PNNL–SA-35467

Mechanical Damage Characterization in Pipelines

P.D. Panetta
A.A. Diaz
R.A. Pappas
T.T. Taylor
R.B. Francini
K.I. Johnson

Pacific Northwest National Laboratory

Richland, WA 99352

October 2001

Prepared for the U.S. Department of Energy

National Energy Technology Laboratory

United States Department of Energy

under Contract DE-AC06-76RLO1830.
 Mechanical Damage Characterization in Pipelines

P. D. Panetta, A. A. Diaz, R. A. Pappas, T. T. Taylor, R. B. Francini, and K. I. Johnson

Pacific Northwest National Laboratory1
Richland, WA 99352

Introduction

The U.S. Department of Energy (DOE) National Energy Technology Laboratory (NETL)
has established the Natural Gas Infrastructure Reliability Program in order to provide
research and technology development for ensuring safe and efficient operational
reliability of the Nation’s natural gas transmission and distribution network. Maintaining
the highest levels of integrity and reliability regarding design, construction and operation
of the Nation’s natural gas infrastructure is of primary importance. Of particular
importance is the accurate prediction of the lifetime of damaged pipelines due to outside
force. In order to accurately predict the remaining life it is essential to accurately
determine the stress and strain in the damaged region. Currently there is a significant
technological gap inhibiting accurate diagnostics and prognostics for pipeline life
assessment. Specifically, there is no accurate quantitative connection between
nondestructive measurements and the state of stress and strain in a damaged region.

There are several categories of outside force incidents that can lead to the failure of
otherwise sound pipe. These include: 1) third-party mechanical damage, usually caused
by construction damage; 2) secondary loads imposed on a pipeline, usually due to soil
movement; and 3) dents due to the pipe resting on rocks. Third-party damage, sometimes
referred to as “dig-in’s” by the industry, accounts for the vast majority of outside force
incidents in pipelines.

Either immediate or delayed failure can occur due to mechanical damage. Immediate
failure occurs when construction equipment punctures the pipe and produces a leak at the
time of damage. However, mechanical damage more frequently provides an initiation
site for crack formation and eventual failure. Unreported damage can result in delayed
failure due to either slow crack growth through the thickness or hydrogen-stress cracking
of the cold worked and strain-aged steel. Recent failures in Edison, New Jersey and
Bellingham, Washington serve as examples of damage to pipelines that resulted in
delayed catastrophic failure.

Mechanical damage is normally divided into two categories, dents and gouges, which are
deformations in the wall of a pipe that serve as crack initiation sites. Dents typically
result from a purely radial deformation whereas a gouge has a component of deformation
along the surface of the pipe. For example, a pipe impinging on a rock will result in a
dent. If the pipe also slides on the rock a dent with a gouge will result. Third-party

1
Pacific Northwest National Laboratory operated by Battelle for the United States Department of Energy
under Contract DE-AC06-76RLO1830.

1
 PNNL–SA-35467

mechanical damage, usually caused during construction and excavation, is a common

cause of gouges. A gouge normally results in a highly deformed, work hardened surface
layer and may involve metal removal. Mechanical damage can result in immediate
failure of the pipe, delayed failure or no failure over the design life of the pipeline.
Distinguishing between damage that can result in either delayed failure or no failure is
the major concern of damage characterization. Immediate failures are of more concern
from a design standpoint. When predicting the remaining life of the pipeline, there are
two steps, the first is detecting and locating the damage and the second is characterizing
it. Detecting damage can routinely be performed with common nondestructive evaluation
(NDE) techniques. However, characterizing the degree of stress and strain in dents and
gouges is a particularly challenging task because the damage creates spatial and through
thickness stress and strain gradients.

The detection and analysis of mechanical damage has been an ongoing interest for the
pipeline industry for many years. Due to its brevity, this description is not meant to be
complete. It is meant to give an indication of the present state of research in the area, to
identify gaps in the application of NDE techniques to characterize damaged pipelines and
indicate how current research by the authors fit into this body of work. In the remainder
of the document the state of the art of traditional damage detection and characterization
methods will be discussed. This discussion will be followed by a discussion of the state
of the art of ultrasonic damage characterization and will highlight the limits of current
capabilities.

Detection of mechanical damage

At the present time In Line Inspection (ILI) tools are the most commonly used technique
to detect mechanical damage.1 In Line Inspection tools (or smart pigs) are instrumented
pigs that are placed in the pipeline and moved from one location to another along the line
by traveling with the product in the pipe. Almost any type of instrumentation can be
mounted on a pig. The most common techniques for pipeline integrity assessment are
electromagnetic, ultrasonic or dimension measurements. Magnetic Flux Leakage (MFL)
tools use an electromagnetic technique that has been shown to be sensitive to both
geometric and mechanical changes that result from mechanical damage. Caliper tools
measure the pipe radius at multiple locations around the circumference and have been
used to detect dents and estimate their geometry.2 Direct assessment is a surface based
technique primarily used to detect coating problems related to corrosion. Some direct
assessment methodologies are also capable of detecting mechanical damage, but they are
unable to distinguish damage from coating holidays.3 Some pigs use a ring of acoustic
sensors to measure the inside profile of the pipeline in liquid filled lines. Ultrasonic
methods have also been used to detect dents and cracks, however, implementation on
pigs is complicated because of the need to couple the sound to the pipe. Therefore,
measurements which utilize alternate forms of transduction are attractive such as, gas
coupling or the use of Electromagnetic Acoustic Transducers (EMATs), which require no
coupling.4

2
 PNNL–SA-35467

Regulatory requirements

Currently CFR 1925 requires the removal of dents in transmission pipelines operating at
pressure that results in a hoop stress of 20 percent or more of the Specified Minimum
Yield Strength (SMYS) if the dent contains a stress concentrator such as a scratch, gouge,
groove, or arc burn, or affects a weld. If the pipeline is operating at greater than 40
percent SMYS, there is the additional requirement that the dent must be removed if it is
more that 0.25 inch deep in pipe 12.75 inches or less in diameter; or more than 2 percent
of the nominal pipe diameter for pipe greater that 12.75 inches in diameter. The CFR
requirements only take the dent depth into consideration and in some cases these
regulations can be very general and inaccurate, and in most cases the pipe has rerounded
due to the internal pressure of the gas and this criteria is not conservative, especially if an
axial gouge is present.6,7,8 In other cases where the pipe has not rerounded, the criteria
may be very conservative. While these regulatory mechanisms are adequate in some
cases there is a need to have a more quantitative measure of the degree of damage to the
pipeline in order to predict the remaining life.

Qualitative Characterization of mechanical damage

Once mechanical damage is detected and located, the degree of degradation to the
pipeline properties must be assessed and the remaining life predicted. Currently there is a
significant technology gap with regard to the characterization of mechanical damage.
Specifically the quantitative determination of stress and strain to accurately provide
diagnostic and prognostic capability is needed to accurately predict the remaining life of
the damaged pipeline. Current techniques involve measurements obtained from pigs and
both pigging and pipeline companies have algorithms for characterizing mechanical
damage based on dent dimensions. The pigging companies usually have 2-3 levels of
defect severity that serve as a basis for digging up the defect.9 In some cases the pipeline
companies use the raw data from the pig runs in their own damage algorithms.
Ultimately, the final judgment is made by digging up the pipeline to visually inspect the
defect. Depending on the companies operating procedure, a subjective judgment based
on inspector experience, field measurements combined with damage algorithms are used
to assess the severity of the damage. Some of the measurements that are made include
dent geometry (length, depth, profile, wall thickness, etc.), gouge geometry, magnetic
particle, eddy current and ultrasonic NDE for cracking and hardness. This process is
highly inaccurate and can be subject to inspector bias. In addition, the dimensions do not
accurately represent the physical state of the damage because the shape changes due to
rerounding as the dent is pushed out due to the internal pressure of the gas.

The parameters used to calculate the structural integrity of a pipe with mechanical
damage are a subject of ongoing research. In most analysis methods, information on both
geometric changes (residual dent depth, amount of wall thinning and shape) and
mechanical changes (residual stresses, plastic deformation and cold working) are
important. Due to the complexity of mechanical damage, most damage algorithms are
empirically based. Two commonly used algorithms include one developed by the
Pipeline Research Council (PRC) of the American Gas Association (AGA) and one

3
 PNNL–SA-35467

developed by British Gas.10 Both of these algorithms include pipe geometry, defect
geometry and material properties to estimate the remaining strength of damaged pipe.
They tend to involve complicated equations with many parameters further decreasing the
accuracy.

The PRC model tends to give a lower bound for the remaining strength whereas the
British Gas model tends toward the mean of the remaining strengths and needs to be used
with a safety factor. More recent work has extended a Ductile Flaw Growth Model
developed by the Pipeline Research Council International (PRCI) from gouged pipe to
pipe with a dent and gouge.11 This elasto-plastic fracture mechanics based model yields
good predictions for both the remaining strength and the remaining life under cyclic
loadings. The difficulty with this model is that it requires knowledge of the residual
stress in the dented area, which is not readily available from field measurements.

The onset of failure in damaged pipe can be accurately predicted if the state of stress and
strain can be determined in a dented pipe and coupled with fracture mechanics based
models. Currently there is a technology gap to implementing this type of procedure;
specifically there are no current NDE techniques for accurate determination of stress and
strain in dented pipelines. Filling this gap is the primary objective of work currently
being performed by the authors for the NETL, Natural Gas Infrastructure Reliability
Program.

Ultrasonic Characterization of Mechanical Damage

Commonly used techniques for nondestructively determining the stress and strain in
damaged materials typically rely on ultrasonic velocity measurements. Ultrasonic
velocity is very sensitive to damage induced stress and strain and offers the potential for
accurate quantitative determination of stress and strain as inputs into fracture mechanics
models. This powerful combination offers a prognostic capability that is not currently
available to pipeline inspection companies and utilities because of the lack of ability to
accurately measure stress and strain. In addition, the main drawbacks include the fact
that velocity is affected by competing sources of velocity shifts due to microstructural
effects such as texture, temperature variations, the necessity for very precise time
measurements, and low spatial resolution. Despite these drawbacks, ultrasonic
measurements are valuable because they allow one to obtain information about the stress
and strain in the interior of the materials as a function of depth. Typical ultrasonic
measurement equipment is relatively inexpensive, portable, quick to set up, and the
ultrasonic velocity data is rich with information. For these reasons the use of ultrasonic
measurements to determine stress and strain has been an active area of research for many
years.

When considering applications to characterizing dents and gouges in pipes it is important

to realize that the stress and strain exhibit spatial variations and depth gradients. There
are several different ultrasonic methods that are appropriate for this application, including
through thickness measurements to characterize spatial gradients and waves with
penetration depths that can be varied to characterize depth gradients. Thompson et al. 12

4
 PNNL–SA-35467

thoroughly reviewed the past several decades of research that utilized ultrasonic velocity
measurements to characterize the stress and strain of materials. They reviewed the
considerable progress in developing the scientific foundation underlying the techniques
and highlighted specific results that have promise for various applications.

One such result that is promising for characterizing inhomogeneous through-plate stress
distributions is the work by King and Fortunko based on the velocity measurements of
horizontally polarized shear waves that were incident to the surface at shallow, grazing
angles.13 The theoretical development exploits the relative insensitivity of the grazing
shear wave propagation mode to texture and other microstructural anisotropies, compared
to other wave propagation modes including bulk and Rayleigh modes. Experimental
measurements were made of 25 mm thick 6061-T6 aluminum under a variety of
conditions including unstressed, under tensile stress, and under a four-point bending
stress. Using two EMATs separated by about 35 mm on the same surface, velocity
difference measurements were combined with the model to provide surface residual
stress values within 5% of the accepted values. These results help to establish the
fundamental scientific foundation and point to the need to improve spatial resolution for
practical applications.

One way to improve spatial resolution is by utilizing immersion measurements with

focused transducers to characterize areas on the order of millimeters by measuring the
Rayleigh wave velocity where the effective wave penetration depth is inversely related to
the frequency. Lavrentyev and Veronesi14 have shown that significant factors that
contribute to the surface wave velocity measurements include surface roughness, near-
surface grain orientation (texture), dislocation density changes, and residual stress. The
authors contend that the immersion method has the potential to provide the stress-depth
profile of a specimen with improved spatial resolution and a high degree of measurement
precision. The study found that in Waspaloy and Al-7075, the Rayleigh wave velocity
decreased with shot peening intensity, with a larger decrease observed at higher
frequencies. From a quantitative standpoint, the study concluded that the velocity
changes caused by stress are significantly smaller than those observed experimentally.
This translates into an important practical consequence that a near-surface residual stress
measurement in shot peened materials is difficult to achieve. One other drawback is that
this technique requires relatively sophisticated ultrasonic equipment and careful
measurement capabilities of the operator.

Results by Thompson et al.15 applied many different types of waves generated by EMATs
to accurately characterize the texture in aluminum and copper plates. Their experimental
results agreed well with theoretical predictions for these plates under stress free
conditions. In addition, the texture parameters obtained ultrasonically agreed well with
x-ray determinations. Thompson et al. presented the theory and applications of the
measurements of stress and texture in biaxially stressed specimens.16 Their
measurements agreed well with theoretical predictions and form the scientific basis for
characterizing biaxial stress states.

5
 PNNL–SA-35467

Initial measurements performed by the authors have shown that through thickness
measurements were strongly correlated with the strain in pipeline steels and experiments
were performed which indicate that the stress and texture may have separable
contributions to the velocity measurements. 17,18 Results were obtained while uniaxially
loading a pipeline steel sample, which showed systematic changes in the shear wave
birefringence during and after loading, which increased with increasing strain.
Compression tests were also performed on the pipe and showed systematic differences in
the birefringence between compression and tension that could easily be distinguished.17

The difficulty with numerically extracting the stress and strain from experimental data
further complicates the quantitative determination of stress and strain. Degtyar et al.
described a method for determining absolute stresses in an anisotropic material and
simultaneously determining the stress-dependent elastic constants from inversion of the
wave equation, where the bulk wave velocities as a function of their measured angular
dependence are used as inputs.19 The authors argue that this method is applicable for
stress measurements in materials that have undergone a complicated history (plastic
deformation) of loading and unloading, and can be used to determine both applied and
residual stresses. They presented experimental results that provide a means for
conducting high precision ultrasonic velocity measurements. In order to measure the
angular dependence of the ultrasonic bulk wave velocity, the authors used a self-
reference bulk wave (SRBW) method for determination of elastic moduli, where the
effects of geometric imperfections were significantly reduced.

Summary

While there has been much progress in qualitatively characterizing mechanical damage, a
significant technological gap inhibiting accurate diagnostics and prognostics for pipeline
life assessment exist that regulatory requirements do not alleviate. Specifically, there is
no accurate quantitative connection between nondestructive measurements and the state
of stress and strain in a damaged region. Ultrasonic measurements are promising for
characterizing damage in pipelines and several applications are currently being developed
or are already in place. However, considerable work needs to be undertaken in
improving the resolution of the techniques, in suppressing the influences of competing
microstructural effects, especially inhomogeneities in density and texture and in
understanding the effects of plastic deformation.

The large body of work in the field provides a strong possibility for providing
quantitative ultrasonic measurement methodologies for determining residual stresses and
strains in damaged pipelines. Furthermore, quantitative ultrasonic measurements could
be integrated with FEM models to completely characterization stress and strain fields of
damaged pipelines. In work being performed by the authors for the NETL Natural Gas
Infrastructure Reliability Program we are extracting the potentially practical
measurements from the science foundation to fill the technology gap that exists to
nondestructively characterize dents and gouges in pipelines.

6
 PNNL–SA-35467

References
1
T.A. Bubenik, J.B. Nestleroth, R.J. Davis, B.N. Leis, R.B. Francini, A. Crouch, S. Udpa, M.A.K. Afzal.
2000. “In-Line Inspection Technologies For Mechanical Damage And SCC In Pipelines - Final Report,”
Contract No. DTRS56-96-C-0010.
2
M.J. Rosenfeld. 1998. “Investigations of Dent Rerounding Behavior,” International Pipeline Conference-
Vol. 1. ASME.
3
T.A. Bubenik, B.N. Leis, D.A. Burgoon, S.W. Rust, E.B. Clark, K.C. Garrity, P.H. Veith, D. van
Oostendorp. 2000. “Direct Assessment And Validation,” GTI/GRI Report Number GRI-00-0231.
4
G.A. Alers and Hirotsugu Ogi. 2000. “EMAT Techniques for Elastic Constant Measurements.” Chapter
9 in Handbook of Elastic Properties, Vol II, eds. A. Every and W. Schase. Academic Press.
5
Pipeline Safety Regulations-Parts 190 thru 199, U.S. Department of Transportation, Research and Special
Programs Administration.
6
B.N. Leis and R.B. Francini. 1999. “Line Pipe Resistance to Outside Force Volume 2: Assessing
Serviceability of Mechanical Damage,” PRCI Catalog Number L51832.
7
R.J. Eiber, W.A. Maxey, C.W. Bert and C.M. McClure. 1981. “The Effects of Dents on the Failure
Characteristics of Line Pipe”, AGA Catalog Number L51403.
8
I. Corder and P. Chatain. 1995. “EPRG Recommendations for the Assessment of the Resistance of
Pipelines to External Damage,” EPRG/PRC 10th Biennial Joint Technical Meeting on Line Pipe Research.
9
For example see the following link for reporting and defect evaluation at PII:
http://www.pipetronix.de/in-line_inspection/analysis_reporting/analysisreport.html
10
R.J. Eiber and B.N. Leis. 1995. “Line Pipe Resistance to Outside Force,” EPRG/PRC 10th Biennial Joint
Technical Meeting on Line Pipe Research.
11
See Reference 2.
12
R.B. Thompson, W.-Y. Lu, and A.V. Clark, Jr. 1996. “Ultrasonic Methods.” Chapter 7 in Handbook of
Measurement of Residual Stresses. Eds. Dr. Jian Lu, pp. 149-178, The Fairmont Press, Inc., Lilburn,
Georgia.
13
R.B. King, and C.M. Fortunko. 1984. Surface-residual-stress evaluation using horizontally polarized
shear waves. J. Appl. Phys. 55(11), p. 3978.
14
A.I. Lavrentyev and W.A. Veronesi. 2001. “Ultrasonic characterization of shot peened metal surfaces,”
SPIE conference, undated.
15
R.B. Thompson, J.F. Smith, S.S. Lee, and G.C. Johnson. 1989. “A comparison of ultrasonic and x-ray
determinations of texture in thin Cu and Al plates,” Metallurgical Transactions A, Volume 20A, pp. 2431-
2447.
16
R.B. Thompson, S.S. Lee, and J.F. Smith. 1986. “Angular dependence of ultrasonic wave propagation in
a stressed, orthorhombic continuum: theory and application to the measurement of stress and texture.” J.
Acoust. Soc. Am. 80(3):921-931.
17
G.A. Alers and J.D. McColskey. 2001. “Measurement of residual stress in bent pipelines.” In Review of
Progress in Quantitative Nondestructive Evaluation, Vol. 21, eds .D.O. Thompson and D. I. Chimenti.
American Institute of Physics.
18
P.D. Panetta and G.A. Alers. 2000. “Characterization of plastically deformed steel utilizing ultrasonic
velocity and attenuation measurements.” In Review of Progress in Quantitative Nondestructive Evaluation,
Vol. 20, eds .D.O. Thompson and D. I. Chimenti. American Institute of Physics pp. 1494-1500. PNNL-
SA-33946, Pacific Northwest National Laboratory, Richland, WA.
19
A.D. Degtyar and S.I. Rokhlin. 1995. “Absolute stress determination in orthotropic
materials from angular dependences of ultrasonic velocities,” J. Appl. Phys. 78(3):1547-1556.