A reliability-based design format for jacket platforms under wave loads

Abstract

When the specification of safety goals for offshore jacket platforms is made in explicit “probability of failure” terms, a design procedure may be developed that rationally derives design load return periods and levels for sizing. One such procedure is described here. Using information on the probabilistic description of annual maximum wave heights, an ultimate-level wave height is first estimated that establishes a load level against which performance will in the end be confirmed. From this ultimate-level load, one derives a lower level of load (and associated design-level wave height) to be employed for sizing members by assuming/anticipating the reserve strength characteristics of the jacket beyond the design level and into the nonlinear post-yield regime. A primary objective in establishing this procedure was that initial member sizing and interaction ratio (IR) checks be conducted using familiar procedures (e.g., API RP2A-WSD [10]. Within the framework of the proposed design procedure, members are first sized in the conventional manner (e.g., by working stress or LRFD procedures). A nonlinear static pushover analysis is then performed to establish the initial design’s ultimate capacity. It is desirable that the proposed design should yield a reliability that meets the target/desired reliability for the jacket. Both, the ultimate-level and the design-level wave loads, are derived with this goal in mind while employing various assumptions regarding the nature of the loads, response, and ultimate capacity of the structure. The initial assumptions are, if necessary, altered iteratively. By way of illustration of the proposed procedure, a jacket platform is adequately designed after a few member-sizing cycles. During these iterative design cycles, critical member sections are adjusted to ensure that the reliability is close to the target level; and the various assumptions made (in deriving load levels and that serve as the basis for load/capacity estimates) are modified and/or verified.