ABSTRACT

Hurricane Andrew passed through some 3,000 offshore structures in the Gulf of Mexico on August 24 to 26, 1992. While most structures were not adversely affected by Andrew, several sustained significant damage and in some cases collapsed. Since platforms were observed to survive as well as fail during the storm, there was a unique opportunity to determine if the present wave force and ultimate capacity procedures could have predicted these "observed" survivals or failures.

Such a study was conducted during a Joint industry Project funded by 12 organizations that closely studied 13 platforms that were affected by Andrew [1]. Each of these platforms was analyzed using nonlinear analysis to determine its performance (survived, damaged, failed) during Andrew. The analytically-based performance was then compared to the observed actual field performance.

A rigorous, probabilistic-based, "Bayesian" updating process was used to make the comparisons. These comparisons resulted in the calibration of a "bias" factor that provides a general indication as to the accuracy of the wave force and ultimate capacity procedures. This paper summarizes the approach used in the project and the results of the calibration process. The results of this work have helped guide API Task Group (TG) 92-5, which is responsible for developing guidelines for assessment of existing platforms.

BACKGROUND

An extreme event such as hurricane Andrew provides an opportunity to determine the effectiveness of existing technology for evaluating the performance of offshore structures by asking the question: Does the present state-of-the-art wave force and ultimate capacity procedures (hereafter called procedures) for offshore steel jacket platforms provide a reasonable estimate of what would really happen in a large storm? Andrew provided an excellent opportunity to test this premise since it was a very large storm which in some regions exceeded the API recommended design wave criteria. In addition, Andrew generated the three types of "observations" (platforms that survived, platforms that were damaged, and platforms that failed) that allow a full scope test of the procedures. Previous similar comparison studies have attempted a similar "calibration" [2, 3] but few had the amount of detailed information, and particularly, the number of damaged and failed structures that were available from Andrew.

GENERAL APPROACH

The calibration process involves a comparison of analytically predicted platform performance to observed platform performance. For steel jacket offshore platforms, there are several parameters in the procedures which can be "calibrated" (i.e., adjusted so that their true values equal computed values). For example, specific parameters of the procedures, such as the drag coefficient or material strength, could be calibrated. However, it was felt that such attempts would require substantial specific detailed information in addition to a large number of platform cases which was not practically feasible. Therefore, the calibration effort was decoupled from specific details of the procedures and instead a bias factor "B" was introduced that affects the "safety margin" of the platform, defined as the ratio of resistance (R) to load (S).

This content is only available via PDF.
You can access this article if you purchase or spend a download.