Optimal maintenance policies for degrading hydrocarbon pipelines using Markov decision process.

The random nature of deterioration of piping systems used to transport hydrocarbons, is identified by periodic or continuous inspection. At a minimum threshold t(min) of pipe wall thickness, the pipeline is in a failed state, with a leak or pipe rapture imminent. We modeled the maintenance policy of a pipeline, based on an infinite horizon Markov decision process, taking into consideration the current state of the pipe, the action performed and the transition probabilities from one state to the other. This provides an optimal decision-making strategy or policy to minimize total maintenance cost.