Research Of Corrosion Prediction For Buried Gas Steel Pipeline Based On Markov Theory

Along with buried gas pipeline service time, the accidents caused by corrosion failure not only affect normal operation, but also cause tremendous waste of energy and economic losses. Especially in city gas pipeline, the accidents badly endanger country common property and the safety of the residents. Therefore, carrying on research of corrosion prediction for buried gas pipeline is meaningful to reliable management. Simultaneously, determining reasonable maintenance about different corrosion grades of gas pipeline can be used to decrease the maintenance cost.In this paper, the research work is corrosion prediction and maintenance decision-making for buried gas steel pipeline. Makov theory is introduced into the research of application in buried gas steel pipeline corrosion process. Markov theory includes Markov chain and Markov decision process. The former is a prediction method for stochastic processes, which can be used to predict corrosion status for pipeline. The latter is a method for decision-making, which can be used to find an optimal solution in the cost of pipeline maintenance. Corrosion status for gas steel pipeline is divided into external anticorrosion coating aging condition and pipe wall corrosion condition.First, according to safety quality status evaluation for external anticorrosion coating, the aging states for anticorrosion coating are divided. The prediction models are established, including a single pipe and the entire pipeline, when Markov chain is applied to predict aging conditions for anticorrosion coating. In a singe pipe coating prediction model, three cases are discussed, including multiple times detection, few times detection and only one time detection. The concrete prediction processes are also given in these cases. In practical example of the last case, the transition probability matrix is confirmed, and the development conditions for anticorrosion coating are predicted. For the entire pipeline anticorrosion coating conditions, the prediction process and practical example are given, which can determine development conditions of states distribution for anticorrosion coating.Secondly, according to evalution for pipe wall corrosion damage, the corrosion states for pipe wall are divided. The prediction models are established, including a single pipe and the entire pipeline, when Markov chain is applied to predict corrosion conditions for pipe wall. In a singe pipe wall corrosion prediction model, the transition probability matrix is confirmed by curve fitting method, and a practical example is given for predicting residual wall thickness. In the entire pipeline wall corrosion prediction model, the transition probability matrix is confirmed by grey prediction method, and a practical example is given for predicting maximum corrosion depth.Finally, corrosion states for gas steel pipeline in maintenance process are determined. The maintenance measures and related cost for gas steel pipeline are also analyzed. By applying Markov decision process, three solving methods are elaborated, including steady-state probability, a policy improvement algorithm and dynamic programming. Taking gas pipeline in Dalian for example, Markov decision process is applied to find the optimal strategy for maintenance. There are four aspects discussed in detail, including corrosion states, maintenance measures and cost, transition probability matrices in different strategies, and solving the problem by a policy improvement algorithm. The optimal strategy can provide some directional suggestions for buried gas pipeline repair, maintenance and replacement.