Probabilistic Structural Design Research Articles

Simulation approaches to probabilistic structural design at the component level

In this paper, structural failure of large nuclear components is viewed as a random process with a low probability of occurrence. Therefore, a statistical interpretation of probability does not apply and statistical inferences cannot be made due to the sparcity of actual structural failure data. In such cases, analytical estimates of the failure probabilities may be obtained from stress-strength interference theory. Since the majority of real design applications are complex, numerical methods are required to obtain solutions. Monte Carlo simulation appears to be the best general numerical approach. However, meaning-ful applications of simulation methods suggest research activities in three categories: methods development, failure mode models development, and statistical data models development.

Some thoughts on the future of probabilistic structural design of nuclear components

This paper presents some views on the future role of probabilistic methods in the structural design of nuclear components. The existing deterministic design approach is discussed and compared to the probabilistic approach. Some of the objections to both deterministic and probabilistic design are listed. Extensive research and development activities are required to mature the probabilistic approach sufficiently to make it cost-effective and competitive with current deterministic design practices. The required research activities deal with probabilistic methods development, more realistic causal failure mode models development, and statistical data models development. A quasi-probabilistic structural design approach is recommended which accounts for the random error in the design models.

Truncation Errors in First-Order Reliability

In a second moment format for a probabilistic structural design code, it is often necessary to estimate the mean and standard deviation of stress and strength, which in turn may be complicated functions of several random variables. In a broader sense, probabilistic analyses in many disciplines require knowledge of the first and second moments of complicated functions of random variables. This study attempts to describe, under special conditions, the errors introduced in the most widely used approximate method for making this calculation.