Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Doctor of Philosophy


Civil and Environmental Engineering


Zhou, Wenxing


Metal loss corrosions and dents are two major threats to the integrity of oil and natural gas pipelines. In the pipeline industry, the Fitness-For-Service (FFS) assessment is commonly employed for pipelines containing these defects. However, FFS assessment usually assumes that a defect has a simple shape, and such a simplification may significantly affect the accuracy of the assessment. Therefore, retaining the actual shapes of defects and incorporating them into the FFS assessment can improve assessment accuracy. The main objective of the present thesis is to extract key information about the sizes, directions, and shapes of corrosions and dents from the measurement of in-service and excavated pipelines, and then improve the accuracy of FFS assessment based on the extracted information.

The first study develops a wavelet transform-based denoising method for the measured inner surface of in-service dented pipelines obtained from caliper tools. Since the inner surface is differently sampled along the longitudinal and circumferential directions, the commonly used denoising methods cannot sufficiently remove measurement errors from the signal. The proposed method is based on overcomplete expansion, and the overcomplete dictionary is constructed from the hyperbolic wavelet transform and stationary transform. The strain estimated from the signal denoised by the proposed method is closer to the actual strain than the other denoising method. An overcomplete dictionary that can effectively denoise the dent signal is then constructed based on the statistics of the measurement of in-service dented pipelines.

The second study explores the vital directional features and length scales of natural corrosion clusters that govern the burst capacity of corroded pipelines. The corrosion depths in a cluster are measured by high-resolution laser scans, and two-dimensional (2D) discrete wavelet transform (DWT) with a suitable wavelet function is employed to decompose the corrosion cluster. A methodology is proposed to determine level- and sub-band-dependent thresholds such that those wavelet coefficients below the thresholds have a negligible impact on the burst capacity predicted by the widely used RSTRENG model and can be ignored for the reconstruction of the cluster. The preserved wavelet coefficients show that longitudinally orientated features with 4 – 32 mm in length have a greater influence on the remaining burst capacity than other features. This facilitates FFS assessment of corroded pipelines.

The third study aims to simulate the corrosion fields whose morphology and marginal distribution are close to the actual corrosion fields from limited information summarized from the ILI data. The corrosion field containing multiple corrosion anomalies is modelled as a nonhomogeneous non-Gaussian random field, where the spatial correlation and marginal distribution of anomalies are estimated from their sizes. The proposed methodology provides realizations of corrosion fields with the RSTRENG-predicted burst capacity closer to the actual burst capacity than the commonly used methodology that idealizes anomalies as cuboids.

The fourth study presents a framework to analyze and simulate nonhomogeneous non-Gaussian corrosion fields on the external surface of buried in-service pipelines by using continuous and discrete wavelet transforms. Continuous wavelet transform (CWT), dual-tree complex discrete wavelet transform (DT-CDWT), and dual-tree complex discrete wavelet with hyperbolic wavelet transform scheme (DT-CHWT) are incorporated into the iterative power and amplitude correction (IPAC) algorithm to extract the features of the natural corrosion field measured by a high-resolution laser scan and generate synthetic corrosion fields. The results indicate that the proposed framework can generate synthetic corrosion fields that effectively capture probabilistic characteristics of the measured corrosion field in terms of the scalogram, textural features, and burst capacity of the pipe segment containing the corrosion field.

Summary for Lay Audience

Pipelines are widely considered the most efficient and safest way to transport large quantities of oil and gas products over long distances. However, the integrity of pipelines is often threatened by corrosion and mechanical damage such as dents. The fitness-for-service (FFS) assessment is generally employed in the pipeline industry to assess a pipeline segment containing corrosions or dents. The characteristics of naturally occurring corrosions and dents are essential to the accuracy of FFS assessment, and these characteristics can be summarized from the measurement of the in-service and excavated pipelines. This thesis aims at improving the accuracy of FFS assessment models for pipelines containing corrosions or dents using the wavelet transform and random field.

The pipeline industry usually runs inline inspection (ILI) tools to detect and size defects along the in-service pipelines. After denoising the ILI signals, the sizes and locations of detected defects are incorporated into some FFS assessment models to determine if the pipeline can work safely with these defects. The present study involves different stages of this process. Some modifications are introduced to the wavelet transform-based denoising to better distinguish the actual shape of dents from the noise caused by measurement errors. The size and direction of the corrosion anomalies that significantly affect the burst capacity are extracted and summarized by the discrete wavelet transform. Based on the statistics of corrosion anomalies, the corrosion defects summarized by ILI tools are modelled as random fields rather than simple deterministic shapes. Introducing uncertainties will make the modelled corrosion defects similar to the actual corrosion defects, thus improving the accuracy of the burst capacity of corroded pipes. If the naturally corroded pipeline surface is detailly measured, a methodology incorporating wavelet transforms and random field modelling can capture the main characteristics of the measured surface and generate synthetic surfaces with these characteristics. These improvements to the different stages provide more accurate assessment results than the commonly used methods in the pipeline industry.