Abstract—The objective of this study is two-fold. The first is to define the significant risk factors pertinent to the corrosion and excavation failures in the natural gas pipeline industry and subsequently check for reliability of the data in hand. The second is to propose a statistical model to predict the odds of survival of the pipeline for the next ten years. The data for the proposed statistical model was obtained from the U.S. Department of Transportation-Pipeline & Hazardous Materials Safety Administration (PHMSA) database of pipeline failure over the past 25 years. A statistical model to predict probability of pipeline failure in the forthcoming decade was developed by processing the data through the JMP software to get the significant parameters contributing to failure. The proposed model predicts that the chance of survival of the pipe, according to the data evaluated, decreases gradually with age for the initial 40 years of the pipeline life and remains within acceptable limits till its nominal expected life. Between years 40 and 60, the survival rate is almost constant and just about at the threshold level, if one assumes a threshold of about 50% survival odds for the forthcoming ten years. The interesting take away from the trend observed is the increase in survival odds beyond the age of 60. The possible reasons contributing to this trend and a detailed analysis of the results based on the proposed model are discussed in the paper.
Index Terms—Pipeline failure, corrosion, excavation, multivariate regression model.
N. Hariharan and H.Sahin are with Texas A&M University, College Station, TX 77840, USA. (e-mail: narain110@tamu.edu, shnhakan@tamu.edu).
M. M. Cobanoglu is with BOTAS, Ankara, Turkey (e-mail: mmcobanoglu@gmail.com).
J. R. Menendez is with Fugro Roadware, USA (e-mail: jmenendez@fugro.com).
G. Saygili is with University of Texas at Tyler, USA (e-mail: gsaygili@uttyler.edu).
[PDF]
Cite: Narain Hariharan, Mustafa Murat Cobanoglu, Hakan Sahin, Jose Rafael Menendez, and Gökhan Saygili, "Multivariate Regression Model to Predict Failure of Pipelines," International Journal of Engineering and Technology vol. 9, no. 5, pp. 398-404, 2017.
