Development of Pile Foundation Bias Factors Using Observed Behavior of Platforms During Hurricane Andrew

Abstract

Abstract The performance of more than 3,000 offshore platforms in the Gulf of Mexico was observed during the passage of Hurricane Andrew in August 1992. ?IMs event provided an opportunity to test the procedures used for platform analysis and design. A global bias was inform for overall platform capacity and loads in the Andrew Joint Industry Project (JIP) Phase 1.1 It was predicted that the pile foundations of several platform should have failed, but did not. ?lime results indicated that the biases specific to foundation failure modes may be higher than those of jacket failure modes. ?Ile biases in predictions of foundation failure modes were therefore investigated further in this study. ?he work included capacity analysis and calibration of predictions with the observed behave for 3 jacket platforms and 3 caissons using Bayesian updating. Bias factors for two foundation failure mess, lateral shear and overturning, were determined feat structure. Foundation capacity estimates using conventional methods were found to be conservative y biased overall. Introduction Hurricane Andrew was a very intense storm that passed through the Gulf of Mexico in August 1992. While most of the Gulf of Mexico platforms were not adversely affected by Andrew, twenty eight steel jacket platforms were significantly damaged. All of these were installed during 1948 to 1969 and were located in water depths up to 143 feet. In addition, forty seven caissons, nstalled during 1979 to 1991 in water depths up to 113 feet, were also significantly damaged. While foundation failure was identified as the primary cause of damage to most caissons, it was identified as the cause for damage of only one steel jacket platform. An extreme event such as Andrew provides a unique opportunity to study offshore structures tested under fill scale field conditions. By reviewing the platforms that survived, were damaged or failed during the hurricane it is possible to improve our understanding of the behavior of platform during large storm. In October, 1993, Phase I of a joint industry project, "Hurricane Andrew - Effects on Offshore Platforms,''1'3 was completed. This project established that there was bias in the safety factor (capacity to load ratio) by combining the analytical and observed behavior of 13 jacket platforms. The inform bias is characterized as a correction needed to bring analytical results to agreement with the observed results. The capacity analysis performed in Phase I of the Andrew JIP indicated that failure of foundation elements should have occurred in a majority of platforms analyzed, Since this was not observed during post-Andrew inspections, it was concluded that the deterministic analysis was conservatively biased. An overall (system) correction factor, with a mean of 1.2 was established. This correction factor was not specific to any particular failure mode in the jacket structure m its foundation but rather included failures of all types.