This website uses non-intrusive cookies to improve your user experience. You can visit our cookie privacy page for more information.

Analysis of investigated offshore hydrocarbon releases

This section updates the report produced last year (OTO/2001/055) in which all releases reported in 2000/01 were investigated as part of the Process Integrity Initiative (See Appendix 1).

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 taxonomy presented in table 5 of last year’s report. The following conclusions can be drawn from this sample.

4.1 Release site (see table 9)

4.1.1 Investigation reports

Again pipework accounts for the majority of releases, 62%, this is comparable with last year’s 61%. However, there are some differences within this grouping between this year’s analysis and that of last year. A greater proportion of this year’s release sites have been assigned to small bore pipework and associated connections, including instruments, 25% as opposed to 18% last year. Perhaps of more significance is 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:

  1. 70% of releases were cracks, splits or holes in the containment envelope. This is an increase compared to the 47% of the previous years.
  2. 23% of releases were from the body of the pipe, vessel or valve. This is in line with 21% for last year. Mostly these resulted from degradation of the containment envelope caused by degradation of material properties.
  3. 25% of releases were associated with small bore piping including instruments, the bulk of these were associated with connections.
  4. Flanges accounted for 15% of releases, which is exactly the same proportion as last year.
  5. 19% of releases were from seals or valve stems. This is an increase over last year’s proportion of 14%.
  6. There were no hose releases in the sample investigated.

4.1.2 Comparison with full set of HCR data

The information provided in the hydrocarbons release database also records the location of release albeit using a different coding framework from that used in OTO 2001 055 (see table 14). This confirms that in most respects the sample of investigated reports is representative of the location of releases for the whole set of 241 releases. Releases from instrument (22%) and pipework (22%) dominate. These results reflect the indication from the sample that there is a slight change towards instrument releases and away from pipework compared to last year’s data.

There were only 12% of releases from valves in the full data set which is in line with last year’s results and indicates that the sample of investigated reports has a higher proportion of these types of release (21%).

The other main difference is in relation to releases from flanges. The sample includes 15% of these releases which is a greater proportion than the whole 241 releases which only had 7% flange releases.

There were no hose releases investigated this year. However, 3% of the total releases were from hoses, this was in line with 4% of releases last year.

[back to top]

4.2 Immediate causes (see table 10)

4.2.1 Investigation reports

As last year, the largest single cause was degradation of material properties which accounted for 28% of the incidents, comparable with last year’s 26%. Incorrect installation at 21% and fatigue/vibration also 21% were the second largest contributors. Both of these are almost double last year’s figures of 12% and 11% respectively, and more significant than corrosion/erosion which accounted for 13%, this is a drop compared to 19% last year. Procedural type causes were all less than 10%.

4.2.2 Comparison with full set of HCR data

A check against the full set of 241 releases on the HCR database shows good agreement for erosion/corrosion causes. The full set of data shows an even greater contribution from mechanical failure and wearout which together account for 36% of the total, whereas fatigue accounted for 12% of the total. Incorrectly fitted at 7% of the total is the most frequent of the procedural failures. Although the percentages differ between the sample and the full set of data, the two sets of data give the same indication of the most prevalent immediate causes, despite using different causation frameworks. One interesting result from this comparison is to confirm the finding that in this year’s releases, a much smaller proportion were due to releases from open end of pipes or valves compared to last year. This is confirmed by the causes “left open” or “opened with hydrocarbon present” which accounted for 7% of releases. This is in line with the 10% of “open end” releases in the sample which is a reduction compared to the 20% of such releases occurring last year.

Degradation of material properties is taken to mean loss of integrity by failure of equipment that was originally fit for purpose and has been operated correctly. This includes failure of flange gaskets and valve packings and might be termed general “wear and tear”. It excludes the more specific causes of failure such as corrosion, erosion, fatigue and vibration. Unlike last year, three of the four major releases had immediate causes of degradation of material properties and only one was due to operator error. Although four is too small a number to draw any far-reaching conclusions.

4.2.3 Immediate cause versus release site

When it comes 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 is a significant increase compared to last year. Degradation of material properties was much less significant at 15%. For small bore tubing and associated connections (25% of releases) fatigue accounts for 33% of causes, incorrect installation a further 25% and degradation of material properties 17%. This is in line with last year’s findings although incorrect installation plays a more prominent role. Open ends only account for 10% of releases in this analysis. The causes are evenly distributed although most are procedural rather than hardware related, which is not really surprising.

4.2.4 Hardware versus Software

The immediate causes can be divided into hardware or software related as follows:

Hardware

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

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%).

[back to top]

4.3 Underlying causes (table 11)

4.3.1 Investigation reports:

The underlying causes of many incidents are complex and more than one cause can be identified in many cases. When these are analysed independently of immediate cause, the largest contribution was from inadequate inspection/condition monitoring in 32% of investigated incidents, closely followed by inadequate design 30%. These are very similar to last year’s figures of 28% and 29%. The next four most significant underlying causes were inadequate procedures in 23% of incidents, incorrect installation 15%, incorrect material specification/usage 13% and inadequate risk assessment 13%. Other underlying causes were identified in less than 10% of investigated incidents.

4.3.2 Comparison with full set of HCR data

Comparison with last year’s analysis reveals that the same two underlying causes dominate but that there is a slightly different contribution from the other causes. Excessive workload, inadequate communication, outdated information and inadequate task specification again made little or no contribution.

[back to top]

4.4 Failed safeguards (table 12)

As with last year’s analysis, inspection/condition monitoring was the most prevalent failed safeguarding system that might have prevented the release of 30% of all incidents investigated. 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 in last year’s analysis.

There was no identifiable pattern for major releases.

[back to top]

4.5 Additional safeguards (table 13)

The UKOOA small bore piping guide continues to be relevant, in 23% of investigated incidents. But in this year’s analysis, the flange verification scheme was less relevant, only in 4% of this sample.

[back to top]

4.6 Conclusions

The sample of investigated releases shows similar results to the analysis of a larger sample from last year. Where there are changes identified it is difficult to be certain that these are real differences or whether they arise from a different interpretation of some of the coding. However, this analysis indicates that important lessons can be learned from analysis of investigation reports, over and above the information that is available from the Hydrocarbons release database.

The results indicate that conclusions drawn from the previous analysis are still valid and therefore the focus of the process integrity intervention project should continue. In order to identify any real change in the causes of hydrocarbon releases, analysis of a larger sample would be necessary which perhaps points to collection of more investigation reports during the remaining years of the project. This information would also be of use in demonstrating the impact of the intervention, not only in driving down the number of releases, but also in tackling some of the immediate and underlying causes.

[back to top]

4.7 HSE Process Integrity Initiatives

Appendix 1 of this report outlines the framework for the OSD Process Integrity Initiative aimed at raising industry awareness of offshore hydrocarbon releases and the need for their reduction. The initiative is structured around 10 key elements that could contribute to the prevention of hydrocarbon releases. Not all elements are applicable to all installations and some elements are of lesser direct impact for releases for many installations. OSD have conducted process integrity inspections on manned production installations commencing in the year 2000.

To the end of 2002, inspection progress has been made on 87% of installations covered by the initiative. The greater activity has been on 7 of the 10 elements that have larger direct bearing on potential releases. These include issues such as process integrity management, documentation and change control, and design and operational issues including isolation & PTW, small bore pipework, plant protection systems and maintenance & verification of safety critical elements. Around half of all possible inspections across all installations in these 7 elements have been completed. Work has also progressed on the remaining elements.

HSE has kept the offshore industry informed of key findings emerging from inspections via joint HSE/industry workshops held at staged points throughout the project (Appendix 1). As information has been passed on, emphasis has moved towards the current position of development of remedial programmes by dutyholders to target their high risk areas and activities. During the coming year, HSE will be targeting efforts towards the poorest performing companies, as measured by the number of releases per manned production installation (see Table 15 in Section 5.0).

OSD will use the existing data on offshore hydrocarbon releases to target inspections on dutyholders and installations that appear to have, in recent years, the poorest performance. This is intended to focus inspection resources towards areas of highest risk, in line with HSE policy. The inspections will use analysis of incident investigations and other sources of information to identify topics for particular attention. Any findings of non-compliance with regulations that arise from the inspections will result in enforcement as set out in existing HSE and OSD policy.

Updated 2009-09-06