed, hydrates
are difficult to remove; prevention is better than cure. Hydrate inhibitor, such as glycol,
methanol or industrial methylated spirits, is used to inhibit the formation of hydrates by removal
of free water. Hazards caused by hydrates include:
• Blockage of pipework, and instrument tappings, causing false readings;
• Plugging of valves, giving operational problems, and potentially compromising ESD

action;
• Hydrate particles travelling at high gas velocities can cause large forces at elbows and
tees; and
• Removal of hydrates may require physical intervention, with associated risks. Sand
particles can erode piping and fittings, particularly chokes and flowlines.
Sampling arrangements
Sampling involves directly breaking into the hydrocarbon containment envelope. Hazards
associated with potential loss of containment, and with static electricity, should be recognised.
Protection against air ingress and flammable mixtures in process plant
Flammable mixtures can form in piping, plant and equipment when air enters systems that
normally contain hydrocarbon, as a result of operational or maintenance activities. Correct
purging and operational procedures will ensure that the risks are minimised.
Segregation of hazardous drains
Open drain systems are typically classified as hazardous and non-hazardous. It is important that
segregation of the drain systems is maintained at all times to prevent migration of hydrocarbons
into safe areas where they may present an ignition risk.
5.2.9 FPSO specific systems
In this part of the manual is discussed:

• The effects of motion on process plant systems;
• Turret arrangements/swivel joints and seals, leakage and recovery;
• Integrity of flexible hoses;
• Inert gas controls/cargo tank blanketing;
• Marine and process systems interfaces; and
33
• HP fuel gas and gas compression.
5.2.10 Process plant construction and commissioning
In construction, important issues include piping, flange joints and small bore pipework/tubing
and the control of change. In commissioning, important issues include cleaning/flushing/drying,
pressure testing/leak testing and function testing.
5.3 ASSESSMENT OF THE MAJOR ACCIDENT HAZARD ASPECTS OF

SAFETY CASES
The topic guidance [9], commonly known as GASCET, discusses a number of situations where
hydrocarbon releases have occurred. These releases are generally discussed under specific
technical issues.
5.3.1 Degradation in Service
This is an issue for pressure vessels, piping and piping components and valves.
To avoid corrosion, piping containing hydrocarbons should avoid 'dead legs' and be designed to
facilitate drainage to prevent trapping of fluid.
Fatigue is a damage mechanism by which cracks can propagate in a structure under the
influence of repeated cycles of stress well below the level capable of causing general yielding.
Fatigue is often characterised as occurring in two phases; the first is that of initiation, i.e. from
manufacture up to the point where a detectable crack is present. The second is the phase of
defect growth, where propagation from the point of detectability to the point of failure occurs.
Fatigue is addressed initially at the design stage. There are a number of methodologies by
which this can be done. However we note that for plant with a limited fatigue load, the codes
normally provide for the exclusion of a full analysis, providing that certain pre-conditions can
be met, i.e. it is established that there will only be a limited number of full pressure cycles etc.
In general though, the fatigue loads from all sources of repetitive stress have to be characterised
both in terms of the stress amplitude and their number. This can be used to determine a fatigue
lifetime for the component.
A suitable demonstration should be provided for the integrity of joints and seals where failure
could lead to a release of hydrocarbons. General information should be provided to indicate
that flanges and other joints have been adequately designed and properly made to avoid
flammable and toxic hazards. Further guidance is available in IP/UKOOA Guidelines for the
Management of Integrity of Bolted Pipe Joints.
The use of fully welded pipework topsides is one of the approaches to adhere to the principle of
inherently safer design. However, for ease of access for operation, inspection, maintenance and
repairs, it is not possible to have fully welded pipework everywhere on topsides plant. The duty
holder should avoid routing of pipework containing hazardous fluid through non-hazardous
areas. If this is unavoidable, then pipework shall be all welded (no flanges) and not located in a
vulnerable position where it may be mechanically damaged.
34
5.3.2 Materials
Materials chosen should be suitable for the application in terms of the process fluid,
environment and applied loading. Again, pressure vessels, piping and piping components, and
valves are relevant here.
The prevention of brittle fracture is addressed within design codes. Prevention involves the
correct choice of materials, operation within strict temperature/pressure limits and monitoring
ageing phenomena such as embrittlement. Ferritic steels are subject to a ductile to brittle
transition as temperature decreases, rendering them highly vulnerable to brittle fracture when
cold. Transition temperatures vary, but are typically below ambient values for offshore
applications. Ageing though can lead to a shift in the transition temperature and render
components more susceptible to brittle fracture. Austenitic steels remain ductile at low
temperatures and may be preferred for application such as blowdown lines.
Brittle fracture is possible whenever low temperatures are involved, in particular low
temperatures associated with gas expansion. This is particularly the case when systems are still
pressurised, although in some circumstances, the differential stresses through the wall of a
vessel by sudden cooling could lead to crack propagation.
The effect of ageing is undoubtedly one of the major integrity issues facing the older
installations. Ageing encompasses degradation mechanisms such as fatigue and corrosion.
There are also other phenomena, for example creep and the deterioration in mechanical
properties such as fracture toughness. The latter phenomenon is associated with changes in
transition temperatures. Provision against these mechanisms is explicitly required, as part of the
design criteria and operational monitoring exists for the express purpose of detecting these
phenomena.
Nevertheless, age-related failures are occurring. The implication of this is that either plant is
being operated beyond its original design life, that conditions have changed because
modification has rendered the initial assumptions invalid, or that inspection regimes are
inadequate.
In recent years, the popularity of risk-based inspection schemes has led to situations where
inspection intervals have been lengthened for some plant. Where such decisions have been
made, the requirements on the knowledge about plant state are high.
5.3.3 Gasketed Plate Heat Exchangers
There is a likelihood of significant hydrocarbon release to the atmosphere on gasket failure.

Shields should normally be fitted to prevent fluids from contacting personnel in the event of
gasket failure. There is a working pressure limitation for gasketed plate heat exchanger of
approx 25 barg.
5.3.4 Hazardous Drains/Caissons
Dip pipes can be subjected to accelerated rates of corrosion at, or just below, the liquid level in
the caisson. Perforation resulting from such corrosion may result in the migration of
hydrocarbon vapour from the caisson into the drains system (this has resulted in a number of
hydrocarbon releases). Confirmation should be obtained that there is an inspection scheme in
place to address this phenomenon.
A number of hydrocarbon releases have resulted from poor design involving inappropriate
interconnections between the closed/flare system and the open drains. Plant blowdown then
35
causes gas to discharge from the open drains. Confirmation should be sought that this
possibility has been examined during the plant HAZOP studies.
5.3.5 External Corrosion
External corrosion of topsides on an ageing installation does not usually receive the same
degree of attention as the management of the internal corrosion. The result of this is that on a
number of installations, the primary threat of hydrocarbon release is from external corrosion. In
addition, a significant number of personnel injuries on such installations are due to falls and
trips resulting from failure of corroded members used as temporary supports or steps. Corroded
walkways have also featured in a number of incidents. Particular issues that should be probed
as part of the safety case assessment include:
• Management of process plant integrity around corrosion traps such as pipe supports,
penetrations, saddles, etc;
• Management of the risks associated with surface preparation and painting on ‘live’ plant;
• Management of corrosion under insulation;
• Management of bolt corrosion;
• Management of pitting and stress corrosion cracking in corrosion resistant alloy piping and
tubing operating in areas exposed to sea spray/deluge. See RR129 Review of external Stress
Corrosion Cracking of 22% Cr Duplex Stainless Steel for further guidance;
• Painting and refurbishment planning systems and performance standards including short-term
remedies;
• Maintenance of spring supports;
• Corrosion management of walkways, hand railings, escape equipment attachment points and
other similar secondary structural components.
5.3.6 Erosion
There have been a number of major hydrocarbon releases recently caused by solids particle
erosion where failure of a number of crucial control measures had occurred. Wall thinning is
usually very rapid and hence prevention rather then control should be the guiding principle.
Operations staff do not always appreciate the impact of the production rate on erosion risk.
Prevention of erosion in the production plant can be achieved by design, whereas for well
servicing and drilling operations process management is usually the only available option.
Erosion tends to be a localised effect, which means that a very good knowledge of the local
rather then global flow velocities is required in order to assess erosion risks. Sand detection
systems have proved to have varying reliability and hence their effectiveness should be explored
as part of the assessment process.
Relevant guidance documents include
RR115 Erosion in Elbows in Hydrocarbon Production systems: Review Document and
SPC/TECH/OSD/19 Offshore Produced Sand Management.
36
5.3.7 Incorrect Material Specification
The guidance notes that confirmation should be obtained that material selection has been based
on a rigorous evaluation of all internal and external environments, operational and non-
operational conditions including upset conditions, design life and system performance
standards, failure modes and consequences, inspection and monitoring requirements and health
conditions. In addition, various standards are listed in the guidance. Most process plant
component standards also cover material performance requirements to some extent.
Although the standards provide a good basis for evaluation of materials selection they are not
all-encompassing and hence the following should also be examined:
• How has the industry experience been captured and fed into the materials selection process?
• What is being done to design out corrosion under lagging?
• Have the various problems with vessel internal coatings experienced by a number of duty
holders been recognised?
• Are risks of preferential weld corrosion adequately addressed?
• How are the significant erosion risks in vessel sand wash drains tackled?
• Are ESD and Control valve trims adequate to maintain seal tightness under the operating
environment?
• Are the limitations and problems in using corrosion allowance approach to manage
degradation recognised?
• Do provisions for testing include the need to demonstrate adequacy of the material’s corrosion
resistance as well as physical properties?
• Have the limitations of 316SS tubing been considered in the material selection process?
• Are the particular requirements for bolting material and its corrosion protection adequately
addressed?
5.3.8 Isolations
Confirmation should be obtained that isolation procedures are in accordance with recognised
standards or codes of practice including Oil Industry Advisory Committee – The safe isolation
of plant and equipment (OIAC / HSE HSG 253). Where a standard or code of practice other
than that listed above has been employed, judgement as to the adequacy can only be made on an
individual basis and the dutyholder should be requested to justify why equivalent standards of
safety should result.
5.3.9 Permit to Work Systems
Although a sub-set of the procedural assessment element within Human Factors, this is so major
it needs to be a topic in its own right, particularly given that failure of permit to work [PTW]
procedure has been a key causal factor in the major offshore accidents and incidents.
Confirmation should be obtained that clearly defined permit to work procedures are in place for
all activities on the installation, and that these comply with a recognised standard or code of
practice. Recognised standards/codes of practice would include:
37
OIAC Permit to work guidance (OIAC/HSE HSG250);
HSG 65 Successful Health and Safety Management;
HSG 48 Reducing Error and Influencing Behaviour.
Where a standard/code of practice is listed above, but an alternative has been employed,
judgement as to the adequacy of the system in place can only be assessed on an individual basis,
and the duty holder should be required to justify why its system will deliver an equivalent level
of health and safety performance.
38
6 DISCUSSION
6.1 COMPARISON WITH HYDROCARBON RELEASE REDUCTION

CAMPAIGN
The campaign report [6] noted problems in the reporting of minor releases. These problems are
not relevant to this project, which was only concerned with major and significant releases.
The trend of releases being concentrated in normal production and start up / reinstatement
operational modes was also found to be true in the current study, although based on differently
specified data sets. When considering the number of releases in various operational modes the
frequency of that mode will affect the frequency of hydrocarbon releases. It may be that the
frequency of releases in start-up and reinstatement is higher than normal production expressed
as the number of releases per hour.
The current study highlighted the number of releases from piping, instruments, storage tanks,
valves and flanges. The campaign report also highlighted these release sites.
The current study showed the importance of releases from piping and flanges, that platform age
was generally significant and that more releases were associated with smaller diameter flanges.
This may be due to population, i.e. are there more small/medium size flanges than large ones?
There may be a higher number of failures involving smaller sizes, but this is not necessarily the
same as saying small flanges have more leaks than large ones. The campaign report also found
that flange failures were sometimes related to degradation of mechanical properties.
The current study found that mechanical failure, erosion and mechanical fatigue were the most
frequent equipment causes. The campaign report highlighted erosion / corrosion, degradation of
mechanical properties and fatigue/vibration as immediate causes. Thus similar causes are being
highlighted.
The current study found that improper operation, improper maintenance, left open and
incorrectly fitted were the most important types of operational causes. The current study found
that deficient procedures, non-compliance with PTW and non-compliance with procedures were
the most important procedural causes. The campaign report highlighted incorrect installation
and operator error as immediate causes. Thus similar causes are being highlighted.
6.2 COMPARISON WITH LOSS OF CONTAINMENT MANUAL
Some of the same issues discussed in this manual [8] could also be identified in the current
project; see the appendices on release causes per year and releases in terms of platform age.
The manual [8] stresses the importance of maintaining process integrity and looking at the
underlying causes behind process incidents. In terms of pressure limits, using the partial
dataset, there were only two cases where the actual pressure exceeded the maximum pressure.
The OIR12 reporting system ensures that the duty holder makes some attempt to look at
underlying causes for hydrocarbon releases.
The manual [8] highlights the importance of guidance to ensure that isolations and PTW are
carried out properly. The current study found that improper operation, improper maintenance,
left open and incorrectly fitted were the most important types of operational cause. Similarly
deficient procedures, non-compliance with PTW and non-compliance with procedures were the
most important procedural causes. Clearly the two OIAC/HSE guidance documents on isolation
and PTW need to be more closely followed.
39
The manual highlights change control. The current study also gives some information on this
topic. As well as normal production, re-instatement and start up are operating modes during
which hydrocarbon releases are occurring. Mechanical fatigue was significant for platforms
between 5 and 10 years old. Mechanical changes may be contributing to this increased
mechanical fatigue. Internal corrosion was significant for platforms over 20 years old. It may
be that the change from dry gas to wet gas was not anticipated in the original design, thus
change control needs more careful management.
The loss of containment manual discusses the control of miscellaneous process fluids. The
current study found that in terms of equipment causes the order (most common first) is
mechanical failure, mechanical fatigue, internal corrosion, erosion, and external corrosion.
Thus internal corrosion seems to be more of an issue than erosion and external corrosion.
However the manual [8] shows that internal and external corrosion and erosion are all related.
The presence of sour gases will increase the rates of corrosion. Sand seems to prevent corrosion
inhibitors from working properly. Hydrate particles cause releases at Ts and elbows. Sand can
also erode pipes and fittings, particularly chokes and flow lines.
6.3 COMPARISON WITH GASCET
Some of the same issues highlighted in Gascet [9] could also be identified in the current project:
see the appendices on release causes per year and releases in terms of platform age. Piping and
flanges were pieces of equipment involved in a considerable number of hydrocarbon releases.
Most piping releases involved platforms over 20 years old. Most flange releases involved
platforms aged between 5 and 10 years.
Internal corrosion was more of an equipment cause for platforms older than 20 years. This may
be because the expected corrosion allowance had been lost over time or because the fluid had
changed its characteristics to become more corrosive. Oil seems to be the main substance that
causes internal pipe corrosion. Gascet states that ‘dead legs’ should be avoided and piping be
designed to prevent trapping of fluid. The current study finding on oil seems to link in with
this. Some wells are becoming sourer as they become depleted. Piping is also the item of
equipment likely to be affected.
Mechanical fatigue was more of an equipment cause for platforms between 5 and 10 years.
Presumably this would be the defect growth phase mentioned in the sub-section on degradation.
No particular patterns were observed which may cause the fatigue, but a significant proportion
involved a design failure of some kind. This also seems to link in with the Gascet comments on
mechanical fatigue on design.
Flange problems were particularly common in platforms in the 5-10 year age range. Ring Type
Joint (RTJ) flange joints had the most problems and the lower to middle sized diameters seemed
to have more problems. Mechanical fatigue and mechanical failure seem to the main equipment
causes for these flanges. Releases from flanges could be due to poor bolt tightening or
unsuitable gasket materials. The UKOOA guidance on bolted joints mentioned in Gascet
should perhaps be consulted here.
Piping problems were particularly common in platforms over 20 years old. Internal corrosion
was the main equipment cause with the mechanical causes (fatigue failure and wear) also
contributing. Smaller diameter steel pipes seemed particularly prone. There seemed to be less
releases for flexible pipes. Guidance on ageing platforms (Dalzell, 2007) is pertinent to the
prevention of failures on older platforms.
Gascet discusses fracture, but this does not appear as an equipment cause on the HCR database
but may perhaps be represented as mechanical failure. Particular types of fracture such as brittle
40
fracture, also discussed in Gascet, are not indicated as such. If brittle fractures are associated
with low temperatures, then it may be possible to relate mechanical failure and equipment that
operates at low temperatures.
In terms of equipment causes, the order (most common first) is mechanical failure, mechanical
fatigue, internal corrosion, erosion, and external corrosion. Thus internal corrosion seems to be
more of an issue than erosion and external corrosion, which are discussed in some detail in
Gascet. Releases from heat exchangers (equipment) and drains (systems and equipment) do
occur but are not particularly significant. Liquid levels in caissons are discussed in Gascet. The
current study’s observations on the relationship between internal corrosion and oil seems
relevant here. Material specification as an equipment cause was not particularly significant.
Normal production is the most common operational mode at failure for platforms between 5 and
10 years old. No particular equipment type seems to dominate. Mechanical issues such as
failure and fatigue result in most releases.
41
7 CONCLUSION AND RECOMMENDATIONS
7.1 CONCLUSIONS
The main findings, based on the sum of significant and major releases, are:
• The total annual number of releases has decreased over the study period. However, this
decrease is due mainly to the reduction of significant releases, as major releases have
stayed fairly constant in number;
• Typically, the number of releases peaks during November, April and August, and
experiences corresponding troughs during May, June, January and September. The
releases per month may be influenced by plant intervention i.e. high interventions in
summer and low in winter;
• Gas releases are the most common followed by oil, non-process and 2-phase (which are
tied) and lastly condensate;
• Older platforms aged 20 years and over experience the most releases;
• Releases from installations in most age ranges seem to be decreasing over time except
the 10-15 age range which seems to be increasing;
Comments on the full set of major and significant releases were:
• Gas compression is the system that results in most releases, then export, then utilities;
• Piping is the most common equipment type that experiences releases, then instruments,
then flanges;
• The most frequent equipment failure cause is mechanical failure, then none then
mechanical fatigue. Note that the subgroup ‘failure’ is a default under mechanical
causes of releases and may include some cases of ‘fatigue’ or ‘wear-out’ not adequately
described in the Form OIR12;
• Incorrectly fitted equipment is the most widespread operational cause that is explicitly
stated, then improper operation – human factors issue. ‘None’ appears most frequently;
• Non-compliance with a procedure is the most common procedural cause that is

explicitly stated, then deficient procedure – human factors issue. ‘None’ appears most
frequently;
• Most releases occur in the normal production operational mode, followed by start up
and reinstatement
Main causes per year
The number of releases for each category is either roughly consistent, or reducing in number
with time. With all causes, except releases due to system failure, the number of blank inputs
has increased in later years, particularly between 2005 and 2007. This is mainly due to the
manner in which the release reports are recorded, so it may be worthwhile checking why these
areas have been left blank. There seemed to be no major issues relating to the year.
42
Age of platform
The oldest platforms at 20+ years have the most instances of releases for export and utilities
systems, perhaps confirming the importance of management of ageing installations. Relatively
new platforms, between 5-10 years of age at the time of release, account for the most gas
compression system releases. In fact, these platforms have high numbers of releases in general
when compared to the surrounding platform ages, so it may be useful to investigate the issues
surrounding these installations.
Considering the equipment, piping is an issue with installations aged 20+, this suggests an issue
with the management of ageing installations. Releases concerning flanges are fairly common
with installations aged 5-10 years. Ring Type Joint (RTJ) flange joints are the most
troublesome, again with 5-10 year old platforms and usually with smaller diameter flanges.
This may be due to the population of smaller sized flanges on installations.
Mechanical failure is the most common equipment cause for releases. There are a significant
number of mechanical fatigue releases, again with platforms aged 5-10 years at the time of
release. Internal corrosion also causes a significant number of releases for platforms aged 20+
years compared with smaller numbers for other ages.
Normal production is the most common operational mode in which releases could occur. The
values over all platform ages are fairly consistent except with the oldest platforms experiencing
the most releases. However platforms aged 5-10 years also show a significant number of
releases during normal production.
Main causes per quarter
There is no obvious pattern that indicates seasonal variance for the number of releases. The
only area with possible questions is the start up operational mode that has the vast majority of
releases in the third quarter (July to September), compared to lower numbers the rest of the year.
Most of the third quarter start up releases involved gas. It may be worthwhile for operators to
investigate if their procedures change during this time or if start up does not usually occur
during the colder months. The number of releases will be influenced by plant interventions, i.e.
high interventions in the summer and low interventions in the winter.
7.2 RECOMMENDATIONS
On the whole, older platforms and processes involving gas result in the most releases. Piping
failures in terms of internal corrosion are also an issue, as is mechanical failure of all equipment
in general. As a result, operators should focus on resolving any outstanding concerns dealing
with these areas; taking account of guidance on ageing platforms (Dalzell, 2007).
It is possible to continue this work in the future in order to monitor offshore safety issues, which
could highlight areas that need further improvement. Further investigation could yield more
patterns. The HSE OSD Incident reports could also be sampled. However, relevant incidents
would first have to be identified by platform name and OIR9B number. The recommendation
on the HSE Intranet version of the HCR database also applies here (see below).
The HSE Intranet version of the HCR database should be modified so that all data filters can be
used to fully identify, by platform name and OIR9B number, relevant releases. At present only
a limited number of data filters can be applied, mostly relating to the type of release, size of
release and date. The current arrangement makes data sampling or full identification of releases
involving, for example, particular types of system, equipment, operational mode and cause type,
very difficult.
43
8 APPENDIX A – SEVERITY CLASSIFICATION
Within the scope of this work, only significant and major releases are considered, but for
completeness the definition and criteria for minor releases is included here.
Definitions
MAJOR: "Potential to quickly impact out with the local area e.g. affect the TR, escape routes,
escalate to other areas of the installation, causing serious injury or fatalities." A major leak, if
ignited, would be likely to cause a "major accident", i.e. it would be of a size capable of causing
multiple casualties or rapid escalation affecting TR, escape routes, etc.
SIGNIFICANT: "Potential to cause serious injury or fatality to personnel within the local area
and to escalate within that local area e.g. by causing structural damage, secondary leaks or
damage to safety systems." A significant leak, if ignited, might have the potential to cause an
event severe enough to be viewed as a "major accident" or be of a size leading to significant
escalation within the immediate area or module.
MINOR: "Potential to cause serious injury to personnel in the immediate vicinity, but no
potential to escalate or cause multiple fatalities." A minor leak, even if ignited, would not be
expected to result in a multiple fatality event or significant escalation, but could cause serious
injuries or a fatality local to the leak site or within that module only.
Criteria
MAJOR:
(i) Gas Releases:
EITHER [Quantity released > 300 kg]
OR [Mass release rate>1kg/s AND Duration >5 mins]
This could result in a jet fire of over 10 m length (>1kg/s) capable of causing significant
escalation after 5 minutes duration, or a flash fire/explosion on reaching LFL. Where 300 kg
equates to approx. 3000 m3 explosive cloud at NTP, enough to fill an entire module or deck
area, and to cause serious escalation if ignited.
(ii) Liquid Releases (Oil/Condensate/Non-process):

EITHER [Quantity released > 9,000 kg]
OR [Mass release rate>10kg/s AND Duration >15 mins]
This could result in a pool fire over 10 m in diameter (>10kg/s) filling a module or cutting off a
deck, hindering escape and affecting more than one person directly if lasting for over 15
minutes duration.
(iii) 2-Phase Releases:

EITHER [Quantity of liquids released > 300 kg]
OR [Liquids mass release rate>1kg/s AND Duration >5 mins]
Combinations of the major gas and liquids scenarios described above are possible, depending on
the gas to oil ratio (GOR) involved.
MINOR:
(i) Gas Releases:
EITHER [Quantity released < 1 kg]
OR [Mass release rate <0.1 kg/s AND Duration < 2 mins]
44
This could result in a jet fire of less than 5 m length (< 0.1 kg/s) which is unstable (< 2 mins
duration) and therefore unlikely to cause significant escalation, or a flash fire/explosion on
reaching LFL. Where <1 kg equates to <10 m3 explosive cloud at NTP, probably insufficient to
cause a significant hazard if ignited.
(ii) Liquid Releases (Oil/Condensate/Non-process):

EITHER [Quantity released < 60 kg]
OR [Mass release rate <0.2 kg/s AND Duration < 5 mins]
This could result in a pool fire smaller than 2 m in diameter (< 0.2 kg/s) unlikely to last long
enough to hinder escape (< 5 mins), but could cause serious injury to persons nearby.

EITHER [Quantity released < 1kg]
OR [Liquids release rate <0.1 kg/s AND Duration < 2 mins]
Combinations of the gas and liquids scenarios described above are possible, depending on GOR
involved.
SIGNIFICANT: (Those between major and minor)

(i) Gas Releases:
Capable of jet fires of 5 to 10 m lasting for between 2-5 minutes, or release rates between 0.1 to
1.0 kg/s lasting 2- 5 minutes giving explosive clouds of between 10 and 3000 m3 in size.
(ii) Liquids Releases (Oil/Condensate/Non-process):
Pool fires between 2 and 10 m in diameter, lasting for between 5 and 15 minutes.
Combinations of the gas and liquids scenarios described above are possible.
45
9 APPENDIX B – RELEASE CAUSES PER YEAR
The table below is based on the full dataset of 35 major releases and 560 significant releases.
The blue bold text indicates the top three causes in each category.
Table 6a Release causes per year - System
Year
System Total
2001 2002 2003 2004 2005 2006 2007
Blowdown 1 0 0 2 1 0 0 4
Drains 1 3 6 3 2 4 2 21
Drilling Equipment 1 0 0 0 2 0 0 3
Drilling Ops 3 1 2 0 0 2 0 8
Export 16 15 13 9 10 6 12 81
Flare 3 3 2 1 0 1 1 11
Flowlines 7 7 4 6 7 7 6 44
Gas Compression 26 23 22 28 18 23 16 156
Import 2 0 1 1 4 2 0 10
Manifold 4 4 4 2 2 3 3 22
Metering 5 6 7 5 2 4 0 29
Processing 3 7 9 5 6 9 3 42
Separation 7 3 9 2 3 4 3 31
Utilities 9 10 8 14 10 7 4 62
Vent 2 0 0 3 5 0 3 13
Well 13 5 2 2 3 3 5 33
Well Ops 10 3 4 2 3 2 1 25
Total 113 90 93 85 78 77 59 595
Table 6b Releases causes per year - Equipment
Year
Equipment Total
2001 2002 2003 2004 2005 2006 2007
Compressors 4 5 3 3 7 6 4 32
Drain Opening 0 2 3 3 1 2 2 13
Drain Plug 0 0 0 1 0 1 0 2
Filters 0 1 2 1 2 3 3 12
Flanges 10 12 9 11 8 3 6 59
Heat Exchangers 2 3 4 3 0 5 1 18
Instruments 25 14 12 14 11 16 15 107
Mud 1 0 0 0 2 0 0 3
Pig Launchers 0 0 2 1 1 2 2 8
Pipelines 2 0 0 0 3 0 1 6
Piping 28 17 22 24 18 11 8 126
Pressure Vessels 2 2 0 0 1 2 0 7
Pumps 4 1 4 1 2 1 2 15
Risers 0 0 0 1 0 1 0 2
Storage Tanks 1 5 6 1 1 0 0 14
Turbines 2 0 1 1 2 2 1 9
47
Valve-actuated 6 8 10 10 8 9 6 57
Valve Manual 3 10 6 6 4 7 7 43
Wellheads 5 3 1 1 0 1 0 11
Xmas Trees 5 2 1 1 3 0 0 12
Blanks 13 5 7 2 4 5 1 37
Total 113 90 93 85 78 77 59 595
Table 6c Release causes per year – Equipment Cause
Equipment Cause Year Total

2001 2002 2003 2004 2005 2006 2007
Awaiting Investigation 5 0 4 2 0 0 0 11
External Corrosion 2 3 3 3 4 0 1 16
Internal Corrosion 9 6 10 7 9 5 4 50
Erosion 3 3 3 3 5 4 1 22
Manufacturing 0 0 1 0 0 0 0 1
Material Defect 2 1 2 1 3 0 1 10
Mechanical Failure 58 26 35 30 27 22 14 192
Mechanical Fatigue 11 11 6 12 8 9 5 62
Mechanical Wear 7 5 6 4 4 5 2 33
None 34 30 23 22 16 15 9 149
Not Known 0 2 0 0 0 0 0 2
Specification 2 3 0 1 0 1 0 7
Blanks 0 0 0 0 2 16 22 40
Total 113 90 93 85 78 77 59 595
Table 6d Release causes per year - Operational Cause
Operational Cause Year Total

2001 2002 2003 2004 2005 2006 2007
Adverse Weather 1 0 0 0 0 0 0 1
Dropped Object 2 1 4 0 1 0 0 8
Improper Inspection 3 5 2 2 2 0 1 15
Improper Maintenance 10 1 10 3 5 7 1 37
Improper Operation 12 13 11 6 10 5 0 57
Improper Testing 1 1 0 0 0 0 0 2
Incorrectly Fitted 13 11 8 15 6 6 4 63
Left to Open 7 9 7 9 8 10 6 56
None 60 48 45 48 41 41 26 309
Opened 1 1 4 1 4 2 2 15
Specification 1 0 0 1 0 2 0 4
Blanks 0 0 0 0 1 4 19 24
Total 113 90 93 85 78 77 59 595
48
Table 6e Release causes per year – Procedural Cause
Year
Procedural Cause Total
2001 2002 2003 2004 2005 2006 2007
Deficient Procedure 15 11 5 8 10 3 1 53
Non-compliance with a
13 10 11 10 5 9 4 62
Procedure
Non-compliance with a
2 1 2 1 3 1 2 12
Permit to Work
None 82 68 73 64 59 59 43 448
Quality Control 1 0 0 0 0 1 0 2
Blanks 0 0 0 0 1 4 9 14
Total 113 90 93 85 78 77 59 595
Table 6f Release causes per year – Operational Mode
Operational Mode Year Total

2001 2002 2003 2004 2005 2006 2007
Blowdown 0 4 2 2 1 0 1 10
Cleaning 0 1 2 1 0 0 0 4
Drill gas 1 0 0 0 0 0 0 1
Drill Oil 1 0 0 0 0 0 0 1
Drill Operation 3 1 2 0 0 2 0 8
Flushing 0 2 3 0 0 0 1 6
Inspection 0 0 1 2 2 0 0 5
Cold Maintenance Work 0 0 0 1 2 0 0 3
Draining for Maintenance 0 2 0 0 0 0 0 2
Normal Production 62 41 50 48 37 46 36 320
Pigging 1 0 2 1 0 1 3 8
Reinstatement 11 9 3 10 6 10 3 52
Replacement 0 2 1 3 0 0 0 6
Routine Maintenance 2 3 0 3 1 1 2 12
Sampling 0 1 0 0 0 2 0 3
Shut down 3 1 1 1 3 0 0 9
Shutting down 3 2 3 2 1 1 0 12
Start Up 9 11 11 6 15 10 7 69
Testing 2 1 5 1 3 1 1 14
Top up 1 1 2 0 0 0 0 4
Well Services 4 3 1 2 1 0 0 11
Well Operation 10 3 4 2 3 2 1 25
Well Operation with Xmas 0 2 0 0 3 0 0 5
Tree in Place
Blanks 0 0 0 0 0 1 4 5
Total 113 90 93 85 78 77 59 595
49
10 APPENDIX C – RELEASES IN TERMS OF AGE OF
PLATFORM
Tables 7a to 7g illustrate the number of releases that occur for each area of interest, for example,
system and equipment. Table entries in red bold are investigated further in section 10.1 to 10.5.
Table 7a System releases per age of platform
Age (years) Type Gas compression Export Utilities

<5 System 20 9 6
5-10 System 48 20 15
10-15 System 36 12 11
15-20 System 21 10 9
20+ System 31 30 21
Table 7b Equipment releases per age of platform
Age Type Piping Instruments Flanges

<5 Equipment 13 9 8
5-10 Equipment 23 20 23
10-15 Equipment 23 27 6
15-20 Equipment 15 19 9
20+ Equipment 54 32 13
Table 7c In-depth analysis of flanges from the previous table
Flange joints Sizes (Inches)

Age Compressed (Grayloc) Spiral
RTJ Chicsan D<=3 3<D<=11 D>11
joint clamp wound
<5 5 3 5 2 1
5-10 11 6 4 1 1 7 10 6
10-15 4 1 1 3 2 1
15-20 3 1 3 2 3 5 1
20+ 7 5 1 6 6 1
Table 7d Cause of equipment failure per age of platform
Mechanical Internal Mechanical

Age Type None
failure corrosion fatigue
Equipment
<5 18 20 2 7
cause
Equipment
5-10 52 34 6 27
cause
Equipment
10-15 43 25 6 5
cause
Equipment
15-20 30 24 3 9
cause
Equipment
20+ 49 46 33 14
cause
50
Table 7e Procedural cause per age of platform
Non-compliant Deficient
Age (years) Type None
procedure procedure
<5 Procedural cause 43 8 4
5-10 Procedural cause 108 11 11
20+ Procedural cause 146 19 20
Table 7f Operational cause per age of platform
Incorrectly Improper
Age (years) Type None
fitted operation
<5 Operational cause 29 10 7
5-10 Operational cause 81 14 11
20+ Operational cause 99 17 16
Table 7g Operational mode per age of platform
Age Normal
Type Start up Reinstatement
(years) production
<5 Operational mode 31 4 10
5-10 Operational mode 87 10 12
20+ Operational mode 107 27 11
10.1 FURTHER INVESTIGATION – FLANGES
Tables 8a to 8c are obtained by filtering the full dataset to reveal only flange releases on
platforms aged 5-10 years. In total, there are 23 records.
Table 8a Equipment causes for flange releases for platforms aged 5-10 years
Equipment Cause Number of releases

Mechanical failure 7
None 6
Mechanical fatigue 8
Mechanical wear 1
Blanks 1
51
Table 8b Operational mode for flange releases for platforms aged 5-10 years
Operational Mode Number of releases

Normal Production 18
Well operation with xmas tree 1
Flushing 1
Start up 2
Shutting down 1
Table 8c Operational cause for flange releases for platforms aged 5-10 years
Operational Cause Number of releases

Incorrectly fitted 8
None 11
Improper operation 1
Awaiting investigation 1
Blanks 2
In addition, nine of the 23 records indicate gas compression as the most common system failure.
Furthermore, there is one major release out of 23 records, the remainder being significant.
Most are gas releases. Two out of 23 are due to design failures and none are ignited.
10.2 FURTHER INVESTIGATION – PIPING
Filtering all piping releases occurring on platforms aged 20 years and over yields 54 release
records. Of these, six are major and 48 are significant.
Table 9 Equipment causes for piping releases for platforms aged 20+ years
Equipment Number of releases

Piping/steel/D<=3 27
Piping/steel/D>11 7
Piping/steel/3<D<=11 16
Piping/flexible/D<=3 3
Piping/flexible/3<D<=11 1
Of the 54 equipment cause releases, 22 are due to internal corrosion, and 10 are due to some
kind of mechanical fault including fatigue, failure and wear.
There appears to be no solid reason for both operational and procedural causes as none appears
most commonly.
The most common operational mode is normal productions, which accounts for 36 out of 54
releases.
Six out of 54 releases are due to design failures while one release results in ignition.
52
10.3 FURTHER INVESTIGATION – INTERNAL CORROSION
Filtering all platforms aged 20 years and over that experience equipment failure due to internal
corrosion yields 33 records. Of these, there is one major and 32 significant releases.
Table 10a Process types involved in release for platforms aged 20+ years
Process Type Number of releases

Oil 17
2-phase 6
Gas 9
Condensate 1
Table 10b Operational causes for platforms aged 20+ years
Operational Cause Number of releases

Improper maintenance 6
None 17
Specification 1
Improper inspection 4
Opened 1
Left to open 1
Blanks 1
For system releases, six are classified as occurring during export and a further six during
separation.
Piping is the equipment type that fails most, resulting in 22 releases out of 33.
21 out of 33 releases occur during the normal production operational mode.
Two out of 33 releases are due to design flaws while one release is ignited.
10.4 FURTHER INVESTIGATION – NORMAL PRODUCTION
Filtering the complete data set for platforms aged 5-10 years and with an operational mode of
normal production yields 87 release records. Major releases account for six out of 87 records.
In addition, 70 out of 87 are gas releases.
Table 11a Equipment item failing on platforms aged 5-10 years
Equipment item Number of releases

Flanges 18
Piping 15
Instruments 14
Valve-actuated 12
Valve manual 8
53
Table 11b Cause of equipment failure on platforms aged 5-10 years
Equipment cause Number of releases

Internal corrosion 5
Mechanical wear 4
Table 11c Operational cause on platforms aged 5-10 years
Operational cause Number of releases

None 59
Left to open 4
Improper inspection 3
76 out of 87 releases have none as the main procedural cause.
15 out of 87 releases are due to design failures. Only one release results in ignition.
10.5 FURTHER INVESTIGATION – MECHANICAL FATIGUE
Filtering platforms aged 5-10 years that suffer mechanical fatigue as an operational mode yields
27 results from the full set of data. Two releases are classified as major while 21 involve gas.
Table 12 Equipment item failing on platforms aged 5-10 years
Equipment item Number of releases

Flanges 8
Instruments 6
Piping 6
17 releases out of 27 are due to the system operation of gas compression. The main procedural
cause is none, accounting for 26 out of 27 releases. The operational mode of normal production
accounts for 21 out 27 releases, while 23 operational cause releases are classified as none.
11 releases are due to a design failure of some kind. No releases were ignited.
54
11 APPENDIX D – RELEASE CAUSES PER QUARTER
The table below is based on the full dataset of 35 major and releases and 560 significant
releases. The blue bold text indicates the top three causes in each category.
Table 13a Release causes per quarter – System
Quarter
System Total
1 2 3 4
Blowdown 0 1 3 0 4
Drains 3 8 5 5 21
Drilling Equipment 0 2 0 2 3
Drilling Ops 1 3 2 2 8
Export 15 22 29 15 81
Flare 2 3 6 0 11
Flowlines 8 10 8 18 44
Gas Compression 44 31 42 39 156
Import 1 2 5 2 10
Manifold 8 4 5 5 22
Metering 6 6 6 11 29
Processing 11 10 10 11 42
Separation 8 8 6 9 31
Utilities 16 10 17 19 62
Vent 3 3 1 6 13
Well 7 7 11 8 33
Well Ops 5 7 9 4 25
Total 138 137 165 155 595
Table 13b Release causes per quarter - Equipment
Quarter
Equipment Total
1 2 3 4
Compressors 12 4 11 5 32
Drain Opening 1 5 3 4 13
Drain Plug 1 0 0 1 2
Filters 2 4 3 3 12
Flanges 11 16 18 14 59
Heat Exchangers 7 2 6 3 18
Instruments 32 15 27 33 107
Mud 0 2 0 1 3
Pig Launchers 2 4 2 0 8
Pipelines 1 3 0 2 6
Piping 30 32 32 34 128
Pressure Vessels 0 4 2 1 7
Pumps 1 3 3 8 15
Risers 0 0 0 2 2
Storage Tanks 2 5 5 2 14
Turbines 1 0 4 4 9
Valve-actuated 10 17 16 14 57
55
Valve Manual 12 5 14 12 43
Wellheads 5 1 4 1 11
Xmas Trees 1 4 4 3 12
Blanks 7 11 11 8 37
Total 138 137 165 155 595
Table 13c Release causes per quarter – Equipment Cause
Quarter
Equipment Cause Total
1 2 3 4
Awaiting
1 4 5 1 11
Investigation
External Corrosion 3 4 2 7 16
Internal Corrosion 13 15 7 15 50
Erosion 5 4 8 5 22
Manufacturing 0 1 0 0 1
Material Defect 2 2 3 3 10
Mechanical Failure 50 40 53 49 192
Mechanical Fatigue 19 18 15 10 62
Mechanical Wear 8 8 7 10 33
None 30 29 47 43 149
Not Known 1 0 1 0 2
Specification 2 3 1 1 7
Blanks 4 9 16 11 40
Total 138 137 165 155 595
Table 13d Releases causes per quarter – Operational Cause
Quarter
Operational Cause Total
1 2 3 4
Adverse Weather 0 0 0 1 1
Awaiting
0 1 1 2 4
Investigation
Dropped Object 0 3 2 3 8
Improper Inspection 9 0 1 5 15
Improper
8 11 10 8 37
Maintenance
Improper
12 14 19 12 57
Operation
Improper Testing 0 1 0 1 2
Incorrectly Fitted 7 16 24 16 63
Left to Open 14 6 18 18 56
None 78 73 75 83 309
Opened 5 3 4 3 15
Specification 2 1 1 0 4
Blanks 3 8 10 3 24
Total 138 137 165 155 595
56
Table 13e Releases causes per quarter – Procedural Cause
Quarter
Procedural Cause Total
1 2 3 4
Awaiting
1 1 1 1 4
Investigation
Deficient Procedure 14 9 20 10 53
Non-compliance
13 9 20 20 62
with a Procedure
Non-compliance with
2 3 4 3 12
a Permit to Work
None 105 112 113 118 448
Quality Control 1 0 1 0 2
Blanks 2 3 6 3 14
Total 138 137 165 155 595
Table 13f Releases causes per quarter – Operational Mode
Quarter
Operational Mode Total
1 2 3 4
Blowdown 4 2 1 3 10
Cleaning 1 1 0 2 4
Drill gas 0 0 0 1 1
Drill Oil 0 1 0 0 1
Drill Operation 1 3 2 2 8
Flushing 1 0 3 2 6
Inspection 2 0 2 1 5
Cold Maintenance
0 0 2 1 3
Work
Draining for
1 1 0 0 2
Maintenance
Normal Production 75 77 84 84 320
Pigging 2 5 1 0 8
Reinstatement 16 10 14 12 52
Replacement 2 0 3 1 6
Routine Maintenance 5 1 3 3 12
Sampling 1 1 1 0 3
Shut down 0 4 3 2 9
Shutting down 13 3 3 6 25
Start Up 3 10 27 19 59
Testing 2 4 4 3 13
Top up 2 1 1 0 4
Well Services 5 4 0 5 14
Well Operation 2 7 9 4 22
Well Operation with
0 1 1 1 3
Xmas Tree in Place
Blanks 0 1 1 3 5
Total 138 137 165 155 595
57
11.1 FURTHER INVESTIGATION – START UP
Filtering the third quarter of the year with start up in the operational mode field yields 27
releases records, one of which is a major release.
Tables 14a to 14f illustrate the number of releases that occur for each area of interest, for
example, process type or equipment.
Table 14a Process type involved in release on start up
Process Number of releases

Gas 20
Oil 4
2-phase 2
Condensate 1
Non-process 0
Table 14b System failures on start up
System Number of releases

Gas compression 8
Export 5
Table 14c Equipment failures on start up
Equipment Number of releases

Compressors 3
Flanges 4
Instruments 6
Piping 3
Valve-actuated 3
Valve manual 3
Table 14d Cause of equipment failure on start up
Equipment cause Number of releases

None 5
Erosion 1
Mechanical wear 3
Internal corrosion 1
58
Table 14e Operational cause leading to failures on start up
Operational cause Number of releases

Left to open 2
None 12
Table 14f Procedural cause leading to failures on start up
Procedural cause Number of releases

Deficient procedure 3
Non-compliance with a procedure 4
None 20
Four releases are due to a design fault of some kind. No releases are ignited.
59
12 APPENDIX E – RELEASE CAUSES FOR EACH
PROCESS TYPE
The table below is based on the full dataset of 35 major and releases and 560 significant
releases.
Table 15a Release causes per process type - System
Process Type
System Non- Total
Gas Oil Condensate 2-Phase
Process
Blowdown 4 0 0 0 0 4
Drains 12 7 0 1 1 21
Drilling Equipment 2 0 0 0 1 3
Drilling Ops 3 1 0 1 3 8
Export 30 44 5 1 1 81
Flare 10 1 0 0 0 11
Flowlines 33 0 0 11 0 44
Gas Compression 149 0 2 1 4 156
Import 6 1 0 3 0 10
Manifold 16 2 0 4 0 22
Metering 18 9 2 0 0 29
Processing 27 5 3 1 6 42
Separation 15 9 2 5 0 31
Utilities 40 0 0 0 22 62
Vent 10 1 1 1 0 13
Well 26 0 0 5 2 33
Well Ops 17 1 0 7 0 25
Total 418 81 15 41 40 595
Table 15b Release causes per process type – Equipment
Process Type
Equipment Non- Total
Process
Compressors 29 0 0 0 3 32
Drain Opening 11 2 0 0 0 13
Drain Plug 1 0 0 0 1 2
Filters 7 4 0 0 1 12
Flanges 53 2 2 1 1 59
Heat Exchangers 13 5 0 0 0 18
Instruments 87 11 1 6 2 107
Mud 2 0 0 0 1 3
Pig Launchers 3 5 0 0 0 8
Pipelines 2 2 0 2 0 6
Piping 64 30 4 17 13 126
Pressure Vessels 4 2 0 0 1 7
Pumps 1 9 4 0 1 15
61
Risers 2 0 0 0 0 2
Storage Tanks 6 4 0 0 4 14
Turbines 5 0 0 0 4 9
Valve-actuated 52 0 3 2 0 57
Valve Manual 37 1 1 2 2 43
Wellheads 10 0 0 0 1 11
Xmas Trees 8 0 0 3 1 12
Blanks 21 4 0 8 4 37
Total 418 81 15 41 40 595
Table 15c Release causes per process type – Equipment Cause
Process Type
Equipment Cause Non- Total
Process
Awaiting
7 3 0 1 0 11
Investigation
External Corrosion 13 1 0 2 0 16
Internal Corrosion 16 20 4 9 1 50
Erosion 12 3 2 5 0 22
Manufacturing 1 0 0 0 0 1
Material Defect 7 0 0 1 2 10
Mechanical Failure 138 22 7 8 17 192
Mechanical Fatigue 43 7 1 5 6 62
Mechanical Wear 27 3 1 0 2 33
None 118 15 0 8 8 149
Not Known 2 0 0 0 0 2
Specification 6 0 0 0 1 7
Blanks 28 7 0 2 3 40
Total 418 81 15 41 40 595
Table 15d Release causes per process type – Operation Cause
Process Type
Operational Cause Non- Total
Process
Adverse Weather 1 0 0 0 0 1
Awaiting
3 1 0 0 0 4
Investigation
Dropped Object 4 2 0 2 0 8
Improper Inspection 9 4 0 1 1 15
Improper
26 5 1 2 3 37
Maintenance
Improper Operation 35 10 3 3 6 57
Improper Testing 2 0 0 0 0 2
Incorrectly Fitted 53 5 1 1 3 63
Left to Open 46 6 0 0 4 56
None 211 42 10 27 19 309
Opened 11 2 0 2 0 15
62
Specification 2 0 0 2 0 4
Blanks 15 4 0 1 4 24
Total 418 81 15 41 40 595
Table 15e Release causes per process type – Procedural Cause
Process Type
Procedural Cause Non- Total
Process
Awaiting
4 0 0 0 0 4
Investigation
Deficient Procedure 38 7 1 3 4 53
Non-comlipance
48 8 0 0 6 62
with a Procedure
Non-compliance
with a Permit to 11 1 0 0 0 12
Work
None 308 61 14 37 28 448
Quality Control 2 0 0 0 0 2
Blanks 7 4 0 1 2 14
Total 418 81 15 41 40 595
Table 15f Release causes per process type – Operational Mode
Process Type
Operational Mode Non- Total
Process
Blowdown 8 0 2 0 0 10
Cleaning 2 2 0 0 0 4
Drill gas 1 0 0 0 0 1
Drill Oil 0 0 0 0 1 1
Drill Operation 3 1 0 1 3 8
Flushing 6 0 0 0 0 6
Inspection 3 1 0 1 0 5
Cold Maintenance
3 0 0 0 0 3
Work
Drainingfor
0 2 0 0 0 2
Maintenance
Normal Production 219 50 10 20 21 320
Pigging 3 5 0 0 0 8
Reinstatement 47 2 1 0 2 52
Replacement 6 0 0 0 0 6
Routine
5 3 0 2 2 12
Maintenance
Sampling 3 0 0 0 0 3
Shut down 6 1 0 0 2 9
Shutting down 10 1 0 1 0 12
Start Up 53 9 1 6 0 69
Testing 6 3 0 2 3 14
63
Top up 0 0 0 0 4 4
Well Services 9 0 0 1 1 11
Well Operation 17 1 0 7 0 25
WellOperation with
5 0 0 0 0 5
Xmas Tree in Place
Blanks 3 0 1 0 1 5
Total 418 81 15 41 40 595
Blue text Failures applicable to ALL process types
Green text Failures applicable to MOST process types
Red text Failures applicable to one or two particular process types
64
13 REFERENCES
1. Thyer, A. M., Offshore ignition probability arguments, HSL Report FS/04/13, 2004
2. Bale, Evan, HSE - Major Hazards Strategic Programme Plan 2006/7, version 1.0, 2006
3. HID Operating Plan 2006-2007
4. http://www.hse.gov.uk/offshore/index.htm
5. Hydrocarbon Releases System, Internet Help File
6. Summary report on the OSD hydrocarbon release reduction campaign, April 2000 to
March 2004, March 2005, HSE, SPC/TECH/OSD/27
7. Offshore hydrocarbon releases statistics and analysis, HID statistics report HSR 2002
002, Feb 2002
8. Loss of Containment Manual, HSE, HID, Offshore Division, March 2007
9. Guidance on the topic assessment of the major accident hazard aspects of safety cases,
HSE, HID, Offshore Division, April 2006
65
14 BIBLIOGRAPHY
Hydrocarbon leak reduction offshore – Report on the findings of the HSE’s process integrity
National Inspection Project (NIP) 200-2004, HSE, July 2005, Revision 1 (SPC/TECH/OSD/28)
Offshore technology report – OTO 96 956 – Revised guidance on reporting of offshore

hydrocarbon releases, HSE, Nov 1996
OSD hydrocarbon release reduction campaign – Report on the hydrocarbon release incident
investigation project – 1/4/200 to 313/2001, HSE Offshore Technology Report 2001/055
Supplementary guidance for the reporting of hydrocarbon releases, UKOOA, September 2002
Dalzell G A et al Guidance on fire and explosion hazards associated with ageing offshore
installations, HSL Report PS/07/06, 2007
Key Programme 3 - Asset Integrity Programme, HSE Hazardous Installations, Directorate,

Offshore Division, HSE Website, November 2007
66
15 NOMENCLATURE
The bullet points below briefly detail some of the causal and mode terms, which are used in
OIR/12 records, analysed in this report.
• EQUIP_CAUSE - Equipment causation is sorted in the database into six primary

categories, two secondary categories, and keywords to further describe ‘other’
categories. This field contains an abbreviated code word to describe the reported
equipment - related failure as follows:
CORRINT = Internal corrosion

CORREXT = External corrosion
MECHFAIL =Mechanical Failure, other than fatigue or wear out
MECHFAT = Fatigue e.g. due to vibration effects
MECHWEAR = Wear out of component
EROSION = Erosion
MATLDEF = Material defects, e.g. metallurgical deficiencies
MANUFACTURING = Defects due to manufacture, e.g. welding defects
NONE = No equipment failure involved
• OP_CAUSE - Operational causation is sorted into seven primary categories, two

secondary categories, and keywords to further describe ‘other’ categories. This field
contains an abbreviated code word to describe the reported operational failure as
follows:
INCORRFIT = Incorrectly fitted

IMPROPMAINT = Improper maintenance
IMPROPINSP = Improper inspection
IMPROPTEST = Improper testing
IMPROPOP = Improper operation
DROPOBJ = Dropped object
OTHIMPACT = Impact other than a dropped object e.g. struck by something or
someone
LEFTOPEN = Left open whilst containing hydrocarbons
OPENED = Opened up whilst containing hydrocarbons
LIGHTNING = Struck by lightning
SPECIFICATION = Incorrectly specified (but not defective) e.g. inappropriate rating or
material
NONE = No operational cause involved
• PRO_CAUSE - Procedural causation is sorted into four primary categories, one

secondary category, and keywords to further describe ‘other’ categories. This field
contains an abbreviated code word to describe the reported procedural failure as
follows:
NONCOMPROC = Non-compliance with a procedure

NONCOMPTW = Non-compliance with a Permit to Work (PTW)
DEFPROC = Deficient procedure
QUALITY CONTROL = Quality control failure, e.g. poor inspection of incoming
goods, usually linked with manufacturing failure
NONE = No procedural failure involved
67
• OP_MODE - The operational mode in the area at the time of release is sorted into nine
primary categories, seven secondary categories, two tertiary categories, and keywords
to further describe ‘other’ categories. This field contains an abbreviated code word to
describe the reported operational mode in the area as follows:
DRILLOIL = Drilling an oil well

DRILLGAS = Drilling a gas well
WELLOPTREE = Carrying out a Well Operation with the Xmas tree in place
WELLOPEXTREE = Carrying out a Well Operation with the Xmas tree removed
NORMPROD = Normal production
PIGGING = Pipeline pigging operation underway
SHUTTINGDN = In the process of shutting down
SHUTDOWN = Already shutdown
BLOWDOWN = Blowing or blown down
FLUSHING =In the process of flushing out
CLEANING = In the process of cleaning
INSPECTION = In the process of inspection
MAINTHOTWK = Carrying out hot work during maintenance
MAINTCOLD WORK = Cold maintenance work not otherwise specified
MAINTDRAINING = Draining for maintenance activities
TOPUP = Replenishing of stocks (e.g. of fuel oil, etc.)
REPLACEMENT = Replacing equipment during maintenance
ROUTINEMAINT = Routine or planned maintenance not otherwise specified
WELL SERVICES = Servicing or maintaining well equipment
CONSTHOTWK = Carrying out hot work during construction
CONSTCOLD WORK = Cold construction work not otherwise specified
CONSTDRAINING = Draining for construction activities
COMMISSIONING = Carrying out the commissioning of newly installed equipment
INSTALLATION = Installing equipment during construction
REMOVAL = Removing equipment during construction
TEMPORARY = Using temporary equipment
WELLCONST = Construction work on wells not otherwise specified under well
operations
TESTING = Testing equipment
SAMPLING = Sampling fluids, etc
REINSTATEMENT = Reinstatement of equipment after maintenance
STARTUP = Initial start-up of equipment after commissioning or shutdown
Published by the Health and Safety Executive 12/08

Health and Safety
Executive

2001-2008
The offshore industry employs about 28 000 personnel
involved in a wide range of activities. Since the Piper Alpha
disaster in 1988, health and safety issues concerning
offshore platforms have vastly reduced, however, the
work practices involved are not risk free and still have the
potential to cause considerable loss of life when things
go wrong. Increases in oil prices, declining reserves and
an ageing infrastructure have resulted in increased drilling
activity around marginal fields. Operators have looked for
new ways in which to cut costs, which could affect the
health and safety of the workforce.
HSE’s Major Hazards Strategic Programme Plan outlined

targets that hope to reduce the number of major and
significant releases from the 2001/02 baseline of 113 to 67
by the end of 2006/07 (10% year-on-year reduction) and
to 60 by the end of 2007/08 (10% year-on-year reduction).
However, in recent years there has been an increase in
the number of major and significant hydrocarbon releases
on offshore platforms that require investigation. This work
hopes to identify the immediate cause of hydrocarbon
leaks, and determine if there are discernable reasons for the
increasing trends.
Two databases currently used by HSE when dealing with

offshore releases were utilised, namely the Hydrocarbon
Release (HCR) and RIDDOR databases. Cross-referencing
between the two catalogues was expected to yield
complete information including platform location, release
size and type, as well as possible failure causes. When
brought together over a range of different releases, this
information can generate an overall picture of issues
related to increasing release frequency.
Once an analysis had been completed, areas that require

improvement, such as structural limitations, system and
equipment faults as well as failings in procedural and
operational methods, were indicated where possible.
This report and the work it describes were funded by

the Health and Safety Executive (HSE). Its contents,
including any opinions and/or conclusions expressed, are
those of the authors alone and do not necessarily reflect
HSE policy.
RR672
www.hse.gov.uk

Offshore Hydrocarbon Releases 2001-2008

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Offshore Hydrocarbon Releases 2001-2008

Hochgeladen von

Copyright:

Verfügbare Formate

Health and Safety

Offshore hydrocarbon releases

Prepared by the Health and Safety Laboratory

for the Health and Safety Executive 2008

Offshore hydrocarbon releases

First published 2008

All rights reserved. No part of this publication may be

reproduced, stored in a retrieval system, or transmitted

in any form or by any means (electronic, mechanical,

photocopying, recording or otherwise) without the prior

written permission of the copyright owner.

Applications for reproduction should be made in writing to:

Licensing Division, Her Majesty’s Stationery Office,

St Clements House, 2-16 Colegate, Norwich NR3 1BQ

2.1 HCR Database .........................................................................................2

2.2 RIDDOR Database...................................................................................2

3 HYDROCARBON RELEASES OVERVIEW................................................3

3.1 Yearly and monthly release trends ...........................................................3

3.2 Release trends in relation to process type................................................5

3.3 Release trends in relation to platform.......................................................6

3.4 Release trends in relation to platform age ................................................7

3.5 Release trends in relation to system pressure..........................................8

3.6 Other information....................................................................................10

4 HYDROCARBON RELEASE ANALYSIS .................................................15

4.1 Partial data .............................................................................................15

4.2 Full data..................................................................................................24

4.3 Overall analysis ......................................................................................26

5 REVIEW OF OTHER RELEVANT STUDIES / GUIDANCE.......................28

5.1 OSD hydrocarbon release reduction campaign......................................28

5.2 Loss of containment manual...................................................................31

5.3 Assessment of the major accident hazard aspects of safety cases........34

6.1 Comparison with Hydrocarbon release reduction campaign ..................39

6.2 Comparison with loss of containment manual ........................................39

6.3 Comparison with GASCET .....................................................................40

7 CONCLUSION AND RECOMMENDATIONS............................................42

8 APPENDIX A – SEVERITY CLASSIFICATION ........................................44

9 APPENDIX B – RELEASE CAUSES PER YEAR .....................................47

10 APPENDIX C – RELEASES IN TERMS OF AGE OF PLATFORM.......50

10.1 Further investigation – Flanges ..............................................................51

10.2 Further investigation – Piping .................................................................52

10.3 Further investigation – Internal corrosion ...............................................53

10.4 Further Investigation – Normal production..............................................53

10.5 Further investigation – Mechanical fatigue .............................................54

11 APPENDIX D – RELEASE CAUSES PER QUARTER..........................55

11.1 Further investigation – Start up ..............................................................58

Comments on the full set of major and significant releases were:

• Non-compliance with a procedure is the most common procedural cause that is

Main causes per quarter:

2.1 HCR DATABASE

2.2 RIDDOR DATABASE

3.1 YEARLY AND MONTHLY RELEASE TRENDS

Figure 1 Total number of minor, significant and major releases

Major and Significant Releases

2000 2001 2002 2003 2004 2005 2006 2007 2008

Figure 2 Number of major and significant releases per year between

Figure 3 Number of major and significant releases per month between

3.2 RELEASE TRENDS IN RELATION TO PROCESS TYPE

2003 2004 2005 2006 2007

Figure 4 The number of major and significant releases corresponding to

3.3 RELEASE TRENDS IN RELATION TO PLATFORM

Figure 5 The number of major and significant releases corresponding to platform

3.4 RELEASE TRENDS IN RELATION TO PLATFORM AGE

2001 2002 2003 2004 2005 2006 2007