Page 498 - Marine Structural Design
P. 498

474                                                   Part IV Structural Reliability

                 This  implies  that  provided  the  method  of  nonlinear  analysis is  sufficiently accurate,  the
                 variability in system strength is less than 10% for ductile systems.
                 By introducing those considered uncertainties into LSF, the LSF can be re-expressed simply as
                      g(z)=x,x:x,x,   -4%                                           (26.21)
                                                                                    (26.22)
                      g(z)=x,x:x,x,x, -6s
                 The reliability analysis is based on the probabilistic data given in Table 26. I.
                          Table 26.1  Basic Probabilisti  'arameter Descriptions
                           Random Variable           Distribution   Mean    COV
                           Model uncertainty, X,     Lognormal      1.16    0.138
                           Yield strength uncertainty, X,   Lognormal   1.14   0.04
                           Diameter uncertainty, XD   Normal        1.02    0.02
                           Thickness uncertainty, XT   Normal       1.04    0.02
                           Load uncertainty, &       Lognormal      0.90    0.08
                           Strength uncertainty, XR   Lognormal     1.05    0.05



                 26.5.5  Target Safety Levels
                 When  carrying out  structural reliability analysis, an  appropriate safety should be  selected
                 based on factors like consequence of failure, relevant rules, access to inspection and repair,
                 etc., which is termed as target safety level. Target safety levels have to be met in design in
                 order to ensure that certain safety levels are achieved.
                 Any evaluation of safety levels should be based on information about safety level implied by
                 the design codes and component with historical data on reported failure. The safety level of
                 existing tubular joints designed according to traditional procedures may be a good reference
                 for the target level if the reliability is on average satisfactory. It is important to state that this is
                 related to average failure rate only, as there is expected to be a large variability in the real
                 safety from  one  tubular joint  to  another, due  to  differences and  shortcomings in  design
                 practices in the past. The target safety level should hrther be related to the consequences of
                 failure modes as well as the nature of failure, and it may be found that the target reliability
                 level should be increased or even could be decreased concerning specific failure modes.
                 A target safety level should normally reflect the consequences of failure, safety phiIosophy,
                 access to inspection and repair, and behavior of the structural components. Safety classes are
                 generally based  on mainly the  consequences and types of  failure, which can  be generally
                 divided into low,  normal  and high safety class depending on the considered platform and
                 components.
                 Low safety class: where failure of component or tubular joint implies no risk to human safety
                 and environmental damage. When a certain damage is found in this class, its condition can be
                 monitored and no other necessary measures needs to be applied.
   493   494   495   496   497   498   499   500   501   502   503