BLOCKAGE DETECTION IN NATURAL GAS PIPELINES BY TRANSIENT ANALYSIS

Open Access
Author:
ADELEKE, NAJEEM A.
Graduate Program:
Petroleum and Mineral Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
April 12, 2010
Committee Members:
  • Michael Adebola Adewumi, Thesis Advisor
  • Mku Thaddeus Ityokumbul, Thesis Advisor
Keywords:
  • Transient Model
  • Blockage
  • Transient analysis
  • Numerical modelling
  • Hydrate
  • Pipelines
  • Natural Gas
  • FVM
  • Finite Volume Method
Abstract:
Pipelines are the most reliable means for the transportation of natural gas. A major problem of flow assurance in natural gas pipelines is solid deposition that result in partial/complete blockages. Blockages in natural gas pipelines are mostly due to hydrate formation and deposition. The industry has adopted a handful of hydrate formation prevention techniques however there is still no known means for the complete eradication of hydrates that is applicable under all pipeline blockage scenarios. The industry is now resorting to early blockage detection techniques in order to appropriately manage blockage scenarios for economic and safety reasons. Several studies have been conducted to determine the best approach for early blockage detection. Modeling and analyzing natural gas transients for blockage characterization is one of the promising early blockage detection techniques. It is the most economical and least intrusive simply because it requires no additional instrument other than a dynamic pressure gauge which is usually already part of most modern pipeline networks. However, the problem with this technique lies in the resulting mathematical model. Its formulation results in a system of non-linear hyperbolic partial differential equations which have no known generalized solution. Besides, numerical techniques for solving the resulting mathematical model are computationally involved and are subject to numerical stability issues. This study explores the possibility of the use of a simple numerical technique based on finite volume method for blockage characterization. In previous studies, pressure waves were assumed to be propagating at the speed of sound. This assumption is incorporated into the mathematical formulation by the isentropic assumption where the the pressure term within the momentum conservations flux is substituted for a function of the speed of sound. A calculated estimate of the speed of sound is then assigned prior to conducting numerical experiments. This assumption idealizes the model and causes the resulting pressure waves to propagate at sonic speed. The idealized solution and thereby affects the blockage characterization capability of the model when applied to real pipeline blockages or lab flow loop experiments. In this study, we do not make this assumption and instead the pressure one of our unknown variables. Hence the compression wave is expected to travels at its true speed. Additionally, previous studies did not include viscous effects in their mathematical model for blockage characterization (Ahmed, 1996; Eltohami, 1999; Adewumi et al., 2000 and 2003; Chen et al., 2007). This is another assumption of ideality which further makes the solution impractical and reduces the accuracy of blockage characterization analysis. This study evaluates the effect of neglecting viscous effects on blockage characterization and a preliminary alternative equation that accounts for the effect of friction on blockage severity estimation is proposed. Furthermore, previous studies (Adewumi et al., 2000 and 2003) utilized empirical formulations (Dranchuk and Abou-Kassem, 1974) for the estimation of the compressibility factor. In this study, the mathematical formulation is a quasi-compositional Eulerian gas flow model in which the compressibility factor is estimated using the Peng-Robinson equation of state. This provides a more realistic prediction of transport properties since the composition of the gas is put into direct consideration. The numerical techniques implemented here are specialized the for finite volume method of discretization. A staggered three-point stencil upwind scheme and a second-order centered five-point Nessyahu and Tadmor TVD scheme with MUSCL reconstruction are solved implicitly using the Newton-Raphson iterative technique. The fully implicit approach offers model stability for relatively large time steps thereby reducing the overall computational time despite the use of an iterative solver. Blockage location prediction error was found to be reduced by one order of magnitude if the actual speed of the pressure wave that is determined from the inlet pressure profile is used. If viscous effects are considered, blockage severity analysis using the linear theory equation for determining wave reflection ratio that does not account for viscous effects results in large blockage severity prediction errors. However, blockage severity prediction is improved when the analysis is made in consideration of viscous effects.