A significant finding from the performance of jacket platforms in major hurricanes, including Andrew (1992), Roxanne (1995), Lili (2002), Ivan (2004), Katrina (2005) and Rita (2005), is that pile foundations performed better than expected (e.g. Aggarwal et al. 1996, Bea et al. 1999 and Energo 2006 and 2007). Jacket platforms are fixed-base offshore structures that are used to produce oil and gas in relatively shallow water,
generally less than about 500 feet. The foundation systems consist of driven, open-ended, steel pipe piles. While observed damage or failure in the foundation is rare, assessment of jacket platforms subjected to environmental loads greater than their original design loading frequently indicates that the capacity of the structural system is governed by the foundation. In addition, there were several hundred platforms damaged in these hurricanes and yet there were only a few cases of reported pile foundation failures. Therefore, there is a need to better understand and quantify the potential conservatism in foundation design for the purposes of assessing platforms.
The objectives of this project were to identify and analyze the factors that may contribute to the apparent conservatism in foundation design and to provide guidance on how to incorporate this information in assessing existing platforms. The methodology was to compile and analyze data for existing platforms that have been subjected to hurricane loads near or greater than the design capacity for the foundations. An expert panel of practitioners and researchers was convened to guide the work.
The major conclusion from this work is that the performance of platform foundations in recent hurricanes, based on the available but limited information that we have, is consistent with expectations based on their design and there is no direct evidence of excessive conservatism. In the cases we analyzed in detail, the actual performance of the foundation was either expected or could be explained without conservatism in the design capacity of the foundation.
These results do not preclude the possibility of foundation capacities being greater than expected based on design. One of the limitations of this study was lack of cases from recent hurricanes where the ultimate strength of the foundation system was reached or exceeded in hurricanes. Foundations are designed with a factor of safety, such as a factor of 1.5 for axial loading under 100-year hurricane conditions. The magnitude of hurricane conditions required to reach this value is significant, requiring extreme waves and currents, and was not reached in most of the cases studied (i.e., a foundation failure was not expected and indeed was not observed). Of all the cases obtained from MMS files and industry sources, there were unfortunately only a few that met the high loading condition required to truly test the foundation. In addition, there is redundancy in foundation systems so that overload of a pile, either axially or laterally, does not necessarily lead to collapse or even observable damage. Finally, the study was limited to platforms that were not destroyed since little if any information is available about the performance of platforms that were destroyed.
The major factor contributing to potential conservatism is the effect of long-term set-up (or aging) or pre-loading; there is evidence from laboratory and field studies to suggest that both axial and lateral pile capacities may increase with time beyond the values that are assumed for design. However, in the one case we do have of a foundation failure in a hurricane (an axial pile failure in clay), there is no evidence of the capacity being greater than the design value.
Platform foundations can fail both in shear where the piles are failing laterally (plastic hinges form due to bending) and overturning where the piles are failing axially (plunging or pulling out). Therefore, both axial and lateral capacities are significant for pile foundations in platforms. The axial capacity of piles is derived mostly from the soils in the bottom one-third of the piles. The axial capacity of the piles and, therefore, the
overturning capacity of the foundation are approximately proportional to the shear strength of the soils. The lateral capacities of the piles and conductors are derived mostly from the soils in the upper 40 to 50 feet below the mudline depending on their diameters. The lateral capacities of the piles and conductors and the shear capacity of the foundation are much less sensitive to the shear strength of the soils than the axial capacity for typical soil conditions in the Gulf of Mexico.
Structural elements are important to the performance of a foundation system. Well conductors can contribute significantly to the shear capacity of a foundation system, and in some cases to the overturning capacity. The yield strength of steel in the piles also affects the shear capacity of the foundation. Increasing the nominal yield strength to reflect the average value can have a significant effect on the shear capacity for a foundation system, and a much greater effect than increasing the shear strength of the soil by the same amount.
The presence of sand layers contributing significantly to pile capacity was a notable factor in the platforms analyzed herein. Sand is significant because it generally corresponds to a geologic setting where there is significant spatial variability over rather short distances. A soil boring not drilled at the location of a platform when sand is significant to pile capacity, even if the boring is within several thousand feet of the structure, may not provide representative information for the soil conditions at the platform location. In addition, most historical soil borings in the Gulf of Mexico used a Driven Penetration Test to characterize the shear strength of sand layers. This method is generally considered to be outdated, it may not have fully penetrated the sand layers due to sampler refusal, and it has been replaced over the past several decades with Cone Penetration Testing. Finally, pile capacity models when sand layers are present are more complex than for clay layers alone, and we identified numerous cases where sand layers were inappropriately modeled in pushover analyses due to this complexity.
A final conclusion is that general trends in foundation performance cannot be drawn easily based on qualitative assessments. Each platform case, considering the water depth, vintage, structure, geologic setting, hurricane loading and platform performance, is unique and a detailed analysis is required to understand how it performs. The following guidelines are intended to provide a defensible and consistent approach for modeling pile foundations in platform assessments:
- When the foundation controls the assessment, include a geotechnical engineer familiar with platform assessment.
- Include well conductors realistically and explicitly in the structural analyses.
- Consider using mean rather than nominal yield strength for steel piles and well conductors.
- Use static versus cyclic p-y curves for lateral soil capacity.
- Be careful in specifying the input for sand layers.
In addition to the specific guidelines for how foundations are modeled in platform assessments, the following general guidance is provided to improve the overall practice of platform assessment:
- Be careful when relying on a soil boring that was not drilled at the location of the platform or was not drilled using modern methods of sampling and testing (pushed, thin-walled tube sampling for clay layers and Cone Penetration Test for sand layers).
- Try to obtain pile driving records as well as soil boring logs to help estimate the axial pile capacities and as-built conditions.
- When the pile foundation system is governing the capacity of the platform in a pushover analysis, check the sensitivity of the foundation system capacity to the lateral and axial capacity of the piles independently.
- Do not arbitrarily increase the shear strength of the soil to account for perceived conservatism in foundation design.
The following recommendations are intended to improve the API Recommended Practice for Platform Assessment:
- Provide specific guidance for characterizing pile foundations in platform assessments by incorporating the guidelines developed in this study into RP 2SIM.
- Update p-y curves for clay in API RP 2A.
- Better clarify and update design guidance for sand in API RP 2A.
- Appropriately account for pile flexibility when determining the required pile length.
A major limitation of this study was not analyzing platforms that were destroyed by hurricanes. If detailed analysis could be conducted on these platforms, they may provide valuable information about how pile systems and the jackets they support performed under extreme loading conditions and may allow us to refine our conclusions. In addition, there may be additional platforms that survived even though the pile systems experienced loads greater than their capacity. Such platforms should be considered for future study. Finally, performing a Cone Penetration Test at the location of the two case study platforms where we are uncertain about the geotechnical properties could provide important information to better understand why these structures survived.