Axial cracks are a common site of failure for aging oil and gas pipelines. But how can operators know if a crack presents a risk of imminent failure or if it is safe to monitor and continue operations?

Pipeline operators rely on data from inline inspections and risk assessments to make decisions for pipeline repair, replacement and monitoring. Often, these decisions are made using limited information about crack length and depth and rules of thumb developed over decades. Until now, there has been limited empirical data about axial crack development to back up these recommendations. As a result, operators have relied on frequent testing, inspection, and often overly conservative analyses to guide repair and replacement decisions.

This leads to higher costs and, paradoxically, can compromise safety if not properly informed. If every repair is marked as equally urgent, resources will be wasted repairing stable or less severe cracks that could have waited while cracks presenting a higher risk of failure sit unattended.

Over the last several years, Battelle has conducted studies with the U.S. Department of Transportation (DOT) to better understand how axial cracks develop under different operating conditions. Optimizing both physics-based models and empirical data, researchers created a leading-edge tool to predict crack growth under different operating scenarios. Use of this tool, entitled PipeAssess PI, can help operators reduce reliance on frequent, unoptimized inline inspection and hydrostatic tests, as well as reduce maintenance costs by informing prioritization of repair and replacement decisions.

High Cost Of Cracks

More than 2.6 million miles of oil and gas transmission and distribution pipelines traverse the U.S., delivering trillions of cubic feet of natural gas and hundreds of billions of tonnes per mile of liquid petroleum products each year. All of these pipelines are aging; in fact, more than 45% of crude oil pipelines to date are more than 50 years old, and there are still pipelines in use that date to the 1920s and 1930s.

According to the U.S. Pipeline and Hazardous Materials Safety Administration (PHMSA), there were 6,335 reported pipeline incidents between 2005 and 2014, resulting in more than $5 billion in total damages and 152 fatalities. These failures result from a variety of causes, ranging from corrosion to cracking and accidental physical damage.

Axial cracks are a significant factor in many pipeline failures; material and weld failures were implicated in 36% of pipeline incidents between 2006 and 2010. This risk increases as pipelines continue to age. Vintage electric resistance welded (ERW) pipe prior to the 1970s, which is still used for many oil and gas transmission pipelines, is especially vulnerable to axial crack growth along the weld bondline as the asset ages.

Anatomy Of An Axial Crack

Axial cracks extend lengthwise along the pipe. Many of these cracks are invisible to the naked eye, but over time they can threaten the integrity of the pipe and raise the risk of pipeline failure. A small, shallow crack typically presents little to no risk to immediate operations. But once cracking has initiated, continued stress and strain from normal pipeline operation may cause the crack to continue to develop until failure occurs.

There are several ways that axial cracks can develop initially. Thermal and mechanical fatigue caused by cyclical operating pressures and cycles of freezing and thawing may weaken pipes and cause cracking over time.

Older ERW and flash-welded pipes are especially prone to this cracking and associated failures along the seam caused by thermal and mechanical stress. Those cracks that propagate in or along the seam include cold welds, hook cracks, and selective seam weld corrosion with cracking.

The nearby table summarizes these anomalies. Steel pipes are also susceptible to stress-corrosion cracking (SCC) fields, which can develop on the external side of the pipe when high-stress conditions combine with a highly corrosive environment. Hydrogen-induced cracking (HIC) develops when chemical compounds interact with the pipeline material or coatings, causing blisters to form, which promote crack growth. SCC and HIC may cross into the seam but exist over a cross section, and not just propagate along the seam bondline.

The timeline for development of a crack depends on several variables including crack geometry, crack depth, pipe material, material thickness, operating pressures and environmental conditions. Predictions of crack development must take all of these factors into account.

Testing And Inspection Limits

To reduce the risk of these failures, pipeline operators must monitor their assets for axial cracks and other flaws. This is generally done with a combination of hydrostatic testing, in-line inspection (ILI) and in-the-ditch non-destructive examination (NDE) characterization techniques.

Hydrostatic testing can help detect near-critical axial cracks and other near-critical defects.

In a hydrostatic test, pipelines are tested by pressuring them with water, typically to pressures above their maximum allowable operating pressure (MAOP) for at least 10 minutes or up to 24 hours. The pipe is tested for leaks, crack development and plastic deformation under the higher pressure. Known axial cracks within the pipeline are inspected to determine if the higher pressure causes accelerated development; if no change occurs in the crack during the test, the pipeline is assumed to be safe for continued operation. Periodic hydrostatic testing of certain pipelines is required under PHMSA’s 49 CFR 192 and 195 regulations.

Hydrotesting is useful but expensive and poses environmental and safety challenges. Pipeline operations must be shut down during testing and vast volumes of water must be properly treated and disposed. Hydrotests can be used to monitor known existing flaws and other defects such as SCC and active corrosion cells.

However, it will not detect sub-critical material flaws that remain stable during the hydrotest. At the same time, hydrostatic testing puts additional stress on these undetected, resident flaws that can cause them to grow during testing, which may lead to less time to instability post-hydrotest.

ILI is performed using sensors that are sent through the pipeline with the product. These tests can detect and characterize new flaws. Traditional ILI tools for cracks use Electromagnetic Acoustic Transducer (EMAT) or ultrasonic (UT) technologies for flaw sizing and detection.

These tools can report crack length and maximum crack depth. Newer, more advanced ILI technologies can also measure the defect’s depth profile. Advanced magnetic flux leakage (MFL) ILI tools can provide additional information that can also be useful for integrity assessments such as material binning and reviewing EMAT calls to designate non-injurious features. With respect to material binning, these ILI tools have begun to non-destructively characterize pipe material by noting the differences that may be tied to a change in pipe batch, manufacture, grade, seam type and other material properties.

In-the-ditch NDE characterization techniques can add to the information gathered by ILI and are often utilized for verification of ILI crack sizing. These NDE examination methods can also characterize material properties and report estimates of tensile strength, ultimate strength and/or pipe grade. NDE and ILI can together provide a robust set of data regarding both crack and pipeline characteristics.

Predictive Problem

None of these testing and inspection technologies are predictive, however; it is not always clear whether a detected crack warrants immediate repair or replacement or simply requires continued monitoring. Pipeline operators must use rules of thumb and vintage models, which are often excessively conservative and/or severely limited in applicability, to interpret the data, determine safe operating pressures and predict how soon a given pipeline flaw will require attention.

The consequences of misjudging these operating decisions can range from merely expensive to potentially catastrophic. A pipeline flaw that develops faster than anticipated can lead to pipeline failures resulting in expensive shutdowns, environmental cleanup operations, and potential litigation. Operators may, therefore, rely on frequent hydrostatic testing to monitor crack development. This is expensive, and may paradoxically accelerate development of some pipeline flaws.

Initiating a dig for inspection, repair or replacement is also expensive. Average costs range from $25,000 to $50,000 per dig. Without accurate prediction of crack development, pipeline operators often spend more than they need to on unnecessary digs and unnecessarily frequent inspections and hydrostatic testing.

Building A Better Model

Operators rely on available software tools to model anomaly failure and growth for oil and gas pipelines. While there are now several modeling software options to choose from, few claim to be able to probabilistically predict anomaly failure or growth of any type, and none of the tools currently in common provide robust prediction of crack development under different operating conditions based on physics and empirical data.

This lack of a comprehensive, physics-based model has been a limiting factor for prediction tools of pipeline crack development. Models are only as good as the algorithms or historic data behind them; they cannot make accurate predictions without a clear understanding of the physics involved and the variables that influence crack development.

Battelle conducted empirical studies as part of a DOT-PHMSA project, “Comprehensive Study to Understand Longitudinal ERW Seam Failures.” The goals of the study were to determine the characteristics of ERW seams that make them susceptible to failure and identify the factors that pipeline operators should consider in order to assure that the operation of ERW pipelines is safe. Ultimately, DOT wanted a better integrity management tool to understand how axial cracks impact pipelines and help pipeline operators provide more effective recommendations for pipeline management and risk reduction.

Battelle researchers gathered empirical data using Battelle’s and publicly available databases of oil and gas distribution and transmission pipes. Pipes with known defects were also evaluated using ILI and in-the-ditch technologies as part of the program. Following ILI, pipe sections were then pressurized to failure.

This comprehensive dataset was used to develop a more accurate model of axial crack development for various pipe types—including ERW and FW pipes made of brittle, quasi-brittle and ductile steels—under different operating conditions. Battelle then developed a modeling program to analyze the collected data and build predictive models of pipeline failure.

The PipeAssess PI software allows operators to model stress due to internal pressure with various crack geometries, including cold welds, hook cracks and selective seam weld corrosion with cracks. The software incorporates user-defined hydrostatic test conditions, operating pressure cycles and attributes such as pipe size, material properties and crack geometry (from in-line inspection and/or in-the-ditch non-destructive examination).

It uses Monte Carlo simulation and probabilistic techniques to predict crack growth and ultimate pipeline failure. The model considers multiple mechanisms for growth, including time-dependent cracking during hydrotests as well as fatigue cracking due to pressure cycles. The crack growth physics utilize J-tearing theory for elastic-plastic material behavior and Paris Law behavior accounting for stress-ratio variations and overloads (i.e. hydrotests) for fatigue crack growth analyses. The material constitutive model for time-dependent crack growth under simulated hydrostatic test conditions is the Ramberg-Osgood stress-strain model.

These material and fracture mechanics models, along with state-of-the-art stress-intensity factor solutions, provide highly accurate prediction of future crack development and remaining pipeline life.

A Case Study

A 2016 study by Battelle and T.D. Williamson (TDW) compared pipeline management decisions made using data from traditional ILI technologies to decisions with advanced inspection techniques and leading edge PipeAssess PI to model and predict future crack development. The study showed that using more sophisticated modeling and inspection tools allows pipeline operators to better prioritize inspection intervals and remedial actions, significantly reducing costs.

This study analyzed 34 crack-like seam defects using different ILI and in-the-ditch technologies. All pipes were low-frequency electric resistance welded (LF-ERW) pipe, circa 1950.

The Battelle-TDW study evaluated the data collected from three levels of inspection technologies. Each level was designed to mimic a generic operator’s approach to assessing each crack’s criticality and prioritizing remediation efforts using the known information. With each increasing level came additional material and flaw characterization from ILI and/or NDE:

  • Level 1 – Traditional EMAT ILI;
  • Level 2 – Advanced ILI crack-fleet (i.e. multiple dataset platform [MDS] and EMAT); and
  • Level 3 – Advanced ILI fleet (i.e. MDS and EMAT) and in-the-ditch NDE (i.e. ABI, OES and advanced UT).

Researchers analyzed 34 axial seam cracks with known crack depth profiles and material properties using all three levels of technology. The data was then analyzed using Battelle’s PipeAssess PI software to determine which pipes would be flagged as Priority 1 (immediate action), Priority 2 (further assessment) or Priority 3 (monitoring only) using standard industry best practices.

Different levels of data collection turned out to provide very different recommendations for pipeline management.

The Level 1 assessment method, which simply screened for crack criticality, flagged 97% of all reported flaws as potentially requiring immediate remediation. With an advanced ILI crack tool run (MDS and EMAT), the number of cracks requiring immediate attention was drastically reduced to 12%. Under the most advanced Level 3 assessment, zero cracks were identified as Priority 1, four cracks were identified as Priority 2, and all remaining flaws were classified as Priority 3.

Of the four cracks requiring action, all but one had been identified as critical in previous inspections.

The study showed that operators relying solely on Level 1 assessment data would have initiated 23 more digs than operators using more accurate Level 3 assessment data—with huge financial implications. The results suggest that using better data and more accurate modeling would allow pipeline operators to control costs by optimizing inspection intervals and reducing unnecessary digs.

The accompanying graphic estimates the cost difference for pipeline management decisions made with each level of assessment.

Meeting DOT Requirements

Predictive modeling can help operators improve safety and reduce costs while meeting DOT requirements for assessment of liquid and gas pipelines. The DOT-PHMSA Notice of Proposed Rulemaking (NPRM) Part 192 - §192.607 proposes that all transmission lines have material records that are “verifiable, traceable and complete” in designated High Consequence Areas (HCAs). More accurate modeling will help operators meet existing and proposed rules for pipeline integrity management while controlling the costs and risks associated with repeated hydrostatic testing, ILI and NDE.

Using predictive modeling, operators can reduce reliance on hydrostatic testing and other inspection methods without reducing pipeline safety. Accurate, physics-based models allow companies to better prioritize replacement and repair decisions and direct resources to pipeline flaws that present the most immediate risks first.

Modeling also improves safety by helping operators better understand the risk implications of re-rating MOAP for existing pipelines that will ensure pipeline safety until repairs or replacements can be made.

PipeAssess PI is already in use by industry and is available commercially for pipeline owners, operators and consultants. This predictive modeling software may be used for engineering critical assessments (ECAs) consistent with the requirements of current and pending PHMSA rules, including Code of Federal Regulations (CFR) 49 parts 192 (gas “mega-rule”) and 195 (liquid rule).

Battelle is continuing to refine the models and expand the capability of the software to analyze other mechanisms such as external corrosion. Pending funding, stress corrosion cracking, internal corrosion, and plain denting will also be added.

Researchers are now working on refinements that will allow the software to analyze a large set of anomalies between pump and compressor stations. As physics-based predictive modeling continues to improve, it is likely to become a critical part of the toolbox for completion of ECAs and compliance with PHMSA rules for pipeline integrity management.

Bruce A. Young and Jennifer M. O’Brian are with the Battelle Memorial Institute.