PREDICTIVE MAINTENANCE OF PIPELINES WITH DIFFERENT TYPES OF DEFECTS

Аннотация

The paper describes a tested and proven practical methodology of predictive maintenance of pipelines with two types of defects—«loss of metal» and «pipe wall lamination», detected by the ILI technology.

For the defects of the «pipe wall lamination» type the assessment of their level of danger is conducted only after they are converted to surface «loss of metal» type defects. The paper presents models on how to adequately convert the «pipe wall lamination» type of defects to the «loss of metal» type defects.

A methodology is described on how to rank the defects according to their level of danger (with respect to the rupture type of failure), and how to perform the probabilistic assessment of the residual life of the inspected pipeline. The defects detected by the ILI are divided, depending on their type, size, and the level of safety factor, into three following categories: Dangerous, Potentially dangerous and Not dangerous defects. 

In order to account for «leak» and «rupture» types of failure, a computer based express assessment is developed of the level of severity of each defect. This defect assessment is based on graphs, which restrict the permissible sizes of defects and allow making operative decisions as to which maintenance measures should be taken, regarding each detected defect and the pipeline segment as a whole. The pipeline defects are ranked according to their potential danger, which depends on their location on the graphs. These graphs form five zones, which define the level of the defects danger.

The probabilistic assessment of the residual pipeline life is performed taking into account the stochastic nature of defect growth. In order to achieve this, the maximal γ-percentile corrosion rate is defined over all detected defects. The distribution of the n detected pipeline defects is described by the two-parameter Weibull probability density function (PDF).  As the main decision parameter the gamma-percent operating time is chosen. It is characterized by 1) the safe operating time, and 2) the percentile probability that during this time the pipeline limit state will not be reached.

A detailed example of implementation of the described methodology to a real product pipeline segment operating in a severe corrosion environment is given. The economical effect of the implementation is outlined.