Content uploaded by N.T. Switzner
Author content
All content in this area was uploaded by N.T. Switzner on Feb 25, 2020
Content may be subject to copyright.
Influence of line pipe steel
microstructure on NDE yield
strength predictive capabilities
by
Nathan Switznera, Solver Thorssonb, Jeffrey Kornutab, Peter Veloob, Peter
Martinc, Troy Rovellac, Michael Rosenfelda
a
RSI Pipeline Solutions,
b
Exponent,
c
Pacific Gas and Electric
Pipeline Pigging and Integrity
Management Conference
Marriott Marquis Hotel, Houston, USA
February 17-21, 2020
Organized by
Clarion Technical Conferences and Great Southern Press
Pipeline pigging and integrity management conference, Houston, February 2020
2
Proceedings of the 2020 Pipeline Pigging and Integrity Management conference. Copyright ©2020 by Clarion Technical Conferences,
Great Southern Press and the author(s).
All rights reserved. This document may not be reproduced in any form without permission from the copyright owners.
Pipeline pigging and integrity management conference, Houston, February 2020
3
Abstract
Recent changes to federal rules governing the operation of natural gas pipelines allow operators to
use nondestructive examination (NDE) technologies for materials verification. Since 2013, PG&E has
been evaluating instrumented indentation testing (IIT) to estimate yield strength (YS) using a
qualification sample set of ~100 line pipe features. Comparisons of NDE YS estimates with tensile test
YS results have revealed differing trends based on pipe vintage and manufacturing process. For
example, two steel samples may exhibit similar YS estimates from IIT but different results for YS from
tensile testing. IIT algorithms rely on empirical relationships that are based on the strain-hardening
exponent to estimate YS. These empirical relationships may underrepresent the influence of
microstructure. This influence may be a contributing factor to differences observed between laboratory
tensile test YS and IIT estimates of YS.
Pipeline manufacturing standards, e.g., API 5L, do not control microstructure. Therefore, Operators
are likely to order line pipe to meet composition and mechanical properties specifications, with little
regard for microstructure. Microstructure, however, provides key insight about the deformation
behavior and strengthening mechanisms of the steel, which then governs the strain-hardening rate.
Metallography and microscopy were employed to reveal the influence of microstructure on NDE YS.
Preliminary analysis on a subset of validation samples indicates that IIT predictive capabilities can be
categorized based on sample microstructure.
Introduction
A nondestructive method to accurately estimate line pipe yield strength (YS) would be valuable to
pipeline operators to defray costs, delays, and risks associated with extracting samples from their
pipeline assets for destructive laboratory testing. Newly adopted federal rules governing the operation
of natural gas pipelines allow operators to use nondestructive examination (NDE) technologies for
materials verification. PG&E has been evaluating the NDE technologies including instrumented
indentation testing (IIT) since 2013 for estimating steel line pipe strength. The IIT estimate of YS is
based on the relationship between the indentation response of a material and its stress-strain curve.1
During an IIT measurement, an indenter made of a hard material (tungsten carbide) is sequentially
forced into the softer steel and the travel depth is measured. An algorithm based on Finite Element
Analysis (FEA) is used to relate the indentation response and the material stress-strain curve.
The results generated by PG&E show that IIT estimates of YS for line pipe steels relate moderately
well to YS estimates from tensile tests.2 Figure 1 shows the evaluation data presented by Kornuta et
al. (2019) for average IIT estimates of YS versus average YS from tensile tests. The results reveal a
promising trend of increasing IIT estimate with increasing destructive YS with an R2 value of 0.75 for
the linear regression fit. PG&E utilizes IIT YS in probabilistic MAOP reconfirmation calculations. The
probabilistic calculations retain the precision and accuracy characteristics determined during
validation. Improvements of the IIT estimate of YS, i.e., improving IIT YS prediction precision and
accuracy, will increase the certainty of probabilistic MAOP reconfirmation results. Note that other
1 Kim, J.Y., Lee, K.W., Lee, J.S., Kwon, D., “Determination of tensile properties by instrumented indentation
technique: Representative stress and strain approach,” Surface & Coatings Technology, 201, 2006, pp. 4278-4283.
2 Kornuta, J.A., Martin, P., Louie, M.W., Ames, N.M., Rovella, T., Veloo, P., “An evaluation of instrumented
indentation testing to estimate yield and tensile strength”, Paper #60, The 31st International Pipeline Pigging and
Integrity Management Conference (PPIM), Houston, TX, USA, February, 2019.
Pipeline pigging and integrity management conference, Houston, February 2020
4
researchers identified relatively high scatter (-7% to +16%) for line pipe steels.1 One possible reason for
the scatter is that the algorithm is based on a wide variety of materials, including aluminium alloy,
tool steel, structural steel, line pipe steel, and stainless steel.3 Similarly, all of the PG&E data in Figure
1 were incorporated into one trendline. It is possible, however that the data represent more than one
trend, and that if these trends could be distinguished, the result would be multiple trends with higher
accuracy and a lower R2 value for the multiple linear regressions. For example, different types of steels,
manufacturing processes, compositions, or grain sizes could result in different relationships between
IIT estimate of YS and tensile YS.
This paper will briefly introduce important parameters for both the IIT algorithm and
microstructural developments for pipeline steels. Then results will be provided from two initial studies
indicating that microstructure and steel processing history influence IIT results. The following
discussion will provide evidence that further study is merited of the interactions between steel
processing history, microstructure, strain hardening, deformation, pile-up, and mechanical properties.
Laboratory test scatter and line pipe steel inhomogeneity can be important components of apparent
IIT error but are outside the scope of this paper.
Figure 1. Average YS estimates provided by IIT versus average tensile YS measured
destructively.2
3 Kim, S.H., Lee, B.W., Choi, Y., Kwon, D., “Quantitative determination of contact depth during spherical indentation
of metallic materials—A FEM study,” Materials Science and Engineering A, 415, 2006, pp. 59-65.
Pipeline pigging and integrity management conference, Houston, February 2020
5
IIT Considerations
The IIT estimate of YS relies in part on an accurate estimate of pile-up of material around the
indenter1 as represented in the schematics in Figure 2. For example, pile-up as shown in Figure 2a
may increase the contact area of the indenter and subsequently increase the load necessary for further
indentation. In the same way, sink-in as shown in Figure 2b may decrease the contact area of the
indenter and decrease the load necessary for further indentation. Taljat and Pharr showed that the
amount of pile-up around a spherical indenter depends on the modulus of elasticity, YS, strain-
hardening exponent, coefficient of friction, and depth of penetration.4 Morrison showed the importance
of microstructure by correlating large grains with high strain-hardening exponents and small grains
with low strain-hardening exponents.5 Thus, based on research that correlates high strain-hardening
exponents with sink-in and low strain-hardening exponents with pile-up,6 one might assume that large
grain sizes would result in sink-in behavior and small grain sizes with pile-up behavior. Silva et al.,
however, showed that for a single composition of modern steel, grain size could not account for changes
in strain-hardening exponent, but processing history also needs to be taken into account.7
Figure 2. Schematic showing a) pile-up and b) sink-in of the material around the indenter.
Steel Processing and Microstructure
For engineering materials such as line pipe steels, the term microstructure generally refers to
images, description, or classification of grains, grain boundaries, grain shapes and sizes, phases,
orientations, and textures. Generally, these important features are less than a millimeter in size and
are the result of the specific production practice, composition, and subsequent thermal, mechanical or
4 Taljat, B., Pharr, G.M., “Development of pile-up during spherical indentation of elastic–plastic solids,”
International Journal of Solids and Structures, Volume 41, Issue 14, 2004, Pages 3891-3904, ISSN 0020-7683,
https://doi.org/10.1016/j.ijsolstr.2004.02.033.
5 Morrison, W.B. "The effect of grain size on the stress-strain relationship in low-carbon steel." ASM Trans quart 59,
no. 4, 1966, pp. 824-846.
6 Karthik V., Visweswaran P., Bhushan A., Pawaskar D.N., Kasiviswanathan K.V., Jayakumar T., Raj B., “Finite
element analysis of spherical indentation to study pile-up/sink-in phenomena in steels and experimental
validation,” International Journal of Mechanical Sciences. 2012 Jan 1;54(1):74-83.
7 Silva R., Pinto A., Kuznetsov A., Bott I., “Precipitation and Grain Size Effects on the Tensile Strain-Hardening
Exponents of an API X80 Steel Pipe after High-Frequency Hot-Induction Bending,” Metals. 2018;8(3), pp. 168.
Pipeline pigging and integrity management conference, Houston, February 2020
6
chemical processes applied to the steel. Characterization of these features enable a fundamental
understanding of the mechanical properties and performance of the pipeline.
Microstructure provides insight into the processing and design intent of the steel. If the vintage is
unknown, the microstructure along with composition and knowledge of the evolution of line pipe
manufacturing processes over time can aid in estimating vintage, processing, and manufacturing
methodology. Proper characterization and quantification of the microstructure could aid with
interpretation of the behavior (such as the nature of crack growth for toughness testing) and may be
useful input for grade estimation.8
Thermo-mechanical processes, both bulk and localized, generate the resulting microstructure. Some
important high temperature processes that define microstructure include ingot casting, hot rolling,
pipe-making, welding, and annealing. Processes that change the shape of the steel product also
influence the microstructure, including casting, rolling, coiling, flattening, straightening, leveling,
pipe-making, and cold expansion.
There are three general approaches to steelmaking (as identified in API 5L)9 that are used as inputs
to the pipemaking process: normalizing, thermomechanical controlled processing (TMCP), and
quenching and tempering (Q&T), abbreviated as N, M and Q, respectively in API 5L - 2018. These
steelmaking approaches are schematically presented in Figure 3. For each relative temperature range,
the stable or expected phases are listed at the left in the diagram. Austenite is typically the stable high
temperature phase for carbon and alloy steels. Ferrite and pearlite are the typical stable constituents
for steel at service temperatures.
Figure 3. Schematic simplification of three steelmaking approaches commonly used for
pipelines to show the interaction of rolling temperatures with the stable phases for steel.
Normalized (N) steels are typically hot rolled within the austenite phase region at temperatures
from ~900-1200 °C (~1650-2190 °F), depending on the alloy and processing approach. After hot rolling,
8 Switzner, N.T., Veloo, P., Amend, B., Gould, M., Rosenfeld, M., Ma, J., Rovella, T., “A Methodology to Determine
the Grade for a Vintage Gas Transmission Pipe,” The Iron & Steel Technology Conference (AISTech), Pittsburgh, PA,
USA, May, 2019.
9 API Specification 5L, Line Pipe, Forty-Sixth Edition, American Petroleum Institute, 2018.
Pipeline pigging and integrity management conference, Houston, February 2020
7
the material cools through the transformation temperature range, ~700-900 °C (~1290-1650 °F), and
the austenite transforms to ferrite and pearlite. Ferrite and pearlite are typically easily resolved in the
final microstructures of normalized steels. A wide range of grain sizes are possible, depending on the
austenite grain size after deformation and the cooling rate. In general, the term, “normalization” can
also refer to the final heat treatment in plate or pipe processing whereby the steel is heated to a high
enough temperature that the grains completely transform and recrystallize, and thus reduce or
eliminate the detrimental effects of cold working, strain hardening, and residual stresses.
TMCP (M) steels are typically rolled within the transformation temperature range with careful
control of the deformation temperatures and cooling rates. Microalloying elements such as niobium,
vanadium, and titanium are sometimes added to enable microstructural control and improved
mechanical properties. The deformation temperatures and cooling rates interact with the phase
transformations in such a way that the TMCP approach can produce fine microstructures of polygonal
ferrite, carbide, bainite, and/or acicular ferrite. These engineered microstructures can enable
production of steel plate and sheet with high strength and high toughness. The low rolling
temperatures for TMCP steels demand high rolling forces and careful temperature control.
Q&T (Q) steels are typically hot rolled in the austenite phase, quenched to form a very hard
constituent called martensite, and subsequently tempered to restore ductility and toughness. Because
of the rapid quenching, there is inadequate time for the diffusion required to develop a ferrite/pearlite
microstructure. Instead, the rapid transformation reaction from austenite to martensite typically
results in jagged, irregular grain boundaries. These jagged, irregular grain boundaries may persist
through the tempering process and some second phase may precipitate during the tempering process.
Examples of microstructures typical of normalized, TMCP and Q&T steel line pipes are presented later
in this paper in the section “Study 2: Line Pipe Steels”.
After steelmaking, several different pipe-forming paths may be followed to produce the types of
pipes typical for API 5L including seamless pipe, double submerged arc welded (DSAW) pipe, and
electric resistance welded (ERW) pipe. Depending on the type of steel that enters the pipemaking
process, different microstructures can be expected. Accordingly, a wide variety of microstructures are
observed for various vintages and grades of line pipe steel.
It is important to briefly mention possible approaches to evaluating and quantifying
microstructures. Figure 4 shows examples of two easily quantifiable features in a representative steel
microstructure: pearlite (second phase) and ferrite grains. Grain size can be evaluated using ASTM
E11210, “Standard Test Methods for Determining Average Grain Size.” The fraction of second phase
can be evaluated using ASTM E562, “Standard Test Method for Determining Volume Fraction by
Systematic Manual Point Count”.11 Some other characteristics such as grain morphology (shape) and
ferrite clarity can be observed but are difficult to quantify.
10 ASTM E112, “Standard Test Methods for Determining Average Grain Size,” ASTM International, West
Conshohocken, PA.
11 ASTM E562, “Standard Test Method for Determining Volume Fraction by Systematic Manual Point Count,” ASTM
International, West Conshohocken, PA.
Pipeline pigging and integrity management conference, Houston, February 2020
8
Figure 4. Examples of ferrite grains and pearlite in the microstructure of normalized
steel.
Study 1: Plate Steels
To investigate the influence of steel processing and microstructure on IIT results, an initial study
was performed on steel plates with different compositions and microstructures. The three steel plates
included a high strength low alloy (HSLA) steel, a low carbon structural steel (A36), and a medium
carbon engineering alloy (1045). Details of the experimental methods are omitted here for brevity and
are planned for a future publication. The plates were cross-sectioned such that the microstructure could
be examined. Six laboratory tensile tests were performed for each plate and at least ten IIT
measurements were performed on each plate.
Microstructure
Metallographic samples were prepared in the laboratory for each of the plate samples and example
microstructures are shown in Figure 5. The A36 steel (Figure 5a) had a microstructure with polygonal
ferrite grains with well-defined grain boundaries and a low—but significant—amount of pearlite. The
A36 microstructure was similar to the normalized vintage steels from API 5L. The HSLA steel (Figure
5b) had very fine ferrite grains with interlocking grain boundaries and no pearlite. This microstructure
was somewhat similar to modern M (TMCP) grades from API 5L. The 1045 steel (Figure 5c) had a
mostly pearlitic microstructure with large areas of pearlite surrounded by smaller ferrite grains. The
elevated amount of carbon in the 1045 steel (as evidenced by the high amount of pearlite) would not
likely be found in a pipeline steel as API 5L limits the carbon content to ensure adequate weldability.
Pipeline pigging and integrity management conference, Houston, February 2020
9
Figure 5. Representative microstructures for the plate steels a) A36, b) HSLA, and c) 1045.
Tensile test and IIT results
Multiple tensile tests were performed for each steel plate, and a representative engineering stress
versus strain curve for each alloy is provided in Figure 6. Observe that the stress strain behavior of
the three samples are quantifiably different. The HSLA steel has the highest YS of the three steels
followed by a gradual downward slope. The gradual downward slope is indicative of a very low strain-
hardening exponent. The A36 steel appears to have the lowest YS followed by a slowly increasing and
then slowly decreasing stress versus strain curve. Similar to the previous note about the A36
microstructure, the A36 stress versus strain behavior is the most similar in this study to that of vintage
line pipe steel. Finally, the 1045 steel had a slightly higher YS than the A36, but much higher strain
hardening and the highest ultimate tensile strength (UTS) of the three alloys in this study. Thus, the
diverse compositions and processing histories of the three steels in this study led to microstructures
and strain-hardening behaviors that were quite different from one another.
Figure 6. Engineering stress versus strain curves for the three plate steels.
The quantitative results of tensile tests and IIT measurements are provided in Table 1. The results
are averages of six tensile tests and at least ten IIT measurements for each alloy. For the HSLA steel,
the IIT estimate of YS was ~16 ksi lower than the YS from the tensile test. For the A36 steel, the IIT
estimate appeared relatively accurate for both YS and UTS. For the 1045 steel, the IIT estimate for YS
appeared accurate, but the UTS was underestimated by ~8 ksi.
Pipeline pigging and integrity management conference, Houston, February 2020
10
Table 1. Summary of destructive tensile YS averages and IIT estimate of YS averages and
their differences.
YS (ksi)
UTS (ksi)
Alloy
Tensile
IIT
estimate
Difference Tensile
IIT
estimate
Difference
HSLA
73
57
-16
77
82
+5
A36
51
47
-4
71
72
+1
1045
56
54
-2
96
88
+8
The average YS estimate from IIT versus average YS result from tensile testing is plotted for each
steel in Figure 7a in a unity plot, and the average ultimate tensile strength (UTS) estimate from IIT
versus average UTS result from tensile testing is plotted for each steel in Figure 7b. The error bars
represent the standard deviation of the multiple tests for both IIT and tensile results. Data falling
along the dotted line on this unity plot would represent perfect correspondence between the two tests,
and the A36 and 1045 data are relatively close to the dotted line. Note however, that the IIT estimate
of YS provided an overly conservative estimate of YS for the HSLA steel. It is possible that the different
microstructures and strain-hardening behaviors resulted in quite different pile-up in the IIT test for
the three alloys. Thus, more insight about the pile-up behavior of the three steels might be useful for
distinguishing whether the YS for the three steels could be more accurately estimated by slightly
different IIT algorithms.
Figure 7 Strength comparisons for plate steels: a) average IIT YS estimate versus average
0.5% elongation under load (EUL) YS and b) average IIT UTS estimate versus average
tensile UTS.
Study 2: Line Pipe Steels
In a second study a subset of 12 modern (~2000s vintage) segments of line pipe was used (with
traceable, verifiable, and complete mill test records) from the PG&E IIT validation sample set. Three
of these pipes were normalized steel (API 5L, N designation), which would likely be the most similar
to early vintage line pipes in terms of microstructure and mechanical properties. Six of the pipes were
Pipeline pigging and integrity management conference, Houston, February 2020
11
made from steel produced using TMCP (API 5L, M designation). Three of the pipes were made from
quenched and tempered steel (API 5L, Q designation). Again, the details of the experimental methods
are omitted here for brevity. The pipes were cross-sectioned such that the microstructure could be
examined. Between four and eight tensile tests and at least 30 IIT measurements were performed on
each sample.
Microstructure
Metallographic samples were prepared in the laboratory for each of the pipe samples and example
microstructures are shown in Figure 8. The normalized steel pipe (Figure 8a) had a microstructure
with polygonal ferrite grains with well-defined grain boundaries and a low—but significant—amount
of pearlite, similar to the A36 steel microstructure. Again, this microstructure is expected to be similar
to the normalized vintage steels from API 5L, which are likely include Grade B, X42 and X52. The
TMCP steel pipe (Figure 5b) had very fine ferrite grains with interlocking grain boundaries and no
pearlite. This microstructure was somewhat similar to the previously discussed HSLA steel plate. The
quenched and tempered steel pipe (Figure 8c) had a fully ferritic microstructure with grains of irregular
shapes and jagged edges. No pearlite was observed, but other second phase products (carbides) could
exist between the parallel, elongated regions (lathes) of ferrite.
Figure 8 Example microstructure for each type of pipe steel a) normalized, b) TMCP and
c) quenched and tempered.
Tensile test and IIT results
The average YS estimate from IIT versus average YS result from tensile testing is plotted for each
steel line pipe in a unity plot in Figure 9a, and the average UTS estimate from IIT versus average UTS
result from tensile testing is plotted for each steel line pipe in Figure 9b. The error bars represent the
standard deviation of the multiple tests for both IIT and tensile results.
Pipeline pigging and integrity management conference, Houston, February 2020
12
Figure 9. Strength comparisons for line pipe steels: a) average IIT YS estimate versus
average 0.5% elongation under load (EUL) YS and b) average IIT UTS estimate versus
average tensile UTS.
The normalized pipes ranged from 46-53 ksi for average IIT YS and 44-47 ksi for average tensile
test YS. The average IIT YS was slightly non-conservative for all of the normalized pipes. Additionally
the normalized pipes were on the lower end of the YS spectrum. The Q&T pipes all yielded
approximately the same IIT estimate of YS, from 52-53 ksi, but exhibited very different average tensile
test YS results, ranging from 56-69 ksi. The TMCP pipes exhibited the widest range of average YS
values: 53-68 ksi for average IIT YS and 58-79 ksi for average tensile test YS. Similar to the previous
data from the HSLA steel, the IIT estimate of YS for the TMCP pipes was overly conservative.
It is important to observe the spread for the tensile YS data for a single value of IIT estimate of YS.
When the average IIT YS was 53 ksi, for example, average tensile test YS values of 47 ksi, 57 ksi, and
64 ksi were obtained for a normalized pipe, a Q&T pipe and a TMCP pipe, respectively. This is a spread
of 17 ksi in tensile YS for three pipes that had exhibited the same IIT YS result. As in the steel plate
study, the diverse compositions and processing histories of pipes in this study led to microstructures
that were quite different from one another. It is likely that the different microstructures resulted in
very different strain-hardening behaviors and pile-up in the IIT test, which caused three pipes with
very different YS to appear to have matching YS. It is possible that microstructure could be used to
provide evidence of the processing history of the steel, that could then be used to inform the IIT
algorithm in terms of strain-hardening exponent and pile-up. We propose that a better understanding
of the pile-up behavior for different types of pipe steels might prove useful for identifying whether the
YS could be more accurately estimated by different variants of the IIT algorithm.
Future Work
PG&E plans to perform a unique study on line pipe steel with varying microstructures and strain-
hardening behaviors wherein the amount of pile-up will be precisely measured at multiple steps of the
IIT measurement process. PG&E is also planning collect more metallographic samples to characterize
Pipeline pigging and integrity management conference, Houston, February 2020
13
and evaluate the quantifiable aspects of the microstructures. The microstructural data will be utilized
to distinguish trends within a larger set of IIT data that incorporates YS estimates for vintage line
pipes. Simultaneously, more information about the evolution of manufacturing processes over time will
be gathered to enhance our understanding of how the relationship between steel processing,
microstructure, and properties affect the accuracy and error trends of in-situ, nondestructive strength
measurements.
Concluding Remarks
The results of initial studies for three plate steels and twelve steel line pipes were presented to
elucidate the influence of processing history and microstructure on IIT YS measurements. The present
results suggests that understanding sample microstructure allows the prediction of the performance
of IIT measurements when compared to laboratory tensile test values. A sample’s microstructure is
linked to its strain-hardening exponent and subsequently to IIT pile-up. Specifically, the 1045 and A36
steel plates, and the normalized line pipe, exhibited ferrite-pearlite microstructures and resulted in a
relatively good match between the IIT estimate for YS and the tensile test YS. For HSLA plate, TMCP
pipes, and Q&T pipes tested in this study, average IIT estimates of YS tend to be conservative in
comparison with the tensile YS values. This conservatisms was greater for the samples with a YS > 60
ksi. However, for the normalized line pipe steels, IIT measurements of YS were non-conservative when
compared to tensile test values. Future studies of pile-up from IIT tests are proposed to clarify the
relationships between processing history, microstructure, and pile-up with a goal to improve the NDE
estimate for YS using IIT.
