22 Wiskel PDF
22 Wiskel PDF
22 Wiskel PDF
Ultrasonic testing (UT) is typically used for crack and/or flaw detection in many materials. In
addition, the velocity (and/or attenuation) of the ultrasonic waves can be used to assess the
microstructure of the material. The work presented in this paper explores the effect of steel
type/microstructure on ultrasonic wave velocity. A series of heat treatments – ranging from
water quenching to furnace cooling to obtain a variety of microstructures (i.e. martensite,
bainite and ferrite + pearlite) - was applied to three grades of steel: L80, 4130 and 5160. For
each heat treatment, the longitudinal (L80) and shear velocity were measured. Optical
microscopy and hardness were used to determine the phase(s) present in the steels. In
addition, wave velocities through the thickness of the as-received L80 steel strip were also
measured. X-ray diffraction analysis of the as-received L80 samples was conducted to assess
the magnitude of preferred orientation (texture) in the steel strip. For both the heat treated and
as-received steels, the measured velocities were compared with predicted velocities
calculated using either an isotropic or anisotropic methodology. For the heat treated steels,
the shear velocity decreased appreciably with both a change in microstructure (ferrite to
martensite) and increasing steel hardness. The predicted values of both longitudinal and shear
velocity for martensite compared favourably with the measured values.
Keywords: velocity, L80, steel, heat treated, microstructure, hardness, anisotropy, 4130, 5160
1. Introduction
A number of studies have been undertaken correlating ultrasonic velocity with
microstructural features [1-11] in cast iron, steel, stainless steel, and heat treatable high
strength steels. The work conducted in this paper focuses on ultrasonic velocity
measurements made on L80 steel – a low carbon (0.24 wt% C) heat treatable steel used for
pipe in the oil and gas industry. The L80 steel was examined in the as-received (L80-AR)
condition (through the thickness) and following specific heat treatments (L80-HT) to
generate a variety of microstructures phases including martensite, bainite and ferrite +
pearlite. In addition, ultrasonic velocity measurements were made on two heat treatable alloys
with carbon contents different than the L80 steel– specifically 4130 (≈ 0.3 wt% C) and 5160
(≈ 0.6 wt% C) steels. Both alloys were heat treated in manner similar to the L80 steel. The
effect of phase type (ascertained via optical microscopy and hardness measurements) on the
magnitude of longitudinal and shear velocities was examined. The measured velocities were
compared with predicted velocities calculated using either an isotropic or anisotropic
(Christoffel equation) methodology.
2. Background
The processing of L80 steel for pipe will be briefly introduced followed by a review of
microstructural factors that effect ultrasonic velocity. In addition, correlations for predicting
ultrasonic velocities, for isotropic and anisotropic materials, will be presented.
2.1 L80 steel
L80 steel is a low alloy heat treatable steel used extensively for pipe in the oil and gas
industry. The composition of the L80 studied in this work is 0.24 wt% C, 0.99 wt% Mn and
0.16 wt% Cr. L80 pipe can be manufactured as either seamless or electric resistance welded
(ERW) pipe where the latter is made from hot rolled steel strip. Following forming, the pipe
is austenitized, quenched and tempered to the required mechanical properties and then
ultrasonically inspected. The formation of an entirely martensitic structure during the
quenching cycle is necessary to ensure that the pipe properties are met in the tempering
operation. The evaluation of the quenched martensitic microstructure using existing
ultrasonic systems may be used for quality control purposes.
2.2 Microstructural factors affecting ultrasonic velocity
Microstructural factors affecting the velocity [1, 12-16] of an ultrasonic wave transiting an
elastic material include phase type, grain size, phase anisotropy and crystallographic texture
(preferred orientation).
2.2.1 Elastic constants for different phases in steel
A substantial body of research has been conducted on the effect of steel microstructure (e.g.,
martensite, ferrite and ferrite + pearlite) on ultrasonic velocity, stemming primarily from a
variation in the elastic constants associated with each phase. The elastic constants for the
microstructure types typically observed in steel, including ferrite, pearlite and martensite, are
tabulated in Table 1. It should be noted that pearlite is a lamellar structure consisting of
ferrite (body centred cubic - BCC) and iron carbide (orthorhombic); thus, the elastic constants
shown in Table 1 for this microconstituent are deemed as “effective” values. The crystal
structure for martensite is body centred tetragonal (BCT) due to carbon “trapped” within the
structure during the quenching operation. However, the degree of tetragonality depends on
the carbon content of the steel. As the C content is reduced, the c/a ratio (of the BCT
structure) decreases and at relatively low carbon contents the martensite crystal structure
approaches a BCC structure. Experimentally determined elastic constants [17] for a 0.5 wt%
C martensite showed that C11 ≈ C33, C12 ≈ C13 and C44 ≈ C66. The equivalency of each of these
elastic constants is characteristic of a cubic structure [16]. A cubic assumption for the
martensite phase will be used in subsequent calculations.
2.2.2 Phase anisotropy
where C11, C12 and C44 are the elastic constants defined in Table 1. A value of A = 1 defines
an isotropic material; while A values >> 1 indicate an increasing degree of inherent phase
anisotropy. The A values for ferrite, pearlite and martensite are shown in Table 1.
2.2.3 Grain size effect on velocity
Grain size [12] can affect the measured velocity in a polycrystalline material via the Snell
effect (i.e., refraction of ultrasonic waves as they propagate through individual grains with
different orientations and, hence, elastic properties). The anisotropy inherent in ferrite can
result in a significant Snell effect (depending on grain size); thus, leading to a longer
propagation path with a corresponding decrease in apparent velocity.
Table 1. Elastic constants for 1050 steel [17-18]
Phase C11 C12 C44 A E Density v
(GPa) (GPa) (GPa) (GPa) (kg/m3) (Poisson’s ratio)
Ferrite 232.0 135 116.0 2.39 211.9 7,851 0.289
Pearlite 273.4 110.5 82.2 1.01 210.3 7,835 0.288
Martensite 268.1 111.2 79.1 1.01 203.5 7,709 0.292
where ν is the Poisson's ratio for the material. Similarly, the velocity of a
longitudinal/compression wave (VL-iso) can be calculated [2]:
VL −iso = Vo ⋅
(1 −ν ) (4)
(1 − 2ν ) ⋅ (1 +ν )
2.3.2 Anisotropic velocity – Christoffel Equation
For an anisotropic material, the Young’s modulus is not a true representation of the material’s
elastic properties and the stiffness tensor in its entirety must be used instead. The ultrasonic
velocity equation can be calculated using the Christoffel equation:
(C ijkl )
⋅ d j ⋅ d l − V 2 ⋅ ρ ⋅ δ ik ⋅ p k = 0 (5)
where Cijkl is the stiffness tensor, dj and dl are the wave propagation directions, V is the phase
velocity of the wave, ρ is the density of the material, δik is the Kroenecker delta function and
pk is the particle displacement direction. The equation in matrix form [16, 20] is:
λ11 − ρV 2 λ12 λ13 p1 0
λ12 λ 22 − ρV 2 λ 23 ⋅ p 2 = 0
λ13 λ 23 λ 33 − ρV 2 p 3 0 (6)
where λij are functions of Cij - the elements of the stiffness tensor in the Voigt notation. For a
cubic crystal structure (e.g., ferrite) the stiffness tensor is simplified into three components,
C11, C12 and C44 and thus λij can be calculated from C11, C12 and C44 and the cosine of the
wave direction with respect to the x, y and z axis. For example,
λ11 = l 2 ⋅ C11 + (m 2 + n 2 ) ⋅ C 44 (7)
where l, m and n are the directional cosines and are defined by l = cos(θ1), m = cos (θ2) and n
= cos (θ3) where θ1, θ2 and θ3 are the angles which the directional vector makes with the x, y
and z axes of the cubic system. These angles are illustrated in Figure 1a for the [110]
direction.
The three (3) real roots of the determinant of Equation (6) (described by a cubic function of
V2) are calculated and correspond to three ultrasonic velocities [21], two shear modes (Vs1
and Vs2) and one longitudinal (or compression) mode (VL). The difference between velocity
modes is related to the particle displacement axis relative to the wave propagation direction.
For example, for a wave propagating down the z-axis of a material, the wave is a longitudinal
wave if the particles are displaced along the z-axis. For a shear wave, particle displacement is
perpendicular to the travel direction (i.e., along either the y axis or x axis as shown in Figure
1b [21]). For an isotropic material there is no difference in properties along the x or y axis;
hence, the two shear velocities are the same. However, for an anisotropic material, the
properties are not necessarily the same along the x or y axis. Therefore, the magnitude of the
two shear velocities can differ depending on the particle displacement direction relative to the
propagation direction. Where applicable, the particle displacement parallel with the rolling
direction is defined as Vs1 and particle displacement transverse to the rolling direction is
defined as Vs2.
Figure 1a. Directional cosines Figure 1b. Shear wave particle displacements
∑ϕ i ⋅ Vi
(8)
Vavg = i =1
N
where, Vavg is the velocity of interest (i.e., VL-avg, VS1-avg and VS2-avg), Vi is the longitudinal or
shear velocity for a specific direction (from Equation 6), ϕi is the texture weighting factor
(ϕi= 1 for no texture) and N is the total number of directions considered. For the purpose of
this work, N corresponds to the sequential stepwise (∆θ = 5°) manipulations of θ1, θ2 and θ3
through all angles (see Figure 1a) in the crystallographic cell. Equation (8) assumes the
effects of Snell refraction are negligible and that (as stated earlier) martensite approximates a
cubic structure. It is observed (Table 2) that the predicted velocities (Isotropic, Directional
and Average) for either pearlite or martensite are virtually identical. Due to this similarity,
only the Average calculated values will be used in subsequent comparisons with measured
velocities in both of these phases.
Table 2. Predicted velocities for ferrite, pearlite and martensite
Ferrite (m/s) Pearlite (m/s) Martensite (m/s)
Isotropic
VL-iso 5939 5915 5896
VS-iso 3235 3228 3196
Directional VL-dir VS1-dir VS2-dir VL-dir VS1-dir VS2-dir VL-dir VS1-dir VS2-dir
[100] 5436 3844 3844 5910 3238 3238 5902 3202 3202
[110] 6177 2486 3844 5914 3232 3238 5906 3194 3202
[111] 6405 3007 3028 5915 3234 3234 5907 3197 3197
Average
VL-avg 6161 5914 5906
VS1-avg 2823 3233 3196
VS2-avg 3605 3236 3200
3.1.2 Heat Treated L80, 4130 and 5160 steel ultrasonic velocities and hardness
The average values and standard deviation (S.D.) for VL, VS1 and VS2, Vickers hardness and
the observed microstructure for the L80-HT steel are shown in Table 3. The VS1, Vickers
hardness and observed microstructure for the heat treated 4130 and 5160 steels are shown in
Table 4.
Table 3. Velocity, hardness and microstructure for heat treated L80-HT steel
Table 4. Velocity, hardness and microstructure for the heat treated 4130 and 5160 steel
Sample VS1 Hardness Microstructure
(m/s) (HV)
4130
WQ 3172 566 Martensite
AC 3209 258 Ferrite + Pearlite
FC 3212 163 Ferrite + Pearlite
5160
WQ 3157 702 Martensite
AC 3288 387 Ferrite + Pearlite
FC 3200 290 Ferrite + Pearlite
Figure 3a. L80 near the top of the strip Figure 3b. L80 at the centerline
3.2.2 X-ray diffraction
Quantitative X-ray diffraction (XRD) was undertaken on each through thickness L80-AR
sample as a means to quantify texture (i.e., preferred orientation). The XRD pattern for L80-
S1 is shown in Figure 4. For a completely random grain orientation, the relative intensities of
the (110), (200) and (211) ferrite diffraction peak intensity should be 100, 20 and 30.
However, the relative peak intensities observed in Figure 4 are 100, 12.1 and 16.6. This
difference in peak intensity ratio indicates that the ferrite in the L80 exhibits some
preferential orientation. Iron carbide (Fe3C) is not detected in the XRD pattern due to the low
weight fraction present in the L80 steel. Similar XRD patterns were observed for all other
through thickness L80 samples.
Figure 5a. VL for L80-HT- M and Lit. Figure 5b. VS for L80+4130+5160 and Lit.
4.2 Velocity as a function of hardness for L80-HT steel
The VL and VS2 values for the L80-HT steel only are shown in Figures 6a and 6b,
respectively. The value of VS2 (VS1 is similar) decreases with increasing hardness, while VL
appears to be less sensitive to hardness. This suggests that shear velocity is a better indicator
for microstructure analysis of this alloy.
Figure 6a. VL for L80-HT vs. hardness Figure 6b. VS2 for L80-HT vs. hardness
4.3 Through thickness velocity for L80-AR steel
The longitudinal velocity for the L80-AR steel (a mixture of ferrite and pearlite) as a function
of depth below the rolling face is shown in Figure 7 along with predicted velocity values
(horizontal lines). VL-avg-ROM and VL-iso-ROM were calculated using a rule of mixtures (ROM)
approach on the assumption that the structure of L80-AR consists of 71 wt% ferrite and 29
wt% pearlite). VL-avg-ROM is observed to be significantly higher than the measured data. VL-iso-
ROM is closer to the measured data, but the measured values are consistently lower. The XRD
relative intensities shown in Table 2 indicate that there is some preferred orientation in the
L80, hence the assumption of a randomly oriented grain configuration (used to calculate both
VL-avg-ROM ) may not be valid.
Figure 7. VL-avg-ROM and VL-iso-ROM for L80-AR vs. distance below the surface
5. Conclusions
The measured longitudinal and shear velocities for martensite in L80, 4130 and 5160 steels
compare favourably with literature values. The predicted velocities for martensite using the
Christoffel equation also compare reasonably well with the measured values. XRD patterns
for L80-AR steel indicate a degree of preferred orientation in the steel. The predicted rule of
mixtures (ROM) velocity using the Christoffel equation for a L80-AR steel (consisting of
ferrite and pearlite) significantly over predicts the measured L80-AR steel values. The
ultrasonic shear velocity VS2 measured for bainite (3239 m/s) was observed to lie between the
martensite and ferrite + pearlite velocities.
Acknowledgements
The authours would like to thank NSERC, EVRAZ N.A., Enbridge, TCPL, Alliance
Pipelines and UT Quality for financial assistance.
References
1. E. Papakadis, ‘Ultrasonic attenuation and velocity in three transformation products in
steel”, J. Appl. Phy., 35, pp. 1474-1482, 1964.
2. A. Moro, C. Farina and F. Rossi, ‘Measurement of ultrasonic wave velocity in steel
for various structures and degree of cold working’, NDT Int., Aug., pp.169-175, 1980.
3. R. Prasad and S. Kumar, ‘Study of the influence of deformation and thermal treatment
on the ultrasonic behaviour of steel’, J. Mat. Proc. Tech., 42, pp.51-59,1994.
4. D. C. Hurley, D. Balzar, P. T. Purtscher, and K. W. Hollman, ‘Nonlinear ultrasonic
parameter in quenched martensitic steels’, J. App. Phys.,83(9), pp. 4584-4588, 1998.
5. S. Kim and W. Johnson, ‘Elastic constants and internal friction of martensite steel,
ferritic-pearlitic steel and α-iron’ Mat. Sci. Eng. A, 452-453, pp. 633-639, 2007.
6. C. Hakan Gur and I. Cam, ‘Comparison of magnetic Barkhausen noise and ultrasonic
velocity measurements for microstructure evaluation of SAE 1040 and SAE 4140
steels’, Mat. Char., 58, pp447-454, 2007.
7. T. Łukomski and T. Stepinski, ‘Steel hardness evaluation based on ultrasound
velocity measurements’, Insight, 52(11), pp. 592-596, 2010.
8. V. Feritas et al., Nondestructive characterization of microstructures and determination
of elastic properties in plain carbon steel using ultrasonic measurements’, Mat. Sci.
Eng. A, 527, pp. 4431-4437, 2010.
9. H. Terasaki et al., Experimental Determination of Elastic Modulus during Martensitic
Transformation of Low Transformation Temperature Steel’, ISIJ, 51(9), pp. 1566-
1568, 2011.
10. E. Fonseca and A. Reguly, ‘The Use of the Ultrasound Measurement Technique for
the Evaluation of Mechanical Properties of the ASTM A36 Steels’, J. Braz. Soc.
Mech. Sci. Eng., XXXIII(2), pp. 212-217, 2011.
11. M. El Rayes, E. El-Danaf and A. Almajid, Ultrasonic characterization of heat-
treatment effects on SAE-1040and -4340 steels’, J. Mat. Proc. Tech., 216, pp. 188-
198, 2015.
12. D.R. Allen and C.M. Sayers, ‘The measurement of residual stress in textured steel
using an ultrasonic velocity combinations technique’, Ultrasonics, pp. 179-188, 1984.
13. W. Orlowicz and Z. Opiekun, ‘Ultrasonic detection of microstructure in cast iron’,
Theoretical and Applied Fracture Mechanics’, 22, pp. 9-16, 1995.
14. C. Lane, The Development of a 2D Ultrasonic Array Inspection for Single Crystal
Turbine Blades, Springer Publishing, pp.13-39, 2014.
15. M.G. Silk, ‘Relationships between metallurgical texture and ultrasonic propagation’,
Mat. Sci., 15, pp. 559-565, 1981.
16. B. Castagnede, W.Sachse and M. Thompson, ‘Determination of the elastic constants
of anisotropic materials via laser-generated ultrasound’, Ultra. Int., pp.71-79, 1989.
17. S. Kim and W. Johnson, ‘Elastic constants and internal friction of martensite steel,
ferritic-pearlitic steel and α-iron’ Mat. Sci. Eng. A, 452-453, pp. 633-639, 2007.
18. V. Feritas et al., Nondestructive characterization of microstructures and determination
of elastic properties in plain carbon steel using ultrasonic measurements’, Mat. Sci.
Eng. A, 527, pp. 4431-4437, 2010.
19. R. Ray, M. Butron-Guillen, J. Jonas G. Ruddle, ‘Effect of controlled colling on
texture development in a plain carbon and Nb Microalloyed Steel, ISIJ, 32(2), pp203-
212, 1992.
20. J. Rose, Ultrasonic Waves in Solid Media, Cambridge University Press, 1999.
21. J. Kennedy, “Ultrasonic Evaluation of Microstructure in Pipe Steels”, M.Sc. Thesis,
University of Alberta, 2015.
22. H. Terasaki et al., Experimental Determination of Elastic Modulus during Martensitic
Transformation of Low Transformation Temperature Steel’, ISIJ, 51(9), pp. 1566-
1568, 2011.
23. C. Hakan Gur and I. Cam, ‘Comparison of magnetic Barkhausen noise and ultrasonic
velocity measurements for microstructure evaluation of SAE 1040 and SAE 4140
steels’, Mat. Char., 58, pp. 447-454, 2007.
24. D. C. Hurley, D. Balzar, P. T. Purtscher, and K. W. Hollman, ‘Nonlinear ultrasonic
parameter in quenched martensitic steels’, J. App. Phys.,83(9), pp. 4584-4588, 1998.
25. M. El Rayes, E. El-Danaf and A. Almajid, Ultrasonic characterization of heat-
treatment effects on SAE-1040and -4340 steels’, J. Mat. Proc. Tech., 216, pp. 188-
198, 2015.
26. C. Gur and B. Tuncer, ‘Characterization of microstructural phases of steels
by sound velocity measurement, Mat. Char., 55, pp. 160-166, 2005.
27. S Hsia and Y. Chou, “Assessing the Hardness of Quenched Medium Steel Using
an Ultrasonic Nondestructive Method’, Advances in Materials Science and
Engineering, pp. 1-8 , 2014.