Professional Documents
Culture Documents
P. R. Blackburn
Structural Evaluation of Seamless,
Steel Tubes in Compressed Gas
M. D. Rana
Consultants,
Service
Linde Division, Tube trailers are used in the distribution of industrial gases. The tubes, which are
Union Carbide Corporation, seamless steel, measure 22-in. diameter, about 1/2-in. wall, and 34ft long. The U. S.
Tonawanda, N.Y. 14151 Department of Transportation mandates periodic re-tests of tubes every 5 yr.
Acoustic emission re-tests have been substituted for hydrostatic re-tests. This paper
presents an analysis of flaws using fracture mechanics methods and establishes a
maximum allowable flaw depth. It describes the acoustic emission test method and
presents some typical results. Also, some discussion of the hydrostatic re-test is
included.
Introduction
Seamless steel tubes are used in the distribution of industrial closed and filled with water; and the tube is pressurized.
gases. Often they are mounted on truck trailers for over the Typical test pressure is about 4000 psi (27.5 MPa). During
road transportation (Fig. I). The work described in this paper pressurization the tube expands; the expansion is observed by
pertains to two types of tubes: U. S. Department of Transpor- measuring the water displaced from the jacket with a buret
tation (DOT) specification 3AAX tubes and specification 3T tube.
tubes [1]. These seamless tubes are fabricated from quenched A tube will fail the hydrostatic test if nominal wall stress, as
and tempered, modified AISI 4130 (3AAX) and 4140 (3T) determined from measured elastic expansion, exceeds
type steels. The chemistry, heat treatment, and typical prescribed limits [1]. The relationship between wall stress limit
mechanical properties of the tubes are listed in Table 1. and elastic expansion is [2]
Typical dimensions are 22 in. (55.8 cm) o.d. with minimum 1.7 EE
wall thicknesses of 0.536 and 00415 in. (1.36 and 1.05 cm) for a=----OAp (I)
3AAX and 3T tubes, respectively. Usually they are about 34 ft VoK
(10.3 m) long. Service pressure is typically 2640 psi (18.1 where
MPa). Normally such tubes are hydrostatically re-tested every EE = elastic expansion, cm 3
five years at 1.67 times rated pressure. V o = internal volume (139,000 in. 3 )
Since early 1983, acoustic emission testing has been used, by K = factor (1.296 x 10- 7 )
Linde, in lieu of hydrostatic testing to meet DOT mandated re- (J = wall stress (87,000 psi)
test requirements. This change in re-test method has provided p = test pressure (4100 psi)
much more information, which relates to the structural in-
tegrity of tubes, than did the hydrostatic method. Also, a
substantial reduction in operating cost is realized with the
acoustic emission re-test method.
The method of acoustic emission testing, some results
(which include flaw qualification by ultrasonic inspection),
and structural evaluation of the flawed tubes are presented in
this paper.
Hydrostatic Re-Test
Hydrostatic testing is performed in a water jacket. The
trailer is disassembled; individual tubes are filled with water;
they are placed in the water jacket (usually with the
longitudinal axis in a vertical orientation); the water jacket is
Contributed by the Pressure Vessels and Piping Division for publication in the
JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received by the
Pressure Vessels and Piping Division, February 26, 1985; revised manuscript
received May 9, 1985. Fig. 1 Tube trailer
C Mn P(max) S(max) Si Cr Mo
3AAX 0.25/0.35 0.40/0.90 0.04 0.05 0.20/0.35 0.80/1.10 0.15/0.25
3T 0.35/0.50 0.75/1.05 0.035 0.04 0.15/0.35 0.80/1. I5 0.15/0.25
Heat treatment
3AAX Quenched in oil or other suitable medium and tempered at greater than or equal to 1000°F.
3T Quenched in oil or other suitable medium and tempered at greater than or equal to 1050°F.
Mechanical properties
Ultimate Elongation
Yield tensile in 2-in.
strength, strength, gage length,
ksi ksi percent
3AAX 90/106 105/125 20
3T 115/130 135/155 16
where
/ = wall thickness, in.
oy = yield strength, ksi
Using the toughness value of 120 ksi Vrru (132 MPa Vm) and
equation (2), the critical-flaw-size calculations were made and
results are presented in Fig. 5. As can be seen from Fig. 5, the
0.2 0.3
fracture critical-flaw-depth is close to the through thickness at
CRITICAL FLAW DEPTH, a, INCH the DOT specified, maximum allowable service stress of 57.4
Fig. 5 Stress versus fracture critical flaw depth, 3T tube ksi, (396 MPa), thus indicating leak-before-break per-
formance characteristic of this tube.
Figure 6 shows the fatigue-crack-growth response of a
CRITICAL FLAW SIZE typical 3T tube in the presence of various initial flaws at cyclic
r a°=-41" / stresses of 57.4 ksi, (396 MPa). Equations (2) and (3) were
-Bi-.125"/
/ a,, INITIAL CRACK / used in the fatigue-crack-growth calculations.
/ DEPTH /
0.3- *i - .06" .
— a i = .03" /
Acceptable Flaw Size Determination
0.2 H j£— =i = .oi"
As presented in Figs. 3 and 5, the critical-flaw-sizes of the
3AAX and 3T tubes are fairly large (almost through
0.1-
thickness). Therefore, relatively small flaws do not com-
• ^ ^ .
promise the tubes structural integrity. However, because of
the cyclic pressurization of these tubes, small flaws could grow
1
0 1,000 2,000 3,000 4,000
1
5,000 6,000
i i
7,000
i
8,000 3,000
i
10,000
(Figs. 4 and 6) to a critical size. Therefore, it is necessary to
NUMBER OF CYCLES, N establish a limit on the flaw size, so that the flaw will not grow
I 1 , , , , 1 1 1 , 1
0 10 20 30 40 50 60 70 80 90 100 to a critical size within the 5-yr re-test period. These tubes are
NUMBER OF YEARS typically subjected to 100 pressure cycles per year in the Linde
Fig. 6 Fatigue crack growth versus pressure cycles, 3T tube distribution network.
Use of a safety factor in structural integrity assessment of
pressure vessels via a fracture control program is not uncom-
mon. In this analysis, a safety factor of two has been assigned
to the calculated fracture critical-flaw-sizes of the 3AAX and
3T tubes. Using Figs. 4 and 6 data, the number of fatigue
cycles to reach critical flaw size (for different initial flaw sizes)
are presented in Fig. 7.
As shown in Fig. 7, the 3AAX tube, in hydrogen service, is
the limiting case; and, for the proposed 5-yr re-test period, the
initial acceptable flaw size (measured during AE re-test) is less
than or equal to 0.106 in. (2.7 mm). Although flaws larger
than 0.106 in. (2.7 mm) could be tolerated in 3T tubes, the
same rejection limit is used by Linde for both 3AAX and 3T
tubes.
Thus, based on the foregoing fracture analyses, 3AAX or
3T tubes with flaws greater than 0.10 in. deep, (as detected by
the AE test method and measured by ultrasonic inspection) are
removed from service.
A E Test Method
With AE re-testing, the trailer can be tested while it is filled
with product, at a production facility. It need not be removed
from service.
Immediately prior to a re-test, each tube is instrumented; a
piezoelectric sensor is placed at each end (Fig. 8). Sensors are
attached with silicone adhesive; they are acoustically coupled
to the tubes with silicone grease. Sensors are resonant at
140-150 KHz. Signals are filtered with 100-350 KHz bandpass
INITIAL FLAW SIZE a,, INCH filters, and processed with a multi-channel system. System
Fig. 7 Years to reach fracture critical flaw depth versus initial flaw threshold is 40 dB (referenced to 1 microvolt at the sensing ele-
depth for 3AAX and 3T tubes ment). The 40 dB threshold is obtained by using 40 dB of
i i i i i 1
6 in. (15.2 cm)
4
H ~|
Z
UJ .
ILL
2
J_L
9 10 11 12
01234567 0.00 X A X I A L LOCATION - INCHES
FILTER: LOCATION FOR ARRAYS
FILE: UC1846A. 10/19/83 11:26:11 5 Y R RETEST TRLRS/N Fig. 10 Flaw depth and event count versus axial location
Fig. 9(a) Acoustic emission event count versus axial location
/ \
OQ
A
fe
A
I t I
I I
05
1.25 in.
" (3.2 cm) ~
\- 1.5 in. -| -J I- L-0.8 in J
Fig. 9(b) Flaw depth and event count versus axial location Fig. 11 Flaw depth and event count versus axial location
preamplification, with 40 dB of post amplification, together distribution) is shown in Fig. 9(a). This type of display is nor-
with a 1-V threshold at the analog to digital converter. More mally monitored during re-tests. Units of the horizontal axis
recently, sensors with integral preamplifiers were used. Trailer are inches. Inspection of Fig. 9(a) shows that one location had
pressure is measured with a transducer on the trailer manifold, produced 4 events. This location was subjected to ultrasonic
and is recorded by the AE system. inspection; and a lap, on the outside surface, 1.25 in. (3.2 cm)
The trailer is filled to 110 percent of normal fill pressure long was found.
(whereas, in a hydrostatic test the pressure is 167 percent of Figure 9(b) shows a plot of the depth profile (as determined
rated pressure). This permits re-testing at a pressure above with UT) and immediatgely below a magnified location
that which occurred during the previous 5 yr of service. distribution. Depth is plotted as percent of the "distance
Following the re-test, the excess product is bled to a receiver; amplitude correlation value" (DAC). Actual depth of 0.10 in.
and the trailer returns to service at normal fill pressure. corresponds to 100 percent DAC. When the depth is less than
If sufficient AE events are detected from a specific location, 0.10 in. (100 percent) we have a relative (or approximate)
that location is noted; and, ultrasonic inspection of that loca- depth measurement; but, we do not have a good quantitative
tion is performed (with a shear wave angle beam). Flaws at depth measurement. The AE location distribution cor-
that location are measured (with UT); and, if they are less than responds extremely well with the ultrasonic result in that the
0.10 in. (2.54 mm) deep, the tubes pass the re-test (i.e., they deepest portion of the flaw produced the most AE events.
are returned to service for another 5 yr). Figure 10 shows the depth profile and location distribution
for a crack (on the inside surface) 6 in. (15 cm) long in a tube
which was in oxygen service. Again the deepest portion of the
A E Test Results crack produced the most AE events.
Figure 11 shows similar plots for a tube which was in
An exemption from the requirement to perform hydrostatic hydrogen service. Here within a section of tube, 9 in. (23 cm)
re-testing was granted by the U. S. Department of Transporta- long, there were 6 AE events; and, ultrasonic inspection
tion in March 1983. Since that time hundreds of tubes have showed three small flaws on the inside surface. These flaws
been tested. Results from these tests are being incorporated in- were separated from one another by about 11 in. in the cir-
to a data base. Some examples are discussed in the cumferential direction. Here, it is not completely obvious
forthcoming. which events are associated with which flaws. However, emis-
A plot of AE events versus axial location (a location sion was observed; and, ultrasonic inspection of the area
CH2&4TUBE 2
CH5&6TUBE3
i±=i±S=i±2=± wij9mm»»iJlmmAm
CH 7 & 8 T U B E 4
iiii|iiMnWM|algB •
' CZZZ3C IE=E
E V E N T S = 179
-I 1-
MESSAGES = 10
GRAPH 05 OF 20 Miii^J&im~~*i&&mmwmt*mmJLIm—mArB^Brm-pm£b„
•^^^^miniiiMiiiiMui^^^^BOTaMBmmiiMiinmM nuJff MWW^^MM
Fig. 12 Acoustic emission event count versus axial location for four tubes
10-
CH 1 & 2 T U B E 1 2 " X .02" -,
I N T E R V A L SUM
EVENTS
FIRST A R R I V A L
06.
04-
SURFACE
ROUGHNESS -i
\|
I 1
Iill ll
A "
E V E N T S = 179
02-
V
MESSAGES-10
I \
1 111 11 1 J^l II
G R A P H 0 1 O F 20
1, 1 1
0 20 40 60 80 100
Location (% o f distance between sensors)
A E E V E N T S versus LOCATION
Fig. 13 Acoustic emission event count versus axial location for one tube. Some results of ultrasonic in-
spection are indicated.
Table 2
6,000 Cumulative
Maximum threshold Distance from
4,000
depth crossing No. of sensors
(Percent DAC) counts AE events in. (m)
r
j 1 distribution for tube 1 (this distribution is also shown on the
top portion of Fig. 12).
r~ iJj
! i
tion). Table 2 shows data produced by three different, 3/4 in.
(1.9 cm) long, flaws, all in 3AAX tubes. These data were ob-
tained while the tubes were pressurized from 1800 psi (12.4
MPa) to 2900 psi (20 MPa), with hydrogen. The AE system
N OBMAL SERVICE i threshold was 40 dB (referenced to 1 mv at the sensor).
I1
PRESS URE
1 r The number of counts is directly proportional to the flaw
1,600 1,800 2,000 2,200 2,400 2,600 2,800 3,000
PRESSURE-PSI
depth. These numbers are not corrected for attenuation
(which increases with distance from the sensor). The number
Fig. 14 Threshold crossing counts versus pressure, 3/4 in. long flaws,
in 3AAX tubes, in hydrogen service
of counts produced by a wave packet will vary with distance
between the source and the sensor. This occurs because the
wave will convolve and will be attenuated.
which had produced emission showed that notable flaws were Data from Table 2 is plotted in Fig. 14 as counts versus
present. pressure. The smallest flaw had produced four events; and,
Figure 12 shows location distributions from four tubes. This two of these events had occurred at pressures which are below
plot is equivalent to Fig. 9(a), but was obtained with a dif- normal service pressure. This is typical for data obtained from
ferent, signal processor. Here, AE events are plotted versus ax- these tubes. Usually significant emission occurs at relatively
ial location for four tubes. The distribution at the top (tube 1) low pressure. Measurable emission, from flaws in these tubes,
shows many events; a large flaw was present near one end (at can originate from crack propagation, from rubbing at crack
97 percent of the distance between sensors). Some ultrasonic surfaces, or from fracture of (or rubbing of) oxide scale con-
inspection was performed on these tubes and results are noted tained within a flaw.
on Fig. 13. Figure 13 is an enlarged plot of the location Figure 15 shows a magnifield cross section from a large flaw
Acknowledgments
Fig. 15 Section through flaw In 3T tube. Maximum depth (normal to This project benefited from contributions of many people
surface) In this secllon Is 0.056 In.
within Union Carbide Corporation. Particularly noteworthy
in a 3T tube. Solid oxides (apparently mill scale) fill the in- are: Mr. R. O. Tribolet for management support, Mr. S.
terior of this fold, which was located on the outside surface. Hoffman for liaison with regulatory agencies, and Mr. B. H.
There was substantial emission from this flaw. The section Clark for the ultrasonic inspection technique. Also, the
shown in Fig. 15 shows no crack growth. The measured emis- Dunegan Corporation provided field test service and other
sion was apparently from the oxides contained within the support. And, without the cooperation of the Research and
flaw. Special Programs Administration, U. S. Department of
The re-test method described herein is based upon event Transportation we could not have so rapidly implemented this
count only. Association of flaw characteristics (e.g., length application of AE testing.
and depth) with other AE parameters is possible only if wave
attentuation and convolution (which depend upon distance
from the sensor) are taken into account; and, if consistant References
coupling of the sensors to the tubes is maintained. Coupling is I "Hazardous Materials Regulations of the Department of Transporta-
affected by paint on the tube (thickness and adhesion), by tion," 49 Code of Federal Regulations, Sections 178.37 and 178.45, U. S.
coupling medium (and variation of its characteristics with am- Government Printing Office, Washington, DC, 1983.
2 "Cylinder Service Life, Seamless High Pressure Cylinders," Pamphlet
bient temperature), and by the force with which the sensor is C-5, Compressed Gas Association Inc., Arlington, Va.
attached to the tube (coupling medium thickness). 3 Drouillard, T. F., Acoustic Emission, A Bibliography With Abstracts,
It is pointed out that the aspect (length to depth) ratios of Plenum Publishing Corporation, New York, 1979.
the flaws shown in Figs. 9(b), 10 and 11 are higher than that 4 ASME Pressure Vessel Code, Section XI, 1983 Edition, Article A-3000.
5 Loginow, A. W., and Phelps, E. H., "Steels for Seamless Hydrogen
used in the analysis. However, because the ratio of flaw depth Pressure Vessels," Corrosion-NACE, Vol. 31, No. 11, Nov. 1975, pp. 404-412.
to wall thickness is small, the large aspect ratios do not affect 6 Eagle, R. M., Jr., "Aspect Ratio Variability in Part-Through Crack Life
the crack tip stress-intensity, and hence, the crack growth rate. Analysis," Part-Through Crack Fatigue Life Prediction, ASTM STP-687, 1979,
It is also expected that crack growth by fatigue will initiate pp.74-88.
7 Barsom, J. M., "Fatigue Behavior of Pressure-Vessel Steels," WRC
from a weak point within such long defects; and, the active Bulletin 194, Welding Research Council, New York, May 1974.
flaw, at failure, will probably have, nominally, four to one 8 May, T. S., Stuber, A., and Rolfe, S. T., "Effective Utilization of High
aspect ratio. Thus, the detected sub-critical defects with large Yield Strength Steels in Fatigue," WRC Bulletin 243, Welding Research Coun-
aspect ratio do not affect the specified fracture critical-flaw- cil, New York, Nov. 1978.
9 Kesten, M., and Windgassen, K. F., "Hydrogen Assisted Fatigue of
size rejection limit. Periodically Pressurized Steel Cylinders," Third International Conference on
Effect of Hydrogen on Behavior of Materials, Jackson Lake Lodge, Wyoming,
Conclusions Aug. 1980.
10 "Standard Methods of Test for Plane-Strain Fracture Toughness of
The work described in this paper has shown that substitu- Metallic Materials," ASTM Standard E-399-8I, Part 10, ASTM Annual
tion of acoustic emission testing for hydrostatic testing has Standards.
II Irwin, G. R., Kraft, J. M., Paris, P. C., and Wells, A. A., "Basic Aspects
been very beneficial, both technically and economically of Crack Growth and Fracture," Naval Research Laboratory Report 6598,
beneficial. NTIS No. 663882, Nov. 1967, p. 38.