Statistical Estimation of Tank Stress by Total System Analysis for a Spherical LNG Ship Tank

抄録

Stresses in a LNG tank, which are random variables caused by various types of random wave induced loadings, should be treated statistically.
The Square Root Method (SRM) is a conventional method for the estimation of maximum expected values of the tank stress resulting from these types of loadings. This method gives an accurate solution in a special case where the loadings are statistically independent and stress response is linear with regard to the loadings. However, there are, in general, certain correlations between wave induced loadings and stresses, such as Von Mises' equivalent stress and principal stresses, which are non-linear quantities with regard to the loadings.
This paper explains and discusses an analysis procedure called “Total System Analysis” (TSA) procedure, which has been developed to accurately estimate longterm statistical values of tank stresses. This procedure is an extension of the concept of the commonly-used statistical analysis of the long-term prediction of response of ship's motions and wave induced loadings; the procedure treats non-linear stresses as well as linear ordinary stress components.
In establishing the procedure specific items such as the introduction of stress response functions, the problems of non-linear stresses and the approximation of probability density functions of tank stress local maxima (or minima) to Rayleigh distributions, etc. are discussed. These items are verified by numerical experiments by a computer.
A Hitachi Zosen CBI type spherical tank on a 130, 000 cubic meter LNG carrier has been analyzed by the TSA procedure ; the results show that the correlations between types of wave induced loadings have a large influence on the tank stress, tending to reduce it, and that the non-linear stresses including the influence of correlations between the stress components are smaller than those without the above influence.