Page 183 - Reliability and Maintainability of In service Pipelines

Case Studies on the Application of Structural Reliability 167

5.4.4.2 Sensitivity Analysis

As it was mentioned earlier, for a comprehensive pipeline assessment, the effect
of variables on the failure of the concrete sewer can be analyzed by performing
sensitivity analysis. In view of the large number of variables that affect the corro-
sion process, and hence the limit state functions, it is of interest to identify those
variables that affect the failure most so that more research can focus on those
variables.
Unlike individual failure mode assessment (Section 5.4.1), the concept of rela-
tive contribution can not be used for multifailure mode analysis. It is simply
because Eq. (3.9) has been presented for calculation of relative contribution in
case of individual limit state function. Hence, a new parametric method is devel-
oped and applied for sensitivity analysis of the concrete sewer in case of multifai-
lure mode assessment.
To assess how the change in the values of the six random variables (k, u, j,

[DS], b=P , and A) can affect the service life of the concrete sewer system, the
values for each variable are changed from μ 2 2σ i to μ 1 2σ i (where μ is the
i
i
i
mean of the random variable and σ i is its standard deviation). Assuming a
Gaussian distribution for the random variables, this range corresponds to 95.4 per-
cent of the possible values of the variable. The results of the analysis by using
Monte Carlo simulation method are illustrated in Fig. 5.37(a f).
It can be concluded that among all variables, the effect of [DS], A, and b=P
on the probability of failure of the sewer is highly remarkable. The disparity
shown within the graphs for each of these three variables means that the sensitiv-
ity of the failure of the pipeline is more dependent on their actual value. In such
cases, more concern should be taken in order to determine relevant parameter
values.
For a better comparison of the effectiveness of random variables on the ser-
vice life of the pipeline, the results in are summarized in the form of Fig. 5.38.In
this figure, a range value is defined for each random variable as the difference
between the maximum and minimum values of the probability of failure in each
elapsed time. Therefore a higher range value means wider resultant values for the
probability of failures. This figure illustrates which variables contribute most to
the probability of the failure of the system. To clarify, this means that the vari-
ables have more effect on the service life of the pipeline.

The significance of the three major variables (i.e., [DS], b=P; and A) on the
failure of the concrete sewer has also been concluded in individual failure mode
analysis in Section 5.4.1.2.
Of these three major random variables, the analysis shows the particular sig-
nificance of dissolved sulfide concentration in the probability of failure of the
pipeline, indicating that dissolved sulfide concentration is the most significant
variable on the reliability of concrete sewers.