Page 105 - Reliability and Maintainability of In service Pipelines

94 Reliability and Maintainability of In-Service Pipelines

was developed that simulates degradation until failure under in-service loading
conditions. Simulated failure times were then fitted to a Weibull probability dis-
tribution, which allows the expected time to first failure to be calculated at differ-
ent locations along the pipeline. They found a reasonable agreement between pre-
dicted failure times and recorded failures for a period of 8 years.
Amirat et al. (2006) used reliability analysis to assess the effect of both the
residual stresses generated during manufacturing process and in-service corrosion
of underground steel pipes. During the service life of a pipe, residual stress relax-
ation occurs due to the loss of pipe thickness as material layers are consumed by
corrosion. First they focused on the influence of residual stresses in uncorroded
pipelines in order to identify the sensitivity of system parameters. In the second
step, a probabilistic-mechanical model was used to couple the residual stress
model with the corrosion model, in order to assess the aging effects through the
pipe service life. For long-term corrosion, the reliability analysis incorporated the
residual stress relaxation resulting from wall thickness losses. The probability of
failure of the pipeline was then evaluated for different corrosion rates varying
from the atmospheric baseline to very active corrosion processes.
DeSilva et al. (2006) presented a condition assessment and probabilistic analy-
sis to estimate failure rates in metallic pipelines. A level II FOSM analysis was
combined with condition assessment data to determine the probability of failure.
Davis and Marlow (2008) developed a physical probabilistic failure model for
service life prediction of CI pipelines subject to corrosion under internal pressure
and external loading. A limitation within their study was that the model only con-
sidered internal pressure and in-plane bending; therefore, the resulting failure
mode was only shown with a longitudinal fracture, which is just one type of sev-
eral failure modes. Other failure modes, such as circumferential fractures, were
not considered.
Moglia et al. (2008) looked at the exploration of a CI pipe failure model utiliz-
ing fracture mechanics of the pipe failure process. The first model generated was
simple, which allows explorations of additional model assumptions. Throughout
numerous assumptions, the model improved drastically. An elementary method,
FOSM, was initially used but proved to yield inaccurate results. A new approach
to the model evaluated the nominal tensile strength of pipes, which could deter-
mine the maximum corrosion defect. To account for the uncertainty or random-
ness within the data, a Weibull distribution was utilized adding stochasticity to
the corrosion rate. The proposed model calculated failure rates based on historical
data using a random Poisson statistical process. The maximum likelihood estima-
tor used within the Poisson distribution was used to calculate the failure rate of
the historical data sets. A case study was employed utilizing small diameter retic-
ulation mains. By modeling various assumptions into the simulated model, the
predicted and observed failure rates yielded similar results. Only failure modes by