Date of Award

Summer 2019

Document Type

Open Access Dissertation

Department

Civil and Environmental Engineering

First Advisor

M. Hanif Chaudhry

Abstract

The crude and refined oil, water, and other liquids are widely transported through pipelines over long distances within geographical boundaries of countries or beyond. Pipeline accidents, however, are frequent, affecting operation and reducing targeted performance. The frequency of these accidents presents a growing concern and necessitates the need for an effective health monitoring plan. This plan requires monitoring of flow conditions and warning of any changes or abnormal conditions. Abnormalities in the flow conditions include changes in the piping system performance due to defects, such as leaks and blockages. These changes can be located and quantified utilizing different approaches. Two effective and economical approaches are: i) observation of pressure oscillations and, ii) acoustic signal monitoring. This may be conducted by comparing the data set for an intact pipe and a pipe with a fault to diagnose the defect and prepare an appropriate course of action.

This work develops two practical approaches to detect abnormal conditions caused by a defect in the pipeline. In the first approach, the transfer matrix method is applied to detect a partial blockage or a leak in liquid pipelines using the first four harmonics of the pressure oscillations produced by a sinusoidal movement of a downstream valve. Unlike most relevant available methods which use a large number of harmonics to analyze the pattern of the frequency response diagram, this study uses only the first four harmonics to investigate the effect of a leak or a partial blockage on the amplitude of pressure head oscillations. A relationship between the location of the blockage or the leak and the amplitude of pressure head oscillation is developed for each lower harmonic in the steady-oscillatory flow. The effects of steady and unsteady friction and pressure head oscillation nodes are discussed. The results show a satisfactory agreement with those obtained in the time domain using the method of characteristics and with the experimental data reported in the literature.

In the second approach, a primary zone representing the inspected length of the pipe is scanned for potential defects by installing a set of acoustic sensors on the exterior surface of the wall of the pipe. The acoustic signal emitted due to the interaction between the defect and the liquid flow is recorded and analyzed. The characteristics of this signal depend on the location and size of the defect. For instance, the existence of a partial blockage causes a hump in the plot of the relationship between the accumulated signal strength and the sensor location, the height of which is proportional to the blockage size. Following this approach, an application is developed to detect a partial blockage in a simple reservoir-pipe-valve system. The experimental results are verified by computational fluid dynamics (CFD) simulations including solution of the Reynolds-averaged Navier-Stokes equations using the FLUENT commercial package within the ANSYS software. The numerical results show a spike in the plot of the relationship between the sensor location and a number of selected flow parameters in the vicinity of the blockage location in an agreement with the experimental results.

In summary, the two approaches proposed in the current study represent simple techniques that may be used for detection of defects in short or long pipelines. The first approach including measurements of pressure head oscillations at one location may be used in the case of limited access to the entire length of the pipeline, while the second approach including monitoring of the acoustic signal may be used for short pipes with a full access to the entire length of the pipe. The two approaches may be extended to detect defects in pipelines including flow of different fluids, such as gas and steam.

Share

COinS