The purpose of this paper is to explain and demonstrate how statistical decision theory can be used to determine the optimal level of structural capacity for a proposed offshore platform. This type of analysis can be used to decide whether additional information concerning the likely future hazard environment of a proposed offshore platform should be obtained. This use of statistical decision theory is the main focus of the following discussion. A proposed platform at an offshore southern California site in the vicinity of Tanner Bank has been chosen as an example to demonstrate statistical decision analysis procedures.
Applied statistical decision theory initially was developed and applied in the business world. The methodology is particularly attractive to engineers since it logically models the engineering decision-making process. Engineering problems, however, differ materially from those in the business world,1,2,3,4,5 so that it is necessary to adapt and translate decision theory techniques to the realities of the engineering environment in order that the methodology can be used to make engineering decisions.6
The problem considered in this paper is that of deciding whether to obtain additional information concerning earthquake ground motion and ocean wave characteristics in the Tanner Bank area before deciding how strong to build an offshore platform. If it is decided to obtain additional data, the next step is to define the optimum experimental or investigative program considering the data to be obtained, the cost, and the value to be received. This problem is complex and the decision must be based on all uncertain aspects of the future.
The components of the decision process are shown in Figure 1. The maker of the decision is shown in Box 1. The decision maker is concerned about making a design decision (Box 4), but before making that decision, the first question is whether or not to conduct an investigation (Box 2) in order to obtain more data (Box 3). Once the design decision is made, the platform will be constructed and placed at the site. The decision maker will then observe the performance of the platform for the life of the structure (Box 5), and receive value depending on the entire sequence. The decisions in Boxes 2 and 4 are made in accordance with a plan or decision rule.
The diagram of Figure 1 is sufficiently flexible to accommodate almost any design decision situation. Investigation plans and the result of the investigations must relate to future platform performance. It is obviously not possible to conduct an investigation that involves sampling the future platform performance; thus the investigation must consist of obtaining improved loading data. It is assumed that performance can be satisfactorily forecast once the loadings are probabilistically defined. However, an additional level of uncertainty exists between loading and performance which also must be considered in any decision analysis.
