Atkins’ Ramsay Fraser looks at offshore asset integrity assessment techniques and how acceptance criteria depend on geographic region.
Evaluation of the integrity of an offshore structure is one of the four key functions of an integrity management system – the others being acquisition and management of data; strategy for in-service inspection, mitigation and/or early decommissioning; and program for in-service inspection.
Evaluation requires assessment if an initiator is triggered, such as development projects on existing installations, when new metocean or geotechnical information becomes available and, for older assets, to ensure that degradation due to fatigue, corrosion, and seabed settlement from reservoir compaction is acceptable.
Development projects on fixed installations typically involve a significant increase in topside weight due to new process facilities together with increased wave load due to new risers or conductors.
The quantity of metocean data and quality of data processing and modelling has increased significantly in some geographic regions and has resulted in greater awareness of higher assessed probability for wave-in-deck loads during extreme storms and appreciation of the implications of these higher probabilities.
While the assessed loads typically increase due to the above, the latest computer simulation techniques (such as ultimate strength analysis using software such as ABAQUS1 with the critical components of the structure represented in detail and embedded within the model of the total structure (in order to provide the correct boundary conditions) usually demonstrate that the structure has greater capacity than shown using previously adopted techniques.
The ISO2 and API standards for offshore structures are currently being updated and one of the outcomes is likely to be greater consistency in the assessment methodology and criteria required for the integrity of offshore structures in all regions of the globe. Recent assessments of fixed offshore structures by Atkins in the North Sea, Gulf of Mexico, Arabian Gulf, Caspian Sea, and NW Shelf have indicated that to achieve greater consistency:
1) The industry needs to set acceptance criteria based on reliability studies, structural performance around the globe and experience. The criteria need to be straightforward and limited to linear and non-linear methods of analyses and software. Good guidance is required in the codes to allow most engineering contractors to do good work. This guidance should focus on the structural modelling; loading model and material behavior in component design versus system ultimate capacity checks. 2) Engineers performing structural assessments require a deeper understanding of the assessment methods used by metocean consultants and geotechnical engineers.
The acceptance criteria for the safety level of an offshore structure have developed since the early API codes and in particular in the 1980s in the Gulf of Mexico and the 1990s in Europe.
The 1980s developments came together under the umbrella of API in the early 1990s after Hurricane Andrew in 1992 and led to Section 17 being established in API for assessment of existing structures. The work was very focused on the Gulf where platforms are unmanned for the extreme events and for which therefore reduced criteria, based on economics are acceptable, since life safety was preserved. This is unique to the Gulf of Mexico and has often been missinterpreted and applied incorrectly in other parts of the world.
In the mid 1990s, Efthymiou et al3 showed that different geographic regions have different probabilities of occurrence of an extreme storm load normalized by its 100 year storm load. Regions with fewer larger storms (e.g. cyclonic regions such as the North West shelf) have a greater probability of extreme load than regions with more sustained storm conditions such as the North Sea. The 1990s developments were incorporated in the 2007 publication of ISO 19902.
The ISO 19902 code allows three methods of capacity assessment:
a) DLA (Deign Level Assessment)
b) RSR (Reserve Strength Ratio)
c) SRA (Structural Reliability Assessment).
The DLA determines the forces on each component in the structure from a linear analysis (with non-linear pile-soil interaction) and compares these against the capacities in the code equations for each component and the regional specific load factor γE. The RSR determines the collapse capacity of the whole structural system (as a multiple of the 100-year storm load and using mean or best-estimate resistance) and compares against the code requirement. The SRA determines the return period for failure of the structural system (i.e. the reciprocal of annual probability of failure Pf) and compares against the code implicit probability of failure. Assessment by SRA is more structure specific and thus less conservative than the RSR method, which, in turn, is more structure specific and thus less conservative than the DLA method.
The acceptance criteria for all three methods depend on the geographical region of the installation. However, for a given geographical region, the acceptance criteria for the three methods are related. The (shifted exponential) probability density for the extreme storm load can be plotted as a hazard curve as shown in the figure below. The solid red line represents the hazard curve for the Northern North Sea (NNS) with no uncertainty in load or resistance i.e. it represents the probability of a storm of a given load magnitude occurring in any year at the installation location. The dashed red line represents the hazard curve for the NNS with uncertainty in the load given a specific storm were to occur together with the uncertainty in jacket capacity given the failure mode comprises failure of four or more braces. The dotted red line is the same as the dashed line, but with the uncertainty in jacket capacity given the failure mode is a single leg failure. The dashed dark blue line represents the hazard curve for the Gulf of Mexico with uncertainty in the load given a specific storm were to occur together with the uncertainty in jacket capacity given the failure mode is one of four or more braces failing.
These hazard curves show the relationship between RSR (on the left axis), γE for the DLA (on the right axis) and return period for failure of the structural system for the SRA on the horizontal axis. These curves also assume that the members participating in the failure mechanism are dominated by wave load rather than dead load. As described by Efthymiou et al3, these hazard curves show the relationship between RSR γE and return period for structural failure for the NNS for different modes of failure.
All of the above assume sufficient air gap such that wave-in-deck (WID) load does not significantly contribute i.e. the storm load is “wave-in-jacket” (WIJ) only.
However, before these curves can be determined accurately and used appropriately, the metocean data must be analyzed in a specific and consistent manner using the response-based method developed by Tromans & Vanderschuren4. Atkins has found that the industry’s metocean consultants do not yet routinely use this method and consequently, the resulting calculated structural safety level is not consistent and usually conservative.
The probability distribution for the crest elevation of the extreme waves in a storm together with the measured elevation of the deck above mean sea level are required in order to determine the annual probability of wave-in-deck load.
The figure below shows hazard curves relating return period for failure of the structural system versus the platform’s overturning moment (OTM) capacity normalized by the magnitude of the OTM having a return period of 100 years.
The blue line represents the hazard curve for WIJ load on the platform while the green line represents the hazard curve for WID load on the platform. The red line represents the hazard curve for combined WIJ and WID load on the platform. This curve shows that once the return period is sufficient for WID load to occur, the OTM capacity required increases rapidly in order to achieve a sufficient return period for failure of the structural system. This is to be expected due to high WID loads acting at a significant lever arm from the legs at the lower elevation of the jacket.
The red dotted hazard curve on the earlier diagram was calculated from base shear and thus it would be steeper if it were based on OTM and consequently the code implicit return period to failure would reduce to slightly below 10,000 for NNS jackets designed to ISO 19902 with γE=1.35 and a single leg mode of failure.
The hazard curves give the capacity required in order to achieve a specified return period for failure of the structural system. If significant additional topside load is to be added to a platform, then an additional check on the reliability of the jacket and foundations is required for the still water condition.
The plot below shows the annual probability of failure for a structural component when it is designed to an ISO code utilization ratio of 1.0 and subjected to loads ranging from 100% dead load to 100% wave load. The plot shows that the code equation for the L1 condition governs the Pf for high proportions of wave load, the abnormal 10,000-year code equation governs for members with more than 80% of their load being due to wave load and the still water code equation governs for low proportions of wave load. This plot illustrates that compliance with the ISO code for operational conditions results in excessive conservatism for structural components with a low proportion of storm load in comparison to components with higher proportions of storm load.
Atkins has applied the above methodology to assessment of jacket structures in many global regions. Although some further work is required to establish complete consistency across all regions, the next revision of the ISO code may well achieve this. OE
|Ramsay Fraser is Atkins’ technical director for offshore structures. Having originally managed Atkins’ Aberdeen office across all disciplines, Ramsay now has a global role which sees him work on projects with Atkins teams from across the world. He has a BSc and PhD in Engineering from the University of Aberdeen|