Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Submitted September 2007 From Materials Evaluation, Vol. 66, No. 3, pp: 332-337.

Copyright © 2008 The American Society for Nondestructive Testing, Inc.

Time of Flight Diffraction Testing of Austenitic


Stainless Steel Weldments at Elevated
Temperatures
by R. Subbaratnam,* Saju T. Abraham,* M. Menaka,* B. Venkatraman† and Baldev Raj*

ABSTRACT time of flight diffraction for the testing of such welds. The tempera-
Ultrasonic testing is a mandatory requirement during inservice inspec- tures likely to be encountered during inservice inspection would be
tion of the welds in the main and safety vessels of a prototype fast breeder of the order of 423–473 K. One of the main problems during ultra-
reactor. Presently, conventional ultrasonic techniques have been proposed sonic testing at high temperatures is the decrease in the signal-to-
for use. Two of the main challenges likely to be encountered during ultra- noise ratio, resulting in poor sensitivity. A survey of the literature
sonic testing are the high temperature of the vessels (around 423–473 K indicates that, while time of flight diffraction has been extensively
during shutdown) and the background radiation. Ultrasonic time of flight applied for testing thick carbon steel weldments at ambient temper-
diffraction is now well established as a technique for NDT of thick (greater atures, records of its application for testing stainless steel weld-
than 12.5 mm) weldments. The main advantage of using time of flight dif- ments, especially at high temperatures, are sparse. Compared to
fraction is its increased accuracy for discontinuity detection and fast scan- carbon steel, testing of stainless steel is more complicated due to the
ning times. Conventionally, time of flight diffraction has been applied only anisotropic nature of the material and the weldment. In this work,
for carbon steel weldments and at ambient temperatures. This paper high- the authors explore the feasibility of applying time of flight diffrac-
lights the successful application of time of flight diffraction for testing of tion for testing of thick austenitic stainless steel weldments with
stainless steel weldments at high temperature (423 K). Experimental stud- temperatures up to 473 K and the errors likely in discontinuity
ies reveal that the diffracted signal amplitude decreases and, hence, the time characterization at these temperatures. An attempt has also been
of flight diffraction image contrast decreases with increasing temperature. made to obtain a constitutive relation between temperature and ad-
Additional gain was necessary to compensate for the decreased signal am- ditional gain based on the experimental analysis.
plitude. Analysis of the experimental data indicated that the additional
gain necessary to compensate for the lower signal amplitude varied nonlin- EXPERIMENTAL STUDY
early (quadratically) with temperature. The errors in the discontinuity di- A systematic experimental study was undertaken on two
mensional measurement were observed to be less than 5% compared to austenitic stainless steel weld pads, each with dimensions 200 ¥ 150
measurements at ambient temperature. ¥ 25 mm. The first weld pad had lack of penetration introduced
Keywords: time of flight diffraction, stainless steel, high temperature, during the welding process at two different locations, and the sec-
thermal image, scattering. ond weld pad had slag inclusions at two different locations. The
weld was of the double vee butt joint configuration and was made
INTRODUCTION with the shielded metal arc welding process. A hot plate with pro-
Ultrasonic testing is a mandatory requirement during inservice visions to vary the current using a variac setup was used for heat-
inspection of the welds in the main and safety vessels of a proto- ing the weld pads. The temperature distribution on the steel plate
type fast breeder reactor. Presently, conventional ultrasonic tech- surface, especially in the weld region, could be maintained within
niques have been proposed for use. Two of the main challenges ±4 K during the scanning by adjusting the variac. The uniformity of
likely to be encountered during ultrasonic testing is the high tem- temperature distribution, and also the temperatures in the vicinity
perature of the vessels (around 423–473 K during shutdown) and of the weld and on the weld, was verified using thermal imaging.
the background radiation. Ultrasonic time of flight diffraction is Figure 1 shows the thermal image of the weld pad after it has been
now well established as an NDT technique, and codes such as the
ASME Boiler and Pressure Vessel Code (2007) recommend it for regu-
lar testing of thick (greater than 12.5 mm) weldments. The two
main advantages of time of flight diffraction over conventional ul-
trasonic testing are:
■ time of flight diffraction can characterize the dimensions of lin-
ear discontinuities with better accuracy compared to conventional
pulse/echo ultrasonic techniques
■ the entire weld can be scanned in a single pass with great relia-
bility and within very short time.
The abovementioned points are highly relevant in the context of
inservice inspection. The second advantage is particularly attrac-
tive, since the probes and electronics would be exposed to the ef-
fects of radiation and high temperatures for shorter periods of time.
In view of this, it was decided to explore the feasibility of applying

* Indira Gandhi Centre for Atomic Research, Kalpakkam, Tamil Nadu 603
102, India.
† Quality Assurance Division, Indira Gandhi Centre for Atomic Research,
Kalpakkam, Tamil Nadu 603 102, India; e-mail <bvenkat@igcar.gov.in>. Figure 1 — Typical thermal image of a weld pad.

332 Materials Evaluation/March 2008


heated to the required temperature. It can be observed that the tem- RESULTS AND ANALYSIS
perature distribution is quite uniform in and around the weld re- Figures 4, 5 and 6 are the time of flight diffraction images of the
gion. Quantitative estimation of the temperature was then attempt- reference notch, lack of penetration and slag acquired at ambient
ed after incorporating appropriate corrections for emissivity and and elevated (approximately 423 K) temperatures. It can be ob-
distance. The temperature in and around the weld was observed to served from Figures 4a and 4b, 5a and 5b, and 6a and 6b that the
vary from 421 to 433 K, with the mean temperature being approxi- noise increases and image contrast (signal-to-noise ratio) decreases
mately 426.6 K. Within the weld, the temperature was observed to at high temperature. This is also clearly revealed by the peak ampli-
vary between 429 and 433 K. This variation is primarily due to the tude of the diffracted signal, which decreases as the temperature of
variation in thickness of the weld reinforcement. Figure 2 is the the object under test increases.
temperature profile over the area in and around the weld. At ambient temperature with 50 dB system gain, the peak am-
Prior to testing the actual weld pads, a reference specimen of plitude of the diffracted signal from the side drilled notch was ob-
thickness 30 mm with a side notch measuring 10 ¥ 10 ¥ 1 mm was served to be 39% of the screen height. As indicated above, with the
heated up to 423 K starting from ambient temperature (298 K) in increase in temperature, the peak amplitude was observed to de-
steps of 25 K. The typical experimental setup for the testing of the crease. Additional gain was required to compensate for the loss in
reference specimen with the side drilled notch and for the weld signal amplitude. Thus, the system gain had to be increased corre-
pads is shown in Figure 3. A special probe holder and scanning spondingly in each of the cases. Figures 4c, 5c and 6c are the images
arrangement was designed and developed for this study. Time of of the reference notch and discontinuities at 423 K, and with the
flight diffraction testing was performed using commercially avail- gain correspondingly increased so that the peak amplitude is the
able time of flight diffraction equipment. High temperature, 45° same as observed at ambient temperature.
longitudinal wave probes (capable of being used for testing of com- The additional gain required at various temperatures in all three
ponents with temperatures up to 473 K) were used along with high cases to compensate for the decreased signal-to-noise ratio is sum-
temperature couplant and cable. A preamplifier of 20 dB was used marized in Table 1.
with the equipment. The system gain used was 50 dB. The decrease in signal-to-noise ratio with increasing object
temperature can be explained as follows. It is well known that the
signal-to-noise ratio depends on the attenuation of ultrasonic
waves. Ultrasonic waves are high frequency (in NDT, usually on
the order of 1 MHz) elastic waves that travel through a material
medium. Attenuation of elastic waves in polycrystalline metals is
generally due to elastic hysteresis, thermoelastic relaxation or grain
boundary scattering (Darbari et al., 1968). Scattering is the most im-
portant cause of attenuation of signals at higher temperatures. As
the temperature increases, interatomic vibration increases, produc-
ing large amounts of scattering. Scattering is also the sum of several
effects like reflection, refraction and mode conversion, which de-
pend on the frequency and polarization of the vibrations, the grain
size and degree of anisotropy of the grain. An increase in elastic
anisotropy (this can be achieved by increase in temperature) alone
causes an increase in scattering. Elastic anisotropy is determined by
the elastic constants of the material. Elastic constants in a material
can be expressed in terms of longitudinal sound velocity vl and
transverse sound velocity vt by the following relationships:

(1) longitudinal modulus: Cl = ρvl2

(2) shear modulus: G = ρvt2

Figure 2 — Temperature profile of a specimen along the weld center where


line. r = density of the material.

Figure 3 — Experimental setup: (a) for a reference specimen with a side notch; (b) for one of the weld pads.

Materials Evaluation/March 2008 333


(a) (b) (c)

Figure 4 — Reference notch image: (a) at 298 K with 50 dB gain; (b) at 423 K with 50 dB gain; (c) at 423 K with 67 dB gain.

(a) (b) (c)

Figure 5 — Lack of penetration image: (a) at 298 K with 50 dB gain; (b) at 423 K with 50 dB gain; (c) at 423 K with increased gain (65 dB).

(a) (b) (c)

Figure 6 — Slag inclusion image: (a) at 298 K with 50 dB gain; (b) at 423 K with 50 dB gain; (c) at 423 K with increased gain (65 dB).

Table 1 Additional gain required at different temperatures to increase peak amplitude to the reference level (39% of screen height at
ambient temperature)

Temperature (K) Additional Gain Required (dB) for Additional Gain Required (dB) for Additional Gain Required (dB) for
Reference Specimen Weld with Lack of Penetration Weld with Slag
298 0 0 0
323 3 2 1
348 7 3 4
373 10 11 7
398 14 12 9
423 17 15 15

334 Materials Evaluation/March 2008


4
(3) bulk modulus: B = Cl − G
3 20

9GB
(4) Young′s modulus: E =
G + 3B 15

Additional Gain (dB)


E 10
(5) Poisson ′s ratio: ν = −1
2G

5
Elastic constants depend on temperature as well, since elastic
constants are coefficients that relate stress and strain, which is a Data points
function of the relative position of crystal lattices. As the tempera- 0
Linear fit
ture increases, the relative position of the lattice changes because of
the increased lattice vibration. Temperature dependence of the elas-
tic constants also results from the variation of the elastic potential –5
energy due to anharmonicity. 250 300 350 400 450
As the temperature increases, the Poisson’s ratio n has been ob- Temperature (K)
served to increase (at 473 K, n increases by approximately 2%),
which indicates softening of the material. Scruby and Moss (1993)
have observed that shear modulus and Young’s modulus decrease
(by approximately 7% at 473 K) with an increase in temperature. Figure 7 — Additional gain versus temperature for the reference
This clearly indicates that ultrasonic wave velocity decreases with specimen.
an increase in temperature. Within the material, the temperature
field is not uniform, but varies with depth. This temperature gradi-
ent results in a velocity gradient. Reflection is another cause for at-
tenuation at high temperature. The reflection coefficient ar is given
by (Canella and Monti, 1980)
20
C − C1
(6) αr = 2 18
C2 + C1
16

14
where
Additional Gain (dB)

C1 = the velocity in the first medium (due to the velocity 12


gradient) 10
C2 = the velocity in the second medium (due to the velocity
gradient). 8

6
Thus varying temperature gradients lead to varying reflection coef-
ficients and enhanced scattering. 4 Data
Temperature gradients also result in significant changes in the Quadratic fit
2
beam angle in the material as well as in the contact probe itself
(Brunk, 1988). Large angles change more rapidly than small angles 0
250 300 350 400 450
per unit change in temperature. As a result, refraction and/or re- (a) Temperature (K)
flection occur and this will cause reduction in received energy.
Thus, overall as the temperature of the object increases, the scat-
tered radiation due to the abovementioned causes increases, result-
ing in significant decrease in the signal-to-noise ratio and hence 20
image contrast. 18
Figures 7 and 8 are graphical plots of the additional gain required at
different temperatures to compensate for the decreased signal-to-noise 16
ratio for the reference notch and the discontinuities. Curve fitting was 14
done using commercial software to get the best fit.
Additional Gain (dB)

The data in the case of the reference notch could be fitted with 12
an equation of the form 10

8
(7) ∆g = aT − b
6

where 4 Data
Dg is the additional gain 2
Quadratic fit
T is the temperature in kelvins
a and b are constants. 0
250 300 350 400 450
(b) Temperature (K)
It can be observed from Figure 7 that for the reference notch, the ad-
ditional gain required increases linearly with increasing temperature.
In the case of welds with discontinuities, based on the experi-
mental results (Figure 8), the increase in gain with temperature can Figure 8 — Additional gain required at different temperatures for welds
best be described by the constitutive relation with: (a) slag inclusion; (b) lack of penetration.

Materials Evaluation/March 2008 335


case of slag, is due to the orientation of the discontinuity, since
(8) ∆g = aT 2 + bT + c the pulse/echo technique is highly dependent on the orientation
of the reflector. This can be minimized by using different angle
probes and higher decibel drop techniques. In this investigation,
Equation 8 indicates that in the case of weld discontinuities, the a 45°, 4 MHz longitudinal wave probe was used for time of flight
additional gain varies nonlinearly with increasing temperature. diffraction and, hence, the same was used for pulse/echo testing
This can be explained by the fact that the weld has a different grain on a weld with a 6 dB drop technique. Table 4 gives the percent-
structure compared to the parent material, and the anisotropy in age variation of discontinuity dimension by time of flight diffrac-
the weld region, especially in the case of stainless steel, is much tion compared with radiography. Maximum variation was found
more pronounced compared to the base material — hence this non- to be 2.18% for the first lack of penetration specimen (LP 1),
linear behavior. 1.09% for the second lack of penetration specimen (LP 2), 4.80%
for the first slag inclusion specimen (Slag 1) and 4.60% for the
Discontinuity Dimensional Analysis second slag inclusion specimen (Slag 2).
An important requirement of inservice inspection is dimension-
al measurement of discontinuities. A study was made to estimate Scanning Time: Comparison with Pulse/Echo Technique
the error in discontinuity dimensional measurement at higher tem- It is well known that time of flight diffraction has a faster scan-
peratures. The baseline data were measurements at ambient tem- ning time compared to the pulse/echo technique. In the present
perature. The actual dimensions in the case of the reference notch case, the 200 mm long weldment could be scanned by time of flight
(length and width) were determined using a 3D coordinate mea- diffraction within 10 s after initial setup, while in the case of
suring machine with a least count of 1 mm. In the case of slag and pulse/echo, the scanning could be completed only in 120 s. Thus, to
lack of penetration in welds, radiographic images were used to gen- scan a 3 m weld in the main vessel or safety vessel, time of flight dif-
erate the baseline data. The radiographs of the welds were digitized fraction would require about 3 min while pulse/echo would re-
using an industrial X-ray film digitizer (Class DA as per ISO 14096) quire about 30 min. This means more prolonged exposure of the
at 600 dpi and 12 bits. Digitized radiographs were subjected to con- probes and electronics to radiation and temperature, which can af-
trast stretching and edge enhancement, and through profiling, ap- fect the characteristics/properties of the probe as well as shorten its
propriate calibrations and appropriate corrections for geometric life (and that of associated electronics).
unsharpness and inherent magnifications, the dimensions of the Overall, it can be clearly stated that time of flight diffraction
lack of penetration and slag were determined. would be a better alternative to conventional pulse/echo tech-
It can be observed from Table 2 that, for the reference notch, niques for inservice inspection of the welds. However, one precau-
the maximum deviation observed was 0.38% compared to the tion to be observed while scanning using time of flight diffraction is
physical dimension at high temperature. Table 3 summarizes the that it should be ensured that in each repeat scan, the same line is
results of measurements by radiography, time of flight diffrac- followed by the transducers. Any deviation either in the scanning
tion and pulse/echo testing. It can be observed that the error in line or probe positioning would result in increased error. Once a
determination of the discontinuity size is much less in the case of discontinuity is detected, it is recommended that a repeat scan be
time of flight diffraction compared to the pulse/echo technique. performed such that the discontinuity is lying midway between the
Higher variation in sizing by pulse/echo testing, especially in the two transducers, so that the time taken by the diffracted signals to
reach the receiver will be minimal.
A second precaution is that as the temperature increases, there
could be loss of couplant due to evaporation, which will reduce
Table 2 Percentage variation in sizing of notch by time of flight the transduction of the ultrasonic energy into the material. Hence,
diffraction at different temperatures with respect to the actual a couplant reservoir should be maintained during actual inservice
physical dimension (found to be 10.49 mm) inspection.

Temperature (K) Sizing (mm) Variation (%) CONCLUSION


298 10.50 +0.09 This study clearly establishes the fact that time of flight diffrac-
323 10.49 0.00 tion is better suited for inservice inspection of stainless steel weld-
348 10.51 +0.19 ments at high temperatures, due to its better discontinuity charac-
373 10.51 +0.19
terization capabilities and its faster scanning times compared to
398 10.49 0.00
423 10.53 +0.38 pulse/echo techniques. Faster scanning times also mean less expo-
sure of probes and other hardware to higher temperatures and ra-
diation, thus leading to longer lives for these components.
The main challenge in the inservice inspection of reactor vessel
Table 3 Linear discontinuity dimensions by various techniques* welds would be the high temperatures likely to be encountered,
which can lead to degradation of ultrasonic signals. Analyses of the
LP 1 LP 2 Slag 1 Slag 2 experimental results clearly reveal that the ultrasonic signal ampli-
Linear dimension (mm)
Radiography 17.00 32.00 36.00 30.00
tude degrades rapidly, varying nonlinearly with increasing temper-
TOFD 16.98 32.08 37.73 31.31 ature. With additional gain, this decrease in signal amplitude can be
Pulse/echo 15.00 30.00 25.00 22.00 compensated for. For the case of 25 mm thick weldments at 423 K, it
is found that a total gain of 65 dB (+15 dB over ambient) is required.
* LP = lack of penetration; TOFD = time of flight diffraction This image contrast is quite adequate to quantitatively characterize

Table 4 Linear discontinuity dimensions (mm) and variation (%) in time of flight diffraction sizing with respect to radiography data*

Temperature (K) LP 1 Variation LP 2 Variation Slag 1 Variation Slag 2 Variation


298 16.98 0.12 32.08 0.25 37.73 4.80 31.31 4.36
323 16.98 0.12 32.08 0.25 37.25 3.47 31.36 4.53
348 16.66 2.00 32.35 1.09 37.35 3.75 31.38 4.60
373 16.98 0.12 32.08 0.25 37.60 4.44 31.37 4.57
398 16.63 2.18 32.06 0.19 37.25 3.47 31.38 4.60
423 16.63 2.18 32.07 0.22 37.25 3.47 31.38 4.60

* LP = lack of penetration

336 Materials Evaluation/March 2008


the discontinuity features with the same accuracy, as in the case of REFERENCES
images acquired at ambient temperature. ASME, ASME Boiler and Pressure Vessel Code, New York, American Society of
Mechanical Engineers, 2007.
ACKNOWLEDGMENTS Brunk, John A., “The Influence of Ambient Temperature Changes on Angle-
Beam Ultrasonic Testing of Steel,” Materials Evaluation, Vol. 46, 1988, pp.
The authors are grateful to Y.C. Manjunatha, director of the En-
1148–1153.
gineering Services Group of the Indira Gandhi Centre for Atomic Canella, G. and F. Monti, “Ultrasonic Inspection of Hot Thick Steel Prod-
Research, for his keen encouragement and constant support. The ucts,” NDT International, Vol. 13, 1980, pp. 10–14.
authors also thank all colleagues involved in the fabrication of the Darbari, G.S., R.P. Singh and G.S. Verma, “Ultrasonic Attenuation in Carbon
probe holder at the Fabrication and Machine Shop Sections of the Steel and Stainless Steel at Elevated Temperatures,” Journal of Applied
Central Workshop Division, and all colleagues of the NDT Sec- Physics, Vol. 39, 1968, pp. 2238–2245.
tion, Quality Assurance Division, for their valuable experimental Scruby, C.B. and B.C. Moss, “Non-contact Ultrasonic Measurements on Steel at
support. Elevated Temperatures,” NDT&E International, Vol. 26, 1993, pp. 177–188.

Materials Evaluation/March 2008 337

You might also like